Social Comrehension and Judement

Social Comprehension and Judgment
The Role of Situation Models,

Narratives, and Implicit Theories
Social Comprehension and Judgment

The Role of Situation Models,
Narratives, and Implicit Theories
Robert S. Wyer, Jr.

Hong Kong University of Science and Technology
2004
LAWRENCE ERLBAUM ASSOCIATES, PUBLISHERS

Mahwah, New Jersey
London
This edition published in the Taylor & Francis e-Library, 2008.

To purchase your own copy of this or any of Taylor & Francis or Routledges
collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.
Copyright 2004 by Lawrence Erlbaum Associates, Inc.
All rights reserved. No part of the book may be reproduced in
Any form, by photostat, microform, retrieval system, or any other
means, without the prior written permission of the publisher.
Lawrence Erlbaum Associates, Inc., Publishers

10 Industrial Avenue
Mahwah, New Jersey 07430
Cover design by Kathryn Houghtaling Lacey
Library of Congress Cataloging-in-Publication Data

Wyer, Robert S., Jr.
Social comprehension and judgment : the role of situation models, narratives, and implicit
theories / Robert S. Wyer.
p. cm.
Includes bibliographical references and index.
ISBN 0-8058-4190-3 (alk. paper)
1.
Social perception. 2. Human information processingSocial aspects. 3. Judgment. 4.
Memory. I. Title.
BF323.S63W94 2003
153dc21
2003052865
ISBN 1-4106-0900-6 Master e-book ISBN
Books published by Lawrence Erlbaum Associates are printed on acid-free paper,

and their bindings are chosen for strength and durability.
Dedicated to
Rashmi, Kathy, and Natalie
and
The Social Cognition Group
Contents
Preface
xi
PART I: INTRODUCTION AND BASIC CONCEPTS

1.
INTRODUCTION
Types of Social Knowledge 6
The Representation of Knowledge in Memory 10
Situation Models 16
Generalized Event Representations 17
Generalized Entity Representations 22
2.
THE ARCHITECTURE OF THE INFORMATION-PROCESSING

SYSTEM
Limitations of Wyer and Srulls (1989) Theory 26
Structure of the Processing System 29
Storage and Retrieval Processes 37
Goal-Directed Versus Non-Goal-Directed Processing:
Spontaneous Reminding Processes 41
Concluding Remarks 45
3.
DETERMINANTS AND EFFECTS OF KNOWLEDGE

ACCESSIBILITY
General Considerations 47
Frequency and Recency Effects of Knowledge Activation
on Judgments and Decisions 49
Effects of Awareness 53
Effects of Knowledge Accessibility at Different Stages
of Processing 57
The Effects of Prior Judgments on Subsequent Ones 63
Determinants and Consequences of the Accessibility
of Goals and Motives 65
The Activation of Behavioral Dispositions 72
26
47
vii
viii
CONTENTS
PART II: COMPREHENSION PROCESSES

4.
THE COMPREHENSION AND VALIDATION OF INFORMATION

ABOUT FAMILIAR PEOPLE AND EVENTS: THE ROLE OF
SITUATION MODELS
The Content and Structure of Situation Models 80
The Role of Visual Imagery in Comprehending Social Events 83
Spontaneous Comprehension Processes 88
The Comprehension of Verbal Statements 92
Spontaneous Validity Judgments 98
79
5.
THE CONSTRUCTION OF EPISODE MODELS AND GENERALIZED

NARRATIVE REPRESENTATIONS
The Construction of Episode Models 106
The Construction of Generalized Narrative Representations 116
106
6.
THE IMPACT OF GENERALIZED EVENT REPRESENTATIONS ON

COMPREHENSION AND MEMORY
The Role of Prototypic Event Representations in the
Comprehension of Ones Own and Others Experiences 125
The Effects of Communicating About an Experience
on Memory and Judgment 137
124
7.
PRAGMATIC INFLUENCES ON THE INTERPRETATION

OF STATEMENTS MADE IN A SOCIAL CONTEXT
Responses to Uninformative Messages 155
Reactions to Communications That One Believes to Be Untrue
160
Pragmatic Influences on Impression Formation in Informal
Conversations 165
The Effect of Nonverbal Behavior and Conversational
Style on Impression Formation, Judgment, and Information
Seeking 178
Pragmatic Communication in Close Relationships 186
151
CONTENTS
8.
THE DYNAMICS OF HUMOR ELICITATION: THE EFFECTS

OF INFORMATIONAL CONTEXT ON THE INTERPRETATION
AND ELABORATION OF NARRATIVES
Historical Background 192
A ComprehensionElaboration Theory of Humor Elicitation 198
Cognitive Elaboration Processes 209
Reactions to Disparaging Humor 218
ix
189
PART III: INFERENCE PROCESSES

9.
CRIMES, VACATIONS, AND POLITICAL CANDIDATES:

THE CONSTRUCTION AND USE OF NARRATIVES IN SOCIAL
JUDGMENTS
The Impact of Narratives on Jury Decision Making 232
A Comparison of Narrative-Based Versus Nonnarrative-Based
Information Processing 234
The Influence of Narratives on Political Impression Formation
240
The Role of Visual Imagery on Verbal Information Processing:
Additional Considerations 249
General Effects of Narrative-Based Processing on Explanation
and Prediction 255
231
10.
THE IMPACT OF IMPLICATIONAL MOLECULES AND IMPLICIT

THEORIES ON INFERENCES ABOUT ONESELF AND OTHERS
Effects of Implicit Theories on Reconstructive Memory 274
Motivational Influences on Implicit Theory Construction
and Use 278
The Role of Implicit Theories in Heuristic-Based Judgments 280
Individual Differences in Implicit Theories 286
Implicit Theories of Social Support: The Effect of Perspective
on Theory Activation 288
Cultural Differences in Implicit Theories 290
The Role of Implicit Theories in Marital Satisfaction 295
Effects of Implicit Theories on Behavioral Decisions 299
Final Comment 305
267
CONTENTS
11.
THE ROLE OF AFFECT IN INFORMATION PROCESSING

Affect and Concept Accessibility 312
The Influence of Affect on the Recall and Interpretation
of Information 315
The Use of Affect as Information 322
Affective Versus Descriptive Bases for Judgment 325
Affect-Confirmation Processes 332
Adjustments for Bias 336
Automatic Influences of Affect on Information
Processing 339
Motivational Influences of Affect on Information
Processing 342
A Performance-Feedback Model of Affect as Information 344
Affect, Attitudes, and Behavior 358
Final Comments 363
306
12.
EPILOGUE: THE BOOK, THE AUTHOR, AND PHILOSOPHICAL

RUMINATIONS
365
APPENDIX: SUMMARY OF POSTULATES

REFERENCES
374
380
AUTHOR INDEX
SUBJECT INDEX
405
415
Preface
Authors often become disenchanted with the books they have written
almost as soon as they are published. Perhaps this is inevitable. Although
a book reports the culmination of ones theoretical and empirical work at
the time it goes to press, it nevertheless contains the seeds of discontent.
That is, the author in writing the book becomes painfully aware of the
limitations of his research and the conceptualization underlying it, and
consequently is already motivated to move beyond the confines of this
conceptualization by the time it appears in print.
At least, this has been true of me. A book I published in 1974, entitled
Cognitive Organization and Change, provided a fairly rigorous analysis of
cognitive consistency theories of the way beliefs and attitudes were
organized in memory, individual and situational differences in the
cognitive responses to new information, and the way in which different
features of information were combined to form judgments. The book
assumed that humans were analogous to computers, generating outputs
based on operations that were specified in various software routines
that were activated by the user (in the case of humans, the demands of
the social environment).
By the time the book was published, however, I had already begun to
appreciate the need for a more general conceptualization of mental
representation. A second book (Social Inference and Attribution, with Don
Carlston, published in 1979) proposed a rather complex associative
network model of social memory that took into account the role of
knowledge accessibility in judgments and decisions. In this book, we
recognized the impact that script-like representations of events could have
on social judgment. However, we did not explore this possibility with any
degree of rigor. More generally, the cumbersome nature of the network
representation, coupled with the fact that it provided little insight into
how the information represented was actually used, convinced me of the
futility of using this approach to account for many phenomena that were
being identified in research on person impression formation and the use
of judgmental heuristics.
xi
xii
PREFACE
The third book (Memory and Cognition in Its Social Context, with Thom
Srull) appeared in 1989. This book remedied several deficiencies of the
1979 volume, providing a comprehensive description of social information
processing that specified the activities at several different stages: the
interpretation, organization, and storage of information in memory, its
later retrieval and integration with other relevant knowledge to make a
subjective judgment or decision, and the translation of this subjective
response into an overt one. This rather grandiose formulation not only
generated specific predictions that received empirical support but had the
flexibility to integrate the implications of more specific theories of research
that had been developed.
By the time the book was published, however, I had become aware of
several obvious limitations. For one thing, the title of the book was a
misnomer. The theory I had proposed purported to describe the
comprehension and use of information of the sort that one encounters in
daily life, In fact, however, it was applied almost exclusively to the
processing of linguistic information of the sort people receive in laboratory
situations but are unlikely to encounter elsewhere (e.g., lists of unrelated
behaviors of a fictitious person, verbal descriptions of routine sequences
of events of little intrinsic interest, etc.) In short, although the book
purported to be about cognition in its social context, the social context
was typically absent in much of the research to which the theory was
applied.
Stimulated by our dissatisfaction with the applicability of the theory
to phenomena outside the laboratory, my students and I began to examine
the processing of information of the sort people actually encounter in the
course of their daily lives. This research took many directions. For example,
we investigated the impressions people formed on the basis of a persons
opinions about social policies with which they agreed or disagreed rather
than abstract behaviors that had few implications for matters of personal
concern to the recipients. Moreover, we studied the impressions people
formed on the basis of information conveyed in conversations rather than
written lists of unrelated behaviors described out of context. Moreover,
we explored the way individuals respond to information that violated
social norms of communication and that was conveyed for the purpose of
eliciting amusement rather than to convey an impression of its referent or
to espouse a particular point of view. In doing so, we began to take more
seriously the fact that information conveyed in social situations was
frequently in the form of a narrative, or story, and was often accompanied
by visual images that were either based on pictures or direct observation
or generated by the recipients themselves. Finally, we explored the role of
affect in information processing.
PREFACE
xiii
This book provides a conceptual integration of this work. It proposes

a general theoretical formulation of the way that the sort of information
acquired in the course of daily life is comprehended and represented in
memory, and how it is later used as a basis for judgments and behavioral
decision. In doing so, it takes into account both the spontaneous
comprehension of information about specific persons and events and the
more deliberative, goal-directed interpretation of information that occurs
when information is acquired in a social context. In addition, it considers
not only the representation of this information in memory but also the
way the information is later used as a basis for judgments and decisions.
A major emphasis throughout the volume is on the construction and use
of narrative representations of knowledge and the way visual images
influence the formation of these representations and the judgments that
are based on them. The role of affective reactions in this cognitive activity
is also discussed.
The book is divided into three sections. After a conceptual overview
of the approach to be taken (chap.1), it outlines the general theoretical
framework within which the discussion is housed, focusing on
assumptions about the storage and retrieval of information (chap. 2) and
reviews recent research on the impact of knowledge accessibility on
judgments and decisions (chap. 3). Although the architecture of the
information-processing system we propose is similar to that described in
the 1989 volume, the conceptualization of memory processes provided
here is quite different.
The second section deals specifically with the comprehension of
information. Chapter 4 discusses the mental simulations that are formed
of situationally and temporally specific events (i.e., event models) in
the course of comprehending descriptions of familiar persons and events,
and specifies the conditions in which information that stimulates these
simulations is spontaneously recognized as true or false. Chapter 5
discusses both the representation of situation-specific sequences of events
that are formed spontaneously in the course of comprehending the events
and more abstract representations that constitute implicit theories about
the causal and temporal relations among events of the sort people
encounter in daily life. Chapter 6 discusses the manner in which
generalized narrative representations influence memory for the specific
experiences that exemplify them. In this context, the influence of
socialization practices on the construction of narratives about oneself, and
on the self-concepts that result from these representations, are considered.
The next two chapters discuss cognitive reactions to information that
violates either normative expectations for the type of information that is
conveyed in the social context at hand (chap. 7) or expectations that result
xiv
PREFACE
from previously acquired information about the informations referents

(chap. 8). In the latter context, we consider the effects of expectancy
violations on the elicitation of humor and the extent to which the cognitive
activities that occur in response to these violations can increase or decrease
the humor that is elicited in a social context.
The final section describes the inferences that are based on information
of the sort that is conveyed in social situations. Chapter 9 outlines the
way visual characterizations of events (either pictures or self-generated
images) influence the impact of verbal descriptions of the events. Chapter
10 discusses the effects of peoples implicit theories about their social
world on the inferences they draw from specific experiences. Finally,
chapter 11 analyzes the role of affective reactions in information
processing and the judgments and decisions that result from this
processing.
Many individuals have contributed both directly and indirectly to the
content of this volume. My consideration of the role of narratives in
information processing was stimulated in large part by a multisemester
workshop conducted at Illinois by Julian Rappaport. G. A. Radvanskys
empirical and theoretical work on the nature of situation models was also
a major stimulant. My conceptualization of the role of affect in information
processing was profoundly influenced by my participation in an ongoing
seminar on emotion and cognition conducted by Jerry Clore. Clores work
and that of his students are well represented in this volume.
However, I am particularly indebted to the University of Illinois Social
Cognition group. This group, an informal consortium of faculty,
postdoctoral students, and graduate students from several disciplines, met
weekly over a 20-year period to discuss theoretical and experimental issues
in the general area of information processing. Members of this group, both
individually and collectively, have had a profound influence on the ideas
represented in this volume and on my professional career more generally.
I am particularly indebted to two subsets of individuals whose work is
widely cited in this volume. Galen Bodenhausen, Lee Budesheim, Bob
Fuhrman, Deborah Gruenfield, Alan Lambert, Victor Ottati, Jong Won Park,
L. J. Shrum, and Dave Trafimow, participants in the Group during the
period in which my views on social information processing were rapidly
changing, were not only inspirational colleagues but close friends, and
are largely responsible for many of the changes in my thinking that
occurred. The stimulation provided in later years by Dolores Albarracin,
Stan Colcombe, Carol Gohm, Linda Isbell, Eric Mankowski and Michaela
Wnke was equally influential. The research and ideas of all of these
individuals are readily apparent throughout this book. I should also
acknowledge the contributions of Donnel Briley, Catherine Yeung, and
PREFACE
xv
Candy Fong, who worked with me at the Hong Kong University of Science
and Technology, whose work is also represented.
I owe a special note of appreciation to Rashmi Adaval. Both as a
collaborator and as a wife, she has contributed in both tangible and
intangible ways both to this book in particular and to my personal and
professional life more generally. Her research on both the role of narratives
in memory and judgment, and on affect and information processing, is
prominent throughout this book. However, her personal support and
encouragement has been equally invaluable. It is extremely difficult for
anyone to be both a colleague and a spouse. Her willingness to play these
two often conflicting roles, her tolerance for my idiosyncratic work habits,
and her love and support throughout the past decade, has been
inexpressibly gratifying.
Finally, I want to acknowledge my deep appreciation to Larry Erlbaum
and his staff at Lawrence Erlbaum Associates. The support he has given
me over the past 30 years has been a constant source of gratitude. Larrys
commitment to excellence, and his sincere dedication to the enhancement
of basic research in psychology, have won him the respect and admiration
of all psychologists who have had the opportunity to work with him. I
can only hope that this volume confirms his confidence in me and
approximates the standards of excellence for which he strives.
Robert S. Wyer, Jr.
P A R T
I
INTRODUCTION AND
BASIC CONCEPTS
C H A P T E R
1
Introduction
Human memory is a collection of thousands of stories we remember

through experience, stories we remember by having heard them, and stories we remember by having composed them. Any story in memory could
have gotten there in one of these three ways. The key point is that, once
these stories are there in our memory, we rely upon them for all that we
can say and understand. . . .
Schank and Abelson (1995, p. 3)
Schank and Abelson (1995) argued that virtually all the meaningful social
knowledge we acquire is in the form of stories. This claim is probably overstated (Brewer, 1995; Rubin, 1995). Nevertheless, the role of narratives in
the acquisition and transmission of information in daily life is undeniable.
The content of everyday conversations almost invariably includes stories
we tell about ourselves and personal acquaintances, descriptions of books
we have read or movies we have seen, and jokes that we tell to one another. Narratives are also invoked in the course of understanding the
causes or likely consequences of real and hypothetical social events (e.g.,
the outcome of a presidential election, or the IndiaPakistan dispute over
Kashmir). Persuasive messages and television commercials often stimulate
us to imagine the sequence of events that might result from taking a particular course of action. Life itself is in the form of a narrative, consisting of a
sequence of temporally related events that we experience as either participants or observers. In short, narratives are fundamental to an understanding of ourselves and of the world in which we live.
The importance of narrative forms of knowledge is recognized in virtually every area of psychology. Research and theory on prose comprehen3
CHAPTER 1
sion and learning have long been concerned with the comprehension and
mental representation of stories and temporally ordered sequences of
events (Graesser, 1981; Mandler & Johnson, 1977; Stein & Glenn, 1979). Developmental psychologists (e.g., Miller, 1994; Nelson, 1993) have identified
the role of stories in parentchild interaction and socialization more generally. Narrative forms of knowledge can often constitute implicit theories
that people use both to explain their own and others past experiences and
to predict the future (Dweck & Leggett, 1988; Ross, 1989). The influence of
these theories on behavior and judgments has been identified in research
on topics as diverse as personal achievement (Dweck, 1975), predictions of
future life events (Oishi, Wyer, & Colcombe, 2000), marital relations
(Holmes & Murray, 1995; Murray & Holmes, 1996), personality development
(McAdams, 1988) and communityindividual relations (Mankowski &
Rappaport, 1995).
Despite the consideration of narratives in these specific areas of research, few attempts have been made to develop a broader theoretical formulation of social information processing that permits their influence to be
conceptualized in relation to that of other types of knowledge that people
encounter in their daily lives. Several types of knowledge representations
are likely to exist (Pennington & Hastie, 1993), and their role in information
processing can differ. One must understand the factors that govern the retrieval of this knowledge, how it is used to comprehend and construe the
implications of new information to which it is relevant, and the nature of its
influence on judgments and behavioral decisions about its referents.
This book attempts to provide this understanding. In particular, it is concerned with the comprehension and use of information of the sort people
encounter in the course of their daily lives. This information can be obtained through direct experience, from movies and television, from reading
newspapers, or from conversations with friends or strangers. The information can be transmitted in writing but also visually or acoustically. Moreover, it can be conveyed in several modalities simultaneously. Finally, although the information can sometimes concern hypothetical people and
situations, it more often refers to actual persons and events about which recipients already have substantial knowledge. Finally, the information is often conveyed in a social context, and can elicit affective or emotional reactions toward either its referent or, in some cases, the communicator. Much
of this information, particularly that which is acquired through direct experience, is necessarily transmitted in the form of a narrative, that is, a temporally related sequence of events.
The conceptualization we propose addresses the way this information is
comprehended and represented in memory, and how it is later recalled and
used in making judgments and decisions. Our focus on real-world information processing does not imply that the research we bring to bear on it
INTRODUCTION
has not been conducted in controlled laboratory settings. In fact, most of

the research reported is of this variety. However, the type of information of
concern in this research, the referents of the information, the context in
which it was presented, and the type of judgments and decisions for which
it had implications, is similar to that encountered in daily life situations.
The results of this research, and the formulation we bring to bear on it, are
consequently more likely to generalize to nonlaboratory situations than
work we have done in the past.
In this introductory chapter, we review the different types of knowledge
people acquire in their daily lives, and discuss briefly the various ways this
knowledge is represented in memory. Chapter 2 describes the theoretical
formulation of information processing within which the research and theory discussed later is conceptualized. This formulation has features in common with the earlier formulation proposed by Wyer and Srull (1989). However, it is modified in several important ways that permit it to account for
the comprehension and use of information about familiar persons and
events of the sort we acquire in daily life. Chapter 3, another introductory
chapter, discusses the determinants and effects of knowledge accessibility,
which is central to much of the theory and research presented in later
chapters.
Part II of the book is devoted to the representation of knowledge about
social events and its role in comprehension. Chapters 4 and 5 examine the
sorts of representations that people form specifically in the course of comprehending social events that they either read about or directly experience.
Chapter 5 also discusses the more abstract event representations that are
formed in the course of deliberative goal-directed activity. Chapter 6 addresses the extent to which generalized event representations can influence the representations that are formed of situation-specific occurrences,
and the impact of forming abstract representations of an experience on
later memory for the experience. Chapter 7 considers the effects of the situational context in which information is presented on its comprehension.
Finally, chapter 8 applies the conceptualization developed in previous
chapters to a phenomenon that has received surprisingly little attention in
research on social information processing but is a central ingredient of
communication outside the laboratory: namely, the reactions to humor.
Part III of this volume examines the use of social information to make judgments and behavioral decisions. Chapter 9 focuses specifically on the way
verbal and visual information combine to affect judgments under conditions
in which narrative representations of information are and are not likely to be
constructed. Chapter 10 considers more generally the impact of narrative
representations of knowledge on judgment, focusing largely on the impact of
implicit theories that people construct and use as a basis for inferences
CHAPTER 1
about specific events. Finally, chapter 11 addresses an issue largely ignored

in previous chapters but nevertheless of central importance in understanding the effects of information in situations outside the laboratory, namely, the
impact of affective reactions on judgments of the persons, objects, and
events to which these reactions are (or, in some cases, are not) relevant.
The conceptualization proposed in this volume has been largely stimulated by research conducted in our own laboratory during the past 15
years. However, its implications for phenomena identified in other areas
are considered as well. We therefore hope that the conceptualization provides a perspective on the dynamics of information processing in everyday
life situations that is not only useful in integrating a variety of known phenomena but suggests new directions for research and theorizing.
TYPES OF SOCIAL KNOWLEDGE

Social knowledge can vary in terms of its referent (what the knowledge is
about), its modality (verbal, visual, or affective), the generality of its implications, and the way in which it is acquired (through direct experience, or
through cognitive operations that are performed on other cognitive material). Two types of knowledge should be distinguished at the outset. Declarative knowledge concerns the referents of our everyday life experience
(e.g., persons, objects, events, social issues that we read about, or oneself).
In contrast, procedural knowledge concerns the sequence of actions that
one performs in pursuit of a particular goal (driving a car, using a word
processor, etc.). Whereas declarative knowledge is reflected in the information we can recall about an entity or that we implicitly draw on in the
course of attaining a particular objective, procedural knowledge is reflected
in the sequence of cognitive or motor acts that are actually performed in
the pursuit of this objective. People can of course have declarative knowledge about how to attain a particular objective, and might sometimes consult this knowledge for use as a behavioral guide. Once the procedure is
well learned, however, it may often be applied automatically, with little if
any conscious cognitive mediation.
These automated procedures can be conceptualized as productions of
the sort suggested by J. R. Anderson (1982, 1983; see also E. R. Smith, 1990).
Thus, they may metaphorically have the form of If [X], then [Y] rules in
which [X] is a configuration of perceptual or cognitive stimulus features
and [Y] is a sequence of cognitive or motor acts that are elicited automatically when the eliciting conditions are met. These productions, which are
acquired through learning, are strengthened by repetition, and can ultimately be activated and applied with minimal cognitive mediation. The rou-
INTRODUCTION
tines involved in driving a car (e.g., putting in the clutch, turning on the ignition, putting the car in gear, gradually releasing the clutch while stepping
on the gas, etc.) initially require conscious thought. However, they ultimately come to be performed without consultation of ones declarative
knowledge about the sequence of steps involved, and require few if any
cognitive resources (Schneider & Shiffrin, 1977). As Bargh (1997) asserted, a
very large amount of our social behavior is likely to involve the use of these
automatically activated productions.
Procedural knowledge plays a very important role in social information
processing, as I indicate frequently throughout this volume. For the present, however, I will concentrate my discussion on the content and structure
of declarative knowledge. I first consider the referents of this knowledge,
and the modality in which it is acquired. I then turn briefly to the different
ways in which this knowledge is represented in memory, and discuss the
need to understand these matters in conceptualizing the processing of information of the sort we receive in the course of daily life.
Referents of Social Knowledge
The stimulus information that impinges on us in a given social situation can
often far exceed our capacity to assimilate its implications. In a conversation, for example, several persons typically communicate to one another
both verbally and nonverbally, and in doing so, are likely to convey information about not only the topic under discussion but also about themselves. Moreover, this information is often transmitted in the context of
many other situational stimuli that are largely irrelevant to either the topic
at hand or the individuals who are discussing it. We are usually not passive
recipients of this information. Rather, we actively try to determine its meaning. Furthermore, we might construe its implications for the attainment of a
particular goal that we have in mind. In doing so, however, we necessarily
pay more attention to some features than to others. The features that receive the greatest attention depend in part on our perception of the referent (i.e., what the information is supposed to be about).
The referent of a piece of information can be either an entity (a person or
object, a place, oneself, etc.) or an event. However, most events involve persons and objects and occur in a specific (although perhaps unspecified) location. Furthermore, ones knowledge about an entity often concerns
events in which the entity has been involved. As a consequence, the referent of information is often subjective, depending on not only the context in
which the information is received but also the interests and objectives of
the recipient. For example, consider the statement, Muhammad Ali
knocked out George Foreman in Zaire. This statement could be about Ali,
Foreman, or the event itself. In some cases, of course, perceptions of the
CHAPTER 1
referent of information are determined by the context in which the information is received. Thus, someone who hears the aforementioned statement
during a discussion of Muhammad Ali is likely to identify Ali as its referent,
whereas someone who reads the statement in a magazine article about boxing might consider the event itself to be the referent. In addition, however,
individual differences in prior knowledge and motivation (e.g., whether one
is a fan of Muhammad Ali or George Foreman) can also influence perceptions of a statements referent. These situational and individual differences
in perception of the referent can have an impact on how the knowledge is
represented in memory and the likelihood that the knowledge comes to
mind later. For example, a person who learns that Ali knocked out Foreman
in a conversation about Ali, and consequently perceives Ali to be the referent, may not recall the information later when information about George
Foreman is being discussed. (For further discussion of this possibility, see
Wyer & Srull, 1989.)
Furthermore, when the information conveyed in a particular situation is
detailed and complex, recipients are likely to think more extensively about
features that pertain to its referent (as they perceive it) than other features.
Thus, the referent-related features are likely to be retained, whereas more
peripheral features seem to be forgotten. For example, a man who learns
that V. S. Naipaul won the Nobel Prize for Literature might remember the
specific context in which he acquired the information a short time after it
was conveyed. Over time, however, the information becomes part of his
general knowledge and the conditions that led to its acquisition are not
recalled.1
Affect and Emotion. Although the referents of knowledge are usually
external to the organism, internally generated stimulation can also be a
source of information. (See Strack and Deutsch, 2002, for a theoretical analysis of these alternative sources of information.) For example, external
stimuli can elicit affective or emotional reactions, and these reactions can
be used as bases for both thoughts and overt responses to the stimuli
(Schwarz & Clore, 1983, 1996; Wyer & Carlston, 1979). This does not mean
1
The dissociation of knowledge from the context in which it is learned does not mean that
contextual features are erased from memory. In the previous example, I learned about Naipauls
award from my wife, who stopped into my office to tell me about it one day at work, and I wrote
the preceding paragraph shortly afterwards. I would normally not be able to recall where and
when I acquired knowledge of the award. However, being reminded of the example while going
over this manuscript several months later was sufficient to stimulate a relatively detailed mental image of the event, including my own and my wifes physical locations in the room. Thus, although these contextual features became lost in the interim, they continued to exist in memory and could be retrieved by a subset of cues that identified the event and its use as an example
in this book. The theory to be proposed in this volume attempts to account for this and related
memory phenomena.
INTRODUCTION
that affective reactions per se are part of stored knowledge. Rather, they
are conditioned or unconditioned responses to cognitions about their referent, and can sometimes be the referents of cognitions, but they are not
themselves part of the cognitive system (Wyer, Clore, & Isbell, 1999). This
view, which contrasts with that proposed by Wyer and Srull (1989; see also
Bower, 1981), is discussed more fully in chapter 11.
Relational Knowledge. Much of the knowledge we require about entities and events is relational. These relations may be physical or conceptual.
For example, we know not only that a nose and eyes are parts of a face but
also their positions in relation to one another. Furthermore, we know that
our bedroom is down the hall from the living room, that Chicago is west of
New York, that lions are larger than rabbits, and that the earth goes around
the sun. We also know that people graduate from high school before going
to college, that smoking is a cause of lung cancer, and that people have to
order a meal in a restaurant before they eat it. This knowledge requires not
only understanding of the entities being related but the relations themselves (larger than, causes, above or below, etc.). Note that in many
instances, the mental representation of these relations requires visual imagery.
Modalities of Knowledge
Information is conveyed in many different modalities and formats. For example, it can be transmitted verbally in the form of propositions. It can also
be conveyed visually as in a picture or movie, or acoustically, as in music.
Our knowledge can likewise be in different modalities. That is, our knowledge of a person could consist in part of verbal descriptions of the persons
traits, and our knowledge of restaurants could include abstract characterizations of the events that usually take place there. However, we also have
mental images of people we meet and places we visit, and might have
acoustic representations of a persons voice, or of the Beatles rendition of
Hey Jude. Social experiences are often a configuration of verbal, visual,
and acoustic stimulation that combines to form our knowledge of what occurred.
There has been some controversy about whether information is actually
coded into memory in the modality in which it is conveyed or is represented in a common format that can be decoded into different modalities,
depending on situational demands (J. Anderson, 1978; Kosslyn, 1980; Pylyshyn, 1973). Certainly we can easily translate many visual stimuli into linguistic terms, and can form visual images of people and events on the basis
of verbal descriptions of them. Images can be acoustic as well as visual.
(This is apparent from the fact that people who have seen a foreign movie
10
CHAPTER 1
are later unable to remember whether actors statements to one another

were conveyed in subtitles or were dubbed.) Linguistic representations of
the actors utterances become integrated with the visual representation of
the movie and the quality of the speakers voices in a way that provides a
memory of the events very similar to the memory one would have if the
words had actually been spoken.
Be that as it may, it is most convenient in the present volume to assume
that codings of information in different modalities do exist in memory and,
in some cases, can be stored independently. Several studies we describe in
this volume attest to the importance of visual and auditory information
processing and the impact of visually coded knowledge on memory and
judgment. The way in which visual and verbal codings of knowledge theoretically combine to influence the representations that are formed, and how
these representations are later used, are detailed in chapters 4 and 5.
THE REPRESENTATION OF KNOWLEDGE

IN MEMORY
General Considerations
As the preceding discussion testifies, it is virtually impossible to discuss
the type of knowledge we acquire without making implicit assumptions
about the representation of this knowledge in memory. However, the mental representation of information can differ in several ways from the information itself. Some features of the information we receive are not stored in
memory at all, whereas other features that were not specified in the information may be added (Barclay, 1973; Loftus, 2000). Furthermore, detailed
descriptions of an event can often be interpreted in terms of abstract (e.g.,
attribute) concepts that are stored in memory independently of these descriptions (Carlston, 1980; Higgins, 1996; Wyer & Srull, 1989). In addition, information might be mentally reorganized into a form that is relevant to the
goal one happens to be pursuing (cf. Hamilton, Katz, & Leirer, 1980; Wyer &
Bodenhausen, 1985). Moreover, information that is transmitted visually may
be coded into memory linguistically, and verbal information may sometimes elicit visual images.
It might nevertheless be argued that people extract similar implications
from information regardless of the mental representation they form of it. If
this were so, an understanding of the mental representations that are
formed of information would be of little importance in predicting its impact.
We would simply need to characterize clearly the stimulus information to
which people are exposed, and to obtain a reliable indication of which as-
INTRODUCTION
11
pects of this information they retain (as evidenced by the aspects of the information they can recall). This would be sufficient to determine its impact.
In fact, however, research in social and cognitive psychology since the
1970s suggests this contention is nave. That is, a conceptualization of the
cognitive operations that are performed on information, and of the mental
representations that result from these operations, is essential for understanding its influence. Because the comprehension and mental representation of experiential knowledge is a central focus of this volume, some empirical support for this conclusion may be worth summarizing.
1. Communication and Persuasion
One of the first demonstrations in social psychology of the need to take
into account the cognitive activities that people perform in response to information was constructed by Greenwald (1968). He showed that the influence of a persuasive message was often unrelated to the implications of
those aspects of the message that participants could recall. Rather, its impact was governed by the implications of the thoughts they had about the
message at the time it was received. These cognitive responses, which included elaborations of the message content, counterarguments, and
thoughts that were unrelated to the message itself, presumably reflected recipients comprehension of the message and their evaluation of its validity
as well as the implications of other previously acquired knowledge that
they brought to bear on the position advocated.
In a study by Loken (1984), participants read a list of information items
pertaining to a college professor with instructions either to form an impression of the individual or to decide whether they wanted to take a course
from him. The information consisted of either favorable or unfavorable trait
descriptions, and either favorable or unfavorable consequences of taking
the course. In some cases, individual pieces of information had implications
for both attitudes toward taking the course and attitudes toward the professor (e.g., Dolan is interesting, Taking Dolans course is time consuming,
etc.). In other cases, they had implications for only one of these attitudes
(e.g., Dolan is clumsy, Taking Dolans class fulfills a graduation requirement, etc.). Later, participants reported their attitudes toward both the
person and the behavior, and then recalled the information they had received. The type of information that participants recalled was influenced by
the task objectives they were given. Nevertheless, the effects of experimental variables on attitude judgments did not mirror their effects on recall.
Moreover, correlations between attitudes and the evaluative implications
of attitude-relevant information that participants recalled were very low,
ranging from .05 to .24. Loken concluded that participants spontaneously
formed attitudes toward both the professor and taking the course at the
12
CHAPTER 1
time the information was received, and later reported these attitudes independently of the information they could remember at the time of judgment.
2. Person Impression Formation
In a study by Carlston (1980), participants read descriptions of a target
persons behavior that exemplified two traits with evaluatively different implications. In one condition, for example, behaviors (e.g., told his girlfriend
that her hair-do was ugly) could be interpreted as both honest and unkind.
Participants judged the target with respect to one of the two traits and
then, either a few minutes or several days later, evaluated him with respect
to the other trait. Participants second judgments were evaluatively biased
by the first. In other words, participants who initially evaluated the targets
honesty (and thus judged him to be honest) later reported him as being
more kind than participants who had not made this initial judgment. Correspondingly, participants who made kindness judgments at the outset (and
thus rated the target as unkind) judged him to be less honest than other
participants did. Moreover, these effects increased over time. Participants
apparently interpreted the information in a way that was relevant to the
first judgment they made and formed a mental representation of the target
as the type of person who had the trait being judged. Then, having done so,
they based their second judgment on the implications of this representation rather than reconsidering the original information they had received.
In a series of studies by Srull and Wyer (1983), some participants read descriptions of a persons behavior that were either predominantly favorable
or predominantly unfavorable with instructions to form an impression of the
actor. Then, after either a short or long delay (5 min or 45 min), they were
asked to report their liking for him. Other participants, who were asked to recall the behaviors they had learned, remembered fewer behaviors after a
long delay than after a short delay. However, participants rated the target
more extremely in the long delay condition (e.g., as more likable or more
dislikable, depending on the predominant implications of the behaviors). Apparently, these participants formed a generally favorable or generally unfavorable impression of the target on the basis of the predominant theme of
the behaviors they had read. After a delay, their recall of the specific behaviors diminished, but their overall impression remained. Therefore, their judgment, based on this impression alone, was more polarized.
3. Assimilation and Contrast
In a study by Higgins and Lurie (1983), subjects rated the harshness of
sentence imposed on a criminal by a target judge (Jones) along with the
sentences recommended by three context judges. Participants rated
Joness sentence as more lenient when sentences recommended by the
INTRODUCTION
13
context judges were long than when they were short. The critical conditions of the study, however, occurred a week later. At this time, participants
were given a sample of the sentences imposed by a different group of context judges that were again either long or short. Then, after receiving the information, they were asked to recall the actual sentence that Jones had imposed, as conveyed in the first session. Instead of recalling this sentence
directly, however, participants apparently recalled their rating of him as
harsh or lenient, and then reinterpreted the implications of this rating with
reference to the sentence imposed by the context judges in Session 2.
Therefore, they recalled the targets sentence to be longer when the sentences imposed by the second set of judges were long than when they were
short, and this was true regardless of the sentences imposed by the context
judges in Session 1.
4. Impact of Information That One
Is Told to Disregard
In a study by Wyer and Budesheim (1987; see also Wyer & Unverzagt,
1985), participants read a series of behaviors of a target person with instructions to form an impression of him. In some cases, the first behaviors
were favorable and the last behaviors were unfavorable. In other conditions, the order was reversed. On the pretense that an error had been
made, some subjects were told after reading the first set of behaviors that
these behaviors actually pertained to a different person than the one they
were supposed to consider and to disregard them. Other participants read
the entire set of behaviors and then were told to disregard the last set they
had considered. Finally, participants were asked both to judge the target
with respect to the traits to which the information pertained and to recall
the information they received. When participants had been told to disregard the first behaviors in the series, their judgments were not influenced
by these behaviors although they could recall the behaviors quite well.
When participants had been told to disregard the last behaviors in the series, these behaviors had an impact on their judgments although their recall of the behaviors was poor. Thus, instructions to disregard the information had opposite effects on recall and judgments in the two conditions.
These studies suggest that the mental representations that participants
form of information can exert a major influence on their judgments independently of the information itself, and that the magnitude of this influence
can increase over time. Greenwald and Banaji (1995) described several additional situations in which people are influenced by information they cannot recall and, in some cases, are unaware that they ever received (see also
Bargh, 1997). Thus, neither the implications of the information that people
receive nor the implications of the information they can remember provide
14
CHAPTER 1
much insight into the impact of this information on judgments. Rather,

the impact is mediated by the cognitive operations that people perform
on the information and the mental representation that results from these
operations.
A Conceptual and Methodological Approach
to Investigating the Influence of Mental Representations
To the extent that mental representations of information have an impact on
judgments, a question arises as to how their content and structure can be
determined. Any characterization of the mental representations that people
form from information is obviously metaphorical and does not reflect the
physical representation of the information in the brain. For one thing, information is not stored in particular locations, but rather, is distributed
throughout the cerebral cortex. Thus, all theoretical models of mental representation must be viewed in terms of their utility and not whether they
are true or false.
The general logic underlying theory and research on mental representations can be conveyed with reference to Fig. 1.1. This figure distinguishes
between input processes, recall processes, and judgment processes. Both
the type of stimulus information presented and the recipients processing
objectives at the time they encounter it can influence the cognitive operations they perform on the information and, therefore, the mental representation they construct. This representation is then stored in memory. If participants are later asked to recall the information, they may activate a
procedure for extracting this information from the representation they
formed earlier. If, on the other hand, they are called upon to make a judgment or behavioral decision, they may engage in search and retrieval operations that are relevant to this objective, and may use the material they extract from the mental representation as a result of these operations to
compute their response.
Two things are worth noting, however. First, the objectives that exist at
the time the information is received may differ from those that occur later,
at the time the information is retrieved for use in attaining a particular objective. Second, the search, retrieval, and computational operations that
are performed at this second stage depend on the particular judgment or
decision to be made and, therefore, are likely to differ from the operations
that are performed when peoples objective is to recall the information they
received. This means there may be little relationship between the implications of the information people recall and the judgments they make, as the
research cited earlier suggests.
This leads to the logic underlying research on social information processing in general, and much of the research to be summarized in this vol-
15
FIG. 1.1. Relations among various components of social judgment. Observable (independent
and dependent) variables are enclosed by solid lines, and mediating variables by dashed lines.
Rectangles denote states and ovals denote processes.
16
CHAPTER 1
ume more specifically. That is, the mental representation that is formed
from information is a hypothetical construct and cannot be observed directly. However, given a specific set of assumptions about the retrieval operations that occur when people are asked to recall the information, and a
metaphor for describing the associations among features of the information
that exist in memory, one can use the amount, type, and order of information to conceptualize the content and structure of the mental representations that was formed at the time the information was received. Then, once
this is established, one can hypothesize and investigate empirically the way
this mental representation is used to attain specific objectives to which it is
relevant. Much of the research and theory summarized in this volume is
based, implicitly or explicitly, on this general approach.
The social knowledge we acquire can be represented in memory in several ways. The nature of these representations, when they are constructed,
and how they are used, are a major focus of this book. Two types of mental
representations that occupy our greatest attention are worth noting briefly.
These representations, to be denoted situation models and generalized representations, are distinguished in terms of both their content and structure
and the point at which they typically come into play in social information
processing. Although we consider both types of representations in detail in
subsequent chapters, a general overview of the representations is perhaps
desirable.
SITUATION MODELS
We assume that people who receive information about situation-specific
events, either verbally or through direct experience, form mental simulations of these events in the course of comprehending them that indicate not
only the people and objects that are involved in the events but their spatial
and temporal relatedness. These mental simulations, or situation models
(Johnson-Laird, 1983, 1989; Kintsch, 1998) are often represented in memory
in terms of mental images of the events and, in some cases, propositionally
as well. To the latter extent, the representations are somewhat analogous
to a picture plus caption. Moreover, a temporally and thematically related
sequence of events might be represented by a multiple-segment representation somewhat analogous to a comic strip. (For a similar metaphor in
conceptualizing the mental representation of event sequences, see
Abelson, 1976.)
We assume that situation models are often formed spontaneously in the
course of comprehending information (see chap. 4 for a discussion of these
processes). This information could be either obtained through direct experience or conveyed verbally. However, a necessary condition for the con-
INTRODUCTION
17
struction of these models is that the events or states to which they refer are
temporally and spatially specific. That is, the phenomena must occur at a
specific (although not necessarily specified) place and point in time. This
constraint is implicit in the assumption that a mental image is formed.
Thus, the statements, The boy kicked the ball and The book is on the table describe events or states of affairs that existed at a particular time and
place, and a mental picture can be formed of them. In contrast, the boy
owns a ball, or the book is interesting, are not spatially or temporally
constrained, and a mental picture of the situations is not constructed
(Radvansky, Wyer, Curiel, & Lutz, 1997). Rather, a generalized representation is formed on the basis of these statements that does not have an image
component.
GENERALIZED EVENT REPRESENTATIONS

More complex generalized representations can also be formed in the
course of higher level goal-directed processing that occurs once information has initially been comprehended. For example, people who wish to tell
someone about a movie they have seen might describe the things that happened in more abstract terms than they were shown in the movie and, in
doing so, might construct a more general, linguistic representation of the
sequence of events that occurred. Generalized event representations can
also be formed in the course of evaluating the implications of a persuasive
message, describing someones personality, or generating a conclusion on
the basis of logical reasoning. Although these goal-directed representations
might also be based on previously formed situation models, they are stored
in memory independently of these models.
The Dimensionality of Event Representations

Generalized event representations can vary along several dimensions.
Generalizability. A generalized event representation might refer to a
particular experience that involves specific individuals and occurs at a
given time and place. Alternatively, it could pertain to a general class of individuals, situations, or events, instances of which occur frequently at different types. Thus, for example, I might have a representation of going to
dinner at Zorbas with Rashmi at 7 p.m. last Thursday. In addition, I might
have a representation of going out to eat at restaurants that generalizes
over times and places.
18
CHAPTER 1
Abstractness. A narrative representation could be encoded in terms of

specific concepts that apply to single entities. Thus, the representation of
my dinner at Zorbas might specify where I sat, what I ate, and the details
of the conversation I had over dinner. Alternatively, it could be coded in
very abstract terms similar to those used to convey the example provided
in the preceding paragraph.
Personal Relevance. Event representations can be about oneself and
ones personal experiences, about acquaintances, or about unknown or fictitious individuals. Similarly, they can describe either events one has personally encountered (e.g., ones observation of a teachers interchange with
an obnoxious student) or those that occur in a remote and unfamiliar clime
(e.g., the border dispute between India and Pakistan).
Time Frame. A representation could refer to people and events in either the past, the present, or the future, or could extend over two or more
of these time periods.
Types of Generalized Event Representations
Several different types of event representations can be distinguished in
terms of values along these dimensions. In this volume, we refer to all representations as narratives. Certain types of narratives are nevertheless worth
distinguishing, including scripts, stories, histories, and implicit theories.
1. Scripts and Plans
Scripts are prototypic sequences of events that occur in a particular situational context (Schank & Abelson, 1977). They can refer to either specific
persons (e.g., my wifes ritual of getting ready for a dinner party) or people
in general (a visit to a restaurant). However, although scripts can refer to
particular individuals and situations, they apply to routinized sequences of
actions that involve these persons and situations that typically occur under
the circumstances at hand.
A plan is a special type of script in which the events described produce a
particular (typically favorable) end state. These constructions are typically
applied to new situations, often serving as behavioral guides. For example,
an individual who has formed a script of changing a flat tire as a result of
observing others engage in the activity could use this script as a plan in determining how to change a tire when personally required to do so.
Most research on the representation of scripts in memory (Bower, Black,
& Turner, 1979; Graesser, 1981; Nottenburg & Shoben, 1980; Trafimow &
Wyer, 1993) has focused on mundane event sequences that are very familiar to members of a Western culture. As noted earlier, however, individuals
INTRODUCTION
19
can form routinized sequences events that are specific to themselves, or

their family (e.g., I might have a representation of the sequence of acts my
spouse routinely performs when getting ready for a party). As we see in
chapter 5, there is some question about how often these representations
are actually constructed (cf. Colcombe & Wyer, 2002). Nevertheless, several
such constructions undoubtedly exist.
The fact that the events that compose scripts are routinized and generally mundane distinguishes them from other types of narratives in two
ways. First, they are likely to have little evaluative significance for the persons involved. (A visit to a restaurant, for example, which terminates with
paying the bill and leaving, has relatively neutral implications.) Second,
scripts are unlikely to be the topic of conversation. That is, the events that
compose them are expected to occur, and so mentioning them has little information value.
2. Stories
Stories, unlike scripts, concern sequences of events at least one of which
is unique. That is, it deviates from expectations based on prior knowledge
about the types of people and situations involved. In addition, the events
have some evaluative significance for the protagonists (Mandler & Johnson,
1977; Stein & Glenn, 1979). This conceptualization of a story is more restricted than Schank and Abelsons (1995). In fact, a substantial literature in
cognitive psychology has been concerned with both the appropriate definition of stories and their representation in memory (cf. Black, Galambos, &
Read, 1984; Brewer, 1995; Brewer & Lichtenstein, 1981; Mandler & Johnson,
1977; Stein, 1982). In this work, stories are postulated to have certain common features that are captured in large part by a story grammar (Mandler
& Johnson, 1977; Stein & Glenn, 1979). Although these grammars differ in detail (see Stein, 1982, for a comparison of alternative story grammars), they
share the assumption that stories have a definable beginning and end along
with an event that provides a transition between these states.
Other features that are sometimes used to distinguish stories from other
representations are of less utility. For example, Brewer and Lichtenstein
(1981) argued that a story must also elicit some degree of affect or emotion
in the recipient. However, it seems undesirable to define a story in terms of
responses of the recipient rather than attributes of the story itself. (Brewer
and Lichtensteins conceptualization implies that the same narrative might
constitute a story to one person but not to another, depending on the affect
it happens to elicit.)
Brewer and Lichtenstein (1981) also assumed that the function of stories
is to entertain. However, the same narrative might be communicated for the
purpose of being entertaining in one context but for the purpose of making
20
CHAPTER 1
a point or exemplifying a principle in another context. It would be conceptually confusing to define the same narrative differently, depending on the
conditions in which it is conveyed.
It therefore seems more reasonable to define a story as simply a narrative that (a) describes events involving specific individuals or groups that
are localized in time and place, (b) contains at least one event that deviates
from expectations based on prior knowledge of the type of individuals and
situations involved, and (c) whose end state has emotional significance for
one or more of the protagonists.
3. Histories
A history, like a story, consists of a sequence of events that are specific
to a particular person or group and are localized in time and place. Unlike
stories, however, a history does not necessarily contain unique or unexpected events. Nor does it necessarily have emotional significance for the
protagonists. Histories can concern real-world events in which one is not involved. However, it can also refer to oneself or an acquaintance. The mental
representation of a personal history is clearly not a single sequence of
events that occurs over the course of a lifetime. Rather, an individual is
likely to construct several histories, each of which is restricted to life
events that are relevant to the purpose for which it is formed (communicating about oneself to another, comprehending ones own or anothers personal experience, etc.). Once constructed, these sets of segments are likely
to be stored independently of one another in memory.
As these observations imply, histories can contain overlapping segments. Moreover, a history does not usually cover all the events that occur
during the time period to which it pertains. Thus, one might construct different histories that cover the same or overlapping time period. For example, a personal history of how one became a psychologist could temporally
overlap a history of ones romance with an ex-call girl. However, the events
that compose these mental constructions might be totally different.
According to this conceptualization, stories and histories are not mutually exclusive constructs. A history would be a story to the extent it contains unexpected events and is emotionally significant for one or more of
the parties involved. However, not all histories are stories, and not all stories are histories.
4. Implicit Theories and Implicational Molecules
A theory, like a script, is composed of a prototypic sequence of events.
However, the events are causally as well as temporally related. Moreover,
the events are often defined more broadly. A theory is not localized in time
and, therefore, can be applied in many different circumstances. It might
INTRODUCTION
21
nevertheless concern a particular individual (e.g., oneself) or general category of individuals (African Americans, fraternity members, etc.) as well as
people in general. Similarly, it could be specific to a particular type of situation (living in public housing, studentfaculty interactions) or could apply
more generally. In all cases, however, the events that compose a theory are
likely to recur at different times.
For example, one might have an implicit theory that (a) people with alcoholic fathers feel insecure, (b) these feelings lead them to pursue unrealistically high goals, (c) their failure to attain these goals leads to depression
and ultimately to drinking, and (d) they wind up becoming alcoholic like
their fathers. Each segment encompasses a large number of more specific
events and functions as a general concept that can be used to interpret and
organize new information. To this extent, a theory can function as an event
stereotype.
Theories play a very important role in processing new information about
the world in which we live and making inferences on the basis of it. They
are used both to explain events that have occurred in the past and to predict the future. Moreover, they can be used to fill in gaps in a sequence of
events about which one has incomplete information.
In this regard, some theories can consist of only two or three segments
that, in combination, constitute a general principle. In these instances, the
principle could function as a linguistically coded title, or header of the
theory to which it refers. For example, the principle people get what they
deserve could be the title of a two-segment theory composed of the events
A person does something bad and The person gets punished. These theories can be conceptualized as implicational molecules of the sort suggested
by Abelson and Reich (1969; see also Wyer & Carlston, 1979, 1994). That is,
they consist of a set of general propositions that have become associated
through learning. For example, the propositions associated with the principle that people get what they deserve could compose the molecule
[P does something good (bad); Something good (bad) will occur to P].
The use of an implicational molecule is theoretically governed by a completion principle. That is, if the information one receives about a specific situation instantiates all but one of the propositions that compose a molecule,
an instantiation of the remaining proposition will be inferred. Thus, people
who have formed the implicational molecule described in the preceding example might predict that a particular person who has done a bad thing will
be punished. However, they might also infer that someone who has experienced misfortune has done a bad thing or for other reasons deserves his or
her fate. Several other examples of the use of implicational molecules, and
implicit theories more generally, are discussed in detail in chapter 10 of this
volume.
22
CHAPTER 1
GENERALIZED ENTITY REPRESENTATIONS

Generalized representations are obviously not restricted to events. They
can also pertain to persons, to groups of persons, to objects, or to places. In
fact, most of the mental representations that have typically been investigated in social information processing research (for reviews, see Carlston &
Smith, 1996; Wyer & Carlston, 1994) are of this type. For example, people
who receive descriptions of someones traits and behaviors with the goal of
forming an impression of them may think about the individual pieces of information in relation to one another in order to understand their relatedness or, perhaps, to reconcile apparent inconsistencies among them. In the
course of this cognitive activity, associations are presumably formed
among the pieces of information that influences the ease of recalling the information later.
The entity representations that are constructed in the course of goaldirected processing can often be conveyed metaphorically in terms of an
associative network. That is, concepts and units of knowledge (e.g., situation models) are represented by nodes in memory. When a person thinks
about one concept or knowledge unit in relation to another, an association
is formed between them that is represented by a pathway. Thus, suppose
people are asked to form an impression of a person, John, on the basis of
set of behavior descriptions. If they interpret a particular behavior of the
person (e.g., returned a lost wallet) as honest, this establishes an association between a concept denoting the behavior and the trait concept honest. If they interpret several other behaviors (refused an opportunity to
cheat on an examination, told his girlfriend that her hair-do was ugly,
etc.) in terms of the same trait, an association is formed between these behaviors and the trait concept as well, leading to a traitbehavior cluster as
shown in Fig. 1.2a.
Alternatively, suppose both these and other behaviors (e.g., honks at
slow drivers, reads poetry, etc.) are thought about with reference to
John without construing their trait implications. Then, associative pathways are formed between the behavior concepts and the person concept,
John, as shown in Fig. 1.2b. General traits that are attributed to John as a
result of processing these behaviors can also become associated with this
concept. If people happen to think about some of the behaviors in relation
to others, associative links may be formed between the behaviors as also
shown in the figure. Finally, if they think extensively about Johns engaging
in some behaviors but not others, the former behaviors may become more
strongly associated with John than the latter ones do. This difference is
conveyed by the thickness of the associative links between the behaviors
and John (cf. Wyer & Carlston, 1979), as also shown in Fig. 1.2b.
INTRODUCTION
23
FIG. 1.2. (a) Metaphorical trait-based representations formed as a result of interpreting behaviors in terms of trait concepts. (b) Metaphorical representation of a person, John, described by behaviors and traits. Thicker lines denote stronger associations.
The representations of the sort shown in Fig. 1.2 are theoretically the result of conscious cognitive activity (but see Chartrand & Bargh, 1996, 2002,
for a different conclusion). Moreover, the associations shown in the figure
reflect the implications of specific assumptions about the nature of this activity. Given assumptions about the processes that underlie the extraction
of specific features from the network, these implications can be evaluated
24
CHAPTER 1
empirically. One set of assumptions, espoused by Hastie and Kumar (1979)

and others (Srull, 1981; Srull, Lichtenstein, & Rothbart, 1985; Wyer & Srull,
1989), which captures the processes involved in person impression formation on the basis of linguistically coded traits and behaviors, is discussed
briefly in chapter 7. However, an associative network metaphor has been
applied successfully in other domains as well (cf. Park, Yoon, Kim, & Wyer,
2001; Park & Wyer, 1993; Wyer & Carlston, 1979).
The conceptualization described previously has typically been applied
to representations that are formed of fictitious entities. Representations of
actual people and objects may be similar with two exceptions. First, the
representation of a familiar person might often include the referents name
and also a mental image of his or her physical appearance. Second, the features that are associated with the individual might be added to the representation over a period of time, as new information about the person is acquired. These additional features simply become linked to the central
concept of the referent, much as the ones that were conveyed at the outset.
On the other hand, once features have become part of a representation,
they cannot be modified or erased. (See chap. 2 for an elaboration of this
assumption.) When an existing representation appears to be inappropriate,
a new representation is constructed whose features differ from the original.
Thus, people might form different representations of the same person or
object that are stored independently in memory, each with a different (although perhaps overlapping) set of associated attributes.
Finally, representations might pertain to a group or category as well as a
single entity. For example, people might form a representation of the persons in a particular social group, of U.S. Presidents, or of African Americans. These representations might consist not only of features that are typically associated with members of the group or category as a whole, but also
names of the members that belong to the group and have been thought
about in relation to it. In this regard, the name of a particular person who
belongs to several groups could be contained in several different representations. However, these representations would not be interconnected but
would exist independently in memory. This assumption distinguishes the
present conceptualization from more general associative network models
(e.g., Collins & Loftus, 1975; Wyer & Carlston, 1979), which assume that all
representations with common features are interconnected.
An assumption underlying the conceptualization is that once a mental
representation is formed (either spontaneously in the course of comprehension or as a result of goal-directed processing), this representation is
stored in memory as a single unit of knowledge and is later retrieved as a
whole. Thus, in our example, a representation of John would presumably
be stored in memory as a unit and might later be retrieved as a whole when
information about John is sought. Other representations of the individual,
INTRODUCTION
25
based on information obtained at different times, might also be formed and

stored independently. These latter representations might include situation
models of the sort described earlier. This means than when information
about a person or object is sought, the type and implications of the information retrieved depends on which of several potentially applicable knowledge representations come to mind. This, in turn, depends on its accessibility in memory. This matter is discussed more fully in chapter 2 within the
framework of the general information-processing formulation we propose.
C H A P T E R
2
The Architecture of the
Information-Processing System
A complete theoretical formulation of information processing must specify

the content and structure of memory, the storage and retrieval processes
that are involved in the transmission of knowledge throughout the cognitive system, and the way this knowledge comes into play in the pursuit of
specific goals to which it is relevant. Wyer and Srulls (1989) theory attains
many of these objectives, and several features of this earlier theory are retained in the conceptualization I now propose. However, several assumptions of the earlier formulation must be modified in order to account for the
processing of information about familiar people and events that is acquired
through direct experience. Before describing these modifications, certain
limitations of the earlier model are worth noting. (These limitations are
shared by other, more restricted conceptualizations of social information
processing as well; see Brewer, 1998; Carlston, 1994; Smith, 1990; Wyer &
Carlston, 1979.)
LIMITATIONS OF WYER AND SRULLS (1989)
THEORY
The Comprehension and Validation of Information
About Familiar People and Events
Perhaps the most general deficiency of the original model is its failure to
specify the operations that govern the comprehension of information at the
time the information is first encountered, before goal-directed cognitive ac26
ARCHITECTURE OF THE INFORMATION-PROCESSING SYSTEM
27
tivity is initiated. Wyer and Srull simply assumed that input information is
spontaneously encoded in terms of low level semantic concepts that are
applied independently of the modality in which the information is received.
They provided no indication of how this is done. Furthermore, their assumption that all incoming information is encoded semantically constrains
the models ability to address the comprehension of information that pertains to people and events about which one already has substantial knowledge. The model assumed that referent-specific knowledge does not come
into play until a later, goal-directed stage of processing.
One implication of this assumption is that the validity of information
about well-known people and events is not assessed until some time after
the information is comprehended, and may not occur at all unless people
have a specific goal to which this assessment is relevant. This seems intuitively unlikely. Most Americans would almost immediately recognize that
the assertion Abraham Lincoln was President of the United States is true
in the course of comprehending it. By the same token, they are equally
quick to recognize that Mao Tse Tung was President of the United States
is false. These validity assessments appear to occur spontaneously in the
absence of any specific goal that requires them.
Not all information is spontaneously verified at the time it is comprehended, however. The statement Abraham Lincoln visited Boston is likely
to be comprehended without any conscious recognition that it is either true
or false. Statements that refer to unknown persons may be similarly comprehended without spontaneously validating them. The conditions in which
spontaneous validity assessments do and do not occur in the course of
comprehending information is not specified by either the Wyer and Srull
(1989) formulation or others.
Information is likely to be processed quite differently when it is comprehended with reference to previously acquired knowledge about its referents than when it is not. For example, suppose people receive information
that a person is kind, followed by information that he both (a) bought dinner for a street person who did not have enough money to eat and (b) berated his secretary for coming 10 minutes late to work. Recipients are likely
to think more extensively about the second behavior, which is ostensibly
inconsistent with the trait description than about the first one, leading this
behavior to be better remembered later (Hastie & Kumar, 1979; Srull, 1981;
Wyer & Gordon, 1982). However, suppose people receive information that a
conservative Republican has both organized a sit-in at a nuclear power
plant and picketed an abortion clinic. In this case, recipients are less likely
to think about the behaviors in terms of their consistency with the general
characteristics of the actor. Rather, they think about them with reference to
their knowledge about the social implications of the behavior and their perception that these implications are desirable. Consequently, they have
28
CHAPTER 2
better recall of the behaviors that reflect policies they consider to be incorrect or undesirable (Wnke & Wyer, 1996; Wyer, Lambert, Budesheim, &
Gruenfeld, 1992). In short, information that has little relevance to ones
prior knowledge about its referents is likely to be thought about in terms of
the semantic consistency of its implications. However, information about
persons and situations about which one has prior knowledge is usually
thought about with reference to this knowledge, and so its semantic consistency has less impact.
The aforementioned studies provide only few examples of the role of referent-specific knowledge in comprehending and validating new information. The effects of spontaneously recognizing information as true or false
in the course of comprehending it, and the reactions to the implications of
these assessments, are central to an understanding of humor elicitation
(Wyer & Collins, 1992), emotional communication (Gaelick, Bodenhausen, &
Wyer, 1985; Scott, Fuhrman, & Wyer, 1991) and persuasion (Gruenfeld &
Wyer, 1992; Wyer & Gruenfeld, 1995). This becomes clear in later chapters
of this volume.
The Nonverbal Coding of Social Events

A second limitation of Wyer and Srulls (1989) theory is its focus on linguistically coded information processing. The information we acquire in daily life
is often obtained nonverbally, through direct observation. Although linguistically coded information (e.g., verbal utterances) can accompany our observations, this information is also conveyed in the context of visual and
acoustic features that are important for comprehending it. (Ones interpretation of a colleagues comment You idiot! is clearly different, depending
on the speakers facial expression and tone of voice.) The interpretation of
nonverbal behavior, both in isolation and in the context of verbal behavior,
has received much attention (for a recent review, see DePaulo & Friedman,
1998). However, although the multiple modalities in which information can
be represented in memory have been recognized in cognitive psychology
(A. Anderson, Garrod, & Sanford, 1983; Graesser, Singer, & Trabasso, 1994;
Schank & Abelson, 1977, 1995; Zwaan, Magliano, & Graesser, 1995), the implications of this fact has rarely been considered in social psychological theory and research. Carlston (1994) noted that visual and linguistic processing systems can interact in the course of attaining a specific processing
objective (i.e., impression formation). However, the nature of this interaction is left ambiguous in his model. Wyer and Srull (1989) allowed for the
construction of mental representations in multiple sense modalities, but
they also fail to articulate the nature of these representations and the conditions in which they are formed.
29
In this regard, the modality in which information is presented is not necessarily the modality in which it is reported. Visual information can sometimes be encoded into memory linguistically. Moreover, linguistically coded
information can elicit mental images (Bransford, Barclay, & Franks, 1972;
Garnham, 1981; Radvansky, Gerard, Zacks, & Hasher, 1990; Reyes, Thompson, & Bower, 1980; Wyer, Adaval, & Colcombe, 2002). These processes,
however, are not articulated in the theory proposed by Wyer and Srull
(1989).
Other, more specific problems with the original theory will be noted
presently in the context in which they become relevant. The conceptualization we propose in this and later chapters is intended to address the earlier
deficiencies while at the same time preserving the general architecture of
the information-processing system of the original model and, therefore,
many of its implications. In the next section, we outline the general structure of the proposed system, noting the assumptions that differ from those
of the original model. We then summarize the assumptions we now make
concerning information storage and retrieval, which differ significantly
from those of the original formulation.
STRUCTURE OF THE PROCESSING SYSTEM

The architecture of the social information-processing system we propose is
described in Fig. 2.1. The system, which is structurally very similar to that
postulated by Wyer and Srull (1989), consists metaphorically of four information storage units (the Sensory Store, the Work Space, the Permanent
Storage Unit, and a Goal Specification Box), four special-purpose processing
units (the Comprehender, Encoder/Organizer, Inference Maker and Response Selector), and an Executor that directs the transmission of cognitive
material from one unit to another. The original theory differs from the
model to be proposed in terms of (a) the organization of memory in Permanent Storage, (b) the functions of the Comprehender, and (c) the processes
of information retrieval.
Storage Units
Of the four storage units the theory postulates, the Sensory Store is least
important for the issues of concern in this volume. This unit simply records
the configuration of sensory input material that impinges on the cognitive
system at any given movement, in fairly veridical form. However, the trace
of this information decays rapidly (in a manner of seconds), and if it is not
involved in further processing, it is lost (cf. Sperling, 1960). The postulation
of this storage unit recognizes that information in multiple sense modalities
30
FIG. 2.1. Schematic representation of the information processing system. Circles and ovals refer to processing units. Rectangles denote storage units. Arrows indicate the direction of transmission of cognitive material between these units.
31
can simultaneously impinge on the processing system and can serve as input to the Comprehender (see Fig. 2.1). However, its importance for the issues of concern here is limited, and the unit will generally not be discussed
further.
1. The Work Space
The structure and function of the Work Space is similar to that of the
original model. This unit functions as a temporary store of all information
that has been encoded by the Comprehender along with other material
that is or has recently been involved in higher level goal-directed processing. As shown in Fig. 2.1, input to the Work Space can include material that
has been interpreted and transmitted to it by the Comprehender, concepts and knowledge representations that have been retrieved from Permanent Storage for use in attaining specific processing objectives, and the
results of this processing. The capacity of the Work Space is large, and as
long as its contents are relevant to the attainment of a processing objective that is active, it can be retained indefinitely. However, if a processing
goal has been attained, or if information has not been used for some time,
the Work Space may be cleared to permit easier access to material that is
relevant to other, more immediate objectives. This may be done either automatically, as new material is deposited, or volitionally (for evidence, see
Srull & Wyer, 1983). Phenomenally, this simply means that people can intentionally stop thinking about material of relevance to one objective and
concentrate on a second.2
These considerations become important in the context of an additional
assumption of both the Wyer and Srull model and the present one. That is,
only the output of higher order goal-directed processing is transmitted to
Permanent Storage. Therefore, once the Work Space is cleared, any input
information from the Work Space that has not been involved in goaldirected cognitive activity is irretrievably lost.
2. Permanent Storage
A major modification of the present model in relation to that proposed
by Wyer and Srull surrounds the conceptualization of long-term memory or,
in other words, Permanent Storage. We first describe the assumptions of the
original model, and then discuss the modifications to be made.
2
On the surface, this implication appears inconsistent with the result of research on thought
suppression (Bodenhausen & Macrae, 1998; Wegner, 1994), which suggests that conscious attempts not to think about a concept increases its accessibility in memory. However, this inconsistency is illusory, as we point out later in this volume.
32
CHAPTER 2
The Original Model. Wyer and Srull conceptualized long-term memory

metaphorically as a number of content-addressable storage bins. One, semantic bin, was likened to a cognitive dictionary that contained semantic
concepts necessary to interpret incoming stimulus information. Referent
bins, which in combination functioned as a cognitive encyclopedia, contained representations of a particular referent (entity, event, etc.) that are
formed in the course of goal-directed cognitive activity. Each bin was identified by a header, or set of features that defined or circumscribed its contents. Different representations could be formed of a referent at different
points in time, being stored in a bin as separate units of knowledge in the order they are constructed. Thus, more recently formed representations were
on top. These representations, which might differ in form and modality,
could theoretically be retrieved independently of one another in the course
of a top-down search of the bin contents. This means that the most recently
formed and deposited representations are most likely to be identified later.
Different bins were theoretically formed to contain information about
referents at different levels of specificity. Thus, an individual might form a
George W. Bush bin containing information about Bushs actions and policies while in office, but also a U.S. President bin containing more general
information about U.S. Presidents, the names of specific exemplars, and so
on. Note from this example that pieces of knowledge in one bin (e.g., the
name of a specific U.S. President) can include features in the header of another bin. Moreover, not all information about a referent is necessarily
stored in the bin that pertains to this referent. Thus, suppose ones objective at the time a piece of information is received about George W. Bush
pertains to presidents in general rather than to Bush in particular. Then, it
would be stored in a U.S. Presidents bin. Consequently, it would not be retrieved if one is later searching in the Bush bin for relevant knowledge
about George W. Bush.
The Proposed Modification. The bin construct permitted the Wyer
and Srull model to predict and explain a number of phenomena that are difficult for alternative formulations to capture (e.g., see Srull & Wyer, 1980,
1983; Wyer & Bodenhausen, 1985; Wyer & Budesheim, 1987; Wyer & Unverzagt, 1985). At the same time, the conceptualization is inadequate to account for the comprehension of information about known people and
events. One problem derives from the aforementioned assumptions that (a)
the information contained in a referent bin is only accessed in the course of
goal-directed cognitive activity, and (b) a top-down search of the bin is required in order to identify it. Because of these assumptions, the conceptualization has difficulty accounting for the spontaneous comprehension of
information about known people and events, or for the spontaneous recognition that the information one receives is true or false.
33
In the revised theory, therefore, we drop the construct of bins. Instead,

following multiple-trace theories of long-term memory (cf. Hintzman, 1986),
we assume that each mental representation that is formed of a given body
of information (either at the time the information is first received or later, in
the course of higher order, goal-directed processing) has its own memory
trace. Each of these representations can become involved in the processing
of information at the initial comprehension stage as well as at later stages,
in ways that we specify presently. This modification of the theory permits
new information about a known referent (e.g., George W. Bush) to be comprehended immediately in terms of a specific representation of this referent
without requiring goal-directed cognitive activity.
3. Goal Specification Box
The Goal Specification Box is a temporary store of the procedures that
are currently being used to attain the particular goals that are actively being pursued at the time. These procedures, or goal schemas, are drawn from
Permanent Storage, and are consulted by the Executor for instructions
about the cognitive steps required to attain these goals. More than one goal
schema can be stored in the Goal Specification Box at one time. This means
that the system has the capability of pursuing more than one goal simultaneously. However, the size of the Goal Specification Box is limited. Consequently, if a particular procedure is complex, the number of other procedures that can be stored in the Goal Specification Box, and, therefore, the
number and complexity of other ongoing cognitive activities that can be
pursued simultaneously, decreases.
Special Purpose Processing Units
The Wyer and Srull conceptualization postulates four special purpose processing units, each of which performs a particular function. The first, Comprehender, is an initial lower level encoding device that interprets stimulus
input information that enters the processing system from the Sensory
Store. Unlike other processing units, the Comprehender is not controlled by
the Executor. That is, it receives stimulus information from the Sensory
Store and automatically interprets it in terms of concepts and knowledge
structures that are drawn from Permanent Storage (Fig. 2.1). This is done independently of any more specific processing objective that might exist.
In the original theory, the activities of the Comprehender were restricted
to the encoding of information in terms of semantic concepts. In the present conceptualization, however, the functions and capabilities of the Comprehender are expanded in ways that permit it to interpret information in
terms of previously acquired referent-specific knowledge as well. These
processes are articulated in detail in chapter 4.
34
CHAPTER 2
The other three processing units come into play in the course of specific
goal-directed activity, and are activated on the basis of instructions in a
goal schema that is used by the Executor to monitor this activity. The characteristics of these units are not appreciably different from those described
by Wyer and Srull (1989). The Encoder/Organizer interprets stimulus information transmitted to it from the Comprehender in terms of more abstract
concepts that are relevant to the goal being pursued, or organizes the information with reference to a pre-existing configuration of features that are
specified in the information. The Inference Maker generates specific subjective inferences on the basis of new or pre-existing information using logical,
algebraic or other combinatorial rules, or, in some cases, heuristics. The
Response Selector then translates subjective inferences into an overt response, which could be a behavioral decision, a judgment along a category
response scale, or a verbal utterance.
Each processing unit is equipped with a library of routines (conceptually
equivalent to cognitive productions, as mentioned in chap. 1) that it draws
on to perform its functions. The nature of these productions can depend on
both the goal being pursued and the type of information being processed.
Thus, for example, the procedures performed by the Inference Maker depend on whether the goal is to form a general impression or to make a comparative judgment. On the other hand, they can also depend on whether
the information is conveyed visually or verbally, or whether it is conveyed
in a list or a narrative (cf. Adaval & Wyer, 1998).
The Executor
The Executor monitors the flow of cognitive material between memory storage units and special purpose processing units. It has no mind of its own,
but instead, takes instructions from a pre-existing goal schema that is active in the Goal Specification Box. These schemas presumably provide stepby-step descriptions of the sequence of steps that must be performed in order to attain the objective at hand.
For example, suppose people receive descriptions of a person for the
purpose of forming an impression. The goal schema that guides the attainment of this objective might indicate that the individual behaviors should
be interpreted in terms of trait concepts, that the evaluative implications of
these traits should be assessed, and the implications of these assessments
should then be combined to form an inference of the persons general likeableness, and that this evaluation should be communicated to another verbally. The Executor would presumably recognize the first two (trait and
evaluative) encoding tasks to be the responsibility of the Encoder/Organizer and would direct the information to this unit with instructions to perform the encoding. Once the results of the evaluative encoding are avail-
35
able, the Executor would transmit these encodings to the Inference Maker
with instructions to combine their implications into an overall inference.
Then, once this inference is computed, the Executor would transmit the inference to the Response Selector with a request to generate an overt response in a language that was applicable (a verbal utterance or, if the inference is to be reported in an experiment, a numerical value along a response
scale).
Note that the routines contained in the libraries of the various processing units, as well as the goal schemata used by the Executor to govern the
flow of information in pursuit of specific processing objectives, are generally not specified by the conceptualization we propose. These routines and
procedures are the subject of research on specific types of cognitive activity of the sort described by Wyer and Srull (1989) as well as later chapters
of the present volume. However, many social psychological theories derived from this research can incorporated into the more general conceptualization we have proposed. For example, the model of person impression
formation and memory proposed by Srull and Wyer (1989; for earlier versions, see Hastie & Kumar, 1979; Srull, 1981), which specifies the cognitive
processes involved in forming an impression and the cognitive representations that are formed as a result of these processes, can be viewed as a
specification of an impression formation goal schema along with a specification of the organizational processes performed by the Encoder/Organizer
and stored in the Encoder/Organizers library. The algebraic integration
processes postulated by Norman Anderson (1971; 1981), several of the cognitive heuristics identified by Tversky and Kahneman (1974), and the attribution processes postulated by Bem (1972) and Jones and Davis (1965), can
be viewed as theories about the routines performed by the Inference Maker
and the conditions in which they are applied. The impact of contextual
stimuli on the use of category response scales (Lynch, Chakravarti, & Mitra,
1991; Ostrom & Upshaw, 1968; Parducci, 1965; Wyer, 1974) and opinion survey responses (Schwarz, 1994; Strack, 1994) are essentially theories about
the processes performed by the Response Selector. These and other theoretical formulations can in principle be incorporated into the general framework we propose.
Effects of Multiple Goal Schemata
Two more general aspects of the conceptualization are worth noting at this
point. First, more than one goal schema might be relevant to the attainment
of a given objective. An alternative to the sequence described in our example might involve a comparison of the configuration of trait encodings to
the pre-existing knowledge representation of a known person, or alternatively, a prototype, and, if there is sufficient match, an inference of the per-
36
CHAPTER 2
sons likeableness from a pre-existing evaluation of the person to which this

representation refers (Fiske & Pavelchak, 1986). (Thus, for example, the
configuration of features might match those of ones roommate or, alternatively, the typical college professor, leading the evaluation to be similar to
that associated with this individual.) In this case, which alternative schema
is applied could depend on its relative accessibility in memory and the ease
of performing the cognitive operations it specifies (Logan, 1988).
In this regard, several recent formulations of information processing
have been viewed as dual processing models (e.g., see Chaiken & Trope,
1999). These models recognize that different processes are likely to govern
memory and judgment, depending on a number of situational and motivational factors that happen to exist. These alternative processes are generally implicit in the conceptualization we propose and also the Wyer and
Srull theory, both of which postulate that several different cognitive processes may be specified in the alternative goal schemata that can potentially
be brought to bear on the attainment of a particular processing objective as
well as in the libraries of various processing units. Although the present
conceptualization does always not specify the nature of these processes, it
clearly allows for their existence. Moreover, the conceptualization takes
into account the possibility that different processes can operate simultaneously. Several alternative processes of encoding, organization, and inference are described in detail throughout this volume.
Conscious Versus Nonconscious Processes
Second, Wyer and Srull assume that consciousness resides in the Executor. Thus, the Executor is aware of the general processing stages to be
performed, as specified in the goal schema. Furthermore, it is aware of the
inputs to processing units, the material that is drawn from Permanent Storage to use in operating on these inputs, and the output of processing at
each stage. On the other hand, the cognitive operations performed by the
various processing units, which are governed by procedures stored in the
libraries of these units, are not subject to awareness.
For example, suppose a man is asked his impression of a woman who
reads Tolstoy and pickets abortion clinics. He might judge the woman to be
both intellectual and feminist on the basis of these behavioral descriptions. Furthermore, he might be aware that the characterization reminded
him of a college professor he knew in the 1960s, and that he based his liking
for the woman on his feelings about this professor. This is because the output of processing at each stage is retained in the Work Space and, as such,
is accessible to the Executor. On the other hand, the man would have little
insight into why he interpreted the behavior in terms of these particular
traits rather than others, or into the processes that led to his inference that
37
the woman was similar to the professor. We consider the results of automatic and deliberative processing later in this volume. For the time being, it
is sufficient to note that the theory can account in principle for differences
in conscious and nonconscious processing and can specify the conditions
in which each type of processing may operate.
STORAGE AND RETRIEVAL PROCESSES

The assumptions surrounding the storage and retrieval of information are
intimately connected to assumptions about the structure of memory. In the
original theory, information in Permanent Storage was identified by (a)
compiling a set of features (probe cues) that specify the type of information desired, (b) identifying a referent bin whose header contains these features, and (c) conducting a probabilistic, top-down search of memory for a
knowledge representation that contains these features. These assumptions
imply that the likelihood of identifying a particular knowledge representation will increase with both the recency of its acquisition and use (and,
therefore, its proximity to the top of the bin) and the frequency of its acquisition and use (i.e., the number of times it is represented). In the present
theory, we retain the construct of probe cues. However, the bin construct
is eliminated and with it, the sequential search processes that Wyer and
Srull (1989) postulated. Consequently, the effects of recency and frequency
on accessibility are accounted for in other ways.
Storage
Our assumptions about the storage of information can be stated simply in
terms of a postulate:
Postulate 2.1. Each knowledge representation that is formed, either in the
course of comprehension or in the course of higher level, goal-directed processing, constitutes a separate unit of knowledge that is stored independently
of other knowledge units.
This assumption has broad implications. For example, it suggests that if

two conceptually related representations of a referent are formed at different times, they will be stored separately in memory. Consequently, one representation can later be retrieved independently of the other. Thus, for example, suppose people who have formed a detailed representation of a
personal experience are later asked to describe this experience to others.
They may do so by retrieving the representation and encoding it in terms of
38
CHAPTER 2
more general, semantic concepts that can be communicated verbally. This

more general, verbally coded representation is then stored separately from
the more detailed representation on which it is based, and can correspondingly be retrieved independently of the latter representation for use in later
processing.
This latter processing could lead still other representations to be constructed. In the preceding example, people might later recall their abstract
representation of the experience and use it to make a trait inference about
the protagonists. This inference, in turn, could constitute yet another representation that is stored independently of either of the others. Note that according to Postulate 2.1, a higher order representation does not replace the
original. That is, all of the knowledge representations that have been
formed coexist in memory. Whether or not a given representation is retrieved, however, depends on factors described shortly.
Direct evidence for the independent-storage postulate is somewhat limited. However, a study by Klein and Loftus (1993) is suggestive. Participants
were asked both to report whether they had each of several traits and, in
addition, to recall a specific behavior that exemplified the traits. However,
the order of responding to the two requests was varied. The time that participants required to verify that they had a trait did not depend on whether
they had previously recalled a behavior that exemplified it. By the same token, the time to report a trait-related behavior did not depend on whether
or not participants had previously reported having the trait implied by this
behavior. Thus, participants did not base their trait judgments on the relevant behavior that was salient to them at the time. Furthermore, making a
trait judgment that a behavior exemplified did not facilitate their later recall
of this behavior. These findings suggest the independence of traits and behavior in memory. The independence of trait representations and behavioral representations may be more characteristic of self-knowledge than of
knowledge about others. However, other evidence of the independence of
conceptually related knowledge representations are discussed in chapter 6
as well as elsewhere in this volume.
This conceptualization obviously requires the assumption that Permanent Storage has unlimited capacity. That is, once a representation has
been encoded into memory, it is never erased. Given the virtually infinite
number of experiences that individuals have over the course of a lifetime,
this assumption may seem nonsensical. An alternative assumption is suggested by distributed memory or connectionist models, which assume that
memories are not localized but rather, are represented by an array of activated features that, in different strengths and combinations, can capture
the sum total of experiences to which the system has been exposed. In principle, a very large number of unique memories can theoretically be captured by a small number of feature nodes. (E.g., 10 nodes, each of which can
39
be assigned 10 different strengths, could potentially represent 1010 unique

memories.)
Several distributed memory models of social information processing
have been proposed (e.g., Humphreys & Kashima, 2002; Smith & DeCoster,
1998, 1999), and it is conceivable that such a conceptualization of Permanent Storage would be viable in the present context. However, all theories
of mental representation are inherently metaphorical, as noted in chapter
1, and should be evaluated on the basis of their utility and not their validity.
The assumptions underlying the conceptualization we propose provide a
more parsimonious account of the phenomena being considered in this volume than the distributed memory models proposed to date.
Retrieval Processes
The retrieval processes we postulate are borrowed in part from a previous
formulation by Ratcliff (1978) and make use of a tuning fork metaphor (for
related conceptualizations, see Albrecht & Myers, 1995; Kintsch, 1998).
Many implications of this conceptualization are somewhat similar to those
of a spreading activation, associative network model (cf. J. Anderson, 1983;
Collins & Loftus, 1975; Wyer & Carlston, 1979). In contrast to these latter
models, however, our formulation requires no assumptions about the organization of these knowledge representations in memory.
We assume that when a configuration of stimulus features is activated,
they vibrate and this causes the configuration to resonate with knowledge representations (memory traces) that contain some or all of these features. If the resonance of a knowledge representation exceeds a threshold
value, it is also activated. When a knowledge representation is no longer involved in processing, it becomes deactivated. Like a tuning fork, however,
the resonance does not dissipate instantly, but decreases gradually over
time. These processes can be stated more formally in terms of five additional postulates.
Postulate 2.2. The activation of a set of stimulus features causes all knowledge representations that contain these features to resonate. The resonance
of a given representation, k, increases with (a) the similarity of its features to
those in the stimulus set and (b) the length of time the stimulus set has been
activated. The subjective similarity of the stimulus features to the representation, and thus the rate at which its resonance increases, is estimated by the
quantity:
Res(k) =
n(pI k )
,
n(k )
where n(k) is the number of features in the representation and n( p I k ) is the

number of features that are common to both the representation and the stim-
40
CHAPTER 2
ulus set. If the resonance reaches a threshold value, T (k), the representation
is activated.
One implication of Postulate 2.2 is worth noting at this point. That is, when
features of the probe set are contained in more than one representation,
the resonance at the representation with the fewest number of noncommon
features will reach threshold faster. Thus, the probe football game is
more likely to activate an abstract representation of football games than a
specific representation of the 2002 Super Bowl.
Postulate 2.3. (a) If two or more representations resonate with a given set of
probe cues, the one that reaches threshold most quickly is identified. (b) If
several representations reach threshold simultaneously, a composite of the
features is activated. The weight of each feature in the composite is a positive
function of the proportion of times it occurs in the set of representations involved.
Postulate 2.3a is consistent with the assumptions of the independent

trace model proposed by Logan (1988). Postulate 2.3b is suggested by
Hintzmans (1986) conception of the process of abstracting general concepts from exemplars. For example, it implies that when several relevant
knowledge representations are activated by a given set of stimulus features, the features that are typical of these representations as a whole are
most likely to come to mind and other features, which characterize only a
few of these representations, tend to be filtered out. Thus, suppose a baseball fan who encounters the statement Nolan Ryan pitched a no-hitter has
a single instance of this event represented in memory. In this case, the features Nolan Ryan and no-hitter in combination should activate a detailed representation of the game in question. In contrast, a fan who has observed all six of Ryans no-hit games is likely to activate a more general set
of features that are common to these games without thinking about any
particular one of them. But suppose the same two individuals encounter
the general statement a ball player pitched a no-hitter. In this case, both
fans might activate a general set of features that do not specify any given
baseball pitcher or behavior.
The preceding analysis suggests that if baseball fans encounter the statement Nolan Ryan pitched a no-hitter, they will typically comprehend it in
terms of a previously formed representation of Nolan Ryan engaging in this
activity rather than in terms of more general semantic concepts associated
with baseball players and no-hit games. That is, they will comprehend the
statement in terms of different concepts than they would if they encountered the statement a baseball player pitched a no-hitter. Although these
observations may seem intuitively obvious, Wyer and Srulls (1989) model
would not make these distinctions.
41
Determinants of Knowledge Accessibility

The next two postulates, which are implied by numerous studies of the determinants and effects of knowledge accessibility (for reviews, see Bargh,
1997; Higgins, 1996; Wyer & Srull, 1989), become important in conceptualizing the effect of previously using a knowledge representation on their subsequent reactivation and use at a later time.
Postulate 2.4. If a knowledge representation has resonated with a set of
probe cues, this resonance does not dissipate immediately but rather, decays
gradually over time.
Postulate 2.5. The activation threshold of a knowledge representation in
memory, T(k), is an increasing function of the number of times the representation has been activated in the past.
In effect, Postulate 2.4 permits the theory to account for the effect of recency of activation on concept and knowledge accessibility that is consistently observed in research on priming effects (Higgins, Bargh, & Lombardi,
1985; Srull & Wyer, 1979). Postulate 2.5 permits the theory to account for the
effect of frequency of activation that is also observed in these studies. One
implication of these postulates is worth noting. Specifically, the effect of recency of activation, which is a function of the residual resonance that exists
at a memory location after the knowledge representation at the location
has been deactivated, is transitory. That is, the effect decreases over time
as the residual resonance dissipates. In contrast, the effect of activation frequency, which is localized in the level of the activation threshold, is relatively permanent. These implications of the theory are discussed more fully
in chapter 3.
GOAL-DIRECTED VERSUS NON-GOAL-DIRECTED

PROCESSING: SPONTANEOUS REMINDING
PROCESSES
Postulate 2.2 specifies the rules that govern the retrieval of a particular
knowledge representation from Permanent Storage, given that a set of features are identified for use as probe cues. However, it does not indicate
the processes that underlie the selection of these features. These processes
typically come into play once information has been comprehended and
transmitted to the Work Space and, therefore, are governed by the Executor. The nature of this cognitive activity depends in part on whether or not
the information is processed in pursuit of a particular goal.
42
CHAPTER 2
Goal-Directed Retrieval Processes

If previously acquired knowledge is required for a particular purpose, the
following steps are assumed:
1. The Executor compiles a set of features that exist in the Work Space,
giving greater weight to those that have been activated and used more recently. These features (probe cues) are likely to be ones that (a) are relevant to the objectives one is pursuing and (b) have recently been transmitted to the Work Space by the Comprehender. However, the number of
features that are compiled may depend on the specificity of the objective being pursued. When this objective is relatively nonspecific, features may be
fortuitously included in the set that are not relevant to it.
2. The probe cues identified in Step 1 vibrate, leading them to resonate
with knowledge representations in either the Work Space or Permanent Storage that contain them. The representation(s) that reach activation threshold
most quickly, based on criteria specified in Postulate 2.2, are identified, and
copies of them are transmitted to the Work Space for use in goal-directed activity.
3. If no representation is identified in Step 2, the set of probe features is
recompiled and the search is repeated.
These processes are somewhat similar to those proposed by Norman
and Bobrow (1979) in accounting for the retrieval of information from very
long-term memory. The fact that goal-irrelevant features are sometimes included in the probe set has two noteworthy implications. First, the inclusion of these features can sometimes prevent the identification of a goalrelevant representation that exists in memory. This may be experienced
subjectively as a tip of the tongue phenomenon, where one is temporarily
unable to identify the name of a well-known acquaintance or a movie one
has seen. In these instances, extraneous probe cues may interfere with the
retrieval of a representation containing this information, and so the name
may not come to mind unless the set of probe cues is recompiled.
Second, the configuration of probe cues that results from the presence
of a goal-irrelevant feature can sometimes lead to the retrieval of knowledge that would not normally come to mind. This can account for the spontaneous recall of a person or event that had not been thought about for
many years. To borrow an example from Wyer and Srull (1989), a person
who is asked to describe his hometown might normally use the name of the
town (e.g., Delhi, New York) as a probe cue, activating a general representation of its physical characteristics and the type of people who live there.
However, suppose the individual had been discussing religion just before
being asked to provide this description. Then, a feature activated by this
43
discussion (e.g., church) might be fortuitously included in the probe set.

This feature, in combination with the towns name, might stimulate a memory of adjourning to the basement of the church while the collection was
being taken in order to avoid putting money in the plate. This behavior may
not have been thought about for 40 years, and might have been buried in
memory forever if the two probe features had not been fortuitously activated in combination.3
Information Processing During the Free Flow
of Thought
Much of our cognitive activity in daily life appears to occur without any particular objective in mind. We frequently engage in daydreaming, or let our
minds wander aimlessly from one topic to another in an ostensibly haphazard manner. In the course of this free flow of thought, however, we may be
reminded of something we have to do, or of a problem we need to solve,
and we suddenly find ourselves immersed in thoughts about how to accomplish these objectives. At other times, our cognitive musings can stimulate
a memory of a past experience or acquaintance that we have not thought
about in years. A complete formulation of information processing in situations outside the laboratory must be able to specify the processes that surround the free flow of thought and the transition of these processes into
more deliberative, goal-directed cognitive activities.
To account for these processes within the present conceptualization, we
assume that when individuals do not have a specific goal in mind, the processing system enters into a continuous feedback loop that continues indefinitely until a goal is identified, either as a result of external demands (i.e.,
transmitted by the Comprehender) or internally generated cognitive activity. These processes can be described easily with reference to Fig. 2.2. The
3
In comparing the retrieval processes described here with the processes postulated by
Wyer and Srull (1989), one further implication should be noted. The earlier model assumed that
in the pursuit of goal-relevant information, the Work Space was searched first, before information is retrieved from Permanent Storage. In the present conceptualization, this assumption is
an implicit by-product of Postulate 2.5. That is, a set of probe cues is likely to resonate to the
greatest extent with existing knowledge representations that have been activated and used
most recently. Because the knowledge representations contained in the Work Space have typically been used recently in the pursuit of processing objectives, they would normally be reactivated more quickly than representations in Permanent Storage for use in attaining goals to
which they are relevant. To this extent, the two conceptualizations are quite compatible. On the
other hand, the content of the Work Space typically consists of representations that have either
been (a) recently recalled from Permanent Storage, or (b) recently formed and transmitted to
Permanent Storage by the Executor. These representations would be activated in Permanent
Storage according to criteria assumed by Postulate 2.2. To this extent, the implications of the assumption that the Work Space is searched first are largely redundant with those of the retrieval
postulates that we have already stated.
FIG. 2.2. Flow diagram of information processing in the absence of a specific

objective (e.g., during the free flow of thought). Circles denote the processes
involved. Diamonds indicate conditions that give rise to different types of
processing.
44
45
first two stages of processing are similar to Steps 1 and 2 of the process assumed to underlie goal-directed retrieval. That is, the Executor identifies a
set of features that happen to exist in the Work Space (either transmitted
from the Comprehender or resulting from prior cognitive activity). These
features in combination activate a representation in Permanent Storage, a
copy of which is sent to the Work Space. The Executor scans this representation for a goal specification and, if none is found, repeats the process just
described, recompiling a set of probe cues and using them to activate another representation in Permanent Storage. This continues until a goal
specification is identified as a result of the content of the material retrieved
from Permanent Storage or transmitted from the Comprehender.
Phenomenologically, this simply means that when people do not have a
particular goal in mind, the configuration of cognitions they happen to be
thinking about cues the retrieval of a previously formed representation.
Features of this representation, once retrieved, may combine with other activated concepts to cue the retrieval of yet another representation, and so
on until a goal is identified.
Thus, for example, suppose a student overhears a conversation in which
one person comments that John gave Peter an answer during the exam. A
representation of the statement that is constructed in the course of comprehending it is transmitted to the Work Space. Features of this representation (e.g., exam) along with other features that happen to be in the Work
Space at the time (e.g., psychology) are sampled, and a set of features is
compiled (e.g., psychology, exam). These features may resonate with a
representation in Permanent Storage that contains them (e.g., the thought
that the final exam in psychology is only 3 days before Christmas). As this
knowledge does not have goal implications, a new set of probe features is
compiled which might now include Christmas. These features then resonate with representations in Permanent Storage that contain this feature,
perhaps leading a representation to be retrieved that concerns buying a
Christmas present for ones boyfriend. In this case, however, the Executor
recognizes this as a goal. Consequently, the system leaves the feedback
loop. That is, the Executor identifies a goal schema pertaining to the objective and deposits it in the Goal Specification Box, and then uses the content
of the schema to direct processing relevant to attainment of the goal (e.g.,
the construction of a plan about what to buy and where to purchase it).
CONCLUDING REMARKS
The information-processing model outlined in this chapter provides a
framework for conceptualizing much if not all of the theory and research
discussed in the remainder of this volume. This will be particularly clear in
46
CHAPTER 2
our analysis of the comprehension of personal experiences, and of information about actual persons and events of the sort we encounter in the course
of daily life. Several important implications of the conceptualization are localized in Postulates 2.4 and 2.5, which govern the factors that influence the
accessibility of knowledge in memory and, therefore, the likelihood that
this knowledge is brought to bear on the processing of new information. Indeed, knowledge accessibility is a primary consideration in all phases of information processing from the initial comprehension of information to the
generation of an overt response. Before embarking on a more detailed analysis of comprehension and judgment, therefore, a more general discussion
of the determinants and effects of knowledge accessibility will be useful.
The next chapter provides this discussion.
C H A P T E R
3
Determinants and Effects
of Knowledge Accessibility
The retrieval postulates outlined in chapter 2 come into play at many

stages of processing and therefore play a role in our discussion throughout
this volume. However, they have particular implications for the factors that
influence the accessibility of concepts and knowledge in memory. The important role of knowledge accessibility in information processing is recognized in numerous areas of social psychology, including the comprehension of information in terms of semantic concepts (Higgins, Rholes, & Jones,
1977; Srull & Wyer, 1979, 1980), stereotyping (Bodenhausen & Macrae, 1998),
consumer judgment (Adaval & Monroe, 2002), cultural influences on judgment (Briley & Wyer, 2002; Hong, Morris, Chiu, & Benet-Martinez, 2000), and
overt behavior (Bargh, Chen, & Burrows, 1996; Dijksterhuis & van
Knippenberg, 1998). Higgins (1996) provides an extensive review of this research, and a duplication of his efforts is unnecessary here. In this chapter,
we restrict our attention to representative phenomena for which the proposed conceptualization has implications.
GENERAL CONSIDERATIONS
The general effects of concept and knowledge accessibility are intuitively
obvious. That is, people who are called upon to make a judgment or decision seldom consider the implications of all of the knowledge they have acquired that bears upon it. Instead, they use the first relevant judgment or
decision criterion that comes to mind, ignoring other criteria that might be
equally or more applicable. Taylor and Fiske (1978) were among the first to
47
48
CHAPTER 3
recognize this possibility, arguing that many judgments and decisions are
made off the top of the head. A more formal statement of this phenomenon was proposed by Chaiken (1980, 1987). She argued that people who
have to make a judgment or decision first apply the criterion that comes to
mind most quickly and easily and then assess their confidence that the implications of using this criterion are sufficient to attain the goal at hand.
Only if this confidence is below some minimal confidence threshold do
they search for additional criteria. Thus, they may search further if they are
highly motivated to make a correct decision, or if for some reason they
believe that the criterion they initially applied may be biased. Normally,
however, they base their judgment or decision on the first relevant criterion they happen to think about.
The implications of Chaikens conceptualization can be formalized in the
following, sufficiency postulate:
Postulate 3.1. People retrieve and use only the amount of information that
they consider to be sufficient for attaining the processing objective they are
pursuing at the time. The information they use is likely to be the first goalrelevant information that comes to mind or, alternatively, permits their objective to be attained most quickly and easily.
Chaikens formulation has been applied primarily in conceptualizing responses to persuasive communications (Chaiken, 1987; Chaiken, Wood, &
Eagly, 1996). However, it has more general applicability to information processing (Chaiken, Liberman, & Eagly, 1989), and many of the phenomena
reviewed later in this volume can be viewed from the perspective she suggests. The discussion in this chapter is restricted to representative research that bears on three main points to which the formulation outlined in
chapter 2 is particularly relevant:
1. The primary determinants of knowledge accessibility are (a) the recency with which knowledge has been acquired and/or used in the past and
(b) the frequency with which it has been applied. In the latter regard, knowledge that has been employed in many situations over a long period of time
can become chronically accessible.
2. The accessibility of knowledge representations for use in attaining a
particular processing objective can be activated by extraneous factors that
are objectively irrelevant to this objective and of which persons may not be
consciously aware. In fact, increasing awareness of these extraneous factors
is likely to decrease their influence.
3. The effects of differences in knowledge accessibility are apparent at
several stages of processing, including the spontaneous comprehension of
information at the time it is first received, the more deliberative interpreta-
DETERMINANTS AND EFFECTS OF KNOWLEDGE ACCESSIBILITY
49
tion of information that occurs in the course of attaining a particular goal, inference processes, and behavioral decision making. Differences in knowledge accessibility can also influence the goals that individuals pursue under
conditions in which more than one objective is potentially relevant, and the
procedures that are spontaneously activated and used to attain a particular
goal (e.g., productions of the sort postulated by Anderson, 1983).
FREQUENCY AND RECENCY EFFECTS

OF KNOWLEDGE ACTIVATION ON JUDGMENTS
AND DECISIONS
The earliest research in social cognition to examine the impact of knowledge accessibility was conducted in the area of person impression formation. When people are asked to form an impression of someone on the basis
of the persons behaviors, they often interpret these behaviors in terms of
traits. This activity is presumably performed by the Encoder/Organizer
based on concepts that are transmitted to it by the Executor. Many behaviors are ambiguous. That is, they have implications for values along several
different trait dimensions or, alternatively, for different values along the
same dimension. Under these conditions, the interpretation of the behavior
can depend on the particular concept that is brought to bear on it. This, in
turn, can depend on both the recency and the frequency of its use in the
past and, therefore, its accessibility in memory (Postulates 2.4 and 2.5).
Once the behavior of a person is interpreted, this interpretation provides
the basis for a more general characterization of the person and an evaluation of his or her general likeableness.
Early Demonstrations
Higgins et al.s (1977) well-known study was among the first to demonstrate
the effects of recency. Participants were asked to form an impression of a
target person who ostensibly wanted to cross the Atlantic in a sailboat, was
well aware of his ability to do things well, felt that he didnt need to rely on
others, and rarely changed his mind once he had made a decision. Before
reading a description of these characteristics, however, participants were
unobtrusively exposed to a number of trait adjectives in the course of performing a color-naming (Stroop) task. These adjectives could all be used to
interpret the targets behaviors. In some cases, however, the adjectives
were favorable (adventurous, confident, independent, and persistent) and in other cases, they were unfavorable (reckless, conceited,
aloof, and stubborn). Participants apparently interpreted the information in terms of concepts activated by the trait adjectives to which they had
50
CHAPTER 3
been exposed in the color-naming task, and these interpretations influenced their evaluations of the target. That is, they evaluated the target
more favorably when they had been exposed to favorable trait adjectives
than when they had been exposed to unfavorable ones.
In a similar series of studies (Srull & Wyer, 1979, 1980), participants performed a sentence construction task that required the use of concepts associated with hostility. Later, they were asked to form impressions of a target
person whose behaviors were ambiguous with respect to this trait (e.g., refused to pay the rent until the landlord painted his apartment, demanded
his money back from a salesperson, etc.). Both the number of hostilityrelated items presented in the sentence-construction task and the time interval between this task and the impression formation task (immediately, 1
hour or 1 day) were manipulated. Both of these factors influenced the likelihood that participants interpreted the targets behavior as hostile, indicating that both frequency and recency played a role. Moreover, a supplementary study indicated that the effects only occurred when participants were
exposed to trait concepts before they read the target information (Srull &
Wyer, 1980). When the concepts were activated after the information was
presented, they had no effect at all. This indicates that the effects of primed
concepts on judgments were mediated by their impact on the interpretation of the trait information at the time the information was received and
did not have a direct influence on judgments independently of this information.
The Relative Persistence of Frequency
and Recency Effects
Although the frequency and recency of prior activation both have an impact on the use of trait concepts to interpret information, the underlying basis for these effects theoretically differ. As noted earlier, the effects of recency of activation implied by Postulate 2.4 are transitory, whereas the
effects of activation frequency, which impacts on the activation threshold
(Postulate 2.5) are more enduring.
This was demonstrated empirically by Higgins, Bargh, and Lombardi
(1985). Participants were initially exposed to adjectives exemplifying both
favorable and unfavorable trait concepts (adventurous vs. reckless)
that could each be used to interpret the same behavior (e.g., crossing the
Atlantic in a sailboat). However, the relative frequency and recency of exposure to the adjectives were varied. When participants were asked to interpret the behaviors a very short time after exposure to the trait adjective,
they tended to use the more recency primed trait concept rather than the
frequently activated one. When a longer period of time had elapsed, however, the effects of activation frequency predominated over recency, as im-
51
plied by Postulates 2.4 and 2.5. (For an alternative conceptualization of

these effects, see Higgins et al., 1985.)
The enduring influence of activation frequency has important implications. That is, concepts that have been frequency used in the past can become chronically accessible and, therefore, can influence comprehension
and judgments in the absence of any immediate situational factors that activate them. A study by Anderson, Reynolds, Schallert, and Goetz (1977) provides an example. Participants were asked to read the following story:
Every Saturday night, four good friends get together. When Jerry, Mike, and
Pat arrived, Karen was sitting in her living room writing some notes. She
quickly gathered the cards and stood up to greet her friends at the door. They
followed her into the living room but as usual, they couldnt agree on exactly
what to play. Jerry eventually took a stand and set things up. Finally, they began to play. Karens recorder filled the room with soft and pleasant music.
Early in the evening, Mike noticed Pats hand and the many diamonds. As the
night progressed, the tempo of play increased. Finally, a lull in the activities
occurred. Taking advantage of this, Jerry pondered the arrangement in front
of him. Mike interrupted Jerrys reverie and said, Lets hear the score. They
listened carefully and commented on their performance. When the comments
were all heard, exhausted but happy, Karens friends went home.
After reading the story, participants were asked what the protagonists had
commented about.
The story is normally interpreted as the description of a card game.
However, it could also describe the rehearsal of a woodwind ensemble.
Some participants, who were students in a weight-lifting class, typically indicated that the protagonists had commented on how well they were playing cards. In contrast, students who were planning a career in music education indicated that the protagonists had commented on the sound of their
music. Thus, concepts and knowledge that were chronically accessible to
participants as a result of their curriculum and interests apparently affected their interpretation of the story and this interpretation influenced
their answers to questions about it.
The Additivity of Situationally Induced
and Chronic Knowledge Accessibility
A further implication of Postulates 2.4 and 2.5 is of particular interest in this
regard. That is, the effect of activation frequency (which are mediated by
changes in activation threshold) and the recency of activation (which are
mediated by the amount of residual resonance that remains at a knowledge
representation once it is no longer activated) are independent. This independence was demonstrated by Bargh, Bond, Lombardi, and Tota (1986).
52
CHAPTER 3
They showed that the chronic accessibility of a trait concept (inferred from
individual differences in the a priori likelihood of using the trait to describe
acquaintances in a previous experiment) and situationally induced accessibility of the trait (resulting from exposure to the trait in a laboratory task)
had independent and additive effects on the use of the concept to describe
a target person on the basis of information about the targets behavior.
Although the independence of these interpolated effects are implied by
Postulates 2.4 and 2.5, they are difficult to explain on the basis of the bin
conceptualization proposed by Wyer and Srull (1989). According to the latter conceptualization, priming trait concepts should make them equally accessible (either in the Work Space or in the semantic bin) regardless of
the frequency of their prior activation, and consequently should lead these
concepts to be used to an equal extent regardless of their chronic accessibility. In other words, recency should override the effects of chronic accessibility. Therefore, Bargh et al.s (1986) findings are embarrassing to the
Wyer and Srull model. On the other hand, they are quite consistent with the
retrieval postulates of the present conceptualization.
Generalizability of Concept Activation
Over Attribute Dimensions
In the study by Higgins et al. (1977), activating favorable and unfavorable
trait concepts that were descriptively inapplicable for interpreting the targets behavior had no influence on participants evaluations of him. This
finding suggests that the effects of activating trait concepts do not have a
direct impact on judgments of the target, but rather, influence judgments
only through their mediating influence on how the information about the
persons behaviors is interpreted. As we noted earlier, however, impression
formation is a two-stage process. Once a target persons behaviors are interpreted and a general impression is formed of him, the person may be inferred to have other attributes that are characteristic of the sort of individual implied by this impression. Thus, as Srull and Wyer (1979, 1980) found,
activating concepts associated with hostility led the target person not only
to be described as hostile but also to have other attributes that are typical
of a hostile individual. (For a more recent explication of these two processing stages, see Trope, 1986.)
The conclusion that activated trait concepts influence the interpretation
of only those behaviors to which they are descriptively applicable may nevertheless need to be qualified. Stapel and Koomen (2000) found that when
stimulus attributes with extreme evaluative implications (malevolent,
warm, etc.) are primed, they do appear to influence the interpretation of
behaviors that are descriptively unrelated to them. Apparently, adjectives
with extreme implications spontaneously activate more general evaluative
53
concepts (good or bad) that, once activated, influence the interpretation of information along evaluative as well as descriptive dimensions.
EFFECTS OF AWARENESS
Many effects of previously acquired concepts and knowledge on the interpretations of information can occur without awareness of the factors that
led them to become accessible in memory. This possibility is implicit in the
conceptualization we propose. That is, consciousness theoretically resides
in the Executor. The processes performed by the Comprehender, which are
not controlled by the Executor, occur without awareness. Correspondingly,
the knowledge representations that are involved in these spontaneous
comprehension processes can later be activated and applied without consciousness of why they came to mind or, for that matter, without awareness
that these representations had recently been used. For similar reasons, the
representations that are later identified by the Executor for use in goaldirected processing are unlikely to contain cognitions about the recency or
frequency with which the representations have been formed or used in the
past, as these cognitions are typically irrelevant to the objectives for which
the representations were formed. Therefore, these representations are
likely to be used as bases for judgments and decisions without considering
the reasons for their accessibility. Consequently, as studies cited in the preceding section testify, factors that fortuitously make knowledge accessible
in memory can increase the use of this knowledge. Moreover, this can occur without awareness of the factors that influence its accessibility.
Nonconscious Effects of Knowledge Accessibility
The first compelling demonstration of the nonconscious influence of concepts and knowledge was reported by Bargh and Pietromonaco (1982). In
this research, participants were exposed to a computerized text of perceptual vigilance that required them to press a button as soon as a light appeared on the computer screen. In actuality, the points of light were words
that were presented subliminally. In some conditions, a number of words
conveyed hostility (hostile, kill, etc.). After performing the task, participants as part of a different experiment were asked to form an impression of
a person on the basis of behaviors that were ambiguous in terms of the hostility they conveyed. Participants were unable to distinguish between
words that were presented and those that were not presented in a recognition memory task after the experiment. Nevertheless, they judged the target to be more hostile if they had been subliminally exposed to hostilityrelated words than if they had not.
54
CHAPTER 3
Even when people are conscious of the conditions in which knowledge

has been acquired, they may not include a specification of these conditions
in the representation of this knowledge that they store in memory. Consequently, they may later retrieve and use the representation as a basis for
judgment without considering the conditions that led the knowledge to be
acquired. Two studies are interesting to view in this light. In a study by
Hasher, Goldstein, and Toppin (1977; see also Kelley & Lindsey, 1993), participants completed a general knowledge questionnaire in which they
judged the validity of several statements about unfamiliar events and situations. Then, in a second session of the experiment, they completed a similar
questionnaire containing some of the same statements they had considered
earlier. Participants reported stronger beliefs in the validity of these statements than they had in the first session. Representations of the belief statements that participants had considered in the first session were apparently
stored in memory without any indication of the conditions in which they
were encountered. Consequently, these representations were retrieved
when participants encountered the statements in the second session, leading the statements to seem familiar. However, participants attributed the
statements familiarity to information they had received outside the laboratory rather than to the presence of the statements in the original questionnaire. As a result, they inferred that the statements were likely to be true.
In a conceptually similar study, Jacoby, Kelley, Brown, and Jasechko
(1989) exposed participants to a number of names of persons, some of
whom were familiar and others of whom were fictitious. Then, 24 hours
later, they were given a second list that included a subset of the fictitious
names they had seen earlier and were asked to indicate which names in the
list referred to well-known persons. Participants were more likely to misidentify fictitious persons as well known if they had seen their names in the
earlier list than if they had not. Thus, both this study and Hasher et al.s research suggest that knowledge that is activated in one context can later be
recalled and used as a basis for judgments without considering the reasons
that led it to be familiar.
The general tendency for representations of knowledge to be stored in
memory independently of the conditions in which the knowledge was acquired has particularly important implications for the impact of daily life
experiences on perceptions of social reality. As Shrum and his colleagues
(OGuinn & Shrum, 1997; Shrum, Wyer, & OGuinn, 1998) have found, people
often confuse the persons and events they encounter in the real world with
those they have seen on television. Consequently, heavy television viewers
tend to overestimate the incidence of events that occur frequently in soap
operas and crime shows. We elaborate this research more fully in chapter
10 in the context of discussing the role of implicit theories on judgments
and decisions.
55
Effects of Awareness on Avoidance of Bias

In fact, when people are aware that the concepts they used when performing an ostensibly unrelated task might influence their interpretation of the
information they receive later, they may consciously avoid the use of
these concepts. In a study by Lombardi, Higgins, and Bargh (1987), for example, participants were exposed to trait concepts in a priming task that
were associated with earlier (a) adventurousness and self-confidence or
(b) recklessness and conceitedness. They were then asked to report their
impressions of a person whose behaviors could exemplify both sets of
traits and, finally, were asked to recall the priming words to which they
had been exposed. Participants who could not remember the priming
words were positively influenced by them in much the same way observed by Higgins et al. (1977) in the study described earlier. However,
participants who could remember the words showed a boomerang effect;
that is, they were less inclined to interpret the targets behavior in terms
of the primed concepts than in terms of the alternatives. Similar results
were obtained by Strack, Schwarz, Bless, and Kuebler (1993) in a different
research paradigm.
Even when people are aware of the reasons why concepts and knowledge come to mind, they may not always avoid using them. This avoidance
should only occur when the individuals are both motivated and able to
search for alternatives. Martin, Seta, and Crelia (1990) demonstrated these
contingencies under conditions in which trait concepts that were applicable for interpreting a targets behavior were blatantly primed. When participants were both able and motivated to search for alternative concepts to
use, boomerang effects of priming occurred; that is, participants were less
inclined to use the primed concept than they were in the absence of priming. However, when participants were distracted while performing the evaluation task and were less able to search for alternative concepts to apply,
primed concepts had a positive impact on judgments similar to those observed by Higgins et al. (1977) and others.
A second study by Martin et al. (1990) showed that primed concepts only
had a positive influence on target judgments when participants had little intrinsic motivation to think about the information received (i.e., low need for
cognition; see Cacioppo & Petty, 1982) but had boomerang effects on judgments by participants whose motivation (need for cognition) was high. Particularly interesting are the results of a third study in which participants
were led to believe they were forming impressions of the target person either as individuals or as members of a group. Primed concepts had boomerang effects on judgments by individuals who believed they were participating as individuals. However, participants who thought they were
participating as a group engaged in social loafing (Harkins, Latane, & Wil-
56
CHAPTER 3
liams, 1980). That is, they used the primed concepts as a basis for interpreting the information without bothering to seek alternatives.
The Effect of Thought Suppression
on Concept Accessibility
It therefore seems reasonable to conclude that when people are aware
that the concepts and knowledge that come to mind at the time they make
a judgment, they often actively avoid the biasing influence of these cognitions and, in doing so, are actually less likely to use them than they might
otherwise be. However, the effects of this cognitive activity can be somewhat ironic. Suppose individuals consciously avoid the use of a concept or
knowledge representation in responding to a stimulus. This conscious suppression takes cognitive effort. Moreover, it is impossible to consciously
avoid using a concept without thinking about the concept one is trying
to avoid. Consequently, active attempts to avoid using a concept can actually increase its accessibility in memory, making the concept more likely to
be used later under conditions in which attempts to suppress it are no longer made. Thus, as Wegner (1994) noted, instructing people to avoid thinking about a white bear may stimulate conscious thoughts about the bear,
thus making it more accessible in memory than it would have been in the
absence of this cognitive activity and, therefore, more likely to come to
mind later.4
An intriguing series of studies by Bodenhausen, Macrae, and their colleagues demonstrated the effects of conscious thought suppression in the
use of stereotypes (for a review, see Bodenhausen & Macrae, 1998). In a representative set of studies (Macrae, Bodenhausen, Milne, & Jetten, 1994),
participants were given information about a skinhead with instructions to
form an impression of him. However, some participants were told not to
rely on stereotypes in arriving at their impression, whereas others were not
given this instruction. The former participants, relative to those who had
not been told to avoid using the stereotype, (a) responded more quickly to
stereotype-related words in a subsequent lexical decision task, (b) described a second skinhead more stereotypically once the request to avoid
using stereotypes was no longer imposed, and (c) sat further away from a
skinhead while waiting to participate in a later part of the experiment. Thus,
4
A game we used to play as children anticipated Wegners insight. That is, we would tell
nave playmates to try not to think of a white horse and then, after ensuring that they were doing
so, would ask them to report the direction in which the horse was facing. They invariably found
this question quite meaningful and responded to it quickly. Attempts to suppress thoughts
about the horse typically elicited an image of the horse that individuals found easy to describe
despite their assurances that they were trying not to think about it.
57
the attempt to suppress the use of the stereotype in the initial task increased rather than decreased the accessibility of concepts associated with
the stereotype. Consequently, it increased the use of the stereotype as a basis for judgments and behavioral decisions in situations in which the sanctions against using it were removed.
EFFECTS OF KNOWLEDGE ACCESSIBILITY

AT DIFFERENT STAGES OF PROCESSING
The retrieval processes implied by Postulates 2.1 through 2.5 apply at all
stages of cognitive activity, from the initial interpretation of information to
the generation of an overt response. The role of knowledge accessibility in
these processes is central to our discussion throughout this volume, being
particularly evident in conceptualizing the impact of generalized representations on memory for exemplars (chap. 5), the role of implicit theories in
explanation and prediction (chap. 10), and the impact of affective reactions
on judgments and decisions (chap. 11). In anticipation of this discussion, a
brief review of the effects of knowledge accessibility at different stages of
processing may be helpful. This review is not intended to be exhaustive (for
more extensive reviews, see Bargh, 1997; Higgins, 1996). Instead, we provide
representative examples of the effects that occur, several of which raise
considerations that become important in our later discussion.
Input Processes: Selective Attention
and the Activation of Bipolar Concepts
As Bargh and Pietromonacos (1982) findings indicate (see also Srull &
Wyer, 1979, 1980), information that is ambiguous in terms of its implications
for a given attribute (e.g., hostility) is more likely to be interpreted as exemplifying this attribute when concepts associated with this attribute are easily accessible in memory than when they are not. However, many attribute
concepts are bipolar (hot vs. cold, hostile vs. kind, etc.). Moreover, these
concepts may become associatively linked in memory by virtue of having
been thought about in relation to one another. If this is so, exposure to one
bipolar concept (e.g., honest) is likely to activate the other (dishonest)
as well and, therefore, to increase the likelihood that the latter concept is
used to interpret information to which it applies.
To see the implications of this, suppose two pairs of bipolar concepts,
honest/dishonest and kind/unkind, are associatively linked as shown in
Fig. 3.1. Suppose further that the behavior, refuses to give an answer to a
friend during an exam is associatively linked to both honest and un-
58
CHAPTER 3
FIG. 3.1. Hypothetical associative network showing the relations among behaviors, bipolar trait concepts (honest, dishonest, kind, and unkind), and the
priming stimuli that activate these concepts (Ph, Pd, Pk, and Pu, respectively).
kind. Then, priming honest should increase the likelihood of interpreting

the behavior as honest, whereas priming unkind should increase the
likelihood of interpreting it as unkind. However, because dishonest and
kind are associatively linked to honest and unkind, respectively, priming these concepts should have similar effects. In other words, priming
either bipolar concept along a trait dimension should increase the likelihood of using a value along this dimension to interpret information to
which the dimension is relevant.
Park, Yoon, Kim, and Wyer (2001) confirmed these hypotheses. In one experiment (Park et al., 2001, Experiment 1), participants performed a scrambled sentence task that required the use of one of four concepts: kind (e.g.,
offered carry to help elder and luggage), unkind (refused carry to help elder and luggage), honest (e.g., homework to copy refused anothers) or
dishonest (e.g., homework to copy tried anothers). Later, they read a
story containing descriptions of a target persons behavior that could be interpreted as both honest and unkind (told a girlfriend that her new hair
style looked terrible, insisted that a friend leave his address on a car he
had barely scratched, etc.). Participants read the behaviors and then both
described the target in their own words and, after doing so, evaluated the
person.
The proportions of participants who used the trait exemplified by the behavior (i.e., either honest or unkind) to describe the target are shown in Table 3.1 as a function of priming conditions. When primed concept was directly applicable for interpreting the targets behavior, participants were
more likely to use this concept (pooled over the two relevant priming conditions, M = .294) than to use the equally applicable but unprimed concept
59

TABLE 3.1
The Proportion of Times the Target Was Described
as Honest and Unkind as a Function of Priming Conditions
Primed Concept
Applicable for
Describing Target
Proportion of times the

target was described as
Honest
Unkind
Primed Concept
Inapplicable for
Describing Target
Honest
Unkind
Dishonest
Kind
.280
.160
.115
.308
.407
.074
.231
.231
Note. Based on data from Park et al. (2001).
(M = .137). However, this was also true when the bipolar opposite (inapplicable) concepts were primed (.315 vs. .152, respectively).5
Finally, these interpretations influenced participants overall evaluations of the target. That is, participants evaluated the target more favorably when they had been primed with a trait pertaining to hostility (M =
4.55) than when they had been primed with a trait pertaining to kindness (M
= 3.76), and this was true regardless of whether the primed trait was directly applicable for interpreting the behavior (4.08 vs. 3.58 when honest
and unkind were primed, respectively) or when the bipolar opposites
were primed (5.02 vs. 3.94, when dishonest and kind were primed, respectively).
The behavioral information that participants received in the preceding
study pertained to two different dimensions, but its implications along each
dimension were very clear. An additional experiment (Park et al., 2001, Experiment 2) showed that when the implications of information along a dimension are ambiguous, activating concepts can influence not only the dimension along which the information is evaluated but also the value it is
assigned along this dimension. In this study, participants first performed a
task that required the use of concepts associated with either good health,
bad health, good taste, or bad taste. Later, they were given the description of a milk product as having more artificial sweeteners. Although
this description clearly implied that the product was bad for the health, its
implications for tastiness were ambiguous. That is, it could imply that the
5
This latter effect appears more pronounced when dishonest was primed than when kind
was primed. However, this difference results from a more general tendency to judge the target
as honest rather than as unkind, which inflates the effect of priming dishonest but decreases
the effect of priming kind. Pooling over the two trait replications, the overall effect of priming
was similar in magnitude regardless of whether the primed concept was applicable for describing the target or its bipolar opposite.
60
CHAPTER 3
product was either good tasting (sweet) or bad tasting (too sweet). Participants who had been exposed to health-related concepts described the
product as unhealthy regardless of whether good health or bad health
was primed. In contrast, participants who were exposed to taste-related
concepts described it as having either good taste or bad taste, depending
on which taste-related concept had been activated. Thus, primed concepts
influenced not only the dimension along which the information was interpreted but also the value it was assigned along this dimension.
Inference Processes: Belief Formation and Change
People who are asked to report their belief that a statement concerning it is
true are unlikely to search memory for all of the belief-related knowledge
they have acquired. Rather, they are apt to rely on the first belief-relevant
knowledge that comes to mind (Postulate 3.1). That is, they search memory
for a possible reason why the statement might or might not be true and, if
they find such a reason, use its implications as a basis for their judgment
without further ado. To this extent, the reason they identify and, therefore,
the belief they report, may depend on the recency with which the reason
had been thought about. We discuss this possibility in more detail in chapter 10 of this volume. However, two examples are worth mentioning in the
present context.
Participants in a study by Henninger and Wyer (1976) were asked to report their beliefs in a series of propositions. Some propositions were syllogistically related. That is, one, informational proposition, A (e.g., The
army is recruiting people of below average intelligence), had implications
for the validity of a second, target proposition, B (e.g., The quality of the
peace-time army is deteriorating). In some cases, beliefs in the informational proposition were reported early in the questionnaire, and the target
proposition was not considered until later. In other conditions, the target
proposition was considered at the outset. Participants use of their belief in
the informational proposition as a basis for inferring the validity of the target was inferred on the basis of a quantitative model of syllogistic inference
proposed by Wyer and Goldberg (1970; see also McGuire, 1960, 1981).6 Par6
If beliefs are defined in units of probability (along a scale from 0 to 1), the model asserts that
P(B) = P(A)*P(B/A) + P(~A)*P(B/~A)
Where, P(B) is the belief that Proposition B is true, P(A) and P(~A) are the beliefs that A is and is
not true, and P(B/A) and P(B/~A) are beliefs that B is true if A is and is not true, respectively.
This model provides an accurate quantitative description of the relations among these beliefs,
and also the effect of changing beliefs in A on beliefs in B (Wyer, 1970, 1974; for summaries of evidence, see Wyer, 1974; Wyer & Hartwick, 1980). The implications of this model are discussed
more fully in chapter 10.
61
ticipants based their beliefs in the target on their beliefs in the informational proposition when they had evaluated this proposition earlier in the
questionnaire. When they evaluated the target proposition at the outset,
however, they typically identified a different subset of previously acquired
knowledge to use as a basis for reporting their belief, and so this belief was
inconsistent with their belief in the informational proposition they considered subsequently.
Interestingly, when participants reported their beliefs again a week later,
their beliefs in the target were consistent with their beliefs in the informational proposition in both order conditions. Apparently, the informational
proposition, having been reported in Session 1, was accessible to both
groups of participants in Session 2, and so both groups of participants used
it as a basis for inferring the validity of the target.
This interpretation obviously assumes that no other relevant information bearing on the validity of the target was received or thought about during the 1-week interval between sessions. The target propositions pertained
issues that participants were unlikely to think about spontaneously, and so
this assumption seems reasonable. In an unpublished study by Wyer and
Hartwick (see Wyer & Srull, 1989), however, some participants were asked
to report their beliefs that drinking coffee was desirable after having been
exposed to a proposition that either implied that this was true (Coffee
keeps you alert) or implied that it was false (Caffeine destroys nerve
cells). Others were asked to report their beliefs that student use of the university health center would increase after exposure to a proposition suggesting that the event was likely to occur (Dental services are scheduled to
be provided at the health center) or unlikely (Some doctors at the health
center are about to lose their licenses). In both cases, participants based
their beliefs in the target proposition on the implications of the particular
information proposition to which they had been exposed. However,
whereas the effects of informational propositions on participants beliefs
about the use of the health center were still evident in a second experimental session 1 week later, their effects on beliefs in the desirability of drinking
coffee were not. The reason for this difference is rather obvious. That is,
participants in the 1-week interval between experimental sessions were
likely to have numerous coffee-drinking experiences with implications for
the desirability of this activity, and these more salient implications overrode the effects of the knowledge they had selectively activated and used in
the first experimental session. In contrast, participants had few if any experiences with implications for the use of the health center during the 1-week
interim, and so the knowledge they had activated in the first session continued to have an impact.
The studies described in this section demonstrate the effects of activating prior knowledge about an issue on beliefs and opinions. However, this is
62
CHAPTER 3
not the only type of knowledge that people bring to bear on these judgments. An equally important source of information is provided by ones
own past behavior toward the persons and objects to which the issue pertains. Bem (1972), for example, noted that people often use their past behavior (e.g., a decision to advocate a position publicly) to infer their belief
that the position is valid (see also Albarracin & Wyer, 2000). These effects
are likely to be mediated by an implicit theory that people have acquired
about the causal relatedness of their beliefs and their behavior under the
particular situational conditions at hand. We explore this possibility more
fully in chapter 10.
Response Processes: The Use of Activated Concepts
as Standards of Comparison
The effects of accessible concepts on the interpretation of information typically occur at the time the information is first received. At the time a judgment is made, the same concepts can have other effects. In particular, judgments are typically made with reference to a standard of comparison.
Moreover, these standards may be applied spontaneously, without awareness. Thus, we may judge a baby to be big and a house to be small without feeling any inconsistency in these judgments at all.
The effects of knowledge on standards of comparison can sometimes offset or override the effects that occur at the comprehension stage. A study
by Herr (1986) is of interest in this context. Participants were initially exposed to the names of persons who were known to be either moderately
hostile (e.g., Muhammad Ali or Howard Stern) or extremely hostile (e.g.,
Adolf Hitler or Genghis Khan). Later, they were asked to form impressions
of a person whose behaviors were ambiguous with respect to this trait. It
seems reasonable to suppose that the priming stimuli employed in this
study had two effects. First, they activated trait concepts that were associated with the individuals being primed, leading participants to interpret the
targets behaviors in terms of these concepts. At the same time, they activated concepts of the individuals themselves that participants later used as
standards of comparison in evaluating the target. These effects could potentially offset one another.
Herrs results were consistent with this possibility. Participants who
were exposed to moderate exemplars of hostility judged the target as more
hostile than participants in a control condition, confirming the findings obtained by Srull and Wyer (1979) and by Bargh and Pietromonaco (1982).
However, participants who were exposed to extreme exemplars judged the
target to be less hostile than control participants did. The exemplars to
which participants were exposed may have been used as standards of comparison in both conditions. When the exemplars were only moderately ex-
63
treme, however, the effect of using them as standards was not sufficient to
override the effect of the activated trait concepts on interpretation of the
information at the time it was received. When the exemplars provided a
more extreme standard of comparison, the effect of using them in this capacity predominated, producing a contrast effect. In this regard, standards
of comparison for evaluating a person are generally less likely to be activated by trait concepts per se than by the exposure to persons who possess the traits. Thus, as Stapel, Koomen, and van der Pligt (1997) found (see
also Moskowitz & Skurnik, 1999), contrast effects of priming are more likely
to be evident in the latter case than the former.
The activation and use of stimuli as standards of comparison can occur
without awareness. This was demonstrated in a series of studies by Adaval
and Monroe (2002). In some of their studies, participants were exposed subliminally to either high or low numbers before being asked to judge a target
product whose price fell between the two sets of priming stimuli. Participants judgments of the target product were lower when high numbers had
been subliminally primed than when low numbers had been primed. That
is, the target was judged as less expensive, but also less desirable, in the
former condition than in the latter. Thus, priming stimuli appeared to induce a disposition to use either a high or low value as a standard of comparison regardless of the dimension along which judgments were made.
THE EFFECTS OF PRIOR JUDGMENTS

ON SUBSEQUENT ONES
In most research on the effects of knowledge accessibility, the concepts
and knowledge that participants bring to bear on their judgments have
been activated by situational and individual differences factors that are ostensibly irrelevant to the judgment or decision they are called upon to
make. However, concepts and knowledge are often activated in the course
of cognitive activity that is directly relevant to the persons or objects being
judged. For example, people in the course of judging an object may retrieve
concepts and knowledge that are specifically relevant to this judgment. If
these cognitions remain accessible at the time a second judgment or decision is made, they may influence the latter response as well.
Studies described earlier can be interpreted in this light. For example,
participants in an early study by Carlston (1980) were given behaviors that
exemplified by kindness and honesty (e.g., told his girlfriend that her new
hair style was ugly), and were asked to judge the person with respect to
one of these traits. In the course of making this judgment, participants presumably formed a general impression of the person, based on their prior
knowledge about the sort of individual who typically possesses the trait
64
CHAPTER 3
they are asked to judge. Once this impression was formed, it may have influenced their later judgments independently of the behavioral information
they had received earlier. Thus, in our example, participants who were
asked to judge the targets honesty judged him as honest. Having done so,
they later judged the target as more kind than they would have done if they
had not made the initial judgment.
A quite different series of studies by Thomas Mussweiler and Fritz Strack
(for a review, see Mussweiler, 2003) provide a further example of these effects and confirm assumptions underlying their interpretation. In a typical
study (Mussweiler & Strack, 2000b), participants were first asked to indicate
whether Mt. Everest was greater or less than a specific anchor value (e.g.,
Is Mt. Everest higher or lower than
feet?). The anchor value was either high or low. In all cases, these values were sufficiently extreme that
participants would normally consider the target to fall in between them. In
some cases, however, the values were fairly plausible (e.g., 45,000 ft. vs.
2,000 ft.), whereas in other cases, the values were totally implausible (e.g.,
300,000 ft. vs. 65 ft.). Participants after making the comparative judgment
were asked to estimate the actual height of Mt. Everest. Participants typically judged the actual height of Mt. Everest to be greater when they had
compared it to a high value than when they had compared it to a low one.
Moreover, this tendency was even greater when the anchor values were implausible (115,728 ft. vs. 9,271 ft., respectively) than when they were plausible (36,106 ft. vs. 27,783 ft., respectively).
On the surface, these results appear to resemble the anchoring-andadjustment phenomenon identified by Tversky and Kahneman (1974). That
is, people in judging a stimulus arbitrarily identify a high or low anchor
value along the judgment dimension and then adjust upward or downward
in relation to this standard. However, they do not adjust enough, with the
result that their judgment is displaced toward the anchor they have arbitrarily used (for examples of this effect in social judgment, see Schwarz &
Wyer, 1985). Mussweiler and Strack (1999a) raise an alternative possibility,
however. That is, people who estimate a stimulus in relation to a standard
may activate a body of general knowledge that is consistent with it. Although they may ultimately reject the anchor value as implausible, the subset of knowledge and its implications, having been made accessible, is then
used as a basis for their own estimates.
To distinguish between the two alternative interpretations, Mussweiler
and Strack (1999a) asked participants after making their comparative judgments to perform a lexical decision task requiring them to identify words
that were semantically associated with either high anchor values, low anchor values, or irrelevant. Participants were quicker to identify words that
were associated with the anchor to which they had been exposed than the
opposite anchor, confirming the assumption that exposure to the anchor in-
65
creased the accessibility of anchor-consistent knowledge in memory. The

anchoring-and-adjustment processes assumed by Tversky and Kahneman
(1974) would not predict this difference.
The phenomena identified by Mussweiler and Strack generalize over a
variety of content domains (Mussweiler & Strack, 1999a, 1999b, 2000b). A
particularly intriguing implication of their conceptualization surrounds the
effects of comparing oneself to a standard on self-judgments. In one study
(Mussweiler & Strack, 2000b), participants were first asked whether the
number of simple mathematical computations (e.g., 2 + 5 = ?) they could
perform in 1 minute was greater or less than either a high value (62) or a
low one (10). After doing so, they estimated the actual number they could
do and also rated their intelligence along a scale from 1 to 7. Consistent
with results of earlier studies, participants estimated that they could do
more problems when they had been exposed to the high anchor than when
they had been exposed to the low one. On the other hand, they rated their
general intelligence to be lower in the first condition.
This combination of results makes salient two different effects that standards of comparison can have. On one hand, a standard can activate a selective subset of general knowledge that is consistent with it, thus biasing
the criteria that people bring to bear on their numerical estimates. At the
same time, the standard can influence the range of subjective values with
which these numerical estimates are associated, as suggested by Parducci
(1965) and Upshaw (1969), leading persons to assign lower values to these
estimates when the anchor is high than when it is low (for an analysis of
these effects and the processes underlying them, see Ostrom & Upshaw,
1968; Wyer & Srull, 1989).
DETERMINANTS AND CONSEQUENCES

OF THE ACCESSIBILITY OF GOALS AND MOTIVES
The preceding discussion focused on the accessibility of concepts and
knowledge that are used to interpret and make inferences about an external referent. As noted in chapters 1 and 2, however, goals and motives are
also components of the cognitive system. As such, the activation of these
motives and their influence on judgments and decisions should be subject
to similar processes. People are obviously motivated by a wide variety of
general and specific goals, and the behaviors that are required to attain
these goals in any given situation may not be compatible. In this case, peoples behavioral decisions may depend in part on which goal happens to
predominate at the time the decisions are made.
In some cases, the relative accessibility of a particular goal or motive is
likely to be activated by features of the stimuli to which the goal is relevant.
66
CHAPTER 3
Shiv and Fedorikhin (1999), for example, found that people were inclined to
choose to eat chocolate cake rather than fruit salad when they did not have
the opportunity to think about their decision, whereas the reverse was true
when they were able to deliberate. Presumably, the visual stimulus features
of the cake led hedonic goals to be activated very quickly, whereas the
competing goal of good health took more time to be activated.
In many instances, however, people may acquire more general motivational orientations that, once activated, can influence their judgments and
behavior in a variety of specific situations. Such general motivational dispositions are often activated by features of a stimulus situation to which they
are relevant. However, their accessibility can also be influenced by extraneous factors that are not directly relevant to the situation at hand. Two quite
different bodies of literature provide evidence of this possibility. One concerns the desire to maintain beliefs in a just world (Lerner, Miller, &
Holmes, 1976). A second surrounds the relative emphasis placed on positive versus negative consequences in making a behavioral decision (cf. Higgins, 1997).
The Motivation to Believe in a Just World:
Implications for Reactions to Rape
Motivational influences on attitudes and opinions, which were called to the
attention of social psychologists by Festinger (1957), are well established.
The nature of these effects obviously depends on the particular motive involved (e.g., see Arkin, Gleason, & Johnson, 1976; Baumeister & Newman,
1994; Giner-Sorolla & Chaiken, 1987; Psyzczynski & Greenberg, 1987; Zanna
& Cooper, 1976). One motivational determinant of judgments, proposed by
Lerner et al. (1976; Lerner & Miller, 1978), is the desire to believe that the
world is just. Lerner et al. assume that people have a need to maintain beliefs in a just world in order to reassure themselves that they (who are
righteous) will personally not fall victim to misfortune for reasons beyond
their control. One manifestation of this motivational disposition is that people are not only disposed to believe that individuals who do something
wrong will be punished but also, that people who encounter misfortune are
responsible for their fate. In other words, individuals not only get what they
deserve but deserve what they get.
Lerner et al. (1976) provided several examples of this tendency. Walster
(1966), for example, found that people judged victims of an automobile accident more negatively if the victims had been seriously injured than if they
had not. Similarly, Lerner and Simmons (1966) found that observers of a
learning experiment in which a confederate was ostensibly shocked for
making errors disparaged the learner more when the shocks were severe
than when they were mild.
67
Beliefs in a just world are not the only possible basis for judgments and
decisions, of course. To this extent, the use of this criterion should be increased by situational factors that increase the accessibility of these beliefs
and the motive to maintain them. This possibility was demonstrated by
Wyer, Bodenhausen, and Gorman (1985) in a study of the cognitive mediators of reactions to rape. We were interested in whether activating aggression-related concepts in one context might influence peoples reactions to
rape situations that they encountered in a different, ostensibly unrelated
context. To do so, we asked participants to engage in two ostensibly unrelated experiments. The first study was ostensibly concerned with the things
shown in the media that college students find objectionable. On this pretense, we exposed participants to slides of 12 pictures. Nine of these pictures showed objects and events that participants were unlikely to consider offensive. The other three pictures varied. In one condition, these
slides showed aggressive acts of the sort that occurred frequently (e.g., police subduing a criminal, a boxing match, etc.) and, therefore, activated concepts that aggression is normal and socially sanctioned. In a second condition, the pictures portrayed severely negative outcomes of aggression that
activated the concept that human beings were cruel and inhumane (e.g., a
lynching episode, a dead soldier with a hole in his head, etc.). The third,
control set portrayed stimuli that might be considered unpleasant but
were unrelated to aggression (e.g., deformed babies, a smoking advertisement, etc.).
Participants rated each of the 12 pictures in terms of how objectionable
it was. Then, they were told that the experiment (which took about 10 min)
was over, but that because there was time remaining, we would like them to
help out another faculty member who was conducting a study in a different
room down the hall. The new study was introduced as an investigation of
the factors that people consider important in judging criminal cases. On
this pretense, participants were asked to read descriptions of four rape
cases, and in each case, to report several reactions. Two questions concerned (a) whether the defendant should be convicted and (b) whether he
actually was convicted. Three others concerned the victims responsibility
for the incident (e.g., whether she provoked the rape, whether she could
have avoided it, etc.).
We assumed that the pictures presented in the first experiment would
activate concepts that participants would use to construe the implications
of the rape scenarios they encountered in the second experiment. On a priori grounds, several hypotheses seemed plausible. First, if pictures of socially sanctioned acts of aggression activated concepts that aggression is
normal, they might decrease beliefs that the defendant should be convicted. On the other hand, if pictures of extremely negative consequences
of aggression activate concepts that people are cruel and inhumane, they
68
CHAPTER 3
might increase beliefs that the defendant should be convicted. Moreover,

they might increase sympathy for the victim, decreasing beliefs that she
was responsible for the incident.
An alternative possibility, however, is suggested by the assumption that
exposure to instances of aggression threaten peoples beliefs that the world
is just and, therefore, activate a motive to reaffirm this belief. This motive,
once activated, may influence participants reactions to the rape incidents
they encounter subsequently. If this is so, exposure to severe consequences of aggression should increase beliefs that the defendant was punished (i.e., that he got what he deserved). However, it should also increase
beliefs that the victim was responsible for the incident (that she deserved
what she got).
Results summarized in Table 3.2 confirm the latter hypothesis. Exposure
to acts of aggression had little influence on judgments of either the defendant or the victim. (If anything, they increased beliefs that the defendant
should be convicted, contrary to the assumption that these beliefs would
be mediated by perceptions that aggression was normal and socially sanctioned.) Moreover, exposure to severe consequences of aggression also
had no effect on participants beliefs that the defendant should be convicted. However, exposure to these stimuli increased beliefs that the defendant was convicted and, at the same time, increased beliefs that the victim
was responsible for the incident. Thus, these results are consistent with the
hypothesis that exposure to extreme consequences of aggression activated
concepts that threatened participants perceptions that the world is just.
Therefore, it motivated them to reaffirm these perceptions by believing not
only that the defendant got what he deserved, but also that the victim deserved what she got.
These results could have implications for the impact of the media on
perceptions of rape and reactions to aggression more generally. The effects
TABLE 3.2
Judgments of the Defendant and Victim as a Function
of Concepts Activated by Priming Stimuli
Priming Stimuli
Belief that defendant should be convicted

Belief that defendant was convicted
Belief that victim was responsible
Control
Socially
Sanctioned
Aggressive Acts
Severe
Outcomes of
Aggression
8.97
3.95
2.97
9.62
3.47
3.07
8.70
5.10
4.20
Note. Judgments are reported along a scale from 0 (not at all likely) to 10 (very likely). Based
on data from Wyer, Bodenhausen, and Gorman (1985).
69
of situationally primed concepts on judgments are often of short duration

(but see Srull & Wyer, 1979, for evidence of priming effects over a period of
24 hours). However, the effects of frequent exposure to stimuli on concept
accessibility are much more enduring (Higgins, 1996; Higgins, Bargh, &
Lombardi, 1985). It therefore seems reasonable to assume that frequent exposure to extreme violence could induce a chronic tendency to maintain
beliefs in a just world that is manifested in a variety of contexts. To this extent, it could have effects similar to those that Wyer, Bodenhausen, and
Gorman (1985) observed.
The Motivation to Avoid Negative Outcomes
People who are confronted with a decision are likely to base their judgments on their perception of its possible consequences. In doing so, they
might potentially consider both the positive consequences that might result from their decision and the negative consequences it could have. However, both individual and situational differences may exist in the relative
emphasis that people place on these consequences.
This possibility is recognized by Higgins (1997) conceptualization of regulatory focus. Higgins postulates that individuals can have either a promotion focus (which is manifested by a concern with positive consequences of
ones behavior) or a prevention focus (characterized by a concern with
avoiding negative decision outcomes). These motivational orientations may
reflect general dispositions that, once activated, generalize over situations.
Although chronic individual differences can exist in these orientations,
their relative influence can also be determined by situational factors that influence their relative accessibility.
1. Chronic Differences in Promotion
Versus Prevention Focus
Chronic differences in the motivation to attain positive outcomes or
avoid negative ones are suggested in a study by Briley, Morris, and
Simonson (2000). At the same time, these motives may not be activated
spontaneously. Briley et al. speculated that European Americans were disposed to be promotion focused, whereas Asians were more inclined to be
prevention focused. To examine this possibility, representatives of both
cultural groups were asked to choose between (a) an alternative with a
very favorable value along one dimension (A) and a very unfavorable value
along a second dimension (B), (b) an alternative with a very unfavorable
value along A and a very favorable value along B, and (c) an alternative
with moderate values along both dimensions. Thus, the first two alternatives had both more favorable and more unfavorable features than the
third, compromise alternative. Briley et al. reasoned that participants
70
CHAPTER 3
who were primarily motivated by the desire to attain positive outcomes

would prefer either of the first two alternatives to the third. However, individuals who were motivated to avoid negative outcomes would prefer the
compromise alternative. They found that when participants were asked to
make decisions without having to justify their choices, cultural differences
in these choices did not emerge. However, asking European Americans to
give a reason for their decision increased their preferences for the alternatives with extremely favorable values, whereas asking Asians to give a reason increased their preferences for the compromise alternative. Thus,
chronic cultural differences in promotion and prevention focus appear to
exist. However, these differences are only evident if participants are stimulated to think about the basis for their decisions.
2. Situational Influences on Promotion
Versus Prevention Focus
Feelings of Membership in an Ad Hoc Group. Situational factors that

induce concerns about avoidance of negative consequences can override
chronic differences in this tendency. A series of studies conducted by
Donnel Briley (Briley & Wyer, 2002) investigated this possibility and, in doing so, confirmed the generalizeability of the motivational orientations that
Higgins postulates. Aaker and Lee (2001) found evidence that stimulating
people to think of themselves as part of a group increases their focus to
negative features of a situation relative to positive ones. Based on these
findings, we assumed that stimulating people to feel part of a group would
induce a prevention focus, the effects of which would generalize over a variety of decision situations.
In two studies, participants feelings of group membership were induced
by having them ostensibly participate in either a group or an individual
achievement task. After doing so, participants in Experiment 1 reported
their agreement with implications of a number of proverbs, some of which
emphasized the desirability of balance and equality (e.g., it takes two to
make a quarrel, when the shoulder pole is not secure at both ends, the
load will slip off, the pole is easy to carry if the load is balanced, etc.).
(For evidence that responses in this task are correlated with preferences
for equality in actual decision situations, see Briley et al., 2000; Weber, Hsee,
& Sokolowska, 1998.) We expected that participants with a prevention focus
would prefer situations that minimized the likelihood that one individual
would receive negative outcomes at others expense, and, therefore, would
be more likely to endorse proverbs that advocated outcome equality than
would participants without this orientation. This was in fact the case, as
shown in the first column of Table 3.3 (top); people reported stronger
71

TABLE 3.3
Mean Evaluations of Equality-Oriented Proverbs, and Proportion
of Participants Who Chose Candies of Different Types, as a Function
of Participating as Individuals or in a Group and Regulatory Focus
Endorsement of equality-oriented
proverbs
Individual participation
Group participation
M
Proportion of participants choosing candy of different types
Individual participation
Group participation
M
No Instructions
(Experiments
1 and 2)
Promotion
Focus
(Experiment 3)
Prevention
Focus
(Experiment 3)
5.07
5.60
5.34
5.08
5.07
5.07
5.49
5.60
5.55
.45
.73
.61
.50
.55
.43
.59
.81
.71
Note. Based on data from Briley and Wyer (2002).
agreement with these proverbs when they had participated as part of a

group than when they had taken part as individuals.
In Experiment 2, participants after performing the group task were dismissed. Upon leaving the experiment, however, they were given the opportunity to take two candies as an additional reward for participating. Two
types of attractive candy were available, and so participants could choose
either two candies of the same type or one candy of each type. We expected that individuals with a desire to avoid negative consequences of
their choice would be inclined to choose one candy of each type, thereby
minimizing the risk of experiencing postdecisional regret if their momentary preference for the two types of candy happened to change. Consistent
with this reasoning, participants were more likely to choose candies of different types after participating in the experiment as members of a group
(.73) than after participating as individuals (.45).
To confirm the assumption that the effects of group membership on behavior in these studies were mediated by its impact on prevention focus,
we conduced an additional experiment in which this focus was manipulated
more directly (Briley & Wyer, 2002, Experiment 3). This study was virtually
identical to the first two experiments except that participants were told either that (a) the individual/group who performed best on the achievement
test would receive a monetary reward (promotion-focus conditions) or, alternatively, (b) individuals/groups would receive a reward except for those
who did poorer than average (prevention-focus conditions). Proverb endorsement and candy choices under these two conditions are shown in the
last two columns of Table 3.3. Participants were more inclined to endorse
72
CHAPTER 3
equality-related proverbs, and also more likely to choose candies of different types, under prevention-focus conditions than under promotion-focus
conditions.
Note also that under promotion-focus conditions, proverb endorsement
and candy choice were each very similar to that observed under individualparticipation conditions of Experiments 1 and 2, whereas under preventionfocus conditions, they were similar in magnitude to that observed under
group-participation conditions of the first two experiments. Thus, these
data provide strong confirmation of the assumption that inducing feelings
of group membership activated a prevention focus, and therefore had the
same impact as a more direct manipulation of this motivational orientation.
Feelings of Cultural Identity. A second series of studies (Briley & Wyer,
2002, Experiments 5 and 6) confirmed these conclusions using a quite different means of inducing feelings of group membership, and also different indices of prevention focus. Specifically, participants were both European
American and Hong Kong Chinese college students whose feelings of group
membership were activated by making them aware of their cultural identity. This was done by exposing representatives of each culture to symbols
of their own or a different culture as part of an ostensibly unrelated test of
general knowledge. (Specifically, European Americans were exposed to pictures of Marilyn Monroe, a Dixieland jazz band, the American flag, etc. Chinese were exposed to pictures of the Great Wall, a Chinese musical instrument, people doing calligraphy, etc.).
Following this task, some participants were asked to state their preferences for different combinations of outcomes to themselves and another
in a hypothetical resource allocation task. Others were administered the
product choice task employed by Briley et al. (2000) and described earlier.
Exposing participants to symbols of their own culture not only increased
their preferences for equality in the resource allocation task but also increased their likelihood of choosing the compromise alternative in the
product choice task, and this was true of both European Americans and
Hong Kong Chinese. Thus, making participants aware of their cultural
identity activated a disposition to avoid negative outcomes that overrode
the chronic cultural differences in prevention focus that Briley et al. (2000)
observed.
THE ACTIVATION OF BEHAVIORAL DISPOSITIONS

Behavior that results from promotion and prevention focus is likely to be
mediated by conscious decisions to seek positive outcomes or avoid negative ones. In some cases, however, behavior may be mediated by produc-
73
tions of the sort postulated by Anderson (1983) and described in chapter 1,

which are activated and applied spontaneously, with a minimum of conscious deliberation. These productions, like declarative knowledge representations, could depend on their accessibility in memory at the time.
A series of studies by Bargh and his colleagues are particularly intriguing. In one study (Bargh, Chen, & Burrows, 1996, Experiment 1), participants
completed a sentence construction task that required the use of concepts
associated with either politeness or rudeness. Upon completing the task in
the experimenters absence, they went to another room to turn in their
work. Upon arriving, however, they found the experimenter engaged in a
discussion with a graduate student. The time that participants waited
before they interrupted the conversation was recorded. Participants interrupted more quickly if they had previously been primed with rudenessrelated concepts than if they had been primed with politeness-related concepts. On the other hand, participants post-experiment judgments of the
experimenters own rudeness was not influenced by the priming manipulation. Thus, the primed concepts appeared to influence participants behavior independently of their interpretation of the experimenters actions.
Effects of Stereotype Activation on Own Behavior
Two additional studies by Bargh et al. (1996) demonstrated that the activation of stereotypes can induce stereotype-consistent behavior, and that this
is true even when individuals are not themselves members of the stereotyped group. In a particularly provocative study (Bargh et al., 1996, Experiment 2), participants completed a similar sentence construction that in this
case required the use of concepts associated with the elderly (aged,
bingo, etc.). After completing the task, participants were dismissed from
the experiment, and the time they took to walk to the elevator was recorded. Participants who were primed with elderly related concepts walked
more slowly to the elevator than control participants. This was true even
though the participants themselves were not elderly but rather were college students.
Participants in a third study (Bargh et al., 1996, Experiment 3) were subliminally exposed to pictures of either African-American or EuropeanAmerican faces while they performed a tedious perceptual task on the
computer. Upon completing the task, however, they were told that the
computer had malfunctioned and they would have to perform the task a
second time. Participants nonverbal behavior (facial expressions, etc.)
were recorded and coded for manifestations of irritation and hostility.
Participants manifested significantly more hostility when they had been
exposed to African-American faces than when they had been exposed to
European-Americans.
74
CHAPTER 3
Qualifications and Extensions

Bargh et al.s studies combine to suggest that unobtrusively exposing participants to stimuli can activate sequences of behavior under conditions in
which they are minimally aware of either the priming stimuli or the behavior itself. Moreover, these effects are not mediated by their effects on priming on the interpretation of the situation or the person toward whom the
behavior is directed. To this extent, they are consistent with the assumption that features of the priming stimuli activated a production that elicited
a behavioral sequence automatically, with minimum awareness of the conditions that gave rise to it. It should be noted, however, that the priming
stimuli may not by themselves be sufficient to activate the production. That
is, the eliciting conditions of a production are likely to include features of
the situation as well. It seems unlikely that people who are primed with
rudeness-related concepts in Bargh et al.s first experiment would have
been rude if the experimenters own behavior had not provoked it. Nor
would the exposure to African-American faces activate hostile behavior in
the absence of situational conditions in which the behavior was potentially
appropriate. Rather, both primed features and situational features in combination may be necessary to form the eliciting conditions for a production.
In fact, the same primed features might elicit quite different behavior, depending on the situational features that accompany them.
This possibility is made salient in a series of studies by Stanley
Colcombe (Colcombe & Wyer, 2001). In some conditions of this study, Caucasian participants were first subliminally primed with either African American, Asian, or Caucasian faces using procedures very similar to those employed by Bargh et al. (1996). Then, they completed a shortened version of
the mathematics section of the Graduate Record Examination. African
Americans are stereotyped as unmotivated to perform well in intellectual
activity, whereas Asians are stereotyped as highly motivated to achieve. To
this extent, priming African-American faces should decrease White participants attempts to perform well on the mathematics test, whereas priming
Asian faces should increase the effort they expend. Results shown in the
first column of Table 3.4 confirm this hypothesis. That is, participants attained lower scores on the test when African-American faces had been
primed, and higher scores when Asian faces had been primed, than they
achieved when White faces had been primed.
The fact that the effects of priming stimuli on participants behavior occurred without awareness was made salient by additional conditions in
which the priming stimuli were presented above awareness threshold. (To
justify this, participants were told that the perceptual task had been used in
another, unrelated experiment and that we had been unable to modify the
program to eliminate the stimuli from being shown.) As shown in the righthand column of Table 3.4, participants under these conditions performed
75

TABLE 3.4
Mathematics Test Performance as a Function
of Priming Stimulus and Prime Type
African American faces

Caucasian faces (control)
Asian faces
Subliminal Prime
Overt Prime
420
452
512
502
446
524
Note. Scores are converted to percentiles based on GRE norms. Adapted from Colcombe and
Wyer (2001).
better when they had been exposed to either African-American or Asian

faces than under control conditions. In these conditions, participants who
were exposed to African-American faces apparently worked hard in order to
distance themselves from this stereotypically low achieving group, whereas
those who were exposed to Asian faces were stimulated to work hard in order to emulate this stereotypically high achieving group. Thus, participants
performed better in both conditions, albeit for different reasons.
The effects observed under subliminal priming conditions require further consideration, however. It is not completely clear why activating a
stereotype would have a positive influence on the behavior of individuals
to whom the stereotype does not apply. A conceptualization by Prinz (1990)
suggests one possible explanation. He postulated that in order to comprehend another persons behavior, one must often imagine oneself performing it. This mental simulation could establish an association between a representation of others behavior and a representation of ones own. As a
result of this association, factors that activate concepts of anothers behavior might activate a representation of ones own behavior in similar situations, producing the effects that both Bargh et al. (1996) and Colcombe and
Wyer (2001) observed.
CONCLUDING REMARKS
As the research in this chapter testifies, cognitive responses to situations
are determined to a large extent by the subset of concepts and knowledge
that happen to be easily accessible in memory at the time the responses
are generated. The research we have described does not begin to cover the
numerous ways in which differences in knowledge accessibility can have an
impact. It nevertheless suffices to indicate the importance of taking these
differences into account in conceptualizing information processing of the
sort we discuss in the chapters that follow. This becomes immediately important in the next section, where we analyze the comprehension of social
information of the sort we acquire through direct experience.
P A R T
II
COMPREHENSION PROCESSES
C H A P T E R
4
The Comprehension and
Validation of Information
About Familiar People and Events:
The Role of Situation Models
The comprehension of an event can occur in two stages. The first stage is
spontaneous and automatic, and occurs whenever the event can be easily
understood in terms of concepts and knowledge that comes easily to mind
at the time. It is difficult to observe a boy pounding a nail into a piece of
wood, or to see a book lying on a table, without comprehending this experience. It is equally difficult to read descriptions of these events without comprehending them. For example, try not to understand The boy pounded a
nail into a piece of wood, or the book is on the table. For a native English
speaker, the comprehension of such statements is spontaneous and virtually uncontrollable.
In many cases, however, comprehension requires deliberative cognitive
activity. Just as we comprehend some verbal statements spontaneously, we
immediately recognize that others cannot be easily understood. For example, The nail pounded a tree into the boy is structurally identical to The
boy pounded a nail into a tree. To understand the first statement, however, one must consciously construct a cartoon-like mental image of an animated nail driving the tree into a screaming juvenile. This construction is
deliberative and requires cognitive effort.
Moreover, the comprehension of some events requires an understanding of other events that precede or follow them. This information is particularly necessary when the event is inconsistent with a preexisting conception of the things that typically occur in the situation at hand. For example,
a nuclear physics professor might have difficulty computing a 15% tip at a
restaurant. Or, a persons comment at a party might violate a conversational norm to be polite, informative, or truthful (Grice, 1975). These events
79
80
CHAPTER 4
are likely to stimulate conscious cognitive activity in an attempt to understand why they occurred.
However, this deliberative cognitive activity is only likely to occur when
the comprehension processes that are spontaneously activated by exposure to information are insufficient. Before discussing this activity and the
conditions that give rise to it, one must first understand the processes that
underlie the spontaneous comprehension of information. The present chapter focuses on these processes. We first describe the mental representations that are formed in the course of comprehending events of the sort we
often encounter or read about in daily life. Then, we discuss the processes
that underlie the construction of these representations. Finally, we consider the implications of our conceptualization for the spontaneous recognition of statements about known persons and events as either true or
false. These latter implications are of considerable importance in understanding responses to information outside the laboratory, as our discussion
in later chapters testifies.
THE CONTENT AND STRUCTURE

OF SITUATION MODELS
Most social experiences are dynamic. That is, they take place over a period
of time. Such an experience can be conceptualized as a sequence of states
of affairs along with events (e.g., behaviors) that transform one state into
another. When an experience is described verbally, however, the entire sequence of events that occur is unlikely to be specified. For example, our observation of a boy throwing a ball necessarily includes both the action itself
and features of the situation in which it takes place. It might also include
other events that precede or follow the ball-throwing event per se (e.g., the
ball sailing through the air and being caught by another person). In contrast, a verbal description of the event (e.g., The boy threw the ball) does
not specify its antecedents or consequences. Nor for that matter, does it indicate the nature of the ball, physical characteristics of the thrower, or the
situational context in which the event occurred. To infer these characteristics, therefore, we must relate the verbal description to a past experience
we have had whose features match those of the description but contain
others as well. These latter features may then be added to the mental representation we form.
The comprehension process just described essentially involves a mental
simulation of the event to which the information refers. We conceptualize
this mental simulation as a situation model. The situation model construct is
not new (Johnson-Laird, 1983, 1989). In cognitive psychology, it has been
used to conceptualize the processes that underlie language comprehension
THE ROLE OF SITUATION MODELS
81
(Johnson-Laird, 1983; Kintsch, 1998; Zwaan & Radvansky, 1998), spatial memory (Taylor & Tversky, 1992) and logical reasoning (Schaeken, JohnsonLaird, & dYdewalle, 1996). Other research has examined their role in both
comprehending observed events (e.g., a movie; Magliano, Dijkstra, &
Zwaan, 1996) and imagining experiences that have not yet occurred (cf.
Glenberg, 1970; Glenberg, Kruley, & Langston, 1994; Graesser, Singer, &
Trabasso, 1994). Perhaps surprisingly, however, their role in social information processing is not widely recognized.
The mental simulation of a situation typically requires assumptions
about the relations among the entities (persons and objects) involved in it.
These relations are most easily represented in a mental image. An image
is often coded visually, and thus can be likened to a picture.7 However, it
should not be equated with a photograph. For one thing, not all features
that are found in a photograph are likely to be mentioned. (Thus, the mental image that is formed in response to the statement the boy kicked the
ball might specify the nature of the ball and portray the boy in the act of
kicking it, but might not include a specification of the boys hair color or
what he was wearing.) On the other hand, a mental image, as we conceptualize it, may convey information in other modalities. Thus, a representation
of The man shouted, Get out of my way! may contain an acoustically
coded representation of the mans tone of voice as well as the thing he said.
Two types of situation models are of particular relevance in conceptualizing social experiences. Event models are simulations of specific actions
and, in isolation, are similar to states of affairs. However, a number of temporally and thematically related event models might be combined to form a
multiple-segment episode model that conveys the temporal and causal relatedness of the events as well as the events themselves.
As the preceding discussion implies, all event and episode models are
assumed to have an image component. This is true regardless of whether
the information that leads the model to be constructed is conveyed
nonverbally, orally, or in writing. When information is conveyed verbally,
the representation of it in memory is also likely to have a propositional
component that provides a linguistic description of the event or situation
being represented. To this extent, an event model might be analogous to a
picture plus caption, and an episode model to a mental comic strip.
Note, however, that although the verbal description of an event may spontaneously lead to the construction of a mental image in the course of comprehending it, an observation or a picture of the event can often be compre7
The representation of visual images in memory has been somewhat controversial (cf.
Kosslyn, 1980; Pylyshyn, 1973; Shepard & Metzler, 1971). However, we do not intend to enter into
this debate. The construct of a visual image as used in this volume, like mental representations
more generally, should be evaluated in terms of their utility as metaphors and not in terms of
their validity as depictions of the physiology of the brain.
82
CHAPTER 4
hended without recoding it linguistically. In this case, the recoding may not
normally be performed (see Postulate 3.1).
The preceding discussion can be summarized in two postulates:
Postulate 4.1. Event and episode models are mental simulations of an event,
or sequence of events, that are temporally and spatially constrained. These
models are constructed spontaneously in the course of comprehending
events that occur at a specifiable (although not necessarily specified) time
and place.
Postulate 4.2. An event model can consist of both a mental image and a linguistically coded description of the event or state of affairs to which it refers.
Although the image component of the model is obligatory, the linguistic component is optional.
A critical feature of our conceptualization is indicated in Postulate 4.1.

That is, not all of the representations that people form of social information
are situation models. Situation models are restricted to events or states of
affairs that are localized in space and time. This restriction is important. Although the time and place of an event is not always specified, it must be implicit. Thus, statements such as The book is on the table, or The boy
kicked the ball, refer to states and events that necessarily occurred at a
particular place and point in time. However, many verbal statements that
people encounter are not temporally or situationally constrained. For example, The book is interesting, or The boy owns a ball, are unrestricted
as to time and place. Therefore, the comprehension of these statements
would not involve the construction of a situation model. For example, they
might be represented linguistically, but a mental image of their referents
might not be formed.
Note that although most visual images are situationally and temporally
specific, this is not necessarily the case. For example, a mental picture of an
acquaintance could contain nonverbally coded features of the persons face
and other physical attributes but might be independent of any particular
situation in which the individual is involved. Moreover, a schematic representation of a prototypic situation could provide expectations for the type
of events that are likely to occur without its features being verbally articulated. These schematic representations can influence the spontaneous initial appraisals of a new situation that one encounters in a manner similar to
that postulated by Lazarus (1982, 1991), eliciting feelings that the situation
is benevolent or hostile without a clear understanding of why this is so.
These schematic representations, or situation prototypes, can also be
part of the eliciting conditions of a production of the sort noted in chapter
1, which automatically stimulate behavior under conditions in which they
are activated. Thus, people often behave quite differently in a classroom sit-
83
uation than at a party, and differently with friends than with casual acquaintances. These behavioral dispositions are activated and applied without
conscious analyses of the situational factors that give rise to them.
Headers
Episode models are often assigned a header, or title, that constitutes a generalization of the sequences of events to which the model refers and that
essentially tells what the model is about. The header of a model may be
coded propositionally. Thus, the fiasco at Lydias dinner party might be
the title of an episode model whose segments depict my stumbling into a table and spilling a bloody Mary on a guests new dress, causing her to drop a
dish of guacamole on the hostesss white shag rug. Features of a models
header can serve as retrieval cues for the model itself, activating the model
as a result of the processes described in chapter 2. In this regard, many
general statements that convey attitudes and opinions could be headers of
more specific episode models that exemplify them. For example, Religious
fanaticism is dangerous could be the header of an episode model of the
events surrounding the attack on the World Trade Center.
Not all propositions are model headers, of course. On the other hand, a
given proposition might be the header of more than one model. It is interesting to speculate that the strength of ones belief in a given proposition is
a positive function of the number of episode models for which it serves as a
header. Because episode models are constructed from specific experiences, this proposition would be consistent with Fazios (1990) proposal
that the strength of an attitude (or opinion) is typically greater when it is
derived from personal experience than when it is not.
In this and the following chapter, we describe the construction of event
and episode models and their role in the comprehension of social information. Before doing so, however, it will be useful to discuss more generally a
fundamental assumption that underlies the postulation of these models,
namely, that verbal event descriptions are often spontaneously encoded in
terms of mental images.
THE ROLE OF VISUAL IMAGERY IN

COMPREHENDING SOCIAL EVENTS
According to Postulates 4.1 and 4.2, verbal images are fundamental components of situation models and, therefore, are formed spontaneously in the
course of comprehending events that are temporally and situationally constrained. This is true even when the events are described verbally. On the
other hand, the linguistic coding of the events that compose a situation
84
CHAPTER 4
model is optional. Therefore, this coding is not always performed in the

course of comprehending events that are conveyed visually (e.g., in terms
of pictures).
The Construction of Images in Text Comprehension

Some of the earliest and best-known indications that mental images are
constructed in the course of comprehending text were provided by John
Bransford and his colleagues (cf. Bransford, Barclay, & Franks, 1972;
Bransford & Johnson, 1972). In an early study (Bransford et al., 1972), some
participants were asked to learn sentences that on the surface appeared
anomalous (e.g., The haystack was important because the cloth would
rip, or The notes were sour because the seam was split). Not surprisingly, their memory for these sentences was typically poor. In another condition, however, the same sentences were each preceded by a single word
(in our examples, parachute and bagpipes, respectively). In the context
of these words, memory for the sentences improved substantially. The
most plausible explanation of these results is that the cue word stimulated
the construction of a mental image of an event or sequence of events in
which the statements made sense (e.g., an image of a parachutist landing in
a pile of hay, or a frustrated piper trying to get music out of his defective instrument). Although these results could perhaps be interpreted without recourse to the assumption that mental images were formed of the events, an
explanation in terms of this assumption is the most parsimonious.
A study of recognition memory by Garnham (1981) provided circumstantial evidence of a different sort. In this study, participants after being asked
to learn a number of sentences were given a recognition memory test in
which the test sentences differed slightly from those they had seen earlier.
In some cases, the events described in the two sentences presumably occurred in the same situation; for example, participants were initially exposed to The hostess bought a mink coat from the furrier but were later
asked to verify The hostess bought a mink coat at the furriers. In other
cases, however, the events described by the two sentences occurred in different situations; for example, The hostess received a telegram from the
furrier and The hostess received a telegram at the furriers. In each case,
the learning and test sentences were identical in structure. However, the
events described in the first pair of sentences presumably occurred in the
same situation and, therefore, were likely to elicit similar mental images. In
contrast, the events described in the second pair occurred in different situations and were likely to elicit quite different mental images. Therefore, if
participants constructed these images spontaneously in the course of com-
85
prehending the sentences, they should be more likely to confuse the first
pair of sentences than to confuse the second pair, as reflected in their recognition responses. This was in fact the case.
A study by Glenberg, Meyer, and Lindem (1987) is particularly compelling. Participants read a story in which a sweatshirt either was associated
with the protagonist at the outset (John put on his sweatshirt and went
jogging) or was separated from him (e.g., John took off his sweatshirt and
went jogging). Their recognition of specific features mentioned in the passage (including sweatshirt) was then assessed. The time that elapsed between the mention of the feature in the story and its occurrence in the
recognition list was the same in all cases. Nevertheless, participants recognized the target object more quickly in the first condition than in the second. Participants in the first condition apparently formed a mental image of
John wearing the sweatshirt that persisted throughout the mental representation they formed of the events that occurred later, and this image was salient at the time of recognition. In the second condition, however, this was
not the case.
Finally, a study by Black, Turner, and Bower (1979) suggested that when
people form mental images on the basis of event descriptions, they construct these images from a specific visual perspective. Participants read
passages describing pairs of events that were either likely to be imagined
from the same perspective (e.g., Mary was reading a book in her room.
John came in to talk to her.) or from different perspectives (Mary was
reading a book in her room. John went in to talk to her.). Participants typically took longer to read and comprehend the second type of passage than
the first. In the first case, readers presumably formed an image of Mary
from someone in the room that was maintained while comprehending the
second. In the second case, however, the second sentence stimulated a
mental image from the perspective of someone outside the room, and so
readers were required to shift their perspective in order to comprehend it
in the context of the first sentence. As a consequence, reading and comprehension time increased.
Differences in the Comprehension of Picture
and Text Information
Using quite different methodologies, all four studies described earlier converge on the conclusion that mental images are constructed spontaneously
in the course of comprehending information about situation-specific events.
In all cases, however, the evidence is circumstantial. More direct support
for Postulate 4.2, requires a comparison of the comprehension of verbal descriptions of events with comprehension of the same events when they are
86
CHAPTER 4
conveyed in pictures. Specifically, suppose participants read a verbal description of an event. If this description spontaneously stimulates a mental
image of the event, the event should later be identified just as easily on the
basis of a picture as it would be if a picture of the event had been conveyed
in the first place.
Wyer, Adaval, and Colcombe (2002) examined this possibility. In doing
so, we also evaluated the assumption that although verbal descriptions of
an event are spontaneously encoded in terms of mental images, pictures of
events are not spontaneously encoded linguistically. We constructed two
scenarios of about 14 events each. One scenario, which concerned a day at
school, described a male student getting up in the morning, getting
dressed, eating breakfast, leaving his apartment, finding that his bicycle
had a flat, walking to school, getting to class late, leaving, meeting a friend,
going to the library, and falling asleep at the table. The second, a dating scenario, portrayed the student getting ready for a date, meeting the date, buying tickets for a movie, buying some popcorn, eating it while watching the
show, leaving, going to a pub, observing his date flirt with a friend, sulking
while walking home, but kissing his date goodnight at the door. Four sets of
stimuli were constructed for each scenario, two composed of pictures
and two composed of captions. The two picture scenarios were composed of photographs of the events, with one set taken from different perspectives than the other. The two caption scenarios were also constructed composed of verbal descriptions that were similar in meaning but
different in wording (e.g., discovered his bicycle tire was flat vs. saw that
his bike had a flat tire).
Participants were shown the events composing one of the two scenarios
on a computer screen with instructions that we were interested in how people comprehend everyday life events. This was done in three conditions. In
picture-only conditions, only pictures of the events were presented. In caption-only conditions, only verbal descriptions of the event were conveyed,
and in a third, picture-plus-caption condition, both pictures and verbal descriptions were shown.
After receiving the event descriptions and a short delay, participants
were presented stimuli portraying a number of events with instructions to
decide whether the event had occurred in the sequence they had seen
earlier. Some of the events were described verbally and others were described in pictures. Moreover, some stimuli referred to the same events
that occurred in the original scenario but were from a different perspective (in the case of photographs) or were worded differently (in the case
of verbal descriptions). Other stimuli, however, described different events
entirely. Participants were told that the stimuli they would see would not
be identical to those they had seen earlier, and to base their judgments on
whether the stimulus event was similar in meaning to the one they had
87
seen earlier. In each case, the time that participants took to judge the
stimuli was recorded.
1. Results
Participants under all conditions presumably formed a multiple-segment episode model on the basis of the events they encountered regardless of the modality in which they were conveyed. According to Postulate
4.2, however, participants who received written descriptions of the events
should spontaneously form mental images of them and, therefore, should
have these images in memory at the time of judgment. Therefore, they
should later be able to identify pictures of the events as easily as they
would if pictures of the events had been presented in the first place. Data
shown in the top half of Table 4.1 support this hypothesis. That is, when
participants were asked to verify an event on the basis of a picture, they
responded just as quickly if the event had been described verbally (M =
1.90 s) as they did if the event had been portrayed in a similar but not
identical picture (M = 1.97 s).
However, Postulate 4.2 also implies that people do not spontaneously
code pictures verbally when these codings are not necessary to comprehend the events they portray. If this is the case, however, participants who
are later asked to verify an event on the basis of a verbal description of it
must recode the picture linguistically at the time of judgment in order to
perform the verification. This recoding presumably takes time. Consequently, participants should take longer to verify the event than they would
if a verbal description had initially been presented. This was also the case,
as Table 4.1 indicates. That is, when participants were told to identify an
event on the basis of a verbal description of it, they took longer to do so
when the event had initially been conveyed in a picture (M = 2.53 s) than
when it had been described in words (M = 1.71 s).
TABLE 4.1
Mean Time to Verify Pictures and Captions as a Function of Presentation
Order and the Type of Information Initially Presented
Narrative order
Response time to
Response time to
Scrambled order
Response time to
Response time to
Picture
Verbal
Description
Picture Plus
Verbal
Description
pictures
verbal descriptions
1.97 (19)
2.53
1.90 (14)
1.71
2.05 (20)
1.96
pictures
verbal descriptions
2.16 (19)
2.26
2.50 (18)
2.23
1.91 (16)
2.09
Note. Adapted from Wyer et al. (2002).
88
CHAPTER 4
2. A Qualification
The results just described are quite consistent with Postulate 4.2. There
are nevertheless some qualifications on the validity of this postulate. In
other conditions of Wyer et al.s study, the events were presented in scrambled order rather than in the order they occurred. In this condition, participants could not spontaneously form a multiple-segment episode model of
the sequence as a whole. Rather, comprehension of this sequence required
deliberative cognitive activity. Results shown in the bottom half of Table 4.1
indicate that in this condition, participants could verify an event on the basis of a verbal description of it just as quickly when the event had originally
been conveyed in a picture (M = 2.26 s) as when the event had been described verbally in the first place (M = 2.23 s). However, they took longer to
verify an event on the basis of a picture when the event had been described
verbally (M = 2.50 s) than when it had been described pictorially (M = 2.16).
Although this pattern of response times was unexpected, it has a plausible
explanation. Verbal symbols are easier to manipulate mentally than visual
images are. Consequently, participants who received pictures of the events
in scrambled order may have subjectively translated them into linguistic
terms in order to construct a mental representation of the sequence of
events as a whole. As a result, they could later verify a verbal description of
the events just as easily as participants who had been given verbal descriptions of the events in the first place. When participants received verbal descriptions of the events in scrambled order, however, they relied on linguistic codings alone to construct a representation of the sequence without
including visual images in this representation. Therefore, when they were
later asked to verify a picture, they were required to recode it linguistically in
order to compare it with the linguistic representation they had constructed
earlier, and this took time. As a consequence, they took longer to make this
comparison than participants who had received a picture at the outset.
SPONTANEOUS COMPREHENSION PROCESSES

To reiterate, much of the knowledge we acquire in daily life consists of
event and episode models of specific events that we observe or read about.
This knowledge may often pertain to the activities of actual people and situations that we learn about over a period of time. To this extent, new information about the person and situations is likely to be comprehended in
terms of this referent-specific knowledge rather than in terms of more abstract concepts of persons and events in general. Thus, new information
about Bill Clintons affair with Monica Lewinsky is likely to be comprehended spontaneously in terms of prior knowledge about these specific
protagonists rather than more general concepts pertaining to U.S. Presi-
89
dents and staff associates, or to men and women more generally. As we

noted earlier, the conceptualization proposed by Wyer and Srull (1989)
could not account for this possibility.
Earlier in this chapter, we distinguished three different types of representations that might be constructed spontaneously in the course of comprehending new information. The representations differ in both the type of
input information that stimulates their construction and the terms in which
the representation is encoded. Specifically:
1. Episode and event models may be constructed from either verbal descriptions of situationally and temporally specific events or direct observations of these events. These representations consist of mental images and, in some cases, linguistically coded descriptions of the events
being depicted.
2. Nonspecific representations of persons and events can be constructed
from verbal descriptions that are not temporally or situationally specific (e.g., the boy owns a baseball, John went to college at Yale,
Peter is dishonest). These representations are coded linguistically
but do not contain mental images.
3. Situational appraisals are constructed of the context in which more
specific information is conveyed. These appraisals, and the representations formed of them, are nonverbal.
All three types of representations obviously come into play in social information processing. They are not only formed from new information but,
once constructed, can be used as bases for comprehending experiences
and information that are acquired subsequently. In this chapter, we focus
primarily on the construction of event models in the course of comprehending events involving familiar persons of the kind we are likely to observe or read about, and will defer our discussion of the construction of episode models to the chapters that follow. Much of our discussion, however,
is applicable to the construction of nonspecific person and event representations as well.
Functions of the Comprehender
The comprehension processes we propose are theoretically localized in the
Comprehender (see Fig. 2.1). Our conceptualization of this unit differs from
that of the corresponding unit in the Wyer and Srull (1989) model in terms
of its accessibility to knowledge in Permanent Storage and, therefore, its
ability to comprehend information about specific persons and events with
which one is already familiar. The Comprehender is assumed to have two
components: a Parser and a Simulator. These components and when they
come into play are shown schematically in Fig. 4.1.
90
CHAPTER 4
FIG. 4.1. Theoretical sequence of processing steps involving the Parser and
Simulator in the course of comprehending visual and verbal information.
1. The Parser
The Parser only comes into play when the input information is coded linguistically. In this case, the Parser transforms verbal information into the
equivalent of a SubjectPredicate proposition. The output of this Parser
then resonates with previously constructed mental representations in Permanent Storage. In this regard, a major difference between the conceptual-
91
ization of the Simulator and those of the Comprehender postulated by

Wyer and Srull lies in the range of material in Permanent Storage to which it
has access. In the original theory, the Comprehender had access to only semantic concepts that were stored in a semantic bin. In contrast, the Simulator has access to all previously formed knowledge representations that exist. Thus, information about a specific person or event about which one already
has knowledge is comprehended with reference to aspects of this referentspecific knowledge without requiring any interpolated cognitive activity. This
process is elaborated presently.
If the input information is verbal and a previously constructed representation exemplifies the subjectpredicate proposition that is formed of it, the
Parser retrieves this representation from Permanent Storage and uses it to
form a linguistically coded representation of the new information as well. If
the predicate describes an event or stimulus that is situationally and temporally constrained, however, it is also likely to resonate with a preexisting
nonverbally coded representation (e.g., an event or episode model). In this
case, the Simulator is activated.
2. The Simulator
The basic operations performed by the Simulator are theoretically similar regardless of whether the original stimulus information is conveyed visually (as in a picture or movie) or verbally. It has the ability to (a) identify
referential knowledge in Permanent Storage, (b) evaluate the compatibility
of new information with previously formed event and episode models, (c)
substitute features of the new information for those of preexisting models
to form a new model, and (d) combine individual segments extracted from
previously formed models to form a new episode model that depicts a
unique sequence of events.
The output of the Simulator is largely nonverbal (visual or acoustic). If
both a verbally coded representation has been formed by the Parser and a
nonverbally coded representation has been formed by the Simulator, these
representations are then integrated into a single representation, as indicated in Fig. 4.1.
The general processes outlined in Fig. 4.1 are potentially applicable regardless of the type of input information, and theoretically underlie the formation of all three types of representations described earlier. However, the
most complex comprehension processes occur when verbal information is
presented about situationally and temporally specific events. In the pages
that follow, therefore, we first discuss these processes and present some
empirical evidence bearing on them. In doing so, we note the implications
of the conceptualization for not only the comprehension of information but
also the spontaneous recognition of its validity or invalidity. Finally, we
92
CHAPTER 4
turn briefly to the processes that underlie the comprehension of events

that are directly observed. These latter considerations raise some additional issues that become important in later chapters of this volume.
THE COMPREHENSION OF VERBAL STATEMENTS

Theoretical Considerations
To reiterate, a verbal statement that enters the Comprehender is theoretically transformed by the Parser into the equivalent of a proposition consisting of a subject and a predicate. The subject can refer to either a specific
person (Michael Jordan), an object (Gone with the Wind), or a general
category (basketball players, book, etc.). The predicate can convey a
specific or general act (sank a three-pointer, played basketball), or can
specify a relation between the statements subject and an attribute (e.g., is
an athlete, has brown hair, is on the table). Although our examples typically refer to predicates that describe actions, they apply equally well to
predicates of other types.
Once the Parser identifies the subject and predicate, it uses the linguistic
codings of them as retrieval cues to access concepts in Permanent Storage
that they exemplify and forms a linguistic representation of the statement
in terms of these concepts. In addition, it transmits a specification of the
subject and predicate to the Simulator. The functions of the Simulator are
captured in four postulates:
Postulate 4.3. The subject and predicate identified by the Parser are independently used as probe cues of memory, resonating with all of the mental
representations that contain them (see Postulate 2.2). In each case, the representation whose resonance level reaches threshold most quickly is most
likely to be identified.
One implication of this postulate should be noted at the outset. That is,
the type of representations activated by the subject and predicate are likely
to differ. In particular, the subjects features are likely to activate a generalized person representations of the sort described in chapter 1 (Fig. 1.2).
That is, this representation may consist of a set of general features that are
independently associated with the individual to whom the subject refers.
Although some of these features can be specific behaviors the individual
has performed, they are also likely to include more general descriptions of
the persons personality and behavior.
On the other hand, the features of the predicate, which pertain to behaviors or actions, are likely to activate previously formed event and episode
models of these actions. The actors depicted in these representations could
93
sometimes be the same person that is specified in the subject of the statement to be comprehended. However, they could include many other individuals as well.
In each case, the speed with which a previously formed knowledge representation reaches threshold depends on factors specified in the retrieval
postulates described in chapter 2. That is, it increases with the number of
features of the representation that are also contained in the probe set (Postulate 2.2) and also the degree to which the representation is already vibrating as a result of prior cognitive activity (Postulate 2.4). Note that if a
preexisting knowledge representation exists that contains both the probe
cues pertaining to the subject and the probe cues pertaining to the predicate, it is likely to be activated much more quickly than other representations that contain only one set of cues. Moreover:
Postulate 4.4. If a memory search activates a previously formed knowledge
representation whose features include both the subject probe cues and the
predicate probe cues, the new information is spontaneously recognized as redundant with prior knowledge and a new representation is not formed.
When the two sets of probe cues activate different knowledge representations, it means that a previously formed knowledge representation of the
information does not exist. Then, the process we assume is somewhat more
complicated. In this case, the probe cues associated with the subject are
likely to activate a number of knowledge representations that contain them,
some of whose features are more likely to overlap than others. In addition,
the predicate probe cues activate all of the models that explicate this action or state of affairs. (If the stimulus is Michael Jordan sank a threepointer, for example, it would activate all preexisting situation models
whose predicate was sank a three-pointer or the equivalent.) The subjects of these situation models could vary, but they are likely to have a subset of features in common. In the composite set of subject features that are
activated, the features that are represented most often are weighted most
heavily. Then, the following postulate applies:
Postulate 4.5. If no previously formed model (i.e., knowledge unit) refers to
both the subject and predicate of a target statement, the features that are activated by the targets subject are compared to a weighted composite of the
features associated with the targets predicate. This comparison yields a global
estimate of similarity, S. A new situation model is formed if either (a) S is greater
than a minimum threshold value, TC, or (b) the targets subject contains a subset
of features that are common to the subject of all models activated by the probe.
In the latter case, the new model is constructed by combining the composite of
features activated by the subject and the composite of features activated by
the predicate. In all cases, however, only the features that are weighted heavily in
the composite are included in the new model that is constructed.
94
CHAPTER 4
This postulate obviously requires unpacking. Suppose the target statement is Michael Jordan sank a three-pointer. According to Postulate 4.3,
features of the Subject resonate with all previously stored representations
containing these cues, whereas sank a three-pointer resonates with all
previously formed event models of this activity. However, several of these
models are likely to have Michael Jordan as the subject. Then, according to
Postulate 4.4, these representations should be activated by resonance from
both the targets subject and the targets predicate, and the statement
should be regarded as redundant with prior knowledge.
However, suppose the target statement is Michael Jordan rode a motorcycle. In this case, the predicate may not activate models whose subject is
Michael Jordan. However, features that are frequently associated with motorcycle riders are big, muscular, and athletic. Because these features
are likely to be conveyed in the generalized person representations that are
activated by Michael Jordan, the similarity, S, should be above threshold
TC . Therefore, a new model is likely to be constructed.
In contrast, suppose the target statement were Woody Allen rode a motorcycle. In this case, the person representations activated by Woody Allen are less likely to include big, muscular and athletic. Consequently, the similarity index computed on the basis of a comparison
features and the composite set activated by rides a motorcycle may be
below threshold. Nevertheless, the features of Woody Allen are likely to
include those that are common to all previously formed models of motorcycle riders (e.g., human). Thus, a new model of this statement might still be
formed, but it would not be constructed as quickly as it would if statement
had referred to Mr. Jordan.
Finally, suppose the target statement was A hippopotamus rode a motorcycle. In this case, the similarity index would not only be below threshold, but the features of the subject would not be shared by all previously
formed models of rode a motorcycle. In this case, therefore, a situation
model of the statement would not be constructed spontaneously.
The implications of this analysis for comprehension become clear in the
context of the next postulate:
Postulate 4.6. A statement is comprehended spontaneously if it is redundant
with prior knowledge or if a new model is constructed of it. If neither is the
case, comprehension of the statement is not spontaneous, but requires (Executor-controlled) goal-directed processing.
Thus, to summarize our example, Michael Jordan sank a three-pointer

and Michael Jordan rode a motorcycle would both be comprehended
very quickly. Woody Allen rode a motorcycle would also be comprehended spontaneously, but would take more time to comprehend than the
95
first two statements. In contrast, A hippopotamus rode a motorcycle

would not be comprehended spontaneously, but would require conscious
Executor-controlled processing.
An interesting implication of the preceding analysis derives from Postulate 4.5. When a new situation model is formed, only those features that occur frequently in the representations that are activated by the subject and
predicate are depicted in this model. This means that the greater the amount
of information one has acquired about a referent in the past, the less detailed
the new information is likely to be. Thus, the representation of Michael Jordan riding a motorcycle is likely to be less detailed if one has encountered
numerous instances of people riding motorcycles, or has had exposure to
Mr. Jordan in many different situations, than if ones previous experience is
limited in these respects. More generally, ones initial encounter with an exemplar is likely to be represented more vividly than later encounters that
have been comprehended in terms of more general sets of features.
A related implication surrounds the comprehension of a statement about
members of a general category (e.g., basketball players). According to
Postulate 4.5, comprehending such a statement should depend on the diversity of exemplars who are known to have engaged in the action described.
For example, consider the statement A basketball player wore Nikes. In
this case, a generalized person representation of basketball players may
be activated that refers to Michael Jordan as an exemplar (see chap. 1).
Moreover, previously formed situation models of wears Nikes are likely to
include Michael Jordan as the subject as a result of television commercials
that portray him as such. Consequently, the resonance elicited by basketball player and wears Nikes could lead a previously formed model of Michael Jordan in Nike sports shoes to be activated very quickly, and so the
statement might be comprehended almost as quickly as a statement that refers to Michael Jordan directly.
In contrast, suppose the category does not have a known exemplar that
has engaged in the activity in question. Then, different predictions would
be made. For example, The tennis player wore Nikes should be comprehended less quickly than The basketball player wore Nikes. This should
be true even though Nikes are associated just as strongly with playing tennis as with playing basketball. The reason is that wearing Nikes and playing
basketball are both strongly associated with Michael Jordan, whereas
wearing Nikes is less likely to be associated with a previously formed model
of a specific tennis player.
Empirical Evidence
Wyer and Radvansky (1999) reported evidence in support of this conceptualization. In one study, participants read a series of statements on a computer screen with instructions to press one of two designated keys on the
96
CHAPTER 4
keyboard to indicate whether the statement was comprehensible or not.

Some of the statements were anomalous (e.g., A show played poker). Others were meaningful and of four different types:
a. true exemplarreferent statements (e.g., Jane Fonda does aerobics)8
b. false exemplarreferent statements (e.g., Jane Fonda plays professional hockey);
c. exemplarreferent statements of unknown validity (e.g., Jane Fonda
rides a motorcycle)
d. categoryreferent statements that are likely to be true but for which
there is no known exemplar (e.g., A steel worker does aerobics).
All four of these statements were judged as comprehensible (mean likelihood = .92). However, they differed significantly in the time required to
make this assessment. Of particular interest is a comparison of the time to
comprehend true exemplarreferent statements and the time to comprehend categoryreferent statements. According to the earlier Wyer and Srull
(1989) theory, the subject and predicate of the exemplarreferent statements would have to be instantiated in terms of more general semantic concepts in order to be comprehended. According to this theory, therefore, it
should take longer to comprehend these statements than to comprehend
statements that were already conveyed in terms of these concepts. In contrast, the present theory implies that true exemplarreferent statements
are interpreted in terms of preexisting situation models, and therefore
should be comprehended very quickly.
This was in fact the case. The first column of Table 4.2 (Experiment 1)
shows the mean time to comprehend the four types of statements described earlier. True exemplarreferent statements were comprehended
more quickly (M = 1.35 s) than either categoryreferent statements (M = 1.68
s), false exemplarreferent statements (M = 1.64 s), or exemplarreferent
statements of unknown validity (M = 1.55 s).
Other differences in comprehension time are also consistent with expectations. For example, exemplarreferent statements for which a previously
formed model exists should take less time to comprehend than statements
for which a new model must be constructed. Moreover, false exemplarreferent statements (e.g., Jane Fonda plays professional basketball) should
take longer to comprehend than statements of uncertain validity. This is because the similarity index that is computed on the basis of a global comparison of features activated by their subject (Jane Fonda) with features that
are common to the models activated by the predicate (plays professional
8
At the time of this study, Jane Fonda was a well-known promoter of aerobics on television
and magazines.
97

TABLE 4.2
A Comparison of the Time Required to Comprehend
Statements and the Time to Judge Their Validity
Comprehension
Time
Validation
Time
Difference
1.35
1.64
1.55
1.86
0.20
0.22
1.55
2.07
0.52
1.68
1.88
0.20
1.84
1.99
0.15
2.17
2.23
0.06
Experiment 1
Exemplar-referent statements
Definitely true (e.g., Mozart wrote a Symphony)
Definitely false (e.g., Mozart played rock and roll)
Unknown validity (e.g., Jane Fonda read her
horoscope)
Category-referent statements with no known exemplar (e.g., a Musician did aerobics)
Experiment 2
True exemplar-referent statements (e.g., Jane Fonda
did aerobics)
Category-referent statements with known exemplar
(e.g., An actress does aerobics)
Note. Response times are in seconds. Adapted from Wyer and Radvansky (1999).
basketball) is below comprehension threshold, leading to an additional

stage of processing (Postulate 4.5). Although the difference in time to comprehend the two types of statements (1.64 s vs. 1.55 s) was not statistically
significant, it directionally confirmed this expectation.
In a second study, participants were asked to comprehend either true
exemplarreferent statements (Jane Fonda does aerobics) or statements
that referred to the category to which the exemplar belonged (e.g., an actress does aerobics). According to our assumptions, the second type of
statement should activate the same preexisting event model as the first
type and, therefore, should be comprehended nearly as quickly. This was in
fact the case, as also shown in Table 4.2 (Experiment 2). Although comprehension generally took longer in this study than in the first one, the times to
comprehend true exemplarreferent statements and categoryreferent
statements with a known exemplar did not appreciably differ (1.84 s vs. 2.00
s, respectively).
Priming Effects
Wyer and Radvanskys (1999) second experiment evaluated a more subtle
implication of the theory we propose. These implications are suggested in
part by Postulate 2.4, that the resonance of a previously accessed knowledge representation decreases gradually over time. This means that if the
98
CHAPTER 4
residual resonance has not dissipated at the time resonance is emitted

from other sources, the representation may be activated more quickly than
it otherwise would. This possibility has unique implications for the effects
of exposure to an exemplar-referent statement on the comprehension of a
category-referent statement that refers to the same or a different event.
For example, suppose participants are exposed to an exemplar-referent
statement (e.g., Jane Fonda did aerobics). This statement may increase
the accessibility of other event models that refer to the same subject (e.g.,
Jane Fonda played in a movie). This, in turn, should increase the speed of
recognizing a new statement that refers to this activity, making it easier to
comprehend. Moreover, this should be true even if the new statement refers to someone else. Thus, exposure to Jane Fonda did aerobics should
increase the speed of comprehending not only Jane Fonda played in a
movie but statements about other persons in the same activity (e.g.,
Dustin Hoffman played in a movie). In contrast, the subject of a categoryreferent statement (e.g., An actress did aerobics) should activate a generalized person representation (cf. Fig. 1.2), but is less likely to activate event
models. Consequently, exposure to this type of statement should not influence the comprehension of statements that refer to different behaviors.
The second experiment reported by Wyer and Radvansky (1999) provided some support for this prediction. Participants were first asked to
comprehend either an exemplar-referent statement about a persons behavior (e.g., Jane Fonda did aerobics) or a statement that referred to a category to which the exemplar belonged (e.g., An actress did aerobics).
Then, they were asked to comprehend a target statement that referred to
either the same behavior or a different one with which the exemplar was associated (e.g., played in a movie). Exposure to a category-referent statement decreased the time required to comprehend a target statement only if
the two statements referred to the same behavior. In contrast, exposure to
an exemplar-referent statement decreased the time to comprehend a target
statement regardless of whether it referred to the same behavior or to a different behavior with which the exemplar was also associated.
SPONTANEOUS VALIDITY JUDGMENTS

The most unique implications of Wyer and Radvanskys (1999) formulation
have yet to be considered. As noted earlier, people who receive information about a known person or event may not only comprehend the event
but may judge it to be true or false. Moreover, this determination is often
made automatically in the course of comprehending the statement, in the
absence of a more specific goal that requires its assessment. Thus, the
99
statement Abraham Lincoln wore a beard is comprehended and immediately recognized a true. By the same token, Abraham Lincoln shot John F.
Kennedy is immediately recognized as false. These assessments occur
spontaneously, in the absence of any goal that requires them.
These phenomena and the conditions in which they occur are taken into
account in the formulation we propose. According to the theory, the recognition that a statement is true or false is often a necessary by-product of the
comprehension processes embodied in Postulates 4.4 and 4.5. This is most
obvious when peoples comprehension of a statement activates a preexisting situation model of the event to which it refers. In this case, awareness of the prior existence of the model is typically sufficient to affirm its
validity. There may be exceptions to this. Situation models can obviously
be constructed of fictitious events (e.g., Santa Claus came down the chimney on Christmas Eve). However, the representation of these events is
likely to contain a tag denoting their fictional character. In the absence of
this tag, statements that instantiate an existing situation model may be
spontaneously recognized as true in the course of comprehending them.
The following postulate formalizes this assumption.
Postulate 4.7. A statement is spontaneously recognized as true in the course
of comprehending it if (a) a situation model of the event or state described already exists in memory and (b) no tag has been attached to the model that
designates it as false.
Spontaneous judgments of a statements validity can also occur when a

previously formed model of the situation does not exist. As Postulate 4.5 implies, comprehension may often be a two-stage process. That is, persons
who encounter a statement first compute a global estimate of the similarity
(S) between the statement and the referents of previously formed situation
models of the predicate, and comprehend the statement if this similarity is
above threshold. In the course of this determination, however, they theoretically become aware of the value of S in relation to two other threshold
values that are necessary to confirm or disconfirm its validity. If S exceeds
or falls short of these values, the statement may be identified spontaneously as true or false. To formalize:
Postulate 4.8. A statement about an event is recognized as true in the course
of comprehending it if the similarity of its subject to those of previously
formed models of the event (S) exceeds an upper threshold, TT . The statement is identified as false if S is below a lower threshold, TF . If S falls between
TT and TF , the statement is comprehended but its validity is not spontaneously evaluated.
100
CHAPTER 4
Postulate 4.9. Spontaneous inferences that a statement is true or false are

transferred to the Work Space along with a representation of the statement
that has been formed in the course of comprehending it.
Thus, as an example, An Israeli donated money to a synagogue would

be quickly judged as plausible because the categorical representation of
Israeli is likely to share many features with the subject of these models (S
> TT). In contrast, Hitler donated money to a synagogue is likely to be immediately recognized as false, because few features of Hitler are likely to
characterize the subject of previously formed models of donating money to
a synagogue (i.e., S < TF). Finally, Madonna donated money to a synagogue would not be immediately evaluated as either true or false, because
Madonnas known features are neither highly similar nor highly dissimilar
to those of the relevant models subjects (TF < S < TT). That is, the latter
statement would be comprehended, but its validity would only be assessed
if more specific goal-directed processing required it.
Empirical Evidence
Wyer and Radvansky (1999) obtained support for the implications of these
postulates in the two experiments described earlier. Rather than being
asked to comprehend the statements they were presented in these conditions, some participants in these studies were asked to indicate whether
the statements were true or false. If statements were verified spontaneously in the course of comprehending them, there would be little difference
between the time required to make these verifications and the time required to comprehend them. If verification requires additional processing
subsequent to comprehension, the time to perform the verification should
be greater than comprehension time. According to Postulate 4.8, however,
this latter condition arises only when the subject of a statement is neither
similar nor dissimilar to the subject of preexisting models associated the
predicate (e.g., Madonna donated money to a synagogue).
Data bearing on these possibilities are summarized in Table 4.2. The first
column of data refers to the time required to comprehend different types of
target statements, as indicated in our earlier discussion. The second column summarizes the time to judge whether the same statements were true
or false. Although the statements in Experiment 1 were verified more slowly
than they were comprehended, this difference was appreciable only for exemplarreferent statements of unclear validity. The difference in all other
cases was small and did not differ over the three types of statement, suggesting a general response bias that was independent of the cognitive processes underlying judgment. In Experiment 2, the time to verify statements
101
and the time to comprehend them were even more similar. Thus, these results are quite consistent with the formulation we propose.
Further Implications
The results we have brought to bear on the comprehension and validation
of propositions were obtained in a rather artificial research situation. However, results of research in other areas are consistent with this conceptualization. Of particular interest in this regard are the implications of Postulate 4.7. To reiterate, this postulate states that if a pre-existing situation
model of a statement exists in memory, and if no tag has been attached to
the model that denotes it a false, the statement will become spontaneously
as true. Thus, suppose participants construct a new situation model of a
statement in one context in the course of comprehending it but do not
spontaneously assess its validity. Later, however, they encounter the same
statement in a different context. To comprehend the statement, they presumably retrieve the situation model they constructed of it earlier (Postulate 4.4). However, because the statement is comprehended in terms of this
model, it should be spontaneously identified as true (Postulate 4.7). Put
more simply, the exposure to and comprehension of a statement at one
point in time may increase beliefs in the statements validity at a later time.
The study by Hasher et al. (1977) described in chapter 3 provides direct
support for this prediction. That is, people who had reported their beliefs
in statements about unfamiliar events and situations in an initial experimental session increased their beliefs in the statements when they encountered
them in a second experimental session 1 week later. Presumably, participants comprehension of the statements in the initial session led them to
construct situation models of the events described by the statements. Consequently, they retrieved and used these models to comprehend the same
statements when they encountered them later. However, the pre-existence
of these models in memory should increase inferences of the statements
validity, as Hasher et al.s findings indicate.
These phenomena may pertain not only to the verification of a single
event, but also to sequences of events of which multiple-segment episode
models are constructed. In a study by Green and Brock (2000), participants
read a story about a violent crime. They were clearly aware that the story
was factual. Nevertheless, they later estimated the incidence of story-related events in the real world to be higher than participants who had not
read the story.
In a similar vein, Shrum and his colleagues (OGuinn & Shrum, 1997;
Shrum, Wyer, & OGuinn, 1998) found that heavy television viewers were
more likely than light viewers to overestimate the incidence of crimerelated events in the real world. These quite different sets of studies con-
102
CHAPTER 4
verge on the possibility that people, either in the laboratory or when watching television, construct episode models of the events they encounter and
store these models in memory. Later, they are likely to retrieve the models
when contemplating events that are redundant with their implications, and
therefore estimate the events to be more likely to occur. Similar conclusions can be drawn from research on the effects of generating an explanation for hypothetical events on later beliefs that the event will occur (Ross,
Lepper, Strack, and Steinmetz, 1977; Sherman, Skov, Hervitz, & Stock, 1981).
We discuss these studies in more detail in chapter 9.
An Alternative Conceptualization
The conceptualization proposed in this chapter should be considered in
the context of a formulation proposed by Gilbert (1991) on the role of comprehension on beliefs. Gilbert hypothesized that belief formation is a twostep process, which first involves comprehension and only later involves
verification. Moreover, in order to comprehend a statement, one must necessarily entertain the possibility that it is true. Thus, for example, comprehension of the statement George W. Bush donated money to the Taliban
entails a tentative acceptance of this state of affairs, and only later is the
statements validity rejected. One implication of this conceptualization is
that people who have been asked to learn true and false statements under
conditions of high information load are more likely to recall false statements as true than to recall true statements as false. This is presumably because the heavy processing demands prevent the second (verification)
stage of processing from occurring. Thus, the statements are likely to be
judged as true, based on results of the first stage of processing alone. As a
consequence, objectively true statements are evaluated correctly but objectively false statements are not.
The processes that Gilbert postulates to occur at the comprehension
stage are compatible with those assumed by the conceptualization proposed here. That is, the comprehension of a statement about an event involves the construction of a mental simulation of it (i.e., a situation model).
Moreover, the storage of this representation in memory increases its likelihood that a statement asserting the event it describes will later be judged
as true (Postulate 4.8). On the other hand, this postulate distinguishes between the comprehension of a statement and the criteria that are used to
estimate its validity. Moreover, it specifies the conditions in which the latter assessment is made spontaneously rather than at a later, postcomprehension stage of processing. Our conceptualization and Gilberts are therefore compatible with one qualification. That is, according to the proposed
model, statements that differ substantially from the subset of previously
formed models of the events they describe are likely to be below falsifica-
103
tion threshold (TF). These statements should be judged as invalid spontaneously at the time the statements are comprehended and should later be
reported as false regardless of information processing load. Thus, for example, the proposed theory predicts that the statement Abraham Lincoln
landed on the moon would be spontaneously judged as false regardless of
processing load. Gilberts formulation does not seem to make this prediction. With this exception, however, the two formulations are consistent.
The Comprehension of Observed Events and Situations

The comprehension of observed events is in some respects simpler and in
other respects more complex than the comprehension of verbal descriptions of the events. For one thing, observations typically consist of not just
one but a continuous stream of actions. For another, observations typically
include features of not only the event itself but also the situational context
in which it occurs.
In the first regard, it seems unlikely that people store a continuous
stream of experiences in memory. Rather, as Newtson (1973, 1976) argued,
we are likely to extract visual frames from an ongoing stream of events that
identify the actions that occur. The specific nature of these frames has been
somewhat controversial. Newtson (1973, 1976), for example, assumed that
the frames extracted are located at breakpoints between the end of one
event and the onset of another. As Ebbesen (1980; see also Wyer & Srull,
1989) pointed out, however, this assumption seems rather tenuous, and the
empirical support for it is equivocal.
Following Ebbesen (1980), we assume that the representation formed in
the course of comprehending an ongoing sequence of behavioral events
consists of static frames, each of which corresponds to a meaningful act,
and therefore, constitutes a mental picture of the event that occurred.
Thus, suppose a person enters his house, takes off his coat, goes into the
kitchen, opens the refrigerator and gets a can of beer. Observers do not
need to store a mental movie of these actions in order to comprehend
and remember what went on. Rather, they need only to extract a frame that
is typical of each event and that permits its essential features to be identified. Each of these frames could normally be represented in a mental image
of the sort we assume to compose an event model, and the sequence of
frames as a whole could constitute an episode model.
The operations of the Simulator in this process must nevertheless be
clarified. We assume that as an individual observes a sequence of events,
features of these events resonate with previously constructed event models in memory. Although the features of the observation are constantly
changing, the features that occur within the time interval associated with a
104
CHAPTER 4
meaningful event are all likely to be sufficiently similar to those of a previously formed event model to activate it. Thus, for example, the features extracted from the observation of someone taking off his coat may all resonate with a static event model of this behavior, leading the model to be
accessed and used in much the same manner it is used to comprehend a
verbal description of the event.
However, the features of an observed event sequence also include aspects of the situational context in which the sequence occurred. These features can also resonate with situational features of preexisting event and
episode models in Permanent Storage. The subset of situational features
that are most typical of these models could constitute a situational appraisal of the sort described earlier in this chapter. However, because situational features are common to all of the specific events that are observed,
these features, along with others that occur frequently in the models that
have been activated, may be transmitted to the Work Space as a separate
representation that is independent of the more specific event models that
are constructed. This situational representation could operate in several
ways. First, if the events that are specified in the activated models are relatively homogenous, the situational appraisal defined by these features
could give rise to expectations for the sorts of events that are likely to occur in the situation at hand. Second, the appraisal, once it is transmitted to
the Work Space, might be included among the eliciting conditions of a production of the sort described in chapter 1, and thus could stimulate the observers own behavior in the situation. This possibility, which could help to
account for the different effects of concept activation on behaviors described in chapter 3 (cf. Bargh et al., 1996; Colcombe & Wyer, 2001), is considered further in chapter 7, where we discuss the situational norms that
govern expectations for the communications that are exchanged in social
interaction and the impact of deviations from these expectations on the interpretation of these communications.
CONCLUDING REMARKS
To summarize, the proposed formulation specifies the conditions in which
a statement about a situation-specific event or state of affairs is spontaneously identified as true or false in the course of comprehending it. Our analysis has largely been restricted to statements about events and states that
are temporally and situationally constrained (i.e., those that are comprehended by constructing a situation model). However, similar processes
should guide the comprehension and verification of other types of statements as well. Be that as it may, the implication that many statements are
verified spontaneously in the course of comprehending them has further
105
implications. As implied by Postulate 4.9, statements that are spontaneously verified as true or false are transmitted to the Work Space along with
a designation of their validity, and this spontaneous validity assessment
may be subject to further processing. For example, communications that
are obviously true or obviously false may violate normative principles of
conversation, and therefore may stimulate a conscious effort to understand
why the communication was uttered. This additional processing may sometimes lead to a reconstrual of the statements meaning that differs from the
meaning implied by the situation model that was constructed in the course
of comprehending it. We elaborate this possibility and its implications in
chapter 6.
However, the discussion of comprehension processes in this chapter has
focused on the representations formed of statements about single events
or, alternatively, on event models. Equally important considerations surround the way in which event models are combined to form a multiplesegment episode model of a thematically related sequence of events. The
next chapter is devoted to these matters.
C H A P T E R
5
The Construction of Episode
Models and Generalized
Narrative Representations
Chapter 4 focused largely on the comprehension of statements about single

events of the sort that are represented in memory as event models. Most of
the information we acquire in daily life, however, consists of sequences of
events that theoretically give rise to the construction of multiple-segment
episode models. To understand the conditions in which these models are
constructed, several questions must be answered. These questions concern
the number of episode models that are formed from a given sequence of
events, the role of thematic relatedness in the construction of these models, and the way in which the temporal order of events is reconstructed
when the events are not contained in the same model. We attempt to provide tentative answers to these questions in this chapter.
Not all mental representations of event sequences are episode models,
of course. In the last section of this chapter, we discuss briefly the construction of generalized narrative representations. Then, in chapter 6, we consider the conditions in which these generalized representations are used to
comprehend specific experiences, and also the extent to which abstract
representations of an experience, once constructed, can influence memory
for this experience.
THE CONSTRUCTION OF EPISODE MODELS
How Many Episode Models Are Formed?
People obviously do not construct a single representation of their experiences over the course of a lifetime, or even over the course of a day. Nor do
they construct a single representation of the events they read about in a
106
EPISODE MODELS AND GENERALIZED NARRATIVE REPRESENTATIONS
107
novel or see in a movie. Rather, they are likely to construct several different
episode models, each pertaining to a given sequence of events that are thematically as well as temporally related. Thus, the statements John packed
his bags. John left for the airport. may be represented in a two-segment episode model because they refer to the same person and are both related to
going on a trip. In contrast, the sentences John packed his bags. Mary watered the flowers. are likely to be represented in different models because
they appear to have little to do with one another. These considerations are
captured by a more general postulate:
Postulate 5.1. Once a situation model has been formed, subsequent information is integrated into the model if and only if this information concerns the
same situation as the model or a thematically related one.
This integration is presumably performed by the Simulator. Thus, the

Simulator presumably forms an event model of the first piece of information it receives. When it encounters a second piece, it forms an event model
of this as well and then compares its features with those of the first model.
If the two sets of known features are compatible, features of one model that
are either unspecified or absent are instantiated in terms of specified features of the other. An episode model is then formed composed of both
event models in combination. On the other hand, suppose the features of
the second model are incompatible with those of the first. Then, the integration is not performed.
Thus, suppose people who have formed an event model of George
packed his bags encounter the statement He left for the airport. The referent of he in the second statement is compatible with the representation
of George in the first statement, and so a model of the second statement
is likely to be integrated with the first model, producing an episode model
of the events described two statements in combination. However, suppose
the second statement is She left for the airport. In this case, she cannot
be instantiated in terms of the features of the first representation, and so a
separate model of the event is constructed.
In some cases, there might be an a priori reason to suppose that statements are thematically related despite appearances to the contrary. In this
case, an integration of the two statements might be performed. However,
this would not occur spontaneously by the Simulator, but rather, would involve higher order, Executor-controlled processing. Thus, in our example,
people might speculate that George is actually a woman. Or, they might
infer that she refers to an individual who was mentioned in a previous
statement to which they are not privy. However, these speculations would
not occur automatically.
108
CHAPTER 5
Note that these latter, more deliberative integration processes require a

modification of the representation that was initially formed of the first
statement. As we indicated in chapter 2, new features can be added to a
representation, but existing features can never be erased. Thus, the goaldirected processing described in the previous paragraph would result in
two representations: (a) the initial event model, which contains only the
original information, and (b) a second representation that contains implications of both the original information (as modified to accommodate the new
material) and the new. This assumption, which becomes of particular importance in our discussion of higher order comprehension processes (see
chap. 7), is captured by the following postulate:
Postulate 5.2. When new information requires a modification of an event or
episode model that had been constructed on the basis of prior information,
the original model is retained in memory. However, a second representation
is constructed consisting of a modified version of the first model along with
the implications of the new information.
In this regard, the thematic relatedness of information is not an inherent

property of the information itself. That is, it can depend to a large extent on
the prior knowledge of the recipient. For example, the statements Mary got
a coke. John went to the bathroom. may appear to be more thematically
related to people who know that John and Mary are preparing to see a
movie at a local theater than to people who do not have this knowledge. A
more graphic example is provided by the following passage written by
Nancy McCarrell (see Bransford & Stein, 1984):
Remember Sally, the person I mentioned in my last letter? Youll never guess
what she did this week. First, she let loose a team of gophers. The plan backfired when a dog chased them away. She then threw a party but the guests
failed to bring their motorcycles. Furthermore, her stereo system was not
loud enough. Sally spent the next day looking for a Peeping Tom but was unable to find one in the yellow pages. Obscene phone calls gave her some hope
until the number was changed. It was the installation of blinking neon lights
across the street that finally did the trick. Sally framed the ad from the classified section and now has it hanging on her wall. (p. 51)
To the naive reader, the events described in the passage appear to have little to do with one another. However, a reader who has prior knowledge that
the woman is trying to get rid of her noisy neighbors is much better able to
understand the causal and thematic relatedness of the events.
The question is how to determine empirically whether episode models
are formed of information and, if so, how many. A procedure for determining this was developed by Radvansky and Zacks (1991) in investigating the
109
construction of situation models of states of affairs. In a typical study

(Radvansky & Zacks, 1991), participants were asked to learn statements
concerning an object and its location. In multiple-object, single-location conditions, the statements described 1, 2, or 3 objects in a given location (e.g.,
The book is on the table. The cup is on the table. The pen is on the table. In single-object, multiple-location conditions, the sentences described a
given object in 1, 2, or 3 locations (The book is on the table. The book is
on the chair. The book is on the floor.). Radvansky and Zacks (1991) reasoned that people can easily imagine several objects in a given location
and, therefore, would construct a single situation model under multipleobject, single-location conditions regardless of the number of objects involved. However, an object cannot occupy different locations at the same
time. Therefore, in single-object, multiple-location conditions, participants
would have to form a different model on the basis of each statement.
Participants after learning the statements to criterion were given a recognition memory test, and the time to verify the statement was observed.
Radvansky and Zacks argued that if participants formed a different model
on the basis of each statement, the models would interfere with one another at the time of retrieval, and the magnitude of this interference would
increase with the number of models involved. To this extent, the time to
recognize a given statement should increase with the number of statements
in the set. However, suppose the statements were integrated into a single
situation model. Then, there should be no interference, and so the verification time should not depend on the number of statements. Results were
consistent with expectations based on this assumption. That is, response
times increased with the number of statements presented under singleobject, multiple-location conditions, but were independent of the number of
statements presented in multiple-object, single-location conditions.
Radvanksy and Zacks study not only provided insight into the construction of situation models but also suggested a general methodology for determining the conditions in which they are constructed (for other demonstrations of the effectiveness of the procedure, see Radvansky, Spieler, &
Zacks, 1993). This methodology was therefore applied in investigating the
validity of Postulate 5.1. In this study (Radvansky, Wyer, Curiel, & Lutz,
1997), participants were asked to learn sentences each describing the relationship of a person (e.g., a lawyer) to an object (e.g., a toothbrush). In
some cases, the person and object were related in terms of a specific behavior (e.g., the lawyer is buying a toothbrush), whereas in other cases,
they were related in terms of ownership (e.g., the lawyer owns a toothbrush). Finally, the objects presented in some conditions (e.g., a toothbrush, shaving cream, and aspirin) could all be purchased in a single location (e.g., a drugstore) but in other cases, the objects (e.g., a toothbrush, a
bicycle, and a diamond ring) were typically purchased in different loca-
110
CHAPTER 5
tions. As in Radvansky and Zacks (1991) studies, participants after learning

the statements were asked to indicate whether or not these statements
(along with distracters) had been among those they had been asked to
learn. Response times to presented items were recorded.
Although the design of the study was complex, the predictions were
straightforward. That is, participants should form a single episode model
from statements that describe a single person buying objects in a single location. In this case, therefore, the time to verify such statements should not
depend on the number that were presented. However, separate event models should be formed of a person who buys objects in different locations, or
of different persons who buy the same object regardless of location. Finally,
when the relationship involves ownership, which is not temporally constrained, event models should not be formed at all. Therefore, in all cases
but the first, verification times should increase with the number of statements presented.
Results shown in Fig. 5.1 are quite consistent with these expectations. The
figure shows mean verification times to statements in each experimental condition as a function of the number of statements presented. The top left panel
shows that when the statements described a single persons purchase of objects in a single location, verification time was virtually identical regardless of
the number of statements presented. In all other conditions, however, verification time increased with the number of statements presented.
Thus, these findings confirm the implications of Postulate 5.1. First, event
models are only likely to be formed in the course of comprehending statements about events that are situationally and temporally constrained. Second, event models are likely to be integrated into a single multiple-segment
episode model only if the events occur in a single situation. A qualification
on the latter conclusion, however, is indicated in the next section.
Effects of Thematic Relatedness
Radvansky et al.s (1997) study indicated that events that occur in different
locations may be represented separately in memory rather than integrated
into a single episode model. However, the construction of episode models
may be influenced by other factors, the effects of which could override the
impact of situational differences. If events are causally related, for example,
they may be incorporated into a single model even if they occur in different
locations. By the same token, thematically unrelated events are likely to be
represented in different models even if they occur in the same situation.
An early study by Wyer and Bodenhausen (1985) provided indirect evidence of these differences. Participants read a story about events that occurred at a cocktail party that was written from the perspective of a person
who attended it. Two target episodes were included among several others
111
FIG. 5.1. Mean time to verify a statement as a function of (a) the number of persons associated with a given object (denoted by squares) or (b) the number of objects associated with a given person (denoted by circles).
The two top panels show responses to items pertaining to buying objects that could be found in either a single
location (a drugstore) or different locations. The two bottom panels show responses to items pertaining to
owning these objects. RT = response time. Based on data reported by Radvansky et al. (1997).
112
CHAPTER 5
that took place in the course of the party. One episode consisted of events
that were causally related and occurred in the same situation. Specifically,
someone bumped a guests arm as he was reaching for an hors doeuvre,
causing him to spill his drink on a womans dress. The woman thereupon
called the offender a disparaging name and stalked off to the bathroom to
clean up. In contrast, the events composing a second episode were causally
related but occurred in different physical locations. That is, it concerned a
guests telling a story about Willa, who learned that her father was dying,
packed her bags and got on a plane to San Francisco, had three drinks on
the plane to calm herself, but felt tipsy by the time the plane landed. Consequently, she forgot the name of the hospital where her father was staying
and wound up crying on the streets of San Francisco.
One of the episodes was conveyed nearer the end of the party scenario
than the other. However, the order in which the events composing each episode were described was varied, sometimes being in chronological order
and in other cases being in the reverse order (e.g., Willa had cried on the
streets of San Francisco. She had forgotten the name of the hospital where
her father was staying. She was feeling tipsy as a result of having three
drinks on the plane . . .). Moreover, the events were all mentioned in sequence or were separated by other, unrelated events that occurred at the
party and temporarily distracted the visitor from those involved in the episode being described.
Participants after reading the entire cocktail party narrative and a short
delay were asked to recall what they had read. Participants typically recalled the events composing each target scenario in chronological order regardless of the order in which they had been presented and regardless of
whether the descriptions of them were separated by unrelated events. This
suggests that they integrated the events in each scenario into a multiplesegment episode model that they later used as a basis for recalling what
had occurred, and this was true regardless of whether the events occurred
in the same physical situation or in different situations. However, they recalled the last story mentioned in the narrative before recalling the one
that occurred earlier. Thus, participants formed separate episode models
of the two event sequences despite their occurrence in the same general
situation (the party), perhaps assigning them a common header (e.g., the
cocktail party). Thus, the residual resonance of the more recently constructed episode model led it to be the first one retrieved and reported
(Postulate 2.4).
Temporal Coding of Episode Models
The conclusion that thematically unrelated sequences of events are organized into different episode models and are stored separately in memory is
quite plausible. But if this is the case, how can the temporal order of the se-
113
quences of events be reconstructed? We can often remember going to Bali

on vacation before George Bush was elected President despite the fact that
the events are totally unrelated. How is this accomplished? The order of
such events may in some situations be computed rather than remembered.
For example, the events may each be compared to a landmark event (e.g.,
my wifes having a baby) that is personally relevant and whose temporal
position is fixed (Loftus & Marburger, 1983). Or, people may simply apply a
heuristic, reasoning that the more they can remember about an event, the
more recently it probably occurred (Brown, Rips, & Shevell, 1985). Even
within the same situation, however, the order of events that are ostensibly
unrelated to one another can often be constructed fairly accurately.
It is conceivable that in the latter situations, people may assign temporal
tags to events that denote the order of their occurrence and permit their
order to be reconstructed later. This possibility was examined by Wyer,
Shoben, Fuhrman, and Bodenhausen (1985). People read a version of the
Willa story employed by Wyer and Bodenhausen (1985) and mentioned
earlier. In some cases, however, the first and last scenes were modified to
make them ostensibly unrelated to those that occurred in the middle one.
These events, which were presented in a paragraph, were as follows:
Willa was awakened by the alarm.
She got out of bed.
She had breakfast.
She got on a plane to San Francisco.
She had three drinks.
She felt dizzy.
She was glad when the plane landed.
She went shopping in the city.
She bought a sweater.
It seemed reasonable to suppose that participants would construct three

episode models in the course of comprehending the events (pertaining to
getting up in the morning, the plane ride, and the shopping spree), designating each by a header that refers to what the model is about. They might
also assign a temporal tag to each model (e.g., early, middle, late, etc.)
to indicate when the episode occurred in relation to others. Thus, suppose
participants are later asked which of two events occurred earlier. They
might first identify the episode model(s) in which the events were contained and compare the tags associated with them. If the tags differ, they
can make a quick decision. If the tags are the same, however, participants
must compute the order of the events on the basis of their location in the
episode model itself, and so they take more time to respond. Thus, the
overall decision time, Ttotal , might be given by the equation:
114
CHAPTER 5
Ttotal = t0 + tI + tC1 + tC2 ,

Where tI is the time to identify the episode model in which the events are
contained, tC1 is the time to compare the temporal tags associated with the
models, and tC2 is the time to locate the relative positions of the events
within a given episode model should this prove necessary.
To evaluate this conceptualization, participants after reading the passage were given several different pairs of events and asked in each case to
decide which event in the pair occurred sooner. Participants took more
time to make this decision when one or both events occurred at the boundary of an episode model than when they occurred in the middle, suggesting
that more time was required to identify the model in which an event was
contained (tI ) when its location was relatively ambiguous. Second, participants took more time to compare events if they occurred in the same
model than if they occurred in different ones, and this was true independently of the positions of the events in the story as a whole. This finding is
consistent with the assumption that when the events are contained in the
same model, their order cannot be determined on the basis of the temporal
tags assigned to the models. Therefore, an additional amount of time (tC2 ) is
required to make this determination.9
In additional conditions of Wyer, Shoben, Fuhrman, and Bodenhausen
(1985) study, the three scenes of the story were similar to those employed
by Wyer and Bodenhausen (1985) and, therefore, were thematically related.
In this case, decision times were generally longer than they were when the
scenes were unrelated. Wyer et al. attributed this difference to an increase
in tI (the time to identify the model in which the events were contained).
However, other results were comparable to those that occurred when
the scenes were unrelated, suggesting that different episode models were
formed in this condition as well. This conclusion might seem to contradict
the implications of Wyer and Bodenhausens findings. However, a sequence of thematically related events may be more likely to be integrated
into a single episode model when the events are described in the context of
several other, unrelated episodes, as was the case in the cocktail party
story that Wyer and Bodenhausens participants read. When a sequence of
9
As an aside, it might be noted that the conceptualization predicts that in general, the time to
compare two events will be less when they are far apart in the sequence than when they are
close together. This suggests a symbolic temporal distance effect of the sort observed by
Nottenburg and Shoben (1980) and similar to that obtained in other judgment domains (cf.
Banks, 1977). According to this conceptualization, however, the effect results from the fact that
the likelihood that the events being compared are in different models (and, therefore, tC2 = 0) increases with the distance between them. In fact, although a symbolic distance effect was observed in Wyer, Shoben, and Bodenhausens (1985) study, it was eliminated when the relative
numbers of between-model and within-model comparisons was controlled.
115
events is conveyed in isolation, the different situations appear more distinct, and so different models may be constructed.
Effects of Familiarity
An additional factor that might determine the extent to which a sequence of
events is divided into different episode models is the similarity of the sequence to ones that participants have been exposed in the past and are already represented in memory. For example, it seems intuitively reasonable
to suppose that familiar sequences are encoded into broader conceptual
units than novel ones (Newtson, 1973, 1976). Thus, multiple-segment episode
models may be formed more frequently in the first case than the second.
Stan Colcombe and I (reported in Wyer et al., 2002) examined this possibility using procedures similar to those employed by Radvansky and his
colleagues and described earlier. We constructed four stories, each composed of events that occurred in three related situations. The events in two
stories were fairly novel; one described Willas ill-fated trip to visit her sick
father and a second described a womans encounter with a bear while
camping. The events composing the other two stories were routine, describing a visit to a restaurant (with scenes pertaining to entering, eating,
and paying) and going to a movie (with scenes pertaining to buying tickets,
eating some popcorn, and seeing the show).
Three versions of each story were prepared. Each version described six
events, with the number of events taken from each scene (1, 2, or 3) varying
in a counterbalanced design. (Thus, one version of the restaurant story described one event that occurred while entering, two events that occurred
while ordering, and three events that occurred while paying; another described three events that occurred while entering, one while eating, and
two while paying, etc.) Participants read the stories under one of two conditions. In other-referent conditions, the protagonist in the story was a hypothetical person. In self-referent conditions, participants were told to imagine
themselves experiencing the situations described, and the protagonist was
referred to as you.
After learning the stories to criterion, participants were asked to verify
the events that were mentioned in each story. We assumed that the extent
to which verification time increased with the number of events described in
each scene of the stories indicated the extent to which participants formed
a single episode model of the scene or separate models. Pooled over otherreferent and self-referent conditions, the time to verify novel events increased with the number presented in the scene to which they pertained
(1.23 s vs. 1.35 s when 1 vs. 3 events were presented, respectively), suggesting that a different model was formed of each scene. In contrast, the time to
verify familiar events did not depend on the number presented (1.48 s vs.
116
CHAPTER 5
1.40 s, respectively). These data therefore confirm the expectation that participants are more likely to construct a single episode model from a series
of familiar events than from a sequence of novel ones.
In addition, the effects of set size were contingent on the nature of the
protagonist. Specifically, verification times increased with the number of
events presented in a scene when the protagonist was described as you
(1.32 s vs. 1.41 s, when 1 vs. 3 events were presented, respectively). This
was not the case, however, when hypothetical others were involved (1.40 s
vs. 1.33 s, respectively). Thus, participants were apparently less likely to
construct a single multiple-segment episode model of events that they
imagined themselves experiencing than events that they imagined an unfamiliar other experiencing, and this was true even when the situation described was very familiar. These results suggest that the criteria that people employ in comprehending self-referent events may differ from those
they apply when comprehending information about unfamiliar others. More
direct evidence of this difference (Colcombe & Wyer, 2002) is described in
chapter 6.
THE CONSTRUCTION OF GENERALIZED

NARRATIVE REPRESENTATIONS
Episode models of social experiences are presumably formed spontaneously in the course of comprehending them. However, more abstract narrative representations of temporally related events exist in memory as well.
Unlike episode models, these generalized event representations are unlikely to be constructed automatically. Rather, they are formed in the
course of conscious goal-directed activity that requires them. This activity
can potentially be of several types.
First, people who wish to tell others about an experience they have had or
a movie they have seen may describe the events that occurred in fairly general terms. In pursuit of this objective, they may form a linguistically coded
representation of the events that is more abstract than the episode model
they formed of these events in the course of comprehending them, and may
store the representation it in memory independently of this model.
Second, people who have had a number of very similar experiences in a
given type of situation might form a generalized representation of the
events containing features that the experiences have in common but are
not specific as to time and place. Thus, for example, people might form a
general representation of the events that occur in a restaurant (looking at
the menu, ordering, eating, etc.) based on a number of specific experiences
in which the events have occurred. These generalized representations,
which have been referred to elsewhere as scripts (Schank & Abelson, 1977),
117
event prototypes (Colcombe & Wyer, 2002) or event schemas (Graesser &
Nakamura, 1982; Wyer & Srull, 1989), can function as implicit theories about
the events that occur in the type of situation at hand. As we note in the next
chapter, however, the abstraction of generalized representations from specific exemplars may occur less frequently than is often assumed. Colcombe
and Wyer (2002) showed that people typically do not form generalized narrative representations of events on the basis of exemplars unless they are
exposed to several different exemplars of the generalized narrative in temporal proximity. Even then, the abstraction may not occur unless people
are stimulated to think about the exemplars in relation to one another.
Thus, although generalized event representations undoubtedly exist in
memory, they may be constructed in other ways.
In fact, many generalized event representations might not be based on
personal experience at all. Rather, people might acquire a general understanding of the causal relations among event concepts through social learning, without directly experiencing the individual events that exemplify
these concepts. Thus, a child may be told that people who study hard get
good grades in school, and that these grades help them to get a good job
upon graduation. This sequence of events could constitute an implicit theory that the child accepts without first-hand knowledge of any specific instance in which the theory holds.
Finally, people may construct generalized event representations in the
course of making a judgment to which it is relevant. For example, people often wish to explain why a past event occurred, or to predict a future event.
Alternatively, they might wish to assess the desirability of a particular
course of action. These events could either be situation specific or more
general. For example, one might wish to explain either why a particular
friend got AIDs or, alternatively, why the incidence of AIDs in underdeveloped countries is increasing. Because the processes of constructing these
generalized representations are somewhat less obvious than the process of
constructing other types, these processes and their implications are worth
discussing in somewhat more detail.
To explain or predict a specific event, people might attempt to identify a
pre-existing episode model in which the event is embedded and use segments that precede or follow this event as bases for their inference. When
such a model does not exist, however, a two-stage process may be required. First, people identify a generalized propositional representation of
the event to be explained (i.e., people get AIDs). Then, having done so,
they search memory for propositions that describe possible antecedents of
it and, if they find one, they base their explanation on this proposition.
118
CHAPTER 5
The assumption that people spontaneously think about the antecedents

and consequences of an event is suggested by McGuire and McGuire (1991)
in conceptualizing the content and structure of thought systems. This conceptualization, which was stimulated in part by McGuires (1960, 1981) earlier work on the cognitive organization of propositional knowledge, assumes that individuals who are intrinsically or extrinsically motivated to
construe the likelihood that an event will occur may search memory for antecedents of the event (i.e., factors that will either facilitate or interfere with
the events occurrence). In contrast, individuals who are stimulated to consider an events desirability are likely to ponder its consequences. In support of these assumptions, the McGuires found that when people were
asked to report their spontaneous thoughts about a target event (e.g., a
vaccine to prevent AIDS will be developed), 78% of their thoughts pertained to either antecedents of the event or consequences of it. Moreover,
providing information about others beliefs that an event would occur stimulated participants to generate more thoughts about antecedents of the
event than about its consequences (56% vs. 22%), whereas providing information about others estimates of the events desirability stimulated more
thoughts about its consequences than about its antecedents (55% vs. 31%).
The McGuires research does not indicate which of several antecedents
or consequences happens to be identified in the course of this cognitive activity. However, when several alternative propositions (or, in some cases,
sets of propositions) could potentially describe a cause or effect of a target
event, people are likely to identify and use the one that is most accessible
in memory at the time (see chap. 3). The accessibility of such a proposition
presumably depends on the strength of its previously formed association
with the target or, if no prior association exists, on the frequency or recency with which it has previously been identified and used for other purposes (see Postulates 2.4 and 2.5). In either case, the identification of the
proposition establishes (or strengthens) an association between the proposition and the target, and so the two propositions in combination may be
stored in memory as a unit, thus functioning as an implicit theory of the
sort described in chapter 1. The construction of this theory should have at
least two consequences. First, people who recall one proposition are likely
to recall the other as well. Second, they should be inclined to use the theory that is constructed from the propositions as a basis for judgments to
which it is applicable.
Empirical Evidence
A study with Jon Hartwick (Wyer & Hartwick, 1984) confirmed these predictions. Participants were first given a list of randomly ordered propositions
and instructed to indicate whether the propositions were understandable.
119
Thirty-two of the propositions comprised 16 pairs of assertions that occupied the positions of A and B in a causal statement of the form, If A, then
B. Moreover, the likelihood that A was true, and the clarity of its implications for B, were systematically varied over the 16 pairs. (Thus, for example, a pair in which A was likely to be true and that had clear implications
for B consisted of the statements, Trucks conveying heavy cargo destroy
highways, and The weight limit on truck cargo will be decreased. In contrast, a pair in which A was likely to be true and had unclear implications
for B was composed of the statements, There will be no national borders
by the end of the year 2000 and War will be a thing of the past by the end
of the [20th] century.) This task presumably increased the accessibility of
the propositions in memory and, therefore, increased the likelihood that
participants would later identify and use them for purposes for which they
were applicable.
Participants then completed a two-part questionnaire in which they were
asked to report their beliefs in half of the propositions they had encountered in the comprehension task (i.e., the likelihood that the events described would occur) and to estimate the desirability of the events described in the remaining ones. This was done in such a way that for a given
AB pair, participants reported either (a) their beliefs in both A and B, (b)
the desirability of both propositions, (c) their belief in A and the desirability of B, or (d) their belief in B and the desirability of A. They were then dismissed but returned for a later session 1 week later, at which time they
were asked to recall as many of these propositions as they could.
We expected that participants who were asked to report their belief in B
would search memory for an antecedent of the event. Moreover, they
should use the first relevant antecedent that comes to mind (Postulate 3.1;
see Henninger & Wyer, 1976). Therefore, if A had been encountered in the
comprehension task and had clear implications for B, participants should
identify and use it as a basis for their judgment, thereby establishing an association between A and B that might otherwise not have existed. Similarly,
participants who were asked to report the desirability of A should search
for a consequence and identify B, and this should also establish an association between the two propositions. In contrast, participants who are asked
to report their belief in A should not identify B. Correspondingly, people
who are asked to estimate the desirability of B should not consider A. In
these conditions, therefore, an association between A and B should not be
formed.
The strength of association that participants formed between A and B in
the first session of the experiment should be reflected in their recall of the
propositions later. Specifically, suppose participants had established an association between A and B. Then, if they recall B during the second session,
they should think of A as well. In other words, their likelihood of recalling A
120
CHAPTER 5
TABLE 5.1
Strength of Association Between Syllogistically Related Propositions as a Function
of the Plausibility of the Antecedent (A), the Clarity of its Implications for the
Consequence (B) and the Type of Ratings Made of A and B
Rating Made
of Consequence (B)
Characteristics of Antecedent (A)

Clear implications for consequence
Plausible
Implausible
Mean
Unclear implications for consequence
Plausible
Implausible
Mean
Rating Made
of Antecedent (A)
Likelihood
Desirability
Likelihood
Desirability
.734
.462
.598
.505
.156
.331
.442
.274
.358
.798
.345
.572
.320
.375
.347
.439
.402
.420
.366
.482
.424
.393
.295
.344
Note. Numbers refer to the difference between the probability of recalling A given that B was recalled and the probability of recalling A given that B was not recalled. Adapted from Wyer and Hartwick
(1984).
if they recall B should be greater if they have recalled B than if they have
not. The difference between these two conditional probabilities therefore
provided an index of the strength of association that people had formed between A and B in the first session of the experiment.
This index is shown in Table 5.1 as a function of (a) the a priori likelihood
that A was true, (b) the clarity of As implications for B, and (c) the type of
rating that participants had made of the propositions in the first session of
the experiment (likelihood vs. desirability). When A had clear implications
for B, it was more strongly associated with B when participants had reported either the likelihood that B was true (.598) or the desirability of A
(.572) than when they had reported either the likelihood that A was true
(.358) or the desirability of B (.331). When As implications for B were unclear, however, these contingencies were not apparent. Thus, when As implications for B were not obvious, participants were less likely to recognize
their relatedness during the first session of the experiment, and so they
based their judgments on other criteria. As a result, an association between
the two propositions was not formed, and so their recall of B had no impact
on the likelihood of recalling A.
Other Determinants of Association
Wyer and Hartwicks (1984) findings obviously did not provide a complete
picture of the associations that are formed among general propositions of
the sort that compose an implicit theory or implicational molecule. McGuire
121
and McGuire (1991) identified several other factors that can exert an influence on these associations. For example, they postulated that people are
likely to think about antecedents of events over which they feel they have
control, but are more inclined to think about consequences of events over
which they have little influence. Therefore, because people feel they have
more control over the events in their personal lives than over events in
the society at large, they tend to think about antecedents of personal life
events (e.g., having a paper rejected for publication) but are more likely
to think about consequences of social events (e.g., a terrorist bombing).
For possibly similar reasons, people are disposed to think about the antecedents of events that are likely to occur in the distant future but to think
about the consequences of immediately impending events over which they
have less control.
Other motivational factors could also influence the type of associations
that people form. For example, people may be motivated to think about desirable rather than undesirable consequences of events that are likely to
occur, and to think about factors that facilitate rather than inhibit the occurrence of events they consider desirable. Although the effects of these
factors were not consistently supported by the McGuires data, their potential importance should not be entirely dismissed.
Indeed, a study by Albarracin and Wyer (2001) provided evidence of
these effects. Briefly, participants who had been induced to feel either
happy or unhappy by writing about a past experience read a persuasive
message containing either strong or weak arguments in favor of performing
a particular behavior. This was done in either the presence or absence of
situational distraction. After reading the message, participants reported
their attitudes toward the behavior being advocated, followed by estimates
of the likelihood and desirability of specific behavioral consequences. Participants who were not distracted based their beliefs and evaluations of the
consequences on the content of the message, and these beliefs and evaluations determined their overall attitude toward the behavior. In contrast,
participants who were distracted from thinking about the communication
based their attitudes on the extraneous affect they were experiencing and
misattributed to their feelings about the behavior. Once these attitudes
were formed, however, they had reciprocal effects on outcome-related beliefs and evaluations. For example, participants with favorable attitudes reported stronger beliefs that desirable consequences would occur, and reported that likely-to-occur outcomes were more desirable.
The Construction of Implicational Molecules
Propositions that become associated as a result of the processes described
in this chapter constitute an implicit theory about the causal relatedness of
the events they describe that can be used to comprehend and make infer-
122
CHAPTER 5
ences about more specific experiences. A conceptualization of these inference processes was mentioned briefly in chapter 1. To reiterate, sets of general propositions that become associatively linked in memory can form
implicational molecules that are stored in memory as conceptual units and,
therefore, can be brought to bear on the comprehension of new information that exemplifies them (Abelson & Reich, 1969; Wyer & Carlston, 1979,
1994). In combination, the propositions that compose a molecule can exemplify a general principle that one uses to comprehend and make inferences
about the world in which one lives. Many types of molecules can potentially
exist.10 The molecules of primary concern in the present context consist of
two or more verbally coded propositions, each describing a different event
or state of affairs. In combination, the segments of a molecule exemplify a
generalization about the world in which we live. For example, the generalization that smoking causes lung cancer might be represented in the twosegment molecule:
[P smokes; P has (will get) lung cancer].
Other generalizations may have moral overtones. For example, the generalization that people get what they deserve might be conveyed in a justdeserts molecule; i.e.:
[P does a bad thing; P encounters misfortune]
or, alternatively,
[P does a good thing; P has good fortune].
Once implicational molecules are formed, they can be used to interpret

specific events to which they are applicable. This possibility is particularly
important when not all of the information that is necessary to instantiate
the propositions in a molecule is available. For example, people might use
the smoking molecule described earlier to infer that a person who
smokes will get lung cancer. Alternatively, they might infer that a person
they hear has lung cancer is a smoker. Similarly, they might use the justdeserts molecule to infer that a person who stole money will be punished,
or that a person who was involved in a serious accident was likely to have
10
Not all implicational molecules pertain to events. Some, implied by cognitive balance theory (Heider, 1957), may concern the sentiment relations among a group of individuals (Picek,
Sherman, & Shiffrin, 1975; Sentis & Burnstein, 1979). Other, syllogism molecules may specify
the logical relations among propositions (Loken & Wyer, 1983). These possibilities, and empirical evidence bearing on them, have been summarized elsewhere (cf. Wyer & Carlston, 1979,
1994).
10
123
done something wrong. A more formal statement of these inference processes is provided in chapter 10, where we discuss the impact of implicit theories on judgments and decisions in some detail.
The associations that give rise to the construction of an implicational
molecule can often be formed as a result of the cooccurrence of its constituents in the course of comprehending information and making judgments.
However, they can also be motivationally determined. As Kunda (1990)
pointed out, people can often be motivated to believe that they have desirable attributes and that these attributes will allow them to have a happy
and successful life. Therefore, they may often construct implicit theories
and implicational molecules that confirm this belief. In doing so, they may
selectively attend to new information, and selectively seek and retrieve previously acquired knowledge, that confirms their desire to maintain a favorable view of themselves and their world. This possibility is elaborated in
chapter 10, where we consider the motivational factors that underlie the
construction and use of implicit theories in more detail.
C H A P T E R
6
The Impact of Generalized
Event Representations on
Comprehension and Memory
People construct at least two types of narrative representations of the

experiences they encounter in daily life. On one hand, they spontaneously form episode models in the course of comprehending specific events
they personally experience, observe, or read about. In addition, they may
intentionally form generalized representations of events in the pursuit of
a more specific objective that requires them. These latter representations can sometimes be abstract codings of events that that occur frequently in certain types of situations and, therefore, constitute event prototypes (Colcombe & Wyer, 2002) or schemata (Graesser & Nakamura, 1982).
They can also be linguistically coded descriptions of a specific experience
that people form in the course of communicating about the experience
to others.
Although generalized event representations and episode models are theoretically stored in memory as separate units of knowledge (Postulate 2.1),
they can obviously exert an influence on one another. For example, event
prototypes can be used as a basis for comprehending new experiences that
exemplify them and, therefore, can influence the mental representation that
is formed of these experiences. Moreover, an abstract narrative representation that is formed in the process of communicating about an experience
might later be recalled and used to attain other objectives to which it is relevant, and this could be done without consulting the episode models on
which the representation was originally based.
The present chapter discusses these possibilities in detail and presents
data bearing on them. We first examine the role of prototypic event repre-
124
IMPACT OF GENERALIZED EVENT REPRESENTATIONS
125
sentations on the comprehension and representation of new experiences

that exemplify them. In doing so, we consider the conditions in which these
prototypes are formed and the extent to which they are actually used to
comprehend ones own experiences and those of familiar others. Second,
we discuss the abstract representations that people form of a personal experience in the course of communicating about the experience to others,
and consider the way in which these representations can influence memory
for details of the experience being represented. Finally, we discuss the implications of these communication phenomena for the impact of telling stories about oneself on self-perceptions. In this context, we consider differences in the storytelling that characterizes parentchild interactions in
different societies and discuss their implications for cultural differences in
self-perception.
THE ROLE OF PROTOTYPIC EVENT

REPRESENTATIONS IN THE COMPREHENSION
OF ONES OWN AND OTHERS EXPERIENCES
Many of the events that compose our daily life experiences are fairly routine. That is, they occur repeatedly, with minor variations, at many different
times and in many places. The events that typically occur in the course of
visiting a restaurant, cashing a check at the bank, or changing a flat tire,
and the sequence in which these activities take place, are fairly commonplace, and there is widespread agreement within a particular culture concerning the events and when they occur. People may also engage in sequences of behaviors that, although idiosyncratic, they personally perform
routinely on a daily basis. For example, I typically get up at 5 a.m., brush my
teeth, do 30 sit-ups, and read a novel while drinking a cup of coffee. I then
walk the dog, return home and take a shower before beginning work at
about 8 a.m. This series of behaviors is unlikely to describe anyone but myself. Nevertheless, it is fairly invariant.
It might therefore seem intuitively likely that people construct generalized representations of sequences of events that they personally experience or observe others experience. However, these representations would
not have the character of an episode model. In particular, the events are
likely to be coded propositionally rather than in terms of mental images. To
the extent these event prototypes are formed, they seem likely to influence
the comprehension of new experiences that exemplify them. In fact, however, the conditions in which this influence occurs are more restricted than
one might expect. In the pages that follow, we consider the nature of these
restrictions.
126
CHAPTER 6
Preliminary Considerations
The assumption that people construct event prototypes is quite common in
research and theory on prose comprehension (e.g., Bower, Black, & Turner,
1979; Graesser, 1981; Graesser, Gordon, & Sawyer, 1979; Schank & Abelson,
1977). These representations are unlikely to be formed spontaneously, however. Rather, they are constructed in the pursuit of specific processing objectives that require them. This might be done in a manner similar to that
suggested by Hintzmans (1986) conceptualization of the construction of
schematic representations from exemplars. Once such a representation is
formed, however, people store it in memory, and can later recall and use it
to comprehend new experiences to which it is applicable.
Several theories have been proposed to address these matters (Bower et
al., 1979; Graesser et al., 1979; Graesser & Nakamura, 1982; Trafimow &
Wyer, 1993). Most of these theories employ a principle of cognitive economy similar to that implied by Postulate 4.4. That is, they assume that when
people encounter a sequence of events that is redundant with the events
contained in an event prototype, they usually do not construct a detailed
representation of these events. Rather, they retain only enough features
that permit them to reconstruct the events that occurred on the basis of
the prototype. However, the theories make different assumptions about the
nature of the features that are retained.
For example, Graesser et al. (1979; see also Bower et al., 1979) assume
that when a sequence of events can be predicted on the basis of a more
general event prototype, people typically do not store a representation of
these events in memory at all. Rather, they store only a pointer to the prototype (i.e., a set of retrieval cues that identify it) along with a set of translation rules that instantiate features of the prototype in terms of specific features of the exemplar. Thus, suppose people read that Joe went to Doms
Patio Villa for dinner, ordered lasagna and a bottle of Chianti, and paid
$15.95 for the meal. They might not construct an episode model of the
events. Instead, they might store a pointer to a prototypic restaurant scenario along with the translation rules, customer = Joe, restaurant =
Doms, meal = lasagna, drink = Chianti, etc. These translation rules,
along with the content of the prototype, would be sufficient to reconstruct
what had occurred. They could also be used to infer the occurrence of
other events that were not specified in the information but typically occur
in restaurants (e.g., that John looked at the menu or actually ate the meal).
Note that these inferred events cannot be distinguished in the representation from the events that were actually mentioned. As a result, errors may
occur in recalling the information later (Graesser et al., 1979).
A related conceptualization was proposed by Graesser and Nakamura
(1982). They suggest that when people read descriptions of events the
events that compose a prototype, they comprehend these events in terms
127
of the prototype and copy them into the representation they form, instantiating the generic features of the prototype in terms of specific ones
that are mentioned in the information. However, they do not include segments of the prototype to which the information does not refer. Thus, the
events that were specified in the information and those that were not mentioned can be more easily distinguished.
A limitation of both conceptualizations, however, surrounds the representation of events that are not instantiations of the prototype. The occurrence of these events in daily life is less often the exception than the rule.
For example, a man who visits a restaurant is likely to look at the menu, order the meal, eat and pay the bill. However, he might also ponder a painting
on the wall, say hello to a colleague who is seated at a nearby table, and
spill some of his wine on the tablecloth. The way in which these latter
events are integrated into the representation of the experience as a whole
is not clearly specified. In fact, the theories simply assume that the events
are appended to the representation as tags. To this extent, there would
be no way to reconstruct the point in the sequence that the events occurred. In many cases, however, this reconstruction is obviously possible.
A conceptualization developed by David Trafimow (Trafimow & Wyer,
1993) eliminates this ambiguity. Trafimow, like Graesser et al. (1979), assumes that people who encounter a sequence of events that exemplify a
prototypic sequence do not normally retain these events in the representation they form. However, there is an important exception to this rule. That
is, when a nonprototypic event occurs, people retain a prototypic event
that occurred in temporal proximity to it in order to localize its position in
the sequence. Thus, in our restaurant example, people who learn about
Johns visit to Doms might not normally retain the events that occurred.
However, suppose that while John was ordering his meal, he said hello to a
colleague at the next table. In this case, the person may retain a depiction
of Johns ordering the meal in the representation, thus permitting the point
at which the nonprototypic event took place to be reconstructed.
1. An Empirical Test
Trafimow designed an intriguing series of experiments to evaluate these
possibilities. Suppose that when participants interpret a sequence of events
in terms of a prototype, they only retain prototype-relevant events in the
representation they construct if these events are necessary to localize
nonprototypic ones. Then, the addition of nonprototypic events to a sequence should increase the number of prototypic ones that are retained
and, therefore, should increase the ability to recall these events at a later
point in time. On the other hand, suppose participants do not comprehend
the sequence in terms of a prototype, but rather, construct an episode
128
CHAPTER 6
model of the events. In this case, additional event descriptions are likely to
increase comprehension difficulty and, therefore, to decrease the likelihood of retaining the presented events in the representation that is formed.
In other words, the addition of the nonprototypic events might decrease the
recall of the other, prototype-related ones. (Evidence that the likelihood of
recalling a given event decreases as the total number of events increases is
consistent with the set size effects observed in other research paradigms;
see Anderson & Bower, 1973; Srull & Brand, 1983.)
To evaluate this possibility, Trafimow and Wyer (1993) identified several
situations with which college students were familiar (e.g., cashing a check,
photocopying a piece of paper, making tea, etc.) and constructed six events
that typically occurred in each. The photocopying sequence, for example,
included found the article, got some change, found a machine,
aligned the copy, put in the coins, and pressed the button. However,
several of the events composing each sequence were not unique to the situation at hand, but could occur in other situations as well. Therefore, without a prior indication of the situation being described, the nature of the situation was unlikely to be immediately identified. In addition to the
prototypic events, six unrelated events were identified that could plausibly
occur in the situation of concern (e.g., remembered he had to return a
phone call, ate a piece of candy, saw a person he knew, etc.).
Stimulus stories were then constructed by combining the prototyperelated events and the prototype-unrelated ones. In one experiment (Trafimow & Wyer, 1993, Experiment 2), the stories contained four prototyperelated events and either two or six unrelated ones. In some cases, the
story was introduced with a statement that identified the situation (e.g.,
John needed to photocopy a paper) and in other cases, the situation was
not indicated until the end (After he had photocopied the paper, John
left). Participants read four stories that differed in terms of (a) the number
of prototype-unrelated events presented, and (b) whether the prototype
was identified at the outset or not until the end. Then, after a short delay,
they were asked to recall all of the events they had read about.
Results were quite consistent with expectations. When the situation was
not identified at the outset, participants were unlikely to identify and use a
prototype to comprehend the information. In this case, increasing the number of nonprototypic events presented decreased the proportion of prototypic events that were recalled (from .218 to .195, when 6 vs. 2 unrelated
events were presented, respectively). When the situation was identified at
the outset, however (thus stimulating participants to comprehend in the information in terms of a prototype), increasing the number of nonprototypic
events presented increased the proportion of prototypic events that participants recalled (from .179 to .279 when 2 vs. 6 nonprototypic events were
presented).
129
More subtle implications of the theory were also confirmed. For example, if participants retain prototypic events in the representation in order to
localize the nonprototypic events, the associations they form between
these events should lead one event to cue the recall of the other. To this extent, participants recall of a prototypic event should be preceded by the recall of the nonprototypic event that came before it in the sequence. When
the prototypic situation was identified at the outset, this occurred with a
greater probability than would be expected by chance. When the situation
was not specified until the end of the story, however, this was not the case.
A more general prediction of the theory results from its implication that
when no prototypic-unrelated events are included in the sequence, no prototypic events at all should be retained in the representation that is formed.
Under these conditions, participants theoretically have no way of distinguishing between the prototypic events that were presented and those that
were not. Consequently, participants should not only have poor recall of
the presented events but also should tend to recall events as presented
when they actually were not mentioned. On the other hand, suppose several unrelated events are added to the representation, leading the presented prototype-related events to be retained. Then, these events can
later be distinguished from the prototypic events that were not presented,
and so intrusion errors should decrease. An additional study (Trafimow &
Wyer, 1993, Experiment 3) confirmed these predictions. Participants read
stories composed of four prototype-related events and either four unrelated events or no such events. When the situation was identified at the outset, participants recalled a greater proportion of prototypic events when
four unrelated events had been presented than when none had been presented (.21 vs. .13), confirming the results of the earlier study. At the same
time, the number of intrusion errors was significantly lower in the former
condition than the latter (0.25 vs. 0.96).
2. Methodological Implications
Trafimow and Wyers (1993) findings, which were replicated by Colcombe and Wyer (2002, Experiment 1), provide evidence of the impact of using event prototypes to comprehend situation-specific sequences of events.
However, the conditions in which this occurs may be more limited than one
might expect. For one thing, prototypic event representations are usually
not formed spontaneously. Rather, they are only constructed in order to attain a specific objective that requires them. Moreover, even if these representations have been formed and stored in memory, they may not always
be retrieved and used. As we assumed in chapter 4, people who encounter
a situation-specific sequence of events are likely to comprehend it with reference to a pre-existing episode model of the person and/or type of event in
130
CHAPTER 6
question. To this extent, they might interpret one exemplar of a routinized

sequence of behavior in terms of another exemplar that happens to come
to mind quickly rather than a more abstract representation that is less
easily accessible in memory. (Thus, I might interpret a description of someones visit to McDonalds in terms of an episode model of a personal experience I have recently had at McDonalds rather than a more general representation of fast-food restaurant visits.)
It can sometimes be difficult to distinguish between the use of an exemplar and the use of a more general prototype (Barsalou, 1990). If the conclusions drawn from Trafimows studies are valid, however, the procedure he
employed can be used for this purpose. Specifically, the use of an exemplar
or an event prototype to interpret a sequence of thematically related
events can be inferred from the effect of introducing theme-unrelated
events on the recall of theme-related ones. That is, suppose the introduction of these events increases the likelihood of recalling theme-related
events (as Trafimow and Wyer found when a prototype was not made salient at the outset). This would indicate that an abstract representation of
the sequence was used to interpret. If the introduction of these events decreases the recall of theme-related ones, however, it would indicate that a
more general event representation was not used. The studies described in
the following pages are based on this reasoning.
When Are Prototypes Formed?
As we have speculated, the conditions in which people spontaneously abstract generalized event representations from a number of exemplars could
be quite limited. A large number of exemplars may be necessary to stimulate the construction of a generalized event sequence. A study with Stan
Colcombe (Colcombe & Wyer, 2002, Experiment 2) demonstrated this possibility. Materials for the study were similar to those employed by Gick and
Holyoak (1983) in research in analogical reasoning. Specifically, we constructed six different exemplars of a problem situation that could be resolved by applying the same convergence principle (i.e., several small entities can do the work of one if they converge simultaneously on the task at
hand). One story described a problem of fixing some laboratory equipment:
A woman walked into her physics lab to find that a very rare and expensive
type of light bulb had been left on overnight, fusing the filaments inside the
bulb together. The woman knew that a sufficiently intense blast from an ultrasonic wave generator would be able to separate the filaments and thus render
the bulb useful again. Several of these wave generators were available. Unfortunately, a blast from the generator at an intensity high enough to separate
the filaments would shatter the glass bulb surrounding the filaments. However, the physicist took several generators and set them to a relatively low in-
131
tensity. After carefully positioning the wave generators in a circle around the
bulb, she simultaneously discharged all of them. This short blast from many
different sources at low intensity separated the filaments and preserved the
glass bulb. Thus, the light bulb was repaired, and the womans physics experiments continued successfully.
A second scenario described the dismantling of a bomb:

Security police discovered that a terrorist had planted a bomb at a major
American airport. A bomb squad was called in to deal with the problem. Unfortunately, the type of bomb that the terrorists had planted was so sensitive
that it was almost impossible to defuse. The bomb squad knew that if they
could cool the bomb down to a very low temperature using liquid nitrogen, it
might be possible to defuse. However, if they cooled the bomb down unevenly or too quickly, it would definitely explode. The only cooling device
they had at the airport used a single large hose, and would deliver too much
liquid nitrogen at a single point, thus setting off the bomb. However, the bomb
squad used an attachment to split the single hose into several smaller tubes.
They positioned these tubes in a circle around the bomb. Then, they slowly
released the nitrogen through the many different tubes. Focusing all of the
small tubes on the bomb simultaneously cooled the bomb down slowly and
evenly, and so the bomb was able to be defused.
Three other stories concerned (a) putting out the fire in a burning woodshed, (b) taking over a corporation, and (c) extinguishing an oil well fire.
Participants were exposed to either one or five of these context stories
followed by a target story. The target, which concerned an attack on the
stronghold of an evil dictator, had four versions. Two versions described
four theme-related events and either two or six unrelated ones. Two other
versions described four theme-unrelated events and either two or six related ones. For example, a story containing six theme-related events and
four theme-unrelated events was:
A small country was under the rule of an evil dictator who was despised by
the people. The dictator lived in a fortress with many roads leading to it like
spokes in a wheel. The dictator was cruel. Rebels realized that they had enough
forces to overrun the fortress if they could attack all at once. The rebel general
noted that the weather had been turning colder lately. A rebel spy reported back
that the dictator had planted mines that would allow only a few men to pass
at once along the many roads leading to the fortress. Given that all of the rebel forces must attack the fortress at once, it seemed as though the rebels
plans were foiled. Birds flocked in a nearby tree. However, the rebel general
was very smart and instructed his men to split up into several groups, each
taking a different road to the fortress. A dog howled in the distance. In these
small groups, the men could pass over the mines without setting them off and
132
CHAPTER 6
meet up again at the fortress. Thus, the rebels overthrew the dictator, and the
people in the kingdom lived happily ever after.
Participants were introduced to the study with instructions that we were

interested in how people understand the sort of information they read in stories and novels outside the laboratory, and that they should read the stories
as if they encountered them in a book. Participants read the stories in sequence, with the target story being at the end. Finally, after a 5-min filler task,
participants recalled the target story in as much detail as possible.
If participants constructed an episode model of the target story without
referring to the stories they had read earlier, the addition of theme-unrelated events should decrease their recall of theme-related ones. This was
the case when only one exemplar of the prototype preceded the target
story. That is, participants recalled a greater proportion of theme-related
events when six unrelated ones were presented (M = .50) than when only
two were presented (M = .80).When participants had been exposed to five
exemplars before reading the target story, however, they recalled a greater
proportion of theme-related events in the first condition than in the second
(.83 vs. .60, when two vs. six unrelated events were presented, respectively). Thus, exposure to five exemplars of the convergence problem appeared to stimulate the extraction of a prototype that, once it was formed,
was used to comprehend subsequent stories. However, a single exemplar
was insufficient to accomplish this.
Colcombe and Wyers study provides one of the first empirical demonstrations that new prototypic representations are actually formed on the
basis of accumulated past experience. Research in other domains (e.g.,
Klein & Loftus, 1993; Klein, Loftus, Trafton, & Fuhrman, 1992; Sherman &
Klein, 1994) suggests that the likelihood of using trait concepts rather than
specific behaviors to describe individuals in a given life domain increases
with the amount of experience one has had in this domain. In only one of
these studies, however (Sherman & Klein, 1994), was the amount of experience experimentally manipulated, and in this study, participants were given
an implicit impression formation goal before they received information
about the person to be described. Therefore, these results do not bear on
the extent to which people spontaneously extract new prototypic representations in the course of comprehending information that exemplifies them.
On the other hand, the generalizeability of the conclusions drawn from
Colcombe and Wyers study may also be limited. For one thing, the exemplars to which participants were exposed were presented in temporal contiguity. Although participants were not explicitly told to think about the stories in relation to one another, their juxtaposition in the same situation may
have stimulated them to believe they were supposed to do so. In everyday
life, these situational demands do not arise. Visits to a restaurant, or obser-
133
vations of a spouses actions when getting up in the morning, occur very

frequently. However, these experiences are temporally separated, and may
seldom be thought about in relation to one another. Consequently, although Colcombe and Wyers (2002, Experiment 2) findings suggest that
prototypic event sequences can sometimes be formed from exemplars
when they are encountered within a short period of time, the conditions in
which this occurs in the absence of a specific goal that requires it may nevertheless be more the exception than the rule. This is particularly true in
the case of personal experiences, to which we now turn.
The Role of Episode Models and Prototypes
in Comprehending Personal Experiences
Many of the behaviors we perform in daily life (e.g., getting up in the morning, washing the dinner dishes, etc.) occur repeatedly with minor variations. The behaviors we see others perform (e.g., my wifes getting ready
for a party, or a roommates studying behavior) may also be fairly invariant. It might seem reasonable to suppose that people spontaneously extract a prototypic representation of the routinized behavior that they later
apply in explaining a particular instance of the behavior and predicting its
consequences. People might also use prototypic representations of their
own behavior to predict what others will do (Ross, Greene, & House, 1977).
However, this is not necessarily the case. For one thing, the conditions in
which people form prototypic representations of their own and others behavior may be limited. The construction of these representations is presumably stimulated by the need to abstract features that are common to
several different exemplars. In daily life, however, people are seldom called
upon to consider their own behavior at one point in time in relation to their
behavior at a different time. This is particularly true if the behavior is routine. As we suggested earlier, many routinized behaviors may be governed
by a production (Anderson, 1983) that, once activated, is applied with little
if any conscious cognitive mediation. To this extent, the behaviors that occur at one point in time may seldom need to be thought about in relation to
similar behavior that occurred earlier. Moreover, even when this need
arises, a single prior instance of these behaviors may be sufficient. To know
what to do at a McDonalds restaurant, for example, I only have to recall my
most recent visit to the restaurant. I do not need to retrieve a prototypic
representations of fast-food restaurant visits in general.
A familiar others routinized behavior may also not be comprehended
with reference to an abstract representation of the behavior. That is, an episode model of the behavior may be constructed on the basis of a previously
formed episode model of the sequence that has been stored in memory, as
suggested by Postulate 4.5. (Moreover, if the sequence is redundant with a
134
CHAPTER 6
previously constructed model, a new representation might not be formed

of it at all; see Postulate 4.4.) An episode model of the others past behavior
may also be sufficient for both explaining the individuals present actions
and predicting his or her future ones. Therefore, the conditions that stimulate the construction of a generalized event representation of a familiar
others routinized behavior could also be very limited.
1. The Comprehension of Routinized Events
Involving Self and Others
Two studies by Colcombe and Wyer (2002, Experiments 3 and 4) confirm
these conjectures. In one study, participants in an initial session of the experiment were asked to list both (a) sequences of activities that they performed on a daily basis and (b) sequences that they observed their mother
and their father perform regularly. Then, in a second session 1 week later,
we asked them to read four stories. One story was based on a routinized sequence of behaviors they had personally reported in Session 1, and a second was based on the routinized activities that a different participant had
reported. A third story described activities that either their mother or their
father routinely performed, and the fourth was based on the activities of another participants parent. Each story consisted of four routinized events
and either two or six events that were not part of the routine. Participants
in reading the stories were told to imagine that they (or one of their parents) were performing the activities. Then, after a short delay, they were
asked to recall the stories they had read.
Results summarized in Table 6.1 are very clear. Participants consistently
recalled fewer routinized events when six novel behaviors were contained
in the story (M = .49) than when only two were contained in it (M = .60).
Moreover, this was true regardless of whether the stories pertained to
TABLE 6.1
Proportion of Routinized Events Recalled as a Function
of the Source of the Story, the Referent of the Story,
and the Number of Nonroutinized Events Presented
Source of Story
Stories about self

Two unrelated events
Six unrelated events
Stories about parent
Two unrelated events
Six unrelated events
Self
Different Person
.63
.49
.51
.45
.70
.57
.55
.45
Note. Based on data from Colcombe and Wyer (2002).
135
themselves (.47 vs. .57) or a parent (.51 vs. .62), and regardless of whether
the stories were based on routines that participants themselves performed
(.53 vs. .67) or those that another had engaged in (.45 vs. .53). Thus, these
data suggest that participants did not use a generalized representation to
interpret either stories about themselves or stories about their parent.
2. The Use of Previously Formed Prototypes
to Comprehend Anothers Behavior
There are two possible reasons why participants did not use prototypic
representations in the preceding study. First, as noted earlier, people might
not form prototypic representations of their own and well-known others
behaviors. Another possibility, however, is that these generalized representations may exist in memory, but people may simply not draw upon them
for use in comprehending new experiences involving themselves and familiar others. A second series of studies support the latter possibility. Participants in these studies read stories similar to those employed by Trafimow
and Wyer (1993). That is, each story consisted of four behaviors that exemplified a prototypic script (cashing a check, changing a tire, etc.) and either
two or six prototype-unrelated behaviors. In one study (Colcombe & Wyer,
2002, Experiment 1), participants were told that the protagonist in the stories was a hypothetical person (Joe), as in Trafimow and Wyers original
studies. In a second experiment, however (Colcombe & Wyer, 2002, Experiment 4), they were told to imagine that the story was about their roommate.
(In this study, the protagonist was referred to as he or she, depending
on whether participants were male or female, respectively). In all cases, the
situation to which the sequence pertained was identified at the outset, thus
ensuring that participants aware of the applicability of the prototype for
comprehending it.
Results of the study are summarized in Table 6.2. When the stories referred to a hypothetical person, Trafimow and Wyers (1993) findings were
TABLE 6.2
Mean Proportion of Prototype-Related and Prototype-Unrelated
Events Recalled as a Function of the Number of Unrelated
Events Mentioned in the Story and the Type of Protagonist
Protagonist
Unfamiliar Other
Roommate
.30
.42
.50
.39
Number of prototype-unrelated events mentioned in story

Two events
Six events
Note. Based on Colcombe and Wyer (2002, Experiments 1 and 4).
136
CHAPTER 6
replicated. That is, participants recalled a greater proportion of prototyperelated events when six unrelated events were presented than when two
were presented (.42 vs. .30, respectively), confirming the fact that participants used a prototype to comprehend the story. When participants imagined the events occurring to their roommate, however, their recall of the
prototype-related events was poorer when six unrelated events were presented than when two were presented (.39 vs. .50, respectively). In this
case, therefore, a prototypic event representation was not employed.
Concluding Remarks
Colcombe and Wyers findings suggest that even when a generalized event
representation exists in memory, people do not always use it to comprehend information about themselves or familiar others. This conclusion is
consistent with the retrieval postulates described earlier, and the results
reported by Wyer and Radvansky (1999). That is, people who receive information about the behavior of a well-known person can often construct a situation-specific event model involving this person very quickly. This is particularly true if a previously formed event model of the persons behavior
already exists in memory. In this case, they are likely to use this model as a
basis for comprehending the information rather than relying on more general knowledge that comes less quickly to mind.
In retrospect, this conclusion is not surprising. Unless people have a reason to think about different instances of their own or familiar others behavior in relation to one another, they are unlikely to engage in this cognitive
activity. In making behavioral decisions, a previously formed episode
model of the behavior and its consequences can often serve just as well as
an abstract prototype. Moreover, as noted earlier, routinized behavior may
be governed by productions and performed with minimal cognitive mediation. The influence of these productions is particularly evident when one is
distracted by thinking about something else. (For example, a person who is
thinking about how to revise a paper while driving to the grocery store may
suddenly find himself about to enter his office parking lot. Similarly, a person may often forget whether he has locked the door to the house before
leaving.) As Bargh (1997) contended, a very large proportion of our behavior may be governed by automatic processes, and there is little need to use
a generalized representation of this behavior that we have stored as part of
declarative knowledge.
Circumstances can nevertheless arise in which a prototypic event for example representation is personally useful. When a persons behavior has
undesirable consequences, the individual might be motivated to avoid
these consequences in the future and, therefore, may attempt to explain it.
To generate this explanation, the person might retrieve a number of epi-
137
sode models of past behavior in situations in which a similar outcome has

occurred, and might abstract commonalities in the behaviors antecedents.
Similar attempts might be made when a behavioral outcome is unexpected.
This cognitive activity is unlikely to occur spontaneously, however, but instead, is driven by Executor-governed processing objectives.
THE EFFECTS OF COMMUNICATING ABOUT

AN EXPERIENCE ON MEMORY AND JUDGMENT
Many abstract representations of experiences are formed in the course of social communication. People are often called upon to describe an experience
to others. These communications can sometimes describe a procedure (e.g.,
instructions about how to change a tire or use a photocopier). More generally, however, they concern a situation-specific experience (e.g., a personal
encounter one has had, or a movie one has recently seen) that one considers
of sufficient interest to relate. In order to generate these communications,
people presumably construct a linguistically coded representation of the experience and, once it is formed, store this representation in memory.
The representations that people form in the process of communicating
about an experience are presumably based on episode models that they
formed of the experience at the time it occurred. However, the representations differ in several respects from the models on which they are based.
Features that are tangentially relevant to the communication goals may be
omitted entirely. Others, which were represented visually in the episode
models, are characterized in more abstract, linguistic terms.
The specific content and structure of the representations that are
formed can depend in part on the communication objective at hand. When
the goal is to describe the experience itself, the communication is likely to
consist of verbal descriptions of the events that occurred (e.g., captions
of the sort that sometimes accompany the images contained in an event or
episode model). Suppose, however, that people want to convey their impressions of the events or the individuals involved in them. Then, their
communications might not describe the events themselves, but rather,
might consist of more general trait descriptions along with characterizations of their own attitudes or opinions. In each case, the representation
that people form may reflect the content and structure of the communication they generate, and this representation may be stored in memory separately from the more specific episode models of the original experiences
that were used to construct it (Postulate 2.1).
This possibility has further implications. As noted in chapter 3, people
who are called upon to make a judgment or decision are unlikely to retrieve
and use all of the information they have available in memory that is poten-
138
CHAPTER 6
tially relevant to this decision. Rather, they retrieve only a subset of this information and do not search further unless their confidence in the validity of
the information or its implications is below threshold (Postulate 3.1; see also
Chaiken, 1987). This sufficiency postulate was applied earlier in conceptualizing the effects of knowledge accessibility on judgments. However, it has implications for memory as well. For example, if a generalized representation of
information has recently been formed in the course of attaining a specific
goal, this representation is likely to be used as a basis for not only making
judgments but also reconstructing the events that occurred. That is, the original representation may only be consulted if the implications of the more recently formed goal-specific representation are below confidence threshold.
A classic study by Loftus and Palmer (1974) provides an example. Some
participants who had been shown a picture of an automobile accident were
later asked to indicate how fast they thought the car was going when it hit
the tree. Others were asked how fast it was going when it crashed into the
tree. Participants estimated that the car was going faster in the second condition than the first. Moreover, when later asked to remember details of the
picture they had seen, participants in the second condition were also more
likely to report (incorrectly) seeing broken glass at the scene of the accident. Participants apparently reconstructed the scene of the accident in a
way that was consistent with implications of the question they had been
asked, forming a new mental representation that contained features in addition to those that were conveyed in the picture. Their later memory was
then based on this newly formed image independently of the one they had
constructed at the time they saw the picture.
In Loftus and Palmers study, however, participants processing objective stimulated them to add features to the representation they had formed
initially. When people wish to describe an experience to others, the representation they form is more likely to omit features that are of peripheral relevance to the description they are providing. This could also decrease the
accuracy of recalling information under conditions in which this representation is consulted. In a study by Schooler and Engstler-Schooler (1990),
participants were shown a series of human faces and asked to describe
some of these events verbally. Later, they were given a memory test for the
faces. One might expect that the increased amount of processing involved
in describing the picture would make it more memorable. In fact, however,
participants were less accurate in identifying the faces they had described
than faces they had seen but not described. Participants who wrote an abstract description of a face apparently formed a new representation of it in
which certain specific features that were useful in distinguishing it from
other faces were omitted, and then used this new representation, which
was less detailed than the original representation, as a basis for their later
recognition responses.
139
Note that the abstract representation that is formed in the course of

communicating verbally about an event does not replace the original (Postulate 5.2). However, because it was formed more recently, it is typically
more accessible in memory, and might often be considered a sufficient basis for recognition responses without consulting the more detailed representation on which it is based. If a goal-specific representation is considered irrelevant, or if participants confidence in the judgments implied by
the representation is below threshold, the more detailed representation
would presumably be retrieved and used.
The Impact of Communication Objectives
on Memory for Observed Experiences
Adaval and Wyer (2003) investigated the conditions in which episode models and goal-specific representations of social experiences were likely to be
activated and used as a basis for reconstructing these experiences. In doing
so, we considered the effects of two communication objectives that people
often have in daily life outside the laboratory: communicating a description
of the events to another, and conveying an impression of the persons involved in the events. We compared the effect of introducing these goals at
the time the event were observed with the effect of introducing them later,
after episode models of the events had already been formed.
1. Procedure
Participants in two studies watched the opening 12 minutes of Edward
Albees Whos Afraid of Virginia Woolf. The segment concerns a married couple who return home from a party and prepare for the visit of some lateevening guests. The protagonists engage in an animated conversation in
which both positive and antagonistic comments are exchanged, thus leaving the couples actual feelings toward one another unclear. Although the
protagonists nonverbal behaviors during the interaction (working on a
crossword puzzle, washing dishes, fixing drinks, cleaning up papers in the
living room, etc.) were relevant to a description of the sequence of events
that occurred, they were relatively incidental to the couples personalities.
Participants in some conditions watched the movie with no specific goal
in mind, much as if they saw it in a theatre. After doing so, however, participants in impression-description conditions were asked unexpectedly to write
a 5-min description of their impressions of the protagonists personalities,
whereas participants in event-description conditions were asked to spend 5
min describing the sequence of events that occurred. Two other groups of
participants were told they would be asked to perform these tasks before
they viewed the movie. Finally, participants under comprehension-only conditions watched the movie without being given specific objectives either be-
140
CHAPTER 6
fore or afterwards. (These participants spent 5 min after watching the

movie writing a description of the university campus.)
After performing the writing task, participants were given a recognition
memory test, the nature of which differed in the two experiments. In Experiment 1, participants were given verbal descriptions of statements and behaviors and asked in each case to indicate whether the statement or behavior described had occurred in the movie. In Experiment 2, we extracted
both pictures of events that had occurred and acoustically recorded statements that protagonists made and presented these stimuli (along with pictures and utterances from other parts of the movie) on a computer with instructions to indicate whether the stimuli had occurred in the movie or not.
2. Predictions
Participants were expected to construct a multiple-segment episode
model of the movie in the course of comprehending it. In addition, participants with a communication objective should construct a more general, linguistically coded representation in the process of generating their communication, the content of which depends on the particular goal being
pursued. Specifically, persons who described the sequence of events that
occurred were expected to include characterizations of not only things the
protagonists said but also things they did, as both types of events were relevant to this description. However, only protagonists verbal statements
were particularly relevant to an understanding of their personalities. Thus,
participants who were told to convey their impressions of the protagonists
were likely to include trait descriptions in their representations along with
descriptions of protagonists statements that exemplified these descriptions, but were unlikely to include nonverbal behaviors.
When participants in these conditions are asked to verify a recognition
item, they should first retrieve the goal-specific representation they have
formed, which is most accessible in memory. If they consider the content of
this representation to be a sufficient basis for evaluating the item, they
should use it without further ado (Postulate 3.1). If it is not, relevant, however, they should default to the episode models they formed at the time
they watched the movie.
Thus, suppose participants who have written a description of the sequence of events that occurred are asked to verify a recognition item. They
should consider the representation they formed in the course of generating
this description to be a sufficient basis for verifying both protagonists
statements and their nonverbal behaviors. However, because this representation is less detailed than the episode models they had formed, they
should be less able to recognize both types of items than they would if they
had not used this representation. In contrast, participants who conveyed
their impressions of the protagonist should consider the representation
141
they formed in the course of this activity to be sufficient for evaluating protagonists statements but not their nonverbal behaviors. Therefore, they
should resort to the use of episode models to verify these behaviors. Consequently, although their recognition of protagonists statements should be
adversely affected by the goal-specific representation they formed, their
recognition of the protagonists behaviors should not be.
3. Results
The results of both studies were very similar and were generally consistent with expectations. Table 6.3 shows the difference between recognition
accuracy under each communication objective condition and accuracy under comprehension-only conditions.11 Writing a description of the sequence
of events that occurred decreased recognition of both protagonists statements and their nonverbal behaviors, and this was true regardless of
whether participants had watched the movie with this objective in mind or
were not told of the objective until afterwards. In contrast, communicating
an impression of the protagonists only decreased participants recognition
of things the protagonists said and did not influence their memory for
things the protagonists did. Moreover, this decrease was only evident when
participants were not given an impression-formation objective until after
they had seen the movie.
The effects of objectives that participants had in mind at the time they
watched the movie deserve further attention. All of the statements and behavior that participants consider interesting enough to communicate to
others are likely to be of interest to themselves as well and, therefore, to be
represented in the episode models they would normally form spontaneously while watching the movie. Therefore, the content of the goal-specific
representation formed by participants with an event-description objective
should be similar regardless of whether they were aware of this objective at
the time they watched the movie or not. To this extent, the decrement in accuracy that results from their use of this representation as a basis for recognition should not depend on when this objective was imposed. This was
the case, as Table 6.3 indicates.
11
The index of recognition accuracy used in Experiment 1, which corrects for guessing, is
given by the equation:
11
P(Acc) =
P(hit) - P(False Alarm)

,
1 - P(False Alarm)
where P(Hit) and P(False Alarm) are the probabilities of identifying an item as having been presented when it was or was not presented, respectively (Hilgard, 1951). This index could not be
applied to Experiment 2, as the number of nonpresented items was too small to obtain a reliable
index of P(False Alarm).
142
CHAPTER 6
TABLE 6.3
Effects of Communication Objectives, and the Point at Which
These Objectives Were Induced on Recognition of Protagonists
Statements and Nonverbal Behaviors
Experiment 1
Event-description objectives
Induced before watching movie
Induced after watching movie
Impression-description objectives
Induced before watching movie
Induced after watching movie
Experiment 2
Statements
Behaviors
Statements
Behaviors
.189a
.065
.360
.297
.070
.044
.046
.095
.130
.108
.000
.021
.045
.022
.046
.046
Note. Based on Adaval and Wyer (2002).

aCell entries refer to differences between the accuracy obtained under each task-objective
condition and accuracy observed in comprehension-only conditions. In Experiment 1, accuracy
was based on the index proposed by Hilgard (1951); see Footnote 12. In Experiment 2, accuracy
was based on the proportion of correct responses to recognition items.
In contrast, participants who are told at the outset that they would be
asked to describe their impressions may include features in the representation they form while watching the movie that are relevant to their impressions but are not depicted in the episode models they would form if they
were only trying to comprehend what is going on. The implications of these
additional features may then be included in the communication they generate later, and their recognition may benefit, as results also suggest.
Effects of Reminders on Recognition
Suppose people after seeing a movie or observing an interaction are later
reminded of the events that occurred. They are likely to comprehend this
reminder with reference to a previously formed representation of the original experience. In the course of doing so, they presumably construct yet another representation of what went on. According to the conceptualization
we propose, the content of this new representation, and its consequent effects on memory, may depend on the nature of the previously formed representation that was used to interpret the reminder.
Adaval and Wyer (2003) investigated these possibilities in a study that
provides further support for the implications of Experiments 1 and 2. In this
study, participants who had watched the movie with instructions to comprehend it were then asked to describe their impressions of the protagonists. After doing so, however, they were dismissed with instructions to return the next day for a second session of the experiment. In this session,
143
some participants were re-exposed to the audio portion of a 4-min portion

of the movie they had seen without the accompanying video. Others were
re-exposed to the video portion without the accompanying audio. Still others received no reminder at all. Finally, all participants were given the verbal recognition task administered in Experiment 1.
Based on the reasoning underlying our interpretation of the first two experiments, we assumed that participants who received a video reminder of
the protagonists nonverbal behavior would find the goal-specific representation they had formed while reporting their impressions to be inapplicable
for comprehending it and, therefore, would retrieve and use the episode
models they had constructed at the time they first watched the movie. Because this model contains representations of things the protagonists said
as well as things they did, both types of events should be included in the
new representation they form. Therefore, if participants use this recently
formed representation as a basis for recognition, their memory for both
statements and behaviors should be facilitated.
However, suppose participants receive an auditory reminder. These participants should consider the goal-specific representation they formed to
be sufficient for interpreting it, and should use it without bothering to consult their episode models of the original movie. In this case, the new representation they form should be similar in content to the goal-specific representation they had formed earlier, containing only abstract depictions of
the protagonists statements. Consequently, this representation should not
benefit their later recognition of these statements relative to conditions in
which the reminder was not given.
Results are consistent with these conjectures. Relative to conditions in
which no reminder was given, a visual reminder increased the likelihood of
recognizing not only protagonists nonverbal behaviors (mean increase =
.110) but their statements as well (M = .056). However, a visual reminder had
relatively little effect on participants recognition of protagonists statements (M = .035) and actually decreased their likelihood of recognizing protagonists behaviors (M = -.126).
General Conclusions
Adaval and Wyers studies provide evidence of the influence of goal-directed processing of a social experience on later memory for this experience. Some caution should be taken in overgeneralizing our specific findings. Impressions of the protagonists in the movie we presented were
conveyed largely through their statements to one another, and their nonverbal behaviors were relatively incidental to these impressions. In many
social situations, individuals nonverbal behavior is likely to have a greater
impact on the impressions that people form of them. In these cases, com-
144
CHAPTER 6
municating impressions of the individuals might interfere with observers

memory for their behavior as well as the statements they make.
The Effects of Storytelling on Self-Representations
and Social Identity
The effects of goal-relevant event representations on memory of the sort
that Adaval and Wyer identified have potentially much broader implications. People are often called upon to recount a past experience they have
had. In doing so, they may tailor their descriptions of the experience to the
values and expectations of their audience, omitting some aspects they believe the audience may consider unpalatable while elaborating on others
that the recipient may like (Higgins & Rholes, 1978). The representation
they form as a result of this communication may be stored in memory and,
therefore, is likely to be called upon again when they tell about the experience a second time. After repeated iterations of this process, a number of
copies of the more abstract, goal-directed representation are formed that
are more accessible in memory than the original. As a result, their likelihood of recalling this original representation may be low.
In some cases, this process could have psychological benefits. For example, a man who has experienced a traumatic event might have a vivid episode model of the experience that elicits considerable negative affect when
it is activated. In communicating about the event to others, however, he
presumably forms a more abstract representation of the event. A second
description of the event may be based on this recently formed representation, leading a still more general representation to be constructed. As the
representations that are formed in the course of repeated communications
about the event become more abstract, fewer emotion-eliciting features
may be retained in them, and so the man can ultimately recall and discuss
the episode without re-experiencing the emotions that occurred earlier. On
the other hand, the original episode model of the experience is retained in
memory, and can potentially be reactivated by a set of retrieval cues that
are unique to a configuration of the features contained in it. In such an
event, the man might find himself re-experiencing the emotions elicited by
the original events even after a period of time has elapsed.
This process has more general implications for the development of individuals perceptions of themselves and others, and for the norms and
values that guide their behavioral decisions. McAdams (1988) argued that
the stories people tell about themselves provide the main source of their
personal and social identity (see Mankowski & Rappaport, 1995, for a similar view). However, suppose people modify the content of the stories they
tell in order to comply with certain expectations or desires of the recipient, and the representations they form in the course of telling these sto-
145
ries are stored in memory. These stories may then become the basis for
their self-perceptions. This means that the self-perceptions that people develop are likely to depend on the type of situations in which they are typically called upon to tell stories about themselves and the expectations
that others have for the content of these stories. Two bodies of research
and theory bear on these processes. One concerns the impact of parentchild interaction in the construction of self-stories. A second area of
research provides insight into cultural and situational influences on individuals personal and social identity.
The Construction of Narratives

in ParentChild Interaction
The role of narratives in the acquisition of social knowledge is particularly
evident at early stages of social development. Much of the knowledge that
children acquire about themselves derives from the stories they tell about
themselves and others. The role of storytelling in early childhood has been
investigated intensively by both Katherine Nelson (1989, 1993; Nelson &
Greundel, 1981) and Peggy Miller (1994, 1995; Miller, Fung, & Mintz, 1996) in
the context of parentchild interaction. Nelsons research has been concerned with the storytelling process itself and its impact on the construction of social reality. Millers research is based on similar assumptions but
has focused on cultural differences in the stories that are exchanged between parents and children and the consequent impact of these differences
on the norms and values that children acquire.
1. Basic Processes
Both Nelson and Miller note that childrens generalized representations
of their personal experiences are often constructed in the course of interactions with their parents. Young children typically do not have a built-in
mechanism for organizing their memories. Therefore, this organization is
often acquired through parentchild interactions in which the parent stimulates the child to recount aspects of an experience by questioning the child
and providing prompts that fill in gaps in the sequence of events that took
place. In other words, the child and the parent collaborate in constructing
the story the child tells about his or her experience. The co-narrative that
emerges may be retained in memory and thus may constitute the childs
reality. In short, the childs memory for the experience may be largely a
construction of the parent, based on the parents perception of the experience, rather than a true reflection of the experience from the childs own
point of view.
146
CHAPTER 6
Certain consequences of this process could be insidious. As Schank and

Abelson (1995) pointed out, children may learn in the course of collaborative storytelling that certain types of events should be recounted but others
should not. To the extent these edited stories become part of the childs
reality, they may be hard to modify later. The extent to which this is true
is often difficult to disentangle. For example, children may fail to include aspects of a parents abusive behavior when describing the interaction because of the fear of repercussions. To this extent, these instances of child
abuse might later be difficult to retrieve. At the same time, as Ceci and
Brucks (1993) pointed out, a prosecutor who wishes to make a conviction
might harass a child into reconstructing an incident of child abuse that may
in fact not have occurred, leading this false memory to become part of the
story that the child remembers as part of the original experience. (Similar
false memories may result from therapy; see Hyman & Loftus, 2002.)
Although our description of these processes focuses on the construction
of an abstract representation of a single event, prototypic event representations could be formed similarly. A parent, for example, may rehearse the
behavior a child should perform in a given social context in order to teach
the child proper behavior. Furthermore, this could be done without reference to specific instances that exemplify the behavior. The childrens perceptions of causality as well as of morality (what behavior are generally
appropriate) could be learned through these processes. Millers work provides examples of this learning.
2. Cultural Differences in Storytelling
Peggy Miller (1995; Miller et al., 1996; Miller, Sandel, Liang, & Fung, 2001)
has analyzed cultural differences in self-perceptions in terms of differences
in the type of collaborative storytelling that characterized parentchild interaction. In an analysis of storytelling involving both European American
and Taiwanese families (Miller et al., 1996; Miller et al., 2000; Miller, Wiley,
Fung, & Liang, 1997), she found that quite different attention was given to
positive and negative behaviors in the stories that children were encouraged to construct and retain about themselves. These differences are particularly evident in stories about misdeeds. For example, Chinese parents
and their children are more likely than their American counterparts to
make the childs present misbehavior a central part of the story the child
tells, but to relate this behavior to past transgressions as well. Thus, the
self-stories that are likely to emerge from these parentchild co-narrations
were likely to be self-critical and to focus on the childs responsibility for
the misdeed. Furthermore, by relating the childs current misdeed to earlier
ones, the stories encouraged the development of a more general conceptualization of the childs undesirable behavior that had implications for his or
her moral character.
147
However, American parentchild co-narratives were more often characterized by a self-favorability bias (Miller et al., 2001). These stories focused much less frequently on a childs past transgression, and when they
did, the transgression was treated humorously, portraying the child in a
positive light despite his or her misdeed. Thus, for example, one story concerned a child who had written on the wall and lied about it. Although the
mother acknowledged the seriousness of the childs negative act, she at the
same time made light of it in helping the child communicate about the act
to others. Another example:
Tommy and his older brother remembered being punished for some misdeed
committed a few days earlier, but none of the participantsTommys brother
or mothercould remember what they had done wrong, a baffling if not inconceivable state of affairs from a Taiwanese perspective. This practice of down
playing transgressions in the narrative medium seemed to be part of a wider
set of practices that [European American] caregivers used to protect their
childrens self esteem. (p. 168)
An equally or more important difference in Eastern and Asian story telling was evident in the stories that parents told about themselves. Taipai
mothers self-stories typically provided positive examples for the child to
emulate. As Miller et al. (2001) suggested:
. . . Taipai mothers goal seems to be to encourage their children to live up to
their own high standards by citing concrete examples of admirable behavior
from their own or other family members lives . . . [T]eaching by example is
one of several traditional Chinese socializing strategies . . . (p. 170)
In contrast, European-American mothers stories about themselves were

more likely to preserve the childrens self-esteem by reassuring them that
something they had done was not really all that bad, and that even admirable persons were prone to error. Therefore, their stories typically conveyed
their own misdeeds as a way of indicating that they were also not perfect
and to err is human, rather than setting themselves up as standards to emulate. In this regard, many European-American parents were prone to tell
hell rising stories about times in their youth in which they had behaved in
willed and irresponsible ways. In contrast, there was not a single instance
of this type of story by Taiwanese parents.
In summary, the stories that European-American parents encouraged
children to tell about themselves were focused on positive events or, when
they concerned misdeeds, tended to treat them as humorous rather than a
potentially serious defect in the childs general personality. Moreover,
American parents stories about themselves were intended to suggest that
they, like their children, were also not perfect. The implications of these
148
CHAPTER 6
stories, which emphasized positive outcomes while minimizing the seriousness of misdeeds, seem likely to induce a general promotion focus (Higgins, 1997), that is, a tendency to attach importance to positive outcomes
with little concern about negative consequences of behavior.
Unlike Americans, Chinese parents encouraged their children to focus
their stories on their misdeeds and to acknowledge responsibility for them.
Moreover, parents stories about themselves established a high standard of
excellence that their children were encouraged to emulate. In combination,
these socialization practices seem likely to induce a prevention focus,
that is, a concern with negative consequences of behavior. These general
differences seem likely to stimulate the development of different implicit
theories about oneself as responsible for negative outcomes and, moreover, in the desire to avoid these outcomes. In fact, this difference exists between European and Asian adults, and can have a profound impact on judgments and behavior (cf. Aaker & Lee, 2001; Briley et al., 2000; see chap. 3).
Cultural Influences on Socialization
The narrative representations that individuals acquire about themselves
and others are transmitted directly through parentchild interaction. As
Miller et al.s research testifies, the character of these individual narratives
can vary over cultures, and also over communities and subcultures within a
society. However, cultural differences can also exist in more general narrative representations that individuals construct. Mankowski and Rappaport
(1995) noted that the norms and values that pervade societies are governed
by a shared cultural narrative that not only serves to guide behavior in specific situations but influences individuals feelings of personal and social
identity. As they indicated:
. . . the phenomena of identity development and change may be understood in
terms of the appropriation of shared narratives into ones personal life story
on the one hand, and the creation of new narratives or modification of existing narratives on the other. (p. 213)
For example, a narrative that pervades American culture is exemplified

by the Horatio Alger myth of individuals who persevere and behave righteously in the face of poverty and adversity ultimately overcome their hardships and become affluent and respected citizens. This narrative implies a
standard of morality and behavior that influences many Americans values
and behavioral decisions. Stories with similar implications surround the
original settlers in America (e.g., the Mayflower stories) and the settlement of the Wild West. (Interestingly, these stories have been modified to
such an extent over the years that they typically bear little resemblance to
the actual events that occurred; see Zinn, 1980.) The life stories of a reli-
149
gious figure can also provide the basis for shared norms and values in societies in which the religion predominates.
The suggestion that individuals in a society share a dominant cultural
narrative does not imply that the entire narrative exists as such in the mind
of a single individual. Moreover, the influence of such a narrative on the
construction of self-narratives may not be completely conscious. Just as
features of the content of childrens self-stories are modified to conform to
the values of their parents or others to whom they are communicating, a
cultural narrative may imply general norms and values that individuals
consider socially desirable and, therefore, influence the content of the stories that individuals construct about themselves, leading some things to be
emphasized and others to be omitted. This may be done without a clear understanding of the roots of these norms and values. Moreover, the functional autonomy of cultural norms and values from their original content is
well established in Boorstins (1973) sociological analysis of the development of American thought. (Thus, e.g., the conviction that individuals have
the right to own firearms is a remnant of a period in the development of the
American West in which law enforcement was either nonexistent or ineffective, and individuals were required to take law into their own hands in order to survive.)
It seems reasonable to suppose that once self-narratives become tailored to cultural norms and values, they are later retrieved without a conscious consideration of the personal experiences on which they are based.
Thus, they ultimately become regarded as true by the storytellers themselves. As a result, they become part of individuals self-identity (McAdams,
1988) that are told to ones children and others independently of the original narrative representations of the events that were formed at the time
they occurred.
CONCLUDING REMARKS
The influence of generalized narrative representations on memory clearly
depends on the nature of the representation and how it is used. When
event prototypes of the sort referred to elsewhere as scripts (Shank &
Abelson, 1977) or event schemas (Graesser, 1981) are brought to bear on
the comprehension of verbal descriptions of events involving fictitious individuals, they can influence the content of the representations formed of
these events and stored in memory. However, these prototypes are much
less likely to come into play in comprehending events that people imagine
themselves or familiar others experiencing.
Generalized representations of single experiences that people form in
the course of telling or writing about these experiences may have a more
150
CHAPTER 6
profound influence on the memory for these experiences. Moreover, to the

extent that these representations are later retrieved and used as a basis for
judgments of oneself and others, they may come to function as reality,
and thus can have a pervasive influence on self-perceptions, personal
norms, and values. We return to this possibility in chapter 10, where we discuss the impact of generalized narrative representations on judgments and
behavior in some detail.
C H A P T E R
7
Pragmatic Influences on the
Interpretation of Statements
Made in a Social Context
Event and episode models are constructed spontaneously in the course of

comprehending information about social events. These processes are automatic. Try not to understand the statement, The boy kicked the ball. For a
native English speaker, this is very difficult. However, the processes involved
in comprehending such a statement, described in chapter 4, are obviously
not sufficient to extract meaning from the information one receives. This is
particularly true when the information is conveyed verbally. Many statements make use of terms with which the recipient is unfamiliar. Others appear anomalous upon first consideration. For example, statements such as
The haystack was important because the cloth would rip can only be assigned meaning in the context of a larger body of knowledge (e.g., a discussion of the hazards of parachuting; cf. Bransford et al., 1972). When such
statements are encountered, they are theoretically transmitted by the
Comprehender to the Work Space with an indication that they cannot be understood, thus stimulating Executor-monitored comprehension processes.
Still other considerations arise when information is conveyed in a social
context. A persons comments to one another are often made with a particular goal in mind: to convey new information, to entertain, to persuade, to
criticize, or to self-promote. Recipients interpretation of these comments
can depend on their perception of the communicators objective. More generally, recipients must not only comprehend a statements literal meaning
but also identify its pragmatic implications. These implications can concern
not only the meaning that the communicator intends to transmit but also
more general feelings, attitudes, and personality characteristics that the
speaker does not always wish to reveal.
151
152
CHAPTER 7
The meaning that communicators intend to transmit and the meaning

that recipients extract from their messages can often differ. For example, a
person who comments, George W. Bushs foreign policy is certainly making America popular overseas might intend to be sarcastic, and to disparage Bushs policy and not to praise it. However, a recipient might not perceive this intention. Alternatively, a professor who describes his graduate
assistant as about as helpful as a well-trained chimpanzee might intend to
convey that the assistants competence leaves something to be desired.
However, a listener might infer that the professor is rather nasty and an unpleasant person to work for. In short, the pragmatic implications of a communication do not always correspond to either the literal meaning of the
message or the meaning that the communicator would like to transmit.
A conceptualization of communication and comprehension in daily life
situations must take into account both (a) the factors that influence communicators attempts to tailor their comments in a way that will attain
their objectives and (b) the factors that govern recipients comprehension
of these communications (see Wyer & Adaval, 2003). In this chapter, however, we focus primarily on the second set of factors. In doing so, we consider two questions. First, what factors stimulate a recipient to think
about the pragmatic implications of a message rather than taking it literally? Second, how does a recipient identify the pragmatic implications of a
communicators message, and how do these implications, once identified,
affect the recipients reactions to it? As we will see, these reactions can be
reflected not only in judgments that are made on the basis of a message
but also the search for additional information that explains why the message was transmitted.
Peoples perceptions of the pragmatic implications of a message can often
be stimulated by their own goals in the situation at hand. A woman who
must decide whether to accept a date might want to know whether the invitation reflects a genuine desire for her company or only a desire to have
sex. More generally, people who wish to form an impression of a person or
product on the basis of a communicators description of the referent may
need to assess the likelihood that this description is biased. They may attempt to identify the pragmatic implications of the communicators message in order to make this assessment.
On the other hand, people often spontaneously recognize that a statements literal meaning is not the meaning the communicator actually intends to convey. This occurs when the statements literal meaning violates
normative expectations for the content and style of communications that
PRAGMATIC INFLUENCES
153
are typically exchanged in the situation at hand. As Grice (1975) first

pointed out and others have since elaborated (e.g., Green, 1989; Higgins,
1981; Sperber & Wilson, 1986), social communications are governed by a set
of implicit principles that are applied by both the communicator in generating a message and the recipient in construing its implications. In informal
situations, for example, speakers are normally expected to be
1. informativeto convey new knowledge that the recipient does not already have and is likely to find interesting;
2. truthfulto convey correct information as the communicator sees it;
3. relevantto say things that are related to the topic under discussion;
4. politeto avoid offending the individuals to whom one is communicating; and
5. modestto avoid unduly promoting oneself.
These normative principles are obviously not universal. A modesty
norm, for example, is applied less often in job interviews than in informal
conversations, and truthfulness is less likely to guide exchanges between a
customer and a used car salesman than a students discussions with a professor. Moreover, certain types of situations may activate specific expectations for communicators to ingratiate, to try to persuade, or to be amusing.12
It seems reasonable to suppose that when people receive a communication whose literal implications appear to deviate from expectancies for the
content and style of messages that are typically conveyed in the situation in
which they occur, they attempt to reconcile this discrepancy. In doing so,
they often reinterpret the message, or the communicators purpose in
transmitting it, in a way that is more consistent with their expectations.
This, in turn, may require the recipients to consider the communication in
relation to other available information in the situation at hand. As a result,
they may have better recall of this information later, and may be more inclined to use it as a basis for judgment.
The assumption that people spontaneously recognize that a communication violates a normative principle is based in part on Postulate 4.6. According to this postulate, many statements are spontaneously recognized as either redundant with prior knowledge (and thus in potential violation of a
norm to be informative) or false (and thus in potential violation of a norm
12
The expectations for communications in a social situation can pertain not only to the content of the messages that are exchanged but the style in which they are transmitted. The extent
to which parties to a conversation elaborate answers to one anothers questions, or ask questions in return, may influence perceptions of the communicators attitudes toward both the conversation and one another (Wyer, Swan, & Gruenfeld, 1995). Nonverbal behaviors can obviously
play a role as well, as we note later in this chapter.
12
154
CHAPTER 7
to convey valid information) in the course of comprehending them. In the

first two sections of this chapter, we discuss the identification of these two
types of norm violations in some detail and describe research bearing on
responses to these violations. The cognitive dynamics that underlie the recognition of other norm violations are discussed in later sections.
Before embarking on this discussion, however, the basic processes we
assume to underlie the identification are worth reviewing. As we noted in
chapter 4, situation models are likely to be formed of the states or events
described in a verbal communication. In addition, features of the situational
context in which a communication is made can activate representations of
previously encountered situations that share these features. This configuration of common features may be responded to schematically, thus constituting a situational appraisal of the sort suggested by Lazarus (1982, 1991).
The situation-specific representations that give rise to this appraisal, however, contain other features that are not common to all of these representations but nevertheless fall within a circumscribed range of values. These
features, which could pertain to characteristics of statements and behaviors that have occurred in the situation, give rise to expectations for the features that are likely to exist in new situations.
Thus, suppose a specific communication (statement or verbal utterance)
is encountered in a given situation. The Comprehender presumably constructs a representation of the statements literal meaning with reference to
previously formed knowledge representations and computes an estimate of
it similarity to these representations (S; see chap. 4). If this similarity is either very high (S > TT ; see Postulate 4.6) or very low (S < TF), the
Comprehender appends a tag that denotes the statement as true (i.e., as
redundant with prior knowledge) or false (i.e., inconsistent with prior
knowledge) and transmits it to the Work Space along with a representation
of the message itself.
When the Executor encounters a tag that denotes the statement as true
or false, it instructs a special-purpose processing unit (e.g., the Encoder/Organizer) to compare the statements similarity index (S) to the range of values that are specified in the situational appraisal and, if it falls within this
range, accepts the statements literal meaning. If, however, the similarity index falls outside this range, the Executor directs the Encoder/Organizer to
construe other interpretations of the statement that are more compatible
with situation-based expectations.
Stripped of the cumbersome terminology imposed by the theory, this
conceptualization simply means that people may spontaneously recognize
that a statement violates normative principles that govern communications
of the sort that are usually transmitted in the situation at hand. When this
occurs, they attempt to determine the reason for this violation. In doing so,
they may reinterpret the statement, or the purpose for which it was con-
155
veyed, in a way that is more consistent with the principle that is violated
that is, whether the statement is interpreted as redundant with prior knowledge or is interpreted as false. We consider these two possibilities in turn.
RESPONSES TO UNINFORMATIVE MESSAGES

Suppose the representation of a statement that is transmitted to the Work
Space is tagged as true (Postulate 4.6) and, therefore, as redundant with
prior knowledge. If the representation is inconsistent with expectations for
communications that are encountered in the situation at hand, it may stimulate attempts to explain its occurrence. This may involve two steps, as
shown in the right half of Fig. 7.1:
1. Recipients consider why the source of the statement might consider it
to be informative. For example, they might speculate that the source is
unaware of their knowledge. Or, they might infer that the source knows
of a reason why the statement might not be true, thus making a reaffirmation of its validity necessary.
2. If recipients conclude that the statement cannot possibly be false, they
may conjecture that it is not intended to convey information about its
referent per se, but rather, is to convey the sources implicit attitude toward the event or state described in the statement.
Thus, for example, suppose a man hears someone comment, George W.
Bush is not a card-carrying member of the Ku Klux Klan. He might never
before have considered this possibility, and might normally regard the
statement as going without saying. Therefore, if the statement is made under conditions in which communications are expected to be informative,
the recipient might speculate that Bush might actually have been accused
of belonging to the Klan, thus making a denial of his membership informative. This inference, however, could plant a seed of doubt in the recipients
mind about Mr. Bushs status that did not exist before the denial was made.
To this extent, the denial might actually increase the recipients belief that
Bush belongs to the organization or, at least is in sympathy with the organizations principles.
On the other hand, suppose people overhear the comment Americans
are not allowed to vote until they reach the age of 18. To Americans, this is
common knowledge and, furthermore, could not possibly be false. Consequently, recipients who consider the statement to be uninformative are unlikely to speculate that it might actually be invalid. Instead, they are likely
to infer that the communicator is intending to convey an implicit attitude
toward the state of affairs described (Americans are not allowed to vote
156
FIG. 7.1. Theoretical flow diagram of processing steps involved in identifying the implications of
statements that are recognized as true or false. The source and recipient of the statement are
denoted S and R, respectively.
157
until they reach the age of 18, and this is certainly a good thing, or Americans are not allowed to vote until they reach the age of 18, whereas citizens
of other countries can vote much earlier.). To this extent, recipients would
not change their belief in the validity of the statement, but instead might
change their attitude toward the state of affairs that the statement describes. Note, however, that these effects should only occur under conditions in which uninformative communications are counternormative. When
the messages are not intended to convey new information, they should theoretically be taken at face value, and attempts to make them informative
should not be evident.
Effects of Uninformative Statements
on Recipients Beliefs
Gruenfeld and Wyer (1992) investigated implications of the preceding analysis. Participants read a series of statements about social events. In some
cases, the statements asserted the validity of target propositions that college students were likely on a priori grounds to believe to be false (e.g.,
The CIA is engaged in illegal drug trafficking, Lyndon Johnson was responsible for the assassination of President John F. Kennedy, etc.). In
other cases, however, the statements denied the validity of these propositions (e.g., The CIA is not engaged in illicit drug trafficking, etc.) and,
therefore, were uninformative. Finally, some participants believed that that
the statements were taken from newspapers (whose purpose is to convey
new information), whereas others believed they were taken from an encyclopedia (whose purpose is to preserve archival knowledge). Participants
first judged the importance of the information conveyed in the statements.
Then, after doing so, they estimated their beliefs in both the target propositions and related ones (e.g., The CIA engages in other illegal activities,
Lyndon Johnson was an enemy of John F. Kennedy, etc.).
Statements that affirmed the validity of the propositions constituted new
information, whereas statements that denied their validity were redundant
with participants prior knowledge about the persons and events in question. However, we expected that when participants were told that the statements came from encyclopedias, they would accept the statements without
further consideration. In contrast, participants who were told that the statements came from newspapers should assume that they were intended to be
informative and, therefore, might infer that there was some reason to suppose that the propositions being denied were in fact true. To this extent,
they might actually increase their belief in these propositions.
This was in fact the case. Table 7.1 shows the difference between beliefs
in propositions under each experimental condition (reported along a 010
scale) and beliefs reported by control participants who were not exposed
158
CHAPTER 7
TABLE 7.1
Effects of Assertions on Beliefs in Target
Propositions and Related Propositions
Assertion Type
Effect on beliefs in target proposition

Newspaper source
Encyclopedia source
Effect on beliefs in related propositions
Newspaper source
Encyclopedia source
Affirmation
Denial
1.16*
1.81*
1.06*
0.48
1.15*
1.62*
1.20*
0.91*
Note. Cell entries refer to differences between beliefs reported after exposure to an assertion and context-free beliefs reported in the absence of the assertion. Differences denoted by asterisks are significantly greater than 0, p < .05. Adapted from Gruenfeld and Wyer (1992).
to either denials or affirmations. Affirmations that were taken from encyclopedias increased participants beliefs that target propositions were true,
whereas denials had little effect. These effects confirm the assumption that
affirmations constituted new information but denials did not. News headlines that affirmed the validity of the target propositions also increased beliefs that the propositions were true. (This increase was somewhat less
than it was when the statements came from encyclopedias, suggesting that
participants perceived newspapers to be relatively less credible.) On the
other hand, denials of the propositions validity that were taken from newspapers also increased participants beliefs that the propositions were true.
In fact, this increase was just as great as the increase produced by affirmations! Thus, participants attempted to interpret denials in a way that made
them informative, leading them to infer that there might be some reason to
suppose that the propositions were true, and this led the statements to
have a boomerang effect.
Effects of Uninformative Statements on Attitudes
As noted earlier, however, some uninformative assertions cannot possibly
be false. In this case, participants attempts to make them informative could
lead them to infer that they are implicit expressions of opinion. To this extent, the assertions might influence participants estimates of the desirability of the state of affairs being denied by the proposition rather than their
beliefs in their validity. Gruenfeld and Wyer (1992) reported evidence of this
as well. For example, exposure to the news headline, Wearing seatbelts is
required by law in Illinois significantly decreased participants opinion
that wearing seatbelts should be required. Similarly, exposure to the state-
159
ment, Many food products are sold in plastic and styrofoam containers
decreased subjects beliefs that there should be restrictions on the packaging of food product, and exposure to Not all college students drive American automobiles increased beliefs that college students should drive these
automobiles.
General Implications
1. Communication and Persuasion
The conditions constructed by Gruenfeld and Wyer (1992) provide examples of innuendo effects of the sort that are used intentionally by propagandists. Their use is particularly common in the political arena. For example,
a candidate might attempt to plant a seed of doubt about an opponents integrity, not by asserting that the opponent has engaged in illegal activities
but rather, by actually denying it. For example, the assertions I personally
have no evidence that my opponent embezzled funds while serving as an
executive of General Motors, or I do not for one minute believe that my
opponent is a card-carrying member of the Ku Klux Klan, which might normally go without saying, could stimulate people to suppose that the opponent if not an embezzler or a Klan member, is nonetheless likely to have engaged in activities that give rise to these suspicions.
A perhaps more familiar instance of the intentional application of the informativeness principle occurs in advertising. Suppose an advertisement
for Brand X asserts, X contains no hydratropine. Consumers may have no
idea what hydratropine is. In an attempt to make the statement informative,
however, they may infer that there is some reason why promoters of Brand
X made the statement. In particular, they may infer that (a) hydratropine is
undesirable and (b) other products have it. Correspondingly, the equally
meaningless assertion, X contains hydratropine might stimulate them to
infer that hydratropine is desirable and that brands other than X do not
have it.
2. Communication of Emotions
Similar effects occur in interpersonal communication. For example, a
woman might comment to her spouse, Its cold in here. If she has rarely
complained about the room temperature in the past, her husband may regard this as new information and take steps to remedy the situation. However, suppose the wifes remark is one of many similar comments she has
made over a period of time. In this case, the husband would consider the
statement to convey little new information about his wifes feelings. Rather,
he might infer that the statement is a veiled criticism of his compulsion to
save energy by keeping the thermostat at 65F. As this example points out,
160
CHAPTER 7
the recognition of a statement as uninformative, and the construal of its intended meaning, often requires a substantial amount of shared knowledge.
Watzlawick, Beavin, and Jackson (1967) noted that people in close relationships often acquire a private meaning system that permits them to convey
thoughts and feelings of which other listeners are unaware. This shared
knowledge presumably has its impact through its mediating influence on
perceptions of one anothers statements as uninformative and the reinterpretation of their literal meaning that results from it.
REACTIONS TO COMMUNICATIONS
THAT ONE BELIEVES TO BE UNTRUE
Statements that are transmitted to the Work Space by the Comprehender
may often contain tags that identify them as false or, at least, inconsistent
with the implications of prior knowledge (Postulate 4.6). These statements,
like statements that are redundant with prior knowledge, are likely to stimulate cognitive activity in an attempt to understand why they were made.
These processes may be similar to those that underlie responses to violations of the informativeness principle. However, the inferences that result
from the activity are somewhat different.
The processes that seem likely to underlie reactions to untrue communications can be summarized with reference to the left half of Fig. 7.1. Specifically, individuals who hear a statement they recognize as false (S < TF , as
defined in Postulate 4.6) are likely to assess whether the communicator
knows that this is the case. If the source is likely to believe it is true, the recipients may take the statements literal implications at face value. They
might also do so if they think that the source knows the statement is false
but is unaware that they share this knowledge. In this case, they might conclude that the source is trying to deceive them. However, suppose recipients believe that the source is aware of their knowledge. Then, they are
likely to infer that the source does not mean the statement to be taken literally, but rather, intends it to be ironic.
Reactions to ironic statements can depend on both their topic and
whether their literal implications are favorable or unfavorable. In some
cases, ironic statements are intended to be funny. In other cases, however,
they can be indirect expressions of hostility. These motives are sometimes
difficult to distinguish. Suppose a woman whose husband has had three
traffic accidents in the past 2 years comments to her guests, My husband
is a wonderful driver. The comment could be a true expression of opinion,
and an attempt to defend her spouses abilities despite his poor driving record. To the extent it is intended to be sarcastic, however, it could either
convey hostility or be meant as a tease. Which interpretation the guests
161
make is likely to depend on their view of the womans more general feelings
toward her spouse. If they believe that the couples relationship is basically
harmonious, they are likely to infer that the woman is teasing. If the couple
is known to have a history of marital conflict, however, they might interpret
the statement as hostile.
These observations are rather self-evident. The conceptualization outlined in Fig. 7.1b has greater interest, however, when a statements validity
is clearly a matter of opinion. Suppose a person comments, What this
country needs is another Ronald Reagan. This comment is more likely to
be ironic if it is made by a liberal Democrat than if it is made by a conservative Republican. On the other hand, conservative Republicans are less
likely than liberal Democrats to interpret the statement as ironic. That is,
they are more likely to believe that the statement is true, and to accept its
literal meaning as the intended meaning regardless of its source.
Accuracy Violations and Amusement:
The Identification of Witticisms
To reiterate, people often interpret a statement spontaneously as false in
the course of comprehending it (Postulate 4.6). When it is obvious to recipients that the source also believes the statement to be false and, therefore
interpret the statement as ironic, they are often amused. This was demonstrated empirically by Isbell, Wyer, and Collins (2002). Participants were exposed to a number of vignettes in which one person made a comment to another. In some cases, the comment was likely to be true. In other cases, the
comment was literally false, but if interpreted as ironic, was similar in
meaning to the first statement. Finally, in some cases, the intended meaning
of the statement was more favorable than its literal meaning, and in other
cases, it was less so. For example, one vignette described a discussion between two students about where they had spent the summer. In some versions, one student made a true statement that conveyed either a favorable
description of its referent (Switzerland is certainly a great place to spend
the summerall those lakes and high mountains) or an unfavorable description (Champaign, Illinois, is certainly an awful place to spend the summerall that corn and high humidity). In other cases, however, the students statement was literally false (Switzerland is certainly an awful place
to spend the summer . . . , and Champaign, Illinois is certainly a great
place to spend the summer . . .), but its intended meaning was similar to
that of the corresponding true statement. In another scenario, two professors discussed who was the most (or least) intelligent public figure in the
20th century. In this case, the literally true statements were Einstein gets
my vote (for most intelligent) and Dan Quayle gets my vote (for least intelligent) and the corresponding false statements were Einstein gets my
162
CHAPTER 7
vote (for least intelligent) and Dan Quayle gets my vote (for most intelligent).
Participants read one version of each vignette with instructions to estimate how amused they would be if they overheard the statement being
made under the conditions described. Pooled over vignettes, participants
regarded ironic statements as more amusing than literal ones (4.9 vs. 2.7),
and this was true regardless of whether the intended meaning of the statements was favorable (4.6 vs. 1.9) or unfavorable (6.3 vs. 3.5).
It would obviously be incorrect to conclude that all ironic statements are
amusing. Sarcasm can often not be amusing at all. A person who has just
been turned in by a classmate for cheating on an exam might respond to
the other, You are certainly a fine friend, which is intended to convey
quite the opposite. This comment is unlikely to elicit amusement. The necessary conditions for humor elicitation are discussed in chapter 8 (see also
Apter, 1982; Wyer & Collins, 1992). Nevertheless, the evidence that ironic
statements are more likely to elicit amusement than their true counterparts
is worth noting.
Effects of Invalid Communications
on Information Search
The sources of the ironic statements presented in Isbell and Wyers study
obviously believed that their statements were false, and so recipients were
likely to believe that the statements were intended to amuse. In many instances, however, people are uncertain of a sources beliefs in the statements that he or she makes. This is particularly true when the statements
express opinions that are somewhat controversial. People who encounter
statements of opinion that they personally believe to be invalid may often
attempt to seek additional information that helps them to understand why
the statements were made. In addition, they may review their prior knowledge about the issues to which the statements pertain in order to confirm
their belief that the opinion expressed is incorrect.
This cognitive activity is likely to be reflected in recipients ability to recall the information later. Information that is thought about more extensively with reference to prior knowledge is generally easier to recall (Craik
& Lockhart, 1972; Wyer & Hartwick, 1980). If this is so, however, people
should be more likely to recall opinion statements they spontaneously recognize as false than either statements they consider to be true or statements whose validity they have not spontaneously evaluated.
This possibility has not been investigated under conditions in which information is conveyed in conversations. If the proposed conceptualization
is valid, however, the cognitive activities that underlie peoples recall of
opinion statements they recognize as false should occur even when the
163
statements are conveyed out of their social context. Several studies bear on
this possibility. In a study of impression formation reported by Wyer, Lambert, Budesheim, and Gruenfeld (1992), participants received information
about a target person who was characterized as either a conservative Republican or a liberal Democrat and, in addition, was described by either favorable or unfavorable trait adjectives (e.g., kind, hostile, etc.). This
general description was followed by a list of both (a) behaviors that were
both evaluatively consistent and evaluatively inconsistent with the trait descriptions (e.g., invited a charity collector for lunch, shouted and honked
at slow drivers, etc.), and (b) opinion statements that reflected either a liberal or conservative political orientation. Participants after receiving the information were asked to recall the information they had read.
Participants were more likely to recall behaviors that were consistent
with the target persons initial trait description than behaviors that were inconsistent with it, replicating the results of many other studies (Hastie &
Kumar, 1979; for a review, see Srull & Wyer, 1989). However, they were more
likely to recall opinions with which they disagreed than opinions with
which they agreed, and this was true regardless of the consistency of these
opinions with the targets political ideology. Thus, when participants encountered opinion statements about matters with which they were familiar,
they comprehended these statements with reference to their previously acquired knowledge, and devoted a particular amount of time trying to understand the basis for opinions they regarded as incorrect. Therefore, they
had better recall of these opinions regardless of the opinions consistency
with the general attitudes of the individual who expressed them.13
1. Individual Differences in the Processing
of Invalid Information
People obviously differ in the amount of cognitive effort they perform in
an attempt to reconcile information they consider to be invalid. Wnke and
Wyer (1996) speculated that differences in this motivation might vary with
participants own social and political ideology. Participants with a liberal
social and political orientation are inclined to consider the underlying
causes of a persons behavior when evaluating a person (Fletcher, Danilovics, Fernandez, Peterson, & Reeder, 1986). Conservatives, on the other
hand, are more disposed to evaluate a persons behavior in terms of its
13
Wnke and Wyer (1994) found that participants had better recall of behaviors that were
ideologically inconsistent with the liberal or conservative orientation of the actor. However, this
occurred only when the actors ideology was different from participants own (i.e., when the actor was an outgroup member). When the actors ideology was the same as participants (i.e.,
the target was an ingroup member), participants recall of the targets behavior was similar regardless of whether it was inconsistent or consistent with his ideology (.42 vs. .48, respectively).
13
164
CHAPTER 7
consequences. Note that this difference corresponds to different levels of

moral development, as defined by Kohlberg (1976). (For other evidence that
liberals and conservatives differ in this regard, see Emler, Renwick, &
Malone, 1983; Fishkin, Keniston, & MacKinnon, 1973; Nassi, Abramowitz, &
Youmans, 1983.)
Based on this evidence, we expected that liberals would be more inclined than conservatives to consider the attitudes that underlie a persons
behavior when forming an impression of the person. On the other hand,
conservatives should be more disposed to evaluate behaviors in terms of
their desirability without considering the attitudes and opinions they reflect. Furthermore, these individual differences should be particularly evident under conditions that emphasize the use of these different judgmental
criteria.
To evaluate this possibility, participants who had been identified as having a liberal or conservative political orientation were exposed to information about a target person who was initially described as either liberal or
conservative and who had ostensibly engaged in a number of behaviors
that were normatively either liberal (e.g., organized a sit-in at a nuclear
power plant, raised money for a public housing project, etc.) or conservative (tried to prevent a theatre from showing an antireligious movie,
donated $1,000 to the National Rifle Association, etc.). In behavior-focus
conditions, participants were told that we were specifically interested in
how people form impressions of others on the basis of their behavior. In
opinion-focus conditions, they were told that we were interested in how people form impressions of persons on the basis of their opinions as reflected
in their behavior. Participants after receiving the information and evaluating the target person were asked to recall the behaviors they had read and
finally, were asked to evaluate the desirability of the behaviors that were
described in the information they had received.
Table 7.2 shows the proportion of behaviors that liberals and conservatives recalled as a function of their perceptions of the behaviors desirability and the criterion they were asked to use in forming their impressions.
When liberals were told to base their impressions on the target persons
opinions, they recalled a greater proportion of behaviors they considered
to be undesirable than behaviors they considered to be desirable (.64 vs.
.37, respectively). This was also true of conservatives who were told to
base their judgments on behaviors (.66 vs. .35, respectively). However,
this difference was not apparent when liberals were told to focus on the
targets behaviors per se, or when conservatives were told to consider the
opinions underlying these behaviors. In short, participants only thought
more extensively about behaviors whose implications were considered
undesirable when they were told to employ the criterion that they were
likely to use spontaneously to evaluate these behaviors. When they were
165
TABLE 7.2
Mean Proportion of Behaviors Recalled as a Function of Perceived
Desirability, Participants Ideology, and Instructional Conditions
Opinion-Focus Conditions
Desirable behaviors
Undesirable behaviors
Difference
Behavior-Focus Conditions
Liberal
Participants
Conservative
Participants
Liberal
Participants
Conservative
Participants
.37
.64
.27
.39
.33
.06
.46
.54
.08
.35
.66
.31
Note. Adapted from Wnke and Wyer (1996).
told to use a criterion that differed from the one they typically employed,
this spontaneous cognitive activity was disruptive, and so recall differences were not evident.
PRAGMATIC INFLUENCES ON IMPRESSION

FORMATION IN INFORMAL CONVERSATIONS
The pragmatic implications of communications are particularly likely to
come into play when the communications are exchanged in informal conversations. People may spontaneously identify the pragmatic implications
of a persons comments when they perceive the comments to be either uninformative or untrue. However, their identification of these implications
can also be motivated by more specific goals they have at the time they engage in the conversation. People who observe or take part in a social interaction are often disposed to form general impressions of the interaction
participants. In some cases, they are likely to form these impressions spontaneously, without any more specific goal in mind other than to get to know
the individuals better. In other cases, they might do so intentionally in order to interpret the individuals comments about the topic at hand. For example, a persons disparaging remarks about another could convey valid information about the individual being described. On the other hand, they
could also indicate that the speaker is an unpleasant individual who criticizes others indiscriminately regardless of their merits. Similarly, a person
who expresses a favorable opinion of a new television program either could
have discriminating tastes or could like almost anything, including mindless
situation comedies that most people would find dull and offensive. In these
circumstances, listeners are likely to assess a speakers general personality
characteristics in order to construe the implications of his or her remarks.
These possibilities have largely been ignored in much of the research on
the cognitive underpinnings of person impressions (for reviews, see Ander-
166
CHAPTER 7
son, 1981; Carlston & Smith, 1996; Wyer & Carlston, 1994). In this research,
participants are usually asked to form an impression of someone on the basis of trait and behavior descriptions that are conveyed out of their social
context. In doing so, they are expected to take the information at face value,
independently of its source. To this extent, the implications of this research
for person impression formation outside the laboratory may be limited.
The research described in the remainder of this chapter calls attention
to several of these limitations. To provide a perspective on this research,
we first review briefly the conclusions that have been drawn in the typical
research paradigm in which person impression formation has been investigated. We then describe the results of a number of studies in which similar
information is conveyed in the context of informal conversations under
conditions in which the pragmatic implications of the information is taken
into account. Finally, we describe the role of nonverbal and paralinguistic
features of the communications exchanged in a conversation that influence
recipients perceptions of their implications and, therefore, the impressions
they form of both the topic being discussed and the persons who are discussing it.
Impression Formation in Nonsocial Contexts
Extensive research has been conducted on the impressions that people
form of someone on the basis of information that describes the persons
traits and behaviors, and rigorous theories have been proposed to describe
the processes that underlie these impressions (Carlston & Smith, 1996;
Wyer & Srull, 1989). In a typical study (cf. Srull, 1981; Wyer, Bodenhausen, &
Srull, 1984; for a review, see Srull & Wyer, 1989), participants are asked to
form an impression of a target person on the basis of a favorable or unfavorable trait description followed by a number of behaviors. Some behaviors are descriptively and evaluatively consistent with the initial trait description of the person and others are inconsistent with this description.
The target person is typically fictitious, and the trait description is conveyed in a way that suggests that its validity is not open to question. (For
example, the person might ostensibly be a character in a novel, and the
trait adjectives might have been used by the author to describe him; see
Wyer & Martin, 1986.) The mental representation that is formed under these
conditions, which presumably exemplifies a generalized person representation of the sort described in chapter 1, has been conceptualized using an associative network metaphor (Hastie & Kumar, 1979; Srull & Wyer, 1989).
That is, the concepts formed from information are represented by nodes,
and associations between them (formed by thinking about the concepts in
relation to one another) are represented by pathways. Stronger associa-
167
tions, resulting from more extensive thought, are designated by wider

paths. The accumulated evidence (Srull & Wyer, 1989) suggests the following process.14
1. People who receive an initial trait description of a target person form
an evaluative concept of him on the basis of this description. If they later
learn about specific behaviors the target has performed, they encode the behaviors as either favorable or unfavorable and think about their implications
for the concept they formed of him earlier. This establishes associations between the behaviors and this concept. Thus, suppose a person is described
by a set of favorable traits, followed by the following sequence of favorable
and unfavorable behaviors: b+, b+, b-, b+, b-, b-, b+. Then, a favorable concept of the person, P+, is formed on the basis of the traits, and the behaviors
become associated with it as shown in Fig. 7.2.
2. When a behavior is encountered that is evaluatively inconsistent with
the central person concept, people attempt to understand why the behavior
occurred. In doing so, they think about the behavior with reference to other
behaviors the person has performed, establishing associations between the
inconsistent behavior and others. Assuming that each inconsistent behavior
(b-) in our example is thought about with reference to the two behaviors
that precede it in the sequence, the interbehavior associations would be
those shown in Fig. 7.2.
3. In addition, people who encounter behaviors that are inconsistent with
their general concept of the target attempt to bolster their confidence in the
validity of this concept. They do this by reviewing behaviors that are consistent with it, strengthening the association of these behaviors with the concept, as also shown in the figure.
Numerous studies provide support for this conceptualization and its implications. For example, note that as a result of peoples attempts to reconcile the occurrence of a behavior that is inconsistent with their initial impression of the person, the behavior becomes more interconnected to
other behaviors than the consistent behaviors are. To this extent, the behaviors may be more easily recalled. Bolstering, however, strengthens the
14
Srull and Wyer (1989; see also Gordon & Wyer, 1987) postulated that in addition to the processes described here, people interpret the behaviors they read about in terms of the traits they
exemplify, leading trait-behavior clusters to be formed and stored separately from the more
general person representation. The role of these clusters is of considerable importance in conceptualizing a number of phenomena, including the mental representation information about
commercial products (Park & Wyer, 1993) and the impact of information one is told to disregard
(Wyer & Budesheim, 1987; Wyer & Unverzagt, 1985). However, these representations are not
central to the issues of concern in this chapter, so they are not described in detail.
14
168
CHAPTER 7
FIG. 7.2. Metaphorical representation formed of a target person described by

a set of favorable traits and a series of behaviors that are evaluatively consistent (b+) or inconsistent (b-) with the person concept formed from these
traits. Thicker lines denote stronger associations between the elements involved. Each inconsistent behavior is thought about in relation to the two behaviors that precede it in the sequence.
association of evaluatively consistent behaviors with the central concept,

making these behaviors more accessible in memory. Thus, the influence of
inconsistency resolution and bolstering should be reflected by the ease of
recalling of inconsistent and consistent behaviors, respectively.
Many studies (e.g., Hastie & Kumar, 1979; Srull, 1981; Srull, Lichtenstein,
& Rothbart, 1985; Wyer, Bodenhausen, & Srull, 1984) show that inconsistent
behaviors are recalled better than consistent ones, suggesting that inconsistency resolution takes priority over bolstering. (See Abelson, 1959, for a
similar assumption in conceptualizing the processes of resolving inconsistencies among beliefs and attitudes.) However, when people are given more
time to think about the information they receive, either as it is presented or
subsequently, bolstering may predominate, leading consistent behavior to
gain a recall advantage (Wyer, Budesheim, Lambert, & Martin, 1989; Wyer &
Srull, 1989; see also Wyer & Martin, 1986).
Effects of Social Context on Impression
Formation Processes
The person memory model developed by Srull and Wyer (1989) has been
applied successfully in conceptualizing numerous social phenomena, including the impact of information that one is told to disregard (Wyer &
169
Budesheim, 1987; Wyer & Unverzagt, 1985), the combined influence of stereotypes and idiosyncratic trait descriptions on person impressions (Wyer
& Martin, 1986), and the different impressions that are formed of persons
and groups (Srull, 1981; Wyer & Gordon, 1982). Nevertheless, the processes
implied by the theory are restricted to conditions in which information is
presented out of its social context. Quite different considerations arise
when a persons traits and behaviors are described in the context of a conversation. For one thing, an acquaintances trait descriptions of someone in
the course of a conversation are matters of opinion and, as such, may not
be valid. Consequently, listeners may attempt to establish the validity of
these descriptions. To do this, they may construe the implications of the
descriptions for characteristics of the speakers who provide them as well
as for the individual being described.
Wyer, Budesheim, and Lambert (1990) examined this possibility. In their
studies, the trait and behavior descriptions that participants received were
very similar to those presented in more traditional impression formation
research (e.g., Wyer, Bodenhausen, & Srull, 1984). However, the information
ostensibly pertained to an actual person and was provided by acquaintances of the person in the course of a conversation. Specifically, participants were told they would listen to a tape-recorded conversation between
a male and a female student about a mutual acquaintance. The tape began
with instructions to the two speakers to select the person they wanted to
talk about and then, after doing so, to write down three general attributes
they would use to describe this person. At this point, the tape stopped, and
the listeners were given photocopies of the trait adjectives the speakers
had had ostensibly written down. The favorableness of each speakers trait
description of the target and the dimension to which this description pertained (intelligence or kindness) were varied independently.
After participants had read the target descriptions, the taped conversation was restarted, and participants heard the experimenter tell the speakers
to reminisce about things the target person had done that they had either
seen or heard about. The speakers then proceeded to exchange anecdotes,
interspersed with occasional pauses and promptings by the experimenter to
make the conversation seem natural. Over the course of the conversation,
each speaker mentioned 13 behaviors of which 6 were favorable (either kind
or intelligent), 6 were unfavorable (unkind or unintelligent) and 1 (occurring
near the beginning of the conversation) was neutral. These behaviors were
very similar to those presented in studies by Wyer et al. (1984) and Wyer and
Martin (1986), but were conveyed as they might be mentioned in a normal
conversation. An excerpt of one conversation, about a target named Don,
was as follows (for a complete transcript, see Wyer et al. (1990):
Experimenter: Okay, then, why dont we begin. Who wants to start off?
170
CHAPTER 7
M: Well, lets see . . . well, he walks a mile to school each day, even when
its freezing. I guess I thought of that because its so damn cold today.
I asked him if he wanted a ride, but he said he didnt.
F: Yeah, he never seems to get sick. It must be because he works out 3
or 4 times a weekhes in pretty good shape.
M: [Laughs] Better than me, anyway. . . . I remember he won the university chess championship last fall.
F: [Laughs] Yeah. . . . I remember we went out to dinner that night to
celebrate and Don swore at the waiter at Eddies because he didnt
get served as quickly as he wanted.
M: Yeah, I can just see him doing that . . . lets see . . . well, he cant seem
to follow simple directions people give him. Remember that time last
week at Jerrys?
F: Yeah, [Laughs] . . . Oh, by the way, do you know Jerry lost his job at
the Union?
M: No, really? Gee, thats rough.
F: Yeah. I thought of that cause he told me Don had loaned him 20
bucks for a date he had and stuff.
M: [Pause] Jeez, this is a lot harder than I thought. [To experimenter]
You want specific behaviors, right? Like I know hes always smiling
and saying hello to people when I walk down the street. But one particular instance doesnt stand out.
Experimenter: Well, try to be as specific as you can.
M: [Pause] Hmm . . . well, lets see . . . well, someone said hed flunked a
mechanics training course I took last summerI dont know why he
took it, anyway.
The other tape was identical except that the roles of the male and female
speakers were reversed.
We anticipated that people who listened to the conversation would spontaneously form impressions of the speakers as well as the target himself. To
understand the nature of these different impressions, we employed two instructional conditions. In speaker-impression conditions, participants were
explicitly told to form impressions of the speakers, and in target-impression
conditions, they were told to form an impression of the person the speakers
were discussing. In both cases, however, participants after listening to the
tape reported their liking for the target person along a -5 to +5 scale. In addition, they reported their perception of how well each of the speakers
liked the target and rated their own liking for the speakers. Finally, they recalled the behaviors they had heard the speakers mention.
1.
171
Speaker-Impression Conditions
Judgment Data. Participants evaluated each speaker more favorably

when the speakers trait descriptions of the target were favorable than
when they were unfavorable. Moreover, they inferred that the speakers
liked the target to a greater extent in the former conditions. However, the
favorableness of the speakers trait descriptions had contrast effects on
participants own liking for the target. That is, partners liked the target less
when both speakers trait descriptions were favorable than when both descriptions were unfavorable. In combination, these results suggest that participants based their impressions of the speakers on the speakers trait description of the target and then used these impressions as standards of
comparison in judging their liking for the target himself, independently of
the implications of the trait descriptions for his attractiveness.
Recall Data. The recall data were particularly provocative. The proportion of behaviors recalled are shown in the top section of Table 7.3 as a
function of the speaker who mentioned the behaviors and the consistency
of the behaviors with each speakers trait description. When the two speakers descriptions of the target were evaluatively similar, participants had
better recall of behaviors that were inconsistent with these descriptions (M
= .444) than behaviors that were consistent with them (M = .391). When the
two speakers descriptions differed in favorableness, however, participants
had substantially better recall of the behaviors mentioned by a given
speaker that were inconsistent with the trait description of the target by
the other speaker than behaviors that were consistent with this description
(.520 vs. .416). Thus, both sets of data suggest that participants had better
recall of behaviors mentioned by one speaker if they were inconsistent with
the trait description provided by the other speaker than if they were consistent with this description (.482 vs. .401).
Although these data are confusing on first consideration, their interpretation is actually quite clear. That is, participants apparently used the
speakers trait descriptions to form impressions of the speakers, and then
used speakers descriptions of the targets behaviors to confirm the validity
of these impressions. Note that if a behavior mentioned by one speaker is
evaluatively inconsistent with the trait description of the target provided
by the other speaker, this suggests that the description is not a valid characterization of the target but rather, reflects a general disposition of the
speaker to describe people favorable or unfavorably. Thus, these behaviors
confirm the impression that participants formed of the speaker on the basis
of his or her description of the target. Participants who sought to confirm
this impression thought more about these behaviors and, therefore, recalled the behaviors better.
172
CHAPTER 7
TABLE 7.3
Proportion of Behaviors Recalled as a Function of the Speaker Who
Mentioned Them and the Behaviors Evaluative Consistency
With Each Speakers Trait Description of the Target
A. Speaker-impression conditions
Behaviors mentioned by male
Consistent with males trait description
of target
Inconsistent with males trait description of target
M
Behaviors mentioned by female
of target
M
B. Target-impression conditions
Behaviors mentioned by male
of target
M
Behaviors mentioned by female
of target
M
Consistent With
Females Trait
Description of
Target
Inconsistent With
Females Trait
Description of
Target
.430
.514
.412
.403
.416
.438
.476
.421
.351
.430
.390
.527
.439
.451
.441
.489
.396
.437
.416
.422
.409
.424
.430
.423
.437
.528
.482
.409
.423
.472
.500
.441
Note. In each quadrant of the table, diagonal cells are those in which the two speakers trait
descriptions of the target are evaluatively similar, and off-diagonal cells are those in which the
descriptions are evaluatively dissimilar. Adapted from Wyer, Budesheim, and Lambert (1992).
Thus, suppose the female speaker described the target in terms of favorable traits and the male described him in terms of unfavorable traits.
Then, the mental representation that presumably results from the cognitive activity described in the preceding paragraph can be captured in an
associative network of the form shown in Fig. 7.3a, where F+ and M- refer
to concepts of the male and female speakers, respectively, and the favorable and unfavorable behaviors mentioned by the speakers are denoted
b+ and b-, respectively.
173
2. Target-Impression Conditions
In these conditions, participants were told to form an impression of the
target rather than of the speakers. Despite these instructions, the results
were very similar to those observed in speaker-impression conditions. For
example, the favorableness of the speakers initial trait descriptions of the
target influenced both participants perceptions that the speakers liked the
target (1.58 vs. -0.93, when both speakers trait descriptions were favorable
vs. unfavorable, respectively) and also participants liking for the speakers
(1.04 vs. -0.42, respectively). However, in contrast to earlier studies in
which trait descriptions of the target provided the primary basis for evaluations of him (Srull & Wyer, 1989), participants liked the target less when
both speakers described him favorably (M = -0.33) than when they did not
(M = 0.75). Thus, these data, like those obtained in speaker-impression conditions, indicate that participants used the speakers trait descriptions of
the target to form impressions of the speakers themselves, and then, having done so, used their impressions of the speakers as comparative standards in evaluating the target without considering the favorableness of the
speakers trait descriptions for characteristics of the target himself.
The recall data in this study were also similar to those observed in
speaker-impression conditions except that in this case, participants were
primarily disposed to confirm their impression of the female speaker rather
than the male. These data are summarized in the bottom half of Table 7.3.
When both speakers trait descriptions of the target were similar in favorableness, participants had somewhat better recall of behaviors that were
evaluatively inconsistent with these descriptions (M = .448) than behaviors
that were consistent with them (M = .417). When the two speakers descriptions differed in favorableness, participants were more inclined to recall behaviors that were inconsistent with the female speakers description of the
target (M = .482) than behaviors that were consistent with these descriptions (M = .416), and this was true regardless of whether the behaviors were
mentioned by the male or the female. Thus, both sets of data could suggest
that participants organized the information around an evaluative concept
of the female speaker. Thus, they thought more extensively about behaviors she had personally mentioned that were inconsistent with this concept
in an attempt to reconcile their occurrence. On the other hand, they
thought more about behaviors the other speaker mentioned that were inconsistent with this concept and, therefore, bolstered their perception that
the females trait description of the target reflected a general disposition to
judge people favorably or unfavorably independently of the targets actual
attributes. The representation that participants may have formed under
this condition, therefore, would resemble that shown in Fig. 7.3b.
174
CHAPTER 7
FIG. 7.3. Representations formed under conditions in which participants (a)

were told to form impressions of speakers and (b) were told to form impressions of the target. F+ and M- denote favorable and unfavorable concepts of
the female and male speakers, respectively, and b+ and b- denote favorable
and unfavorable descriptions of the targets behaviors, respectively.
3. Summary
Considered in combination, the results obtained under speaker-impression and target-impression conditions suggest that participants who overhear a conversation about a person are inclined to form impressions of the
speakers, and that this is true regardless of whether they are explicitly told
to do so or are asked to form an impression of the person the speakers are
discussing. Moreover, they tend to use the speakers descriptions of the targets behavior to confirm these impressions. Then, once these impressions
are formed, participants use them as comparative standards in evaluating
the person the speakers are discussing rather than basing their impressions on the literal implications of the speakers descriptions of him.
175
The Influence of Modesty and Politeness Norms

on Impression-Related Information Processing
1. General Considerations
Participants in informal conversations are expected not to offend the
people to whom they are communicating. They are also expected to be
modest and not to toot their own horns unnecessarily. Consequently, people who hear someone make a statement that appears to violate one of
these norms may attempt to reinterpret it in a way that is more consistent
with their normative expectations. These processes may be similar to those
that result from violations of informativeness and truthfulness norms.
For example, the interpretation of a statement as hostile or a tease may
depend in part on whether the statement violates a norm to be polite. This,
in turn, may depend on whether the target of the statement is present at
the time the statement is made. Thus, for example, a persons comment
that a colleague has trouble balancing his checkbook is likely to be seen as
a tease if the comment is directed to the colleague himself. In the colleagues absence, however, a politeness norm might not be applicable, and
so the statement is more likely to be interpreted as antagonistic.
However, the effects of politeness norms on comprehension are not independent of its effects on truthfulness. Shortly after the publication of my
1974 book on cognitive organization and change, a friend remarked, Congratulations on your book. Im looking forward to the publication of the
English language version.15 Because I interpreted the statement as a tease
rather than an expression of hostility, I found it amusing. However, this reaction was probably diluted by my lack of confidence in my writing skills
and, therefore, my belief that the statements implications might be valid. If
I had had more confidence in my writing ability, I might have construed the
implications of the statement to be definitely false. I might then have appreciated the irony much more.
A related implication of this analysis is that statements are more likely to
be interpreted as teases if their literal implications are extremely disparaging than if they are less so. This is because they are more likely to be seen
as invalid. I may be more disturbed by a mildly negative comment about my
writing ability than by a comment that my writing is slightly superior to that
of a chimpanzee. This is because I am inclined to believe that the first remark is valid and thus to take it literally, whereas I perceive that the second
statement is clearly invalid and therefore, interpret it as a tease.
15
15
Appreciation is extended to John McCarty for this example.
176
CHAPTER 7
2. Effects of Politeness and Modesty Norms

on Reactions to Information
The conceptualization we propose implies that people who perceive that
a communication violates a norm-based expectation may think about the
communication more extensively in an attempt to understand the reason it
occurred. The fact that people have better recall of opinion statements they
regard as invalid (Wyer et al., 1992) provides indirect evidence of this more
extensive processing when information violates a norm of accuracy. However, similar effects should be evident when other norms are violated. A series of studies by Wyer, Budesheim, Lambert, and Swan (1994) confirmed
this possibility.
Participants listened to a tape-recorded conversation that was ostensibly between two male student volunteers who knew each other well. In contrast to the study by Wyer et al. (1990), however, this conversation was
about one of the participants rather than a third party. Over the course of
the conversation, one person (O) described both favorable and unfavorable
things that the other, target person (T) had performed, whereas the target
conveyed both favorable and unfavorable things that he personally had
done. Some participants were asked to form an impression of O on the basis of Os description of the target (other-impression, other-focus conditions);
others were told to form an impression of the target based on Os description of him (target-impression, other-focus conditions), and still others were
asked to form an impression of the target on the basis of the targets descriptions of himself (target-impression, target-focus conditions).
The tape was constructed in much the same way as the one employed
by Wyer et al. (1990). That is, it began with the experimenter giving instructions to the two conversation partners (referred to in the tape as Don
and Bob). After deciding which of the persons would be the topic of discussion, O was asked to write down three trait adjectives describing the
target, and T was asked to write down three adjectives describing himself.
The tape was then temporarily interrupted, and participants were given
either a favorable or unfavorable trait description of the target that was
ostensibly prepared by the person on whom they were told to focus their
attention. (Thus, they were given the targets trait description of himself
under target-impression, target-focus condition, but were given Os description of the target in the other two conditions.) The tape was then restarted and the conversation was played in which the two participants exchanged anecdotes about Ts behaviors. These behaviors were identical
to those described by the two speakers in Wyer et al. (1990) with minor
variations. Thus, an excerpt analogous to the one employed in the earlier
study was as follows, where T and O refer to the target and the other
speaker, respectively:
177
Experimenter: Okay, then, why dont we begin. Why dont you start, Don?
O: Well, lets see . . . well, you walk a mile to school each day, even when
its freezing. I guess I thought of that because its so damn cold today.
Every time I offer you a ride, you always say no.
T: Yeah, I never seem to get sick. It must be because I work out 3 or 4
times a weekIm in pretty good shape.
O: [Laughs] Better than me, anyway. . . . I remember you won the university chess championship last fall.
T: [Laughs] Yeah . . . I remember we went out to dinner that night to celebrate and I swore at the waiter at Eddies because I didnt get served
as quickly as I wanted.
O: Yeah . . . I remember after we left, and we were supposed to go to
that party at Jerrys, you were trying to follow the directions hed
given you to get to his house and you got us totally lost! . . . Oh, by
the way, do you know Jerry lost his job at the Union?
T: No, really? Gee, thats rough.
O: Yeah. I thought of that cause he told me you had loaned him 20
bucks for a date he had and stuff.
T: [Pause] Jeez, this is a lot harder than I thought. [To experimenter]
You want specific behaviors, right? Like I know Im always smiling
and saying hello to people when I walk down the street. But one particular instance doesnt stand out.
Experimenter: Well, try to be as specific as you can.
T: [Pause] Hmm . . . well, lets see . . . well, I flunked a mechanics training course I took last summerI dont know why I took it, anyway.
Participants after hearing the conversation were asked to indicate how well
they liked the two speakers, and then were asked to recall as many of the
behaviors as they could.
We expected that participants would have better recall of statements
that O made that violated a politeness norm, but would have better recall of
statements the target mentioned that violated a modesty norm. This was in
fact the case. Table 7.4 shows the mean proportion of favorable and unfavorable behaviors that participants recalled in each instructional condition
as a function of the person who mentioned them. Participants had better recall of behaviors that O mentioned if they were unfavorable than if they
were favorable (.49 vs. .39, respectively), but had better recall of behaviors
the target himself mentioned if they were favorable than if they were unfavorable (.45 vs. .38, respectively). The first difference was more pronounced
when participants were explicitly told to focus their attention on the behaviors that O mentioned, and the second was more pronounced when partici-
178
CHAPTER 7
TABLE 7.4
Proportions of Favorable and Unfavorable Behaviors Recalled as a
Function of Task Objectives and the Person Who Mentioned Them
Task Objectives
Behaviors mentioned by
other (O)
Favorable
Unfavorable
Behaviors mentioned by
target (T)
Favorable
Unfavorable
Target Impression,
Target Focus
Target Impression,
Other Focus
Other Impression,
Other Focus
.42
.45
.40
.54
.35
.49
.47
.45
.50
.54
.37
.34
Note. Adapted from Wyer, Budesheim, Lambert, and Swan (1994).
pants were told to focus their attention on behaviors that T mentioned.

However, the differences were evident in all three instructional conditions
and did not depend on the favorableness of the initial trait descriptions that
the speakers provided. In other words, participants spontaneously thought
more extensively about comments that violated norms to be polite and
modest. Furthermore, this was true even when the comments were not directly relevant to the impression formation objectives that participants
were pursuing, and regardless of whether the statements were consistent
or inconsistent with more general trait descriptions of the target.
THE EFFECT OF NONVERBAL BEHAVIOR

AND CONVERSATIONAL STYLE
ON IMPRESSION FORMATION, JUDGMENT,
AND INFORMATION SEEKING
The studies by Wyer et al. (1990, 1994) provide evidence that people who
hear a conversation pay attention to the pragmatic implications of things
that are said about a person rather than to their literal implications. Moreover, they pay particular attention to statements that violate normative expectations for the sorts of statements that are typically made in the type of
situation at hand. However, these expectations can pertain not only to the
content of the information that is exchanged in the situation but also to how
it is exchanged. People often manifest nonverbal behaviors that lead recipients to infer that the information they convey has implications that differ
from those implied by its content alone. Moreover, the style in which people conduct a conversation (e.g., whether they ask other participants ques-
179
tions or elaborate answers to others questions) could have implications

for both their interest in the conversation and their liking for the persons
with whom they are conversing. These implications, in turn, could influence
peoples impressions of both the topic under discussion and the discussants. Two quite different sets of studies bear on these possibilities.
Nonverbal Influences on Perceptions
of a Messages Implications
The nonverbal behaviors that accompany a persons comments in a social
interaction can obviously have an impact on recipients interpretation of
these comments. An individuals disparaging remark to another is more
likely to be interpreted as a tease if the communicator is smiling than if he
or she is not. The frequency and duration of eye contact can often be used
as an indication of feelings of intimacy and of the intensity of feelings that
are conveyed by a communicators statements and, therefore, can influence
the interpretation of these statements (Ellsworth & Carlsmith, 1968). Other,
less controlled behaviors (e.g., eye contact, voice pitch, etc.) can influence
recipients perceptions of the communicators truthfulness and, therefore,
the validity of the information being communicated (for summaries of the
impact of nonverbal behaviors on these and other perceptions, see DePaulo & Friedman, 1998).
A recipients perception of the validity of a message can also be inferred
from manifestations of uncertainty. Speech style characteristics that reflect
hesitancy (stammering, a high frequency of uhhs and ers, etc.) could result from lack of confidence in how the information should be conveyed. On
the other hand, recipients might misattribute these characteristics to lack
of confidence in the validity of the information itself. For example, a person
who has read a story that advocates a particular point of view might try to
convey its contents to someone who opposes this view in a way that the recipient will not find too offensive. Similarly, a students dissertation advisor
might wish to convey the essence of a colleagues devastating criticisms of
a thesis proposal to the candidate in a way that is substantively correct but
is more dispassionate in tone. The communicators difficulty in attaining
these objectives could be reflected in characteristics of his or her speech
style. However, these characteristics could be misinterpreted by the recipient as uncertainty about the validity of the information being conveyed.
Gruenfeld and Wyer (1993; reported in Wyer & Gruenfeld, 1995) demonstrated these effects under conditions in which the difference between communicators objectives and the ostensible purpose of the original information was unobtrusively manipulated. Participants read a passage describing
an unfamiliar disease. They were told that the passage had been taken from
either a newspaper (whose objective is presumably to convey new and in-
180
CHAPTER 7
teresting information) or an encyclopedia (whose goal is to preserve archival knowledge). Regardless of the passages source, however, participants
after reading it were told to convey its content to a nave recipient in a way
that was either as interesting as possible or as accurate as possible. Thus,
in some cases, the assumed objectives of the messages original source and
those of the communicator were ostensibly similar, and other cases, these
objectives differed. After delivering the message (which was tape recorded), both communicators and nave recipients evaluated the seriousness of the disease and reported their specific beliefs about its prognosis,
cause, and treatment. These latter beliefs were coded so that more positive
values indicated stronger beliefs in the content of the original passage.
Communicators own beliefs and opinions about the disease were not affected by either the source of the passage or their own objectives in communicating it. Moreover, a content analysis of communicators speech revealed no differences in the actual content of the information they
conveyed in different conditions. Nevertheless, communicators exhibited
more characteristics of poor speaking style (e.g., a greater frequency of
uhhs and errs), and took longer to deliver the speech (an indication of
rambling), when their communication objectives differed from those of the
original message source than when they were similar. Correspondingly, recipients reported less strong beliefs and opinions about the disease and its
seriousness in the former condition than the latter.
These results are perhaps not too surprising. However, their importance
in the present context lies in their implications that communicators take
into account not only their own objectives in conveying a message but also
the pragmatic implications of the source of the information they are transmitting, and that these factors combine to influence recipients perceptions
of the validity of the information being communicated.
The Impact of Conversational Style on Liking
for the Communicator
The speech style characteristics identified by Gruenfeld and Wyer seem
likely to influence reactions to communications exchanged in an actual conversation. In this context, they might influence participants impressions of
one another as well as the topic under discussion. Additional factors are
likely to come into play, however. People who engage in an informal conversation are typically expected to converse in a way that will make it easy
for others to respond. One obvious way of accomplishing this is to ask
questions of one another. A related technique is to elaborate answers to
one anothers questions, thereby increasing the likelihood that the response will stimulate a relevant idea in the listener. These techniques may
be acquired through social learning and be applied spontaneously, without
181
much deliberation. When the techniques are not applied, however, recipients are likely to find the conversation awkward and difficult to conduct,
and these reactions could affect their liking both for the conversation itself
and for the other participants.
These effects are particularly likely to occur in get-acquainted conversations, where participants are motivated to learn about one another and to
identify areas of mutual interest. For example, suppose a participant in
such a conversation asks, Where are you from? If the other responds,
Chicago. How about yourself? the first person finds it quite easy to continue the dialogue. Alternatively, if the other responds, Well, I live in Chicago now, but I just moved here from Florida, the questioner might be
stimulated to ask about Florida and why the person moved, or might describe his or her own recent visits to the state. However, suppose the other
simply responds Chicago, without elaborating or asking a question in return. Then, the first individual is burdened with the task of finding a different topic to discuss or a different question to ask. This burden increases
the difficulty of continuing the conversation and this difficulty, in turn could
decrease the persons liking for both the conversation and the interaction
partner.
But other factors can enter into the picture as well. An individuals failure to elaborate answers to anothers questions, or the failure to ask questions in return, could also be interpreted as an indication that the individual
is not really interested in interacting with the questioner. This inference
could also decrease the questioners liking for the individual independently
of the effects of these conversational characteristics on the ease of conducting the conversation per se.
Wyer, Swan, and Gruenfeld (1995) investigated these effects in a getacquainted conversation. College students who were previously unacquainted took part in the study in same-sex pairs. However, only one member of each pair was assigned the role of the actual subject. The subjects
partner was recruited as an accomplice. Subjects and their partners arrived
at the experiment in different rooms so they would not see one another
prior to the experiment. Subjects were told we were interested in how people engage in get-acquainted conversations, and that they would be asked
to take part in a 5-minute conversation with another student much as they
might if they met the person for the first time at lunch or on a break between classes. However, we indicated that to ensure that some of the information exchanged in each conversation was similar, we would like them to
ask their partner five questions at some point during the conversation:
1. Where are you from?
2. Do you like school?
3. Where do you live on campus?
182
CHAPTER 7
4. What is your major?

5. How are you doing in your classes?
Subjects were given a list of the questions as a reminder, but were told they
could ask them in any order and to work them into the conversation in a
way that seemed natural. They were further told that their partner did not
receive any questions but was being told simply to conduct the conversation as naturally as possible.
Partners were given similar instructions about the studys purpose. However, we went on to indicate that people vary in the amount of information
they give when they are asked about themselves and in whether they ask
questions in return, and that we were interested in how these factors influenced impressions. We then gave the partner the same list of five questions
we had given to the subject and indicated that when the subject asked
these questions, he or she should respond in one of four ways. In elaboration, question-reciprocation conditions, partners were told:
We would like you to do two things when your partner asks you these questions. First, rather than simply answering the questions with a yes-or-no answer, elaborate your answer in one or two sentences. Then, after doing so,
ask the other person the same question in return. For example, when your
partner asks you if you like school, you might say, Yes, I like the social life
but the tests are hard. How about you? Or, when you are asked where you
are from, you might say, Im from the North side of Chicago, but before that I
lived in Detroit. Where are you from? Answer the questions truthfully, but
dont go into too much detail. Keep your answer to one or two sentences, and
then ask your partner the same question that you were just asked. Except for
your responses to the five questions that I have passed out to you, however,
carry on the conversation as naturally as possible.
Under no-elaboration, question-reciprocation conditions, partners were

told to restrict their answers to one or two words but then to ask the other
the same question in return. Instructions in elaboration, no-question-reciprocation and no-elaboration, no-question-reciprocation conditions, were similar
to those in the first two conditions except that partners were told not to ask
a question in return.
The subject was then ushered into the partners room and seated on the
other side of a partition so they could not see one another. The participants
were then introduced and left alone for 5 minutes to engage in their conversation. Then, after completing the interaction, participants were asked to infer how much they enjoyed the conversation, how easy it was to conduct it,
and how much they thought they would like their partner if they got to
know him or her better. They also predicted how interested their partner
was in them as a person and how much they thought the partner liked
183
them. Finally, they judged the partner with respect to several personality
attributes. Of these, the most relevant to his or her behavior in the conversation was aloofness.
We expected that both partners elaboration and their reciprocation of
questions would influence participants ease of conducting the conversation, and that this, in turn, would infer their liking for the conversation and,
as a result, their liking for the partner. However, we also speculated that if
subjects found that their partner did not elaborate and reciprocate their
questions, they would interpret this behavior as an indication that their
partner had little interest in the interaction and might not like them, and
that these perceptions might also influence their liking for their partner. In
fact, both possibilities were evident.
The effects of conversational style can be seen most easily from path
analyses. Figure 7.4 shows the significant paths connecting the two communication style characteristics to the participants estimates of the ease of
conducting the conversation (Ease), their enjoyment of the conversation
(Enjoy), their perceptions that their partner (O) was interested in them (O
int S) and liked them (O like S) and their liking for the partner (like O). Females perception of the ease of conducting the conversation influenced
both males and females liking for their partner through its mediating influence on both their enjoyment of the conversation and their inference that
the partner liked them. However, whereas males perception of the ease of
conducting the conversation was primarily a function of their partners reciprocation of their questions, females perceptions were influenced primarily by the degree to which their partner elaborated responses to their
questions. In addition, partners elaborations also affected subjects perceptions that the partner was interested in them, and this factor also increased
their liking for their partner. On the other hand, partners conversational
style had no impact on participants perceptions of their aloofness, nor did
these perceptions influence liking for the partners.
The gender differences obtained in the study are provocative. As Wyer
et al. (1995) point out, women are typically more motivated than men to establish a sense of connectedness to the persons with whom they interact
(Chodorow, 1979; Gilligan, 1982; Tannen, 1990).Their partners elaboration
of answers to their questions may have increased their feelings of
connectedness, as evidenced by its impact on these subjects perceptions
that their partner was interested in them and liked them. In contrast, males
placed less emphasis on these factors. Perhaps because they are less socially skilled than females, however, they found that the ease of conducting
the conversation was easier when their partners asked them questions.
In summary, partners counternormative communication style influenced both mens and womens liking for their partner through its mediating impact on their perception of ease of conducting the conversation. In
184
CHAPTER 7
FIG. 7.4. Path diagrams of the effects of the partners conversational style on
subjects liking for their partner (O) and the variables that mediate these effects. Elab = Os elaboration of responses to the subject question. Quest = Os
reciprocation of the subjects question. Ease = subjects perception of the ease
of conducting the conversation. OintS = subjects perception that O was interested in them. O like S = subjects perception that O liked them, and like O =
subjects liking for the partner.
addition, females attributed their partners conversational style to a lack of

interest in them and this attribution was also a major determinant of their
liking for her.
Influence of Counternormative Conversational Style
on Information Seeking
We have assumed that when people encounter an expectancy-violating
communication, they are likely to search for other information that might
explain its occurrence. This cognitive activity may be reflected in the recall
185
of the counternormative messages that give rise to it, as suggested by Wyer

et al. (1994). However, the assumption has other implications as well. That
is, if people seek additional information in an attempt to explain counternormative messages, the information they identify in the course of this
search may have greater influence on their judgments than it otherwise
would.
Evidence of this possibility was obtained in a second study by Wyer et al.
(1995) under conditions similar to those described in the experiment just
described. That is, female participants first engaged in a get-acquainted
conversation with a female partner who responded to their questions by either (a) both elaborating their answers and asking them questions in return
(normative-behavior conditions) or (b) neither elaborating nor reciprocating
the participant questions (counternormative-behavior conditions). Before
engaging in the conversation, however, some participants (ostensibly by
mistake) were exposed to a description of their partner that another participant had provided in a previous experimental session. In some instances,
this description indicated that the partner was sociablea trait to which the
partners conversational style was relevant and, therefore, could provide
an explanation for the partners behavior. In other cases, the description indicated that the participant was high in integrityan attribute that was irrelevant to the partners interaction behavior. Participants in a third, control
condition received no trait information at all.
We expected that when the partners responses to subjects questions
were normatively consistent with the behavior that was likely to occur in
get-acquainted conversations, subjects would base their liking for her on
the interaction and that the others description of her would have little impact on these judgments. When the partners behavior was counternormative, however, participants should seek an explanation for it and, therefore,
should pay more attention to the trait descriptions. As a consequence,
these descriptions should have more impact on their liking for their partners than would otherwise be the case.
Results confirmed these hypotheses. Subjects evaluations of the partner
are shown in Table 7.5 as a function of the partners conversational style
and the relevance of the target descriptions provided. When partners conversational style was normative, subjects evaluated her favorably regardless of whether trait descriptions of the partner were available. (In fact,
their evaluations were nonsignificantly less favorable when trait descriptions were provided, suggesting that, if anything, the traits detracted from
the impact of the interaction.) When the partners behavior was counternormative, however, participants evaluated her relatively unfavorably
when no trait descriptions were provided, but trait descriptions substantially increased their evaluations. Note that this was true even when the
traits were irrelevant to the targets interaction behavior. In other words,
186
CHAPTER 7
TABLE 7.5
Subjects Liking for Their Partner as a Function of the Partners
Conversational Style and Initial Trait Description
Initial Trait Description of Partner
ConversationIrrelevant
ConversationRelevant
None
8.56
6.63
7.50
8.78
6.78
7.78
8.95
3.88
6.56
8.76
5.89
Normative conversational style

Counternormative conversational style
M
trait descriptions of the partner only had an influence on subjects evaluations of her when her behavior in the conversation deviated from expectations and, therefore, stimulated subjects to seek other information that
might account for it. Moreover, this search led the information to have an
impact regardless of whether it was relevant or irrelevant to the explanation that subjects were seeking.
PRAGMATIC COMMUNICATION
IN CLOSE RELATIONSHIPS
The discussion in this chapter has focused largely on the pragmatic influences of communications that are exchanged among casual acquaintances.
When people know one another very well, the pragmatic implications of
their shared messages may be difficult for others to discern. As noted earlier, married couples can often convey feelings to one another through a
statement that seems quite innocuous to observers who are unaware of the
pool of shared knowledge that the partners bring to bear on its interpretation (Watzlawick et al., 1967). Thus, for example, a wife who asks her spouse
at a party what time it is might be interpreted as making a serious request
for information. However, it could be interpreted by her husband as an indirect expression of anger because of his failure to leave the party early
enough for her to see David Letterman.
On the other hand, miscommunication often arises between partners in
close relationships as well as casual acquaintances. This is particularly true
in the case of emotions. One reason for this could be that the expression of
some emotions is inherently ambiguous. However, other factors play a role
as well. A study by Gaelick, Bodenhausen, and Wyer (1985) provided insight
into these possibilities. In an initial session of their experiment, married
couples engaged in a 10-minute tape-recorded discussion of a problem they
were having in their relationship. Then, in a second session, each partner
187
viewed the tape of the conversation and identified three statements that
were made in the conversation that had had an important effect on their
feelings toward one another. Finally, in a third session, partners reviewed
the segments of the tape containing the statements that both they and their
spouse had identified. Partners rated statements they had personally made
in terms of the feelings they intended to convey, their expectations for how
the partner would interpret the statements, and how they thought the partner would respond. They rated statements their partner had made in terms
of the feelings their partner intended to convey, how their partner thought
they would respond, and their actual response. Factor analyses of these ratings revealed they fell along two independent dimensions pertaining to love
and hostility.
Several interesting results emerged. In general, partners attempted to
convey the emotion they perceived their spouse had conveyed to them.
However, they were only accurate in perceiving their spouses hostility.
Consequently, feelings of hostility were actually reciprocated, but feelings
of love were not. One implication of this is that hostility was more likely to
escalate over the course of the conversation than feelings of love.
Second, when wives communicated in a way they intended to be
affectively neutral, their husbands interpreted their statements as expressions of hostility. In contrast, when husbands communicated in a way they
intended to be neutral, their wives interpreted their statements as expressions of love. Gaelick et al. (1985) interpreted these miscommunications in
terms of the stereotyped social role expectations that men and women hold
for one another. Specifically, women are expected to be loving and affectionate. Therefore, when women responded in a way they intended to be
neutral, their husbands interpreted this counternormative comment as hostile and, as noted earlier, were likely to respond hostilely in return. Men,
however, are expected to be hostile and aggressive. Consequently, when
they intended to convey neutral affect, their wives interpreted this expectancy-deviant statement as an indication of love, and presumably reciprocated this emotion. Unfortunately, however, because expressions of love
were typically misperceived, these attempts to deescalate the conflict were
unlikely to do much good.
CONCLUDING REMARKS
Although this chapter has covered a lot of ground, it clearly does not provide a complete account of the factors that theoretically influence peoples
perceptions of the pragmatic implications of the information they receive in
social situations. For example, the extensive research on social attribution,
stimulated by theories of Jones and Davis (1965) and Kelley (1967, 1987),
188
CHAPTER 7
bears indirectly on the situational factors that influence the implications

that people draw from a persons statements and behavior. Research on
characteristics of the source of information that affect the impact of this information (for a review, see Eagly & Chaiken, 1993) also makes salient the
need to take into account the influence of pragmatic factors (e.g., the intentions of the communicator) in conceptualizing the nature of this influence.
A discussion of this research is beyond the scope of this volume. Nevertheless, the work we have summarized suffices to point out the importance of
understanding these processes in the context of actual social situations.
At the same time, the pragmatic influences of information cannot be understood independently of the more fundamental processes that underlie
the spontaneous comprehension of information and an identification of its
validity or its redundancy with prior knowledge. The extent to which pragmatic implications of a message are taken into account often depends on
the recognition that the messages literal implications violate expectations
for the content and form of communications that are likely to occur. In the
research we described in this chapter, these expectations typically derived
from peoples prior knowledge about either the topic of the communication
(and, therefore, whether the message was redundant with prior knowledge
or false) or, alternatively, the source of the communication and the type of
situation in which it is conveyed.
Expectations for the meaning of a communication can also be based on
prior communications that have been conveyed in the same situational
context. Moreover, initial portions of a message can establish expectations
that are violated by later portions. The communications we considered in
this chapter were typically single statements (news headlines, witticisms,
responses to questions, etc.). As we emphasized in earlier chapters, however, many communications that we experience in a social context consist
of sequences of thematically related events, or narratives. In these instances, information that occurs early in a narrative provides a context for
interpreting aspects of the narrative that occurs subsequently. However,
when these latter aspects violate violates based on the context that was established by earlier features of the information, special circumstances can
arise that have not yet been considered. The next chapter addresses these
circumstances in some detail.
C H A P T E R
8
The Dynamics of Humor Elicitation:
The Effects of Informational
Context on the Interpretation
and Elaboration of Narratives
The communications we considered in the previous chapter consisted

largely of single statements. The interpretation of these statements was
based largely on previously acquired knowledge about not only the topic to
which the statements pertained but also the type of situation in which the
statements were made. Similar factors come into play when information describes a temporally related sequence of events and is conveyed in the
form of a narrative or story. There are some additional considerations,
however. People theoretically comprehend descriptions of a series of
situationally constrained events or states of affairs by constructing an episode model of the sequence of occurrences as a whole. In doing so, the interpretation they give to each description in the sequence is likely to be influenced in part by the concepts and knowledge that they have used to
interpret the earlier ones.
To give an example, consider the following pairs of statements:
A.
B.
The pilot was killed when his landing gear malfunctioned. Flying airplanes
can be dangerous.
Marys husband and children were killed when a plane overshot the runway and crashed into their home near the airport. Flying airplanes can be
dangerous.
The second statement in B is likely to be interpreted quite differently than

the second statement in A. This is presumably because readers draw upon
different concepts and knowledge in constructing an event model of the
first statement in each pair and, having done so, apply these concepts and
knowledge to the second statement as well.
189
190
CHAPTER 8
In some instances, however, the concepts and knowledge that are activated and used to interpret the events that occur at the beginning of a sequence cannot be applied to the events that come later. In such cases, recipients must reinterpret the initial events in terms of different concepts in
order to construct a complete understanding of the sequence of events as a
whole. In the preceding example, suppose the statement, Flying airplanes
can be dangerous is presented first. An individual who happens to interpret this statement as equivalent in meaning to It can be dangerous to fly
an airplane might then encounter the statement Marys husband and children were killed. . . . To comprehend the information as a whole, the recipient would presumably reinterpret the first statement as referring to the
hazards of living near an airport.
The comprehension processes that are involved in this example are similar to those described in chapter 7 when a statement violates normative expectations. The only difference is that the expectations are not only activated by the social context in which it occurs. In addition, they result from
the interpretation that has been given to other information in the communication in which the statement is embedded.
Reinterpretations can also occur in the course of comprehending direct
experiences. A waiter whose outward appearance gives the impression of
elegance and sophistication but who is observed to spill soup in a customers lap may be reconceptualized as a pretentious oaf. However, reinterpretations of written or oral communications are more common. Particularly frequent examples occur in jokes and stories whose punch line
stimulates a reinterpretation of the events leading up to it. For example,
consider the following story:
A young Catholic priest is walking through town when a prostitute accosts
him. How about a quickie for twenty dollars? she asks.
The priest, puzzled, shakes her off and continues on his way, only to be
stopped by another prostitute. Twenty dollars for a quickie, she offers.
Again, he breaks free and goes on up the street.
Later, as he is nearing his home in the country, he meets a nun outside her
convent. Pardon me, sister, he asks, but whats a quickie?
Twenty dollars, she says, The same as it is in town.
The events described in the punch line of this story stimulate the reinterpretation of two features of the previous information. First, ones characterization of the nun is revised to include her service as a prostitute. Second,
the question, Whats a quickie?, which was first assumed to be equivalent
to What does a quickie mean?, is reinterpreted as equivalent to Whats
a quickie cost?
Not all reinterpretations of information elicit amusement, as our example of flying airplanes testifies. In fact, the cognitive and motivational un-
DYNAMICS OF HUMOR ELICITATION
191
derpinnings of humor elicitation are not widely understood. This is surprising in light of the central role that humor plays in social communication. It is a rare conversation in which at least one participant does not
respond with amusement to another persons comments or behavior.
Jokes, witticisms, and other humorous events are commonplace in social
interaction situations and can have an impact on the quality of the interactions. Although jokes are often told for the purpose of entertaining, they
can also be used strategically to decrease tension in a heated discussion
(Kane, Suls, & Tedeschi, 1977) or to enliven a boring one. In short, the
transmission and comprehension of humor are central features of social
experience.
Theoretical and empirical analyses of social interaction processes have
largely ignored these processes. Several theories of humor elicitation have
been proposed, beginning with the work of Freud (1928, 1960). However,
most research on humor (for summaries, see Chapman & Foot, 1976, 1977;
Goldstein & McGhee, 1972; McGhee & Goldstein, 1983) has focused on its
motivational bases (cf. LaFave, Haddad, & Maeson, 1976; Zillman & Cantor,
1976) and its personality correlates (Lefcourt & Martin, 1986; Mindess et al.,
1985). Yet, only a few attempts have been made to explicate the cognitive
processes that underlie perceptions that something is funny (Apter, 1982;
Long & Graesser, 1988; Suls, 1972, 1977). And of these, only Apters conceptualization is applicable to situations in which humor is elicited spontaneously in informal social interaction.
In this chapter, we provide a theoretical analysis of humor elicitation in
the context of the comprehension processes we have described in earlier
chapters. Our theory avoids many of the deficiencies of existing formulations while at the same time providing an example of the potential applicability of our more general conceptualization of comprehension phenomena
to information processing outside the laboratory. According to this conceptualization, humor elicitation is a by-product of the comprehension processes postulated in chapter 4 and in some cases, of higher order comprehension processes similar to those discussed in chapter 7. As we have
noted, however, merely the reinterpretation of previously acquired information in the light of new information is not a sufficient condition for humor elicitation. Thus, several additional factors not discussed in earlier
chapters must be taken into account. We first circumscribe the conditions
in which humor is spontaneously elicited by information that people receive and consider the effects of more deliberative, elaborative processes
that occur once the humor-eliciting interpretation of information has been
identified. We then apply the conceptualization to a specific area in which
an understanding of humor elicitation is particularly important, namely, the
amusement elicited by jokes and stories that perpetuate a social stereotype
or are likely to be considered offensive for other reasons.
192
CHAPTER 8
HISTORICAL BACKGROUND
To provide a perspective on the conceptualization to be proposed in this
chapter, a brief review of earlier theories of humor elicitation may be helpful. These theories fall into two categories, one of which focuses on the motivational antecedents of humor elicitation and the other of which concerns
the cognitive processes that give rise to it. Considered in isolation, each
theory is insufficient to account for the numerous circumstances in which
amusement may occur. (A possible exception to this rule is Apters [1982]
theory to be discussed presently.) Nevertheless, the factors that the various theories postulate to influence humor elicitation are generally compatible. That is, the factors postulated by several theories could contribute to
humor elicitation, although not necessarily for the reasons the theories assume.
Motivational Theories of Humor
1. Arousal-Reduction Theories
Responses to humor have sometimes been conceptualized in terms of a
release of tension or reduction in arousal. According to Freud (1905/1960,
1928), for example, peoples responses to humor-eliciting stimulus events is
motivated by the need to reduce tension or arousal they are inhibited from
expressing directly. This arousal, which was often assumed to be aggression- or sex-related, might either be induced by the stimulus itself or exist
before exposure to the stimulus. For example, the humor a joke elicits
could vary with the intensity of suppressed emotions that have previously
become associated with the type of stimulus to which the joke is relevant.
Although Freuds conceptualization can account for some humoreliciting experiences, it is clearly not sufficient to explain the humor that is
elicited by all types of communications. For example, the conceptualization
appears to apply only under conditions in which features of the humoreliciting stimuli are similar to those to which the suppressed emotion is relevant. Given the wide diversity of stimuli that a given individual finds humorous, one would have to postulate a very large number of suppressed
emotions in order to argue that this is a necessary antecedent of humor
elicitation.
A conceptualization of humor elicitation by Berlyne (1969, 1971) is
broader in scope. He assumed an inverted-U relation between physiological
arousal and the experience of pleasure. That is, pleasure increases with
arousal up to a point and then decreases, ultimately reaching a level at
which it becomes aversive. Berlyne viewed a joke as a scenario that induces arousal beyond its optimal value, followed by a punch line that de-
193
creases this arousal to a relatively more pleasant level. This increase in

pleasantness is experienced as amusement.
Although broader than Freuds, Berlynes formulation is also limited in
generality. For one thing, arousal cannot be reduced unless it already exists. Thus, Berlynes theory seems applicable only when a situation-specific
build-up of arousal has occurred before the humor-eliciting event is encountered. This might be true in the case of stories that set the recipient up for a
punch line and, therefore, might induce a degree of cognitive tension. However, witticisms and other unexpected events that elicit humor spontaneously in informal conversations occur in the absence of any prior increase
in arousal.
2. Superiority and Disparagement Theories
A second group of theories (Bergson, 1911; LaFave, Haddad, & Maeson,
1976; Zillman & Cantor, 1976) assume that people derive pleasure from feelings of mastery, and that laughter at anothers deformities or misfortunes
reflects an attempt to establish or maintain these feelings. These reactions
are most likely when there are no social repercussions of expressing them.
Thus, they may occur when the target of the disparagement is generally regarded as socially undesirable or is someone to whom people are indifferent. Americans are more inclined to be amused by a joke if it disparages
Texas Aggies (a group that is of no particular interest to anyone but affiliates of Texas A&M) than if it disparages Jews, African Americans, or
women. Disparagement theories are particularly useful in accounting for
sick or ethnic humor. On the other hand, they have trouble accounting for
self-referent humor that is elicited by ones own ineptitude or by other
events that reflect unfavorably on oneself. Although some people undoubtedly do find it difficult to laugh at themselves, this is by no means universal.
In summary, motivation-based theories can potentially account for the
humor that is elicited in certain circumscribed situations. However, they
are insufficient to account for the spontaneous elicitation of humor that occurs under conditions in which the specific motives postulated by these
theories do not exist. Moreover, the theories do not address the cognitive
mechanisms that give rise to humor and that distinguish humor-eliciting
communications from those that are simply disparaging, aggressive, or
arousal-reducing but do not elicit amusement.
Incongruity Resolution Theories
Many theories (cf. Apter, 1982; Koestler, 1964; Suls, 1972) assume that
amusement is stimulated by the awareness that a stimulus event has two alternative interpretations, each of which could potentially apply but are in
some sense incongruous. Identification of the incongruity, however, may of-
194
CHAPTER 8
ten require access to a large body of knowledge about the type of persons
and events described. For example, consider a joke from the Antioch Humor Test (Mindess et al., 1985):
A blind man enters a department store, picks up his dog by its tail and begins
swinging it over his head. A clerk hurries over and says, Can I help you, sir?
No, thanks, the man replies, Im just looking around.
In this joke, two planes of thought are brought together by the punch line,
one of which pertains to the idiomatic meaning of looking around and the
other of which concerns its literal meaning. However, an appreciation of
the joke also requires knowledge that blind individuals often rely on seeingeye dogs to do their looking.
1. Suls Formulation of Humor Elicitation
A more formal statement of incongruity-resolution processes, provided
by Suls (1972, 1983), is consistent with the conceptualization of comprehension processes we proposed in chapters 4 and 5 as well as more general
theories of prose comprehension (e.g., Graesser, 1981). Suls assumes that
when people begin to read a story, the persons and events described activate a body of conceptual knowledge (e.g., event or episode models as conceptualized in this volume) that can be used to interpret it. This knowledge,
once activated, is brought to bear on the interpretation of information that
is conveyed subsequently. However, recipients may later encounter information that cannot be understood in terms of this knowledge. When this occurs, they must identify alternative concepts and knowledge structures that
can be used to comprehend the new information in the context of the old.
This process often requires a reinterpretation of the original information,
and the generation of this reinterpretation elicits humor.
Suls likens joke comprehension to a problem-solving task in which comprehension is analogous to solving the problem and amusement is analogous to the pleasure that people get from arriving at the solution. This analogy has additional implications. Most obviously, people must recognize
that a problem exists (i.e., that the information cannot be interpreted in
terms of previously acquired concepts and knowledge). More important,
the problem must be neither so easy that the pleasure derived from solving
it is minimal nor so difficult that its solution requires excessive effort. This
suggests that humor is more likely to be elicited by jokes that are moderately difficult to comprehend than by jokes that are either too easy or too
hard. We consider this possibility more fully in the conceptualization we
outline later.
195
2. Apters Reversal Theory

Although inconsistency resolution can be an important antecedent of humor elicitation, it is not a sufficient condition for its occurrence. Not all reinterpretations of previously acquired information in light of new information
are amusing. Scientific discoveries, or creative problem solving more generally, often involve the application of a new set of concepts to previously
acquired knowledge in a way that reconciles this knowledge with new information that becomes available. Although these events can be very pleasureful, they are not amusing.
A conceptualization proposed by Michael Apter (1982) is compatible
with Suls (1983) but specifies both the necessary and the sufficient conditions for humor to be elicited. Apter explicitly recognizes that the social experiences that elicit humor include not only the persons, objects, and
events to which a joke or story refers but also the communicator and aspects of the social situation in which the information is encountered. By taking these contextual factors into account, Apter is able to explain the humor that is elicited not only by jokes and cartoons but also by witticisms
and fortuitous life experiences that are not intended to be funny at all.
Apter (1982) assumed that people who encounter a social experience attempt to interpret it in terms of concepts and knowledge about both its referents and the situational conditions that surround it. In doing so, they arrive at a tentative understanding of the experience as it actually exists from
the perspective of the individuals involved in it. To acquire this understanding, they may often infer unmentioned attributes of these individuals or of
the situation more generally. (Thus, if a protagonist is described as a lawyer, he or she might be inferred to have characteristics that are typical of
lawyers, etc.) Moreover, if a communication is conveyed verbally, peoples
understanding of it may take into account not only the situation described
in the communication but also the purpose for which the communication is
transmitted (e.g., the goals of the communicator). Observed behavior is
likewise understood in terms of the presumed goals of the actor and the situational constraints that are placed on the behaviors occurrence.
When an experience is composed of a sequence of events, however, the
concepts and knowledge that are activated and used to interpret initial aspects of the experience may turn out to be inapplicable. (The sentence
about flying airplanes described at the beginning of this chapter provides
an example.) Then, a reinterpretation of the initial features may be required
in order to arrive at a coherent understanding of the experience as a whole.
This interpretation may give rise to the realization that the situation, or the
persons and objects involved in it, are not actually what they were assumed
to be at first.
This process is very similar to the inconsistency-resolution process proposed by Suls (1972). However, Apter postulates that for the reinterpreta-
196
CHAPTER 8
tion of a situation to elicit humor, two things must be true. First, the reinterpretation must not replace the original interpretation that was made. That
is, the perception of the experience that results from the reinterpretation
must not change ones perception of the experience that purported to exist
at the outset. Rather, both interpretations must be taken into account simultaneously. (For a similar assumption, see Koestler, 1964.) Second, the
perception of the experience that is established by the new information
must in some sense be diminished in importance or value relative to the
conditions that were first assumed.
To see the applicability of the diminishment and nonreplacement principles, reconsider the quickie joke described earlier. To reiterate, this joke
actually exemplifies two types of shifts in interpretation. One, purely semantic shift occurs in the interpretation of Whats a quickie? The second
shift occurs from the perception of the nun as a chaste and devout woman
to the perception of her as a prostitute. Note, however, that the reinterpretation of Whats a quickie? that is implied by the nuns response does not
negate the interpretation that was apparently intended by the priest. Moreover, the nun, although turning out to be a prostitute, remains a nun. In
other words, the new perception of the reality of the situation does not invalidate the appearance of the situation that existed before the reality was
revealed. Finally, note that the nuns holiness is diminished as a consequence of being a prostitute on the side. Thus, both of the conditions that
Apter (1982) postulated to be necessary for human elicitation are met.
Diminishment should not be confused with disparagement. Although disparaging reinterpretations may often be diminishing, not all diminishing reinterpretations are disparaging. In the Wizard of Oz, for example, people are
amused by the lion who purports to be ferocious but who turns out instead
to be meek. The lions true attributes are less unfavorable than his purported ones. They are nevertheless more mundane and, therefore, elicit
amusement.
3. Summary
Apters conceptualization can account for a number of humor-elicitation
phenomena that are not easily explained by other formulations. To give
some examples:
Puns. Puns purport to convey interesting information but are revealed
to be silly and, therefore, trivialize the communication itself. For example:
Two Eskimos sitting in a kayak were chilly. However, when they lit a fire in
their craft, it sank, proving once again that you cant have your kayak and
heat it too.
197
A group of chess enthusiasts were standing in a hotel lobby discussing their recent tournament victories. After about an hour, the manager went up to them
and asked them to disperse. But why? they asked. Because, he said, I cant
stand chess nuts boasting in an open foyer.
Mahatma Gandhi, as you know, walked barefoot most of the time, which produced an impressive set of calluses on his feet. He also ate very little, which
made him rather frail and with his odd diet, he suffered from bad breath. This
made him a super callused fragile mystic hexed by halitosis.
In each case, the humor is elicited by the realization that the communication is not intended to be informative but rather, sets up the reader for a
play on words.
Shaggy Dog Stories. These stories are characterized by a lengthy description of events that appear to lead up to something interesting or exciting but turn out to be totally mundane. In this case, the humor does not
arise from an interpretation of the stories semantic content. Rather, it results from a reinterpretation of the situational context in which the information is presented. That is, the storys length and content gives the appearance of leading to something of interest and importance, whereas the
punch line reveals it to have no interest whatsoever.
Slapstick. There are many forms of slapstick humor. A potential challenge to the diminishment assumption could be the Laurel and Hardy variety, in which amusement is elicited by protagonists repeatedly bludgeoning one another with two-by-fours. Certainly hitting someone with a stick is
not inherently funny. The diminishment in this situation presumably arises
from the realization that the protagonists are not, in fact, killed or even
maimed by one anothers actions. That is, they only appear to hurt one another. Thus, the actual situation turns out to be more mundane than the
purported one.
Expectancy Deviations. In a study by Nehrhardt (1976), blindfolded
participants were asked to judge a series of weights. After several exposures to weights that were fairly similar to one another, they encountered
a weight that was either much lighter or much heavier than the others. On
receiving this weight, subjects typically smiled or laughed. Assuming that
participants were amused rather than simply registering surprise, the question is why. It seems reasonable to assume that participants who encountered the deviant weight inferred that they were being tricked and that
the experiment was not a serious study of weight judgment after all. In
other words, participants interpreted the situation as a whole as less im-
198
CHAPTER 8
portant than they had originally interpreted, and this reinterpretation elicited amusement.
In summary, the diminishment assumption not only helps to account for
a number of humor-eliciting phenomena that are hard for other theories to
explain, but also accounts for conditions in which experiences are reinterpreted but do not elicit amusement. For example, scientific discoveries, the
attachment of meaning to a persons dreams, and the reinterpretation of
events in mystery novels all involve a reinterpretation of a situation in light
of new information. In these cases, however, the reality that is implied by
the new information is of greater importance or value than the original, and
so amusement is not experienced.
A COMPREHENSIONELABORATION THEORY
OF HUMOR ELICITATION
The conceptualization of humor elicitation that Wyer and Collins (1992) proposed has much in common with incongruity-resolution theories and borrows particularly heavily from Apters (1982) formulation. At the same time,
it is derived largely from the assumptions surrounding the comprehension
of information outlined in previous chapters. Specifically, we assume that
amusement is the by-product of comprehension processes similar to those
described in chapters 4 and 5. However, the amount of humor that a joke
elicits can depend on the difficulty of comprehending it. Moreover, it can be
increased or decreased as a result of the amount and type of cognitive elaboration that is performed after its humor-eliciting interpretation has been
identified. We first outline the basic assumptions of the formulation and
provide empirical evidence of its implications. We then turn more specifically to the role of cognitive elaboration in humor elicitation and its implications for reactions to stories that people consider offensive. In our discussion, we focus largely on the humor elicited by jokes and stories to which
event and episode models pertain. As Wyer and Collins (1992) indicated,
however, the conceptualization is potentially applicable to the humor that
is elicited by social experiences in general.
Comprehension Processes
Although the comprehension processes we assume to underlie humor elicitation have been elaborated in previous chapters, it may be worthwhile to
reiterate them in the present context. This will again be done in terms of
postulates. The first postulate is simply a recapitulation of the processes
described in chapter 4.
199
Postulate 8.1. New experiences are spontaneously interpreted in terms of

previously formed event representations (e.g. situation models) with which
they have common features. When more than one alternative knowledge representation is applicable for interpreting an experience, the representation
that is most easily accessible in memory is typically applied. The result of this
cognitive activity is a newly formed situation model of the event being comprehended.
When people have no goal in mind other than to comprehend the information they receive, the concepts and knowledge they apply are typically
those they have used most frequently and recently in the past (cf. Higgins,
Bargh, & Lombardi, 1985; Srull & Wyer, 1979; see Postulates 2.4 and 2.5). If
recipients have a more specific goal in mind, however, this goal may activate concepts that are relevant to its attainment, and these concepts may
then be used instead.
The second postulate reiterates the processes that underlie the construction of episode models, and formalizes the effect of previously encountered stimulus events on the processing of information about subsequent
ones.
Postulate 8.2. Once a mental representation has been formed of an experience, this representation, along with the concepts and knowledge that were
activated in the course of constructing it, is used to comprehend thematically
related experiences that occur subsequently.
Thus, for example, information that a man enters a restaurant activates

knowledge about things that are usually found in restaurants and the
events that occur, and this knowledge provides the basis for comprehending events that occur subsequently. In some cases, the information may be
quite redundant with previously acquired knowledge and therefore, as
noted in chapter 6, it may not be retained (Trafimow & Wyer, 1993). Other
events may not be redundant with ones knowledge in the domain of concern but can nevertheless be interpreted in terms of concepts that are
drawn from this knowledge.
Thus, people do not often spill their drinks in a restaurant. Nevertheless,
information that the man in our example spilled a glass of wine on his suit is
likely to stimulate the construction of an event model in terms of concepts
that compose ones knowledge of restaurants and the things that are found
there. On the other hand, suppose the man entered the restaurant, took off
his clothes and started playing the guitar. These behaviors cannot be easily
understood in terms of concepts that compose ones knowledge of restaurants. Thus, they may stimulate a reassessment of the situation being described. (For example, recipients might conjecture that the restaurant was
located in a nudist colony, and the man was an entertainer and not a guest.)
200
CHAPTER 8
Similarly, the initial reference to the nun in the quickie joke mentioned
earlier activates expectations for her attributes that make her later comment to the priest difficult to interpret. Consequently, recipients identify a
different domain of knowledge (e.g., knowledge about prostitutes) in order
to comprehend it, with the result that the nun is reconceptualized as being
a prostitute as well.
These processes are captured by a third postulate, which essentially recapitulates assumptions made by Suls (1972, 1983) and Apter (1982):
Postulate 8.3. When information about an event cannot be interpreted in
terms of concepts drawn from the same domain of knowledge that was applied to previous events, recipients attempt to identify concepts and knowledge in a different domain that is applicable to both the new event and the
preceding ones. If these concepts can be found, the previously learned events
are reinterpreted in terms of them.16
This reinterpretation would not be made spontaneously, but would require

higher order, goal-directed processing.
Postulates 8.1 to 8.3 are generally applicable to the comprehension of information about a thematically related sequence of events. For humor to be
elicited as a result of reinterpretations of the sort described in Postulate
6.3, additional factors must be considered. We describe these factors in
turn along with empirical evidence bearing on their influence.
The Role of Diminishment in Humor Elicitation
1. General Considerations
The first factor we assume to underlie humor elicitation was suggested
by Apter (1982) and has already been discussed.
Postulate 8.4. Humor is elicited by a reinterpretation of a stimulus event only
if the features of one or more features of the event are diminished in value or
importance relative to that implied by the original interpretation.
Like Apter (1982), we further assume that diminishment can occur along
many dimensions and at several levels of generality. For example, the reinterpretation of a stimulus event might paint a more mundane (or, in some
cases, less desirable) picture of a person or event than the original interpretation implied. Alternatively, the reinterpretation could render the event it16
Note that Postulate 8.3 could be applied not only to stories but also to observed experiences or single statements that people make in a social context. To this extent, this postulate
governs processes similar to those described in chapter 7.
16
201
self, or the circumstances surrounding its occurrence, more mundane or

trivial than the original interpretation implied. Thus, the reinterpretation of
a statement describing person As behavior toward person B might diminish the importance of either an attribute of A, an attribute of B, As behavior
toward B, the statement that describes the behavior, or the situational context in which the statement is made. Postulate 6.4 applies at any or all of
these levels.
2. Empirical Evidence
To the extent that a diminishing reinterpretation of the characteristics of
a person or event is unfavorable, Postulate 6.4 would be consistent with disparagement or superiority theories of humor elicitation described earlier
(cf. Zillman & Cantor, 1976). As we have noted, however, not all diminishing
reinterpretations of an event are disparaging. Events that are initially interpreted as extremely aversive can elicit humor if they are later revealed to
be less so. Shurcliff (1968) provided a good example. Some participants
were told they would be asked to pick up and hold a white rat. Others were
led to believe they would extract blood from the rat with a syringe. The latter condition was elaborately staged, with participants being required to
wear a lab coat and being warned that the rat might bite. In both conditions, however, participants upon picking up the rat found that it was made
out of rubber. After having this experience, participants were asked to rate
the humor it elicited. Participants who had expected to extract blood from
the rat rated the situation as more amusing than those who had only expected to handle it. This finding is consistent with Postulate 8.4. That is, the
reinterpretation of the rat as a rubber toy led to a greater diminishment of
the situation as a whole in the first condition, and the greater amusement it
elicited could be attributed to this fact.
An ambiguity in interpreting Shurcliffs findings arises from the fact that
the situations he constructed elicited anxiety. To this extent, his findings
would be consistent with an arousal-reduction theory of humor elicitation.
However, several studies in our own laboratory do not have this ambiguity.
A study by Collins and Wyer (reported by Wyer & Collins, 1992) evaluated
the effects of diminishment employing variations of the quickie joke described earlier. To reiterate, diminishment could come into play in two
ways in comprehension of the story. First, the nuns response to the priests
question (Whats a quickie?), which reveals she is a prostitute, diminishes
her status as a devout and chaste individual. Second, her response stimulates a reinterpretation of the priests question that diminishes its importance. Both factors could contribute to the humor that was elicited.
To evaluate this possibility, Collins and Wyer constructed four versions
of the story. One version was identical to the joke described earlier. In a
second, the nuns response to the priests question was Ill show you, but it
202
CHAPTER 8
will cost you $20, just like in town. Thus, this response preserves the original meaning of the question rather than stimulating a reinterpretation of it.
The third and fourth versions were similar to the first two except that the
person the priest asked was a third prostitute rather than a nun; these versions, then, eliminated the shift in perception of the respondent. Results
shown in Table 8.1 are very clear. That is, the original story, in which both
diminishing shifts in meaning were present, was judged as amusing. However, eliminating either shift in interpretation decreased amusement, and
eliminating both reinterpretations decreased it still further.
In a second study by Collins and Wyer, participants read a story that
could be interpreted in two different ways, one of which was less likely to
be identified than the other. One story, for example, was likely to be interpreted spontaneously as a conversation about the best way to administer
harsh physical punishment to children, but could also be interpreted as a
discussion of the best way to open a jar of pickles. A second story was most
likely to be interpreted as a mans comments to a woman in the course of
making love in the shower, but could also convey his comments in the
course of washing a dog.
No indication was given at the beginning of the story about the nature of
its subordinate theme. In some versions, however, a statement was inserted near the end of the story that was anomalous when considered in
terms of concepts activated by the dominant theme but made sense in
terms of the subordinate one (specifically, But honey, you know theres
nothing tougher than getting into a jar of pickles, and Honey, bring me the
flea powder, in the two stories, respectively). In other versions, this concluding statement was omitted. Participants were told to read the story for
understanding as they would if they encountered it in a magazine or novel.
After doing so, however, they reported how amused they were by it. Participants judged the stories more amusing when the statement that activated
the subordinate, trivializing theme was present than when it was not.
TABLE 8.1
Mean Humor Elicited by Quickie Joke as a Function of the Effects
of the Punch Line on Interpretation of Story Features
Effect of Punch Line on
Perceptions of Nun
Effect of punch line on interpretation of statement

Change in meaning
No change in meaning
Change
No Change
5.60
3.33
3.67
2.18
Note. Judgments reported along a scale from 0 (not at all humorous) to 10 (extremely humorous). Adapted from Wyer and Collins (1992).
203
Considered in isolation, this finding could have several alternative interpretations. For example, humor might be elicited simply by any reinterpretation of the story regardless of whether or not it was diminishing. This
problem was eliminated in a study by Linda Isbell (reported in Isbell, Wyer,
& Collins, 2002). Participants read a version of the pickle story that was
similar to that employed by Collins and Wyer but varied in terms of the relative salience of the two alternative interpretations. The story was given a
title that identified either the serious theme (Disciplining the Children) or
the mundane one (Getting a Pickle). Participants read this story along
with three filler stories with instructions that (a) the stories could each be
about two quite different situations, (b) each story was given a title that
identified one of the situations, and (c) they should generate an alternative
title that reflected a different interpretation. (The three filler stories described situations that were similar in importance; e.g., a game of cards vs.
a woodwind ensemble rehearsal.) After generating an alternative title for
each story, they gave their reactions to the story along several 010 scales,
one of which pertained to the amusement they experienced. As expected,
participants were more amused when their reinterpretation of the story
was more mundane than the original title conveyed (M = 6.12) than when it
was less so (M = 4.29), and this was true regardless of which version of the
story they read.
Thus, the results of these studies converge on the conclusion that stories elicit more amusement if they are reinterpreted in a way that diminishes the importance of either their referents or the stories themselves. At
the same time, the results do not indicate that diminishment is a necessary
condition for humor elicitation. In the preceding experiment, for example,
both reinterpretations that participants generated could have elicited some
humor, differing only in magnitude. The study of peoples responses to witticisms, described in chapter 7, is worth reconsidering in this context. To
reiterate, participants read some scenarios in which a statement was either
likely to be taken literally or likely to be viewed as ironic. Moreover, the intended meaning of the ironic statement was either less favorable than its literal meaning (thereby diminishing the value of its referent) or more so. Diminishing reinterpretations elicited more amusement than enhancing ones
(6.3 vs. 4.6 along a 010 scale), consistent with Postulate 8.4. On the other
hand, both types of ironic statements were interpreted as substantially
more amusing (M = 5.5) than their true counterparts (M = 1.7).17
17
A second aspect of these data is also worth noting. That is, statements with unfavorable implications were judged more amusing than statements with favorable implications, and this was
true regardless of whether the statements were taken literally (3.5 vs. 1.9) or ironic (6.3 vs. 4.6).
This finding, which would be consistent with disparagement theories (Zillman & Cantor, 1976),
suggests that inconsistency resolution is also not a necessary condition for humor elicitation.
17
204
CHAPTER 8
Thus, these findings could suggest that diminishment is not a necessary

condition for humor elicitation. On the other hand, the realization that a
statement does not mean what it says could diminish the value or importance of the statement independently of its implications for the referent and,
therefore, could elicit amusement for this reason. Therefore, in light of the
ability for Postulate 8.4 to account for a wide range of humor elicitation phenomena that would otherwise be difficult to explain (slapstick humor, puns,
etc.), it seems desirable to retain it pending more compelling evidence
against its validity.
Effects of Comprehension Difficulty
1. Theoretical Considerations
The diminishing reinterpretation of information is often quite easy to
identify. In some cases, however, the reinterpretation may be more difficult
to generate, and may depend on recipients special knowledge of the persons and events involved. For example:
Descartes and two of his buddies go into a bar. The two friends both order a
scotch and soda. The bartender turns to Descartes and asks, You, too? Descartes replies, I think not, and immediately disappears.
The humor elicited by this joke presumably derives from the reinterpretation of Descartes reply in the context of his philosophical conclusion, I
think, therefore, I am. However, someone who is not very familiar with Descartes philosophy might find the joke very bewildering.
Even when a jokes humor-eliciting interpretation can be identified, it
might not be perceived as funny if an excessive amount of cognitive activity
is required to understand it. On the other hand, jokes that are too easy to
understand are also unlikely to elicit much amusement. Thus, as suggested
by Suls (1972, 1983) problem-solving analogy, jokes that are either too easy
or too difficult to comprehend may be less amusing than those that are
moderately difficult. To formalize:
Postulate 8.5. The amount of amusement that is potentially elicited as a result of reinterpreting a stimulus event is a nonmonotonic (inverted-U) function of the time and effort that is required to make this interpretation.
In a sense, this postulate is not new. Many years ago, McClelland, Atkinson, Clark, and Lowell (1953) also hypothesized a nonmonotonic relation
novelty of a stimulus (and, therefore, the difficulty of understanding it in
terms of previously formed concepts and knowledge) and judgments of its
pleasantness. The question is why this relationship exists. Freud (1905/
205
1960) suggested that the sheer pleasure of engaging in cognitive activity in

relation to humor contributes to its gratification. Perhaps as comprehension difficulty increases up to a point, recipients feel more challenged, and
their success in comprehending the information is rewarding (White, 1959).
Beyond this optimal level, however, recipients might begin to feel stupid or
incompetent, and so their enjoyment decreases. Therefore, humor elicitation might decrease correspondingly.
2. Empirical Evidence
The results of several studies are consistent with Postulate 8.5. Of particular relevance is a study by Zigler, Levine, and Gould (1967). Participants
read cartoons that systematically varied over four levels of comprehension
difficulty. Both subjects facial expressions as they read the cartoons and
their subsequent preference rating of the cartoons were recorded. An index
of participants difficulty of comprehending the jokes was also obtained.
This latter measure increased as expected over the four difficulty levels.
However, preference rankings of the cartoons were higher when they were
moderately difficult to comprehend than when they were either very easy
to comprehend or very difficult (but not impossible) to comprehend.
Codings of participants facial expressions revealed a similar pattern.
Collins and Wyer (reported in Wyer & Collins, 1992) also supported the
comprehension difficulty postulate using stimulus materials similar to
those described earlier. Participants read a story that was likely to be interpreted spontaneously as concerning a serious situation (either making love
in the shower or abusing children) but could alternatively be interpreted as
concerning a mundane one (giving a dog a bath or opening a pickle jar, respectively). However, the salience of the mundane interpretation was varied in two ways. First, a statement with an interpretation that required concepts associated with the subordinate theme either was or was not inserted
near the end of the story. Second, the story was preceded by a title that disposed subjects to think of either the serious theme or the mundane one.
(Thus, the story that could be interpreted as either love making in the
shower or giving a dog a bath was titled either Marys bath or Spots
bath. Correspondingly, the story that could concern either child abuse or
how to open a pickle jar was titled Getting Out of a Pickle or Getting a
Pickle). According to Postulate 6.5, participants should be more amused
when it was moderately difficult to identify the subordinate, diminishing interpretation than when it was either very easy or very difficult.
This was in fact the case, as shown in Table 8.2. Pooled over the two stories, more humor was elicited when the humor-eliciting theme was suggested in the text than when it was not (6.36 vs. 4.21), suggesting that this
theme was easier to identify in the former case than the latter. However,
when the theme was suggested in the text, mentioning the theme in the title
206
CHAPTER 8
TABLE 8.2
Humor Elicited by Stories as a Function of Allusion
to Subordinate Theme in the Title and Text
Theme Suggested in Title
Lovemaking vs. dogs bath story

Subordinate theme suggested in text
Subordinate theme not suggested in text
Child abuse vs. pickle story
Subordinate theme suggested in text
Subordinate theme not suggested in text
Subordinate
Dominant
6.17
576
6.85
4.32
5.76
3.87
6.67
2.94
Note. Judgments are reported along a scale from 0 (not at all humorous) to 10 (extremely humorous). Adapted from Wyer and Collins (1992).
as well (thus making its identification even easier) decreased amusement

(5.96 vs. 6.76, when the theme was versus was not conveyed in the text, respectively). In contrast, when the theme was not suggested in the text and
so identifying it was more difficult, mentioning the theme in the title increased amusement (4.82 vs. 3.61, respectively).
As an additional test of Postulate 8.5, we constructed different versions
of the following joke:
Q.
A.
What did Adam say to Eve in the Garden of Eden?

Stand back! I dont know how big this thing gets.
The first interpretation of Adams response, which is induced by the warning, Stand back!, connotes danger. The implications of this interpretation
are diminished by the subsequent realization that the thing is only an
erection rather than a real danger and that the warning results from
Adams sexual naivete. To understand the joke, however, one must realize
that the thing is, in fact, an erection, and must know that Adam and Eve
have no prior knowledge of sex. Identifying these concepts and knowledge
requires cognitive effort, the magnitude of which should depend on
whether the relevant information is explicitly provided.
To evaluate this possibility, we constructed eight versions of the joke
that varied in the explicitness of information bearing on three types of
knowledge that were necessary to understand it: (a) the thing referred to
an erection, (b) it was Adams first such experience, and (c) Adam did not
know its size. Thus, the punch line when all three pieces of information
were provided was Stand back! This is my first erection and I dont know
how big it gets! In contrast, the punch line when none of the three pieces of
information was provided was simply Stand back!
207

TABLE 8.3
Humor Elicited by Adam and Eve Joke as a Function
of the Information Mentioned in the Punch Line
Ignorance of
Size Mentioned
Erection mentioned
Erection not mentioned
Ignorance of Size
Not Mentioned
First Time
Mentioned
First Time
Not Mentioned
First Time
Mentioned
First Time
Not Mentioned
5.57
6.00
6.43
8.00
2.63
2.12
3.25
1.00
Note. Judgments are reported along a scale from 0 (not at all humorous) to 10 (extremely humorous). Adapted from Wyer and Collins (1992).
Humor ratings are shown in Table 8.3 as a function of the amount and
type of information conveyed in the punch line. The joke was generally
more amusing when the punch line explicitly stated that Adam did not
know the size of the erection. More important, however, is the fact that
when Adams lack of knowledge was not explicit, providing one or both of
the other pieces of information increased the ease of understanding the
joke and, therefore, increased the humor it elicited. When Adams lack of
knowledge was explicitly mentioned, however, providing the other two
pieces of information presumably reduced cognitive effort below the optimal level required to appreciate the joke. Therefore, it decreased humor
ratings.
3. Situational and Individual Differences
in Comprehension Difficulty
The ease of identifying a humor-eliciting interpretation is not only a function of the explicitness of the information itself. Situational and individual
difference factors can play a role as well. For example, situational variables
that increase the accessibility of comprehension-relevant concepts in memory should increase the humor elicited by difficult-to-comprehend stimulus
experiences but should decrease the humor elicited by easy-to-comprehend experiences. On the other hand, factors that distract recipients from
comprehending the information in terms of these concepts should have the
opposite effects.
Two studies provide indirect support for this possibility. In a study by
Schick, McGlynn, and Woolam (1972), participants were exposed to two sets
of cartoon strips. All strips were originally from the Peanuts series and so
the characters and their personalities were well known. However, half of
the strips were redrawn to make the characters unfamiliar. Participants
were exposed to several strips in succession and asked to indicate how
208
CHAPTER 8
amused they were by each. People are likely to have some difficulty comprehending strips containing unfamiliar characters as they do not have a
previously formed body of knowledge to draw upon for use in construing
the implications of the strips and their situational context. However, comprehension difficulty should decrease as the characters become more familiar, and so the humor the strips elicit should increase. On the other
hand, suppose the characters in the strips are already familiar. Then, increases in the number of strips should not have this effect. In fact, humor
judgments of the redrawn cartoon strips increased as a function of the number of exposures to the strips. In contrast, the humor elicited by the original Peanuts strips was high at the outset and did not change as a function
of the number of exposures.
A perhaps more interesting finding surrounds the combined effects of familiarity and participants chronic anxiety (as inferred from the Taylor Manifest Anxiety Scale). Chronic anxiety might function as internal noise that increases participants difficulty in comprehending the stimuli. If so, it should
increase the humor elicited by familiar-character cartoon strips, which in
the absence of this noise are below the optimal level of comprehension difficulty. However, anxiety might decrease the humor elicited by unfamiliarcharacter strips, which may be above the optimal level of difficulty in the
absence of distraction. This was in fact the case.
It can be difficult to interpret the effects of situational manipulations of
comprehension difficulty without an a priori understanding of the difficulty
level of the stimuli to be comprehended independently of these manipulations. A study by Goldstein, Suls, and Anthony (1972) provides an example.
Subjects were initially shown photographs of either aggressive stimuli or
automobiles as part of an aesthetic preference task, thereby activating concepts that were associated with the domain to which the pictures pertained. Later, in an ostensibly unrelated experiment, participants judged
the funniness of cartoons in either the same domain to which the pictures
they saw were relevant or in the other, unrelated domain. Cartoons elicited
more amusement in the former condition than the latter. This could suggest
that the concepts activated by the photographs facilitated the comprehension of the cartoons that participants encountered later and, therefore, increased the ease of identifying their humor-eliciting features. According to
Postulate 8.4, however, this increase should only occur if the cartoons are
fairly difficult to comprehend in the absence of this facilitating influence. If
the cartoons were very easy to comprehend in the absence of priming, experiences that further increase the ease of comprehending them should
have precisely the opposite effect. Thus, Goldstein et al.s finding would
only be consistent with the conceptualization we propose if the stimuli
were of the first variety.
209
COGNITIVE ELABORATION PROCESSES

The humor that people experience in the course of identifying a communications diminishing reinterpretation theoretically occurs automatically.
However, their spontaneous amusement may be increased or decreased as
a result of deliberate cognitive elaboration that the individuals perform
subsequently. If the individuals have constructed an event or episode
model in the course of comprehending the situation described, for example, they may elaborate the mental image they have formed of the situation
and the unmentioned events that occur subsequently. Their cognitive elaboration of the humor-eliciting aspects of the situation is likely to increase
the amusement that the story elicits.
On the other hand, some stories are considered offensive by the persons
who encounter them. This could occur because the jokes are tasteless, are
socially inappropriate in the situation in which they are told, or cast aspersions on a person or group that the recipients hold in high regard. In such
cases, recipients postcomprehension cognitive activities may not pertain
to the humor-eliciting reinterpretation per se. Rather, recipients may think
about the motives of the person who conveyed the story and, if the story is
disparaging, the storytellers attitude toward the protagonists. This humorirrelevant cognitive elaboration is likely to decrease the amusement the
story elicits. These possibilities are formalized in the following postulate:
Postulate 8.6. The amount of humor that is elicited as a result of reinterpreting
a stimulus event is a monotonic function of the amount of cognitive elaboration
of the event and its implications that occurs subsequent to its reinterpretation.
1. If recipients processing objective at the time is simply to comprehend and
enjoy experiencing the event, cognitive elaboration of the event will typically increase the humor it elicits.
2. If recipients goal is more restricted, the event is elaborated in terms of its
implications for this more specific objective. In this case, cognitive elaboration could either increase, decrease, or have no effect on the amount of humor elicited, depending on whether the humor-eliciting reinterpretation of
the event is relevant to the attainment of this objective.
There is a qualification on the applicability of this postulate, however. The

humor-eliciting aspects of some jokes are easy to elaborate. For example:
A Texas Aggie and two friends are marooned on a desert island without food or
water. Suddenly, the sky opens and a voice says, Each of you may have one
wish.
One friend says, I wish I were in the arms of my loved one. Immediately, he
is gone.
210
CHAPTER 8
The second friend says, I wish I were in the bosom of my family. Immediately, he is also gone.
The Texas Aggie thinks for a moment and says, Gee, I wish my friends were
with me now.
The humor elicited by this story presumably results from the realization
that the Aggies wish essentially negates the effects of the friends wishes,
returning them to the same abysmal situation they had been in before.
Once readers of the story identify these implications, they are likely to form
a mental image of the friends reactions to finding themselves back on the
island again. To this extent, the opportunity to think about the joke is likely
to increase the amusement it elicits. In contrast, consider the following:
Q.
A.
Why did the Texas Aggie want people to save their burned-out light
bulbs?
He needed them for the darkroom he was building.
Although this joke might be mildly amusing, it has little elaboration potential and so thinking about the joke after comprehending it should have little
impact on amusement. (Moreover, note that both jokes might be seen as
disparaging the intelligence of Texas A&M University students, to whom the
name Texas Aggie is typically applied. Students from this university might
be inclined to elaborate the nonhumorous implications of both jokes, and
this could decrease their amusement.)
Effects of Repetition
Some jokes continue to elicit amusement even when they are repeated one
or more times. This can be true despite the fact that the punch line is well
remembered and, therefore, no longer elicits a new interpretation of the
stimulus events being described. Other jokes, however, are less likely to
bear repeating, even though they might have been considered funny at the
time they were first heard. Suls (1972) conjectured that repetition effects reflect a general tendency for novel stimuli to become better liked as they become more familiar (Zajonc, 1968; but see McClelland et al., 1953). If this
were so, however, repetition should have similar effects on all jokes. Thus,
this conceptualization cannot easily explain why some repeated jokes continue to elicit humor but others do not.
It seems more reasonable to interpret repetition effect in terms of Postulate 8.6. If a joke has high elaboration potential, all possible implications of
it are unlikely to be considered at the time the joke is first encountered. To
this extent, repeating the joke might stimulate a different subset of implications than it did the first time, and these new implications could elicit hu-
211
mor. Thus, the desert island joke described earlier is likely to elicit somewhat different mental images and elaborations each time it is told or
thought about. Eventually, with still further repetitions, no new elaborations are likely to come to mind and so humor is no longer elicited. (All
jokes become stale eventually.) However, jokes that have low elaboration
potential to begin with (e. g., the burned out light bulb joke) may elicit little humor when they are repeated even once.
Cognitive elaborations may be aided by external stimulation. In fact,
much of the humor that is generated by professional comedians results
from their ability to stimulate their audience to elaborate the implications
of a humor-eliciting event with which they are already familiar. Moreover,
many movies and stories concern protagonists whose humor-eliciting behavior, although initially unexpected, is repeated in one form or another
throughout. Don Quixote, for example, repeatedly behaves in foolish ways
while appearing to be distinguished and chivalrous. Inspector Clouseau, the
French detective created by Peter Sellers in the Pink Panther movies appears sophisticated but is actually a buffoon. In such cases, the audience
soon becomes aware of the protagonists actual characteristics, and so his
or her later behavior is not unexpected. Yet, the behavior continues to provide amusement. For example, the first Pink Panther movie one sees leads
the ostensibly sophisticated Inspector Clouseau to be reinterpreted as an
inept detective who only appears sophisticated. Once this is established,
new elaborations of the reinterpretation are encountered as the film presents further instances of Clouseau appearing sophisticated but being in
reality a buffoon. The repeated instances of Clouseaus humor-eliciting behavior constitute externally generated elaborations of ones initial reinterpretation of the character and situation and function in much the same way
as self-generated elaborations. These elaborations elicit humor for much
the same reasons that self-generated elaborations do.
These considerations help to conceptualize the humor elicited in a large
number of situations in which the comprehension processes we postulate
might otherwise seem irrelevant. That is, many everyday situations appear
to elicit amusement although they do not require a reinterpretation. A colleague who is already known to be incompetent but who (perhaps like Inspector Clouseau) behaves in a pompous fashion elicits humor whenever
he or she inadvertently says or does something that betrays this incompetence. Neither the new event nor the colleagues attributes are reinterpreted. However, the colleagues behavior essentially constitutes an externally generated elaboration of the implications of a past event (an earlier
instance of the colleagues buffoonery that occurred in the context of his or
her pomposity) that did stimulate a diminishing reinterpretation at the time
it occurred.
212
CHAPTER 8
Situational Influences on Cognitive Elaboration

and Humor Elicitation
Individual differences undoubtedly exist in the disposition to be offended
by jokes that are considered socially inappropriate in the circumstances in
which they are told, or that perpetuate a negative stereotype. These differences may be reflected in the humor the jokes elicit. However, these differences, which presumably result from differences in the type of cognitive
elaboration that people perform after the implications of the jokes have
been comprehended, should only be apparent when recipients are motivated or able to think about the jokes after their humor-eliciting interpretation has been identified. In the absence of this postcomprehension cognitive activity, people should be amused regardless of their disposition to be
offended by them.
A study with Linda Isbell (reported in Isbell, Wyer, & Collins, 2002) suggested this possibility. In this experiment, we obtained participants reactions to jokes that we expected to be more offensive to women than to men
and, therefore, to produce different types of cognitive elaboration. Two
jokes, which described explicit sexual activity (specifically, fellatio and masturbation), were ones that men typically find amusing but women find embarrassing or offensive. A third joke was scatological:
Q.
A.
Why do farts smell?

For the deaf.
Sex differences in reactions to the latter joke were less clear a priori. Females
might find the joke more embarrassing than men. On the other hand, both
men and women might consider the joke to disparage people with disabilities. Moreover, cognitive elaboration of the joke could stimulate unpleasant
olfactory images and might decrease humor for this reason as well.
Be that as it may, we reasoned that if people are encouraged to think
about the implications of these jokes, they should judge a joke to be more
amusing if they elaborate its humor-eliciting aspects than if they think
about its humor-irrelevant aspects. However, suppose participants give
their spontaneous reactions to the joke without engaging in this postcomprehension elaboration. Then, they should be equally amused by the
joke regardless of the sort of elaboration they might otherwise be disposed
to perform.
1. Method
We told participants we were interested in reactions to stories of the
sort they encounter in daily life, and that to study this, we would like them
to react to a number of stories of the sort they might hear in situations out-
213
side the laboratory. With this preamble, participants in spontaneous-reaction

conditions were told we were particularly interested in peoples spontaneous responses to stories, and that after reading each story, they should report their immediate reactions to it without thinking about it carefully. In
contrast, participants in thought conditions were told we were interested in
how people respond to a story after they have had time to think about it,
and were urged to think carefully about the stories before reporting their
reactions.
Participants in each condition read a series of eight stories that included
the three jokes described earlier, two jokes that were void of content that
participants were likely to consider offensive, and two stories describing
situations that were not expected to be amusing. Participants read each
story/joke and then rated it along four 010 scales pertaining to their emotional reactions: sad, angry, amused, and offended.
2. Results
The amusement elicited by inoffensive jokes was similar regardless of participants sex or instructional conditions (M = 5.10). Moreover, encouraging
participants to think about these jokes did not appreciably increase their
amusement. This could indicate that the innocuous jokes we selected were
not sufficiently high in elaboration potential for an increase to be detected.
Reactions to the sexual and scatological jokes are of greater interest.
The amusement elicited by these jokes is shown in the left half of Table 8.4
as a function of participant sex and presentation conditions. As we expected, men were more amused by sexual jokes when they thought about
them than when they did not (7.08 vs. 4.94, respectively), whereas women
were less so (2.62 vs. 4.64, respectively). Therefore, although men were
more amused by the jokes than women under thought conditions (7.02 vs.
2.62, respectively), this difference was not at all apparent under spontaneTABLE 8.4
Mean Ratings of Inoffensive, Scatological, and Sexual Jokes
as a Function of Instructional Conditions and Distraction
Amusement Ratings
Sexual jokes
Males
Females
Scatological joke
Males
Females
Offensiveness Ratings
Spontaneous
Reaction Conditions
Thought
Conditions
Spontaneous
Reaction Conditions
Thought
Conditions
4.94
4.64
7.08
2.62
1.87
1.12
2.07
5.44
5.16
5.63
2.87
3.00
2.62
1.62
1.29
4.78
214
CHAPTER 8
ous-reaction conditions (4.94 vs. 4.64). The interaction of participant sex

and instructional conditions was significant, F(1,42) = 4.55, p < .05.
Gender differences in reactions to the scatological joke were not evident.
In fact, both men and women reported being less amused by the joke when
they were told to think about it (M = 2.94) than when they were told to give
their spontaneous reactions to it (M = 5.40), F(1,42) = 3.97, p < .05, and this
difference did not depend on participant sex (F < 1). Thus, both groups of
participants appeared to elaborate humor-irrelevant implications of the
joke under thought conditions.
The humor-irrelevant cognitive elaboration that occurred when participants had a chance to think about the jokes was presumably stimulated in
part by their perception that the jokes violated social standards of taste or
portrayed activities they considered undesirable. However, the jokes offensiveness was apparently not the only determinant of the elaboration they
performed. Participants offensiveness ratings of the offensiveness of the
three types of jokes are summarized in the right half of Table 8.4. Participants judged both sexual and scatological jokes to be inoffensive when they
were asked to report their spontaneous reactions. However, whereas men
continued to consider these jokes to be inoffensive when they were told to
think about them, women increased their judgments of the jokes offensiveness in these conditions. These conclusions are confirmed by marginally
significant interactions of instructional conditions and participant sex in
analyses of reactions to both sexual jokes and the scatological joke (in each
case, p < .07). However, the fact that men were less amused by the scatological joke under thought conditions despite not being offended by it suggests
that the humor-irrelevant elaboration they performed in this condition was
not a result of the jokes offensiveness per se. Correlational analyses confirmed these conclusions. That is, the correlation between amusement ratings and offensiveness ratings, computed separately for each experimental
condition, was generally low (between -.16 and -.45) and was significant in
only one case.
3. Effects of Distraction
If the different effects of instructions on reported amusement are mediated by differences in the cognitive elaboration that participants performed
when they were told to think about them, distracting participants from engaging in this cognitive activity should decrease the magnitude of these effects. To examine this possibility, we constructed an additional condition in
which participants received instructions similar to those administered under thought conditions of the main experiment. In addition, however, participants were told that we wished to simulate conditions outside the laboratory in which people hear stories when they are preoccupied with other
215
things, and so we would like them to listen to a tape-recorded conversation

while reading the stories. (The conversation was an intrinsically interchange between a man and a woman that was taken from the opening
scenes of Edward Albees Whos Afraid of Virginia Woolf?).
Results were consistent with expectations. Specifically, distracting participants from thinking about sexual jokes reduced the sex difference in
amusement that occurred under thought conditions (7.08 vs. 2.62 for men
and women, respectively; see Table 8.4) to a level (5.12 vs. 3.94, respectively) that did not significantly differ from the level we observed under
spontaneous reaction conditions (4.94 vs. 4.64, respectively). Correspondingly, distracting both men and women from thinking about the scatological
joke, which they reported as unamusing under thought conditions (M =
2.93) increased their amusement to a level (M = 4.25) similar to that observed when participants gave their spontaneous reactions (M = 5.38). It
therefore seems reasonable to conclude that distraction disrupted the cognitive elaboration that participants performed when they were told to think
about the jokes and, as a result, it decreased the effect of this elaboration
on their amusement.
Sexual Humor in Social Interaction
The preceding experiment provides evidence that both situational and individual difference factors combine to influence the type of cognitive elaboration that people perform in response to a joke and the amusement they experience as a result of this elaboration. The conclusions drawn from the
study are nonetheless limited by the failure to take into account the social
context in which the jokes are conveyed. People may respond quite differently to sexual jokes that are told by members of their own sex than to
those that are told by members of the opposite sex. This could be particularly true of females, who may interpret a mans spontaneous relating of
such a joke to be rather boorish and insensitive.
An experiment by James Collins (reported in Isbell, Wyer, & Collins, 2002)
provided evidence concerning this contingency and, moreover, gave insight
into the cognitive processes that underlie reactions to the jokes. The study is
particularly provocative as the jokes were ostensibly conveyed spontaneously under conditions that were not part of the experiment itself.
Specifically, participants were male and female undergraduates who
took part in the experiment in mixed sex groups. The experiment was conducted by either a male or female undergraduate who was ostensibly helping the faculty member obtain reactions to some humor materials he was
preparing for a future experiment. At this point, however, the experimenter
spontaneously remarked, Say, that reminds me. I just heard a new joke.
Maybe he (Dr. Collins) would like this one. Let me try it out on you . . . The
216
CHAPTER 8
experimenter then proceeded to convey one of two jokes. One, sexual joke,
was the Adam and Eve joke described earlier in this chapter (see Table
8.3). The other was irrelevant to sexuality.
After this aside, the experimenter indicated that before going on with Dr.
Collins experiment (s)he would like the participant to perform two unrelated task. One was a filler task that was intended to disguise the relatedness of the studies. The second was a word association task in which participants were asked to select one of three words that was most similar to a
fourth. One alternative in each set had sexual overtones. For example:
Feeling: touch tingle stimulate
Oral: eating sex communication
Excited: winning aroused horny
Finally, ostensibly as part of the main study, participants were asked to

report the amusement elicited by approximately 45 humorous passages, 12
of which had sexual content.
Collins assumed that the sexual joke told by the experimenter would
stimulate male participants to engage in humor-relevant cognitive elaborations regardless of whether the joke was told by a man or a woman. However, females were expected to engage in these elaborations when the joke
was told by another woman, but to engage in humor-irrelevant elaboration
when the joke was told by a man. The nature of these elaborations is likely
to be reflected in the type of concepts that were accessible in memory at
the time participants performed the word association task and, therefore,
the frequency of sexuality-related words they would select. Data bearing on
this possibility are shown in the top half of Table 8.5. Both male and female
participants made a greater number of sexuality-related word associations
TABLE 8.5
Mean Sexuality-Related Word Associations and Amusement Elicited
by Sexual Jokes as a Function Participant Sex, the Experimenters Sex,
and the Type of Joke the Experimenter Told
Sexual Joke Told
by Experimenter
Male
Experimenter
Sexuality-related word associations
Male participants
3.75
Female participants
1.00
Amusement elicited by sexual jokes
Male participants
5.78
Female participants
4.40
Nonsexual Joke Told

by Experimenter
Female
Experimenter
Male
Experimenter
Female
Experimenter
5.40
4.86
3.00
2.00
1.89
1.92
6.28
6.96
5.13
5.52
4.42
3.96
217
when they had heard a female tell a sexual joke than when they had heard
her tell a nonsexual joke, indicating that the females joke activated sexuality-relevant concepts. Correspondingly, as shown in the bottom half of Table 8.5, they were more amused by the sexual jokes they encountered in
these conditions. In contrast, male participants sexuality-related word associations increased only slightly when a male experimenter told a sexual
joke, and female participants sexuality-related associations actually decreased in these conditions. As a result, the male experimenters telling of a
sexual joke only slightly increased males amusement in response to the
sexual jokes they encountered later, and decreased womens amusement.18
Combined Effects of Elaboration
and Comprehension Difficulty
Although we have considered the effects of comprehension difficulty and
cognitive elaboration separately, these effects are actually interdependent.
For one thing, people cannot elaborate the implications of information they
are unable to interpret. Moreover, the experience of difficulty in comprehending the information could itself stimulate cognitive elaborations. For
example, if recipients believe they have had more difficulty understanding
a joke than they should, they might be stimulated to think about themselves and their competence rather than about the humor-eliciting aspect
of the information. Such humor-irrelevant elaborations could offset any
amusement that the joke would otherwise elicit.
A third consideration arises from the fact that people are typically neither willing nor able to devote an unlimited amount of time to the processing of any given stimulus event (Sperber & Wilson, 1986; Taylor & Fiske,
1978). This is particularly true in a social interaction in which a continuous
stream of events commands attention. Suppose people are unwilling to
spend more than a given amount of time thinking about a stimulus event.
To this extent, the more time and effort they devote to comprehending a
stimulus event, the less time they will devote to a subsequent elaboration
of its implications.
If this is so, the combined effects of comprehension difficulty and cognitive elaboration on humor elicitation can be described more precisely.
Suppose the total time that people are willing to devote to (a) the comprehension of an experience and (b) the cognitive elaboration of its humor18
Our interpretation of these results of course assumes that participants cognitive elaborations of the experimenters joke activated concepts that influenced their interpretation of the
sexual jokes they encountered later, thus increasing the humor the jokes elicited. This assumption is supported by a significant correlation between the amusement they reported in response
to these jokes and the number of sexuality-related word associations that participants made (r =
.39, p < .05).
18
218
CHAPTER 8
eliciting implications is a constant. Wyer and Collins (1992) showed that

under these conditions, the elaboration potential of a stimulus not only
has a direct effect on the humor elicited but also influences the relation
between humor elicitation and comprehension difficulty. Specifically, less
comprehension time is required to maximize the humor an experience
elicits when the elaboration potential of the experience is high than when
it is low.
REACTIONS TO DISPARAGING HUMOR

Jokes and stories often perpetuate a negative stereotype or for other reasons disparage the persons or groups to which they pertain. Some people
are amused by such jokes and others are not. Substantial research has
been devoted to the identification of personality and individual difference
factors that account for these differences. Much of this research has been
stimulated by the assumption that laughing at others deficiencies is a manifestation of displaced hostility or aggression (Freud, 1905/1960). Alternatively, it may reflect a desire to maintain a feeling of superiority by derogating others or by otherwise calling attention to their flaws (Wills, 1981;
Zillman, 1983; Zillman & Cantor, 1976). One obvious implication of this assumption is that people are more amused by jokes that disparage individuals who are particularly threatening to their self-esteem. Thus, for example,
jokes that disparage minority group members are more likely to elicit
amusement in persons who feel their economic or social status is being
threatened by these groups than in individuals who feel more secure.
The possibility that humor is elicited by disparagement per se cannot be
entirely discounted (see Footnote 18). According to the conceptualization
we propose, however, the humor elicited by a joke or story is a joint function of the identification of its humor-eliciting implications in the course of
comprehending it, and the amount and type of cognitive elaboration that
occurs subsequently. To this extent, individual differences in responses to
disparaging humor are potentially traceable to (a) the type of previously
formed concepts and knowledge that people bring to bear on their comprehension of a joke and (b) their motivation to engage in humor-eliciting or
humor-irrelevant cognitive elaboration, assuming they have an opportunity
to do so.
Effects of Comprehension Difficulty
People are usually less amused by a joke that disparages a stereotyped
group to which they personally belong than a group to which they are indifferent. This difference could be due in part to the type of cognitive elabora-
219
tion that people perform in the two conditions. However, it could be localized at the comprehension stage as well. For example, jokes often mention
the ethnic group to which they pertain at the outset. People who do not expect to be offended by a disparaging joke about this group may spontaneously activate a body of knowledge about attributes that are stereotypically
associated with the group and use this knowledge to comprehend the rest
of the joke or story. In such cases, activating the stereotype may facilitate
comprehension of the joke and, perhaps, increase the humor that it elicits.
On the other hand, people who identify with the group may anticipate that
the joke is going to offend them, and may think about the motives of its
source. This humor-irrelevant activity could interfere with the identification
of the jokes humor-eliciting interpretation. Whether or not this interference
increases or decreases amusement, however, depends on the difficulty of
comprehending the joke more generally.
The extent to which mentioning a stereotyped group activates humorirrelevant cognitive activity can also depend on characteristics of the storyteller. A joke that pertains to Jews is more likely to stimulate this activity in a
Jewish listener than in a Ku Klux Klan member. On the other hand, it is less
likely to do so if the joke-teller is Jewish than if (s)he belongs to the KKK.
However, not all jokes that refer to members of a stereotyped group are
disparaging. If people who anticipate that a joke will perpetuate a negative
stereotype engage in cognitive activity that interferes with comprehension, it
could affect the amusement the joke elicits even if the joke turns out not to
perpetuate the stereotype at all. Consider the following joke, for example:
A bus passenger notices that the woman sitting beside him is staring at him.
Im sorry to bother you, she says, but I wonder if you would mind my asking you a personal question. Are you by any chance Jewish?
I certainly dont mind your asking, replies the man good-naturedly, but
no, I dont happen to be Jewish.
The woman continues to stare at him, however, and asks again, Are you
sure youre not Jewish?
Yes, Im sure, says the man, now somewhat irritated. My mother is Roman Catholic, and my father is Japanese. Theres really no chance that I am
Jewish.
This still doesnt satisfy her, and as the trip continues, she becomes even
more insistent. Now, I know that youre Jewish. Tell me youre Jewish. Admit
it!
Finally, after repeated denials, the man, only to keep from being pestered
further, says, Okay. Have it your way, Im Jewish.
Well, the woman responds, you certainly dont look Jewish.
Thus, the joke does not disparage Jews, but rather, disparages the stereotype. However, this does not become clear until the punch line. The allusion
220
CHAPTER 8
to the ethnic group in the material preceding the punch line could stimulate
Jewish recipients to engage in cognitive activity that interferes with their
later comprehension. To this extent, they might consider the joke to be less
funny than non-Jewish recipients would, even though the jokes implications are consistent with their a priori values. (This conjecture assumes
that the joke is already moderately difficult to understand. If it were very
easy to understand, the increases in comprehension difficulty resulting
from this activity could increase Jewish individuals appreciation of the
joke rather than decreasing it.)
Effects of Cognitive Elaboration
Although individual differences in attitudes and values undoubtedly influence the comprehension of a joke or story, their primary effects are likely
to result from their impact on the cognitive elaboration that occurs subsequently. Suppose individuals encounter the following joke:
Q.
A.
What is the difference between a Jew and a canoe?

A canoe tips.
This joke is extremely easy to understand. Its humor derives from the
different meanings of tips, coupled with knowledge of the stereotype of
Jews as being tight with money. Thus, differences in the humor the joke
elicits are likely to be due largely to differences in the type of cognitive
elaboration that recipients perform subsequent to its comprehension. That
is, people who are generally offended by the perpetuation of the stereotype
may think about why the joke is told and may question the attitudes and
motives of the storyteller. However, people who personally believe the stereotype, or are not concerned with perpetuating it, might not engage in this
humor-irrelevant elaboration.
There are two qualifications on this prediction, however. First, differences in reactions to the joke should only be evident when people have an
opportunity to elaborate its implications after comprehending it. When this
opportunity does not exist, recipients may be equally amused by the joke
regardless of their feelings about the group being disparaged. (This could
explain why people often laugh at jokes they hear at a party that they
would find unamusing in a less cognitively demanding situation in which
they had more opportunity to think about the jokes implications.)
A second contingency surrounds the elaboration potential of a joke.
When a jokes humor-eliciting reinterpretation evokes visual images, or if
the situation it implies can be elaborated, individuals who are not offended
by the joke may elaborate its humor-eliciting implications and, therefore,
may be more amused by the joke than they might otherwise be. However,
221
jokes of the sort provided in our example, which are simply plays on words,
have low elaboration potential. The amusement elicited by these jokes is
unlikely to increase with the opportunity to engage in elaboration regardless of recipients attitudes toward the stereotyped group. In fact, if individuals have nothing to distract them, any additional postcomprehension
thoughts they might have about the joke could pertain to the source of the
joke, and perhaps the validity of the stereotype, and these thoughts could
occur even if these individuals do not identify with the group being disparaged. As a result, the amusement experienced by these individuals, like persons who are disposed a priori to elaborate the humor-irrelevant implications of the joke, might decrease when they have an opportunity to think
about the joke more extensively.
A second study with Linda Isbell (see Isbell et al., 2002) examined this
possibility. Participants under thought and spontaneous reaction conditions similar to those in Isbells earlier study were asked to read and give
their reactions to six stories. Four of the stories were nonhumorous, but
two others were jokes that had negative implications for the protagonists
intelligence. One of these jokes (e.g., the desert island joke described earlier) had high elaboration potential, whereas the other (e.g., the burnedout light bulb joke) had low elaboration potential. Moreover, the protagonist in one joke was described as Polish and in the other was described as a
Texas Aggie. (The type of protagonist associated with each joke was of
course counterbalanced.)
The midwestern college students who participated in this study were undoubtedly aware of the stereotype of Polish as unintelligent. However, they
were typically not of Polish extraction themselves, and were unlikely to be
offended by the jokes. (Participants reactions to the jokes in the experiment confirmed this assumption; 65% of the participants reported being
not at all offended by these jokes, or 0 along the 010 scale used to report
their judgments.) Although Texas Aggies are sometimes stereotyped as unintelligent as well, this stereotype was generally unknown in the student
population from which participants were drawn. Based on the assumptions
outlined earlier, we expected that encouraging participants to think about
when Polish jokes with high elaboration potential would stimulate them to
elaborate the humor-eliciting aspects of the jokes, leading them to be more
amused by the jokes than they would otherwise be. In contrast, encouraging participants to think about Polish jokes with low elaboration potential
should stimulate them to elaborate the jokes negative implications for the
stereotyped group to which the jokes refer, and thus should lead them to
be less amused by the jokes than they would otherwise be.
Results were consistent with these assumptions. Participants reported
being equally amused by Texas Aggie jokes regardless of whether they
were told to think about them (M = 4.16) or to give their spontaneous reac-
222
CHAPTER 8
tions (M = 4.52). In contrast, thinking about Polish jokes with high elaboration potential increased the amusement the jokes elicited from 4.04 (under
spontaneous reaction conditions) to 5.13 (under thought conditions),
whereas thinking about Polish jokes with low elaboration potential decreased amusement from 4.96 to 3.86. These differences were reflected in an
interaction of elaboration potential and instructional conditions in an analysis of reactions to Polish jokes alone, F(1,40) = 4.27, p < .05.
Thus, the results provide insight into peoples reactions to jokes that
perpetuate a negative stereotype. Thinking about a joke that perpetuates a
negative stereotype can obviously decrease the amusement experienced
by persons who are offended by the joke. However, thinking about such a
joke can also decrease the amusement experienced by persons who are not
offended by the joke, provided the jokes humor-eliciting implications cannot easily be elaborated.
The Effects of Social Context on Responses
to Sexist Humor
Humor-irrelevant cognitive elaborations are particularly likely when participants themselves belong to the stereotyped social category and, therefore,
are apt to be offended by it. However, this is not the only factor that can influence humor-irrelevant elaboration. In some cases, recipients might perceive that a joke-teller is motivated by a desire to ingratiate his or her audience, or alternatively, to convey hostility toward the person or group that
the joke disparages. They might think about the implications of these motives regardless of whether or not they are personally offended by the joke.
Consequently, they might be relatively less amused by the joke than they
would be if the joke-teller were simply trying to be funny. In other cases,
people might be aware of the motives that lead a joke to be told and still
elaborate its humor-eliciting implications. (This is particularly true if the
jokes elaboration potential is high, as suggested by the results of Isbells
experiment described earlier.)
The effects of these contextual factors may be particularly evident (at
least, among college students) when jokes exploit a negative stereotype of
men or women. Consider the following joke:
A man and a woman are in the elevator together. The woman suddenly starts
taking off her clothes and says, Make me feel like a woman. The man takes
off his clothes, throws them on the floor, and says, Here, fold these.
Womens reactions to this joke, which perpetuates a negative stereotype

of women, are likely to differ from mens reactions to it. However, the na-
223
ture of these reactions may depend on both the sex of the communicator
and the social context in which the joke was originally told. The effects of
these factors may be mediated by their influence on recipients assumptions about the speakers motives and, therefore, on the amount and type
of cognitive elaboration that they perform.
1. Perception of the Motives for Telling Sexist Jokes
The motives that people perceive to underlie the communication of sexist jokes were investigated by Isbell et al. (2002). Male and female college
students were introduced to the study with instructions that we were particularly interested in the motives that people attribute to individuals who
tell jokes that convey negative stereotypes of men and women. With this
preamble, we gave participants a short questionnaire containing four (female-bashing) jokes that perpetuated a negative stereotype of women and
four other (male-bashing) jokes that conveyed a negative stereotype of men.
Each joke was preceded by an indication of the sex of both the person who
told the joke and the individual to whom it was told. (Each joke was associated an equal proportion of times with each of the four speakeraudience
configurations.)
Participants were asked to imagine that each joke had been told in the
situational context described and to estimate the likelihood that the joketeller was (a) just trying to be funny, (b) trying to gain social approval, or
(c) being intentionally hostile or aggressive. They also estimated the likelihood that the joke-teller personally believed in the stereotype being conveyed. These estimates, reported along scales from 0 (not at all) to 10 (extremely), are shown in Table 8.6. as a function of the joke-tellers sex, the sex
of the audience, and the type of joke that was told. Between-cell comparisons, shown in the table, suggest the following conclusions:
1. Participants in all cases inferred that the joke-tellers primary motive
was to be funny (M = 7.30). However, they perceived this motive to be particularly strong (M = 8.08) when someone told a joke that disparaged the opposite sex to a member of his or her own sex (i.e., when a man told a femalebashing joke to another man, or a woman told a male-bashing joke to another
woman).
2. Joke-tellers were perceived to have less strong personal beliefs in the
stereotype implied by a joke when the joke disparaged members of their
own sex (M = 3.16) than when it disparaged members of the opposite sex (M =
5.10). However, this difference was evident both when men told a malebashing joke (2.28 vs. 4.97) and when women told a female-bashing joke (4.09
vs. 5.23).
224
CHAPTER 8
3. Joke-tellers were judged more likely to be seeking approval when they

either told a male-bashing joke to a female or told a female-bashing joke to a
male (M = 5.81) than when they told these jokes to members of the sex that
was being disparaged (M = 3.19). This was true regardless of whether the
joke disparaged males (5.75 vs. 3.37) or females (5.87 vs. 3.00).
4. Persons who told male-bashing jokes were attributed less hostility if
they were male (M = 1.37) than if they were female (M = 3.97). In contrast, persons who told female-bashing jokes were seen as hostile regardless of their
sex (3.27 vs. 3.05 when the jokes were told by men and women, respectively).
In addition, joke-tellers were attributed greater hostility if they told a joke to
members of the sex being demeaned (M = 3.32) than if they told it to members of the opposite sex (M = 2.51). This difference was evident regardless of
whether the joke disparaged men (3.07 vs. 2.26) or women (3.56 vs. 2.75).
2. Context Effects on Amusement and Offensiveness
If the motivational attributions described in Table 8.6 stimulate humorirrelevant elaboration, they should decrease the humor the sexist jokes
elicit. As our earlier experiments indicate, however, this elaboration may
not always occur.
To understand the effects of contextual factors on humor elicitation, it is
necessary to distinguish between the audience to whom a speaker intends
to communicate and the actual recipient. To this end, Isbell et al. (2002)
constructed stimulus materials using the same eight jokes employed in the
previous study (Table 8.6). One stimulus replication comprised four filler
TABLE 8.6
Motives Attributed to Tellers of Sexist Jokes as a Function of the
Joke-Tellers Sex, the Sex of the Audience, and Joke Type
Male Joke-Teller
Male
Audience
Motivation for telling male-bashing jokes
To be funny
Belief in stereotype
Social approval
Hostility
Motivation for telling female-bashing joke
To be funny
Belief in stereotype
Social approval
Hostility
Female Joke-Teller
Female
Audience
Male
Audience
Female
Audience
7.67ab
2.72b
3.69b
1.66b
7.47ab
1.84b
6.75a
1.09b
6.59b
5.53a
3.06b
4.47a
8.12a
4.41a
4.75b
3.44a
7.46
3.62
4.56
2.66
8.04a
4.58ab
6.46a
2.79
7.00ab
5.88a
2.62b
3.75
7.08ab
4.17b
5.29a
2.71
6.46b
3.92b
3.38b
3.38
7.14
4.63
4.43
3.16
Note. Means in each row with unlike superscripts differ at p < .05.
225
jokes had nothing to do with sex-role stereotypes, two male-bashing jokes

and two female-bashing jokes. The second replication was composed of the
same filler jokes and the remaining four stereotype-related jokes.
Participants were told we were interested in how people respond to
jokes that were conveyed in social contexts outside the laboratory, and
that to study this, we had collected jokes in a number of informal conversations that had taken place on campus. On this pretense, we asked participants to give their reactions to eight jokes, preceded in each case by a description of the context in which the joke had ostensibly been told. (For
example, one joke had ostensibly been told by a male graduate student to
two female students at a coffee house, another had been told by a female
undergraduate to three undergraduate men at a fraternity party, etc.) Participants were told to imagine they had actually overheard the joke being
told conveyed in this context, and to indicate both how amused they would
be by the joke and how offended they would be. These ratings were made
along scales from 0 (not at all) to 10 (extremely).
Each of the four filler jokes was ostensibly told to a mixed audience. The
joke-teller and audience assigned to the four target jokes were varied in a
partial latin square design so that pooled over participants and stimulus
replications, each target joke was told about the same number of times in
situations that represented each possible combination of speaker sex and
audience sex.
The top half of Table 8.7 shows the amusement elicited by each type of
joke as a function of the sex of the joke-teller, the sex of the audience, and
the sex of the participants. The bottom half of the table conveys the offensiveness of the jokes. The implications of these data are conveyed most
easily by considering separately participants reactions to (a) jokes that disparaged their own sex and (b) jokes that disparaged the opposite sex.
Reactions to Jokes That Disparaged Participants Own Sex. First consider jokes that perpetuate a negative stereotype of males. Men were less
amused by these jokes when they were told by a woman to a group of men
(M = 2.43) than they were in any of the other three joke-teller/audience
conditions (M = 4.19). Moreover, they were nonsignificantly more offended
by the jokes in this condition (M = 3.00) than in the other three conditions
(M = 1.86).
The data summarized in Table 8.6 suggest why this might be so. That is,
male-bashing jokes are perceived to convey greater hostility when women
tell them to a man (M = 4.47) than they are under other conditions (M =
2.06). In addition, joke-tellers are attributed stronger beliefs in the stereotype in the former condition than in others (5.53 vs. 2.99). It seems reasonable to suppose that men who encounter a woman telling a male-bashing
joke to men are offended by the joke as a result of these attributions and
226
CHAPTER 8
cognitively elaborate its humor-irrelevant implications. Consequently, they

are less amused by the joke than they would otherwise be.
Note, however, that male-bashing jokes are also perceived to reflect hostility, and a strong belief in a negative male stereotype, when a woman tells
these jokes to other women (see Table 8.6). Despite this fact, men were not
offended by the jokes in this condition (M = 1.63). Moreover, they were
more amused by the jokes in this condition than they were in other
speakeraudience conditions. Thus, men may have attributed hostility to a
woman who told a male-bashing joke, and may have thought that she personally believed in the stereotype she was perpetuating. Nevertheless, they
were only offended when she told the joke to men and, therefore, appeared
to be intentionally insulting.
Womens reactions to jokes that disparaged females showed a different
pattern. As shown in Table 8.7, women were not very amused by femalebashing jokes in any condition (M = 3.64). Moreover, although women were
somewhat less amused when joke-tellers told these jokes to someone of
their own sex than they told them to someone of the opposite sex, this difference was not reliable (p > .10). Perhaps women are generally threatened
by the perpetuation of a female stereotype and, therefore, are not amused
by female-bashing jokes regardless of the context in which the jokes are
TABLE 8.7
Males and Females Ratings of Jokes as Amusing and Offensive
as a Function of Joke Type, the Sex of the Joke Teller,
and the Sex of the Intended Audience
Male Joke-Teller
Male
Audience
Female
Audience
Female Joke-Teller
Male
Audience
Female
Audience
Amusement Ratings
Male participants
Male-bashing joke
Female-bashing joke
Female participants
Male-bashing jokes
Female-bashing jokes
4.21
4.18
4.03
4.72
2.43
3.51
4.33
5.07
6.17
3.37
6.80
3.77
5.96
4.04
6.89
3.39
Offensiveness Ratings
Male participants
Male-bashing jokes
Female participants
Male-bashing jokes
1.46
1.00
2.50
1.98
3.00
1.93
1.63
1.43
0.90
2.43
0.87
3.70
1.57
3.54
0.39
3.23
227
told. In contrast, men are not particularly threatened by the perpetuation of

a male stereotype. Therefore, they only fail to be amused by male-bashing
jokes when the joke-teller is intentionally insulting.
Jokes That Disparaged the Opposite Sex. Men (but not women) were
less amused by female-bashing jokes when the jokes were told to women
(M = 3.84) than when they were told to men (M = 4.90), and this was true regardless of the joke-tellers sex. Correspondingly, women (but not men)
were less amused by male-bashing jokes when these jokes were told to men
than when they were told to women (6.06 vs. 6.85). Thus, participants appeared to be more amused when they perceived the joke-teller to be seeking social approval (see Table 8.6). Conceivably both men and women recognized this motivation in all conditions. However, they only elaborated the
humor-irrelevant implications of the jokes when the jokes disparaged members of their own sex.
3. Conclusions
The results of Isbell et al.s studies are complex in detail. They nevertheless confirm the general conclusion that the social context in which a
stereotype-perpetuating joke is told can influence the amount of humorirrelevant cognitive elaboration that occurs in response to it and, therefore, the amount of humor that the joke elicits. Moreover, although this
elaboration is likely to concern the reason why the joke was conveyed, it
does not necessarily result from perceptions that the joke is particularly
offensive. In some instances, jokes were reported as funny despite being
offensive. In other cases, jokes were not considered very funny despite being perceived as relatively inoffensive. Perhaps the offensiveness of the
jokes was not a determinant of the humor-unrelated cognitive elaboration
that occurred, but instead, was a consequence of this elaboration. That is,
the social context in which a joke is told might create expectations for the
jokes content, and a violation of these expectations might stimulate attempts to construe the speakers motives. This cognitive activity could decrease the humor the joke elicits independently of the motive that the
speaker is ultimately attributed.
CONCLUDING REMARKS
Humor is a fundamental ingredient of social communication in everyday
life. A specification of the conditions in which it is elicited is therefore a necessary component of any theory of real-world information processing. We
have argued that the experience of amusement is partly a result of the ba-
228
CHAPTER 8
sic processes of comprehension outlined in previous chapters of this book.

That is, it occurs when the episode models that have been formed spontaneously in the course of comprehending a sequence of events must be
modified in order to understand the sequence as a whole, thereby stimulating higher order, goal-directed information processing in order to perform
this modification.
However, although the reinterpretation of information that results from
this processing is a necessary antecedent of humor elicitation, it is not sufficient. In addition, the reinterpretation that occurs must generally diminish
the value or importance of either the events described in the information
being processed, the participants in these events, the situational context in
which the information is conveyed or, in some cases, the information itself.
Moreover, the humor that is elicited by a social experience is also a function of the cognitive elaboration that occurs after its humor-eliciting reinterpretation has been identified. The amount and type of this elaboration,
which depends on both the stimulus information itself and the prior knowledge and values of the recipient, can either add to or offset the amusement
that might otherwise be experienced.
The formulation we have proposed in this chapter does not intend to be
the final word. However, it provides a framework within which future
research and theorizing on humor elicitation can be conceptualized and investigated. At the same time, the comprehension and elaboration processes that underlie humor elicitation can also come into play in responding
to social information more generally and in making judgments and decisions on the basis of this information. The last section of this volume is devoted to a discussion of these processes.
Although our discussion in this chapter has focused on the antecedents
of humor, we have ignored a consideration of its consequences. This concern, which is obviously important, is beyond the scope of this book. One
obvious by-product of amusement, however, is the experience of positive
affect. To this extent, it could have an impact on information processing
similar to that of positive feelings more generally. These effects are also discussed in the final section of this volume.
P A R T
III
INFERENCE PROCESSES
C H A P T E R
9
Crimes, Vacations, and Political
Candidates: The Construction and
Use of Narratives in Social Judgments
The discussion of comprehension processes in the preceding chapters was

focused largely on the construction of narrative representations of social
experience. To the extent that these representations occupy the bulk of our
social knowledge (Schank & Abelson, 1995), they are correspondingly the
primary basis for judgments and decision to which this knowledge is relevant. In this section, we consider the conditions in which narrative representations are brought to bear on judgments and decisions and the processes that underlie their use.
Narrative representations can enter into inference processes in several
ways. As suggested in chapter 3, some inferences occur spontaneously
when comprehending a sequence of events in terms of a previously formed
narrative representation (Spiro, 1977; see also Graesser, Singer, & Trabasso,
1994). However, narrative-based inferences can also be the result of conscious goal-directed activity. In some cases, a narrative account of real and
imagined events might be particularly relevant to the objective at hand
(e.g., planning how to behave in order to please a date, construing the
events that lead up to a murder in a detective novel, or predicting the consequences of taking a new job). In other cases, a narrative representation is
not essential. Nevertheless, information that is conveyed in a form that facilitates the construction and use of a narrative may be easy to comprehend in terms of previously acquired knowledge, and so inferences may be
based on this type of representation rather than other types.
The next two chapters discuss narrative-based inference processes in
some detail. In the present chapter, we consider the conditions in which
narrative representations of new information are constructed and used to
231
232
CHAPTER 9
make inferences about their referents. However, the construction of these

representations often requires the activation and use of previously formed
knowledge structures, many of which constitute implicit theories about the
causes and consequences of social events. The use of these implicit theories, and of previously formed narrative representations more generally,
are discussed in chapter 10.
THE IMPACT OF NARRATIVES

ON JURY DECISION MAKING
One of the first systematic explorations of narrative-based inference processes was conducted by Pennington and Hastie (1986, 1988, 1992) in research on jury decision making. Jurors are presumably motivated to determine whether a person is guilty or innocent of a crime. (This situation is not
one that normally occurs in daily life. However, similar motives are likely to
operate when reading a news story, watching a movie, or reading an Agatha
Christie mystery novel.) To make this determination, people may attempt
to reconstruct the sequence of events that led up to the crime as well as the
crime itself and its aftermath.
This reconstruction can sometimes be difficult. For one thing, the events
are not always described in the order they occurred. In movies and mystery novels, for example, events that occurred early in the sequence are often not revealed until near the end. Courtroom testimony is typically organized according the witnesses who provide it, and is not conveyed in the
order in which the events described are likely to occur. In such cases, people are likely to find it cognitively taxing to fit the various pieces of evidence into a scenario that permits their temporal and causal relatedness to
be evaluated. As a result, they may be inclined to use other strategies to
evaluate the evidence. In a jury trial, for example, they might attach more
importance to the credibility of witnesses who provide individual pieces of
evidence rather than considering how well the evidence fits into a coherent
account of the crime.
Pennington and Hastie (1986, 1988, 1992) provided compelling evidence
of narrative-based decision processes and the factors that influence them.
Participants in one study (Pennington & Hastie, 1988) read the transcript of
a murder trial and were asked to think aloud as they arrived at a decision
about the defendants guilt. Content analyses revealed that participants
thought protocols were more likely to have a story structure than to have
other possible (e.g., categorical) structures. A further study confirmed this
conclusion. That is, participants who had read the testimony and reported
CONSTRUCTION AND USE OF NARRATIVES
233
their verdict were given a recognition memory test containing some statements that had actually been made in the course of the testimony and others that had not. Participants were more likely to report a statement as having been mentioned in the story if its implications were consistent with the
narrative-based causal scenario on which they based their verdict than if
its implications were inconsistent with this scenario.
A second series of studies (Pennington & Hastie, 1988) examined these
inference processes more directly. Participants listened to the transcript of
an actual murder trial containing testimony for both the prosecution and
the defense. In witness-order conditions, the evidence favoring a given verdict was organized according to the witness who provided it, as in the original trial. In story-order conditions, the evidence for a given side was presented in the order it became relevant in the sequence of events that had
allegedly occurred. (That is, testimony about events that preceded the incident was presented first, followed by testimony about the incident itself,
the arrest, the autopsy, etc.). This was done in a 2 2 design so that the order of presenting the prosecution testimony and the order of presenting
the defense testimony were varied independently. Participants after receiving the testimony reported both their verdict and their confidence that
their decision was correct.
When the defense and prosecution testimonies were presented in different orders, 73% of the participants recommended the verdict implied by the
evidence that was conveyed in story order. When the evidence favoring
both verdicts was presented in the same way, the percentage of participants who favored a guilty verdict did not depend on how the evidence was
ordered. However, participants were much more confident of their judgments when the evidence was in story order. In short, evidence that was
conveyed in an order that facilitated the construction of a narrative account of the crime was apparently easier to comprehend, and so recipients
were more confident of its implications. Consequently, they were more
likely to base their judgments on this evidence than on evidence conveyed
in other ways.
This conclusion has further implications. For example, if people have
formed a narrative of the sequence of events that occur, they may be more
receptive to additional information that is consistent with the implications
of the narrative as a whole than to information that conflicts with these implications. Moreover, this could be true regardless of the credibility of the
informations source. In contrast, if people are unable to construct a coherent narrative representation of the information, they may construe the implications of each piece separately. In this case, the credibility of source of
this additional information should be given more weight.
234
CHAPTER 9
A COMPARISON OF NARRATIVE-BASED
VERSUS NONNARRATIVE-BASED
INFORMATION PROCESSING
A narrative account of a crime is of particular relevance in assessing the
likelihood that the defendant was motivated or able to commit it. Participants in Pennington and Hasties studies may conceivably have tried to
construct a narrative account of the events that occurred regardless of how
the information was presented, but simply found it easier to attain this objective when the information was conveyed in temporal order. As these authors point out, however (Hastie & Pennington, 1995), many of the decisions
we are called upon to make in daily life are not typically based on narratives. Rather, they are based on spatial, functional, hierarchical, and argumentative criteria. The mental representations that are formed in the
course of applying these criteria may be quite different. Thus, for example,
computer commands might be cognitively organized on the basis of functional considerations, and the arguments bearing on a particular issue
might be organized according to the alternative points of view they support. Goal-related differences in the representations that people form of information, which were discussed in chapter 6, are considered more generally by Pennington and Hastie (1993).
Even within a knowledge domain, a number of alternative strategies
might be used to compute a judgment or decision. For example, the evaluation of a consumer product could be based on either an imagined sequence
of events that surround the use of the product and its consequences or a
piecemeal analysis of the products individual features (cf. Fiske &
Pavelchak, 1986). In such cases, the strategy that people use, and the mental representation they construct in the course of applying it, may depend
on situational or informational factors that exist at the time the information
is received. Two factorsthe format in which the information is presented
and the presence or absence of picturesare of particular interest.
1. Effects of Presentation Format
on Processing Strategies
When information describes a temporally ordered sequence of events
that are situationally constrained, people may spontaneously construct an
episode model of this sequence in the course of comprehending it (Postulate 4.1). Under these conditions, people are likely to base their judgments
on the implications of the model as a whole without analyzing its individual
features. In contrast, suppose individuals receive the same information in
an ostensibly unordered list. In this case, they may not attempt to assess
235
the temporal or thematic relatedness of the events unless this relatedness

is particularly relevant to the goal they are pursuing. Instead, they may
adopt a piecemeal processing strategy. That is, they may assess the
evaluative implications of each piece of information independently and integrate these implications using a mechanistic procedure similar to that described by algebraic models of information integration (cf. Anderson, 1971,
1981; Fishbein & Hunter, 1964).
Both holistic and piecemeal processing strategies may become very well
learned as a result of numerous past experiences in which they have been
applied. Personal experiences are inherently in the form of a narrative, and
people have undoubtedly acquired procedures for construing their implications that they activate and apply automatically to new experiences that
are similar in form. Verbal descriptions of temporally related events of the
sort that compose a personal experience could likewise stimulate a holistic
inference process. However, people also learn procedures for aggregating a
number of unrelated items. This learning could occur in both formal situations (e.g., solving arithmetic problems in school) and informal ones (e.g.,
making a decision on the basis of a consideration of its possible consequences, or estimating whether one has enough money to pay for the groceries one wishes to purchase). Once learned, these procedures could be
used to construe the implications of information that is conveyed in a manner similar to that in which the procedures have typically been applied in
the past. To this extent, the procedures could function as if [X], then [Y]
productions of the sort proposed by Anderson (1983; see also Smith, 1984,
1990), which are activated and applied spontaneously whenever the eliciting conditions for their application [X] are met. The format in which information is received could be one of these eliciting conditions.
2. The Role of Visual Imagery
The mental representations that people form in the course of using these
procedures are likely to differ in an important respect. Event and episode
models consist in part of visual images (Postulate 4.2). Pictures of the
events to which these models refer are likely to facilitate the construction
of the models and to increase the detail with which their referents are encoded into memory. To this extent, the pictures should increase the extremity of the evaluations that are based on these models. In contrast, the
information that is conveyed in a list, which is not temporally related, may
be represented only semantically, as a network of features. Moreover, the
piecemeal integration procedure that people activate and use to integrate
the implications of these semantic features may not normally involve a consideration of mental images. To this extent, pictures might not facilitate a
computation of evaluations on the basis of this strategy. In fact, they could
even interfere.
236
CHAPTER 9
The role of pictures and visual images in the computation of evaluations

is surprisingly unclear. Lampel and Anderson (1968) applied an information
integration model in assessing the way in which pictures and personality
trait descriptions combined to influence female participants judgments of a
mans datability. Their data were consistent with an averaging model in
which the weight attached to the mans physical appearance was inversely
related to its favorableness. (More generally, women appeared to perceive
the mans physical attractiveness to be a necessary but not sufficient condition for dating him.)
In Lampel and Andersons study, however, pictures were an independent
source of information about the person being rated and, therefore, were
used in addition to the verbal information provided. When pictures simply
elaborate the same attributes that are described verbally, their impact is
not as clear. Studies of consumer behavior suggest that when pictures accompany nonnarrative text (e.g., product attribute descriptions), they often
have little impact on judgments over and above the effect of the text information alone (Edell & Staelin, 1983; see also Costley & Brucks, 1992). In fact,
this impact may only be appreciable when recipients have little personal interest in the verbal information they receive about the products they evaluate (Miniard et al., 1991). (In these latter conditions, recipients might tend
to base their judgments on pictures alone without bothering to assess the
implications of the verbal information at all.)
Two series of experiments by Adaval and her colleagues examined the
interactive effects of pictures and information presentation format in some
detail. In one set of studies (Adaval & Wyer, 1998), participants evaluated
vacations of the sort they might personally experience. In a second set
(Adaval, Isbell, & Wyer, 2003), participants evaluated politicians on the basis of information about events that occurred over the course of their lives.
Thus, the two sets of studies differed in the extent to which situationspecific episode models were likely to be formed in the course of comprehending the event descriptions and, therefore, in the likelihood that the
events would elicit visual images spontaneously.
The Role of Narratives in Judgments
of Imagined Personal Experiences
In the studies by Adaval and Wyer (1998), participants read two travel brochures. One brochure described a vacation in India and the other, a vacation in Thailand. Each brochure began with a general overview of the vacation, followed by descriptions of 12 places that the vacationers would visit.
These events were always conveyed in the same order. In one brochure,
however, the events were described in a narrative, and in the other, they
237
were conveyed in an ostensibly unordered list. In narrative-format conditions, for example, the India vacation brochure began:
On your vacation, you will start out from the capital of India, Delhi, and move
on to see the Taj Mahal. Later, you will go west and see the palaces and temples in the colorful deserts of Rajasthan . . . before heading south. Further
south, you will visit the beaches of Goa, the tropical forests and backwaters of
Kerala, and . . . complete your trip at the southernmost tip of India.
This was followed by written narrative descriptions of the places to be visited. For example:
Only a short trip from Delhi is Agra, the home of the Taj Mahal. A mausoleum
built by Shah Jahan for his empress, the Taj is widely regarded as the most
beautiful man-made structure in the world. Remarkable at all times of the day,
you will visit as the sun rises above the early morning mists and return on a
moon-lit night when the Taj seems to float unattached above the blue-green
pools in front of it.
In contrast, the India brochure under list-format conditions described aspects of the vacation without indicating the order in which the events described would occur. This brochure began with a simple listing of highlights:
Some features of your vacation experience are:
a visit to the capital, Delhi
the cool mystery of the forests
the forest tribes
the Taj Mahal at Agra
palaces and temples in Rajasthan, etc.
Moreover, the individual places and situations were described in a similar manner:
Agra, home to one of the most beautiful man-made structures in the world.
the Taj Mahala mausoleum built by Shah Jahan for his empress.
a beautiful spectacle both when the sun rises above the early morning mists
and on moonlit nights when the Taj seems to float above the blue-green
pools in front of it.
238
CHAPTER 9
1. Experiment 1
In the first study, an attractive picture accompanied the verbal description of each event. However, the relative dominance of the pictures and verbal descriptions varied. In some cases, the pictures were large and the verbal descriptions were in small (10-point) font. In other cases, the pictures
were relatively small and the verbal descriptions were in large (18-point)
font. To ensure that format and picture-word dominance were not confounded with information content, the particular vacation representing
each combination of these variables was counterbalanced. Participants after reading both brochures evaluated the extent to which the brochure pertaining to each vacation made them want to go there. Then, they reported
the difficulty they had had in imagining each of the two vacations and finally, recalled the events contained in each brochure.
Processing Efficiency. The ease of processing the information under
each presentation format condition was inferred from both participants
self-reported ease of imagining each vacation and their memory for events
that would occur in it. Data pertaining to these indices are shown in the
first two sections of Table 9.1. When the text information was dominant,
participants found it fairly easy to imagine the vacation regardless of format, and had relatively better recall of the places they would visit when the
information was conveyed in a list. When the information was conveyed in
a narrative, however, increasing the dominance of the picture decreased
the difficulty of imagining the vacation and increased recall of the places to
be visited. When the same information was presented in a list, on the other
hand, increasing the dominance of the pictures had the opposite effects. In
other words, pictures appeared to facilitate the processing of information
that was presented in a narrative. However, they interfered with the semantic, piecemeal processing of this information when it was conveyed in a list.
Effects on Judgments. The combined effects of format and pictures in
evaluations of the vacations are consistent with the processing differences
just described. The bottom section of Table 9.1 shows evaluations of the
first vacation that participants considered.19 When the text was dominant,
participants rated the vacation somewhat less favorably when the information about it was conveyed in a narrative than when it was described in a
list. However, increasing the dominance of pictures increased evaluations
19
Participants appeared to use the same criteria to judge the second vacation they considered that they used to judge the first one, even though the information was conveyed in a different format. Thus, the effects of format on judgments of the second vacation were the mirror image of its effects on judgments of the first. (This was true in the second experiment to be
reported as well.)
19
239

TABLE 9.1
Effects of Information Presentation Format and
Picture-Text Dominance on Recall and Judgments
Difficulty of imagining vacation

Pictures dominant
Text dominant
Proportion of places recalled
Pictures dominant
Text dominant
Evaluations of first vacation presented
Pictures dominant
Text dominant
Narrative Format
List Format
3.2
4.5
6.3
4.7
.52
.48
7.3
6.5
.37
.55
4.1
7.9
Note. Adapted from Adaval and Wyer (1998).
of the vacation in the first case and decreased it in the second. As a result,
evaluations of a vacation when pictures were dominant were much more favorable when the vacation was described in a narrative than when it was
described in a list.
2. Experiment 2
A second experiment confirmed these conclusions. In this study, the
text-dominant condition was replaced by one in which participants received only text information and no pictures were presented at all. In addition, some participants were explicitly told to imagine themselves actually
having the experience described on each page of the brochure. We reasoned that if the effects of picture results from the mental images they stimulate explicitly telling participants to form their own mental images of the
events might have similar effects.
Results confirmed this reasoning. Participants evaluations of the first vacation they considered are shown in Table 9.2 as a function of presentation
format, the presence of pictures and imagination instructions (see Footnote
20). When the vacation was described in a narrative, participants evaluated
it more favorably when both pictures were presented and they were told to
imagine the events described (M = 8.1) than they did under other conditions (M = 6.4). When the vacation was described in a list, however, participants evaluated it somewhat less favorably in the first condition than in
other conditions (5.8 vs. 6.1). Thus, both studies converge on the conclusion that people form mental images of events that are conveyed in a narrative, and that situational conditions that encourage the construction of
these images increase the extremity of their judgments. On the other hand,
these same conditions interfere with the piecemeal processing that under-
240
CHAPTER 9
TABLE 9.2
Evaluations of Vacations as a Function of Presentation Format,
Presence of Pictures and Imagination Instructions
Narrative Format
List Format
8.1
6.3
5.8
6.8
6.4
6.5
6.1
5.5
Imagination instructions
Pictures presented
No pictures
No imagination instructions
Pictures presented
No pictures
Note. Adapted from Adaval and Wyer (1998).
lies peoples computation of evaluations on the basis of listed information

and, therefore, decrease the extremity of these evaluations.
3. Effects of Information Inconsistency
Research in other domains suggests that people who are asked to judge
a stimulus give greater weight to unfavorable information about the stimulus than to favorable information about it (Birnbaum, 1974; Skowronski &
Carlston, 1989; Wyer, 1973, 1974). However, this may only be true when each
piece of information is considered in isolation. If the information is conveyed in a narrative, people presumably base their evaluations on the implications of the sequence of events as a whole, and so unfavorable information that is embedded in otherwise favorable information is likely to
have less impact. Supplementary data obtained in Adaval and Wyers first
experiment confirmed this speculation. In some conditions, descriptions of
two relatively unpleasant aspects of the vacation (e.g., poor living accommodations, an arduous bus ride, etc.) were included among the favorable
event descriptions. Although adding these undesirable features decreased
participants evaluations of the vacation, this decrease was much greater
when the information was listed (from 6.0 to. 4.7) than when it was conveyed in a narrative (from 6.9 to 6.4).
THE INFLUENCE OF NARRATIVES

ON POLITICAL IMPRESSION FORMATION
We speculated that the format in which information is presented could activate a cognitive production that is applied automatically, with little
awareness. On the other hand, it could also induce an implicit expectation
for the type of processing strategy that should be applied, and participants
might consciously attempt to comply with this expectation. In Adaval and
241
Wyers studies, it is unclear which was the case. It was also unclear whether
the facilitating effects of pictures are restricted to conditions in which episode models are likely to be formed from the information presented or
whether their effects generalize to conditions in which the narrative information is unlikely to activate these models (e.g., abstract descriptions of
events that are not temporally or situationally constrained).
Adaval, Isbell, and Wyer (2003) provided evidence of this generalizability. Some of the more important judgments that people are called upon
to make outside the laboratory are their evaluations of candidates for public office. These judgments are presumably based in part on the candidates
stands on social and political issues. Ever since the 1960 KennedyNixon debates, however, it has been widely recognized that voters evaluations of
politicians are influenced at least as much by the global images of the candidates that are largely unrelated to the candidates political ideology or
specific issue positions (Englis, 1994). These images are undoubtedly stimulated in part by a candidates physical appearance, as conveyed in a magazine or on television. However, they can also be influenced by descriptions
of the individuals activities over the course of his or her lifetime that have
implications for intelligence, morality, or steadfastness. Thus, information
that a candidate had smoked marijuana in college, or performed a heroic
deed in World War II, can have an impact on inferences about the candidates character and, therefore, judgments of his or her suitability for office
at a much later point in time.
Adaval et al. (2003) examined the processes that underlie the construction of the images that people form of politicians using procedures similar
to those employed in the vacation studies. The first two studies confirmed
Adaval and Wyers earlier findings and, in doing so, provided evidence that
the processing strategies that participants employed under different presentation format conditions were activated and applied automatically, with
little awareness. Two other studies used recognition response times to understand more specifically the cognitive representations that were formed
of information under conditions in which the strategies were employed. A
final study explored the role of both pictures and mental imagery in political information processing and clarified further the processes that contributed to judgments in the earlier experiments.
Participants in an initial study were told we were interested in the impressions people form of famous personalities. On this pretense, they
read brochures about two political figures who had been well known during the period they held office. Each brochure began with an overview of
the major events in the politicians life, followed by 12 additional pages
each describing an event or situation in more detail. As in the vacation
studies, the information in one brochure was conveyed in a narrative and
the information in the other was conveyed in the form of a list. In narra-
242
CHAPTER 9
tive-format conditions, for example, the brochure describing one politician (Thomas Winters) began:
Thomas Winters was a well-known political figure between 1950 and 1975. He
was a veteran of World War II and served as an executive of General Motors
before becoming Governor of Michigan. He then served two years as a U. S.
Senator, and ended his career as a special envoy to China.
This paragraph was followed by a series of paragraphs, each describing a

different event that occurred during the politicians career and the point at
which it occurred. For example:
He left General Motors to become Governor of Michigan. There, he showed
sensitivity to public interests. Upon assuming office, for example, he went on
television to oppose the construction of a nuclear waste processing plant
near Detroit that would contaminate the citys water supply.
Other activities included urging the government to halt the bombing in Vietnam, donating his summer home for use by a charitable organization, hosting the Pope during his visit to America, and helping to revise the state budget to provide support for crime prevention.
In contrast, the brochure under list-format conditions described the
events in the politicians life in bullet form and did not indicate their temporal relatedness. Thus, the brochure pertaining to Winters began:
Thomas Winters was a well-known political figure between 1950 and 1970. He
was:
a member of the U. S. Senate
a World War II veteran
a General Motors executive
Governor of Michigan
special envoy to China
Although the individual events were conveyed in the same order they were
presented in narrative-format conditions, they were also conveyed in bullets that had no temporal implications:
Was sensitive to the interests of the public while Governor of Michigan.
Went on television to oppose the construction of a nuclear waste processing
plant that would contaminate the citys water supply.
Information about each life event was presented on a different page of

the brochure. In picture conditions, the information was accompanied by a
black-and-white photograph. In some cases, the picture showed the politi-
243
cian engaged in an activity related to the event (giving a speech, talking to

someone, etc.) whereas in other cases, it portrayed an event-related situation in which the individual was not involved. (Thus, for example, a statement that the politician had headed a committee to investigate how to decrease violent crime was accompanied by a policeman at the scene of a
murder.) The pictures were taken from books and magazines. (Pictures of
Henry Kissinger were used for one of the two politicians, and pictures of
Robert McNamara were used for the other. Pretesting indicated that neither
politicians face was familiar to the college student population from which
participants were drawn.)
Participants after reading the brochures reported the favorableness of
their impressions of each politician along a scale from -5 (very unfavorable)
to +5 (very favorable) and estimated the difficulty of imagining the sequence
of events that occurred in the persons life. Finally, they were asked to recall the events they had read about.
As the aforementioned examples indicate, many of the events described
in the information were not situationally and temporally constrained. Moreover, they were widely separated in time and occurred in unrelated situations. Therefore, the events were unlikely to be integrated into a single episode model. Nevertheless, our results were similar to those observed by
Adaval and Wyer (1998). As shown in the top half of Table 9.3, participants
in the absence of pictures evaluated politicians less favorably when the information about them was conveyed in a narrative than when it was listed.
However, introducing pictures increased evaluations in the former condition and decreased it in the latter. As a result, evaluations of the politicians
when pictures were presented were more favorable when the information
was conveyed in a narrative than when it was listed. Although these differences were small in magnitude, the interaction of format and pictures was
reliable (p < .05). Recall data, shown in the bottom half of the table, had a
similar pattern. That is, pictures increased the recall of events that were described in a narrative but not when they were listed, suggesting that they
TABLE 9.3
Impressions of Politicians and Number of Events Recalled as a Function
of Format, the Presence of Pictures, and Presentation Order
Impressions of politician
Pictures
No pictures
Number of items recalled
Pictures
No pictures
Narrative Format
List Format
3.98
3.58
3.63
3.95
5.79
5.04
4.92
4.98
Note. Adapted from Adaval et al. (2002, Experiment 1).
244
CHAPTER 9
facilitated memory for the events in the former case but not the latter. In
contrast to Adaval and Wyers studies (see Footnote 20), these effects were
evident for both the first politician that participants considered and the
second.
Deliberative Versus Spontaneous Influences of Format
on Information Processing Strategies
If participants intentionally use a holistic or piecemeal strategy when they
encounter information in a narrative or a list, respectively, explicit instructions to employ the strategies should override the effects of format. On the
other hand, suppose the strategies have the status of cognitive productions
that are activated and applied automatically when information to which
they are applicable is encountered. Then, they may have an influence regardless of participants conscious attempts to employ other criteria.
To evaluate these possibilities, participants in a second experiment were
exposed to stimulus materials identical to those in the first study. In this
case, however, we told participants explicitly which processing strategy we
would like them to use. Under piecemeal-instruction conditions, we indicated:
Sometimes people form impressions of a person by focusing on specific
events in a persons life (e.g., winning a lottery). At other times, however, they
try to think about the persons life as a whole and whether it was admirable,
successful, happy, etc. In this study, we would like you to use the first strategy. That is, when reading about the political figures described in the brochures,
try to imagine the specific events that occurred in each politicians life and their
implications for your impression of the politician. Use these individual events as a
basis for your impression.
Under holistic-instruction conditions, participants were told:

Sometimes people. . . . happy, etc. In this study, we would like you to use the
second strategy. That is, when reading about the political figures described in the
brochure, try to imagine each politicians life as a whole and use this as a basis for
your impression.
1. Judgments and Recall

In fact, telling participants what processing strategy they should use had
little impact on the judgments they made. Table 9.4 shows participants
evaluations as a function of instructions, presentation format, and the presence of pictures. Results replicated those of the first experiment. That is,
245

TABLE 9.4
Impressions of Politician as a Function of Format, Task Demands,
the Presence of Pictures, and Presentation Order
Schematic instructions
Pictures
No pictures
Piecemeal instructions
Pictures
No pictures
Mean
Pictures
No pictures
Narrative Format
List Format
3.94
3.46
3.61
4.03
3.76
4.00
3.12
4.02
3.85
3.73
3.37
4.03
adding pictures slightly increased evaluations of politicians who were described in a narrative and decreased evaluations of politicians based on information conveyed in a list. As a result, the advantage of a narrative over a
list format was significantly greater when pictures were presented (3.85 vs.
3.37) than when they were not (3.73 vs. 4.02), F(1,56) = 7.49, p < .01. These effects did not significantly depend on the instructions that participants were
given concerning the strategy they should use (p > .10).
The effects of task demands cannot be ruled out entirely, however. Note
that pictures facilitated impressions to the greatest extent when the information was conveyed in a narrative and participants were told to apply a schematic criterion (3.94 vs. 3.46 under picture and no-picture conditions, respectively). Correspondingly, they interfered with impressions to the greatest
extent when participants were told to base their judgments on specific
events and the information as conveyed in a list (3.12 vs. 4.02, respectively).
These data therefore suggest that participants did attempt to apply the criteria they were told to use, and these conscious efforts added slightly to the
spontaneous effects that were induced by the format of the information.
However, neither the effect of task demands nor its interaction with pictures
was statistically significant. Thus, the impact of these demands was very
small in relation to the effects of presentation format per se.
Recall data also confirmed the results of the first experiment. That is,
participants recalled a significantly greater number of events when the information was conveyed in a narrative than when it was conveyed in a list
(4.91 vs. 4.22), and this effect was greater when pictures were presented
(4.99 vs. 3.85) than when they were not (4.82 vs. 4.58). However, these effects also did not depend on instructional set.
246
CHAPTER 9
2. Self-Report Data
The failure for participants judgments to be appreciably influenced by
explicit instructions concerning the criteria they should use suggests that
the strategies that are elicited by different presentation formats were applied unintentionally. In fact, the use of these strategies, and the impact of
pictures on their effectiveness, may occur without awareness. Several aspects of the supplementary data we collected suggest this conclusion. For
one thing, participants self-reports indicated they attempted to apply the
same strategy in evaluating the second politician they saw that they had applied in judging the first one, even though the information about the second
politician was conveyed in a different format. In fact, however, their actual
judgments of the two politicians were quite dependent on format in much
the same way.
Second, the effect of format on participants actual evaluations of the
politicians depended on whether pictures were presented but did not depend significantly on task demands (see Table 9.4). In contrast, its effect on
the criteria that participants reported using as a basis for these evaluations
depended on task demands but not on the presence of pictures. Moreover,
analyses of participants estimates of the extent to which pictures facilitated the impressions they formed did not significantly depend on either
task demands or presentation format.
These results therefore argue against the possibility that the effects of
presentation format reflect a conscious attempt to comply with implicit format-based expectations concerning the criteria to be used in making judgments. Rather, different presentation formats appear to activate different
processing strategies that participants apply automatically, and that have
effects independently of the strategies they actively try to use. The presence of pictures may nevertheless facilitate or interfere, depending on the
particular strategy that participants actually employ.
The Role of Pictures in Narrative-Based Processing:
A Reconceptualization
Although the interactive effects of pictures and presentation format in the
preceding studies were very similar to those observed by Adaval and Wyer
(1998), they nevertheless raise questions about the content and structure of
the mental representations that underlie these effects. The assumption that
the event descriptions conveyed under narrative-format conditions were
stored in memory as a sequence of temporally related events was confirmed by Adaval et al. (2003) in a study of recognition memory. Participants were exposed to stimuli under conditions comparable to those constructed in the first experiment by Adaval et al. (2003). Rather than judging
the politicians, however, they were given a timed recognition memory test
247
for the verbal descriptions of the events presented. The order in which the
recognition items were presented was systematically varied. We assumed
that if two presentation items were associated in memory, exposure to one
of these items would spontaneously activate the second, and so the speed
with which the second item could be identified would be increased. When
the original information had been conveyed in a narrative, participants
were quicker to identify an event if it was preceded by a description of the
event that had come before it in the presentation sequence (M = 3.68 s) than
if it was preceded by a different item (M = 4.16 s). When the items had been
conveyed in a list, however, this difference was negligible (4.21 s vs. 4.18 s).
These data suggest that the events were organized in memory in a temporal sequence when they were conveyed in a narrative, but that this was not
the case when the events were presented in a list.
The influence of pictures on the representations that participants
formed is less clear. We assumed that when information was conveyed in a
narrative, participants would form mental images of the events and that
these images would become part of the representations of these events
that they stored in memory. Unlike Adaval and Wyers (1998) studies, however, many of the pictures used in the present experiment were only indirectly connected to the event descriptions they accompanied. (E.g., the verbal description of Harrison as displaying courage under enemy gunfire in
World War II was accompanied by a picture that showed him in an army
uniform but did not portray him as actually engaged in combat. Similarly,
the verbal description of him as showing sensitivity to the needs of the
poor and disabled, and as initiating legislation to provide assistance to
homeless children, was accompanied by a picture of him seated a desk
rather than as actively involved in this behavior.) Thus, although the pictures may have facilitated the construction of mental images of the events
presented, their specific content may not itself have been part of the representation that was formed and stored in memory.
A second recognition memory study confirmed this speculation. In this
study, both pictures and verbal event descriptions were used as test stimuli. Participants did not recognize an event description any more quickly if
it had been preceded in the recognition series by the picture that had accompanied it in the original stimulus materials than if it had been preceded
by a novel picture, and this was true regardless of whether the events had
been conveyed in a narrative (1.84 s vs. 1.81 s) or in a list (1.66 s vs. 1.60 s).20
In other words, the specific picture that accompanied the event description
20
The generally shorter recognition times in this study than the preceding one is attributable
to the fact that American students participated in this study and Hong Kong Chinese students in
the earlier one. The latter students generally took longer to read the English phrases than Americans did.
20
248
CHAPTER 9
was not associated with the description in memory under either format
condition.
These data therefore suggest that pictures played a somewhat different
role in the representations formed under these conditions than they played
in Adaval and Wyers (1998) research. The nature of this role is suggested
by a conceptualization of visual information processing by Barsalou (1993).
That is, a photograph of a person might stimulate the formation of a perceptual symbol that people mentally manipulate along with other symbols
to construct a visual image of the persons activities in not only the situation to which the picture pertains but other situations as well. (This could
occur in much the same way that a previously formed mental picture of a
colleagues physical appearance stimulates an image of his or her actions in
a story that someone tells.) Moreover, this perceptual symbol may be applied across situations, establishing a stronger connection between the
events that occur in a narrative than would otherwise exist. Pictures of a
person could create such a symbol and, therefore, could facilitate the construction of a visual image of events in which the individual is involved
even if they are not themselves directly related to these events.
If this is the case, however, pictures of the person should facilitate the
construction of a coherent image-based narrative representation of the persons life even when the pictures do not become associated with the specific events that accompany them. For that matter, pictures of the politician
should have similar effects even if they do not accompany the written event
descriptions at all, but are presented at the outset, before any of these descriptions are provided.
These speculations were confirmed in an additional study described in
the next section. In this study, pictures did not accompany the verbal descriptions of the politicians life events. In some cases, however, four photographs of the politician were provided at the outset, before the event descriptions were presented. Each picture provided a clear image of the
politicians physical appearance but contained few other indications of the
context in which the pictures were taken.
Relative to conditions in which no pictures at all were presented, introducing pictures at the outset increased the favorableness of participants
evaluations of the politicians, and this was true both when the information
was conveyed in a narrative (3.56 vs. 2.88) and when it was listed (3.35 vs.
2.56). Thus, pictures that were presented at the outset had the same influence under narrative-format conditions that they had when they were conveyed in the context of the event descriptions. Furthermore, presenting pictures before the event descriptions in list-format conditions eliminated the
interference effects that occurred when the pictures accompanied these descriptions, leading the pictures to have a positive influence on judgments
under these conditions as well.
249
These results indicate that the influence of pictures in Adaval et al.s

(2003) other studies did not derive from the information they provided about
the events themselves. Nor did it result from the evaluative implications of
the pictures per se; considered out of context, the pictures were evaluatively
neutral and conveyed very little information about the politicians they depicted. Rather, the pictures stimulated the formation of perceptual symbols
(Barsalou, 1993) that people used to construct mental images of the politicians engaging in the events they later read about, in much the same way
that the picture of a character in a novel helps the reader to imagine verbal
descriptions of the characters actions in the text. This facilitation can occur
both when the events are conveyed in a narrative and when they are not. On
the other hand, accompanying listed events with pictures appears to interfere with the piecemeal integration processes that are activated by this presentation format and occur on-line, as the information is presented and a
judgment is computed. The adverse effects of this interference override the
beneficial effects that the pictures might otherwise have.
THE ROLE OF VISUAL IMAGERY ON VERBAL
INFORMATION PROCESSING: ADDITIONAL
CONSIDERATIONS
The conclusions drawn from the results just described call attention to the
need to examine more closely the impact of visual images on the processing of verbal information. Two quite different studies in the political domain
bear on this impact. The first identifies the influence of individual and situational differences in the tendency to form visual images on the impact of
verbal event descriptions. The second shows the way in which visual images can influence the implications that people extract from other types of
verbal information.
Individual and Situational Differences in Visual Imagery
and Its Effects on Information Processing
To reiterate, the influence of pictures on judgments appears to occur because the pictures facilitate the construction of mental images of events
that are described verbally. If this is so, other factors that influence the construction of images should have similar effects. Some evidence of this was
obtained in the vacation study we described earlier (Adaval & Wyer, 1998,
Experiment 2). That is, the effects of instructions to imagine the events
were similar to the effects of presenting pictures. This possibility was examined in additional conditions of the experiment described at the end of the
previous section. In these conditions, we took into account by both instructional demands to form mental images of the events and chronic individual
differences in the disposition to construct them spontaneously.
250
CHAPTER 9
To reiterate, participants made judgments of politicians on the basis of

verbal materials identical to those we employed in the earlier studies. In
this case, however, pictures of the politicians were either presented at the
outset, before participants read verbal descriptions of the politicians life
events, or not at all. In addition, participants in imagination conditions were
explicitly told to imagine what the person must have looked like and what
he must have done at each stage of his life. . . In other words, try to imagine
that you are watching a movie . . . and form a picture in your mind of the
events described. In contrast, participants under no-imagination conditions
received instructions similar to those in the earlier experiment, without further elaboration.
Finally, participants after making judgments of the politicians were administered a questionnaire developed by Sheehan (1967) to assess individual differences in information-processing style several items of which assessed the disposition to use visual images in responding to information
(e.g., I like to picture future events or situations in my mind, When I think
of someone I know, I often picture in my mind what they look like, My
thinking often consists of mental pictures or images, etc.). Participants
whose scores were above and below the mean response to these items
were classified as visualizers and nonvisualizers, respectively.
Participants impressions of each politician were analyzed as a function
of picture conditions, imagination instructions, presentation format, and
the disposition to form visual images. No effects of format were significant
(p > .10). However, the effects of pictures, imagination instructions, and
chronic dispositions to form images were highly interactive, F(1,47) = 5.11, p
< .03. These effects are summarized in Table 9.5. Visualizers evaluated a politician more favorably when they were told to imagine the events that ocTABLE 9.5
Evaluations of Politicians as a Function of Imagination Instructions,
the Presence of Pictures, and Disposition to Form Visual Images
Visualizers
Pictures
Narrative format
List format
Mean
No pictures
Narrative format
List format
Mean
Nonvisualizers
Imagination
Instructions
No Imagination
Instructions
Imagination
Instructions
No Imagination
Instructions
3.33
3.50
3.42
3.08
2.83
2.96
2.73
3.07
2.90
4.04
3.87
3.96
3.08
3.71
3.40
3.10
2.67
2.88
3.83
3.54
3.68
2.67
2.46
2.56
251
curred in his life than when they were not, and this was true regardless of
the format in which the information was conveyed. Moreover, pictures had
no effect whatsoever on these participants evaluations. That is, visualizers
apparently formed clear images of the events even in the absence of pictures, and so making the pictures available had little additional effect.
In contrast, pictures had a positive influence on nonvisualizers evaluations when these participants were not explicitly told to imagine the politicians life events. Furthermore, imagination instructions had a positive effect provided no pictures were presented. When nonvisualizers were asked
to imagine the politicians life events and pictures were presented, however, the two sources of mental images appeared to interfere with one another, thus eliminating the positive effect that each factor had in isolation.
As noted earlier, the verbal information we presented typically described abstract events that were unlikely to elicit visual images spontaneously. Nevertheless, chronic visualizers found it easy to construct images
when they were asked to do so, and the presence of pictures neither helped
nor hindered them. Therefore, the images that these participants formed increased the extremity of their judgments regardless of whether or not pictures were presented. Nonvisualizers, who do not normally form visual images on the basis of verbal information, were also benefited by doing so
provided no pictures were presented. When pictures were presented, however, these individuals apparently found it difficult to integrate their implications into the images they formed, and this led them to form less clear impressions of the politicians than they might otherwise have done.
Effects of Image-Based Representations
on the Processing of New Information
The preceding series of studies focused on how visual and verbal information about a politician combined to affect his image. Once this image is
formed, however, it can have an impact on the way other information is processed and consequently on the implications drawn from this information. A
study by Wyer, Budesheim, Shavitt, Riggle, Melton, and Kuklinski (1991) is of
interest in this regard. Nonacademic employees were recruited for a study of
the way people make judgments of political candidates on the basis of information of the sort they receive during an election campaign. On this pretense, they received two types of information about a member of the U.S.
House of Representatives who had recently run for the Senate in a neighboring state. First, participants were shown a videotaped nonpolitical speech of
the candidates remarks at a bicentennial celebration at which he was asked
to present an award to a local dignitary. The speech, delivered by an accomplished character actor, was identical in content in all conditions but was delivered in either a forceful, articulate manner that conveyed a favorable im-
252
CHAPTER 9
age or in a bumbling manner, with inappropriate pauses, fidgeting, and other

mannerisms that conveyed an unfavorable image.
The second type of information was conveyed in an audiotaped portion
of a radio program that had ostensibly been sponsored by the League of
Women Voters. In this program, commentators reported the candidates
votes on ten bills that had recently come before the House. Votes on six of
these bills (e.g., a proposal to increase military spending by 15%, a proposal
to allow prayer in public schools) conveyed either a consistently conservative or consistently liberal ideology. (Votes on the remaining four bills had
no ideological significance.)
This information was conveyed in three conditions. In no-delay conditions, participants listened to the radio program describing the candidates
issue stands immediately after they watched the videotape of his speech.
Then, after doing so, they reported their impression of the candidate along
a 100-point feeling thermometer. The procedure in two other conditions
was similar. In delayed-information conditions, however, a 24-hour delay was
introduced between the videotaped speech and exposure to the candidates issue positions, with judgments being made immediately after exposure to the issue positions. In delayed-judgment conditions, participants listened to the candidates issue stands immediately after seeing the speech,
as they did under no-delay conditions. However, a 24-hour delay was induced between this information and judgments.
In all conditions, participants after evaluating the candidate reported
their personal positions on each of the issues to which the candidates
votes pertained. Finally, they indicated their party affiliation and ideological orientation. These latter data, in combination with the candidates issue
stands, were used to define two independent variables. First, participants
reported ideology was coded as either similar or dissimilar to the candidates, as implied by the liberal or conservative orientation of his issue
stands. Second, participants were classified as either generally in agreement or generally in disagreement with the candidates specific issue positions, based on the proportion of ideologically relevant issue stands with
which they agreed. (Each participant agreed with at least one liberal and
one conservative issue position, regardless of his or her general ideology.
Therefore, ideological similarity and agreement level were relatively independent.)
Evaluations of the candidate are shown in Table 9.6 as a function of each
informational variable and delay conditions. These evaluations were obviously affected by the favorableness of the candidates image as conveyed in
the videotape. Moreover, this effect was greater when the image was salient
at the time of judgment (i.e., under no-delay conditions) than when it was not.
The effects of agreement and ideological similarity are of greater interest,
however. When the candidates image was not salient at the time information
253

TABLE 9.6
Candidate Evaluations as a Function of Delay Conditions, Image,
Agreement With Candidates Issue Stands, and Ideological Similarity
No-Delay
Conditions
Candidates image
Favorable
51.0
Unfavorable
35.6
Participants agreement with issue stands
Agree
42.3
Disagree
33.2
Ideological similarity to candidate
Similar
49.6
Dissimilar
26.0
Delayed-Information
Conditions
Delayed-Judgment
Conditions
53.8
44.6
47.7
42.9
64.5
30.9
48.1
40.1
48.1
47.3
57.3
31.0
about his issue stands were presented (i.e., under delayed-information conditions), participants based their evaluations of the candidate on their agreement with his issue positions, and the candidates similarity to them in ideology had virtually no effect. However, when participants learned of the
candidates issue stands immediately after they viewed his image-inducing
speech (i.e., under no-delay and delayed-judgment conditions), they based
their evaluations on the candidates general ideology, and their agreement
with him on specific issues had little influence. This was true under both nodelay and delayed-judgment conditions. Thus, unlike the direct effects of the
candidates image on judgments, the indirect effects of image on the processing of issue information were not a function of its salience at the time of judgment. Rather, they depended on the salience of the candidates image at the
time the issue stand information was conveyed.
In short, the salience of the candidates image at the time his issue
stands were learned altered the way in which the implications of these issue stands were construed. When a global image of the candidate was not
salient to them, participants assessed their agreement with his stands on
specific issues and based their judgments on this criterion independently of
the ideological implications of the candidates positions. When the candidates image was salient to participants at the time they heard about his issue positions, however, they applied a global criterion in assessing the implications of his issue positions as well. Consequently, their agreement with
the candidate on specific issues had relatively little effect.21
21
An alternative interpretation of these results might be that participants experienced overload when the candidates videotaped speech and his issue stands were conveyed in temporal
proximity and, therefore, they devoted less cognitive effort to an assessment of the candidates
21
254
CHAPTER 9
Summary and Conclusions

The research described in the preceding sections provides several insights
into the impact of narrative descriptions of a person or object and the influence of visual images on the processes that underlie this impact. At the
same time, it places constraints on the conditions in which these processes
operate. To summarize:
1. When peoples processing objectives require knowledge of the temporal sequence of events that occurred in a particular situation (e.g., the events
surrounding a crime), presenting information in a narrative facilitates the
construction of this sequence. Consequently, it increases recipients confidence in the conclusions they draw from the information and, therefore, increases the impact of this information on judgments.
2. When an evaluation does not require the construction of a temporal sequence of events, this sequence may only be constructed if the events are
described in a narrative that conveys their temporal relatedness. Otherwise,
recipients may employ other strategies for combining the implications of the
information that can often be equally effective (e.g., they may make a piecemeal assessment and integration of the individual pieces of information).
3. Pictures increase the impact of verbally described events when the
events are conveyed in a narrative. This is true regardless of whether the
events are situationally and temporally constrained (i.e., if they stimulate
the construction of an episode model) or refer to more general experiences
that take place over a period of time. However, pictures interfere with the
piecemeal integration of the implications of events that are conveyed in a
list, and consequently decrease the impact of these events on judgments.
4. Pictures can increase the impact of verbal event descriptions when
they are conveyed at the outset as well as when they accompany these descriptions. Furthermore, this increase can occur regardless of the format in
which the descriptions are presented.
5. The information-processing strategies that are stimulated by different
presentation formats may be activated and applied spontaneously, with little conscious awareness. Therefore, their effects are independent of explicit
instructions or situational demands to employ alternative strategies. The influence of pictures on the use of these strategies also appears to occur without awareness.
issue positions. If this were so, however, they would presumably have been inclined to use the
candidates image as a heuristic, leading it to have greater effect on judgments than it otherwise
would. In fact, however, the candidates image had no greater effect under delayed-judgment
conditions (when the two types of information were presented together) than under delayedinformation conditions. Therefore, this alternative interpretation does not seem viable.
255
6. Instructions to imagine the events that are described verbally can influence the impact of these events in much the same way as pictures. However,
when pictures and imagination instructions are both provided, people who
are not normally disposed to form visual images may have difficulty integrating pictures into the images they form when they are explicitly asked to do
so. In this case, therefore, pictures can decrease the impact of the information relative to conditions in which the pictures are not presented.
7. Narrative representations that are formed on the basis of observed
events can induce a holistic processing strategy that, once activated, influences the processing of verbal information that is conveyed subsequently. In
particular, it stimulates the use of more global criteria for judgments, and
correspondingly decreases the use of criteria that require more analytic,
piecemeal processing.
These conclusions may have more general implications. For example, information is likely to have greater impact on judgments and decisions if it
stimulates individuals to form a narrative-based account of an experience,
and to form mental images of the events that occurred, than if it does not.
This possibility, which was suggested by results of the study by Reyes et al.
(1980), was also recognized by Nisbett and Ross (1980). To borrow their example, a person who is considering the purchase of a particular brand of
dishwasher may be less influenced by statistical evidence of its superior
maintenance and repair record than by a neighbors vivid description of her
unique experience with the machine, which broke down 2 weeks after she
purchased it, flooded her kitchen and dining room, and ruined a valuable
rug. As noted earlier, narrative-based information processing may not always
generate more extreme evaluations than other well-learned processing strategies. Nevertheless, as this anecdotal example suggests, when narrativebased representations are easy to construct and are likely to elicit visual images, they are generally more likely to have an impact than abstract information whose implications are more difficult to assimilate and interpret.
GENERAL EFFECTS OF NARRATIVE-BASED

PROCESSING ON EXPLANATION AND
PREDICTION
As Pennington and Hasties research exemplifies, narratives are often constructed in the service of both prediction and explanation. Many examples
of this possibility can be drawn from research in traditional areas of psychology, including (a) the impact of constructing explanatory narratives on
predictions of future events, (b) counterfactual reasoning, and (c) attribution phenomena. Much of this research was not conducted for the purpose
256
CHAPTER 9
of delineating the role of narratives in judgment. Nevertheless, a conceptualization of the research from the perspective proposed in this chapter has
implications for the situational and informational factors that determine
which of several alternative narrative representations are likely to be constructed and used.
The Impact of Constructing Explanatory Narratives
on the Prediction of Future Events
People are likely to predict that an event is likely to occur if they can construct a plausible narrative of the events that might lead up to it. This implies that increasing the accessibility in memory of a plausible narrative account of the event should increase beliefs that the event will occur. Several
studies bear on this possibility. Ross, Lepper, Strack, and Steinmetz (1977)
asked some participants to read a clinical case study with instructions to
explain why the individual might have committed suicide after leaving therapy. Others read the same passage with instructions to explain why the
protagonist might have donated a sizable sum of money to the Peace
Corps. Participants after generating their explanations were asked to predict the likelihood that the protagonist had engaged in a number of behaviors after leaving therapy, one of which was the event they had explained.
Participants were told that the case study had been taken from a textbook
and that neither the experimenters nor anyone else had any knowledge of
what had actually happened to the individual. Nevertheless, participants
typically predicted that the event they had explained was more likely to
have occurred than the events they had not explained. Participants who
were asked to explain an occurrence presumably constructed a plausible
narrative of the sequence of events that led up to it and then, having done
so, used this narrative as a basis for their later predictions.
Results of other studies are consistent with this interpretation. Sherman,
Skov, Hervitz, and Stock (1981), for example, arbitrarily asked participants
to explain why they might either succeed or fail on an anagrams task and
then, after generating their explanation, asked them to predict their actual
performance on the task. Participants predicted the occurrence of the outcome they had arbitrarily been asked to explain. Moreover, their actual
task performance confirmed their prophecy. (That is, participants who had
been asked to explain why they might succeed, and predicted success, performed better than those who had been asked to explain why they might
fail.) Participants apparently used the implications of their narrative-based
prediction as a standard at the time they actually performed the task, motivating them to attain the performance level it implied.
Narrative-based explanations are likely to be generated spontaneously
when individuals receive information that is inconsistent with expectations.
257
In a study by Ross, Lepper, and Hubbard (1975), participants were asked to

distinguish between true and false suicide notes. They were given bogus
feedback on each trial concerning whether they were correct or incorrect.
Some participants were led to believe they had been correct on 12 of 15 trials, whereas others were told they had been correct only three times. Later,
however, participants were debriefed, being told that the feedback they received bore no resemblance to their actual performance. The debriefing
was quite compelling, with participants being shown the preprepared
schedule the experimenter had used to deliver the trial-by-trial feedback. It
was therefore impossible for participants to believe that the feedback they
had received was contingent on their actual performance. Nevertheless,
participants were more likely to predict they would do well on a similar
task in the future if they had been led to believe that they had performed
well on the first task than if they had been led to believe they had done
poorly.
Direct evidence of the cognitive processes that produced these effects
was not obtained in this study. It nevertheless seems reasonable to suppose that participants who received feedback that deviated from their expectations spontaneously attempted to explain it and, in doing so, selectively retrieved a body of self-knowledge about their past experiences that
provided a plausible narrative-based causal account of it. Later, they used
this representation as a basis for their predictions without considering the
validity of the feedback that stimulated its construction.
The Influence of the Ease of Constructing
a Narrative on Its Impact
The use of a narrative as a basis for judgment may depend on not only its
accessibility in memory but also the ease with which it can be applied. Examples of this contingency are suggested by research on the impact of
counterfactual reasoning. People who have a negative experience are particularly likely to ponder the reasons for its occurrence. In doing so, they
may construct a scenario of how the event might have been avoided. If the
story they construct permits them to conclude that the event occurred for
circumstances beyond their control, they may be inclined to dismiss the
event without further ado. However, if they can easily construct a scenario
whereby they could have avoided the unpleasant experience, they may ruminate about their failure to have done so, and may experience upset and
regret.
A well-known experiment by Kahneman and Tversky (1982) provided an
example. Suppose Al and Bob are scheduled for different flights that leave
at the same time. They travel from town in the same taxi, are caught in a
traffic jam and arrive 30 minutes after the scheduled departure. Upon
258
CHAPTER 9
reaching the terminal, Al learns that his fight left on time. However, Bob
finds that his flight was delayed by 25 minutes and left only 5 minutes before he got there. Who is more upset? Ninety-six percent of the participants
who are asked this question agree that Bob would be more upset than Al.
Presumably, the reason is that Bob can easily construct a scenario whereby
he could have saved 5 minutes and caught the plane, whereas Al is less easily able to do so.
For similar reasons, people are more likely to be bothered by a mishap if it
results from an atypical behavior than if it results from a typical one. For example, people imagine being more upset by having a traffic accident while
driving home on a route they take infrequently than by having an accident
while driving home on their normal route. Moreover, people may be more
upset by getting robbed by a hitchhiker if they only rarely pick up hitchhikers than if they regularly do so (Kahneman & Tversky, 1982). In each case,
the difference results from the fact that it is easier to imagine how the event
could have been avoided in the first instance than it is in the second.
The ease of constructing a narrative can also come into play in other
judgments and decision phenomena. Research by Miller (1977) provided an
example. That is, it is easier to imagine how ones donation of money would
help a particular family whose home has been destroyed by fire than to
imagine how it would help to alleviate the suffering of thousands of people
whose homes have been destroyed in Bosnia. Therefore, people are more
likely to contribute money in the first case than the latter. Millers (1977) research provided evidence that this is the case. Evidence that this contingency is recognized outside the laboratory is suggested by the appeals of
charitable organizations to adopt a child, a whale, or even a highway.
This strategy presumably stimulates the construction of a narrative in
which ones own behavior helps an individual person, animal, or highway
maintenance project. Because this narrative is easy to construct, it may increase the willingness to provide assistance.
When two or more alternative narratives are equally easy to construct,
the one that is generated may depend on a number of factors that affect the
relative accessibility of the knowledge that enters into their construction.
Several of these factors, which could be informational, situational, or motivational, are discussed in the sections that follow.
Informational Determinants of Narrative Construction
1. Framing Effects
The narrative that is constructed from information for use in making an
inference can depend to a large extent on features of the information itself
that make different sets of concepts accessible in memory. Examples of this
259
contingency abound. To give a well-known example from research on decision making, consider the following two choice situations:
1. One thousand people are in danger of being infected by a deadly virus.
You have the choice of administering two serums. One serum, A, is effective but in short supply. If you administer A, 350 persons will be
saved. If you choose B, there is a probability of .65 that everyone will be
saved but a probability of .35 that no one will be saved.
2. One thousand people are in danger of being infected . . . If you administer A, 650 persons will die. If you choose B, there is a probability of .35
that everyone will die but a probability of .65 that no one will die.
The choice alternatives are actually identical in each case. In the first
case, however, the description of the alternatives activates concepts associated with saving lives, and so people are likely to construct scenarios in
terms of these concepts. As a result, they are more likely to prefer A, which
guarantees that lives will be saved, than B. In the second case, on the other
hand, the descriptions emphasize the loss of life and activate concepts associated with the avoidance of this negative outcome. In this case, therefore, people typically prefer B, which describes a situation in which no one
will die, than A.
The implications of the narratives that people use to make inferences on
the basis of verbal information can also depend on the perspective from
which the narrative is formed. (For evidence of perspective differences in
the mental representations of verbal information, see Black, Turner, &
Bower, 1979, described in chap. 4.) A study by Read (1985) provides an example. Consider the following vignette:
Helen was driving to work along a three-lane road, where the middle lane is
used for passing by traffic from both directions. She changed lanes to pass a
slow-moving truck, and quickly realized that she was headed directly for another car coming in the opposite direction. For a moment, it looked as if a collision was inevitable. However, this did not occur. Please indicate in one sentence how you think the accident was avoided.
The situation is objectively similar for both cars. However, the story as written focuses the readers attention on Helen, leading a narrative to be constructed in which she is the central figure. Therefore, people who are called
upon to explain how the accident was avoided typically attribute the responsibility of avoiding the accident to Helen rather than the other driver.
If the story had been written from a different perspective, this tendency
would presumably have been less likely.
260
CHAPTER 9
2. Effects of Conversational Norms

on Narrative Construction
A quite different informational influence arises from considerations elaborated in chapter 7. That is, the representation that participants construct of a
social interaction is likely to be guided in part by normative expectations
concerning the type of information they expect to be exchanged (Grice, 1975;
Higgins, 1981; Sperber & Wilson, 1986). These expectations can influence the
concepts and knowledge that listeners use to interpret not only the statements a person asks during a conversation but also the responses to these
statements. This interpretation can often require the construction of a narrative. Kahneman and Miller (1986) recognized this possibility in the context of
norm theory. For example, consider the following example:
Q.
A.
Why did Joan pass the history examination?

She went to bed early the night before.
People who overhear this conversation are likely to assume that the
questioner is seeking new information. That is, they may infer that Joans
passing the exam is somewhat unexpected and, therefore, requires an explanation. They may further assume that the explanation is informative;
that is, it describes behaviors or situational factors that do not normally exist. In interpreting the answer, therefore, they may construct a narrative of
Joan as someone who typically stays out late before exams and is usually
too tired to perform adequately, but who in the present instance went to
bed at a reasonable hour and was sufficiently rested the next morning to do
well. In contrast, consider the conversation:
Q.
A.
Why did Joan pass the history examination?

She stayed up late the night before.
This conversation might stimulate the construction of a scenario of Joan as

someone who normally doesnt study much but, in this particular instance,
decided to burn the midnight oil and consequently performed well. These
narrative-based explanations, which are generated in an attempt to comprehend the situation, are likely to influence judgments of the protagonist
as well as predictions of her behavior in other situations.
Situational Determinants of Narrative Construction
1. Effects of Perspective on Attributions
of Responsibility
The differences in narrative construction suggested by the preceding examples result from the way in which events are described verbally. However, situational variables may have analogous differences when narrative
261
representations are formed on the basis of direct observation. Early studies

of social attribution phenomena (Regan & Totten, 1975; Storms, 1973) provide examples. In Storms (1973) study, some participants watched a videotape of a get-acquainted conversation between themselves (the target) and
another person. The tape was also watched by persons who were not themselves involved in the conversation. The videotape was made from a vantage point that focused attention on either the target or the other. Participants after viewing the tape were asked whether the targets behavior was
attributable to characteristics of the target (personality, mood, etc.) or the
situation (e.g., the other participant). Participants were more likely to attribute the targets behavior to dispositional factors when the target was visually prominent in the tape than when (s)he was not. This was true regardless of whether the participants were themselves in the role of the target.
Analogous conclusions can be drawn from a study by Regan and Totten
(1975). In this case, participants observed a videotape of a conversation involving two other persons with instructions either to observe the target
person or to empathize with this person. Thus, the same information was
available to all participants. Nonetheless, participants who were told to observe the target presumably constructed a representation of the interaction
in which the target was the focus of attention. In contrast, observers who
were told to empathize with the target were likely to imagine interaction
from the targets perspective and, therefore, to construct a narrative that
was focused on the conversation partner. The content of these narratives
then provided the basis for their attributions. Thus, the former participants
attributed the targets behavior to dispositional characteristics, whereas
the latter participants attributed his behavior to situational factors.
Motivational factors can also influence the perspective from which a narrative is constructed. In a study by Wolfson and Salancik (1977), observers
watched an actor perform an achievement task. Some observers expected
to engage in the task themselves at a later time, whereas others did not.
The first group of participants presumably formed a representation of the
situation that focused on the task, whereas the second group formed a representation that focused on the actor. Consistent with this possibility, the
first group of participants was more inclined than the second group to attribute the actors behavior to the task, and were less inclined than the second group to attribute it to the actor himself.
2. Contextual Influences on Narrative-Based
Explanations
The type and implications of narratives that individuals use as a basis
for explanation and prediction can be influenced in part by situational factors that activate different sets of concepts and knowledge for use in con-
262
CHAPTER 9
structing these narratives. Norenzayan and Schwarz (1999) provided provocative evidence of these differences in an analysis of the expectations
that participants generate in psychology experiments. Participants read a
New York Times article about either (a) a disgruntled postal employee who
went on a mass murder rampage or (b) a man who was accused of murder
for bombing a federal office building. The letterhead of the questionnaire
being administered suggested that the study was being conducted by either
the Institute of Personality Research or the Institute of Social Research.
In each case, participants after reading the story listed five reasons why the
protagonist might have committed the crime.
It seems reasonable to suppose that people who believed that the study
was sponsored by a personality research institute would activate concepts
associated with characteristics of the actor and, therefore, would construct
a narrative account of the individual in terms of these characteristics. In
contrast, persons who were told that the study was being conducted by a
social research institute are more likely to activate concepts about social
(i.e., situational) factors and to construct a narrative in terms of these
features. These narratives should then influence the explanations that participants generate. This was in fact the case. Participants gave more dispositional explanations when they believed that the study was conducted
by a personality research institute (M = 4.4) than when they believed it was
conducted by a social research institute (M = 2.6), and gave fewer situational explanations in the former case than the latter (5.8 vs. 9.4).
It is important to mention these differences occurred not only when the
nature of the research institute was identified before participants read the
story but also when it was identified afterwards. As the authors point out,
this could indicate that the differences in attributions reflected a tendency
for participants to tailor their responses to the apparent interests of the individuals to whom they were communicating, and might not reflect a difference in the underlying representation they formed of the interaction. This
possibility cannot be discounted. However, it seems equally likely that participants mentally reviewed and reconstructed the crime after they had
read the story, and that they based their explanations on this reconstruction rather than the representation they had formed earlier.
Be that as it may, these considerations have implications for the interpretation of numerous psychology experiments in which experimenters implicitly or explicitly induce expectations for the criteria that should be
applied in providing information and, therefore, the narrative-based representations they are likely to construct for use as a basis for their responses.
As Schwarz, Strack, Hilton, and Naderer (1991) argued, studies that are ostensibly conducted by psychologists can dispose participants to construct
explanations in terms of personality characteristics of the protagonists
rather than employing other criteria that are less relevant to psychology. If
263
the studies were conducted in a different situational context, this might not
be the case. To demonstrate this, Schwarz et al. introduced some participants to a judgment task with instructions similar to those employed in an
earlier experiment by Kahneman and Tversky (1973):
A panel of psychologists have interviewed and administered personality tests
to 30 engineers and 70 lawyers . . . On the basis of these interviews, thumbnail
descriptions of the [100 individuals] have been written. In your form, you will
find 5 descriptions, chosen at random. . . . For each description, please indicate the probability that the person described is an engineer. . . . the same
task has been performed by a panel of experts who were highly accurate in
assigning probabilities . . .
A second condition was similar except that the first line of the instructions referred to researchers rather than to psychologists, and the last
line referred to statisticians rather than experts. Following these instructions, participants read a sketch of the following type;
Hans K. is 45 years old. He is married and has four children. He is generally
conservative, careful and ambitious. He shows no interest in social and political issues. He spends most of his time on his many hobbies, which include . . .
solving mathematical puzzles.
The authors speculated that participants who believed that the story was
associated with psychologists would construct an account of the individual
in terms of personality and life style characteristics and would use this account as a basis for predicting his vocation. In contrast, associating the task
with statistics was expected to lead participants to use a statistical criterion
(e.g., base rates) as a basis for prediction. This was in fact the case. That is,
.76 of the participants predicted that the individual was an engineer when the
study referred to psychologists, whereas only .55 of the participants made
this prediction when the instructions referred to statisticians.
These data obviously do not provide evidence that a narrative representation of the target person was constructed in one case but not the other.
The results nevertheless provide a general indication that contextual factors can play an important role in the criteria that underlie judgments and,
to this extent, the likelihood that narrative-based explanations are used.
Motivational Influences on Explanation-Based
Narrative Constructions
The scenarios that people construct in the course of explaining events may
not be entirely dispassionate. This is particularly true when the events are
ones in which people are personally involved in the events and the explana-
264
CHAPTER 9
tions they give could reflect positively or negatively on themselves. Individuals in Western cultures typically make defensive attributions. That is, they
take responsibility for their own success but attribute their failures to other
factors that have relatively few negative implications for their self-esteem
(Harvey & Weary, 1984; Zuckerman, 1975). These factors may determine the
narratives that people construct.
The self-serving motives that underlie the construction of scenarios
about oneself may be attributed to others as well. To give yet another example from Tversky and Kahneman (1982):
Tom and Jim were both eliminated from a tennis tournament. Both were eliminated on a tiebreaker. Tom lost when his opponent served an ace. Jim lost on
his own unforced error. Who will spend more time thinking about the match
that night?
Tom can presumably explain his loss in terms of consequences beyond

his control. Jim, on the other hand, may continue to rehearse his error and
to generate alternative ways in which it could have been avoided. Therefore, as Tversky and Kahnemans findings imply, participants expect Jim to
be more likely to think about the match than Tom.
The assumption that people are more motivated to generate scenarios
that reflect positively on themselves, and avoid constructing narratives
that reflect negatively, is indirectly confirmed by Arkin, Gleason, and
Johnston (1976). Participants in this study were induced to administer therapy to another person to eliminate a mild phobic reaction. The therapy was
actually the same in all conditions. However, some participants were led to
believe that the treatment they would use was the only one available
whereas others were ostensibly given several alternatives from which they
could choose. In all cases, participants were told that the treatment they
would use had been typically used successfully or unsuccessfully by others. However, participants after administering the treatment themselves
were told they had either succeeded or failed. After receiving this feedback,
they estimated the extent to which they were personally responsible for the
outcome.
In principle, participants should believe they are more responsible for
the outcome if they personally chose the therapy they use than if they are
required to use it, and more responsible if their outcome differs from others (e.g., if they succeed when others fail, or fail when others succeed) than
when it does not. As shown in Table 9.7, however, this was only true when
participants were told they had failed. Participants took responsibility for
success regardless of whether they had freedom of choice over the therapy
they used and regardless of the outcomes that others had attained.
265

TABLE 9.7
Mean Attributions of Responsibility for Success and Failure
in Administering Therapy as a Function of Freedom
of Choice and Others Outcomes
Personally Succeeded
Choice over therapy

To be administered
No choice
Personally Failed
Others Succeeded
Others Failed
Others Succeeded
Others Failed
6.63
6.38
6.13
6.75
7.13
4.13
4.25
4.63
Note. High scores indicate greater attributions of responsibility to oneself. Adapted from
Arkin, Gleason, and Johnston (1976).
This study obviously does not provide direct evidence that participants
attributions were mediated by a narrative representation that they constructed of the situation. It is nevertheless plausible to assume that participants who did well spontaneously activated a scenario of themselves as
competent individuals who typically succeed in achievement activity and
considered this scenario to be a sufficient explanation for the outcome.
Therefore, because the scenario reflected positively on themselves, they
used it as a basis for judgments without searching for alternative explanations. In contrast, participants were reluctant to construct a scenario that
suggested they were responsible for failure if a plausible alternative could
be generated. Therefore, they only accepted responsibility for their failure
when they had personally chosen the therapy and had failed when others
had done well.
Arkin et al.s study provides only one example of a more pervasive tendency for individuals to construct representations of themselves and others that confirm a perception of themselves as admirable individuals and of
the world as a place in which they can be happy and successful. This tendency, which has been documented in detail by Kunda (1990), is discussed
more fully in the next chapter.
CONCLUDING REMARKS
The studies reviewed in this chapter confirm the assumption that people often attempt to construct a narrative representation of events on the basis
of information they receive about them, and that their judgments are based
on implications of this representation. The events may be ones they imagine themselves or another person experiencing, events they either read
about, or ones they directly observe. It would obviously be wrong to assume that all social inferences are based on these representations. At the
266
CHAPTER 9
same time, the influence of narrative constructions of social experiences on

judgments is incontrovertible.
However, narrative accounts of a new experience, and the inferences
that are made on the basis of these narratives, do not occur in a cognitive
vacuum. These accounts are constructed, and implications are drawn
from them, on the basis of previously formed representations that specify
the temporal and causal relations among events of the sort under consideration. The use of these representations, which may often constitute implicit theories about the world in which one lives, is discussed in the next
chapter.
C H A P T E R
10
The Impact of Implicational Molecules
and Implicit Theories on Inferences
About Oneself and Others
The construction of narrative representations from new information obviously requires prior knowledge about the type of persons, objects, and
events to which the information refers. This knowledge, which often has implications for features that are not specified in the information presented,
can sometimes include episode models of specific past experiences that have
features in common with the new one. However, it can also consist of generalized narrative representations that are composed of object and event concepts that the new information exemplifies. As noted in chapter 1, these latter representations often function as implicit theories that individuals apply in
comprehending new experiences and construing their implications.
This chapter reviews theory and research in a number of areas in which
implicit theories are likely to play a role, including reconstructions of the
past, the use of judgmental heuristics, perceptions of social support, the
correlates of marital satisfaction, and the effects of stereotypes on overt behavior. Before embarking on a discussion of specific inference phenomena,
however, it may be useful to provide a simplified conceptualization of the
processes we assume to underlie the use of previous acquired knowledge
representations in making inferences.
The generalized narrative representations that function as implicit theories
may be abstractions of a number of specific instances of the sequence of
events they characterize (although, as we indicated in chap. 6, these repre267
268
CHAPTER 10
sentations may be formed less often than one might intuitively expect). Others may be formed in the course of goal-directed cognitive activity that
leads the events that compose them to be associated. Many theory-related
narrative representations may be similar to story skeletons of the sort
proposed by Schank and Abelson (1995). Other, simpler theories may be
conceptualized as implicational molecules of the sort described earlier in
this volume. These molecules theoretically consist of two or more verbally
coded segments that combine to describe a temporally and causally related
sequence of events or states of affairs. Although the nature of implicational
molecules has been discussed in earlier chapters, their centrality to the issues of concern in the present chapter arranges a brief recapitulation of
their structure and use in making judgments and decisions.
In combination, the segments of a molecule often exemplify a generalization about the world in which we live. To reiterate an earlier example, the
generalization that smoking causes lung cancer might be represented in the
two-segment molecule:
[P smokes; P has (will get) lung cancer].
Similarly, the generalization that people get what they deserve might be
conveyed in a just-deserts molecule; i.e.:
[P does a bad thing; P encounters misfortune]
or, alternatively,
[P does a good thing; P has good fortune].
The use of such molecules to make inferences is theoretically governed

by a completion principle. This principle, which is consistent with our previous discussion of narrative-based comprehension processes, can be conveyed as a postulate:
Postulate 10.1. If the information available about a specific situation instantiates all but one of the propositions that compose a molecule, an instantiation of the remaining proposition will be spontaneously inferred.
The completion principle theoretically holds regardless of which propositions are instantiated by the information available and which must be inferred. Thus, in the previous example, people who learn that a person has
committed a nefarious act might use a just-deserts molecule to infer that
the person will be punished or will otherwise experience misfortune. However, they might also use this molecule to infer that a person who has expe-
THE IMPACT OF IMPLICATIONAL MOLECULES
269
rienced misfortune has done a bad thing or is reprehensible for other reasons. This possibility is consistent with the just world hypothesis (Lerner et
al., 1976; Lerner & Simmons, 1966). In support of this hypothesis, Walster
(1966) found that people evaluate the innocent victim of an automobile accident more unfavorably if the individual was seriously injured than if he or
she was only slightly harmed. Research with similar implications was reported by Lerner and Simmons (1966) and also by Wyer, Bodenhausen, and
Gorman (1985) in a study described in chapter 3.
A study by Spiro (1977) also suggested an application of the completion
principle. Protagonists read a story about an engaged couple with instructions that the experiment concerned reactions to situations involving interpersonal relations. In some versions of the story, the man informed the
woman that he didnt want children, whereupon the woman expressed considerable upset and a bitter argument ensued. After reading the story, participants engaged in routine activities unrelated to the experimental task.
During this period, the experimenter, who ostensibly knew the couple incidentally noted that they eventually married and were still happily together.
Participants who are motivated to comprehend the couples relationship
are likely to construct a scenario of how this unexpected outcome might
have occurred. In doing so, they might invoke a previously formed kissand-make-up molecule of the form:
[People love one another; people fight; people resolve conflict; people make
up].
To this extent, they may infer an instantiation of the proposition people resolve conflict, which is not implied by the information available. For example, they might speculate that one of the partners had a change of heart,
that the woman found she couldnt have children, or that some other unforeseen event occurred that stimulated the couple to resolve their conflict.
The inferred event might then be added to the mental representation that
participants construct of the couples relationship and store in memory.
Participants after completing the activities they were asked to perform
were dismissed. However, they returned for a second experimental session
a few days or several weeks later, at which time they were asked to recall
the story they had read. They were explicitly told to report only things that
were mentioned in the story and not to include any personal reactions or
inferences they might have made. Nonetheless, participants made frequent
errors, the number of which increased over time. These errors were typically of the sort one might expect as a result of attempts to reconcile the experimenters incidental comment with the original information. For example, one person recalled that the problem was resolved when they found
that [the woman] could not have children anyway. Another reported that
270
CHAPTER 10
although one person thought the matter was important, the other did not
(see Spiro, 1977, for additional examples).
The completion principle has potentially broad applicability. However,
additional considerations arise in applying it to the inference phenomena of
concern in this chapter. For one thing, more than one implicit theory can
potentially be brought to bear on a particular observation. Or, in terms of
our present discussion, a proposition may be contained in more than one
molecule. This is particularly likely when an event has more than one possible cause. For example, peoples theories about poor academic performance might be reflected in the molecules:
[P doesnt study; P gets poor grades]
and
[P is stupid; P gets poor grades].
In such instances, the information that a particular person received a poor

grade could activate either of these molecules. Assuming that each molecule is equally applicable, the molecule that comes to mind first is most
likely to be applied.
A second consideration is that the inferences made on the basis of an implicit theory may often be probabilistic. In some cases, the information
available may not be sufficient to determine a theorys applicability. (In the
case of an implicational molecule, for example, more than one segment
might be uninstantiated.) In other cases, the information to which a theory
is applied might not necessarily be valid. People who hear that an acquaintance did not study for an exam could use the molecule described earlier to
predict that the person will get a bad grade. At the same time, they are undoubtedly aware that a failure to study does not always lead to bad grades.
Moreover, they might not be completely convinced that the information
they have received about the individuals study behavior is valid. Therefore, although they might use an implicational molecule to generate a prediction, they may not be completely certain their prediction is correct.
When an implicational molecule consists of only two propositions, a formal model of probabilistic inference proposed by Wyer and Hartwick (1980;
see also Wyer, 1974; Wyer & Goldberg, 1970) takes this possibility into account. The conceptualization, which is essentially an extension of
McGuires (1960) probabilogical model, provides a quantitative description
of the relations among beliefs in causally related propositions. Suppose an
implicational molecule has been formed as a result of associating two propositions, A (e.g., war is good for the economy) and B (America is likely to
go to war within the next 5 years). If this molecule is accessible in memory
271
at the time an individual is asked to predict the likelihood of B, the individual is likely to identify and use A as a basis for this inference. Suppose, however, that the individual is uncertain that A is true. Then, the person is
likely to consider not only the likelihood that B would occur not only if A is
true but also if A is not true. Thus, in our example, people who are asked the
likelihood that American will go to war might base their belief on not only
the likelihood that this will occur if war is good for the economy, but might
also take into account the possibility that America would go to war even if
war were not good for the economy. The relative weight attached to these
two conditional beliefs may be a function of the relative likelihood that A is
or is not true (i.e., that war is or is not good for the economy, respectively).
The quantification of this process proposed by Wyer (1974; Wyer &
Goldberg, 1970) assumes that if beliefs are defined in units of probability,
then the belief in the event described by one proposition, B, is related to beliefs in a possible cause of this event, A, according to the equation:
P(B) = P(A)*P(B/A) + P(~A)*P(B/~A),
[10.1]
where P(B) is the belief that A is true, P(A) and P(~A) [= 1 - P(A)] are beliefs
that A is and is not true, respectively, and P(B/A) and P(B/~A) are beliefs
that B is true if A is and is not true, respectively.
This equation provides a remarkably accurate quantitative description
of the relations among these beliefs and, therefore, the influence of beliefs
in one proposition on beliefs in a second (for summaries of this evidence,
see Wyer & Carlston, 1979; Wyer & Hartwick, 1980). In one study, for example (Wyer, 1970), participants read scenarios composed of two parts. The
first part led participants to believe that a particular event (A) was unlikely
to occur and, in addition, implicitly conveyed the likelihood that a second
event, B, would occur if A did and did not occur. The second part of the
story provided additional information that increased beliefs that A would
occur without changing its implications for B. In neither part of the story,
however, was the target event itself explicitly mentioned. One story, for example, concerned the likelihood that a student riot will occur at a particular
university (A), and the likelihood that the university president would be
fired (B). The first part of the story described the president as coercive, establishing a low initial belief that a riot would occur. However, the second
part provided information that substantially increased the belief that a riot
would occur without affecting perceptions of whether the president would
remain in office.
After reading each part of the story, participants reported their belief
that B would occur, followed by their belief in A and the two conditional beliefs defined in Equation 10.1. These estimates, along a scale from 0 (not at
all likely) to 10 (extremely likely), were divided by 10 to convert them to
272
CHAPTER 10
units of probability. The predicted belief in B, P(B), was then computed for
each participant separately by combining his or her beliefs in the manner
described by the right side of Equation 10.1. Mean obtained values of P(B)
are plotted as a function of mean predicted values in Fig. 10.1. Changes in
beliefs in B after reading the second part of the story are also shown in this
figure. In each case, the standard error of the difference between mean predicted and mean obtained values was less than .05 (half a scale unit). This
accuracy was obtained without the use of any ad hoc curve-fitting parameters.
Further evidence bearing on the validity of Equation 10.1 as a description
of the relations among beliefs in causally related propositions is provided
by research on the Socratic effect (McGuire, 1960; Rosen & Wyer, 1972).
That is, peoples a priori beliefs in a set of propositions may not conform to
the relations specified in Equation 10.1 because the propositions have not
recently been thought about in relation to one another. However, if peoples beliefs in the propositions are called to their attention in close temporal contiguity, they may recognize their inconsistency and, consequently,
may modify one or more of the beliefs to eliminate this inconsistency. This
should be indicated by an increase in the quantitative accuracy of Equation
10.1. Rosen and Wyer (1972) found evidence that this is the case. That is,
participants reported their beliefs in causally related propositions in two
experimental sessions 1 week apart. The accuracy of Equation 10.1 in describing the relations among these beliefs was significantly greater in the
second session than it was at first, suggesting that the consistency of the
beliefs increased over time.
A later study (Henninger & Wyer, 1976) is potentially more relevant to
the concerns of this chapter. Participants in some conditions were asked to
report their belief in a proposition occupying the position of A in Equation
10.1 (i.e., an antecedent) before reporting their belief in B (the consequent).
These participants appeared to use A as a basis for their belief in B when
they encountered it later in the questionnaire (as inferred from the accuracy of Equation 10.1 in describing their relationship). Other participants,
however, reported their belief in B at the outset, when A was not salient to
them. These participants typically used a different criterion for evaluating
the validity of B, and so their belief in this proposition was inconsistent
with the belief in A that they reported later.
Henninger and Wyers findings suggest that Equation 10.1 might be used
to determine empirically the criteria that people bring to bear on their beliefs in a proposition they are asked to evaluate. Suppose two or more
implicational molecules could potentially provide the basis for an inference. A comparison of the quantitative accuracy of Equation 10.1 in describing the relations among the beliefs in the propositions composing each molecule could provide an indication of which molecule was actually activated
273
FIG. 10.1. Beliefs in a Proposition B, and changes in these beliefs, as a function of predicted values based on Equation 10.1. Based on data reported by Wyer (1970).
274
CHAPTER 10
and used. The utility of the equation is obviously restricted to theories that
concern only two causally related events or states. Many implicit theories
are more complex. However, if these theories can be broken down into subsets of causally related propositions, the conceptualization could be useful
in evaluating the assumptions underlying the application of these theories
as well.
EFFECTS OF IMPLICIT THEORIES

ON RECONSTRUCTIVE MEMORY
Implicit theories may sometimes be more accessible in memory than the
actual events that led them to be formed. To this extent, they may often be
retrieved and used as a basis for reconstructing the events that occurred in
a particular instance without consulting the original memory trace of the
events at all. That is, suppose people are asked to recall a specific event to
which an implicit theory is relevant. They may find it easier to employ the
theory to reconstruct the event than to retrieve the event itself, which may
be relatively less accessible in memory. (This speculation is consistent with
the effect of generalized representations of an observed experience on
memory for the experience; see Adaval & Wyer, 2003, discussed in chap. 6.)
Reconstructing the Past
Several intriguing demonstrations of the role of implicit theories in memory
have been reported by Michael Ross (1989) in the context of discussing the
impact of these theories on the recall of personal experiences. Two studies
are particularly provocative. In one study, female participants who had previously reported their typical emotional reactions during the period of their
menstrual cycle were asked to keep a daily diary of their moods over the
course of a month. At the end of the month, they were asked to recall the
moods they had experienced during this period. Participants recall was
better predicted by their implicit theories about their emotional reactions
during the time of their menstrual cycle than by the actual feelings they had
reported experiencing at this time.
Students in a second study (Conway & Ross, 1984) participated in a program that they believed would increase their study skills. After participating, they were asked to recall their preprogram estimates of their skills.
Their recall was governed primarily by their implicit theories that the program would be effective. Thus, participants whose skills after participating
did not actually change over the course of the program recalled their preprogram skills as lower than they actually were, consistent with their theory that they had improved.
275
In a study by Goethals and Reckman (1973), students participated in a

group discussion of bussing. The discussion was dominated by a confederate whose opinion contrasted with the attitudes that participants had reported in an earlier session. The confederates view had a substantial influence on not only participants postdiscussion attitudes but also their recall
of the attitudes they had reported earlier. Thus, participants apparently
employed an implicit theory that their attitudes were stable over time, and
therefore used their postdiscussion attitudes to infer what their attitude
must have been before the discussion took place. This interpretation was
confirmed by Ross (1989). Specifically, participants after reporting their attitudes toward the position advocated in a persuasive message were asked
to list the thoughts they had had while recalling the attitude they had reported 1 month earlier. Responses of more than 50% of the participants suggested the use of an implicit temporal consistency theory (e.g., I answered the question now and assumed that my opinion probably hadnt
changed in a month or so.).
Self-Perception Theory
Ross (1989) was among the first to elaborate the role of implicit theories in
reconstructive memory and judgment. However, Bems (1972) well-known
theory of self-perception is based on very similar assumptions. Bem argued
that people who are asked to report their attributes, or alternatively, their
attitude toward a social issue, do not perform an exhaustive review of the
large amount of self-knowledge they have stored in memory that bears on
this characteristic. Rather, they retrieve the most recent judgment-relevant
information that comes to mind and base their judgments on the implications of this information alone. In many instances, the most accessible judgment-relevant information is a recent behavior they have performed. Under
these circumstances, people construe the implications of this behavior for
the judgment they are asked to make, and resort to additional information
only if they consider its implications to be unclear or unreliable.
Bem further assumes that people use the same principles to infer their
own attributes that they use to infer others. Or, in terms of the conceptualization we are proposing, both inferences about oneself and inferences
about others may be based on the same implicit theory about the reasons
people engage in attribute-relevant activity. For example, suppose one person agrees to write an essay in favor of capital punishment even though he
is given the opportunity to write about something else. In contrast, a second person is told to write the essay without being given the chance to refuse. An observer of the events is likely to activate an implicit theory that
people voluntarily advocate positions with which they agree, as reflected in
the molecule:
276
CHAPTER 10
[P believes X; P chooses to advocate X].
In the second case, the observer may activate an implicit theory that people do things because they have to; e.g.,
[A controls P; A desires X; P advocates X].
The first theory has implications for the actors belief, but the second theory does not. Consequently, the observer may infer that the first person is
more likely to believe in capital punishment than the second.
However, now suppose the two individuals in our example are asked to
report their own beliefs in capital punishment. If people use similar implicit
theories to infer their own attributes from their behavior, the first person
should report a stronger belief in capital punishment than the second. Although there are obviously alternative interpretations of this difference
(Festinger, 1957), Bems conception is certainly plausible.
Bem and McConnell (1970) constructed an intriguing empirical demonstration of self-perception processes that also has implications for the role
of implicational molecules in reconstructive memory. Some participants
wrote a counterattitudinal essay under conditions in which they were given
the right to refuse, whereas others were told to write the essay without being given a choice. Then, some participants in each condition were asked to
report their attitude toward the position they had advocated. These participants reported more favorable attitudes toward the position under freechoice conditions than under no-choice conditions, consistent with numerous other studies.
Other participants, however, were asked to recall the attitude they had
reported in an earlier experimental session. These participants recall of
their prebehavior attitudes was affected in the same way that postbehavior
attitudes of the first group of participants were influenced. That is, participants under free-choice conditions recalled their prebehavior attitudes as
similar to the position they advocated in the essay they had written subsequently. This was not the case under forced-choice conditions. Participants
in this study, like those in Rosss studies and the research by Goethals and
Reckman (1973), may have invoked an implicit theory about the attitudinal
antecedents of their behavior and used this theory to infer not only their
postbehavior attitudes but also what their attitudes must have been before
engaging in this behavior.
Further Considerations
The interpretation of research on the effects of behavior on attitudes is often equivocal. That is, it is often difficult to know whether these effects occur after the behavior has been performed or in the course of deciding to
277
engage in it. In the latter case, the results would not indicate an effect of the
behavior per se, but rather, would simply reflect the results of cognitive deliberations that were involved in deciding whether or not to engage in it.
A study by Albarracin and Wyer (2000) circumvented this ambiguity. Participants were told that the experiment was designed to test a new computerized procedure for identifying unconscious behavioral tendencies. The
procedure ostensibly consisted of presenting statements about social issues subliminally and having participants respond to them without being
consciously aware of their content. Participants were told that although the
stimulus statements would appear to be only flashes of light, they would
elicit unconscious feelings that would give rise to a more conscious intuition. They were informed that to provide a measure of their behavior,
they should follow their intuition and generate a yes or no response
to each statement, which would then be interpreted by the computer as a
vote either in favor of or against the issue to which it pertained.
With this preamble, participants were ostensibly exposed to 14 statements that ostensibly concerned seven different university policies. They
were told that the statements might express either support for or opposition to a policy and that the computer program would take this into account in determining the implications of their votes. After the 14 statements had been judged, participants received computerized feedback
about both the policies to which they were subliminally exposed and their
responses to these policies. Participants a priori positions on six of the policies (e.g., maintaining civil liberties on campus, receiving free tickets to
sports events, raising tuition, etc.) were self-evident. Participants were
told that they had voted in favor of or in opposition to each of these policies, depending on which vote was normative. (This established the credibility of the assessment procedure.) The seventh policy, instituting comprehensive examinations on campus, was one on which participants held
varying opinions. Independently of these opinions, however, some participants were told they had voted in favor of instituting the exams and others
were told they had voted against it.
Participants after receiving the feedback were told that to understand
their unconscious decisions, we needed to know their personal feelings
about the policies and, on this pretext, asked them to report their attitudes
toward instituting the exams. In addition, they estimated the likelihood and
desirability of a number of possible consequences of instituting the exams.
Finally, we indicated that because participants had an opportunity to think
about comprehensive examinations during the experiment, we wanted to
see how students might actually vote in a referendum that was likely to be
held later in the semester. Participants were then left alone with instructions to select a slip of paper that represented their choice and to place it in
a secured ballot box. (The box appeared to be partially full. In fact, how-
278
CHAPTER 10
ever, it contained only blank slips of paper, and so we could determine after
the experiment how the participants had voted.)
Thus, the procedure ensured that participants had not thought about
comprehensive examinations before they received feedback about their ostensible voting behavior. Nevertheless, participants reported more favorable attitudes toward the institution of the exams if they had ostensibly
voted in favor of instituting them than if they had voted against. Furthermore, this difference was evident regardless of the attitudes that participants had reported before to the experiment. Finally, their postbehavior attitudes, once formed, were used as bases for their actual voting decisions
that they made at the end of the experiment. (Path analyses indicated that
behavior feedback did not have an effect on participants judgments of specific behavioral consequences, nor did it have direct influence on participants voting behavior. Rather, this latter influence was mediated by the impact of the feedback on participants attitudes.)
This study provided no direct evidence of the cognitive mediators of the
effects we observed. It nevertheless seems reasonable to assume that participants invoked an implicit theory that people publicly advocate positions
they favor, as embodied in the implicational molecule:
[P favors X; P votes in favor of X].
Participants may have used this molecule not only to infer their attitude toward comprehensive examinations from their unconscious behavior, but
also, once their attitude was inferred, as a basis for the overt behavioral decision they made at the end of the experiment.
MOTIVATIONAL INFLUENCES ON IMPLICIT

THEORY CONSTRUCTION AND USE
As we have suggested, the associated cognitions that compose the theories
that people form and use can often result from the experience of these
cognitions, or their referents, in temporal proximity. However, this is certainly not the whole story. As Arkin et al.s (1976) study indicated, people
are often motivated to interpret their experiences in a way that reflects favorably on themselves and their ability to be successful in the world in
which they live. This may be a manifestation of a more general tendency to
construct and apply implicit theories about oneself and the world that confirm the desire to have a happy and successful life.
Kunda (1990) summarized substantial evidence in support of this possibility. For example, although college students know that 50% of all first mar-
279
riages end in divorce, they nevertheless are convinced that they personally
will remain married to their first spouse for life (Kunda, 1987). This suggests
that individuals who are motivated to believe that they will be successful in
marriage, construct implicit self-theories that support this belief. To demonstrate this, Kunda (1987) gave participants information about a target person who was either happily married or divorced and whose demographic
and personality characteristics either matched or did not match those of
the participants themselves. After receiving this information, participants
indicated which of the targets attributes were most likely to contribute to
his or her marital status. Participants were more inclined to attribute the
success of happily married targets to characteristics that matched their
own than to characteristics that did not. Correspondingly, they were less
likely to attribute the failure of divorced targets to the former characteristics. Thus, they constructed theories about themselves that were consistent with the outcome they wished to attain.
People who are motivated to construct and maintain a self-serving theory may selectively search memory for information that supports it. In a
study by Santioso, Kunda, and Fong (1990), some participants were told
that extroversion was conducive to success after leaving college, whereas
others were told that introversion was more conducive to success. Then, in
an ostensibly unrelated study, they were asked to list all of the behaviors
they had performed in the past along a related trait dimension (shy vs. outgoing). Participants listed more behaviors that were congruent with the
trait they were told was conducive to success than behaviors that were incongruent with this trait. In a second study, participants after exposure to
the first task were shown a series of behaviors and were asked in each case
to press a button as soon as they thought of a personal experience that exemplified it. Participants responded more quickly to behaviors that exemplified the success-related trait than to behaviors that exemplified the opposite. Thus, people who are told that a particular trait is associated with
success selectively searched memory for personal experiences that confirmed their possession of this trait. Consequently, these experiences came
to mind more easily when they were called upon to report instances of
their behavior at a later point in time.
Research in other paradigms has similar implications. For example, people are more likely to report engaging in a particular activity (drinking coffee, brushing ones teeth, etc.) if they are told that the activity is healthy
than if they are told it is not (Ross, McFarland, & Fletcher, 1986; B. Sherman
& Kunda, 1989). These studies do not provide direct evidence that these responses were mediated by the construction of an implicit theory. However,
it seems reasonable to suppose that the construction of self-serving theories is a result of similar processes.
280
CHAPTER 10
THE ROLE OF IMPLICIT THEORIES

IN HEURISTIC-BASED JUDGMENTS
The use of heuristics in making inferences is well established (for reviews,
see Kahneman, Slovic, & Tversky, 1982; Nisbett & Ross, 1980; Sherman &
Corty, 1984). As Wyer and Srull (1989) noted, many of these heuristics may
reflect a general tendency to treat conditional relationships as biconditionals. That is, people assume that if A implies B, then B implies A (for evidence, see Wyer, 1977). Thus, for example, if people believe that members
of a certain category have a certain cluster of attributes, they infer that individuals with this cluster of attributes belong to the category. Alternatively,
if they believe that events that are likely to occur are easy to imagine, they
infer that events that are easy to imagine are likely to occur.
The latter, availability heuristic (Tversky & Kahneman, 1973) is of particular interest in the present context. This heuristic can be conceptualized
as an implicit theory in the form of an ease-of-retrieval molecule; specifically:
[X has occurred frequently (infrequently); instances of X are easy (difficult) to
recall].
Thus, people may use this molecule not only to infer the ease of recalling instances of an event from knowledge of how often it has occurred, but also
to infer the frequency of an events occurrence from the difficulty of recalling instances of it.
Two quite different sets of phenomena can be conceptualized as resulting from an application of this molecule. One concerns the effects of television on perceptions of social reality. The other concerns the effects of ease
of retrieving knowledge about oneself and others on judgments of the attributes to which this knowledge is relevant. Research on each set of phenomena is discussed in turn.
Perceptions of Social Reality
Americans watch an average of more than 4 hours of television each day
(Nielsen, 1995). As a result, television is a major source of the episode models that people form. On the other hand, people often fail to distinguish
clearly between the sources of the knowledge they acquire (Hasher, Goldstein, & Toppin, 1977; Jacoby, Kelly, Brown, & Jasechko, 1989; Johnson,
Hastroudi, & Lindsay, 1993). As a result, they are likely to use the episode
models they form while watching television as a basis for judgments without considering the context in which these models were constructed.
L. J. Shrum and his colleagues (OGuinn & Shrum, 1997; Shrum, OGuinn,
Semenik, & Faber, 1998; Shrum, Wyer, & OGuinn, 1998) have examined im-
281
plications of this possibility in the context of their research on the cultivation effect (Gerbner, Gross, Morgan, & Signorielli, 1994). That is, people
tend to overestimate the occurrence of events in the real world that are disproportionately represented on television. Moreover, the magnitude of
their overestimation increases with the amount of television they watch. Instances of events that occur frequently on television are easily accessible in
memory, and this is particularly true for heavy viewers. Therefore, if individuals are asked to infer the frequency of occurrence of these events, and
if they apply the ease-of-retrieval molecule, heavy users are likely to make
higher estimates than light viewers do. Thus, for example, heavy viewers
are relatively more likely than light viewers to overestimate the proportion
of Americans who belong to a country club or have a swimming pool in
their back yard (OGuinn & Shrum, 1997). On the other hand, they are also
more likely to overestimate the incidence of violent crime and the size of
the police force.
Several studies by Shrum and his colleagues support this interpretation.
In one study (OGuinn & Shrum, 1997), frequent and infrequent viewers of
soap operas were asked to estimate the incidence of characteristics that
are associated with an affluent life style (belonging to a country club, owning a luxury car, having a swimming pool, etc.). Frequent viewers not only
estimated the incidence of these characteristics to be higher than infrequent viewers did, but also made their estimates more quickly. These findings are consistent with the hypothesis that frequent television viewers find
instances of these characteristics easier to retrieve from memory and estimate their incidence to be greater as a result.
Other interpretations of these effects should of course be considered.
For example, a third variable, such as education level or socioeconomic
class, could have independent effects on both television watching and frequency estimates, producing an artifactual relation between these variables that would not otherwise exist. However, OGuinn and Shrum (1997;
see also Shrum, Wyer, & OGuinn, 1998) found that the relation between frequency estimates and television viewing was evident even when the potentially confounding effects of these variables were controlled. Moreover,
Shrum et al. (1998) found that when participants were asked to report their
television viewing habits before making frequency estimates (thus calling
their television watching to their attention), the impact of television viewing on frequency estimates was significantly reduced. The effects of watching television on frequency estimates can also be reduced or eliminated by
increasing participants motivation to make correct judgments (Shrum,
1999). Thus, people appear able to distinguish between events they see on
television and those they learn about from other sources, and thus can
avoid the bias produced by television when they are motivated to do so.
Normally, however, this motivation does not exist.
282
CHAPTER 10
Although the effects of television on perceptions of social reality are well

established, a qualification on these effects might be noted. Hamilton and
Gifford (1976) found that the incidence of infrequently occurring events is
overestimated to a greater extent than the incidence of more common
ones. This is presumably due in part to the fact that rare events are thought
about more extensively (for evidence that the recall of information increases with amount of processing, see Craik & Lockhart, 1972). In some
cases, this effect could override the effects identified by Shrum and his colleagues. On the other hand, people are typically passive viewers of television, and are likely to encode and store the information they receive with
minimal cognitive elaboration. To this extent, the effects identified by Hamilton and Gifford (1976) may not be pronounced.
The effects of television on perceptions of social reality have potentially
important implications. For example, individuals who are frequently exposed to television violence may come to view violence and aggression as
normal aspects of life and, therefore, may have more tolerance of aggression than they otherwise would. Zillman and Bryant (1982) found that exposing nave college students to heavy doses of sadomasochistic pornography over a period of several weeks increased their tolerance of rape.
Moreover, these effects were still evident several months after the main
study had been completed. Other effects of the media on attitudes and values are worth examining from this perspective.
Effects of Ease of Retrieval on Perceptions

of Self and Others
Shrum et al.s research suggests that the ease of recalling instances of an
event increases estimates of its frequency. To the extent people apply the
ease-of-retrieval molecule, difficulty in retrieving instances should have the
opposite effect. An intriguing demonstration of this effect was constructed
by Norbert Schwarz and his colleagues (for a review, see Schwarz, 1998). In
a typical study (Schwarz et al., 1991), participants were asked to recall either 6 or 12 instances of assertive behavior. After doing so, they rated their
assertiveness.
Participants obviously recalled more behaviors when they were asked to
recall 12 than when they were asked to recall 6. Therefore, one might expect participants to judge themselves to be more assertive in the former
condition than the latter. On the other hand, it is much more difficult to generate 12 instances of the attribute than 6 instances. Therefore, to the extent
individuals base their judgments on the ease-of-retrieval molecule, they
should infer that they engage in assertive behavior less frequently in the
former condition than the latter. Thus, they should infer that they are less
283
assertive when they have been asked to recall 12 instances than when they
were asked to recall only 6. This was in fact the case.
The use of an ease-of-retrieval heuristic as a basis for judgment is quite
pervasive (for a review, see Schwarz, 1998). Further examples are described later in this chapter. Its implications can be quite ironic. For example, people may be less likely to believe that a proposition is true if they
have attempted to generate a large number of reasons for its validity than
if they have thought about only a few. Research by Wnke, Bless, and
Biller (1996) supported this speculation. Some participants were asked to
generate either 3 or 7 arguments that either favored or opposed a specific
issue, after which they were asked to report their own position on the issue. Other, yoked participants read the arguments that individuals in the
first group had written. The latter participants reported themselves to be
more in favor of the position advocated when they had read 7-argument
responses than when they had read 3-argument responses, confirming the
assumption that the substantive implications of the 7-argument sets were
relatively more persuasive. Nevertheless, the participants who had actually generated the arguments judged themselves to be less in favor of the
position when they had generated 7 arguments than when they had generated only 3. Thus, the effects of ease of retrieval overrode the effects of actual knowledge.
People do not always ignore the implications of their past knowledge, of
course. However, their computation of a belief on the basis of these implications is cognitively effortful. Consequently, they may only do so when ease
of retrieval is likely to be an unreliable criterion. In other conditions of
Schwarz et al.s (1991) research, for example, participants generated instances of assertiveness in the presence of distracting background music.
In this case, participants apparently attributed their difficulty of generating
instances to the distraction and to their lack of knowledge. In these conditions, therefore, they judged themselves more assertive when they had generated 12 instances rather than 6.
1. Effects of Information About Others Ease
of Retrieving Information
As suggested in our discussion of self-perception processes, people can
use implicit theories to construe the implications of others behavior as
well as their own. Support for this assumption in the present paradigm was
obtained by Menon and Raghubir (1998) in a study of consumer judgment.
Participants read an advertisement that described the attractive features of
a computer. Later, they were asked to recall either 2 or 8 of these features.
Either before or after the recall task, however, they were given information
about the difficulty that others had in recalling the products attributes. If
284
CHAPTER 10
participants use the ease-of-retrieval molecule to infer the implications of

this information, they should infer that the computer has more attractive
features when others find it easy to recall these features than when they
did not, and should judge the product more favorably in the former condition. Support for this prediction is conveyed in Table 10.1, which summarizes participants product evaluations as a function of the number of features they were personally asked to recall, the difficulty that others
ostensibly had in recalling attributes, and the point at which the latter information was provided. Participants generally evaluated the product more favorably when others ostensibly found it easy to recall features than when
they found it difficult. However, participants use of the ease-of-retrieval
molecule to construe the implications of their own behavior was evident
only when they recalled the products features before learning about the
difficulty that others had. In this condition, they evaluated the product
more favorably when they had recalled 2 attributes than when they recalled 8. When they were told about others difficulty at the outset, however, they appeared to ignore the difficulty they personally experienced
and based their judgments on the number of features they could recall,
thus reporting more favorable evaluations when they were asked to recall 8
features rather than 2.
In summary, people appear to use the ease-of-retrieval molecule to infer
the incidence of information in memory when (a) alternative bases for judgment are not easily accessible in memory and (b) the difficulty they have in
recalling the information is unlikely to be due to extraneous situational factors. However, when alternative reasons for their subjective feelings are
called to their attention, or when other bases for judgment (e.g., the difficulty that others have had recalling attributes) are easily available, people
apply other criteria instead.
TABLE 10.1
Effects of Subjective Difficulty of Recalling Product Attributes,
and Perceptions of Others Difficulty, on Product Evaluations
Two Attributes
Recalled
Eight Attributes
Recalled
5.9
5.4
5.6
4.9
4.4
4.9
5.4
4.9
5.3
4.3
4.8
5.9
4.9
5.4
5.6
4.6
Others difficulty mentioned after recall

Others find recall easy
Others find recall difficult
M
Others difficulty mentioned before recall
Others find recall easy
Others find recall difficult
M
Note. Adapted from Menon and Raghubir (1998).
285
2. The Role of Ease of Retrieval in Hindsight Bias

The role of ease of retrieval in the construction and use of implicit theories is demonstrated in a series of studies by Sanna and Schwarz (2003;
Sanna, Schwarz, & Small, 2003; Sanna, Schwarz, & Stocker, 2002). People
who know that an event has occurred often overestimate the likelihood
that they would have predicted it (Fischoff, 1975, 1982; for a review, see
Hawkins & Hastie, 1990). This hindsight bias presumably occurs for reasons similar to that we assumed to underlie the effects reported by L. Ross
et al. (1975). That is, individuals who are told that an event has occurred attempt to generate a plausible explanation for it and, if this can be done easily, may conclude that the occurrence of the event was foreordained. In doing so, they ignore the possibility that if the event had not occurred, they
might have generated an explanation of its nonoccurrence just as easily.
If this is the case, however, stimulating individuals to generate additional
explanations for the event, which could be difficult, might decrease or reverse the effects observed by Fischoff (1975). Correspondingly, stimulating
people to generate explanations of why the event might not have occurred
could increase the magnitude of the hindsight bias. Sanna and Schwarz
(2003) found this to be the case. Participants in one series of studies read a
story about a military conflict and were arbitrarily told that one of the two
antagonists had been victorious. Participants were then told to generate either 2 or 10 thoughts about how this outcome might have been averted.
Participants who generated thoughts about the alternative outcome of the
conflict decreased their belief that this outcome could have occurred. However, this effect was much greater when they had generated 10 thoughts
(which was difficult to do) than when they had generated only 2. Thus, generating explanations for how the alternative outcome could have occurred
increased beliefs that the actual outcome was inevitable, thus magnifying
the hindsight bias that participants manifested. In a second series of studies (Sanna, Schwarz, & Small, 2003), however, participants thought about
why the outcome they were given should have occurred. In this case, participants who generated many thoughts decreased their belief that the outcome was inevitable, thus eliminating the hindsight bias that was otherwise
evident.
In summary, people may often spontaneously generate an implicit theory for why an outcome will occur, and this theory may underlie their belief
that the outcome was foreordained (Fischoff, 1975, 1982). However, generating additional explanations for the outcome, which is difficult, can decrease
their belief that the outcome was inevitable, whereas generating reasons
why the outcome might not have occurred is likely to strengthen this belief.
Thus, as also implied by Wnke et al.s (1996) findings noted earlier, the
greater the number of thoughts that individuals generate in support of a position actually weakens their belief that the position is valid.
286
CHAPTER 10
INDIVIDUAL DIFFERENCES IN IMPLICIT THEORIES

To reiterate, more than one implicit theory can often be used to account for
a given social experience. In such instances, individual and situational differences may arise in the theory that people happen to apply and, therefore, in the conclusions they draw from the experience. These differences
are likely to be reflected in their judgments and behavioral decisions.
Implicit Theories of Personal Achievement
Individual differences in implicit theories of social judgment and behavior
were suggested by Carol Dweck and her colleagues (Dweck, 1991; Dweck &
Leggett, 1988; Erdley & Dweck, 1993) in an analysis of individuals responses
to success and failure. Based on Dwecks earlier work on learned helplessness (for a summary, see Dweck & Leggett, 1988), they postulated that some
individuals believe that performance on an achievement task can typically
be increased by exerting effort, whereas others are disposed to belief that
performance is largely a reflection of abilities that are fixed and unable to
change. The first belief might be captured by an implicit theory of the form:
[P performs task; P fails; P tries harder; P succeeds].
The second, however, could reflect the theory:

[P has low ability; P performs task; P fails; P tries harder; P fails again].
Thus, suppose people perform a task and fail. This event (P fails)
instantiates a segment of each of these theories. Individuals who activate
and use the first theory to interpret the event may infer that they did not
work hard enough and that if they exert more effort, they might ultimately
succeed. As a result, they may be stimulated to try the task again. However,
suppose individuals activate the second theory instead. Then, they are
likely to infer that their failure reflects low ability and that repeating the
task will have the same outcome. Therefore, they may be disinclined to try
a second time. Individual differences in the chronic accessibility of these
theories, and their effects on responses to failure, have been elaborated by
Dweck and Leggett (1988).
Incremental Versus Entity Theories
of Personality and Ability
Dweck (1991; Dweck & Leggett, 1988) suggest that the aforementioned performance-related theories exemplify two more general theories of personality and behavior. Incremental theories imply that peoples attributes and
287
abilities are malleable and can be modified, whereas entity theories imply
that attributes are fixed and resistant to change. These different theories
have numerous implications for judgment and behavior, many of which
have been examined by Dweck, Chiu, and Hong (1995a, 1995b; Dweck, Hong,
& Chiu, 1993; Chiu, Dweck, Tong, & Fu, 1997; Chiu, Hong, & Dweck, 1997;
Hong, Chiu, Dweck, & Sacks, 1997).
Hong et al. (1997), for example, found evidence that people who were
classified as entity theorists on the basis of an independent measure were
more influenced by a target persons personality and ability test scores
than incremental theorists were, suggesting that they attached relatively
more significance to measures that ostensibly assess stable attributes and
abilities. People may also have incremental vs. entity theories of morality
(implying that individuals moral character can or cannot be easily modified) and of the world in which they live. A series of studies by Chiu et al.
(1997) provided evidence that individuals with entity theories of morality
and the world are more inclined than individuals with incremental theories to evaluate others behavior in terms of its fulfillment of obligations
rather than in terms of individual rights. Thus, for example, participants in
one study were asked to give open-ended descriptions of how they would
handle hypothetical classroom situations. Some scenarios described socially desirable behavior (e.g., The teacher asked Jerry to remove old papers and notices on the students notice boards, and Jerry does what she
requested as soon as possible.). Others described undesirable actions
(e.g., The teacher asked Peter to remove old papers and notices. . . . A
week has passed and Peter does not do it.). Entity theorists, who believe
that the fulfillment of responsibilities is normative, were less inclined than
incremental theorists to indicate that Jerry should be rewarded for fulfilling his obligations. Correspondingly, they were more inclined than incremental theorists to recommend that Peter be punished for not meeting his
responsibilities.
As Dweck et al. (1995a, 1995b) pointed out, the implicit theories that individuals apply may often be domain specific. Moreover, several alternative theories may coexist, the implications of which can conflict. The individual differences in implicit theories that Dweck and her colleagues
identified in the aforementioned studies could reflect differences in the
chronic accessibility of these theories. To this extent, situational factors
that make one or another theory more accessible in memory might sometimes override these chronic differences. Although this possibility was
not examined in the studies cited in this section, evidence that the effects
of transitory situational factors can often dominate the effects of chronic
accessibility (at least in the short run) is provided by other research described presently.
288
CHAPTER 10
IMPLICIT THEORIES OF SOCIAL SUPPORT:

THE EFFECT OF PERSPECTIVE
ON THEORY ACTIVATION
The implicit theories that people acquire are often the result of accumulated past experiences in a particular type of situation that have been sufficiently involving to stimulate the construction of a generalized event representation of the sequence of events that take place. Individuals are
therefore likely to differ in the nature of these theories, depending on the
past experiences they have typically had. However, the implications of
their theories can also depend on the perspective from which the experiences in question were viewed. An interesting example of these contingencies is provided by Eric Mankowskis research on perceptions of social support (Mankowski & Wyer, 1996). Imagine the following situation:
Russell has learned he got a D on the psychology midterm. He is near tears
when he gets back to his room where his roommate, Chris, is studying. He
tells Chris the bad news. If I dont get good grades, Ill have to drop out of
school. . . . I cant seem to do well here.
Wow, Chris replies, What do you say we go out for a pizza and beer? I
havent had a good day, either.
They go out for pizza. . . . Russell tells Chris, Remember that girl I met at
your party? Shes right over there. . . . We were supposed to go to a movie together, but she never showed up. . . . I really thought she was interested in
me. . . . I dont know if I can eat with her sitting there.
Oh, she wont even notice us here, Chris says. You should get fixed up
with someone. . . . Mark might have a suggestion. He sure seems to have no
problem getting dates.
This story is ambiguous in terms of the social support it describes.

Chriss suggestion to get a pizza, for example, could be viewed as either a
desire to distract Russell from thinking about his bad grade or, alternatively, as insensitivity to Russells need for sympathy and reassurance.
Readers perception of the storys implications could therefore depend in
part on the semantic concepts that they bring to bear on its interpretation.
However, their construal of the storys implications could also depend on
the similarity of the situation it describes to their own past experiences.
For example, people could use a theory about their own experiences as a
standard of comparison in evaluating others. Thus, they might perceive
Chriss behavior to be relatively more supportive if they have often experienced rejection themselves than if they have not. This tendency, on the
other hand, is likely to depend on the whether the new experience is
viewed from the perspective of the potential support provider or the
point of view of the recipient.
289
These contingencies were identified by Mankowski and Wyer (1996). Participants were told that the study was concerned with person impression
formation. On this pretense, they were asked to read a story similar to the
one just described, in which one person was in need of support and the
other was in a position to provide it. In third-person perspective conditions,
the story was written from the perspective of a disinterested observer. In
provider-perspective conditions, however, the provider was referred to as
you and the recipient was identified by a name that was ambiguous with
respect to gender (Chris or Terry). In recipient-perspective conditions, the recipient was referred to as you and the potential provider was identified
by name. After reading the story, participants in each perspective condition
wrote down three adjectives that they would use to describe the provider
and then rated the provider with respect to several traits that were relevant to social support (supportive, warm, rejecting, accepting, sensitive,
etc.). Finally, participants were completed a general index of the support
they perceived to be personally available (Cutrona & Russell, 1987).
Participants perceptions of support availability could reflect two
things. First, it could indicate the frequency with which participants had
personally encountered support-relevant situations in the past and, therefore, the likelihood that semantic concepts associated with social support
were chronically accessible in memory. To this extent, participants perceptions of support availability should have a positive impact on their use
of support-related concepts to interpret the providers ambiguous behavior, and this should be true in all three perspective conditions. On the
other hand, participants perceptions of support availability could also indicate the likelihood that they have constructed an implicit theory of
themselves as recipients of support. If participants who read the story
from the recipients perspective activate and use this theory as a standard
of comparison in construing the storys implications, they should perceive
the situation o be less supportive when their self-referent theory implies
that they typically receive support than when it does not. This contrast effect of the storys implications on judgments could offset the positive effect of semantic concepts on its interpretation that occurs at an earlier
stage of processing.
Results were consistent with this analysis. Table 10.2 shows the supportiveness implied by both participants open-ended trait descriptions of the
support provider and their ratings of the providers supportiveness. When
participants took the perspective of either the provider or third person,
their perceptions of the providers supportiveness increased with their perception of their own support availability. When they took the role of the recipient, however, this was not the case. In these latter conditions, the contrast effect that resulted from using an implicit theory of their personal
experiences as a standard of comparison appeared to offset the positive ef-
290
CHAPTER 10
TABLE 10.2
Perceptions of the Providers Supportiveness as a Function
of Perspective and Social Support Availability
Perspective
Supportiveness of trait descriptions

High support availability
Low support availability
Difference
Supportiveness of trait ratings
High support availability
Low support availability
Difference
Provider
Third Person
Recipient
0.39a
0.18
0.21
0.14
0.26
0.40
0.44
0.26
0.18
0.81b
0.35
0.46
0.26
0.44
0.70
1.07
1.16
0.09
Note. Adapted from Mankowski and Wyer (1996).

aScores reflect the mean difference between the number of trait descriptions that implied
support and the number that implied lack of support.
bScores reflect the mean difference in ratings of traits implying support (reported along a
010 scale) and ratings of traits implying rejection.
fects of semantic concepts on the interpretation of the scenario that was evident in other perspective conditions.
CULTURAL DIFFERENCES IN IMPLICIT THEORIES

Individual differences in the chronic accessibility of implicit theories are
likely to result in part from differences in the socialization practices to
which individuals have been exposed (cf. Miller, 1994; Nelson, 1993; see
chap. 6) and, therefore, the frequency with which the theories have been
applied in the past. To this extent, differences may exist in the theories that
are typically applied by members of different subgroups within a society
(e.g., men vs. women, Blacks vs. Whites, etc.). More general cultural differences may exist as well. The precise nature of this difference is somewhat
controversial (cf. Briley & Wyer, 2001; Heine, Lehman, Markus, & Kitayama,
1999; Markus & Kitayama, 1991; Triandis, 1995). However, certain general
differences appear to exist.
For example, members of Western and Asian cultures are typically characterized as differing in terms of individualism and collectivism (Hofstede,
1991; Triandis, 1995). Individualism is characterized by a focus on individual
freedom and independence, a concern with personal goals with little consideration of others, independently of others, and competitiveness. In contrast, collectivism is typified by a focus on oneself in relation to the groups
to which one belongs, and a concern with group well-being rather than per-
291
sonal goals. As Briley and Wyer (2001) pointed out (see also Ho & Chiu,
1994; Rhee, Uleman, & Lee, 1996; Triandis & Gelfand, 1998), the individualismcollectivism construct is multidimensional, and the dimensions that
are often assumed to distinguish individualists and collectivists (i.e., individuality, cooperation, competitiveness, self-sacrifice, etc.) are not strongly
related (Briley & Wyer, 2001). Nevertheless, a general cultural difference in
the tendency to think of oneself as an independent individual vs. a member
of a group is fairly pervasive (Markus & Kitayama, 1994).
However, the implicit theories that reflect these different cultural orientations are not the only ones that people acquire. As Hong et al. (2000) suggested, representatives of a given culture are often exposed to norms and
values that pervade other cultures as well. Moreover, many individuals are
bicultural, having lived for many years in more than one country, or being
residing in a country (e.g., Hong Kong) in which both Western and Asian
norms and values are pervasive. Hong et al. (2000) conceptualize culture as
a dynamic construct, the effects of which can depend on situational factors
that influence the extent to which cultural norms and values are accessible
in memory. Several studies provide evidence of this dependence.
Attributions of Causality
Hong et al. (2000) showed that the implicit theories that Hong Kong participants bring to bear on the interpretation of information and the inferences
they draw from this information can be induced by exposing participants to
cultural symbols of Western and Asian cultures prior to the judgment task.
For example, Westerners (individualists) typically acquire an implicit theory of causality that places responsibility on the individual, whereas Asians
(collectivists) acquire theories that place responsibility on the society or
social context in which the behavior occurs. Evidence supporting this difference was reported by Morris and Peng (1994; see also, Choi, Nisbett, &
Norenzayan, 1999). Nevertheless, both implicit theories are likely to exist in
memory, and the theory that is activated and applied may be influenced by
situational factors that affect its accessibility.
To demonstrate this, Hong et al. constructed stimulus materials consisting of a school of fish in which one fish was swimming ahead of the others.
Thus, the situation could be interpreted as an indication that either the first
fish was leading the others (implying a dispositional cause of the fishs behavior) or, alternatively, that the first fish was being chased by the others
(a situational cause of its behavior). Hong Kong Chinese undergraduates
were exposed to these pictures with instructions to explain why the one
fish was swimming ahead of the others. Before doing so, however, they
were exposed to either a series of American cultural icons or a series of
Chinese icons as part of an ostensibly different experiment. Exposure to
292
CHAPTER 10
American icons increased the tendency to assign a disposition cause to the

fishs behavior relative to control conditions, and correspondingly decreased the tendency to assign a situational cause to it. However, exposure
to Chinese cultural icons had the opposite effects.
Theories of Personal Causality
A series of studies by Oishi, Wyer, and Colcombe (2000) identified cultural
differences in the situational factors that activate implicit theories about
the determinants of ones own behavior, and demonstrated that these differences are reflected in peoples use of their current life satisfaction to predict the future. People are likely to use this criterion if they activate a person-focused theory of causality (i.e., a theory that life experiences are
determined by stable personal characteristics that persist over time). However, they are less likely to do so if they apply a situation-focused theory
(i.e., the theory that life experiences are influenced by unforeseen situational factors). Like Hong et al., we reasoned that people are likely to have
both implicit theories in memory regardless of their cultural background.
However, we expected that the situational factors that activate these theories would differ.
Specifically, Heine and Lehman (1997; Heine et al., 1999) noted that representatives of individualistic cultures exhibit strong self-enhancement motives, as evidenced by a desire to take responsibility for positive experiences but to attribute negative experiences to transitory situational factors
(Harvey & Weary, 1984; Zuckerman, 1975; see also the study by Arkin et al.,
1976, described in chap. 9). If this is so, calling European Americans attention to positive experiences they have had in the past (for which they presumably take responsibility) should activate a theory that people are personally responsible for the events that befall them. Therefore, it should
increase the likelihood that these individuals see themselves as responsible
for both their current and future life situation, and should increase their
tendency to use their current life satisfaction to predict the future. However, calling these individuals attention to negative life events (which they
attribute to situational factors) should activate a theory that life events are
the result of unpredictable factors over which they have little control, and
so it should decrease their tendency to use current life satisfaction to predict the future.
These predictions should not hold for Asians, however. Members of
Asian societies tend to be self-effacing, and to take responsibility for personal failure but attribute success to external circumstances (Fry & Ghosh,
1980; Yamauchi, 1988). If this is so, calling Asians attention to positive and
negative life experiences should have the opposite effects that it has on
Americans. That is, thinking about positive life events should activate a sit-
293
uation-based theory, whereas thinking about negative life events should activate a person-based theory. Consequently, Asians should be less inclined
to use their current life satisfaction to predict their future in the former condition than the latter.
Oishi et al. (2000) confirmed these predictions. Participants were European Americans and Asian Americans whose life satisfaction was assessed
in an ostensibly unrelated experiment. Participants first completed a lifeevent inventory in which they were asked to write about either a positive
life experience they had had or a negative one. Then, after doing so, they
were given a list of 10 positive and 10 negative events (getting the best
grade on an exam, finding that your friends were talking about you behind
your back, etc.), and estimated the likelihood that each event would occur
to them in the near future.
Multiple regression procedures were used to analyze the favorableness
of participants predictions of the future as a function of a number of orthogonal contrasts corresponding to the main and interactive effects of life
satisfaction (a continuous variable), cultural background and the valence of
the life experience that participants recalled. This analysis yielded a significant interaction of all three variables. The implications of this interaction
can be seen in the left half of Table 10.3, which shows estimated values of
the favorableness of participants future outlook as a function of culture,
the valence of the stories that participants wrote about themselves, and
current life satisfaction. European Americans life satisfaction had a greater
effect on their future outlook when they had written about a positive life experience than when they had written about a negative one. In contrast,
Asian Americans life satisfaction had more influence on their future outTABLE 10.3
Regression-Based Estimates of Future Outlook as a Function
of Current Life Satisfaction, Cultural Background, and the
Valence of Stories About Self and Other
Valence of Stories
About Self
European Americans
High current life satisfaction
Low current life satisfaction
Difference
Asian Americans
High current life satisfaction
Low current life satisfaction
Difference
Valence of Stories
About Others
Favorable
Unfavorable
Favorable
Unfavorable
0.36
2.04
2.40
1.26
0.26
1.52
1.44
1.18
0.26
1.84
0.38
2.22
0.80
0.32
1.12
1.54
1.10
2.64
1.74
0.56
1.18
0.98
0.24
0.74
Note. Adapted from Oishi et al. (2000).
294
CHAPTER 10
look when they had written about a negative experience than when they
had written about a positive one.
A second study was analogous to the first but focused on the effects of
activating concepts about others. If European Americans are motivated by
a desire to be superior to others, they may tend to attribute others successes to situational factors but others negative outcomes to dispositional
factors (Snyder, Stephan, & Rosenfield, 1976). In contrast, if Asians are typically self-effacing and tend to promote others interests at their own expense, they may be inclined to attribute others successes to dispositional
causes (Yamauchi, 1988). To this extent, a person-based theory of causality
should be activated in European Americans by stimulating them to think
about others failures, but should be activated in Asian Americans by inducing them to think about others successes.
To explore this possibility, participants participated in a study that was
very similar to the first. Rather than writing about a personal experience,
however, they read a newspaper story describing either positive or negative job prospects for the current years college graduates. (Participants
were freshmen, and so the story was not personally relevant.) Regression
analyses of these participants future outlook as a function of experimental
variables yielded a three-way interaction of cultural background, life satisfaction, and the valence of others experiences, the implications of which
are shown in the right half of Table 10.3. The pattern of these data is exactly
the opposite of the first set of data. That is, European Americans future outlook was most strongly influenced by their current life satisfaction when
they had read about others negative experiences, but Asians future outlook was most strongly influenced by life satisfaction when they had read
about others positive experiences.
Cultural Differences in Inference Strategies
The preceding discussion suggests cultural differences in the content of the
implicit theories that are applied by members of different cultures. However,
more general differences may exist in the conditions in which these theories
are used. Choi et al. (1998) suggested that general cultural differences between Western and Asian cultures exist in the tendency to use analytic reasoning in generating explanations for social events. To this extent, the inference processes that are implicitly assumed in much of our discussion of the
use of implicational molecules might not generalize over cultures.
Some support for this possibility was obtained in a study by Norenzayan
and Kim (2000) in research on the Socratic effect (McGuire, 1960). In an initial experimental session, European Americans and Asians completed a belief questionnaire containing causally related propositions of the sort to
which Equation 10.1 is applicable. They then returned 1 week later and com-
295
pleted the questionnaire a second time. Based on research and theorizing

by McGuire (1960; Rosen & Wyer, 1972), Norenzayan and Kim assumed that
participants might show inconsistencies in their beliefs initially, but that
making the beliefs salient in temporal proximity would lead them to revise
these beliefs to make them more consistent. Thus, their beliefs should be
more consistent in the second session than they were at first, based on
Equation 10.1. This was true of European Americans, thus replicating findings by Rosen and Wyer (1972). However, Asians beliefs were just as inconsistent in the second session as they were at the outset. This suggests that
Asians, unlike Westerners, either did not perceive their beliefs in the propositions to be inconsistent or, alternatively, employed inference processes
that differed from those described by Equation 10.1.
THE ROLE OF IMPLICIT THEORIES

IN MARITAL SATISFACTION
The theories we construct and apply in evaluating our life experiences can
often have social consequences. For example, people often have narrativebased theories about close relationships. However, substantial individuals
exist in the nature of these theories. Some people, for example, may believe
that marriage partners fall in love quickly but that these feelings inevitably
deteriorate over time, ultimately leading to separation or divorce. Others
may believe that love takes a long time to develop but that once established, it is maintained. Still others may believe that peoples feelings
change nonmonotonically. For example, partners affection for one another
decreases over the early years of a relationship as romantic love dissipates
but then increases again in later years as the relationship matures.
Several theorists have conceptualized the role that narrative representations can play in the dynamics of close relationships (Forgas, 1991; Holmes
& Murray, 1995; Miller & Read, 1991; Schank & Abelson, 1995). Murray and
Holmes (1996), for example, found that marriage partners often reconstruct
stories about their personal relationships in order to make them consistent
with the implications of their narrative-based theories about relationships
in general (see also Holmberg & Holmes, 1994). However, it seems likely
that if partners have different theory-based expectations for the typical
progress of close relationships, they may differ in how they evaluate their
own relationship. These differences could create marital conflict and dissatisfaction.
Research by Gohm and Wyer (1998) bears indirectly on this possibility.
We assumed that individuals theories about the prognosis of relationships
over time could be inferred from their perceptions of change in the feelings
that characterize these relationships, and that differences in marital satis-
296
CHAPTER 10
faction might be related to these theories. To this end, we identified 32 married couples who had been in their relationship for an average of 9 years,
and whose marital satisfaction (based on the LockeWallace, 1959, scale of
marital adjustment) had been determined in an earlier study. In 16 of these
couples, both partners expressed satisfaction with their relationship, and in
the other 16, both partners reported being dissatisfied. Participants were
told that the studys purpose was to understand peoples perceptions of
how interpersonal relationships were likely to change over time. On this
pretense, participants plotted six graphs showing the changes that typically occurred in marriage partners feelings toward one another during the
first 10 years of their relationship. Three graphs described the typical
mans feelings of romantic love, commitment to the marriage, and affection
for his partner. Three others described the typical womans feelings along
the same dimensions. To aid them in constructing their graphs, participants
were given axes in which the x-coordinate pertained to the year of the relationship and the y-coordinate referred to the attribute being rated along a
scale from 0 (none) to 10 (very much). Partners were told to place an x
above the number denoting each year of the relationship to indicate their
perception of the partners feelings during that year.
Satisfied partners theories about the prognosis of marital relationships
were expected to be more similar to one another than dissatisfied partners
theories. Data pertaining to this hypothesis are conveyed in Figs. 10.2 and
10.3, which show the composite graphs drawn by satisfied and dissatisfied
partners along each dimension. Analyses of these data revealed an interaction of relationship satisfaction, the partner being rated, and attribute dimension, F(2, 60) = 4.29, p < .05 and an higher order interaction involving
these variables, participant sex, and time, F(18, 540) = 1.88, p < .01. These interactions are most clearly interpretable by considering data for satisfied
and dissatisfied couples separately.
As expected, satisfied husbands and wives had generally similar theories
about how a typical marriage partners feelings were likely to change. For
example, they agreed that the typical partners feelings of romantic love decreased to a much greater extent over time than his or her feelings of affection. Moreover, they agreed that the typical mans feelings of love would
tend to increase during the later years of the relationship without a corresponding increase in affection, whereas the typical womans feelings of affection were likely to increase in later years without a corresponding increase in feelings of romantic love.
In contrast, dissatisfied partners perceptions of the typical marriage differed much more dramatically. These differences were reflected in an interaction of respondent sex, the sex of the partner being rated, attribute dimension, and time, F(18, 270) = 2.80, p < .01, in an analysis of judgments by
dissatisfied partners alone. Three aspects of these data are noteworthy.
297
FIG. 10.2. Satisfied partners perceptions of the typical husbands and typical
wifes feelings over the first 10 years of a relationship.
1. Dissatisfied husbands and wives both perceived the typical marriage

partners feelings of romantic love to decrease substantially over the first 8
years of the relationship. However, whereas dissatisfied wives expected
these feelings to increase in later years, dissatisfied husbands did not.
2. Dissatisfied husbands perceived the typical mans commitment during early years of the relationship to correspond more closely to his feelings of romantic love than to his feelings of affection. In contrast, dissatisfied wives perceived the typical mans feelings of commitment to parallel
his feelings of affection, and to be maintained despite a decrease in his feelings of romantic love.
298
CHAPTER 10
FIG. 10.3. Dissatisfied partners perceptions of the typical husbands and typical wifes feelings over the first 10 years of a relationship.
3. Dissatisfied husbands believed that the typical womans commitment

to her relationship corresponded more closely to her feelings of romantic
love than to her feelings of affection. In contrast, dissatisfied wives perceived
the typical womans commitment to parallel her feelings of affection and to
differ markedly from her feelings of romantic love.
The implications of these findings for an understanding of marital relationships are necessarily equivocal. It is unclear, for example, whether the
differences in dissatisfied partners perceptions are a cause or an effect of
their feelings about their own relationship. Be that as it may, however, the
299
results demonstrate the utility of comparing the implicit theories that are
likely to characterize happy and unhappy couples. Differences in these theories may be worth more careful attention in future research.
EFFECTS OF IMPLICIT THEORIES

ON BEHAVIORAL DECISIONS
The preceding sections discuss the impact of peoples implicit theories on
their judgments of self and others. Perhaps the most important influences
of these theories, however, are on behavioral decisions. This influence is
evident in Dwecks work on learned helplessness, noted earlier. That is,
people who apply an incremental theory of ability are likely to conclude
that they can improve their performance on a task by exerting more effort
and, therefore, they may persist. In contrast, people who invoke an entity
theory may attribute their failure to an inherent lack of ability and conclude
that they are likely to fail again regardless of their effort. Therefore, they
may not bother trying (Dweck, 1975; Dweck & Leggett, 1988).
We have already noted that when two or more implicit theories can potentially be brought to bear on judgments, the theory that is applied may
be determined not only by its chronic accessibility but also by transitory
situational factors that activate concepts and knowledge with which they
are associated. This is true of theories that have implications for behavior
as well. Many situational influences on behavior can be conceptualized in
terms of the mediating effects of situational variables on the activation of
different implicit theories. In this section, we restrict consideration to three
quite different areas in which implicit theories potentially come into play in
behavioral decisions. Although the research reported in each case does not
unequivocally demonstrate the mediating influence of these theories, a consideration of the research from this perspective often raises additional issues that might not otherwise be identified.
Helping Behavior
Several factors that underlie decisions to provide help have already been
noted in our discussion of the just world hypothesis. That is, people are
more likely to provide help if they can construct an image-based episode
model of how their behavior might benefit the recipient than if they cannot
(Miller, 1977). However, other factors also enter into the picture. As noted
by Hong et al. (2000), people may have implicit theories about not only the
dispositional antecedents of their behavior but also situational influences,
and both type of theories could affect their own behavioral decisions as
well as their perceptions of others. These theories can have different impli-
300
CHAPTER 10
cations. For example, if people are asked to help another person, two types
of theories could influence their decision to help. One, person-focused theory might be captured by the molecule:
[P is a kind person; O asks P for assistance; P helps].
A second, recipient-focused theory might have the form:

[O requests assistance; Os request is (is not) legitimate; P helps (does not
help)].
Note that if people who are asked to help activate the first theory and consider themselves to be kind, they are likely to provide help independently
of other considerations. If they activate the second theory, however, their
decision to help may depend on his or her perception of the requests
legitimacy.
An imaginative study by Langer and Abelson (1972) can be conceptualized in terms of these considerations. Shoppers were approached outside a
supermarket by a woman who had ostensibly hurt her leg. In some cases,
the woman asked the shopper to call her husband to come and pick her up.
This request was likely to be considered legitimate under the circumstances. In other cases, she asked the shopper to call her boss and tell him
she would be latesomething that was not sufficiently important to justify
asking a stranger. The second variable manipulated in this study was the
order in which the woman (a) described her adversity and (b) made her request. Specifically, under request-first conditions, the woman said: Would
you do something for me? Please do me a favor and call my husband to ask
him to pick me up. My knee is killing me. I think I sprained it. Under needfirst conditions, however, the womans words were: My knee is killing me. I
think I sprained it. Would you do something for me? Please call my husband
and ask him to pick me up.
Thus, the descriptive content of the womans statements was identical in
the two cases. Under request-first conditions, however, the victims initial
statement (Would you do something or me . . .) focuses attention on the
recipient and, therefore, is likely to activate a person-focused theory of the
sort described earlier. To this extent, shoppers who typically consider
themselves to be kind are likely to grant the request regardless of other
considerations. In contrast, the womans initial statement under need-first
conditions (My leg is killing me . . .) focuses attention on the victim and,
therefore, is likely to activate the second theory, in which the legitimacy of
the request comes into play. In this case, therefore, shoppers should comply with the request only when the request appears legitimate. Langer and
Abelsons findings are consistent with this analysis. That is, the likelihood
301
of helping the victim increased with the requests legitimacy under needfirst conditions but not under request-first conditions.
Effects of Anticipated Outcomes on Decisions
People who are confronted with a decision are likely to base it on an implicit theory about the consequences of the alternative courses of action
available. These theories are often domain specific. Peoples theories about
the consequences of betting on a horse race are likely to differ from their
theories about the consequences of studying for a final examination. However, more general theories can exist as well. As we discussed in chapter 3,
Higgins (1997) postulates two general motivational orientations that can underlie goal-directed information processing. One, promotion focus, is characterized by an emphasis on the rewards that can potentially result from a behavioral decision irrespective of the costs that might be incurred. The
other, prevention focus, is characterized by a concern with avoiding negative consequences of ones behavior independently of its potential benefits.
These different orientations could be embodied in two implicit theories,
each of which has implications for a different behavioral objective:
[P performs A; A has positive consequences; P attains benefits],
and
[P performs B; B avoids negative outcomes; P avoids misfortune].
In many instances, the behavior that maximizes the likelihood of positive

outcomes also increases the risk of negative ones. (E.g., Investing in stocks
can potentially reap substantial benefits but runs the risk of a substantial
loss as well.) In these instances, the two theories can have different implications. For example, suppose an individual is confronted with a choice between (a) a vacation in which one is likely to have new and exciting cultural
experiences but will require living in sleazy hotels and eating poor food, and
(b) a vacation that is much less exotic but where the accommodations are
not elegant but are adequate. A person who uses the promotion-focus molecule to predict the individuals behavior should infer that the vacationer will
choose the first option. However, a person who activates the prevention molecule should predict the vacationer to choose the second alternative.
This possibility becomes of particular interest when people apply the
theories to their own behavioral decisions. That is, the decision of an individual who is personally confronted with the two vacations just described
might depend on which of the two alternative theories happens to be applied. Several situational and individual difference factors could influence
302
CHAPTER 10
the relative accessibility of the theories and, therefore, the likelihood they
are applied. One factor, discussed in chapter 3, can be the salience of ones
group membership. That is, consciousness of belonging to a group appears
to activate feelings of responsibility to others and, therefore, to increase attention to negative consequences of behavior. Thus, it may increase the relative accessibility of a prevention-focus molecule that, once activated, influences decisions in both interpersonal and individual choice situations
(Briley & Wyer, 2002). Situational factors that emphasize the importance of
success as opposed to the importance of avoiding failure could have a similar influence. Finally, chronic individual differences in the accessibility of
the two theories can also exist (Aaker & Lee, 2001; Briley et al., 2000). (This
research is also discussed in chapter 3.)
However, the implicit theories on which decisions are based can pertain
to not only the objective consequences that are likely to occur but also the
emotional reactions that occur in response to these consequences. People
may anticipate feeling happy if their decision has positive consequences
but disappointed if it has negative consequences. In addition, they might
anticipate experiencing regret if they decide against an option that would
have benefited them and might expect to feel relief if they decide to avoid
an option that would have turned out badly if they had taken it. The decision they make might depend to a greater extent on their implicit theories
about the occurrence of these reactions than by the outcomes that elicit
them (cf. Bell, 1982; Loomes & Sugden, 1986; Mellers, Schwartz, & Ritov,
1999). However, both situational and individual differences undoubtedly occur in the conditions in which these theories are invoked.
Fong and Wyer (2003) examined these differences in simulated choice
situations similar to those that participants were likely to encounter in daily
life. Both North American and Hong Kong Chinese students participated. In
one study, participants imagined a financial situation in which they could
decide either (a) to make an investment that could potentially yield a considerable profit but also ran the risk of taking a substantial loss, or (b) to
make a bank deposit that would yield a low rate of interest but would avoid
the risk of loss. In a second study, they imagined an academic situation in
which they could decide whether (a) to study a particular topic that would
ensure a high grade on an exam if the topic actually appeared but a low
grade if it did not, or (b) to study other topics, guaranteeing an average
grade regardless of whether or not the special topic was on the exam. In
each experiment, the magnitude of the gain or loss that might occur as a result of taking the risk, and information about the decisions that others had
typically made, were varied over conditions.
Participants after reading the scenario reported the likelihood that they
would choose the risky option and estimated the risk of doing so. Then,
303
they estimated the emotional reactions they would experience if they (a)
took the risk and benefited, (b) took the risk and incurred misfortune, (c)
did not take the risk and avoided misfortune, or (d) did not take the risk
and missed out on an opportunity to benefit. These emotional reactions
were analogous to happiness, disappointment, relief and regret, respectively.
Participants decisions in each study were typically influenced by their
perception of the risk involved in doing so, the importance of the consequences, and the decisions that others ostensibly made. However, the effects of these variables were largely mediated by their influence on participants anticipated emotional reactions. Regression analyses indicated that
participants anticipated reactions accounted for a sizable proportion of
variance in decisions over and above the effects of situational variables,
whereas the latter variables contributed only a small and often nonsignificant proportion of variance in decisions over and above the effects of
anticipated emotional reactions. In other words, participants implicit theories about the emotional reactions they would experience in response to alternative decision outcomes were the primary determinant of their choices.
However, the relative impact of the four anticipated reactions varied
over both scenarios and cultural groups. For example, decisions in the financial situation were based primarily on the consequences of taking the
risk (i.e., happiness and disappointment), whereas decisions in the academic situation were influenced by anticipated relief and regret as well.
Moreover, Americans decisions in the latter situation were influenced to
the greatest extent by the happiness they anticipated if they benefited as a
result of taking the risk, and to the least extent by the relief they anticipated
if they avoided a negative outcome by not taking the risk. In contrast, Chinese participants decisions were influenced least by the happiness they anticipated as a result of receiving a positive outcome and most by the relief
they anticipated as a result of avoiding a negative one. These results suggest that chronic cultural differences exist in the tendency to invoke theories that focus on positive versus negative decision outcomes. To this extent, they confirm conclusions drawn by Briley et al. (2000; see also Aaker &
Lee, 2001).
Effects of Stereotypes on Social Behavior
One of the most heavily researched areas in social psychological research
surrounds the determinants and effects of social stereotypes (for reviews
of theory theoretical analyses, see Bodenhausen & Macrae, 1998; Fiske,
1998; Hamilton & Sherman, 1994). In most of this research, a stereotype has
been conceptualized as a cluster of traits and behavioral dispositions that
are associated with a social or ethnic category (women, African Americans,
304
CHAPTER 10
fraternity members, etc.). Generalized representations of this sort presumably influence persons attitudes toward individual members of the category and the attributes they assign to these members (Hamilton & Sherman, 1994). Moreover, a trait-based stereotype that is activated in one
situation can later influence the interpretation of information about other
individuals to whom the stereotype is objectively inapplicable (Devine,
1989; Lepore & Brown, 1997). Finally, active attempts to suppress the use of
the stereotype can have boomerang effects, actually increasing the use of
the stereotype as a basis for judgment once these conscious suppression
mechanisms are deactivated (Bodenhausen & Macrae, 1998).
An understanding of the impact of trait-based stereotypes on social judgments, and the conditions in which they are applied, is of obvious importance. As Reid and Wyer (1998) noted, however, there is surprisingly little
evidence that these stereotypes directly influence individuals actual behavior toward members of the stereotyped group. The classic study by
LaPiere (1934) provides an example of a situation in which individuals behavioral decisions in a social context do not correspond to the attitudes
and behavioral dispositions they express outside of this context. In fact, it
seems somewhat implausible that people who encounter a member of a stereotype group would expend the cognitive energy required to activate a
trait-based stereotype and construe its implications in anticipation of deciding how to behave toward members of the group. Instead, they are likely to
activate a situation stereotype of the events that are likely to occur in the
situation at hand, and to base their decision on the implications of this
event representation.
A situation stereotype is likely to be activated by a configuration of features that exist in the situation rather than by individual attributes. Thus,
its activation can depend on features of persons involved in a situation and
features of the situation in combination, rather than on either set of features
in isolation. For example, suppose a woman sees an African-American man
approaching her late at night on an empty Chicago street. She is unlikely to
activate a trait-based stereotype of African Americans as aggressive or hostile before deciding to cross the street. Rather, she is more likely to activate
a prototypic event representation of the events that occur in this particular
situation (i.e., People who see a woman walking alone late at night are
likely to mug or rape them, and so it is best to avoid contact with them if
possible.) This stereotypic event representation may function as an implicit theory that is activated and applied in the particular situation at hand.
If the woman encountered this same individual in a university library she
would be far less likely to engage in avoidance behavior. This is because a
different situation stereotype is activated that has quite different implications for behavior.
305
FINAL COMMENT
The research reviewed in this chapter has cut across a very wide variety of
concerns, ranging from counterfactual reasoning and reconstructive memory to perceptions of social reality, cultural differences in future outlook,
perceptions of social support, marital satisfaction, and stereotyping. The
role of implicit theories in these phenomena is admittedly somewhat conjectural. That is, few studies provide direct evidence of the cognitive basis
of the judgments and decisions that were made. To the extent these theories exist, however, they are a useful construct in integrating a diversity of
empirical findings within a common conceptual framework.
Much of our discussion was guided by the assumption that many different theories can often be brought to bear on a judgment or decision, and
that the theory that is activated and applied is determined by its relative accessibility. The determinants of knowledge accessibility summarized in
chapter 3 obviously come into play in predicting the conditions in which
these theories are used.
Although the phenomena discussed in this and the preceding chapter do
not pretend to be exhaustive, they are representative of those in which narrative representations are likely to be constructed and used as a basis for
judgments and decisions in daily life. However, an obvious and important
factor that influences judgments and behavior in social situations has been
totally ignored, both in these chapters and elsewhere in this volume. We
now turn to a consideration of this factor.
C H A P T E R
11
The Role of Affect
in Information Processing
Our focus throughout this volume has been on the way we comprehend information about people, objects, and events of the sort we encounter in
daily life. In our discussion, we have emphasized the role of narrative representations of knowledge in conceptualizing new information and in making
judgments of the people and events to which the representations refer. We
have also considered the role of pictures and visual images in the comprehension and use of information. In doing so, however, we have virtually ignored an obvious but important factor that pervades information processing outside the laboratory.
Specifically, many situations we encounter in daily life elicit affective reactions. These reactions can be stimulated by direct experiences with a
person, object, or event with which they are associated. However, imagining these stimuli, recalling past experiences in which they occurred, or writing stories about them can also elicit these reactions. When these affective
reactions are experienced, they can influence judgments and overt behavior toward the stimuli that gave rise to them. Thus, we behave antagonistically to a person who makes us angry, and attempt to increase our contact
with someone who makes us feel happy.
The cognitive determinants and consequences of affect and emotion have
been a major concern of social psychological research and theorizing for
more than 25 years. The issues investigated have ranged from the influence
of affect on the encoding and organization of new information in memory
(Bower, 1981, 1991) to its impact on the retrieval of previously acquired concepts and knowledge (Blaney, 1986; Niedenthal & Setterlund, 1994), its use as
a basis for judgments and decisions (Clore, 1992; Schwarz & Clore, 1983,
1988), and its influence on the strategies that individuals use when perform306
THE ROLE OF AFFECT
307
ing cognitive tasks (Isen, 1987; Mackie & Worth, 1989; Nisbett, Schwarz, &
Bless, 1991). These issues have been investigated in research on not only
judgment and decision making but also creativity (Isen, Daubman, & Nowicki,
1987; Isen, Johnson, Mertz, & Robinson, 1985), communication and persuasion (Albarracin & Wyer, 2001; Bless, Bohner, Schwarz, & Strack, 1990; Mackie
& Worth, 1989l; Wegener, Petty, & Smith, 1995); stereotyping (Bodenhausen,
1993; Bodenhausen, Sheppard, & Kramer, 1994), self-evaluations (Levine,
Wyer, & Schwarz, 1994); political judgment (Abelson, Kinder, Peters, & Fiske,
1982; Ottati & Isbell, 1996; Ottati & Wyer, 1993), and consumer behavior
(Adaval, 2001; Pham, 1998; Pham, Cohen, Pracejus, & Hughes, 2001).
Despite the diverse conditions in which affect has been shown to influence information processing, the cognitive mechanisms through which it
exerts this influence may be more limited than is often assumed. These
mechanisms have been difficult to isolate. However, the research reported
to date converges on two general conclusions:
1. Positive and negative affect can be preconditions for cognitive operations for cognitive productions of the sort that compose procedural knowledge (J. Anderson, 1983; E. R. Smith, 1990) and that govern behavior in specific types of situations to which they are relevant. Once activated, these
productions can influence responses to new information without conscious
awareness.
2. The affective reactions that one experiences at a given moment can be
used as information about ones attitude toward oneself, other people, situations with which one is confronted, the outcomes of behavior, or the appropriateness of certain strategies for attaining behavioral outcomes. It can also
influence perceptions of the validity of other affect-eliciting information and,
therefore, can influence the weight that is attached to this information in
making judgments.
The validity of these conclusions is discussed with reference to research
and theory in several different areas, with particular emphasis on the impact of affect on behavioral decisions. In this context, a performance feedback model is proposed that accounts for the influence of affect on behavior and judgment in a number of different situations that occur in both the
laboratory and daily life.
Definition
Wyer, Clore, and Isbell (1999) conceptualized affect as a configuration of
positively or negatively valenced subjective reactions that a person experiences at a given point in time and perceives as either pleasant or unpleas-
308
CHAPTER 11
ant feelings. These feelings can be elicited by proprioceptive cues (e.g., facial expressions; see Strack, Martin, & Stepper, 1988; Zajonc, Murray, &
Inglehart, 1989), physical stimulation, or drugs. More commonly, however,
they are internal responses to a set of new or previously formed cognitions.
Several general classes of affective reactions can be distinguished (for a detailed discussion of these distinctions, see Clore, Schwarz, & Conway, 1994;
Ortony, Clore, & Collins, 1988; Wyer et al., 1999).
1. Moods and Emotions
Wyer et al. (1999) assumed that emotions are affective states that occur
in response to cognitions about persons, objects, and events, and that are
consciously attributed to a specified source. In contrast, moods are typically experienced in the absence of specific cognitions about their source
and are not attributed to any particular cause unless one is called upon to
do so. Thus, the clusters of subjective reactions that exemplify emotions
and moods are similar except that one is consciously attributed to a given
source and the other is not. As Wyer et al. (1999) suggested, feelings that
are elicited by a stimulus event take time to dissipate. Therefore, feelings
that are initially experienced as an emotion can become a mood once their
source is no longer considered.
The preceding remarks suggest an important distinction between emotions and affect per se. That is, emotions, unlike moods or affective reactions more generally, have a strong cognitive component. Forgas (2002), for
example, assumed that emotions are defined in terms of a cognitive appraisal of a situation, and that physiological and autonomic reactions characteristic of that emotion are responses to this appraisal (p. 104). Thus,
emotions, unlike other affective states, are necessarily accompanied by
cognitions about the situational factors that give rise to them. To this extent, there may be concepts in memory whose features include a representation of ones affective reactions as well as alternative names of both the
emotion and its behavioral manifestations.
The concept angry, for example, might be metaphorically represented as
shown in Fig. 11.1. The concept itself is denoted by a central node to which
the features that define and characterize it (a cognitive appraisal of the conditions that elicit the emotion, a mental representation of the affective reactions that are experienced, names, behaviors, etc.) are associatively linked.
The concept of a mood might be similar to that shown in the figure except
that a characterization of its eliciting conditions would not be included
among its features. In each case, thinking about one or more of the features
associated with it could presumably activate the mood or emotion concept.
Thus, consciousness of the cluster of physiological reactions that define an
emotion concept might activate it, in much the same way as the features of
THE ROLE OF AFFECT
309
FIG. 11.1. Metaphorical mental representation of the emotion concept angry.
any concept can activate this concept. Note, however, that these physiological reactions per se do not activate the concept. This activation occurs
only if the physiological reactions are thought about or otherwise cognitively represented.
2. Affect Versus Evaluation
It is important to distinguish between affective reactions to a stimulus
and evaluations of the stimulus along a goodbad dimension (e.g., attitudes;
see Eagly & Chaiken, 1993, 1998). Although ones feelings about a person or
object can often be the basis of such an evaluation, not all evaluative judgments are based on this criterion (cf. Adaval, 2001; Wyer et al., 1999; Zanna
& Rempel, 1988). The distinction between affect and evaluation is not always made in theories of affect-based information processing (cf. Abelson
et al., 1982; Bargh, Chaiken, Govender, & Pratto, 1992; Fiske & Pavelchak,
1986). In the present context, however, this distinction is important to keep
in mind. For one thing, evaluations are cognitions and, as such, can be features of the mental representations of the entities to which they refer. Affective reactions, however, may not be.
3. Affect Versus Cognition
As implied by our discussion of moods and emotions, people have concepts of different types of affect, and cognitions about different configurations of subjective reactions can be features of these concepts (see Fig.
11.1). These latter cognitions, however, may not always have verbal labels.
Moreover, it is important to distinguish between cognitions about ones affective reactions and the affective reactions per se. Although affective reactions may sometimes be learned responses to cognitions, they are not part
310
CHAPTER 11
of the cognitive system. This distinction is made by LeDoux (1994) in conceptualizing the relation of emotion and memory:
It is important to distinguish emotional memory [the learned or conditioned
emotional significance of an object that automatically triggers an emotion]
and memory for emotion. The latter is a declarative, conscious memory of an
emotional experience. It is stored as a fact about an emotional episode. Emotional memory (mediated by the amygdala) and memory for emotion (mediated by the hippocampus) can be reactivated in parallel on later occasions,
thereby giving new declarative memories an emotional flavor. . . . Because
emotional memory and memory for emotion are stored through different
brain systems, it is possible that the information stored will not always coincide. (p. 312)
Thus, experiencing affect is not the same thing as conceptualizing it. In

fact, we often have subjective affective reactions that we do not consciously identify at all. And even if we are aware of the reactions, we might
not attempt to explain their occurrence unless the reactions are intense or
there is some particular reason to do so. Put another way, affective reactions may be responses to cognitions, but are not themselves cognitions.
Because specific clusters of affective reactions are experienced as feelings, a cognitive representation of these feelings can often be a feature of a
more general concept of an emotion or mood state (angry, happy, etc.), as
shown in Fig. 11.1. Once such a concept is activated, it could theoretically
be used by the Parser to interpret a new experience to which it is applicable, depending on whether features of the experience instantiate it. Thus,
an emotion concept could be used to interpret the feelings one is personally experiencing. As suggested by Fig. 11.1, however, the concept could
also be activated by other features, including those of the situational context in which these reactions occur (cf. Schachter & Singer, 1962).
Nevertheless, these considerations do not imply that affective reactions
themselves are part of an associative network of cognitions that compose semantic and episodic knowledge. Affective reactions can be responses to concepts and cognitions that are formed of a new stimulus event, to previously
formed cognitions about a past event, or to thoughts about an imagined (e.g.,
future) event (LeDoux, 1994). Moreover, affective reactions can be the referent of concepts and cognitions about ones own internal state, and these
cognitions, once activated, can enter into subsequent information processing. However, the reactions themselves are no more part of the cognitive system than the referent of the concept chair. This means that if a mental representation is not formed of the affective reactions one experiences, the
influence of these reactions on information processing may be quite limited.
These distinctions might seem self-evident. However, they appear to be
ignored in much of the theory and research on the impact of affect on infor-
THE ROLE OF AFFECT
311
mation processing. Several theories of affect and cognition (e.g., Bower,

1981; Forgas, 1995; Isen, Shalker, Clark, & Karp, 1978), as well as introductory textbooks (e.g., Baron & Byrne, 2000), appear to assume that affect
functions as a concept in an associative network whose role in information
processing is similar to that of other concepts. That is, affect can increase
the accessibility of other concepts and knowledge that are directly or indirectly associated with it (i.e., concepts that are similar to it in valence). As a
consequence, it can influence the interpretation of new information to
which these concepts apply. For similar reasons, affective reactions can
cue the retrieval of previously acquired knowledge representations whose
features are similar in valence. Thus, the impact of affective reactions on information processing is far reaching.
In contrast, if affective reactions are not themselves part of the cognitive
system, their influence may be more limited. If these reactions are represented cognitively, they might increase the accessibility of the specific concepts they exemplify. However, they would not necessarily influence the activation and use of concepts and knowledge to which they are not directly
related. Moreover, affective reactions should not even have this effect unless persons are conscious of the reactions and are stimulated to interpret
them in cognitive terms (see Fig. 11.1).
There are two ways, however, in which affective reactions per se are
likely to play a role. First, affective reactions can be part of the precondition
of an If [X], then [Y] production. That is, the precondition [X] could include not only cognitions but also proprioceptive stimuli that, in combination, could activate a behavioral sequence that is performed with minimal
conscious awareness.
Second, many judgments or decisions are based on feelings. For example, if we experience positive or negative affective reactions when we encounter or think about a particular person, we might attribute these reactions to our feelings about this person and must use these feelings as a
basis for estimating how much we like this individual. More generally, subjective reactions can often be used as a source of information about the
person, object, or event that ostensibly elicits them. This assumption,
which is implicit in early research and theory on attitude assessment
(Thurstone, 1959) and interpersonal attraction (Byrne, 1971; Clore & Byrne,
1974), has been stated more formally in a theory of affect as information
proposed by Schwarz and Clore (1983, 1988, 1996; for an early version of the
theory, see Wyer & Carlston, 1979). There are several contingencies on the
use of affect as a basis for judgment that are noted later in this chapter. For
example, affective reactions are only likely influence judgments for which
they are considered on a priori grounds to be relevant. The informational
influence of affective reactions is nonetheless pervasive, having been demonstrated to occur in judging life satisfaction and self-esteem, jokes and car-
312
CHAPTER 11
toons, task performance, consumer products, and political candidates.

These effects are discussed in some detail later in this chapter.
Methodological Considerations
Before discussing the implications of research and theory on the influence of
affect in information processing, a methodological problem is worth noting.
In much of this research, the affective reactions that people experience have
been manipulated experimentally. This is done in several ways. That is, people might be asked to write about a past experience that had elicited positive
or negative feelings at the time it occurred and that re-elicits these feelings
when thinking about it (Schwarz & Clore, 1983), to watch a movie that elicits
positive or negative emotions (Martin, Achee, Ward, & Wyer, 1993), or to read
aloud statements that reflect strong emotions (e.g., This is the greatest day
of my life!, I wish I could go to bed and never get up, etc.; see Velten, 1968).
Alternatively, participants might be given positive or negative feedback
about their performance on a test (Forgas & Bower, 1987).
These procedures are typically successful in inducing the affective reactions they are intended to stimulate. As the aforementioned examples indicate, however, the procedures activate semantic concepts and knowledge
that could also have an influence on judgments and behavior. Consequently, the interpretation of research that employs these procedures is often equivocal. This contention is elaborated in the following discussion.
AFFECT AND CONCEPT ACCESSIBILITY

The influence of affective reactions (or, more accurately, a mental representation of these reactions) on the accessibility of concepts and knowledge in
memory is presumably governed by the processes described in chapters 2
and 3. In particular, the representation of a particular configuration of reactions should only activate concepts and thoughts that the configuration exemplifies. There is no reason to suppose that the reactions will influence
the accessibility of concepts and knowledge to which they are not directly
related. Two quite different bodies of research bear on this matter, each of
which is discussed in turn.
Effects of Priming Affect on Semantic
Concept Accessibility
Research by Niedenthal and her colleagues (Niedenthal, Halberstadt, &
Setterlund, 1997; Niedenthal & Setterlund, 1994) is consistent with this conclusion. In these studies, participants were induced to feel either happy or
THE ROLE OF AFFECT
313
sad by having them listen to upbeat or gloomy music for 12 to 15 minutes.

Pretesting established that the music elicited specific feelings of happiness
and sadness rather than positive or negative affect more generally. In the
main experiments, however, participants were led to believe that the music
was being played to study the relation between auditory and visual perception, and were given no indication of the type of affect that the music was
intended to induce. Thus, the procedure avoided the interpretative ambiguities described in the preceding section.
If participants who hear the music are aware of the feelings it elicits, they
may form a mental representation of these feelings, and this representation
could increase the accessibility of the specific mood or emotion concept
the feelings exemplify. However, it should not influence the accessibility of
concepts pertaining to other types of feelings, or to positively or negatively
valenced concepts more generally. Results confirmed these predictions. After listening to the music, participants performed a word recognition task in
which a number of positively and negatively valenced words were presented. However, some of these words (e.g., delight, joy, weep, despair, etc.) pertained to happiness and sadness whereas others (e.g., wisdom, luck, crime, injury, etc.) did not. Participants quickly identified words
that described the particular type of affective reactions they were experiencing. However, the time they took to identify words that described other
types of positive or negative affect (Niedenthal et al., 1997), or to identify
positively or negatively valenced words that were not directly related to affective reactions (Niedenthal & Setterlund, 1994), was unaffected by their
reactions.
In short, Niedenthals research suggested that if people experience affective reactions that exemplify a particular mood or emotion concept of the
sort described in Fig. 11.1, these reactions can increase the accessibility of
other features of this concept (e.g., the name of the mood or emotion, or of
behaviors that exemplify it). However, these reactions do not increase the
accessibility of other concepts and knowledge to which the emotion concept is descriptively unrelated.
In fact, even the effects that Niedenthal et al. observed could be of limited generality. In these studies, both participants affective reactions and
the words they were given to identify in the lexical decision task may have
combined to activate the mood or emotion concept they exemplified. In Fig.
11.1, for example, both the representation of the physiological reactions
that define the concept angry and the name of this concept may resonate
with the concept, and the resonance from these two sources in combination may activate it. In the absence of the activation produced by the concepts name, however, it is conceivable that participants representation of
their affective reactions might not produce sufficient resonance to activate
the concept. As we noted earlier, people may often experience feelings
314
CHAPTER 11
without necessarily labeling or interpreting them. This labeling may only

occur spontaneously when the feelings are sufficiently intense that people
are motivated to understand and explain them.
Automatic Evaluation Effects
Niedenthal et al.s findings are consistent with the conceptualization we
propose. At the same time, they conflict with conclusions drawn by Forgas
(1995, 2002). Forgas proposed that priming a particular affect-related concept should increase the accessibility of other concepts with which similarly valenced affective reactions are associated. Research on the automatic evaluation effect (cf. Bargh, Chaiken, Govender, & Pratto, 1992;
Bargh, Chaiken, Raymond, & Hymes, 1996; Fazio, Sanbonmatsu, Powell, &
Kardes, 1986) might appear consistent with this possibility. In this research,
participants are typically exposed to a positively or negatively valenced
word (e. g., lovely, putrid, etc.) at below recognition threshold. Immediately
afterwards, participants in some studies are asked to categorize a second
word as either favorable or unfavorable (Bargh et al., 1992). Participants in
other studies are simply asked to pronounce the second word, without being given any indication that its valence was of concern (Bargh et al., 1996).
In each case, participants typically respond more quickly to the second
word when its evaluative implications are similar to those of the word that
preceded it than when its evaluative implications are dissimilar (Bargh et
al., 1992, 1996). These findings could be interpreted as evidence that the
first word elicits an affective reaction and that this reaction cues the retrieval of other words that are associated with similar reactions. Thus, participants identify the second word more quickly or more slowly, depending
on whether the affective reactions the word elicits are similar to or different
from the reactions elicited by the word that preceded it.
As noted earlier, however, a distinction should be made between the affective reactions that a stimulus elicits and its evaluative implications. If affective reactions were involved in the processes identified by Bargh and his
colleagues, observed effects should increase in magnitude with the extremity of the evaluative implications of the words involved and, therefore, the
intensity of the affect they elicit. In fact, this is true under very limited conditions (Bargh et al., 1992; Chaiken & Bargh, 1993). Moreover, responses to
valenced semantic concepts are influenced by exposure to normatively
positive and negative stimuli that participants have never before seen
(Duckworth, Bargh, Garcia, & Chaiken, 2002). This finding indicates that the
effects do not reflect a previously learned association between the concepts and the affect they elicit.
Wyer, Clore, and Isbell (1999) argued that the effects obtained by Bargh
et al. are likely to reflect a categorization process that people spontaneously
THE ROLE OF AFFECT
315
perform at the time they encounter stimuli that is not affect eliciting in and
of itself. That is, people may spontaneously categorize stimuli as either
good or bad. (This automatic categorization process could reflect an unlearned, evolution-based disposition to identify stimuli as potentially harmful or not harmful in order to survive.) However, affective reactions might
not provide the basis for this categorization. In fact, these reactions might
only be elicited by a cognitive appraisal of the stimulus and its implications
that occurs after the stimulus has been categorized. If this is so, the effects
observed by Bargh et al. (1992; see also Fazio et al., 1986) would not conflict
with the implications of Niedenthal et al.s (1997) findings.
This conclusion can be summarized in the following postulate:
Postulate 11.1. Affective reactions may activate a specific affect-relevant concept that can be used to interpret them. However, affective reactions do not
spontaneously influence the accessibility in memory of other similarly
valenced semantic concepts and declarative knowledge.
This postulate, however, runs counter to claims that affective reactions do

have an influence on the recall and interpretation of information that is
evaluatively congruent with it (Bower, 1981; Forgas, 1995). In the following
section, we review research that is often cited in support of these claims.
Although the review is not exhaustive, it is sufficiently representative to
convey the ambiguities that generally exist in the literature bearing on this
matter.
THE INFLUENCE OF AFFECT ON THE RECALL

AND INTERPRETATION OF INFORMATION
The Recall of Previously Acquired Knowledge
If affective reactions are features of a mental representation that people
have formed and stored in memory, the experience of similar reactions at
some later time should cue the retrieval of this representation. This effect
should occur for the same reasons that linguistically and perceptually
coded stimuli cue the retrieval of previously formed cognitions (see chap.
2). If affective reactions are not represented in memory, however, the experience of these reactions per se would not have this effect. Rather, these reactions should influence the recall of information only if the experience of
these reactions is accompanied by perceptually or linguistically coded features similar to those of the knowledge representations to be retrieved.
Thus, for example, a person who experiences a certain cluster of affective
reactions might be stimulated to interpret this configuration in terms of an
affect-related semantic concept such as sad, or down in the dumps. This
316
CHAPTER 11
concept, once activated, could cue the retrieval of a past experience to

which the concept had been applied. In the absence of this mediating cognitive activity, however, the experience is unlikely to be recalled.
This conclusion is not widely accepted. In fact, the assumption that affect does indeed cue the retrieval of valenced concepts and knowledge
from memory is quite pervasive (cf. Bower, 1981; Forgas, 1992, 1995, 2002;
Isen, 1984). However, empirical support for this assumption is very limited
(Blaney, 1986). In a study by Bower, Gilligan, and Monteiro (1981), participants in some conditions read a story that described both positively
valenced and negatively valenced events that occurred to the protagonists.
Then, they were induced to feel either happy or sad by describing a life
event that had elicited these emotions. If affective reactions per se cue the
retrieval of previously acquired knowledge that is similar in valence, participants should recall aspects of the story that are evaluatively congruent
with their feelings at the time of recall. In fact, however, the affect they were
experiencing had no influence on the type of information they recalled.
Mueller, Grove, and Thompson (1991) induced participants to feel either
happy or sad after exposing them to a list of positively and negatively
valenced words. This study also found no evidence that the affect that participants were experiencing influenced the valence of the information they
recalled.
In contrast, participants in a study by Bower (1981) were asked to recall
childhood experiences after having been induced hypnotically to feel
happy or sad. Participants recalled experiences that were evaluatively congruent with their mood. At the time of mood induction, however, participants were explicitly told what feelings they should try to experience.
These instructions presumably activated semantic concepts of the feelings
being induced, and these concepts, rather than the affective reactions to
which they referred, could have cued the recall of experiences whose features had been encoded in terms of them.
Our interpretation of these findings gains plausibility in light of a series
of experiments by Parrott and Sabini (1990). In one study, participants listened to either upbeat or depressing music with explicit instructions to use
the music to get into a happy or sad mood, respectively. They were then
asked to recall an experience they had in high school. In these conditions,
they were more likely to recall a pleasant high school experience when
they were happy than when they were sad, consistent with Bowers (1981)
earlier findings. In a second study, however, participants were told that the
music was being played to investigate aesthetic preferences, and no mention was made of the feelings it would elicit. In this case, however, the effect
of participants feelings on recall was not only eliminated but reversed; that
is, the first event that participants recalled was less pleasant when the music had made them feel happy than when it had not. Moreover, this effect
THE ROLE OF AFFECT
317
was replicated in nonlaboratory settings in which participants feelings

were induced by naturally occurring events. For example, participants recalled less pleasant high school experiences on sunny days than on rainy
ones, and less pleasant experiences after learning they had done well on an
examination than after learning they had done poorly.
Parrott and Sabinis (1990) results should be evaluated in the context of
Niedenthal et al.s (1997) finding that listening to music increased the accessibility of mood-related words in a lexical-decision task even when explicit
references to the mood or emotion being induced by the music were not
made. As noted earlier, words that denote an emotion concept may activate
this concept more quickly when the feelings that define the concept are
also being experienced. In the absence of semantic labels, however, the
feelings may not be sufficiently intense to activate the concept, as Parrot
and Sabinis (1990) findings indicate.
This does not explain why the influence of induced affect on recall was
actually reversed under conditions in which the affect was not labeled. The
reason for this reversal can only be speculated. Unhappy individuals recall
of positive events could result in part from a desire to distract themselves
from their negative mood state by thinking about something that makes
them happy (Parrott & Sabini, 1990; Rusting & DeHart, 2000). However, this
would not explain the relatively better recall of negative events by persons
who feel happy. Wyer and Srull (1989) suggested that when people are
asked to recall an experience they have had at an earlier point in time (e.g.,
high school), they spontaneously activate a global concept of this period in
their lives. This concept might spontaneously elicit affective reactions that
differ from those they happen to be experiencing for other reasons, and
this discrepancy may stimulate thoughts about the implications of this difference. In the conditions considered by Parrott and Sabini (1990), for example, the affect elicited by the high school might on average be less positive than the feelings that happy persons were typically experiencing at
the time and, therefore, might spontaneously stimulate thoughts such as I
felt less happy in high school than I do now. This thought, in turn, could
cue the retrieval of a negative experience that confirms it. Similarly, high
school might stimulate unhappy individuals to think, I felt happier in high
school than I do now, and this cognition may cue the retrieval of a positive
experience.
Be that as it may, however, Parrott and Sabinis (1990) results clearly indicated that affective reactions per se do not increase the recall of evaluatively congruent knowledge representations. This influence occurs only
when the reactions are accompanied by affect-related semantic concepts
that serve as retrieval cues for descriptive knowledge representations that
contain them. In the laboratory, these concepts are often activated by the
procedures that are used to induce different affective reactions. However,
318
CHAPTER 11
affect-related semantic concepts could also be activated spontaneously if

the affect is intense and persons are stimulated to think about why they are
experiencing it. In this regard, Bower and Mayer (1985) suggested that the
affect induced in the laboratory is often not sufficiently intense to stimulate
affect-congruent retrieval. Thus, the effects are often not detected (cf.
Mueller et al., 1991). This speculation is consistent with our own contention
that affective reactions are often not represented in memory, and that even
if a representation is formed of these reactions, it might not be encoded in
terms of a mood or emotion concept. This encoding might only be made if
the feelings are intense or, alternatively, other features of the emotion concept are present as well.
Selective Attention and Encoding
If affect increases the accessibility in memory of semantic concepts and
knowledge representations that are similar to it in valence (cf. Bower, 1981;
Forgas, 1995, 2002), it should also increase the likelihood of comprehending
and thinking about information that can be interpreted in terms of these
concepts. This increased attention might be reflected in (a) longer time
spent reading and thinking about the information, (b) better memory for
the information, and (c) greater influence of the information on judgments
to which it is relevant (Forgas, 2001). Evidence bearing on each of these factors is discussed in turn.
1. Effects on Processing Time
In a study by Forgas and Bower (1987), participants were induced to feel
either happy or unhappy by giving them positive or negative feedback
about their performance on a measure of social adjustment. Then, they
were told to form an impression of someone on the basis of both favorable
and unfavorable behaviors pertaining to competence, intelligence, likeableness, and happiness. Participants spent more time reading and thinking
about behaviors if the behaviors were evaluatively congruent with their
feelings than if they were incongruent. Although data for behaviors pertaining to each attribute separately were not reported, the effects apparently
generalized to behaviors that had nothing to do with affect per se but were
only evaluatively consistent with participants affective state (e.g., behaviors concerning competence, intelligence, etc.).
If Postulate 11.1 and the implications of Niedenthal et al.s (1997) research are valid, Forgas and Bowers findings must be due to factors other
than the affect that participants were experiencing per se. In retrospect,
these factors seem fairly obvious. That is, feedback about ones social adjustment is likely to activate concepts about attributes that reflect this adjustment. As noted earlier, many of the behaviors that participants encoun-
THE ROLE OF AFFECT
319
tered in the later impression formation task were descriptively related to

such attributes (likeableness, competence, etc.). It seems likely that these
concepts increased participants attention to behaviors that exemplified
them independently of the affect they were experiencing. To this extent,
this increased attention could also facilitate the later recall of the behaviors. Forgas and Bowers findings support this prediction as well.
2. Effects on Selective Encoding
The best-known demonstration of the role of affect in the selective encoding of information was performed by Bower et al. (1981) in a study mentioned earlier. In other conditions of this study, participants were induced
under hypnosis to feel happy or sad by recalling a past experience, and
were given a posthypnotic suggestion to retain these feelings while they
read a story. (Thus, unlike the conditions described earlier, participants
were induced to feel happy or sad before reading the story rather than afterwards.) The story concerned one or two characters who experienced a
series of both positive and negative life events. Participants had better recall of events that were evaluatively congruent with their feelings than
events that were incongruent.
The authors did not distinguish between events that were descriptively
related to the affect participants were experiencing and events that were
only similar in valence. Consequently, the consistency of their results with
Postulate 11.1 cannot be directly evaluated. One aspect of their methodology is noteworthy, however. To induce affect, participants were explicitly
told to experience the particular feeling state to which they were assigned
(i.e., to be happy or sad) and to continue to do so once they were no longer
hypnotized. Therefore, semantic concepts of these feeling states were
clearly activated by the instructions. Semantic concepts that are activated
during hypnosis remain accessible in memory despite instructions to forget the experiences that led to their activation (Kihlstrom, 1980). It is therefore likely that the semantic concepts associated with participants feelings,
rather than the feelings themselves, produced the selective encoding effect
that Bower et al. observed.
Even when the induction of affect explicitly requires the activation and
use of affect-related semantic concepts, it does not always influence the selective encoding of information. In a series of six experiments, Mueller et
al.(1991) induced participants to feel happy or sad by having them utter
statements that express these feelings (Velton, 1968). Then, they asked the
participants to learn positively and negatively valenced words and, finally
to recall the words. The statements that were used to induce different feeling states undoubtedly activated semantic concepts pertaining to these
states. Nevertheless, the manipulation had weak and inconsistent effects on
participants later recall of the words they were asked to learn.
320
CHAPTER 11
The Interpretation of Ambiguous Information

As Postulate 2.4 implies and research summarized in chapter 3 indicates,
concepts that have recently been activated in memory as a result of using
them in one context can influence the interpretation of information presented in a later, ostensibly unrelated context. Therefore, if the affect that
people experience spontaneously activates semantic concepts with similar
evaluative implications, it should influence peoples interpretation of information through its mediating impact on the accessibility of these concepts.
According to Postulate 11.1, however, affective reactions only influence the
accessibility of the particular mood or emotion that the reactions exemplify. To this extent, the effects of inducing these reactions should be quite
restricted if they occur at all.
To examine these possibilities, Forgas, Bower, and Krantz (1984) induced
participants under hypnosis to feel either happy or sad by having them recall a past experience that elicited these feelings. Then, participants observed a previously videotaped interaction between themselves and another person with instructions to identify behaviors that exemplified either
social skill or the lack of it. Happy participants interpreted more behaviors
as socially skilled, and fewer behaviors as unskilled, than unhappy participants did. However, the affect-induction task explicitly referred to the feelings that participants were to experience, creating interpretative ambiguities similar to those surrounding the study by Bower et al. (1981) noted
earlier. Moreover, many of the experiences that participants recalled were
likely to have concerned interpersonal relations, and consequently to activate concepts that had implications for social skills. These concepts could
also have influenced participants interpretation of the behaviors they observed.
A study by Innes-Ker and Niedenthal (2002) avoided this ambiguity. Suppose affective reactions influence the interpretation of information through
their mediating effect on the accessibility of concepts to which the information is relevant. Then, similar effects should result from activating these
concepts directly without inducing affective reactions. To examine this possibility, some participants were induced to feel either happy or sad as a
result of listening to music. Others, however, were required to use emotionrelevant semantic concepts in the course of performing a sentence-construction task similar to that used to prime trait concepts in other studies
(e.g., Srull & Wyer, 1979, 1980). Then, as part of an ostensibly unrelated experiment, participants read about a person whose behaviors could be interpreted as conveying either happiness or sadness; specifically:
The sound of rain on the street was relentless. She thought that maybe it was
a day to stay inside and read a novel, or chat on the phone with friends. . . .
THE ROLE OF AFFECT
321
The cat still had not come back from his night on the prowl, and M wondered
where he was. Then the phone rang. . . . It was an old friend . . . who had recently left town. He was calling to say that things were going fine and that he
liked his new job. . . . He would not be back in town over Thanksgiving but
would probably be back over Christmas break. he said he had a great gift for
her in mind.
As expected, participants who personally experienced an emotion judged

the protagonists feelings as consistent with this emotion. However, when
participants were exposed to emotion-related semantic concepts without
actually experiencing the emotions themselves, these concepts had no influence on their judgments of the protagonists feelings.
Taken at face value, these results are quite provocative. That is, they
seem to suggest that the feelings people personally experience influence
the interpretation of other persons emotional experiences, whereas the semantic concepts that are activated by these feelings do not. This conclusion would directly contradict Postulate 11.1. However, the interpretation of
Innes-Ker and Niedenthals findings is not unequivocal. For one thing, the
failure for primed semantic concepts to influence the interpretation of the
information must be reconciled with the abundant evidence in other research that concepts activated through similar procedures do influence the
interpretation of information (see Higgins, 1996, and chap. 3 of this volume).
Note that primed semantic concepts normally influence the interpretation
of information only if these concepts are descriptively applicable (Higgins,
Rholes, & Jones, 1977; but see Stapel & Koomen, 2000, for an exception). In
Innes-Ker and Niedenthals study, this condition might not have been met.
For example, the emotion words employed in the priming stimuli were ecstatic, pleasure, exuberance, and peaceful on one hand, and lonely,
dejected, melancholy, sorrow, and despondent on the other. Although the concepts defined by these adjectives are not incompatible with
the thoughts and behaviors of the protagonist in the story participants
read, they are not so clearly applicable that readers are likely to use them
spontaneously, even if these concepts are accessible in memory. This could
account for the failure of the semantic task to have an impact.
If this interpretation is correct, however, why did the feelings that participants personally experienced have an impact on their interpretations? It is
conceivable that participants judgments were not mediated by their interpretation of the target information at all. Rather, their judgments may have
been based directly on the feelings they were experiencing independently
of the information in the story. That is, participants may have projected the
feelings they were experiencing onto the target independently of the story
content. This informational influence of affect is elaborated in the next
section of this chapter.
322
CHAPTER 11
THE USE OF AFFECT AS INFORMATION

Preliminary Considerations
The research summarized in the previous sections calls into question the
assumption that affective reactions per se exert an influence on judgments
and behavior through its mediating influence on the accessibility of previously acquired concepts and knowledge in memory. These effects are restricted to concepts that are descriptively applicable to the specific emotion or mood that participants are experiencing, and do not extend to
positively or negatively valenced concepts and knowledge in general. To
this extent, the impact of affective reactions on judgments and decisions
may occur for other reasons.
Perhaps the most widely accepted influence of affective reactions is in
one sense the most obvious. Many evaluations of a person, object, or event
are based largely on our feelings about this referent. Thurstones (1959) attitude scaling procedures were an attempt to assess individuals affective
reactions to stimuli independently of their specific beliefs about its attributes. The assumption that attitudes toward an object reflect ones affective reactions toward it also underlies Fishbeins (1967) theory of attitude
formation. More generally, our liking for something is often an expression
of the feelings we experience when encountering it.
In this regard, Schwarz and Clore (1988) postulated that peoples evaluations of a stimulus are often based on a how-do-I-feel-about-it? heuristic. They suggest that affect is used as a basis for judgment in order to
simplify an otherwise difficult task (see also Forgas, 1995). As Wyer et al.
(1999) pointed out, however, affect can often be the primary basis for an
evaluation rather than only being used when other relevant information is
unavailable.
Peoples affective reactions can be used as information about a wide variety of things and, as such, can have a broad impact on information processing. For example:
1. Affect can be used as a basis for evaluating a particular person, object,
or event.
2. Affect can provide the basis for an initial impression of an object that
guides the attitude toward and processing of other available information about the object.
3. Affective reactions can often influence ones perceptions of whether
one is successful in attaining a goal one is seeking and, therefore, can
influence the likelihood of persisting in goal-directed activity.
The remainder of this chapter discusses these possibilities in detail.
323
THE ROLE OF AFFECT
An Initial Test
An initial demonstration of the use of affect as information was performed
by Schwarz and Clore (1983, Experiment 1). This study not only established
a methodology for identifying the impact of affective reactions on judgments, but also identified conditions in which the reactions were likely to
be employed.
Schwarz and Clore (1983) recognized that the affect people experience at
any given time can come from different sources. However, people often cannot distinguish easily between the affect elicited by one source and the affect elicited by another. Consequently, if they base their judgments of a
stimulus on their affective reactions to it, they are likely to confuse these reactions with affect they are experiencing for other, irrelevant reasons, and
so the latter affect can have an influence on these judgments. On the other
hand, suppose people do not base their judgment on affect. Then, extraneous affect should have no impact in any condition.
To examine these possibilities, Schwarz and Clore asked participants to
write about a happy or sad past experience while they were seated in a
strange-looking soundproof booth. Some participants were told that people
often found that the booth made them feel mildly elated after they had been
in it for a while. Others were told that the booth made people feel mildly depressed, and others were given no information about the booths effects.
Participants after writing their essays were then asked to judge their satisfaction with their life as a whole. These judgments are summarized in Table
11.1 as a function of the type of experience participants wrote about and
the ostensible effects of the booth.
Participants who were given no information about the booths effects reported greater life satisfaction when they had written about a happy experience than when they had written about an unhappy experience. Considered
in isolation, this difference might be attributed to the content of the story
that participants wrote rather than the affect they were experiencing. If this
were the case, however, the alleged effects of the booth on their feelings
should have had no impact on their judgments. In fact, with participants who
TABLE 11.1
Judgments of Life Satisfaction as a Function of the Affective Quality of a
Recalled Personal Experience and Ostensible Characteristics of the Booth
Ostensible Effect of Booth
Happy recalled experience

Sad recalled experience
Depressing
No Effect
Elating
9.6
8.6
8.6
5.7
9.7
4.4
Note. Adapted from Schwarz and Clore (1983).
324
CHAPTER 11
had written about a sad experience, telling them that the booth would make
them feel sad eliminated the negative impact of the story on their judgments
of life satisfaction, whereas telling them that the booth would make them
happy increased these judgments relative to no-information conditions. In
other words, participants who were told that the booth made them unhappy
attributed their feelings to the booth and did not use it as a basis for judgment. In contrast, participants who experienced unhappiness despite being
told that the booth would make them happy believed that they must be particularly unhappy about their life as a whole and, therefore, judged themselves to be even less satisfied than they would otherwise.
It is interesting to note that the effects of the booth were not apparent
when participants had written about a positive life experience and, therefore,
were feeling happy. These participants appeared to base their judgments of
life satisfaction on their feelings regardless of the information they were
given about the booth. The asymmetric effects observed in the two affectinduction condition could be due to factors similar to those underlying Arkin
et al.s (1976) research described in chapter 9. That is, participants whose affect had favorable implications for themselves were inclined to accept these
implications and use them as a basis for judgment without seeking alternative explanations for their feelings. In contrast, people whose affect had undesirable implications resisted using it as a basis for judging themselves unless no alternative explanation for their feelings was available.
The Role of Narratives in Affect Elicitation
Affective reactions can be elicited in many ways, as noted earlier. Affect
that is induced by proprioceptive feedback (cf. Strack, Martin, & Stepper,
1988; Zajonc, Murray, & Inglehart, 1989) can also influence judgments to
which it is relevant (e.g., the amusement elicited by cartoons). The role of
narratives in affect elicitation was confirmed in a series of studies by Strack,
Schwarz, and Gschneidinger (1985). They found that when people listed
past experiences in abstract terms that were unlikely to elicit affect, they
used these experiences as standards of comparison in judging their current
life satisfaction, reporting themselves to be less satisfied when the past experiences were favorable than when they were unfavorable. When participants wrote detailed narratives of their past experiences, however, this activity elicited affect, and this affect had a positive influence on satisfaction
judgments.
In a particularly provocative study (Strack et al., 1985, Experiment 3),
participants were either asked to explain why a particular experience occurred or to describe how it occurred. Writing explanations for happy and
sad events did not elicit affect, and the favorableness of the experiences
that participants explained had contrast effects on the life satisfaction they
THE ROLE OF AFFECT
325
reported later. However, writing descriptions of how the experience occurred did elicit affect and consequently had effects on judgments similar
to those reported by Schwarz and Clore (1983).
AFFECTIVE VERSUS DESCRIPTIVE BASES

FOR JUDGMENT
As noted, not all evaluations that people make are based on affective reactions (Zanna & Rempel, 1988). Many judgments are based on purely descriptive considerations. Thus, although we might like a pair of jeans because
they feel good when wearing them, we might like a computer because it is
fast and has a lot of storage. Similarly, we might evaluate a dinner partner
favorably because her physical attractiveness and effervescent personality
elicit positive affect, we might evaluate a research assistant favorably because she gets work done on time and records data accurately. Other judgments can be based on either affective or descriptive criteria, depending on
which happen to be more accessible (Postulate 3.1). This latter contingency
has been identified in research in quite different areas, concerning judgments of personal competence in one case and evaluations of commercial
products in the other.
Judgments of Personal Competence
and Global Self-Esteem
General characteristics such as life satisfaction or global self-esteem are
typically not based on a single event, but rather, on more global feelings
that reflected the accumulated influence of many experiences. Consequently, affective reactions may be a primary basis for judging these characteristics. In many cases, however, descriptive criteria are potentially applicable as well. In such instances, whether descriptive or affective criteria
are used can depend on the relative accessibility of these criteria in memory. If the content of a narrative has direct implications for a judgment and
if these implications are accessible in memory, the effects of these implications can override the influence of affect.
Levine, Wyer, and Schwarz (1994) provided an example. Participants
wrote about a happy or an unhappy past experience they had had in either
an achievement situation or a social situation. After doing so, they estimated their competence in both types of situations and their social competence as well as their global self-esteem. We reasoned that judgments of
competence in the domain to which their story directly pertained would be
based on the descriptive implications of the story, as these implications
were highly accessible in memory. However, their competence in the do-
326
CHAPTER 11
main to which their narrative was descriptively irrelevant should be based

on the affect they were experiencing.
This hypothesis is difficult to evaluate when descriptive and affective criteria for judgment have similar implications. However, a content analysis of
participants narratives showed that although participants typically took responsibility for both positive and negative achievement experiences they
wrote about, they attributed responsibility for both good and bad social
experiences to situational factors (e.g., the other persons involved in the
situation). Based on these differences, we predicted that participants judgments of their achievement competence would be influenced by the favorableness of the achievement experience they wrote about, but that their
judgments of social competence would not be influenced by the type of social experience they described. On the other hand, participants judgments
of both achievement competence and social competence should be influenced by the affect that was generated by writing about an experience in a
domain to which the competence being judged was unrelated.
Results shown in Table 11.2 confirm these predictions. That is, participants judgments of their achievement competence were influenced by the
favorableness of the achievement experience they had written about,
whereas their judgments of their social competence was not influenced by
the favorableness of the interpersonal situation they had described. On the
other hand, participants judgments of both achievement competence and
social competence were influenced by writing about an affect-eliciting experience in the domain to which the competence they were judging was irrelevant. (Thus, somewhat ironically, judgments of social competence were
more influenced by writing about an objectively irrelevant affect-eliciting
experience than about a relevant one.) Finally, participants overall judgments of their self-esteem were influenced by the favorableness of the description they had written about regardless of the domain, suggesting that
these global judgments were based on affect in all cases.
TABLE 11.2
Judgments of Competence as a Function of the Domain
and Affective Quality of Participants Recalled Experiences
Recalled experience in same domain

Happy experience
Unhappy experience
Recalled experience in different domain
Happy experience
Unhappy experience
Judgment of
Achievement Competence
Judgment of
Social Competence
4.77
4.37
4.38
4.38
4.87
4.67
4.46
3.46
Note. Adapted from Levine, Wyer, and Schwarz (1994).
THE ROLE OF AFFECT
327
Consumer Judgment
Judgments of competence can often be based on either descriptive or affective criteria, and so participants are likely to use whichever criterion comes
to mind most easily at the time judgments are made. This may also be true
in the consumer domain. Evaluations of commercial products, for example,
are often based on utilitarian considerations as well as hedonic or affective
criteria. A refrigerator, for example, is likely to be evaluated on the basis of
its ability to perform the functions for which it is designed. On the other
hand, a cashmere sweater, or strawberry shortcake, could be judged on the
basis of either affective criteria (how it feels or tastes) or utilitarian considerations (durability, healthfulness, etc.). In the latter case, people are most
likely to base their evaluations on affect if (a) the stimulus is one that is typically judged on the basis of affective (hedonic) criteria, (b) the information
about the stimulus is conveyed in a way that spontaneously elicits positive
or negative feelings, or (c) peoples objectives at the time of judgment require the use of affective criteria.
The latter contingency was demonstrated by Pham (1998). Participants
who had been induced to feel happy or unhappy by recalling a past experience were asked to read a description of a new movie and indicate if they
would like to attend a preview of it. Before reading the description, some
participants were given instructions that emphasized hedonic criteria for a
decision (e.g., having a good time, enjoying life, etc.), whereas others were
given a utilitarian reason for seeing it (its relevance to participating in a
subsequent study for pay). Participants in the first condition expressed
stronger intentions to see the movie if they were feeling happy than if they
were not, suggesting that they used the affect they were experiencing as a
basis for their judgment. In contrast, participants intentions in the second
condition were unaffected by the affect they were experiencing.
The use of hedonic or utilitarian criteria can also depend on their relative accessibility at the time of judgment. However, unless situational factors make descriptive criteria particularly relevant (as in Levine et al.s 1994
study), these criteria may be relatively inaccessible. In contrast, the affective reactions can be activated very quickly, before a detailed cognitive
analysis of the utilitarian features of the description is performed. (For a
more general argument that affective responses to a stimulus as a whole occur more quickly than responses to its individual features, see Zajonc,
1980.) One implication of this is that if a stimulus can potentially be evaluated on the basis of both hedonic and utilitarian criteria, hedonic criteria
may often take priority. That is, utilitarian considerations may come into
play only when people have both the motivation to evaluate these criteria
and the ability to do so.
A study by Shiv and Fedorikhin (1999) supported this contention. Participants were given a choice between a chocolate cake and a fruit salad as a
328
CHAPTER 11
reward for participating in the study. The authors assumed that the sight of
the chocolate cake would elicit positive affect, and that the choice of this
option would reflect the use of these reactions as a basis for their decision.
In contrast, the choice of a fruit salad was more likely to reflect the use of
utilitarian (health-related) criteria. The authors further speculated that the
latter criteria would come into play only if participants had the motivation
and ability to think carefully about their decision.
To evaluate this possibility, some participants were placed under high
cognitive load at the time they made their choice by being asked to remember a multiple-digit number for use in a later part of the study. Others were
put under low load. Participants in the latter condition tended to choose the
fruit salad, suggesting that they based their decision on utilitarian (i.e.,
health-related) criteria. Participants who were distracted from thinking about
their decision, however, were more inclined to choose the chocolate cake.
The implications of these findings may extend beyond the domain of
consumer judgment. For example, a faculty member who wishes to hire a
research assistant may consider utilitarian criteria (background, experience, etc.) to be particularly relevant. In a personal interview, however, a
candidates personality and physical attractiveness might elicit affective reactions that are used as a basis for judgment in addition to other considerations. Moreover, if the interviewer meets the candidate before receiving
descriptive information about his or her qualifications, these affective criteria may predominate, and descriptive criteria may be taken into account
only if the interviewer has the motivation and ability to think more carefully
about the candidates credentials.
Effects of Initial Appraisals on the Use

of Affective Versus Descriptive Criteria
In most research on the use of affect as a basis for judgment (for reviews,
see Clore, Schwarz, & Conway, 1994; Schwarz & Clore, 1996; Wyer, Clore, &
Isbell, 1999), the influence of affect is assumed to occur at the time of judgment. If people make a spontaneous appraisal of a stimulus at the time they
first encounter it (Lazarus, 1982), and if this appraisal elicits affective reactions, these reactions can give rise to an affect-based initial impression of
the stimulus that influences participants later judgments over and above
the effects of more specific information they receive later.
This possibility was confirmed by Catherine Yeung (2003; Yeung & Wyer,
in press). She noted that peoples first encounter with a product often occurs when they notice it in a store window or see a picture of it in a magazine, and that they spontaneously appraise the product on the basis of this
encounter. If this appraisal elicits affect, it not only can have an impact on
329
THE ROLE OF AFFECT
evaluations of the product but also can bias the search for additional information about it.
1. Experiment 1
To examine these possibilities, participants in one experiment took part
in three conditions. In no-picture conditions, participants who had been induced to feel either happy or unhappy as a result of writing about a life experience were given verbal descriptions of a pair of sports shoes consisting
of three favorable and three unfavorable features. Some participants were
told to base their judgment on hedonic criteria (how it would feel to wear
the shoes), and others were told to use utilitarian criteria (durability, workmanship, etc.). Participants evaluations of the products in these conditions, shown in the top left panel of Table 11.3, indicate that the extraneous
affect that participants were experiencing influenced their evaluations
when they were told to use hedonic criteria but had little effect when they
were told to use utilitarian criteria. These data, which are consistent with
those reported by Pham (1998), confirmed the assumption that when participants are told to use hedonic criteria as a basis for evaluations, they imagine the feelings they would experience when using the product, and the affect they were experiencing for other reasons influence their assessment of
these feelings. When participants were told to use utilitarian criteria, however, they did not consult their feelings and so the extraneous affect they
were experiencing had little effect.
In two other conditions of this study, however, participants were
shown an attractive picture of the product before they received descripTABLE 11.3
Product Evaluations and Recall of Attribution Information
as a Function of Induced Affect, Judgment Criterion,
and Information Presentation Conditions
Affect Induced
Before Picture
No Picture
Positive
Affect
Negative
Affect
Product evaluation (Experiment 1)

Hedonic criterion
6.50
Utilitarian criterion
4.00
Product evaluation (Experiment 2)
Hedonic product
7.11
Utilitarian product
3.41
Number of positive attributes recalled
Hedonic product
2.55
Utilitarian product
1.14
Affect Induced After

Picture
Positive
Affect
Negative
Affect
Positive
Affect
Negative
Affect
5.25
5.25
3.75
3.78
3.46
3.89
3.50
4.33
4.67
6.38
3.78
4.88
(Experiment 2)
1.32
2.13
1.80
1.62
4.38
3.63
5.78
4.11
5.30
4.48
1.32
1.29
1.68
1.32
1.59
1.26
4.10
4.12
Note. Adapted from Yeung and Wyer (in press).
330
CHAPTER 11
tions of its attributes. Moreover, the pictures in affect-picture conditions

were shown after participants had been induced to feel happy or unhappily, whereas in picture-affect conditions, the pictures was conveyed at the
outset. We assumed that the picture would stimulate participants to make
a spontaneous appraisal of the product that elicits affective reactions, and
that these reactions would then provide the basis for an initial impression
of the product that they would use as a basis for their evaluation of the
product they reported later. Moreover, this should be true regardless of
the specific criteria they would apply in the absence of this affect-eliciting
appraisal. Therefore, if participants are experiencing affect for other reasons at the time they made this appraisal, this affect should also influence
the impression they form and the judgments they ultimately make on the
basis of it. On the other hand, suppose extraneous affect is not induced
until after participants have made an appraisal of the product and formed
an impression on the basis of it. Then, this affect should have no impact
on their initial impression and, therefore, should not influence the evaluation they report later.
Data in the last two sections of Table 11.3 (top) confirm these predictions. That is, the extraneous affect that participants were experiencing had
a positive impact on judgments when it was induced before pictures were
presented, and this was true regardless of the criteria they were told to use.
When extraneous affect was not induced until after participants saw the
product, however, it had no influence on judgments whatsoever.
2. Experiment 2
A second experiment confirmed these findings and, in doing so, provided further evidence bearing on the assumptions underlying their interpretation. The design of this study was similar to the first with two exceptions. First, rather than explicitly instructing participants as to which
criterion they should use, we asked them to evaluate products that they
were likely to evaluate spontaneously on the basis of either hedonic criteria
(salad dressing) or utilitarian criteria (backpacks). Second, participants after receiving the product information were told to recall all of the products
features they could remember. (Because most participants could recall all
six pieces, we considered only the first three features they recalled, and
used the number of these features that were favorable as an index of the
relative attention paid to these features in relation to unfavorable ones.)
Product evaluations, shown in the middle third of Table 11.3, confirm
conclusions drawn from the first study. That is, when a picture of the product was not presented, participants judged the hedonic product more favorably when they were feeling happy than when they were feeling unhappy,
whereas they evaluated utilitarian products similarly regardless of the ex-
THE ROLE OF AFFECT
331
traneous affect they were experiencing. When participants had received a

picture of the product, however, the extraneous affect they were experiencing at the time they saw the picture had an impact on their judgments of
both types of products, whereas affect that was not induced until after they
saw the picture had no impact. Note that this same pattern of effects was
reflected in the favorableness of the attribute information that participants
recalled, as shown in the bottom third of Table 11.3.
The parallel effects of experimental variables on recall and judgments
has two possible interpretations. First, participants who formed affectbased impressions selectively attended to attributes of the product that
confirmed the evaluative implications of these impressions, and then used
this attribute information as a basis for judgments. Second, participants
who formed an initial impression of the product based their judgments directly on this impression independently of the specific attribute information presented. However, they thought more about information they received that was consistent with this impression in order to confirm its
validity (for evidence of such confirmatory information seeking, see
Bodenhausen & Wyer, 1985; Chernev, 2001; Snyder & Cantor, 1979). Although these alternative possibilities cannot be clearly distinguished on
the basis of the data available, the correlation between product evaluations
and the favorableness of the attribute information recalled was very low (r
= .14), suggesting that in fact, the two measures are relatively independent.
In any event, the two studies in combination clearly suggest that when
participants read verbal descriptions of a products attributes without having seen a picture of it, they only based their evaluation of the product on
their affective reactions if they believed that hedonic criteria were relevant
to this evaluation. When they saw a picture of the product, however, the
picture stimulated a spontaneous appraisal of the product that elicited affective reactions, and participants formed an initial impression of the product on the basis of these reactions. Then, they used this affect-based impression as a basis for their later evaluations of the product regardless of
the criteria they would otherwise have applied.
3. A Contingency
Yeung and Wyers (in press) results suggested that affect can exert its
impact on judgments of a stimulus through its influence on the spontaneous appraisal that people make of the stimulus at the time they first encounter it. Not all initial appraisals elicit affect, however. Whether this is so
could depend in part on previously learned associations of affect with features of the stimulus on which appraisals are based. When an appraisal
does not elicit affect, subsequent judgments could still be determined by
the initial impression that is formed on the basis of it. However, affective reactions may not have much influence.
332
CHAPTER 11
Yeung (2003) obtained evidence of this contingency in a study that paralleled the experiments described in the previous paragraphs. Rather than
giving participants a picture of the product, however, she told them that
the product they would judge was being produced by a company that was
well known for either hedonic products (Pepsi) or utilitarian ones (Bonaqua, a brand of distilled water). Either before or after conveying this information, participants were induced to feel happy or unhappy by writing
about a past experience. Finally, they were given information about a pair
of sports shoes that were ostensibly about to be produced by the company
whose name had been provided earlier.
When the brand was associated with a product that was judged on the
basis of affective criteria, results paralleled those obtained under comparable conditions of the earlier studies. That is, extraneous affect had a positive influence on participants evaluations of the brand extension when it
was induced before participants learned of the brand, but not when it was
induced afterward. When the brand was associated with a product that was
judged on the basis of utilitarian criteria, however, the extraneous affect
that participants were experiencing had no influence regardless of when it
was induced.
These results further strengthen the conclusion that judgments of a stimulus are often induced by an initial appraisal that is made at the time the
stimulus is first encountered independently of more specific information
about the stimulus that is conveyed afterwards. In these conditions, however, affective reactions only enter into this process when the appraisal
elicits affect. Moreover, extraneous affect that participants happen to be experiencing only has an impact when it is experienced at the time the appraisal is made and, therefore, becomes confused with the affect that is automatically elicited by the appraisal-inducing stimulus.
AFFECT-CONFIRMATION PROCESSES
In the research by Yeung (2003; Yeung & Wyer, in press) affect was presumably elicited by a spontaneous global appraisal of the product as a whole
that was independent of any specific product features. However, affect can
obviously be elicited by specific features of a stimulus as well. A politicians
position on a particular issue, for example, could elicit positive or negative
affective reactions in someone who favors or opposes this position. In the
product domain, a sweaters material can convey how nice it would feel to
wear it, and could elicit affect for this reason. Individuals who wish to evaluate such stimuli on the basis of affective considerations are likely to pay attention to features that elicit affect and use these features rather than others as a basis for judgment. If the individuals are also experiencing affect
THE ROLE OF AFFECT
333
for other reasons, however, this affect could have two effects. First, they
might confuse their affective reactions to these features with the affect they
are experiencing for other reasons, and consequently could interpret these
features as more or less favorable. On the other hand, their extraneous feelings could influence the importance that they attach to the affect-eliciting
features and, therefore, the weight they attach to these features when computing a judgment.
This latter possibility is formalized in the affect-confirmation theory proposed by Adaval (2001). According to this theory, participants who base
their judgments of a stimulus on hedonic considerations assess not only
their affective reactions to the stimuluss individual features but also the extent to which their perceptions of these reactions are valid. If extraneous affect that they happen to be experiencing is similar to the affect that is elicited by a stimulus feature, it may appear to confirm their reactions to this
feature and, therefore, may give them more confidence in the implications
of these reactions. On the other hand, if extraneous affect they are experiencing differs from that elicited by a stimulus feature, they might interpret
this difference as ambivalence and thus might decrease their confidence in
the validity of their assessment of the features affective implications. Consequently, if individuals base their judgments of the stimulus on their affective reactions to it, they should attach greater weight to features that elicit
affect similar to the feelings they are experiencing for other, unrelated reasons than to features that elicit different reactions.
Empirical Evidence
Two studies by Adaval (2001) evaluated these hypotheses. In one, participants who had been induced to feel either happy or unhappy received information about a pair of jeans consisting of either a favorable or an unfavorable brand name (Levis vs. Rustler) and either a favorable or an
unfavorable attribute (preshrunk vs. not preshrunk). In some cases, participants were told to base their evaluations of the jeans on hedonic criteria
(how it would feel to wear them) and the attribute was described in a way
that emphasized its implications for comfort. In other cases, they were told
to base their judgments on utilitarian criteria, and the attribute description
emphasized its implications for the quality of construction. Participants after reading the product description evaluated the product and then reported their confidence that this evaluation was correct.
Note that differences in the impact of the attribute information on product evaluations under different affect-induction conditions could indicate either that the attribute information was weighted more heavily in making
their judgments or, alternatively, that the attribute information was interpreted differently (i.e., as more or less extreme) under these conditions. To
334
CHAPTER 11
avoid this ambiguity, Adaval inferred the relative weight attached to the attribute information from the impact of the brand information that accompanied it. If affect information receives relatively greater weight when its
evaluative implications are consistent with the extraneous affect that participants were experiencing, the relative influence of the brand information
that accompanies it should be relatively less.22
Data summarized in the top half of Table 11.4 confirm this hypothesis.
That is, when participants were told to evaluate the product on the basis of
hedonic criteria, brand information had less impact on their judgments
when the attribute description was evaluatively consistent with the extraneous affect that participants were experiencing than when it was inconsistent, indicating that the attribute information received greater weight in the
former condition. When participants were told to use utilitarian criteria,
however, brand had greater influence in the former condition than the latter. This reversal suggests that participants in this condition may have intentionally avoided basing their judgments on affective criteria and, in fact,
overcompensated for its influence, attaching less weight to the attribute information than they otherwise would. Interestingly, participants reported
greater confidence in their judgments when the evaluative implications of
the attribute information and those of the extraneous affect they were experiencing were consistent, and this was true regardless of the objective they
were pursuing. Thus, participants were not only more confident about using their affective reactions to the attribute information as a basis for judgments when their evaluations were based on hedonic criteria but were also
more confident about avoiding its use when the basis for their judgment
was utilitarian.
A second experiment confirmed these predictions under conditions in
which the criterion that participants used was activated spontaneously by
the type of product that participants were asked to consider. Normative
data suggested that in the absence of explicit instructions, participants
were likely to evaluate jeans on the basis of hedonic criteria and backpacks
on the basis of utilitarian criteria. A second experiment was conducted using these two product categories but without instructing participants about
the criteria they should employ. The bottom half of Table 11.4 shows that as
expected, brand had less impact on evaluations of jeans when the
evaluative implications of the attribute information were consistent with
the extraneous affect that participants were experiencing, whereas it had
greater impact on evaluations of backpacks in this condition.
22
The use of this criterion assumes that judgments are a weighted average of the implications of the two pieces of information considered separately (cf. N. Anderson, 1971, 1981). Although this assumption is often questionable (cf. Wyer & Carlston, 1979), the validity of the assumption in product evaluation situations is viable (cf. Bettman, Capon, & Lutz, 1975; Troutman
& Shanteau, 1976).
22
335
THE ROLE OF AFFECT

TABLE 11.4
Effects of Brand Favorableness on Product Evaluations as a
Function of Processing Criterion and Attribute-Affect Consistency
Attribute Consistent
With Induced Affect
Experimentally-induced judgment criterion (Experiment 1)
Hedonic
2.21a
Utilitarian
2.83
Spontaneously activated judgment criterion (Experiment 2)
Jeans (hedonic)
3.0
Backpacks (utilitarian)
2.0
Attribute Inconsistent
With Induced Affect
2.56
1.85
4.2
1.3
Note. Adapted from Adaval (2001).

aEffect of brand favorableness is inferred from the difference in product evaluations when
the brand was favorable and evaluations when the brand was unfavorable.
Implications for Political Judgment

Adavals (2001) findings, and her affect-confirmation theory more generally,
are noteworthy in the context of considerations raised earlier in this chapter. That is, the greater weight of affect-consistent information in judgments
has often been attributed to the greater accessibility of affect-consistent semantic concepts and knowledge in memory and, therefore, the greater likelihood of encoding this information into memory (cf. Forgas & Bower, 1987).
However, only two pieces of product information were presented in
Adavals studies, and this information was available at the time of judgment. In other words, the differential accessibility of the information in
memory cannot plausibly account for the effects. Moreover, the effects are
unlikely to result from the encoding of the information in terms of evaluatively consistent semantic concepts at the time it was received (Niedenthal & Setterlund, 1994).
To the extent affect-confirmation processes account for the results of
Adavals studies, it seems likely that the influence of affect-consistent information observed in other research studies are influenced by similar processes. A study by Isbell and Wyer (1999) is worth considering in this regard. Happy and sad participants were asked to evaluate a political
candidate on the basis of a news article that described his stands on a number of social issues. In addition, their intrinsic and extrinsic motivation to
think carefully about the candidates issue positions was varied.23 Partici23
Intrinsic motivation was inferred from responses to a self-report index of the strength of
participants partisanship (i.e., their party identification). To induce extrinsic motivation, participants were told that the individual they would evaluate was a current candidate for office in
their area, and that we were consequently interested in whether they would vote for him. In contrast, low motivation was induced by telling participants we were primarily interested in
whether the article was clearly written, independently of its specific content.
23
336
CHAPTER 11
pants after reading the news article reported their impressions of the candidate and then indicated their personal agreement with each of the issue
stands he had espoused, along with an equal number of positions on which
the candidates stands were unknown.
When participants were neither intrinsically nor extrinsically motivated
to evaluate the candidate, they judged him more favorably when they were
feeling happy than when they were not. Moreover, they reported stronger
agreement with the candidates issue stands in the former case than the latter. (In contrast, their agreement with issue positions that were not associated with the candidate were not influenced by the affect they were experiencing.) Thus, participants mood influenced their perceptions of the
favorableness of the candidates issue stands and these perceptions, in
turn, determined their liking for the candidate. This conclusion was confirmed by covariance analyses. Specifically, the effects of experimental variables on candidate evaluations were eliminated when their effects on issue
agreement were covaried, whereas the effects of these variables on issue
agreement remained significant when evaluations of the candidate were
covaried.
Isbell and Wyer (1999) concluded that participants interpreted their reactions to the candidates issue stands as generally more favorable when
they were feeling happy than when they were not. However, affectconfirmation processes could also underlie these findings. For example,
happy participants may have had more confidence in the validity of their
reactions to issue stands that elicited positive affect (i.e., stands with which
they agreed) than in the validity of their reactions to issue stands that elicited negative affect (stands with which they disagreed). Unhappy participants, on the other hand, may have had less confidence in their reactions
to the former statements than their reactions to the latter. The differential
weighting of the issue stands that resulted from these differences in confidence could have influenced candidate evaluations for much the same reasons suggested by Adaval (2001). Isbell and Wyers study does not distinguish between these interpretations.
ADJUSTMENTS FOR BIAS

The aforementioned results by Isbell and Wyer (1999) were evident only
among participants who were neither intrinsically nor extrinsically motivated to make accurate judgments. When people are more motivated to be
correct, they may attempt to distinguish between the various sources of affect they are experiencing. Furthermore, if they believe their perception of
their feelings toward the stimulus they are judging to be contaminated by
THE ROLE OF AFFECT
337
extraneous sources of affect, they might try to adjust for this biasing influence. As noted in chapter 10, however, people are often unclear how much
adjustment is necessary. If people do not adjust enough, the extraneous affect they are experiencing would continue to have a positive influence on
their judgments. If they adjust too much, on the other hand, it would appear
to have a contrast effect.
The latter appeared to be the case in Isbell and Wyers study. That is, although the extraneous affect that participants were experiencing had a positive impact on their judgments when they were unmotivated to make accurate evaluations, it had precisely the opposite effect on judgments by
individuals who were either intrinsically or extrinsically motivated to be
correct (see footnote 24).
Effects of Ability
Even if people are motivated to adjust for the biasing influence of extraneous affect, they may be unable to do so. Ottati and Isbell (1996), for example, found that extraneous affect had a positive influence on evaluations of
a candidate by participants with relatively little knowledge about politics,
but had a contrast effect on evaluations by participants with high political
expertise. Perhaps both groups of participants adjusted for the biasing influence of the extraneous affect they were experiencing. However, participants with little expertise had difficulty evaluating the implications of the
candidates issue positions and the cognitive resources they had to devote
to this activity prevented them from adjusting sufficiently. On the other
hand, experts found it easy to evaluate the implications of the candidates
issue positions and so they could devote more effort to distinguishing their
reactions to these stands from other sources of affect. However, these participants adjusted too much, resulting in a contrast effect of extraneous affect on their evaluations.
Adjustment Versus Discounting
Adjustments for bias are most likely to occur when people consider their affective reactions to the primary basis for their judgments of an object but
cannot distinguish clearly between these reactions and affect from other
sources. In many instances, however, nonaffective criteria for judgment
may be as or more appropriate bases for judgment. In these instances, people who believe that the affective reactions are biased by extraneous factors might ignore their feelings entirely and default to other, nonaffective
criteria. In this case, the affect they are experiencing would not have a contrast effect on judgments. Rather, it would have no influence at all.
338
CHAPTER 11
One example of discounting was obtained by Schwarz and Clore (1983,

Experiment 2). In this study, residents of downstate Illinois were contacted
by telephone on either a sunny or a rainy day and asked to participate in a
short telephone interview in which they reported their general life satisfaction. Participants judged themselves to be more satisfied with their life on
sunny days than on rainy ones, suggesting they based their judgments in
part on the affective reactions activated by the weather. In other conditions, however, the interviewer, who was ostensibly calling from Chicago,
incidentally asked participants, By the way, hows the weather down
there?, thereby calling participants attention to an extraneous source of
their affect. In this condition, participants apparently ignored the affect
they were experiencing and based their judgments on other criteria (e.g.,
their knowledge about life events that they retrieved from memory). Consequently, weather-induced affect had no impact on their judgments at all.
Calling participants attention to an extraneous source of affect does not
always lead participants to discount their feelings, however. When participants in Schwarz and Clores (1983, Experiment 1) study had written about
an unhappy life event in a booth they believed would make them happy,
they apparently inferred that they must be feeling even more negatively
about themselves than did participants who were not given information
about the booth. Consequently, they reported lower life satisfaction than
other participants did. A study by Gasper and Clore (1998) found that
chronically anxious individuals do not discount the influence of situational
factors on the negative affect they are experiencing. Calling these individuals attention to extraneous sources of negative affect apparently made salient to them that their feelings were not due to transitory factors alone.
Consequently, it increased the influence of the negative affect they were experiencing rather than decreasing it.
In this regard, it is undoubtedly an oversimplification to assume that
when both affective and nonaffective criteria for judgment are available,
people may use one set of criteria or the other but not both. More generally, people use both types of criteria, with the relative weight attached to
each depending on perceptions of its relative validity and also the time and
effort that are required to construe its implications. As noted earlier, affective reactions to a stimulus can often occur spontaneously upon exposure
to it, and so this criterion is likely to be weighted heavily when participants
motivation or ability to think about other available criteria are low. However, if participants have the time and motivation to apply other relevant
criteria, they may use these criteria as well. The impact of situational factors on the relative use of alternative criteria for judgment was identified by
Shiv and Fedorikhin (1999) and also by Albarracin and Wyer (2001) in a
study of communication and persuasion. The latter study is described in
more detail presently.
THE ROLE OF AFFECT
339
AUTOMATIC INFLUENCES OF AFFECT

ON INFORMATION PROCESSING
Peoples affective reactions can influence not only their judgments but also
their overt behavior and decisions. These effects could be either deliberative or automatic. In the first regard, people are sometimes motivated to behave in a way that will maintain a positive mood or, alternatively, will rid
themselves of negative feelings. As noted earlier, however, the affective reactions one experiences, like other internally generated stimuli, can activate a production of the sort described in chapter 1 (J. Anderson, 1983; E. R.
Smith, 1984, 1990). That is, they can form part of the configuration of stimuli
[X] in a production of the form if [X], then [Y], where [Y] is a sequence of
responses that are elicited automatically once the preconditions are met.
These effects may be captured in a postulate:
Postulate 11.2. The affective reactions that one experiences, along with concepts and cognitions that are stimulated by an appraisal of a stimulus situation,
can compose a precondition for a previously learned sequence of responses
(i.e., a cognitive production). This response sequence may be activated spontaneously, with little conscious cognitive mediation, under conditions in
which the precondition is met.
Despite the vagueness of Postulate 11.2, some general predictions seem

tenable. One particularly interesting hypothesis concerns the use of holistic
vs. piecemeal processing strategies to interpret information and make judgments. The impact of affect on the use of these strategies is described in the
pages that follow.
The Influence of Affect on the Use
of Categorical Criteria
Bless (2001; Bless et al., 1996) postulated that positive affect is associated
with a disposition to use broad concepts and general knowledge structures
to perform experimental tasks, whereas negative affect is associated with attention to situational details. These differences could result from social learning. That is, positive affect may become associated with conditions in which
success has not required a careful scrutiny of the situation or task to be performed. In contrast, negative affect could become associated with failure and
negative events that can only be avoided by attention to situational details
(cf. Schwarz, 1990). As a result of these learned associations, the experience
of positive affect can often stimulate a production that involves the use of
general (e.g., categorical) criteria in responding to an achievement task,
whereas negative affect may activate a production that involves a more detailed analysis of the task requirements and the information bearing on them.
340
CHAPTER 11
Furthermore, these alternative productions might be activated spontaneously when the affect-based preconditions for their occurrence are met, and
might guide behavior with little if any conscious cognitive mediation.
The different productions that are activated by positive and negative affect require different types of knowledge. The procedures activated by positive affect involve the use of general concepts and knowledge structures. In
contrast, the procedures activated by negative affect may require more
specific concepts that can be used to interpret informational details. This
difference should be reflected in the type of concepts and knowledge that
are used to interpret and organize new information. At least two studies
support this possibility.
Bless et al. (1996) reasoned that if people who experience positive affect
spontaneously activate general knowledge structures to use in processing information, they should be able to process this information more efficiently.
Moreover, their use of these knowledge structures should free cognitive resources for use in other, unrelated cognitive activity. To evaluate this possibility, participants who were induced experimentally to feel happy or unhappy listened to two tape-recorded stories about daily life events (e.g.,
making a telephone call from a public phone booth) while they performed a
clerical task. Both recognition memory for the events mentioned in the audiotapes and accuracy in performing the clerical task were recorded.
As expected, happy participants were more likely than unhappy ones to
use a prototypic script (Schank & Abelson, 1977) to comprehend the taped
information. That is, happy participants were more likely than unhappy
ones to report prototype-consistent events as having been mentioned, but
were less likely than unhappy participants to report atypical events. To the
extent that participants apply a general prototype in comprehending the
tape-recorded information, however, they should be able to devote more
cognitive resources to the clerical task. This was also the case. Happy participants performed better on the secondary, clerical task than unhappy
participants did.
A quite different study by Dienes (1996) has similar implications. In this
study, happy and unhappy participants were asked to learn a list of words
that exemplified either easily identifiable concepts (animals, countries, etc.)
or obscure ones (e.g., things made of wood). Learning the list could be facilitated by (a) attending to the details of the individual words, (b) mentally
organizing the words in terms of the concepts they exemplify, or (c) both.
(For evidence that the organization of information items into categories improves learning, see Hunt & Einstein, 1981.) Dienes reasoned that when the
concepts exemplified by the stimulus words were obvious, both happy and
unhappy participants would spontaneously identify and use these concepts
to organize the stimuli. When the concepts were more obscure, however,
only happy participants should identify them spontaneously, so only these
participants learning of the stimulus words should be facilitated. Consis-
THE ROLE OF AFFECT
341
tent with this reasoning, happy participants had better memory than unhappy participants for the hard-to-categorize words but not for the easy-tocategorize words.
The conclusions that persons who experience positive affect bring categorical and general knowledge structures to bear on tasks they perform
does not imply that affective reactions per se cue the retrieval of these
knowledge structures. Rather, this retrieval is mediated by cognitive productions that require different types of declarative knowledge. The preconditions for activating a production include situational features as well as
affective reactions. In some cases, the cognitive productions that are activated by situation-specific features might not require the types of knowledge of concern in the studies by Dienes and Bless et al. In other cases, affective reactions might not be a feature of the preconditions that elicit goalrelevant productions. Consequently, there are many instances in which affect is likely to have little influence on the strategies that people use in responding to information.
In this regard, Murray, Sujan, Hirt, and Sujan (1990) found that although
happy participants used fewer categories to classify stimuli than control participants when they were told to focus on similarities, they used relatively
more categories to classify the stimuli when they were told to focus on differences. These latter findings qualify the conclusion that happy persons typically use broader categories to classify stimuli than unhappy persons. Further experiments by Murray et al. (1990) indicated that differences in
categorization observed in both task conditions were likely to result from a
more general tendency for positive affect to increase the number of attribute
dimensions that participants brought to bear on the particular task they
were asked to perform. Thus, it increased their ability to make distinctions
among the stimuli they were considering as well as to identify similarities. As
Murray et al. suggest, these results could reflect mood-induced differences in
the motivation to perform the task at hand rather than differences in the cognitive productions that participants spontaneously apply under the conditions being investigated. We consider this possibility in more detail shortly.
Implications for Judgment
If positive affect increases the disposition to rely on categorical criteria for
comprehending and responding to stimuli, this disposition should be reflected in judgments to which the criteria are relevant. However, this influence could be localized at two stages of processing. First, if the information
that is available for judging a stimulus consists of both categorical descriptions and more detailed features, positive affect might increase the relative
weight that is attached to the categorical information at the time the implications of the two types of information are combined to form a judgment. An al-
342
CHAPTER 11
ternative possibility, however, is that positive affect influences the attention

that is paid to categorical information at the initial, comprehension stage.
That is, if individuals who experience positive affect think more extensively
about the implications of categorical information at the time the information
is received, they may interpret the implications of this information as more
extreme, and this could also lead the information to have more effect on
judgments that individuals report later. (For more general evidence that
thought increases the extremity of judgments, see Tesser, 1978.)
These two alternative possibilities are difficult to separate. That is, both
effects imply that positive affect will increase the apparent influence of categorical information on judgments. However, a series of studies by Adaval
(2003) suggested that, in fact, the effect is localized at the initial comprehension stage. In two studies, happy and unhappy participants evaluated a
product on the basis of an unfavorable or unfavorable brand name (a categorical criterion) and a favorable or unfavorable specific attribute description. Judgments were also obtained of products described by each attribute
feature separately. Brand information had more influence when participants were feeling happy than when they were not. To determine the reason for this difference, Adaval assumed a weighted averaging model of information integration of the sort proposed by Anderson (1971, 1981; see
Footnote 23), and applied a parameter-estimation procedure to identify
both the absolute weight attached to each piece of product information and
its scale value (i.e., its evaluative implications). Analyses of these parameters, computed for each participant separately, showed that positive affect
had no impact on the weight attached to brand information. However, it increased the extremity of the scale value assigned to it (That is, happy participants interpreted favorable brands as more favorable, and unfavorable
brands as more unfavorable, than unhappy participants did.) Further studies provided confirming evidence of this conclusion using different criteria.
These results are consistent with the results obtained by Bless et al.
(1996), that people attend more to categorical bases for interpreting the
stimuli they encounter. Moreover, the effects of this attention occur at the
time the information is first received and interpreted and not at the time
the information is used to make a judgment.
MOTIVATIONAL INFLUENCES OF AFFECT
ON INFORMATION PROCESSING
One body of phenomena to which Postulate 11.2 is applicable has been
identified in research on communication and persuasion. Much of this research suggests that people who feel happy are less influenced by the quality of the arguments contained in a persuasive message than unhappy people are (cf. Bless, Bohner, Schwarz, & Strack, 1990; Mackie & Worth, 1989).
THE ROLE OF AFFECT
343
Several factors could have been hypothesized to account for this difference. Mackie and Worth (1989), for example, proposed that positive affect is
likely to activate a large number of concepts and knowledge representations
in memory, taxing participants capacity to provide new information and
making them less able to attend to details of the information available. This
interpretation, however, is called into question by Niedenthal and Setterlunds (1994) finding that affective reactions activate only those concepts that
the specific type of reactions being experienced exemplify, and do not increase the accessibility of valenced concepts and knowledge in general.
A somewhat more viable conceptualization is consistent with the use of
affect as information. Specifically, Schwarz (1990; Schwarz, Bless, & Bohner,
1991) argued that because positive affect is typically associated with pleasant states of affairs, people who experience these feelings in a new situation
may appraise the situation as benign and, therefore, may not pay careful attention to details. In contrast, negative affect is associated with aversive
states of affairs and so the experience of it stimulates an interpretation of a
new situation as potentially problematic. Although these associations could
give rise to cognitive productions that are applied automatically, as noted
in a previous section, they could potentially stimulate deliberative cognitive activity as well. Consequently, people may be less motivated to pay attention to the details of a message they receive when they feel happy than
when they feel unhappy, and so the quality of the arguments contained in
the message has less impact on their judgments.
Although this interpretation seems plausible, it is contradicted by evidence that people engage in as much cognitive activity in responding to a
persuasive message when they are in a good mood as when they are in a
bad mood. In Bless et al.s (1990) study, for example, thought listing data indicated that the overall amount of this activity did not depend on the mood
that participants were in. That is, participants in a good mood tended to
counterargue both weak and strong arguments (and thus were equally influenced by both),whereas participants in a bad mood elaborated strong arguments and counterargued weak ones, thus being differently persuaded
by them. The total number of thoughts generated in the two mood conditions was the same in both cases, however. Research on impression formation by Isbell (2000) also suggested that happy and sad individuals do not
differ in the amount of cognitive activity they perform in response to information, but differ only in the nature of this activity.
A third conceptualization is based on the notion that people who are feeling happy are motivated to maintain their mood and, consequently, avoid activities that are likely to bring them down. Unhappy individuals, on the other
hand, are motivated to divest themselves of the negative feelings they are experiencing, and engage in activities that distract themselves from thinking
about these feelings and the events that gave rise to them. Because persua-
344
CHAPTER 11
sive messages are typically counterattitudinal, thinking about them is particularly likely to destroy ones good mood. Consequently, people who feel
happy tend to avoid this cognitive activity and, as a result, are insensitive to
differences in the quality of arguments contained in the messages.
If this interpretation is correct, processing differences should be eliminated or even reversed by leading happy persons to believe that thinking
about the information they receive will be enjoyable and, therefore, will
maintain or enhance their positive mood state. Wegener, Petty, and Smith
(1995) showed this to be the case. Participants listened to one of two communications. One message advocated a position with which participants
were inclined to disagree and was introduced with instructions that it was
likely to make them unhappy. The quality of arguments contained in this
message had less impact on participants attitudes when they were happy
than when they were not. The other message advocated a position that participants typically favored and was introduced with a comment that reading
it was likely to be enjoyable. In this case, the quality of the arguments had
greater effect on participants attitudes when they were happy.
Wegener et al.s (1995) results therefore qualified the conclusion that people who experience positive affect are generally unmotivated to engage in extensive cognitive processing. There are also qualifications on the assumption
that people who experience negative affect are motivated to process new information extensively. Cognitive activity may only help to eliminate negative
affect if the affect is mild and nothing can be done about the events that produced it. Extremely unhappy individuals are likely to ruminate about the
events that created their unhappiness instead of distracting themselves by
engaging in irrelevant activity (Ellis & Ashbrook, 1988; Klinger, 1975; Martin &
Tesser, 1996). To this extent, these individuals may be just as insensitive to
details of affect-irrelevant information as happy people are.
The information processing implied by this interpretation is deliberative.
To this extent, it would not contradict the possibility that affective reactions can stimulate cognitive procedures of the sort noted in the previous
section, which are performed automatically in situations in which they apply. However, the deliberative processes that are suggested by moodmaintenance and mood-repair motives have an alternative interpretation.
That is, these processes could be the byproduct of a more general tendency to use affect as information. We now turn to this possibility.
A PERFORMANCE-FEEDBACK MODEL
OF AFFECT AS INFORMATION
In most research that has been performed to evaluate the use of affect as information, people are assumed to make judgments of a particular person,
object, or event, and the affect they experience at the time of judgment is
THE ROLE OF AFFECT
345
used as information about their reaction to the stimulus they are judging.
However, peoples affective reactions can provide information about other
things as well. In situations outside the laboratory, people may use their affect as a basis for behavioral decisions. For example, they may decide to engage in a behavior that gives them pleasure, while avoiding activities they
consider to be aversive. In achievement situations, on the other hand, peoples decisions to continue on a task can also be influenced by their feelings
about the quality of their performance. A woman who is writing a term paper, for example, might have little objective basis for deciding whether her
output is sufficient to attain the grade she is seeking. Therefore, her decision to continue working on the paper may be based on her subjective feelings about the work she has done. The extraneous affect she happens to be
experiencing at this time could influence her perceptions of these feelings
and, as a consequence, her decision to stop working or continue.
These possibilities were formalized by Wyer et al. (1999) in a performance-feedback model of affect. This conceptualization, which was stimulated in large part by Martin, Achee, Ward, and Wyers (1993) previous
work, assumes that people spontaneously monitor their behavior and the
outputs it generates to determine if they have satisfactorily attained the
goal they are pursuing. If they conclude that this is the case, they terminate
their goal-directed activity. Otherwise, they continue processing, using the
same or a different strategy, until they generate an output they consider to
be satisfactory.
When objective criteria exist for deciding whether a goal has been attained, the decision to engage in goal-directed activity is obviously based
on these criteria. More often than not, however, the criteria for goal attainment are ambiguous. When this is the case, people may base their decision
on the affect they happen to be experiencing and attribute to their feelings
about their attainment of the goal they are pursuing.
The most interesting implications of this analysis derive from the fact
that peoples perceptions of their affective reactions depend on the particular goal they happen to be pursuing. In the preceding example, the womans
goal was to write a paper of high quality. In this case, she may ask herself if
she feels the paper meets this standard and is more likely to stop work if
her feelings are positive than if they are negative. On the other hand, suppose the woman focuses her attention on whether she is having fun rather
than the quality of the paper she is writing. Then, her feelings are likely to
have precisely the opposite effect. That is, she may be more inclined to
stop writing if her feelings are negative than if they are positive.
Two studies by Martin et al. (1993) provided direct support for these possibilities. In one, happy or sad participants were asked to form an impression of someone on the basis of a series of behaviors. Each behavior was
printed on a separate card of a deck that participants were given. Partici-
346
CHAPTER 11
pants under performance-criterion conditions were told to continue turning

the cards and reading the behaviors until they had enough information to
form an impression. In contrast, participants under enjoyment-criterion conditions were told to continue as long as they were enjoying what they were
doing. The first group of participants should implicitly ask themselves as
they read the behaviors if they feel they have enough information. Therefore, they should be more likely to answer this question affirmatively, and
to stop reading sooner, if they are happy than if they are sad. The second
group of participants, however, should ask themselves if they are enjoying
the task, again responding more affirmatively if they are happy than if they
are sad. In this case, therefore, happy participants should continue reading
behaviors longer than sad participants. Results confirmed these predictions. That is, happy participants read fewer behaviors than sad participants in performance-criterion conditions (24 vs. 43, respectively), but
more behaviors than sad participants in enjoyment-criterion conditions (47
vs. 36, respectively).
A second study replicated these findings using a different task. In this
case, participants were asked to generate a list of birds. Some participants
were told to continue until they thought it was a good time to stop, whereas
others were told to continue as long as they were enjoying themselves.
Happy participants persevered less long on the task than unhappy participants when they were told to use a performance criterion (140 s vs. 218 s,
respectively), but longer than unhappy participants when they were encouraged to employ an enjoyment criterion (240 s vs. 145 s, respectively).
Individual and Situational Differences
in Goal Orientation
In the conditions just described, the criterion for stopping or continuing
was explicitly stated by the experimenter. When the criterion is unspecified, both individual and situational differences could influence the criteria
that participants use. Persons with high need for cognition (Cacioppo &
Petty, 1982), for example, typically find intellectual activity to be intrinsically enjoyable. They may therefore be likely to adopt an enjoyment criterion, whereas persons with low need for cognition may be guided to a
greater extent by performance criteria. If this is so, the experience of positive affect might increase the persistence of high need-for-cognition individuals on response-generation tasks, but might decrease the persistence of
low need-for-cognition individuals on such tasks. In fact, additional data obtained by Martin et al. (1993) confirmed this prediction.
The relative emphasis on performance and enjoyment is also likely to
vary with the nature of the task and the context in which it is performed. In
laboratory studies, participants are often given a particular task objective
THE ROLE OF AFFECT
347
by the experimenter. Under these conditions, participants may spontaneously employ a performance criterion. That is, they may base their decision
to persevere on whether they feel that their responses fulfill these externally imposed task requirements. This hypothesis was supported in a third
condition of Martin et al.s second experiment in which participants were
simply told to stop generating birds when they wanted to. The results in
this condition were very similar to those obtained when participants were
given an explicit performance criterion; that is, happy participants stopped
sooner than unhappy ones. These results confirm the speculation that participants in many experimental situations spontaneously focus their attention on whether they have fulfilled the implicit task objectives they are
given. Nevertheless, when an experimental task is particularly interesting,
or when achievement goals are de-emphasized, an enjoyment criterion
might be applied as well.
Some evidence in support of the latter possibility was reported by
Murray et al. (1990) in a study mentioned earlier in this chapter. Some participants were asked to list similarities between two popular television programs, whereas others were asked to list differences between the programs. The task was intrinsically interesting to college-age participants and,
therefore, was likely to stimulate them to use an enjoyment criterion in deciding how much effort they should expend. To this extent, participants
should use the affect they were experiencing as a basis for inferring their
enjoyment. Consistent with this assumption, happy participants reported
enjoying the task more than control participants did. Correspondingly,
happy participants listed more similarities between the television programs when they were asked to focus on similarities, and listed more differences between the programs when they were focused on differences, than
control participants did. These results would be hard to explain on the basis of either the assumption that positive affect generally increases the accessibility of global concepts in memory. Nor is it consistent with the assumption that positive affect reduces cognitive resources. It seems more
reasonable to suppose that happy participants inferred from the affect they
were experiencing that they were enjoying the task, and therefore devoted
more effort to an identification of both similarities and differences in the
stimuli they were judging than other participants did.
Murray et al.s finding that happy participants reported greater enjoyment of the task than other participants raises an additional consideration.
That is, when several goals are available in a situation, the affect that persons happen to be experiencing could influence the goal they choose to
pursue as well as their inference that they have or have not attained it (for
a discussion of this possibility, see Schwarz & Clore, 1996). For example,
suppose happy individuals are generally more inclined to adopt an enjoyment objective than unhappy individuals are (Isen, 1987). If this is so, happy
348
CHAPTER 11
persons may persist on a task because they infer they are enjoying it,
whereas unhappy persons, who adopt a performance criterion, may persist
because they feel they have not yet attained an adequate level of achievement. To this extent, happy and unhappy persons might persist to an equal
extent, albeit for different reasons.
Despite this ambiguity, the influence of affect on behavior in a large number of situations both in and outside the laboratory can be conceptualized
in terms of its mediating influence on the answers to questions that persons
implicitly ask themselves about how they feel about their behavior and its
outcomes. The broader implications of this conceptualization can be summarized in two postulates:
Postulate 11.3. When people engage in behavior for the purpose of attaining
an external goal and objective criteria for inferring the attainment of this goal
do not exist, they use the affect they are experiencing to infer whether they feel
the strategy they have used in pursuit of this goal, and the outputs they have
generated by employing it, are appropriate. Therefore, they are more likely to
infer they have met the criterion for satisfactory performance if they are experiencing positive affect than if they are not.
Postulate 11.4. When peoples primary reason for engaging in behavior is enjoyment, they use the affect they are experiencing as a basis for inferring this
enjoyment. Therefore, they will continue the activity longer if they are experiencing positive affect than if they are not.
An application of these postulates calls attention to several contingencies in the impact of affect that have not been investigated in the research
performed to date. Several areas of research are worth reviewing from this
perspective.
Creativity
A creativity task often employed in the laboratory requires participants to
generate instances of a concept. This task is similar to that constructed by
Martin et al. (1993, Experiment 2) except that the concept involved has unclear boundaries, thus allowing participants to generate novel responses.
Thus, for example, participants might be asked to generate uses of a brick
or things they could use to sit on. In this situation, like that constructed by
Martin et al., participants might assume they should stop when they have
generated enough good responses to fulfill the demands of the task. However, the task itself might often be intrinsically enjoyable, as it requires a
degree of imaginativeness. Thus, many participants might spontaneously
apply an enjoyment criterion as well. Consistent with this speculation, Isen
and Daubman (1984) found that happy participants generated more re-
THE ROLE OF AFFECT
349
sponses when performing such a task than control participants did. This
suggests that an enjoyment criterion predominated in their study.
A somewhat different interpretation of Isen and Daubmans results is
suggested by the fact that happy participants also tend to generate more
novel responses than other participants do. When the criterion for an appropriate response is unclear, participants may inhibit reporting responses
they feel are poor exemplars of the concept they are considering. Specifically, they may subjectively generate a response (e.g., a potential use for
a brick), ask themselves if they feel their response is a good one, and only
report it if they answer this question affirmatively. Thus, they are more
likely to record their response if they are happy than if they are not. Evidence that happy persons perform better then others on the Remote Associates Test (Isen, Daubman, & Nowicki, 1987) is consistent with this interpretation. Furthermore, the interpretation could explain the different
results obtained by Isen and Daubmans (1984) study and those reported by
Martin et al. (1993). That is, the criterion for membership in the category
that participants considered in Martin et al.s experiment (birds) was
clearly defined, and so the use of affect to evaluate the appropriateness of
individual exemplars was less likely.
Impression Formation
In impression-formation research (for a review, see Wyer & Carlston, 1994),
participants typically receive information about several behaviors a person
has performed with instructions to form an impression of the person. A large
amount of information is typically presented (e.g., 2530 items; see Wyer &
Martin, 1986), and so it is difficult to form an impression that takes all of it
into account. Consequently, participants in these experiments may often ask
themselves at various points if the information they have received is a sufficient basis for a concept of what the individual is like. If they answer this selfgenerated question affirmatively, they may base their concept of the person
on this information without taking into account the implications of the information they receive later (see Postulate 3.1). If this is so, and if happy participants are more likely than unhappy participants to respond affirmatively to
such a question, they should use relatively less information to form their impressions. Moreover, the information they use should typically occur at the
beginning of the sequence rather than at the end.
Martin et al.s (1993, Experiment 1) results confirmed the assumption
that happy participants with an impression-formation objective use less information to form an impression than sad participants do. The prediction
that happy persons are more likely than unhappy ones to base their impression on the first information they receive was confirmed in research by
Isbell (2000; Isbell, Clore, and Wyer (1998). Participants who were feeling ei-
350
CHAPTER 11
ther happy or sad were asked to form an impression of someone who was
described at the outset as either extraverted or introverted. This general
description was followed by a series of behaviors, an equal number of
which exemplified each of these traits. After receiving the information, participants judged the targets extraversion and then recalled the behaviors
they had received.
Isbell expected that the general description of the target that participants received at the outset would activate a concept of either an extravert
or an introvert, and that participants would implicitly ask themselves if
they felt this concept was an appropriate basis for their impression. Happy
participants, who should answer this question affirmatively, should be
more inclined than unhappy participants to apply the concept to the target
and to base the judgments they report later on this concept. This was in
fact the case, as indicated earlier. That is, the initial target description had a
substantial effect on happy participants judgments of the targets extraversion but had very little impact on unhappy participants judgments.
The fact that happy participants based their judgments on the initial description of the target does not mean that they failed to think about the behavioral information they received later. To the contrary, persons who
apply this concept to the target should think about the behavioral information they receive with reference to it (Srull & Wyer, 1989). Moreover, they
should think more extensively about behaviors that are inconsistent with
the concept in order to reconcile their occurrence (Hastie, 1980; Srull &
Wyer, 1989; Wyer & Srull, 1989). As implied by this reasoning, these participants recalled more inconsistent behaviors than consistent ones. Unhappy participants, who did not apply the concept activated by the initial
trait description to the target, did not engage in this inconsistency resolution. Consequently, they recalled behaviors equally well regardless of their
consistency with the initial trait description.
In many cases, peoples initial impressions of someone are based largely
on their feelings toward this person. This is particularly likely if their first
encounter with the person elicits a spontaneous appraisal on the basis of
his or her physical appearance (cf. Yeung & Wyer, in press). In such cases,
the affect they experience at the time they judge the person could potentially have a dual function. First, it may convey judgment-relevant information about the individual being judged (Schwarz & Clore, 1983). Second, it
could be used as an indication of whether this affect is a sufficient basis for
making the judgment. Thus, people who experience positive affect at the
time they are asked to judge someone might not only infer they feel favorably toward the target, but also that these feelings are a sufficient basis for
the judgment they are asked to make. In contrast, people who experience
negative affect may infer they feel negatively toward the target but also
may conclude that these feelings are not a sufficient basis for the judgment.
THE ROLE OF AFFECT
351
This implies that the effects of positive affect on impression judgments may
often be greater than the effects of negative effect.
There is a contingency in these predictions. That is, the contention that
happy people are inclined to decide that the initial information they receive
is a sufficient basis for judgment assumes that the implications of this initial
information, considered in isolation, are unambiguous. If these implications
are mixed or unclear, they may not be considered sufficient bases for judgments by either happy or unhappy individuals. A recent study by Isbell
(2003) bears on this contingency. She found that happy participants were
more likely than unhappy participants to base their evaluations of the target on a general trait characterization of him even when this characterization was presented after the targets behaviors were described. In this
study, however, the behavior descriptions were both favorable and unfavorable, and so a clear initial impression of the target was difficult to form
on the basis of these descriptions alone. Therefore, participants were unlikely to consider the behaviors to be a sufficient basis for judgment regardless of their affective state. The greater use of categorical information by
happy participants under these conditions could reflect their generally
greater tendency to apply broad categories in interpreting information, as
suggested by Adaval (2003) and Bless et al. (1996) and noted earlier.
Stereotyping
Our interpretation of Isbell et al.s (1998) findings also has implications for
the effects of affect on the use of stereotypes. In a study by Bodenhausen
(1993; see also Bodenhausen et al., 1994), for example, happy and sad participants were asked to judge the culpability of a defendant on the basis of information about the defendants background and the conditions surrounding the crime. The description of the crime was preceded by the targets
name and hometown, which either identified him as Hispanic or provided
no clue as to his ethnicity. It seems reasonable to suppose that the defendants ethnicity spontaneously activated a stereotype, and that participants
implicitly asked themselves if they felt this was an appropriate basis for
their judgment, answering affirmatively if they were happy and negatively if
they were sad. Consistent with this reasoning, the stereotype had more impact on happy participants judgments than on unhappy participants. Note
that if the stereotype had not been activated until after the detailed crimerelated information was presented, participants would not be in a position
to ask themselves at the outset whether the stereotype was a sufficient basis for judgment. If anything, they should ask themselves if they felt the
crime-related information was sufficient. To this latter extent, happy participants in this condition might be more influenced by the crime-related infor-
352
CHAPTER 11
mation, and less influenced by the stereotype, than sad participants are. To
my knowledge, this possibility has not been examined.
Self-Evaluation Processes
The determinants and effects of self-judgments have been a major focus of
social-psychological research for many years (cf. Greenwald & Pratkanis,
1984; Klein & Loftus, 1993; McGuire & McGuire, 1988). Because of the general
interest in these matters, the implications of the performance-feedback
model for self-judgment processes are worth considering in some detail.
The influence of affect on four different types of self-judgmentsjudgments
of global life satisfaction, self-worth, specific trait judgments, and estimates
of personal riskare therefore considered from the perspective of this
model.
Several alternative cognitive processes could underlie these judgments,
each of which can involve the use of a different type of self-knowledge. The
effects of these procedures can be conceptualized in terms of the general
processing model outlined at the beginning of this volume. Specifically,
when a self-judgment is required, the goal specification may resonate with a
number of different cognitive procedures for generating this judgment.
These processes could include (a) the search for a previously formed judgment-relevant trait description, (b) the identification of a judgment-related
past experience and a construal of its implications for the judgment to be
made, and (c) an assessment of the implications of the affective reactions
one is experiencing. Situational or individual difference factors that make
one process easier to use than another will increase the likelihood that this
process is applied. An additional consideration, however, is implied by the
performance-feedback conceptualization. That is, participants who have
generated a response on the basis of the winning procedure may spontaneously ask themselves if they feel that this response is a sufficient basis
for judgment. participants who are experiencing positive affect may answer
this question affirmatively, whereas those who are experiencing negative
affect may not. Therefore, happy persons are more likely than unhappy persons to accept and apply the first criterion they are able to generate. The
following discussion is based on this assumption.
1. Judgments of Life Satisfaction
A request to judge ones life satisfaction could activate three different
processes. First, respondents might search for a previously formed concept
of their life as a whole that is directly applicable to the judgment. Second,
they might recall one or more previous life events that are relevant to the
judgment and base their response on the affect elicited by these memories.
Finally, they might assess the affect they happen to be experiencing at the
THE ROLE OF AFFECT
353
FIG. 11.2. Processing stages involved in making (A) global judgments of life
satisfaction and self-esteem, and (b) specific trait judgments. Based on Wyer et
al. (1999).
time and use this as an indication of their feelings about life in general without conducting a search of memory at all. These processes, which could occur simultaneously, are summarized in Fig. 11.2a.
On a priori grounds, it is unclear which of these processes will predominate. For example, if participants in an experiment have a previously
354
CHAPTER 11
formed concept of their life as a whole, a request to estimate their life satisfaction might spontaneously activate this concept and so a judgment can
be quickly computed on the basis of the affect it elicits. On the other hand,
people are infrequently called upon to make this type of judgment in daily
life. Consequently, they usually do not have a previously formed concept of
their life as a whole stored in memory, and must compute a judgment at the
time they are asked to report it. In this case, it would normally take less
time to compute a judgment on the basis of the transitory affect one is experiencing at the time of judgment than to do so on the basis of other, less
easily accessible criteria, and so judgments are more likely to be based on
the former computation. Evidence that people use transitory affect as a basis for life satisfaction judgments independently of any specific self-knowledge they have available (Schwarz & Clore, 1983; Strack, Schwarz, &
Gschneidinger, 1985) is consistent with this hypothesis.
However, the performance-feedback model we have proposed raises a
further consideration. People who experience positive affect for reasons
that are objectively unrelated to their life situation might not only use these
feelings as a basis for judgment but also consider this criterion to be sufficient. In contrast, people who experience negative affect might be disposed
to question the sufficiency of this criterion. Therefore, they might either
seek other judgment-relevant information or, if no alternative criteria are
readily available, might adjust for the bias that results from their use of feelings as information.
Schwarz and Clores (1983, Experiment 1) findings support this conjecture. To reiterate, happy persons appeared to use their feelings as a basis
for judging their life satisfaction regardless of the situational factors that
had implications for their validity as a basis for judgment. In contrast, unhappy participants adjusted their judgments to compensate for the effects
of these factors. Perhaps both happy and unhappy participants spontaneously considered the affect they were experiencing as a potential basis for
judgment. However, whereas happy participants concluded that these feelings were a sufficient basis for judging without thinking about situational
factors that might have influenced them, unhappy participants considered
that an assessment of their feelings per se was insufficient and took situational factors into account as well.
2. Judgments of Self-Worth
Estimates of self-worth, like estimates of life satisfaction, may typically be based on feelings. These feelings are likely to be estimated by
either (a) retrieving a previously formed concept of oneself with which affective reactions have become associated through learning, (b) recalling
descriptive self-knowledge to which affective reactions have become condi-
THE ROLE OF AFFECT
355
tioned and combining the implications of these reactions to arrive at an

overall estimate of ones feelings; and (c) assessing the implications of the
transitory affect one happens to be experiencing for reasons that may or
may not actually be relevant. If a previously formed concept of self exists
and is activated by the request to evaluate ones self-worth, the affective reactions elicited by this concept might be used as a basis for judgment. If
such a concept has not previously been formed, however, and if the affective reactions based on descriptive self-knowledge take time to compute,
the transitory affect one happens to be experiencing for other reasons
should normally be activated more quickly and, therefore, is more likely to
be used.
The aforementioned study by Levine et al. (1994) bears on this possibility. In this study, the descriptive implications of participants recalled experiences only affected their judgments of competence in the domain to
which these experiences were relevant. However, participants estimates of
their global self-esteem were influenced by the affect elicited by the life
events they had recalled regardless of its descriptive implications. In terms
of our conceptualization, this suggests that participants in Levine et al.s
study did not have a previously formed concept of themselves easily accessible in memory. Therefore, they based their judgments on the affect they
were experiencing at the time, regardless of whether or not the event that
elicited the affect was self relevant.
3. Trait Judgments
Judgments of the specific traits one possesses are more likely to be
based on descriptive rather than affective criteria. The processes that underlie these judgments are summarized in Fig. 11.3b. Specifically, people
who are asked if they have a specific trait may recall and report a previously formed concept of themselves with respect to this trait. Alternatively,
they may infer the trait from the descriptive implications of trait-relevant
self-knowledge that they have acquired (e.g., behaviors that exemplify the
trait, descriptions of oneself by others, etc.)
For affective reactions to have an informational influence on trait judgments, mediating computations must be performed in order to transform
these reactions into descriptive terms. For example, people who experience
negative affect might judge themselves to be incompetent based on the
implicational molecule
[I am incompetent; I feel badly about myself].
The retrieval and use of this molecule, however, is likely to take longer than
computations based on descriptive criteria. This is particularly true if a
trait-relevant experience is easily accessible in memory at the time of judg-
356
CHAPTER 11
ment. Levine et al.s (1994) findings, described earlier, are consistent with
this hypothesis. That is, the descriptive implications of participants recalled life experiences were used as a basis for judgments of attributes to
which they were relevant. The affective implications of recalled experiences only influenced judgments of attributes in a domain to which the experiences were unrelated.
The implications of the performance-feedback model become of interest
in this context. If trait judgments are computed more quickly on the basis of
descriptive criteria than on the basis of affective criteria, people who experience positive affect should be likely to infer that these criteria are sufficient, whereas people who experience negative effect are likely to consider
additional (e.g., affective) criteria. Thus, in contrast to judgments of life satisfaction and self-worth, trait inferences are more likely to be influenced by
negative affect than by positive affect.
Communication and Persuasion
Suppose people read a persuasive communication for the purpose of deciding if the position it advocates is valid. They are likely to have an initial impression of the positions validity position before they evaluate the communications contents. This impression could be based on the source of the
message, on surface features of the communication itself (e.g., writing
style), or on a previously formed attitude toward the position in question. If
people who have made formed this impression ask themselves if it is an appropriate basis for accepting the position advocated, they should be more
inclined to answer this question affirmatively if they are happy than if they
are sad. This has two implications. First, happy persons are more likely
than unhappy persons to rely on the source of a message as a basis for accepting the position advocated. Second, happy persons are less inclined
than sad persons to make a careful evaluation of the arguments contained
in a message. Consequently, their acceptance of the position advocated is
less likely to be influenced by the quality of these arguments. Support for
the first hypothesis has been obtained by Roselli (1995), and the second hypothesis is consistent with results of several studies described previously
(Bless et al., 1990; Mackie &Worth, 1989).
A performance-feedback conceptualization of these results suggests two
contingencies, however. First, the different cognitive activities performed
by happy and unhappy persons should only occur if these persons perceive their feelings to have implications for the validity of their initial impression. If participants attribute their feelings to other, impression-irrelevant factors, the difference in processing should theoretically not occur.
Results reported by Sinclair, Mark, and Clore (1994) confirmed this prediction. They found that the effect of argument strength on participants accep-
THE ROLE OF AFFECT
357
tance of a persuasive message was less on sunny days (when participants

were happy) than on rainy days (when they were unhappy). However, calling participants attention to the weather (thereby making salient the true
source of their feelings) eliminated this difference.
Second, our analysis assumes that recipients objective at the time they
read the message is to evaluate the validity of the position being advocated. In contrast, suppose persons have an enjoyment objective. Then,
precisely the opposite influence of affect might occur. The aforementioned
study by Wegener et al. (1995), suggesting that happy persons are more influenced by the quality of arguments in a message than other participants
when they are led to believe that the message will be enjoyable, confirms
this possibility.
Accuracy
The performance-feedback conceptualization can apply to judgments of
physical stimuli as well as social ones. People who are called upon to judge
a stimulus may spontaneously make a holistic appraisal of the stimulus at
the time they are first exposed to it before they scrutinize its individual features. They may then ask themselves if this appraisal provides a sufficient
basis for judgments. To this extent, happy individuals should be more likely
than unhappy ones to base their judgments on this initial appraisal alone.
One implication of this is that when accurate judgments are likely to require a careful analysis of the individual stimulus features happy participants are less likely to be accurate than unhappy ones. Evidence reported
by Sinclair and Mark (1992), showing that happy participants were less accurate than sad participants in estimating the correlation between two variables on the basis of scatter plots, confirms this speculation.
Behavior in Nonlaboratory Situations
Many situations arise outside the laboratory in which persons behavioral
decisions are likely to be mediated by their feelings about the behavior and
its implications. Students who prepare for an exam must decide if they have
studied long enough. A person who wants to ask someone for a date must
decide if it is the right time to do so. Or, a person who wishes to buy a car
must decide if he or she has enough information about it to make a decision. In such instances, people should stop studying earlier, ask persons for
a date sooner, and make a purchasing decision more quickly, if they feel
happy than if they do not.
For similar reasons, happy persons may be more inclined to behave impulsively, or in ways that upon reflection they might consider undesirable.
A series of studies by Forgas (1998) are interesting to consider in this con-
358
CHAPTER 11
text. In three studies, participants were asked to indicate how they would
phrase a verbal request in a hypothetical social interaction. It seems reasonable to suppose that people who want another person to do something
(e.g., to repay a loan) will spontaneously generate a verbal phrasing of the
request that corresponds directly to the desired action (e.g., Give me the
money you owe me) and may implicitly ask themselves if it is appropriate
to express the request in these terms. If they answer this question positively, they are likely to alter the phrasing in a way that corresponds to normative standards of politeness. Otherwise, they might not. If this is so, the
performance-feedback conceptualization we propose suggests that happy
persons are more likely to express their requests impolitely than unhappy
persons are. This, in fact, is what Forgas (1998) found. Although his studies
were performed in the laboratory, they seem likely to generalize to other
situations as well.
It would nevertheless be hazardous to conclude that happy persons are
generally less likely to engage in extensive information processing than unhappy persons are. This difference should theoretically exist only when individuals have an implicit performance objective. In nonlaboratory situations, information processing is often performed for enjoyment. In these
circumstances, the performance-feedback implies that happy persons will
persevere longer, and will process information more analytically, than sad
persons will. Thus, for example, happy individuals might work harder than
sad individuals on crossword puzzles, and may read news articles in relatively greater depth. Moreover, they might make less hasty decisions about
other persons on the basis of their initial impressions of these persons, in
contrast to results observed in the laboratory (Isbell et al., 1998). The possibilities, which are not implied by other conceptualizations of the influence
of affect on performance, should be kept in mind when drawing conclusions
on the basis of experimental research in which performance objectives are
implicitly emphasized.
AFFECT, ATTITUDES, AND BEHAVIOR

As Thurstone (1959) assumed, attitudes are often expressions of the affective reactions we experience toward a person, object, event, or social policy. To this extent, the situational and informational factors that influence
the use of affect as information have direct implications for the determinants of attitude formation and change. However, much of the theoretical
and empirical interest in this topic derives from the assumption that attitudes provide the basis for behavioral decisions. In fact, there are many
contingencies on the validity of this assumption (cf. Fazio, 1986; Fishbein &
Ajzen, 1975; Miniard & Cohen, 1981). In light of the centrality of the attitude
359
THE ROLE OF AFFECT
construct in social psychological research, however, the influence of affect

in attitude formation and change deserves special attention.
In fact, the role of affect in attitude formation and change has already
been discussed in several earlier sections of this chapter, and does not require reiteration. Here, we attempt to outline more formally the processes
that underlie the construction of an attitude and the situational and informational factors that influence it. Empirical evidence bearing on this analysis is then described.
The ultimate objective of most persuasive communications is to influence
recipients to behave in a way that the communicator considers to be desirable (to purchase a particular product, to endorse a social policy, or to stop
smoking, etc.). However, it is useful to distinguish between two types of
communications that could be constructed for this purpose. One type attempts to alter a persons beliefs, or estimates of the likelihood that an
event will occur, or that a particular proposition is true (e.g., that Americans bombing of Afghanistan has resulted in the deaths of thousands of innocent victims, or that air and water pollution will extinguish life on earth
by the end of the 21st century). As suggested in chapter 5 (Wyer &
Hartwick, 1984), communications that are intended to focus on beliefs in a
proposition are likely to emphasize antecedents, that is, propositions about
events that, if they are true, have implications for the truth of the target
proposition. Affective reactions are likely to influence the impact of such a
communication only indirectly, through the extent to which they influence
the ability or motivation to evaluate the arguments conveyed in the communication or their implications for the validity of the target proposition
(for evidence that affect can have this influence, see Fiedler, 1988.)
The other type of message, which is of primary concern in the present
context, is intended to influence an attitude, or evaluation of a person, object, or event (Americas bombing of Afghanistan, recycling, etc.). These
messages are likely to focus on consequences of the event or state of affairs
toward which the attitude is directed. The cognitions that typically underlie
the construction of an attitude are captured by the model proposed by
Fishbein (1963) or, more recently, Fishbein and Ajzen (1975). Ones attitude
toward a behavior, for example (Ab), is captured by the equation
Ab = Sbi ei ,
[11.1]
Where bi is the belief that the behavior will have a consequence i, and ei is
the evaluation of this consequence. (Not all possible consequences of the
behavior are taken into account, of course, but only a subset that is most
360
CHAPTER 11
salient at the time the attitude is formed.) To this extent, a communication should be effective in changing an attitude if it either increases beliefs
that the behavior has consequences that the recipient considers to be desirable or, alternatively, increases the evaluation of consequences that the
recipient already believes are likely to occur.
Affective reactions are likely to play a quite different role in the processing of attitude-focused communications than in the processing of belieffocused messages. Moreover, this role may depend in part on whether the
attitude in question is based on hedonic or utilitarian criteria (Zanna &
Rempel, 1988). When the criterion is hedonic, the role of affect could depend on whether (a) recipients think sufficiently about the consequences
described in a message for the affect associated with these consequences
to be elicited, (b) whether they can effectively combine the implications of
these consequences into a single attitudinal response in the manner specified in Equation 11.1, and (c) whether they can distinguish this affect from
the feelings they are experiencing for other, unrelated reasons. Once an attitude is formed in this manner it may be reactivated at the time the course
of action is actually contemplated (see Fazio, 1986, for a discussion of the
factors that influence this activation).
As noted earlier, however, the specific consequences of the position advocated in a communication may not be the only determinants of the attitude that recipients report. Affect could also be stimulated by the intrinsic
pleasantness or unpleasantness of the behavior being advocated, independently of any specific information about it. (The thought of taking comprehensive examinations, for example, might spontaneously elicit negative
affect independently of any specific consequences the activity might have.)
Finally, extraneous affect that recipients happen to be experiencing may be
misattributed to their feelings about this activity. Affect from all of these
sources could contribute to peoples perceptions of their attitude toward
the behavior being advocated.
Experimental Evidence
The role of affect in responses to attitude-focused persuasive messages was
evaluated in a series of studies by Albarracin and Wyer (2001). We told participants we were concerned with how people process information in natural settings, and that to do this, we would like them both to transmit and receive information in a simulated coffee shop situation. On this pretense, we
induced them to feel happy or unhappy by asking them to write a letter to a
friend describing a positive or negative life experience. Then, we asked participants to read a communication, ostensibly taken from a local newspaper, which summarized either strong or weak arguments in support of instituting comprehensive examinations and urging recipients to vote in favor
THE ROLE OF AFFECT
361
of the exams in a forthcoming referendum. Participants read the message in

the presence of background noise similar to that encountered in a coffee
shop. In low-distraction conditions, the noise was of low volume and not particularly disturbing. In high-distraction conditions, however, it consisted of a
high volume get-acquainted conversation between two students of the sort
that might take place in a coffee shop situation. Participants after reading
the message reported their attitudes toward instituting the exams and their
intentions to vote for them. Then, they reported their beliefs that the various consequences described in the message would occur and the desirability of these consequences.
1. Effects of Distraction on the Processing
of Message Content
We expected that the influence of the persuasive message would result
from cognitive activities at four different stages of processing: the computation of beliefs that the consequences specified in the message would occur,
evaluations of the desirability of these consequences, an integration of
these beliefs and evaluations to form an attitude in the manner described
by Equation 11.2, and the transformation of this attitude into a behavioral
intention. We further expected that the effects of distraction would be more
evident at later stages of processing (e.g., the computation of an attitude or
behavior intention) than at earlier ones. This was in fact the case.
The effect of argument strength on judgments was inferred from the difference between judgments when arguments were strong and judgments
when arguments were weak. These differences are shown in the top half of
Table 11.5 for judgments that were presumably formed at each stage of
processing. Distraction had a significant effect on the impact of argument
strength on both attitudes and behavioral intentions, which were formed at
later stages of processing. However, it had a much smaller effect on the impact of argument strength on evaluations of individual consequences (ei ),
and had no effect at all on its impact of argument strength on beliefs that
the consequences would occur (bi ).
2. Impact of Distraction on the Influence
of Extraneous Affect
If participants evaluations of the consequences mentioned in the message are based on the affective reactions that were elicited by thinking
about them, Equation 11.1 should describe the results of integrating the implications of these reactions. However, distraction could not only prevent
participants from performing this integration but could also prevent them
from distinguishing between their affective reactions to the consequences
and the affect they are experiencing for other reasons. If this is, so, the ex-
362
CHAPTER 11
TABLE 11.5
Effects of Argument Strength and Extraneous Affect on Behavioral
Intentions, Attitudes, and Beliefs and Evaluations of Specific
Behavioral Outcomes as a Function of Distraction
Effect of argument Strength

High distraction
Low distraction
F(1, 73)
Effect of affect
High distraction
Low distraction
F(1, 73)
Outcome
Beliefs (bi )
Outcome
Evaluations (ei )
Attitudes
(Ab)
Behavior
Intentions
2.10a
2.08
<1
3.22
4.37
2.23
0.62
2.61
5.82*
1.77
3.91
3.69*
0.75b
0.92
4.87*
1.48
0.52
6.80*
1.79
0.73
9.39*
2.09
0.60
5.79*
Note. Adapted from Albarracin and Wyer (2001).

aThe effect of argument strength is inferred from the difference between judgments when arguments were strong and judgments when arguments were weak.
bThe effect of affect is inferred from the difference between judgments when participants
were feeling happy and judgments when they were not.
traneous affect that participants were experiencing should have greater effect. This was also the case, as shown in the bottom half of Table 11.5. When
distraction was high, extraneous affect had an positive influence on not
only participants attitudes and intentions but also their estimates of the
likelihood and desirability of individual consequences. When distraction
was low, however, extraneous affect had slight contrast effects on these
measures.
3. Reciprocal Effects of Attitudes
on Outcome Beliefs and Evaluations
When participants were not distracted, they presumably computed beliefs and evaluations associated with each consequence described in the
persuasive message and then integrated these estimates to form an attitude
in the manner implied by Equation 11.1. To this extent, their outcome beliefs and evaluations should have multiplicative effects on their attitudes.
Under high-distraction conditions, however, participants presumably based
their attitudes on the extraneous affect they were experiencing rather than
the content of the persuasive message. Having done so, they may have attempted to justify the affect-based attitude they have formed using processes suggested by McGuire and McGuire (1991; see also Janis & King,
1954). For example, they may engage in rationalization (evaluating the desirability of the consequences in a manner that is evaluatively consistent
with their attitude) and wishful thinking (increasing their beliefs that these
THE ROLE OF AFFECT
363
consequences would occur). These latter effects, however, should be independent.

These speculations were supported by supplementary regression analyses of attitudes as a function of predicted values based on Equation 11.1
(i.e., Sbi ei ) and indices of beliefs and evaluations considered separately (Sbi
and Sei ). Under low-distraction conditions, the multiplicative effects of beliefs and evaluations implied by Equation 11.1 accounted for 6% of the variance in attitudes over and above the effects of beliefs and evaluations considered independently. Under high-distraction conditions, however, the
multiplicative effects contributed only a trivial amount (.01%) of the variance over and above the component measures.
In summary, Albarracin and Wyers (2001) research suggested that when
participants have ample time and motivation to think carefully about the
content of an attitude-focused persuasive communication, they may assess
its implications for the likelihood and desirability of alternative consequences of the position advocated in the message, and then may integrate
them in a manner suggested by Fishbein and Ajzen (1975). When they are
distracted, however, they are more likely to base their attitudes on their
affective reactions to the position advocated independently of the communication content. Moreover, these attitudes, once formed, can have reciprocal effects on the beliefs and evaluations associated with specific consequences of the position being advocated in order to justify this attitude,
engaging in wishful thinking and rationalization of a sort similar to that suggested by McGuire and McGuire (1991).
FINAL COMMENTS
As noted at the outset, peoples affective reactions can come into play in
judgments and behavioral decisions in virtually every domain of social experience, both in and outside the laboratory. The inordinate length of this
chapter testifies to the diversity of its possible effects and the conditions in
which they occur. We have argued, however, that affective reactions per se
exert their influence primarily through the information they provide about
persons, objects, and events to which these reactions are directed, and the
likelihood that ones information-processing objectives have been satisfactorily attained.
The performance-feedback conceptualization suggested by Martin et
al.s (1993) findings can account for many effects of affect on judgments and
behavior. It therefore seems worthy of continued consideration in future research. Its ultimately utility is nonetheless something of a promissory note.
That is, a rigorous application of the conceptualization requires a task analysis of the cognitive processing steps that underlie a judgment or behav-
364
CHAPTER 11
ioral decision and the mental operations that are performed at each step.
An example of this analysis was provided in our analysis of the impact of affect on self-judgments. Similar analyses of other information processing situations may be fruitful.
Our analysis of the influence of affect on judgments and behavior has
largely been independent of our discussion in previous chapters. As noted,
however, the stories that people write about themselves, as well as stories
they encounter in conversations or in the media, are very likely to elicit affective reactions. To this extent, many of the inference process that are
based on narrative representations of knowledge (see chaps. 9 and 10)
could be based in large part on affect rather than the substantive implications of the information per se. The impact of pictures and visual images on
narrative-based processing could likewise be mediated by their influences
on peoples affective reactions and the use of these reactions as bases for
judgment. Therefore, although we have not explicated these possibilities in
this chapter, the relevance of the phenomena we have described to issues
raised elsewhere in this volume is hopefully apparent.
C H A P T E R
12
Epilogue: The Book, the Author,
and Philosophical Ruminations
The Book
The information we acquire in daily life can come from many sources: direct observation, conversations with friends and acquaintances, movies
and television, newspapers, and books. Some of this information is about
unknown or fictitious people and events. In many cases, however, we already know a lot about the things that are described, and we spontaneously recognize the information as either true or false. When information is
conveyed in pictures or is acquired through observation, we often form visual images and retain them in memory as part of our knowledge about
their referents. However, verbal descriptions of events can stimulate the
construction of visual images as well. Finally, the information we receive
can often elicit affective reactions both at the time it is acquired and when
we are later reminded of the events to which it refers, and these reactions
can be used as a basis for evaluating the events and the people or objects
involved in them. The information from all of these sources, and in all of
these modalities, combines to form our accumulated knowledge about the
world in which we live, and serves as a basis for our judgments and behavioral decisions.
This volume has provided a conceptualization of the way in which this
information is comprehended and stored in memory and how it is later retrieved and used. Much of our discussion has focused on the construction
and use of narratives (that is, temporally and thematically organized sequences of events that are stored in memory as a unit). These mental representations include episode models of specific events that we happen to en365
366
CHAPTER 12
counter, linguistic representations of these events that we form in the

course of communicating about them, and implicit theories that we use to
understand our experiences and to construe their antecedents and consequences. In conceptualizing the role of narrative representations in comprehension and judgment, however, two other ingredients of real-world
knowledge were taken into account: the visual images that people construct on the basis of their experiences and the affective reactions that the
experiences elicit.
Although my concern has been with the way in which information about
real people and events is comprehended and used in daily life, the research I
have brought to bear on these matters has largely been conducted in the laboratory. Nevertheless, the processes identified in the research seem likely to
generalize to daily life situations. The comprehension and validation processes described in chapter 4 were confirmed empirically using statements
about actual people and events with which participants were already quite
familiar. The conceptualization of episode models was applied in understanding not only peoples comprehension of information about hypothetical individuals (chap. 5) but also their representations of their own experiences and
those of close acquaintances (chap. 6). Our discussion of the comprehension
of information in a social context (chap. 7) emphasized the processes of identifying the pragmatic implications of information of the sort one encounters
in newspapers or acquires in informal conversation. It also examined the cognitive processes that underlie reactions to jokes and other humor-eliciting
experiences of the sort we encounter in social situations outside the laboratory (see chap. 8). In discussing inference processes, I examined the role of
both visual images and narrative forms of information in judgments (chap. 9),
and also explored the impact of implicit theories that people construct about
themselves and the world in which they live (chap. 10). This latter discussion
covered research on a wide variety of topics including counterfactual reasoning, cultural influences on judgments and behavior, marital satisfaction, and
perceptions of social support.
At the same time, I attempted to place the discussion of these matters in
the context of a general theoretical analysis of social information processing that permits comprehension and judgment phenomena to be interpreted within a common conceptual framework. This formulation, which is
captured by a series of formal postulates (for a summary, see Appendix),
has several features in common with the earlier theory proposed by Wyer
and Srull (1989). However, it differs from this theory in ways that increase
its potential applicability to real-world information processing without compromising its ability to account for a wide range of phenomena to which the
1989 formulation was applicable.
The formulation I have proposed is broader in scope than most other
theories of social information processing. In fact, theories that focus on
EPILOGUE
367
more circumscribed sets of phenomena can often be viewed as components of this formulation. (That is, they specify the operations that are performed by one or another special purpose processing unit and represented
either in a goal schema or in a processing units library.) In this regard,
many recent conceptualizations of social cognition have recognized that
different cognitive processes can underlie responses to information (e.g.,
see Chaiken & Trope, 1999). However, the assumption that several alternative processes can be brought to bear on a judgment or decision, depending on social, informational, and individual difference variables that determine their applicability and relative accessibility, is also a central feature of
both the Wyer and Srull (1989) formulation and the present one.
The proposed conceptualization is not without flaws. For one thing, it is
metaphorical, and does not pretend to be an exact depiction of the physiology of the brain. As noted in chapter 1, however, most formulations of information processing are inherently metaphorical, and should be judged on
the basis of their utility and not their validity. The present formulation appears to be a useful one, and establishes directions for future research in
social information processing. If this book stimulates even a few researchers to pursue these directions, its objectives will be accomplished.
The Author
In the concluding remarks of his recent book, Constructing social psychology,
Bill McGuire (1999) commented: It may be best if each of us . . . makes his
or her own way . . . but ones own way should be taken deliberatively and
egosyntonically, not stumbled into by chance (p. 432). I wish that I could
say that my own career exemplifies this sage advice. I have, undoubtedly,
made my own way. But this way was hardly deliberative. Even my decision
to become a psychologist was determined by largely fortuitous events. Although my reflections about these events is perhaps less out of place in a
volume that emphasizes the role of narratives than it might otherwise be,
they are likely to be of little interest to anyone but myself. At the risk of
abusing a privilege, however, I would like to take advantage of this opportunity to indulge myself in these reflections.
When I was in high school, I wanted to be either a jazz musician or a
sportswriter. I was discouraged from both pursuits by my father. On one
hand, he pointed out that my skills as a clarinet and saxophone player
would probably qualify me for no better than third chair in a mediocre
dance band. He further argued that to be a sportswriter, I would inevitably
have to start off as a copy boy at a city newspaper and would have to work
my way up gradually over a period of many years. Moreover, he claimed
that this would be true regardless of whether I went to college or not.
Therefore, he argued that if I decided to go to college at all, I should major
368
CHAPTER 12
in something more useful. So, having had my primary options dashed, and
having received a scholarship to Rensselaer Polytechnic Institute, I became
an electrical engineer.
Four years in engineering school did little to stimulate my interest in the
field. Nevertheless, I took a job at Bell Laboratories, which offered me full
salary for going to school 3 days a week and working only 2. Although winding up 2 years later with a Masters degree, I spent most of my time improving my table tennis skills and imbibing in the intellectual atmosphere that
pervaded Greenwich Village in the late 1950s. These activities convinced
me that I should get a liberal arts education.
I took a battery of ability and interest tests, which indicated that the only
things for which I was suited were architecture and psychology. I hardly
knew what psychology was, having barely escaped a D in my single undergraduate course in the area. I nevertheless decided to apply to graduate
schools in psychology in physical locations I might enjoy, without knowing
anything at all about the schools themselves, and not really thinking that I
might actually go. Inevitably, one school that met my enjoyment criterion
was the University of Colorado. The application form, however, listed three
areas that were completely foreign to me: experimental (which sounded
like engineering), clinical (which sounded like medicine) and socialpersonality. Because I thought I understood what social and personality
meant, I checked the third area as my preference. By chance, this was the
one area at Colorado in which Public Health Service fellowships were available, and I was offered one. My bluff being called, I accepted it, knowing
nothing at all about what I was getting into and not really caring, but assuming that I would have the opportunity to expose myself to liberal arts in general.
This was in fact the case. The graduate program at Colorado in 1960 was
hardly rigorous, and I was able to get through the program in 3 years with a
total of only 27 hours of psychology (most of which was for thesis and independent study). I actually audited more courses in humanities, sociology,
history, and philosophy than I took in psychology. What Colorado lacked in
substance, however, it made up for in stimulation, due largely to OJ Harvey
(one of the most inspirational individuals I had ever met) and Bill Scott (my
dissertation advisor, who gave me an appreciation for the need to think rigorously as well as innovatively). In general, Colorados reactions to my
graduate school career were probably exemplified by one faculty members
prediction that unless I was always right, I would never succeed as a psychologist because I never listened to anything that anyone else was saying.
I obviously havent always been right. Whether that reflects a change in my
chronic disposition, however, is debatable.
I spent an additional year at Colorado in 1962 with Glenn Terrell, a developmental psychologist and Associate Dean, on research on nonintellective
EPILOGUE
369
correlates of scholastic achievement. This was a year of political upheaval

in Colorado, leading to the election of an ultra-conservative Board of
Trustees and, consequently, the resignation of the entire University administration. Terrell moved to the University of Illinois at Chicago Circle to become Dean of Arts and Sciences (later Dean of Faculties). I had never seriously thought of an academic career, assuming that I would wind up back in
industry doing human factors research. However, I was offered a job at
the University of Iowa Institute for Child Behavior and Development, which
didnt seem to care whether I knew anything about children, either personally or professionally. So, I accepted it.
At that time, Spence learning theory dominated psychology at Iowa.
Mental representations of knowledge were not in vogue. After 2 years of
indoctrination, I escaped to rejoin Terrell at Chicago Circle to complete the
research on academic achievement and to teach part time in Psychology.
Three years past my PhD, I was still woefully ignorant of theory and research in social psychology. However, I had the good fortunate to have an
office next to Harry Upshaw, who blended the charisma of OJ Harvey with
the rigor of Bill Scott, and who perhaps as much as anyone gave me an appreciation of what social psychology was really all about. Despite this chaotic start, and despite being an active participant in the anti-war movement
of the 1960s, I somehow developed a program of research. The four books I
have published over the past three decades are the result.
The Philosophical Ruminations
The four books I have written over the course of the past 30 years have all
concerned one or another aspect of social information processing. This
might give the impression of a deliberative, single-minded approach to psychological research and theorizing. This impression would be misleading,
however. Both theoretically and methodologically, I have throughout my career simply explored the things I have found intellectually exciting without
giving much thought to the reason for this excitement and how my explorations might fit into the general scheme of things. The fact that they have
turned out to fit such a scheme has been as surprising to me as to anyone.
The specific areas of research in which Ive been involved have concerned the nonintellective determinants of scholastic achievement, persistence in goal-directed activity, conformity, two-person games, algebraic
models of impression formation, stereotyping, belief organization and
change, response language effects on judgments along category scales,
emotional communication, knowledge accessibility, the mental representation of persons and social events, pragmatic factors in social communication, humor elicitation, communication and persuasion, country of origin effects on consumer judgments, affect and cognition, and cross-cultural
370
CHAPTER 12
influences on decision making. Much of this work was inspired by students

whose diverse interests and ideas captured my imagination. It probably
takes a lot of imagination to identify the conceptual and methodological underpinnings of this work.
However, there is usually a reason why something seems interesting,
even if one is not aware of why this is the case. The reason may only be
identified after the fact, on reflection. This is certainly the case with me.
The common thread that connects my interests is a concern with the mental processes that mediate judgments and behavior. These processes operate at several stages: attention, comprehension, the organization and storage of information in memory, inference, and response generation. Almost
all social phenomena implicate one or another of these stages. Correspondingly, I have found myself interested in almost everything. However, the realization that this is what brings it all together has been a personal discovery to me, and was not the result of a premeditated approach or carefully
thought out research paradigm.
Having said this, I have had a desire to develop comprehensive conceptualizations of social phenomena that not only integrate a lot of diverse empirical findings but are sufficiently rigorous to permit the derivation of new,
empirically testable hypotheses. This motive has led me to think both
broadly and formally about the cognitive underpinnings of memory, judgment, and behavior. This penchant for formalism as well as breadth has
sometimes led to overly grandiose conceptualizations, composed of mindboggling sets of postulates and assumptions that few colleagues have had
the stamina to digest. But Ive found it hard to think in any other way.
Three somewhat diverging comments I encountered early in my professional life have remained with me over the years and have had a major impact. In an informal graduate seminar at Colorado, Nan Wertheimer, an epidemiologist, observed that people should attempt to avoid becoming
committed to any particular theory for as long as possible, because once
that commitment was made, their contribution to psychology was over.
That is, they would spend the rest of their lives defending their theory and
would blind themselves to new ideas and approaches.
The second, which seems contradictory on the surface, comes from a
chapter by McGuire in 1972. He likened the search for knowledge to that of
a boy who is lost in the woods:
If the boy strikes out in any one direction and keeps making progress in that
direction, it is likely that he will eventually find his way out of the woods. It
might be that if he had pursued a different line of progress, he would have
gotten out even sooner, but almost any line of advance will suffice if pursued
sufficiently long. The only real danger is that one will not persist in any direction but will wander around at random and never get out of the woods. Analogously, if the empirical scientist is lost in a complex area, his pursuing the im-
EPILOGUE
371
plications of any reasonable paradigm in a steady direction will probably led

him to some ultimate clarification of the area. If in stead he drops each theory
as soon as the slightest negative evidence crops up, there results the danger
that he will wander around in circles and not obtain any clarification. The researcher who keeps the faith and pursues his paradigm to ultimate enlightenment may find that there is a much better theory he could have chosen initially. But his persistence will also have demonstrated the truth of Blakes
proverb that if the fool would persist in his folly, he would become wise.
(p. 138).
Finally, I read an article in graduate school by Edward Tolman, one of the

foremost animal learning theorists of his era, whose views, radical at the
time, have since been implicitly vindicated by both cognitive and social
psychologists. In reviewing his career, he noted that he, like McGuires boy
in the woods, might not have taken the most direct route to understanding,
and that if he had it to do over again, he might have taken a different path.
However, he had had fun along the way, and that was what mattered.
All three of these comments have had an impact on my philosophy of
life, both professionally and personally. Following McGuire, I have tried to
develop a conceptual perspective on social phenomena to guide my theoretical and empirical understanding of both new and old information. At the
same time, I have tried to follow Nan Wertheimers advice not to identify
too closely with this perspective, not to take myself too seriously, to take
my theories with a grain of salt and to challenge them as harshly as I challenge others, and to be receptive to new and better ways of viewing the
world. Finally, I, like Tolman, might do things differently if I had them to do
over. (Indeed, I might choose to ignore my fathers advice and try my hand
at sports writing or clarinet playing.) However, I have also had fun along the
way, and I, again like Tolman, think that this is what really matters.
A Debt of Gratitude
Many individuals have had a major influence on my career. Some are
more obvious than others. Unquestionably one of the major influences
has been Bill McGuire. His empirical and theoretical work in both cognitive organization and persuasion was among the most exciting I have ever
encountered, and captivated me at an early stage in my career. Moreover,
his encouragement to me while serving as the editor of JPSP instilled the
confidence I needed at a time when I was conflicted with many self-doubts.
Although I have personally met McGuire less than a half dozen times in
my life (two of which came from festschrifts in one anothers honor that
we attended), I consider him to be my primary mentor, whose influence
on me over the years has extended far beyond the specific research and
theorizing I have done.
372
CHAPTER 12
I am also in debt to Norman Anderson, who also encouraged me at this

early point in my career. Although my approach to understanding information processing ultimately diverged from Normans, his work as well as
McGuires stimulated my interest in the underpinnings of social judgment.
Andersons most enduring influence on my career was rather fortuitous,
however. He conducted a 3-week workshop at La Jolla in 1973. During the
course of this workshop, Tom Ostrom, Dave Hamilton, Reid Hastie, Ebbe
Ebbesen and I, none of whom had ever before met, became good friends as
well as admirers of one anothers work. Stimulated by our intellectual exchanges during the workshop, we began to meet informally at irregular intervals over the next several years (usually in Ostroms gigantic house), to
discuss alternative approaches to understanding person impression formation. (This group, along with Don Carlston and Thom Srull who joined later,
were the original nucleus of the Person Memory Interest Group, now over
70 members strong, which continues to flourish.) These weekend-long discussions established a new direction to my theorizing and my research on
the mental representation of social information.
During the same period, I had occasion to serve as a journal editor and,
in this capacity, became a fan of both Tory Higgins and Jim Sherman. The
first stimulated my interest in input processes (e.g., the role of knowledge
accessibility) and the second stimulated my interest in output processes
(the cognitive determinants of overt behavior). More recently, Jerry Clore
provided similar stimulation to my interest in affect. Their influence, as well
as the influence of Ostrom et al., is readily apparent in my work over the
past two decades.
However, the major influences on my work over the past 25 years have
been the graduate and postdoctoral students with whom I had contact at
the University of Illinois. Don Carlston and John Lynch, early sources of
stimulation, both have become outstanding researchers in psychology and
consumer behavior, respectively. In 1980, a small group of students consisting of Thom Srull, Sallie Gordon, and Jon Hartwick began meeting weekly to
exchange ongoing research they were conducting in various areas related
to social cognition. This group grew over the years to become the University of Illinois Social Cognition Group, a consortium of faculty, postdoctoral
students, and graduate students who met informally each week to exchange theoretical ideas and empirical findings in social cognition. My participation in this group, and my collaboration with its individual members,
has unquestionably been the most personally rewarding experience of my
professional life.
Although all of the members of this group over the 20 years of its existence are too numerous to mention, there are several to whom I am particularly indebted. Lenny Martin, Norbert Schwarz, and Fritz Strack, who visited
Illinois as postdoctoral students, have since become leaders in the field. I
EPILOGUE
373
continue to be profoundly influenced by their work. Thom Srull, first as student and then as colleague, was a major collaborator for many years. Other
students whose contributions to my career are particularly noteworthy include Galen Bodenhausen, Lee Budesheim, Bob Fuhrman, Alan Lambert,
Victor Ottati, and Dave Trafimow. Dolores Albarracin, Deborah Gruenfeld,
Linda Isbell, and L. J. Shrum were also major collaborators. Their research
on communication and persuasion, the pragmatic implications of information, affect, and perceptions of social reality, respectively, has had an obvious impact on my own. Rashmi Adaval, Stan Colcombe, and Eric Mankowski, in quite different ways, stimulated my interest in the role of
narratives in information processing. Most of these individuals have been
close friends as well as stimulating colleagues, making our collaboration
doubly rewarding.
It has been a long trip, and the journey is not quite ended. However,
these individuals and others too numerous to mention have made the trip
exciting and enjoyable beyond belief, and I will always be in their debt.
A P P E N D I X
Summary of Postulates
Retrieval Postulates (chap. 2)

Postulate 2.1. Each knowledge representation that is formed, either in
the course of comprehension or in the course of higher level, goal-directed
processing, constitutes a separate unit of knowledge that is stored independently of other knowledge units.
Postulate 2.2. The activation of a set of stimulus features causes all

knowledge representations that contain these features to resonate. The resonance of a given representation, k, increases with (a) the similarity of its
features to those in the stimulus set and (b) the length of time the stimulus
set has been activated. The subjective similarity of the stimulus features to
the representation, and thus the rate at which its resonance increases, is
estimated by the quantity:
Res(k) =
n( pI k)
,
n( k)
where n(k) is the number of features in the representation and n( pI k) is

the number of features that are common to both the representation and the
stimulus set. If the resonance reaches a threshold value, T(k), the representation is activated.
374
SUMMARY OF POSTULATES
375
Postulate 2.3. (a) If two or more representations resonate with a given

set of probe cues, the one that reaches threshold most quickly is identified.
(b) If several representations reach threshold simultaneously, a composite
of the features is activated. The weight of each feature in the composite is a
positive function of the proportion of times it occurs in the set of representations involved.
Postulate 2.4. If a knowledge representation has resonated with a set of
probe cues, this resonance does not dissipate immediately but rather, decays gradually over time.
Postulate 2.5. The activation threshold of a knowledge representation
in memory, T(k), is an increasing function of the number of times the representation has been activated in the past.
Sufficiency Postulate (chap. 3)
Postulate 3.1. People retrieve and use only the amount of information
that they consider to be sufficient for attaining the processing objective
they are pursuing at the time. The information they use is likely to be the
first goal-relevant information that comes to mind or, alternatively, permits
their objective to be attained most quickly and easily.
Comprehension and Verification Postulates (chap. 4)
Postulate 4.1. Event and episode models are mental simulations of an
event, or sequence of events, that are temporally and spatially constrained.
These models are constructed spontaneously in the course of comprehending events that occur at a specifiable (although not necessarily specified)
time and place.
Postulate 4.2. An event model can consist of both a mental image and a
linguistically coded description of the event or state of affairs to which it refers. Although the image component of the model is obligatory, the linguistic component is optional.
Postulate 4.3. The subject and predicate identified by the Parser are independently used as probe cues of memory, resonating with all of the mental representations that contain them (see Postulate 2.2). In each case, the
representation whose resonance level reaches threshold most quickly is
most likely to be identified.
376
APPENDIX
Postulate 4.4. If a memory search activates a previously formed knowledge representation whose features include both the subject probe cues
and the predicate probe cues, the new information is spontaneously recognized as redundant with prior knowledge and a new representation is not
formed.
Postulate 4.5. If no previously formed model (i.e., knowledge unit) refers to both the subject and predicate of a target statement, the features
that are activated by the targets subject are compared to a weighted composite of the features associated with the targets predicate. This comparison yields a global estimate of similarity, S. A new situation model is formed
if either (a) S is greater than a minimum threshold value, TC, or (b) the targets subject contains a subset of features that are common to the subject
of all models activated by the probe. In the latter case, the new model is
constructed by combining the composite of features activated by the subject and the composite of features activated by the predicate. In all cases,
however, only the features that are weighted heavily in the composite are included in the new model that is constructed.
Postulate 4.6. A statement is comprehended spontaneously if it is redundant with prior knowledge or if a new model is constructed of it. If neither is the case, comprehension of the statement is not spontaneous, but
requires (Executor-controlled) goal-directed processing.
Postulate 4.7. A statement is spontaneously recognized as true in the

course of comprehending it if (a) a situation model of the event or state described already exists in memory and (b) no tag has been attached to the
model that designates it as false.
Postulate 4.8. A statement about an event is recognized as true in the

course of comprehending it if the similarity of its subject to those of previously formed models of the event (S) exceeds an upper threshold, TT . The
statement is identified as false if S is below a lower threshold, TF . If S falls
between TT and TF , the statement is comprehended but its validity is not
spontaneously evaluated.
Postulate 4.9. Spontaneous inferences that a statement is true or false

are transferred to the Work Space along with a representation of the statement that has been formed in the course of comprehending it.
377
Construction of Episode Models (chap. 5)

Postulate 5.1. Once a situation model has been formed, subsequent information is integrated into the model if and only if this information concerns the same situation as the model or a thematically related one.
Postulate 5.2. When new information requires a modification of an
event or episode model that had been constructed on the basis of prior information, the original model is retained in memory. However, a second
representation is constructed consisting of a modified version of the first
model along with the implications of the new information.
Humor Elicitation Postulates (chap. 8)
Postulate 8.1. New experiences are spontaneously interpreted in terms
of previously formed event representations (e.g., situation models) with
which they have common features. When more than one alternative knowledge representation is applicable for interpreting an experience, the representation that is most easily accessible in memory is typically applied. The
result of this cognitive activity is a newly formed situation model of the
event being comprehended.
Postulate 8.2. Once a mental representation has been formed of an experience, this representation, along with the concepts and knowledge that
were activated in the course of constructing it, is used to comprehend thematically related experiences that occur subsequently.
Postulate 8.3. When information about an event cannot be interpreted
in terms of concepts drawn from the same domain of knowledge that was
applied to previous events, recipients attempt to identify concepts and
knowledge in a different domain that is applicable to both the new event
and the preceding ones. If these concepts can be found, the previously
learned events are reinterpreted in terms of them.
Postulate 8.4. Humor is elicited by a reinterpretation of a stimulus
event only if the features of one or more features of the event are diminished in value or importance relative to that implied by the original interpretation.
Postulate 8.5. The amount of amusement that is potentially elicited as a
result of reinterpreting a stimulus event is a nonmonotonic (inverted-U)
function of the time and effort that is required to make this interpretation.
378
APPENDIX
Postulate 8.6. The amount of humor that is elicited as a result of reinterpreting a stimulus event is a monotonic function of the amount of cognitive elaboration of the event and its implications that occurs subsequent to
its reinterpretation.
1. If recipients processing objective at the time is simply to comprehend
and enjoy experiencing the event, cognitive elaboration of the event will typically increase the humor it elicits.
2. If recipients goal is more restricted, the event is elaborated in terms of
its implications for this more specific objective. In this case, cognitive elaboration could either increase, decrease, or have no effect on the amount of humor elicited, depending on whether the humor-eliciting reinterpretation of
the event is relevant to the attainment of this objective.
Inference Postulate (Completion Principle; chap. 10)
Postulate 10.1. If the information available about a specific situation
instantiates all but one of the propositions that compose a molecule, an
instantiation of the remaining proposition will be spontaneously inferred.
Affect Postulates (chap. 11)
Postulate 11.1. Affective reactions may activate a specific affectrelevant concept that can be used to interpret them. However, affective reactions do not spontaneously influence the accessibility in memory of other
similarly valenced semantic concepts and declarative knowledge.
Postulate 11.2. The affective reactions that one experiences, along with
concepts and cognitions that are stimulated by an appraisal of a stimulus
situation, can compose a precondition for a previously learned sequence of
responses (i.e., a cognitive production). This response sequence may be activated spontaneously, with little conscious cognitive mediation, under conditions in which the precondition is met.
Postulate 11.3. When people engage in behavior for the purpose of attaining an external goal and objective criteria for inferring the attainment of
this goal do not exist, they use the affect they are experiencing to infer
whether they feel that the strategy they have used in pursuit of this goal,
and the outputs they have generated by employing it, are appropriate.
Therefore, they are more likely to infer that they have met the criterion for
satisfactory performance if they are experiencing positive affect than if they
are not.
379
Postulate 11.4. When peoples primary reason for engaging in behavior

is enjoyment, they use the affect they are experiencing as a basis for inferring this enjoyment. Therefore, they will continue the activity longer if they
are experiencing positive affect than if they are not.
References
Aaker, J. L., & Lee, A. (2001). I seek pleasures, we avoid pains: The role of self-regulatory goals in
information processing and persuasion. Journal of Consumer Research, 27, 3349.
Abelson, R. P. (1959). Modes of resolution of belief dilemmas. Conflict Resolution, 3, 343352.
Abelson, R. P. (1976). Script processing in attitude formation and decision making. In J. S. Carroll
& J. W. Payne (Eds.), Cognition and social behavior (pp. 3346). Hillsdale, NJ: Lawrence
Erlbaum Associates.
Abelson, R. P., Kinder, D. R., Peters, M. D., & Fiske, S. T. (1982). Affective and semantic components in political person perception. Journal of Personality and Social Psychology, 42, 619630.
Abelson, R. P., & Reich, C. M. (1969). Implicational molecules: A method for extracting meaning
from input sentences. In D. E. Walker & L. M. Norton (Eds.), Proceedings of the International
Joint Conference on Artificial Intelligence.
Adaval, R. (2001). Sometimes it just feels right: The differential weighting of affect-consistent and
affect-inconsistent product information. Journal of Consumer Research, 7, 207245.
Adaval, R. (2003). How good gets better and bad gets worse: Understanding the impact of affect
on evaluations of known brands. Journal of Consumer Research.
Adaval, R., Isbell, L. M., & Wyer, R. S. (2003). Political impression formation: The role of pictures and
information presentation format on evaluations of politicians. Unpublished manuscript, Hong
Kong University of Science and Technology.
Adaval, R., & Monroe, K. B. (2002). Automatic construction and use of contextual information for
product and price evaluations. Journal of Consumer Research, 28, 572588.
Adaval, R., & Wyer, R. S. (1998). The role of narratives in consumer information processing. Journal of Consumer Psychology, 7, 207245.
Adaval, R., & Wyer, R. S. (2003). Memory for social interactions: Effects of post-information processing
objectives on the recognition of protagonists statements and nonverbal behaviors. Unpublished
manuscript, Hong Kong University of Science and Technology.
Albarracin, D., & Wyer, R. S. (2000). The cognitive impact of past behavior: Influences on beliefs,
attitudes, and future behavioral decisions. Journal of Personality and Social Psychology, 79,
522.
Albarracin, D., & Wyer, R. S. (2001). Elaborative and nonelaborative processing of a behaviorrelated persuasive communication. Personality and Social Psychology Bulletin, 27, 691705.
380
REFERENCES
381
Albrecht, J. E., & Myers, J. L. (1995). Updating a mental model: Maintaining both local and global
coherence. Journal of Experimental Psychology: Learning, Memory, and Cognition, 19,
10611070.
Anderson, A., Garrod, S. C., & Sanford, A. J. (1983). The accessibility of pronominal antecedents
as a function of episode shifts in narrative text. Quarterly Journal of Experimental Psychology,
35A, 427440.
Anderson, J. R. (1978). Arguments concerning representations for mental imagery. Psychological
Review, 85, 249277.
Anderson, J. R. (1982). Acquisition of cognitive skill. Psychological Review, 89, 369406.
Anderson, J. R. (1983). The architecture of cognition. Cambridge, MA: Harvard University Press.
Anderson, J. R., & Bower, G. H. (1973). Human associative memory. Washington, DC: V. H. Winston.
Anderson, N. H. (1971). Integration theory and attitude change. Psychological Review, 78, 171206.
Anderson, N. H. (1981). Foundations of information integration theory. New York: Academic Press.
Anderson, R. C., Reynolds, R. E., Schallert, D. L., & Goetz, F. T. (1977). Frameworks for comprehending discourse (Tech. Rep. No. 12). Urbana, IL: Laboratory for Cognitive Studies in Education,
University of Illinois at Urbana-Champaign.
Apter, M. J. (1982). The experience of motivation: The theory of psychological reversals. San Diego,
CA: Academic Press.
Arkin, R. M., Gleason, J. M., & Johnston, S. (1976). Effect of perceived choice, expected outcome
and observed outcome of an action on the causal attributions of actors. Journal of Experimental Social Psychology, 12, 151158.
Banks, W. P. (1977). Encoding and processing of symbolic information in comparative judgments. In G. H. Bower (Ed.), The psychology of learning and motivation (Vol. 11). New York: Academic Press.
Barclay, J. R. (1973). The role of comprehension in remembering sentences. Cognitive Psychology,
4, 229254.
Bargh, J. A. (1994). The four horsemen of automaticity: Awareness, intention, efficiency, and control in social cognition. In R. S. Wyer & T. K. Srull (Eds.), Handbook of social cognition (2nd ed.,
Vol. 1, pp. 140). Hillsdale, NJ: Lawrence Erlbaum Associates.
Bargh, J. A. (1997). The automaticity of everyday life. In R. S. Wyer (Ed.), Advances in social cognition (Vol. 10, pp. 162). Mahwah, NJ: Lawrence Erlbaum Associates.
Bargh, J. A., Bond, R. N., Lombardi, W., & Tota, M. E. (1986). The additive nature of chronic and
temporary sources of construct accessibility. Journal of Personality and Social Psychology, 50,
869878.
Bargh, J. A., Chaiken, S., Govender, R., & Pratto, F. (1992). The generality of the automatic attitude activation effect. Journal of Personality and Social Psychology, 62, 893912.
Bargh, J. A., Chaiken, S., Raymond, P., & Hymes, C. (1996). The automatic evaluation effect: Unconditionally automatic attitude activation in a pronunciation task. Journal of Experimental
Social Psychology, 32, 104120.
Bargh, J. A., Chen, M., & Burrows, L. (1996). Automaticity of social behavior: Direct effects of trait
construct and stereotype activation on action. Journal of Personality and Social Psychology,
71, 230244.
Bargh, J. A., & Pietromonaco, P. (1982). Automatic information processing and social perception:
The influence of trait information presented outside of conscious awareness on impression
formation. Journal of Personality and Social Psychology, 43, 437449.
Baron, R. A., & Byrne, D. (2000). Social psychology (9th ed.). Boston, MA: Allyn & Bacon.
Barsalou, L. W. (1990). On the indistinguishability of exemplar memory and abstraction in category representations. In T. K. Srull & R. S. Wyer (Eds.), Content and process specificity in the effects of prior experiences: Advances in social cognition (Vol. 3, pp. 6188). Hillsdale, NJ: Lawrence Erlbaum Associates.
Barsalou, L. W. (1993). Flexibility, structure, and linguistic vagary in concepts: Manifestations of
a compositional system of perceptual symbols. In A. F. Collins, S. E. Gathercole, M. A.
382
REFERENCES
Conway, & P. E. Morris (Eds.), Theories of memory (pp. 29102). Hillsdale, NJ: Lawrence
Erlbaum Associates.
Baumeister, R. F., & Newman, L. S. (1994). How stories make sense of personal experience: Motives that shape autobiographical memories. Personality and Social Psychology Bulletin, 20,
676690.
Bell, D. E. (1982). Regret in decision making under uncertainty. Operations Research, 30, 961981.
Bem, D. J. (1972). Self-perception theory. In L. Berkowitz (Ed.), Advances in experimental social
psychology (Vol. 6, pp. 162). New York: Academic Press.
Bem, D. J., & McConnell, H. K. (1970). Testing the self-perception explanation of dissonance phenomena: On the salience of premanipulation attitudes. Journal of Personality and Social Psychology, 14, 2331.
Bergson, H. (1911). Laughter: An essay on the meaning of the comic. New York: Macmillan.
Berlyne, D. E. (1969). Laughter, humor and play. In G. Lindzey & E. Aronson (Eds.), Handbook of
social psychology (Vol. 3, pp. 795852). Reading, MA: Addison-Wesley.
Berlyne, D. E. (1971). Aesthetics and psychology. New York: Appleton-Century.
Bettman, J. R., Capon, N., & Lutz, R. J. (1975). Multiattribute measurement models and
multiattribute attitude theory: A test of construct validity. Journal of Consumer Research, 1,
115.
Birnbaum, M. H. (1974). The nonadditivity of personality impressions. Journal of Experimental
Psychology, 102, 543561.
Black, J. B., Galambos, J. A., & Read, S. (1984). Comprehending stories and social situations. In
R. S. Wyer & T. K. Srull (Eds.), Handbook of social cognition (Vol. 3, pp. 4586). Hillsdale, NJ:
Lawrence Erlbaum Associates.
Black, J. B., Turner, T., & Bower, G. H. (1979). Point of view in narrative comprehension, memory,
and production. Journal of Verbal Learning and Verbal Behavior, 11, 717726.
Blaney, P. H. (1986). Affect and memory: A review. Psychological Bulletin, 99, 229246.
Bless, H. (2001). Mood and the use of general knowledge structures. In L. L. Martin & G. L. Clore
(Eds.), Theories of mood and cognition: A users guidebook (pp. 926). Mahwah, NJ: Lawrence
Erlbaum Associates.
Bless, H., Bohner, G., Schwarz, H., & Strack, F. (1990). Mood and persuasion: A cognitive response analysis. Personality and Social Psychology Bulletin, 16, 331345.
Bless, H., Clore, G. L., Schwarz, N., Golisano, V., Rabe, C., & Woelke, M. (1996). Mood and the use
of scripts: Does being in a happy mood really lead to mindlessness? Journal of Personality and
Bodenhausen, G. V. (1993). Emotions, arousal, and stereotypic judgments: A heuristic model of
affect and stereotyping. In D. M. Mackie & D. L. Hamilton (Eds.), Affect, cognition and stereotyping: Interactive processes in group perception (pp. 1337). San Diego, CA: Academic Press.
Bodenhausen, G. V., & Macrae, C. N. (1998). Stereotype activation and inhibition. In R. S. Wyer
(Ed.), Advances in social cognition (Vol. 11, pp. 152). Mahwah, NJ: Lawrence Erlbaum Associates.
Bodenhausen, G. V., Sheppard, L. A., & Kramer, G. P. (1994). Negative affect and social judgment:
The differential impact of anger and sadness. European Journal of Social Psychology, 24, 4562.
Bodenhausen, G. V., & Wyer, R. S. (1985). Effects of stereotypes on decision-making and information-processing strategies. Journal of Personality and Social Psychology, 52, 871880.
Boorstin, D. (1973). The Americans: The national experience. New York: Random House.
Bower, G. H. (1981). Emotional mood and memory. American Psychologist, 36, 129148.
Bower, G. H. (1991). Mood congruity of social judgments. In J. P. Forgas (Ed.), Emotion and social
judgments (pp. 3153). Oxford, England: Pergamon.
Bower, G. H., Black, J. B., & Turner, T. J. (1979). Scripts in memory for texts. Cognitive Psychology,
11, 177220.
Bower, G. H., Gilligan, S. G., & Monteiro, K. P. (1981). Selectivity of learning caused by affective
states. Journal of Experimental Psychology: General, 110, 451473.
REFERENCES
383
Bower, G. H., & Mayer, J. D. (1985). Failure to replicate mood-dependent retrieval. Bulletin of the
Psychonomic Society, 23, 3942.
Bransford, J. D., Barclay, J. R., & Franks, J. J. (1972). Sentence memory: A constructive versus interpretative approach. Cognitive Psychology, 3, 193209.
Bransford, J. D., & Johnson, M. K. (1972). Contextual prerequisites for understanding: Some investigations of comprehension and recall. Journal of Verbal Learning and Verbal Behavior, 11,
717726.
Bransford, J. D., & Stein, B. S. (1984). The ideal problem solver: A guide to improving thinking, learning, and creativity. New York: Freeman.
Brewer, M. (1988). A dual process model of impression formation. In T. K. Srull & R. S. Wyer
(Eds.), Advances in social cognition: A dual process model of impression formation (Vol. 1, pp.
136). Hillsdale, NJ: Lawrence Erlbaum Associates.
Brewer, W. F. (1995). To assert that essentially all human knowledge and memory is represented
in terms of stories is certainly wrong. In R. S. Wyer (Ed.), Advances in social cognition (Vol. 8,
pp. 109119). Hillsdale, NJ: Lawrence Erlbaum Associates.
Brewer, W. H., & Lichtenstein, E. H. (1981). Event schemas, story schemas, and story grammars.
In J. Long & A. Baddeley (Eds.), Attention and Performance, IX (pp. 363379). Hillsdale, NJ: Lawrence Erlbaum Associates.
Briley, D. A., Morris, M., & Simonson, I. (2000). Reasons as carriers of culture: Dynamic versus
dispositional models of cultural influence on decision making. Journal of Consumer Research,
27, 157178.
Briley, D. A., & Wyer, R. S. (2001). Transitory determinants of values and decisions: The utility (or
nonutility) of individualism and collectivism in understanding cultural differences. Social
Cognition, 19, 198229.
Briley, D. A., & Wyer, R. S. (2002). The effect of group membership salience on the avoidance of
negative outcomes: Implications for social and consumer decisions. Journal of Consumer Research, 29, 400415.
Brown, N. R., Rips, L. J., & Shevell, S. K. (1985). The subjective dates of natural events in verylong-term memory. Cognitive Psychology, 17, 139177.
Byrne, D. (1971). The attraction paradigm. New York: Academic Press.
Cacioppo, J. T., & Petty, R. E. (1982). The need for cognition. Journal of Personality and Social Psychology, 42, 116131.
Carlston, D. E. (1980). Events, inferences and impression formation. In R. Hastie, T. Ostrom, E.
Ebbesen, R. Wyer, D. Hamilton, & D. Carlston (Eds.), Person memory: The cognitive basis of social perception (pp. 89119). Hillsdale, NJ: Lawrence Erlbaum Associates.
Carlston, D. E. (1994). Associated Systems Theory: A systematic approach to cognitive representations of persons. In R. S. Wyer (Ed.), Advances in social cognition (Vol. 7, pp. 178). Hillsdale,
NJ: Lawrence Erlbaum Associates.
Carlston, D. E., & Smith, E. R. (1996). Principles of mental representation. In E. T. Higgins & A. W.
Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 184210). New York:
Guilford Press.
Ceci, S. J., & Bruck, M. (1993). Suggestibility of the child witness: A historical review and synthesis. Psychological Bulletin, 113, 403439.
Chaiken, S. (1980). Heuristic versus systematic information processing in the use of source versus message cues in persuasion. Journal of Personality and Social Psychology, 39, 752766.
Chaiken, S. (1987). The heuristic model of persuasion. In M. P. Zanna, J. M. Olson, & C. P. Herman
(Eds.), Social influence: The Ontario Symposium (Vol. 5, pp. 339). Hillsdale, NJ: Lawrence
Erlbaum Associates.
Chaiken, S., & Bargh, J. A. (1993). Occurrence versus moderation of the automatic attitude activation effect: Reply to Fazio. Journal of Personality and Social Psychology, 64, 759765.
Chaiken, S., Liberman, A., & Eagly, A. H. (1989). Heuristic and systematic information processing
within and beyond the persuasion context. In J. S. Uleman & J. A. Bargh (Eds.), Unintended
thought (pp. 212252). New York: Guilford Press.
384
REFERENCES
Chaiken, S., & Trope, Y. (1999). Dual-process theories in social psychology. New York: Guilford
Press.
Chaiken, S., Wood, W., & Eagly, A. H. (1996). Principles of persuasion. In E. T. Higgins & A. W.
Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 702742). New York:
Guilford Press.
Chapman, A. J., & Foot, H. C. (Eds.). (1976). Humor and laughter: Theory, research and application.
New York: Wiley.
Chapman, A. J., & Foot, H. C. (Eds.). (1977). Its a funny thing, humor. Elmsford, NY: Pergamon
Press.
Chartrand, T. L., & Bargh, J. A. (1996). Automatic activation of impression formation and memorization goals: Nonconscious goal priming reproduces effects of explicit task instructions.
Journal of Personality and Social Psychology, 71, 464478.
Chartrand, T. L., & Bargh, J. A. (2002). Nonconscious motivations: Their activation, operation
and consequences. In A. Tesser & D. Stapel (Eds.), Self and motivation: Emerging psychological
perspectives (pp. 1341). Washington DC: American Psychological Association.
Chernev, A. (2001). The impact of common features on consumer preferences: A case of confirmatory reasoning. Journal of Consumer Research, 27, 475488.
Chiu, C., Dweck, C. S., Tong, J. Y., & Fu, J. H. (1997). Implicit theories and conceptions of morality.
Chiu, C., Hong, Y., & Dweck, C. S. (1997). Lay dispositionism and implicit theories of personality.
Chodorow, N. (1979). Feminism and difference: Gender, relation and difference in psychoanalytic perspective. Socialist Review, 46, 4264.
Choi, I., Nisbett, R. E., & Norenzayan, A. (1999). Causal attribution across cultures: Variation and
universality. Psychological Bulletin, 125, 4763.
Clore, G. L. (1992). Cognitive phenomenology: Feelings and the construction of judgment. In L. L.
Martin & A. Tesser (Eds.), The construction of social judgments (pp. 133163). Hillsdale, NJ:
Clore, G. L., & Byrne, D. (1974). A reinforcement affect model of attraction. In T. L. Huston (Ed.),
Foundations of interpersonal attraction (pp. 143170). New York: Academic Press.
Clore, G. L., Schwarz, N., & Conway, M. (1994). Affective causes and consequences of social information processing. In R. S. Wyer & T. K. Srull (Eds.), Handbook of social cognition (2nd ed., Vol.
1, pp. 323417). Hillsdale, NJ: Lawrence Erlbaum Associates.
Colcombe, S. J., & Wyer, R. S. (2001). The effects of image-based priming on lexical access, conceptual activation, and behavior. Unpublished manuscript, University of Illinois at UrbanaChampaign.
Colcombe, S. J., & Wyer, R. S. (2002). The role of prototypes in the mental representation of temporally-related events. Cognitive Psychology, 44, 67103.
Collins, A. M., & Loftus, E. F. (1975). A spreading-activation theory of semantic processing. Psychological Review, 82, 407428.
Conway, M., & Ross, M. (1984). Getting what you want by revising what you had. Journal of Personality and Social Psychology, 47, 738748.
Costley, C. L., & Brucks, M. (1992). Selective recall and information use in consumer preferences.
Journal of Consumer Research, 11, 464474.
Craik, F. I. M., & Lockhart, R. S. (1972). Levels of processing: A framework for memory research.
Journal of Verbal Learning and Verbal Behavior, 11, 671684.
Cutrona, C., & Russell, D. (1987). The provisions of social relationships and adaptation to stress.
In W. H. Jones & D. Perlman (Eds.), Advances in personal relationships (Vol. 1, pp. 3767).
Greenwich, CT: JAI.
DePaulo, B. M., & Friedman, H. S. (1998). Nonverbal communication. In D. T. Gilbert, S. T. Fiske, &
G. Lindzey (Eds.), Handbook of social psychology (4th ed., Vol. 2, pp. 340). New York: McGrawHill.
REFERENCES
385
DePaulo, B. M., & Kirkendol, S. E. (1989). The motivational impairment effect in the communication of deception. In J. Yuille (Ed.), Credibility assessment (pp. 5170). Belgium: Kluwer.
Devine, P. G. (1989). Stereotypes and prejudice: Their automatic and controlled components.
Dienes, B. P. A. (1996). Mood as information: Affective cues for cognitive processing styles. Unpublished doctoral dissertation, University of Illinois at Urbana-Champaign.
Dijksterhuis, A., & van Knippenberg, A. (1998). The relation between perception and behavior, or
how to win a game of Trivial Pursuit. Journal of Personality and Social Psychology, 74, 865877.
Duckworth, K. L., Bargh, J. A., Garcia, M., & Chaiken, S. (2002). The automatic evaluation of novel
stimuli. Psychological Science, 13, 514520.
Dweck, C. S. (1975). The role of expectations and attributions in the alleviation of learned helplessness. Journal of Personality and Social Psychology, 31, 674685.
Dweck, C. S. (1991). Self-theories and goals: Their role in motivation, personality, and development. In R. Dienstbier (Ed.), Nebraska Symposium on Motivation: Vol. 38. Perspectives on motivation (pp. 199235). Lincoln: University of Nebraska Press.
Dweck, C. S., Chiu, C., & Hong, Y. (1995a). Implicit theories and their role in judgments and reactions: A world from two perspectives. Psychological Inquiry, 6, 267285.
Dweck, C. S., Chiu, C., & Hong, Y. (1995b). Implicit theories: Elaboration and extension of the
model. Psychological Inquiry, 6, 322333.
Dweck, C. S., Hong, Y., & Chiu, C. (1993). Implicit theories: Individual differences in the likelihood
and meaning of dispositional inference. Personality and Social Psychology Bulletin, 19, 644656.
Dweck, C. S., & Leggett, E. L. (1988). A social-cognitive approach to motivation and personality.
Psychological Review, 25, 109116.
Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. Fort Worth, TX: Harcourt Brace
Jovanovich.
Eagly, A. H., & Chaiken, S. (1998). Attitude structure and function. In D. T. Gilbert, S. T. Fiske, & G.
Lindzey (Eds.), Handbook of social psychology (4th ed., Vol. 1, pp. 269322). Boston, MA:
McGraw-Hill.
Ebbesen, E. B. (1980). Cognitive processes in understanding ongoing behavior. In R. Hastie, T.
Ostrom, E. Ebbesen, R. Wyer, D. Hamilton, & D. Carlston (Eds.), Person memory: Cognitive basis of social perception (pp. 179226). Hillsdale, NJ: Lawrence Erlbaum Associates.
Edell, J. A., & Staelin, R. (1983). The information processing of pictures in print advertisements.
Ellis, H. C., & Ashbrook, P. W. (1988). Resource allocation model of the effects of depressed mood
states on memory. In K. Fiedler, & J. Forgas (Eds.), Affect, cognition, and social behavior (pp.
2543). Toronto: C. J. Hofgrefe.
Ellsworth, P. C., & Carlsmith, M. (1968). Effects of eye contact and verbal content on affective response to a dyadic interaction. Journal of Personality and Social Psychology, 10, 1520.
Emler, N., Renwick, S., & Malone, B. (1983). The relationship between moral reasoning and political orientation. Journal of Personality and Social Psychology, 45, 10731080.
Englis, B. G. (1994). The role of affect in political advertising: Voter emotional responses to the
nonverbal behavior of politicians. In E. M. Clark et al. (Eds.), Attention, attitude, and affect in response to advertising (pp. 223247). Hillsdale, NJ: Lawrence Erlbaum Associates.
Erdley, C. A., & Dweck, C. S. (1993). Childrens implicit personality theories as predictors of their
social judgments. Child Development, 64, 863878.
Fazio, R. H. (1986). How do attitudes guide behavior? In R. M. Sorrentino & E. T. Higgins (Eds.),
Handbook of motivation and cognition (pp. 204243). New York: Guilford.
Fazio, R. H. (1990). Multiple processes by which attitudes guide behavior: The MODE model as
an integrative framework. In M. P. Zanna (Ed.), Advances in experimental social psychology
(Vol. 23, pp. 75109). San Diego, CA: Academic Press.
Fazio, R. H., Sanbonmatsu, D. M., Powell, M. C., & Kardes, F. R. (1986). On the automatic activation
of attitudes. Journal of Personality and Social Psychology, 50, 229238.
386
REFERENCES
Festinger, L. (1957). A theory of cognitive dissonance. Palo Alto, CA: Stanford University Press.
Fiedler, K. (1988). Emotional mood, cognitive style, and behavior regulation. In K. Fiedler & J.
Forgas (Eds.), Affect, cognition and social behavior (pp. 100119). Toronto: Hofgrefe International.
Fischoff, B. (1975). Hindsight foresight: The effect of outcome knowledge on judgment under
uncertainty. Journal of Experimental Psychology: Human Perception and Performance, 1,
288299.
Fischoff, B. (1982). For those condemned to study the past: Heuristics and biases in hindsight. In
D. Kahneman, P. Slovic, & A. Tversky (Eds.), Judgment under uncertainty: Heuristics and biases
(pp. 332351). New York: Cambridge University Press.
Fishbein, M. (1963). An investigation of the relationships between beliefs about an object and attitude toward that object. Human Relations, 16, 233239.
Fishbein, M. (1967). A behavior theory approach to the relations between beliefs about an object
and the attitude toward the object. In M. Fishbein (Ed.), Readings in attitude theory and measurement (pp. 389400). New York: Wiley.
Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to theory
and research. Reading, MA: Addison-Wesley.
Fishbein, M., & Hunter, R. (1964). Summation versus balance in attitude organization and
change. Journal of Abnormal and Social Psychology, 69, 505510.
Fishkin, J., Keniston, K., & MacKinnon, C. (1973). Moral reasoning and political ideology. Journal
of Personality and Social Psychology, 27, 109119.
Fiske, S. T. (1998). Stereotyping, prejudice, and discrimination. In D. T. Gilbert, S. T. Fiske, & G.
Lindzey (Eds.), Handbook of social psychology (4th ed., Vol. 2, pp. 357414). New York: McGraw
Hill.
Fiske, S. T., & Pavelchak, M. A. (1986). Category-based versus piecemeal-based affective responses: Developments in schema-triggered affect. In R. M. Sorrentino & E. T. Higgins (Eds.),
Handbook of motivation and cognition (pp. 167203). New York: Guilford.
Fletcher, G. J. O., Danilovics, P., Fernandez, G., Peterson, D., & Reeder, G. D. (1986). Attributional
complexity: An individual differences measure. Journal of Personality and Social Psychology,
51, 875884.
Fong, C. P. S., & Wyer, R. S. (2003). Cultural, social and emotional determinants of decisions under uncertainty. Organizational Behavior and Human Decision Processes,
Forgas, J. P. (1991). Affect and cognition in close relationships. In G. J. O. Fletcher & F. D. Fincham
(Eds.), Cognition in close relationships (pp. 735). Hillsdale, NJ: Lawrence Erlbaum Associates.
Forgas, J. P. (1992). Affect in social judgments and decisions: a multiprocess model. In M. P.
Zanna (Ed.), Advances in Experimental Social Psychology (Vol. 25, pp. 227275). San Diego, CA:
Academic Press.
Forgas, J. P. (1995). Mood and judgment: The affect infusion model (AIM). Psychological Bulletin,
117, 3966.
Forgas, J. P. (1998). Asking nicely? The effects of mood on responding to more or less polite requests. Personality and Social Psychology Bulletin, 2, 173185.
Forgas, J. P. (2001). Feeling and doing: Affective influences on interpersonal behavior. Psychological Inquiry, 13, 128.
Forgas, J. P. (2002). Feeling and doing: Affective influences on interpersonal behavior. Psychological Inquiry, 13, 128.
Forgas, J. P., & Bower, G. H. (1987). Mood effects on person perception judgments. Journal of Personality and Social Psychology, 53, 5360.
Forgas, J. P., Bower, G. H., & Krantz, S. (1984). The influence of mood on perceptions of social interactions. Journal of Experimental Social Psychology, 20, 497513.
Freud, S. (1928). Humor. International Journal of Psychoanalysis, 9, 16.
Freud, S. (1960/1905). Jokes and their relation to the unconscious. New York: Norton.
REFERENCES
387
Fry, P. S., & Ghosh, R. (1980). Attributions of success and failure: Comparison of cultural differences between Asian and Caucasian children. Journal of Cross-Cultural Psychology, 11,
343363.
Gaelick, L., Bodenhausen, G. V., & Wyer, R. S. (1985). Emotional communication in close relationships. Journal of Personality and Social Psychology, 49, 12461265.
Garnham, A. (1981). Mental models as representations of text. Memory & Cognition, 9, 560565.
Gasper, K., & Clore, G. L. (1998). The persistent use of negative affect by anxious individuals to
estimate risk. Journal of Personality and Social Psychology, 74, 13501363.
Gerbner, G., Gross, L., Morgan, M., & Signorielli, N. (1994). Growing up with television: the cultivation perspective. In J. Bryant & D. Zillmann (Eds.), Media effects: Advances in theory and research (pp. 1741). Hillsdale, NJ: Lawrence Erlbaum Associates.
Gick, M., L., & Holyoak, K. J. (1983). Schema induction and analogical transfer. Cognitive Psychology, 15, 138.
Gilbert, D. T. (1991). How mental systems believe. American Psychologist, 46, 107119.
Gilligan, C. (1982). In a different voice: Psychological theory and womens development. Cambridge,
MA: Harvard University Press.
Giner-Sorolla, R., & Chaiken, S. (1997). Selective use of heuristic and systematic processing under defense motivation. Personality and Social Psychology Bulletin, 23, 8497.
Glenberg, A. M. (1997). What memory is for. Behavioral & Brain Sciences, 20, 155.
Glenberg, A. M., Kruley, P., & Langston, W. E. (1994). Analogical processes in comprehension:
Simulation of a mental model. In M. A. Gernsbacher (Ed.), Handbook of psycholinguistics (pp.
609640). San Diego, CA: Academic Press.
Glenberg, A. M., Meyer, M., & Lindem, K. (1987). Mental models contribute to foregrounding during text comprehension. Journal of Memory and Language, 26, 6983.
Goethals, G. R., & Reckman, R. F. (1973). The perception of consistency in attitudes. Journal of Experimental Social Psychology, 9, 491501.
Gohm, C. L., & Wyer, R. S. (1998). Narrative representations of long-term relationships: An evaluation
of assessment procedures and implications for marital satisfaction. Unpublished manuscript,
University of Illinois at Urbana-Champaign.
Goldstein, J. H., & McGhee, P. (Eds.). (1972). The psychology of humor. New York: Springer-Verlag.
Goldstein, J. H., Suls, J. M., & Anthony, S. (1972). Enjoyment of specific types of humor content:
Motivation or salience. In J. Goldstein & P. McGhee (Eds.), Its a funny thing, humor (pp.
159171). New York: Springer-Verlag.
Gordon, S. E., & Wyer, R. S. (1987). Person memory: Category-set-size effects on the recall of a
persons behaviors. Journal of Personality and Social Psychology, 53, 648662.
Graesser, A. C. (1981). Prose comprehension beyond the word. New York: Springer-Verlag.
Graesser, A. C., Gordon, S. E., & Sawyer, J. D. (1979). Memory for typical and atypical actions in
scripted activities: Test of a script pointer + tag hypothesis. Journal of Verbal Learning and
Verbal Behavior, 18, 319322.
Graesser, A. C., & Nakamura, G. V. (1982). The impact of a schema on comprehension and memory. In G. H. Bower (Ed.), The psychology of learning and motivation: Advances in research and
theory (Vol. 16, pp. 66109). New York: Academic Press.
Graesser, A. C., Singer, M., & Trabasso, T. (1994). Constructing inferences during narrative text
comprehension. Psychological Review, 101, 371395.
Green, G. M. (1989). Pragmatics and natural language understanding. Hillsdale, NJ: Erlbaum.
Green, M. C., & Brock, T. C. (2000). The role of transportation in the persuasiveness of public narratives. Journal of Personality and Social Psychology, 79, 701721.
Greenwald, A. G. (1968). Cognitive learning, cognitive responses to persuasion and attitude
change. In A. G. Greenwald, T. C. Brock, & T. M. Ostrom (Eds.), Psychological foundations of attitudes (pp. 147170). New York: Academic Press.
Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: Attitudes, self-esteem, and stereotypes. Psychological Review, 102, 427.
388
REFERENCES
Greenwald, A. G., & Pratkanis, A. R. (1984). The self. In R. S. Wyer & T. K. Srull (Eds.), Handbook of
social cognition (Vol. 3, pp. 129178). Hillsdale, NJ: Lawrence Erlbaum Associates.
Grice, H. P. (1975). Logic and conversation. In P. Cole & J. L. Morgan (Eds.), Syntax and semantics:
Speech acts (pp. 4158). New York: Academic Press.
Gruenfeld, D. H., & Wyer, R. S. (1992). Semantics and pragmatics of social influence: How affirmations and denials affect beliefs in referent propositions. Journal of Personality and Social Psychology, 62, 3849.
Gruenfeld, D., & Wyer, R. S. (1993). A two-set flow model of social communication: The influence of
information source and communication objectives on memory, message content and delivery,
and recipients beliefs. Unpublished manuscript, University of Illinois at Urbana-Champaign.
Hamilton, D. L., & Gifford, R. K. (1976). Illusory correlation in interpersonal perception: A cognitive basis of stereotypic judgment. Journal of Experimental Social Psychology, 13, 392407.
Hamilton, D. H., Katz, L. B., & Leirer, V. O. (1980). Organizational processes in impression formation. In R. Hastie, T. Ostrom, E. Ebbesen, R. Wyer, D. Hamilton, & D. Carlston (Eds.), Person
memory: The cognitive basis of social perception (pp. 121153). Hillsdale, NJ: Lawrence
Erlbaum Associates.
Hamilton, D. L., & Sherman, J. W. (1994). Stereotypes. In R. S. Wyer & T. K. Srull (Eds.), Handbook
of social cognition (2nd ed., Vol. 2, pp. 168). Hillsdale, NJ: Lawrence Erlbaum Associates.
Harkins, S. G., Latane, B., & Williams, K. (1980). Social loafing: Allocating effort or taking it easy?
Journal of Experimental Social Psychology, 16, 457465.
Harvey, J. H., & Weary, G. (1984). Current issues in attribution theory and research. Annual Review of Psychology, 35, 427459.
Harvey, J. H., Weber, A. L., & Orbuch, T. L. (1990). Interpersonal accounts. Cambridge, MA: Basil
Blackwell.
Hasher, L., Goldstein, D., & Toppin, T. (1977). Frequency and the conference of referential validity. Journal of Verbal Learning and Verbal Behavior, 16, 107112.
Hastie, R. (1980). Memory for behavioral information that confirms or contradicts a personality
impression. In R. Hastie, T. Ostrom, E. Ebbesen, R. Wyer, D. Hamilton, & D. Carlston (Eds.),
Person memory: The cognitive basis of social perception (pp. 141172). Hillsdale, NJ: Lawrence
Erlbaum Associates.
Hastie, R., & Kumar, P. A. (1979). Person memory: Personality traits as organizing principles in
memory for behaviors. Journal of Personality and Social Psychology, 37, 2538.
Hastie, R., & Pennington, N. (1995). The big picture: Is it a story? In R. S. Wyer (Ed.), Advances in
social cognition (Vol. 8, pp. 133138). Mahwah, NJ: Lawrence Erlbaum Associates.
Hawkins, S. A., & Hastie, R. (1990). Hindsight: Biased judgment of past events after the outcomes
are known. Psychological Bulletin, 107, 311327.
Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley.
Heine, S. J., & Lehman, D. R. (1997). Culture, dissonance, and self-affirmation. Personality and Social Psychology Bulletin, 23, 389400.
Heine, S., Lehman, D., Markus, H., & Kitayama, S. (1999). Is there a universal need for positive
self-regard? Psychological Review, 106, 766794.
Henninger, M., & Wyer, R. S. (1976). The recognition and elimination of inconsistencies among
syllogistically-related beliefs: Some new light on the Socratic effect. Journal of Personality
and Social Psychology, 34, 680693.
Herr, P. M. (1986). Consequences of priming: Judgment and behavior. Journal of Personality and
Higgins, E. T. (1981). The communication game: Implications for social cognition and persuasion. In E. T. Higgins, C. P. Herman, & M. P. Zanna (Eds.), Social cognition: The Ontario Symposium (Vol. 1, pp. 342392). Hillsdale, NJ: Lawrence Erlbaum Associates.
Higgins, E. T. (1996). Knowledge activation: Accessibility, applicability, and salience. In E. T. Higgins & A. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 133168). New
York: Guilford.
REFERENCES
389
Higgins, E. T. (1997). Beyond pleasure and pain. American Psychologist, 55, 12171233.
Higgins, E. T. (1998). Promotion and prevention: Regulatory focus as a motivational principle. In
M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 30, pp. 146). San Diego,
CA: Academic Press.
Higgins, E. T., Bargh, J. A., & Lombardi, W. (1985). The nature of priming effects on categorization. Journal of Experimental Psychology: Learning, Memory, and Cognition, 11, 5969.
Higgins, E. T., & Lurie, L. (1983). Context, categorization and recall: The change-of-standard effect. Cognitive Psychology, 15, 525547.
Higgins, E. T., & Rholes, W. S. (1978). Saying is believing: Effects of message modification on
memory and liking for the person described. Journal of Experimental Social Psychology, 14,
363378.
Higgins, E. T., Rholes, W. S., & Jones, C. R. (1977). Category accessibility and impression formation. Journal of Experimental Social Psychology, 13, 141154.
Hilgard, E. R. (1951). Methods and procedures in the study of learning. In S. S. Stevens (Ed.),
Handbook of experimental psychology. New York: Wiley.
Hintzman, D. L. (1986). Schema abstraction in a multiple-trace model. Psychological Review, 93,
411428.
Ho, D. Y. F., & Chiu, C. Y. (1994). Component ideas of individualism, collectivism, and social organization: An application in the study of Chinese culture. In U. Kim, C. Kagitcibasi, S. C. Choi, &
G. Yoon (Eds.), Individualism and collectivism: theory, method and applications (pp. 137156).
Thousand Oaks, CA: Sage.
Hofstede, G. H. (1991). Empirical models of cultural differences. In N. Bleichrodt & P. J. D. Drenth
(Eds.), Contemporary issues in cross-cultural psychology (pp. 420). Amsterdam, Netherlands:
Swets & Zeitlinger.
Holmberg, D., & Holmes, J. G. (1994). Reconstruction of relationship memories: A mental models
approach. In N. Schwarz & S. Sudman (Eds.), Autobiographical memory and the validity of retrospective reports (pp. 267288). New York: Springer-Verlag.
Holmes, J. G., & Murray, S. L. (1995). Memory for events in close relationships: Applying Schank
and Abelsons story skeleton model. In R. S. Wyer (Ed.), Advances in social cognition (Vol. 8,
Hong, Y., Chiu, C., Dweck, C. S., & Sacks, R. (1997). Implicit theories and evaluative processes in
person cognition. Journal of Experimental Social Psychology, 33, 296303.
Hong, Y., Morris, M. W., Chiu, C., & Benet-Martinez, V. (2000). Multicultural minds: A dynamic
constructivist approach to culture and cognition. American Psychologist, 55, 709720.
Humphreys, M. S., & Kashima, Y. (2002). Connectionism and self: Distributed representational
systems and their implications for self and identity. In Y. Kashima, M. Foddy, & M. Platow
(Eds.), Self and identity: Personal, social and symbolic (pp. 2754). Mahwah, NJ: Lawrence
Erlbaum Associates.
Hunt, R. R., & Einstein, G. O. (1981). Relational and item-specific information in memory. Journal
of Verbal Learning and Verbal Behavior, 20, 497514.
Hyman, I. E., & Loftus, E. F. (2002). False childhood memories and eyewitness memory errors. In
M. Eisen & J. Quas (Eds.), Memory and suggestibility in the forensic interview (pp. 6384).
Mahwah, NJ: Lawrence Erlbaum Associates.
Innes-Ker, A., & Niedenthal, P. M. (2002). Emotion concepts and emotional states in social judgment and categorization. Journal of Personality and Social Psychology, 83, 804816.
Isbell, L. M. (2000). Beyond heuristic information processing: Systematic processing in happy and sad
moods. Unpublished doctoral dissertation, University of Illinois at Urbana-Champaign.
Isbell, L. M., Clore, G. L., & Wyer, R. S. (1998). Mood-mediated uses of stereotyping in impression formation. Unpublished manuscript, University of Illinois at Urbana-Champaign.
Isbell, L. M., & Wyer, R. S. (1999). Correcting for mood-induced bias in the evaluation of political
candidates. Personality and Social Psychology Bulletin, 25, 237249.
390
REFERENCES
Isbell, L. M., Wyer, R. S., & Collins, J. E. (2002). The cognitive dynamics of humor elicitation: The role
of diminishment, disparagement and cognitive elaboration in responses to jokes and witticisms.
Unpublished manuscript, Hong Kong University of Science and Technology, Hong Kong.
Isen, A. M. (1984). Toward understanding the role of affect in cognition. In R. S. Wyer & T. K. Srull
(Eds.), Handbook of social cognition (Vol. 3, pp. 179236). Hillsdale, NJ: Lawrence Erlbaum Associates.
Isen, A. M. (1987). Positive affect, cognitive processes, and social behavior. In L. Berkowitz (Ed.),
Advances in experimental social psychology (Vol. 20, pp. 203253). New York: Academic Press.
Isen, A. M., & Daubman, L. A. (1984). The influence of affect on categorization. Journal of Personality and Social Psychology, 47, 12061217.
Isen, A. M., Daubman, L. A., & Nowicki, G. P. (1987). Positive affect facilitates creative problem
solving. Journal of Personality and Social Psychology, 52, 11221131.
Isen, A. M., Johnson, M., Mertz, E., & Robinson, G. (1985). Positive affect and the uniqueness of
word association. Journal of Personality and Social Psychology, 48, 14131426.
Isen, A. M., Shalker, T. E., Clark, M. S., & Karp, L. (1978). Affect, accessibility of material in memory and behavior: A cognitive loop? Journal of Personality and Social Psychology, 36, 112.
Jacoby, L. L., Kelley, C. M., Brown, J., & Jasechko, J. (1989). Becoming famous overnight: Limits
on the ability to avoid unconscious influences of the past. Journal of Personality and Social
Janis, I. L., & King, B. T. (1954). The influence of role playing on opinion change. Journal of Abnormal and Social Psychology, 49, 211218.
Johnson, M., Hashtroudi, S., & Lindsay, D. S. (1993). Source monitoring. Psychological Bulletin,
114, 328.
Johnson-Laird, P. N. (1983). Mental models: Towards a cognitive science of language, inference and
consciousness. Cambridge, MA: Harvard University Press.
Johnson-Laird, P. N. (1989). Mental models. In M. I. Posner (Ed.), Foundations of cognitive science
(pp. 469500). Cambridge, MA: MIT Press.
Jones, E. E., & Davis, K. J. (1965). From acts to dispositions: The attributional process in person
perception. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 2, pp.
220266). New York: Academic Press.
Kahneman, D., & Miller, D. T. (1986). Norm theory: Comparing reality to its alternatives. Psychological Review, 93, 136153.
Kahneman, D., Slovic, P., & Tversky, A. (Eds.). (1982). Judgment under uncertainty: Heuristics and
biases. New York: Cambridge University Press.
Kahneman, D., & Tversky, A. (1982). The simulation heuristic. In D. Kahneman, P. Slovic, & A.
Tversky (Eds.), Judgment under uncertainty: Heuristics and biases (pp. 201208). New York:
Cambridge University Press.
Kane, T. R., Suls, J., & Tedeschi, J. T. (1977). Humor as a tool of social interaction. In A. J. Chapman & H. C. foot (Eds.), Its a funny thing, humor. Elmsford, NY: Pergamon Press.
Kelley, C. M., & Lindsay, D. S. (1993). Remembering mistaken for knowing: Ease of retrieval as a
basis for confidence in answers to general knowledge questions. Journal of Memory and Language, 32, 124.
Kelley, H. H. (1967). Attribution theory in social psychology. In D. Levine (Ed.), Nebraska symposium on motivation, 15, 192238.
Kelley, H. H. (1987). Causal schemata and the attribution process. In E. Jones, D. Kanouse, H.
Kelley, R. Nisbett, S. Valins, & B. Winer (Eds.), Attribution: Perceiving the causes of behavior.
Hillsdale, NJ: Lawrence Erlbaum Associates. (Originally published 1971).
Kihlstrom, J. F. (1980). Posthypnotic amnesia for recently learned material: Interactions with
episodic and semantic memory. Cognitive Psychology, 12, 227251.
Kintsch, W. (1998). Comprehension: A paradigm for cognition. Cambridge, England: Cambridge
University Press.
REFERENCES
391
Klein, S. B., & Loftus, J. (1993). Behavioral experience and trait judgments about the self. Personality and Social Psychology Bulletin, 19, 740745.
Klein, S. B., Loftus, J., Trafton, J. G., & Fuhrman, R. W. (1992). Use of exemplars and abstractions
in trait judgments: A model of trait knowledge about the self and others. Journal of Personality
and Social Psychology, 63, 739753.
Klinger, E. (1975). Consequences of commitment to and disengagement from incentives. Psychological Bulletin, 82, 223231.
Koestler, A. (1964). The act of creation. New York: Macmillan.
Kohlberg, L. (1976). Moral stages and moralization: The cognitive-developmental approach. In T.
Lickona (Ed.), Moral development and behavior: Theory, research and social issues. New York:
Holt, Rinehart, & Winston.
Kosslyn, S. M. (1980). Image and mind. Cambridge, MA: Harvard University Press.
Kunda, Z. (1987). Motivated inference: Self-serving generation and evaluation of causal theories.
Kunda, Z. (1990). The case for motivated reasoning. Psychological Bulletin, 108, 480498.
LaFave, L., Haddad, J., & Maesen, W. A. (1976). Superiority, enhanced self-esteem and perceived
incongruity humor theory. In A. J. Chapman & H. C. Foot (Eds.), Humor and laughter: Theory,
research and applications (pp. 6391). New York: Wiley.
Lampel, A. K., & Anderson, N. H. (1968). Combining visual and verbal information in an impression-formation task. Journal of Personality and Social Psychology, 9, 16.
Langer, E. J., & Abelson, R. P. (1972). The semantics of asking a favor: How to succeed in getting
help without really dying. Journal of Personality and Social Psychology, 24, 2632.
LaPiere, R. T. (1934). Attitudes vs. actions. Social Forces, 13, 230237.
Lazarus, R. S. (1982). Thoughts on the relations between emotion and cognition. American Psychologist, 37, 10191024.
Lazarus, R. S. (1991). Emotion and adaptation. New York: Oxford University Press.
LeDoux, J. E. (1994). Memory vs. emotional memory in the brain. in P. Ekman & R. J. Davidson
(Eds.), The nature of emotion: Fundamental questions (pp. 311312). New York: Oxford University Press.
Lefcourt, H. M., & Martin, R. A. (1986). Humor and life stress: Anecdote to adversity. New York:
Springer-Verlag.
Lepore, L., & Brown, R. (1997). Category and stereotype activation: Is prejudice inevitable? Journal of Personality and Social Psychology, 72, 275287.
Lerner, M. J. (1970). The desire for justice and reactions to victims. In J. Macaulay & L. Berkowitz
(Eds.), Altruism and helping behavior (pp. 205229). New York: Academic Press.
Lerner, M. J., & Miller, D. T. (1978). Just world research and the attribution process: Looking
back and ahead. Psychological Bulletin, 85, 10301051.
Lerner, M. J., Miller, D. T., & Holmes, J. G. (1976). Deserving and the emergence of forms of justice. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 9). New York: Academic Press.
Lerner, M. J., & Simmons, C. H. (1966). Observers reaction to the innocent victim: Compassion
or rejection? Journal of Personality and Social Psychology, 4, 203210.
Levine, S. R., Wyer, R. S., & Schwarz, N. (1994). Are you what you feel? The affective and cognitive
determinants of self-judgments. Experimental Journal of Social Psychology, 24, 6377.
Locke, H. L., & Wallace, K. (1959). Short marital adjustment and prediction tests: Their reliability
and validity. Marriage and Family Living, 21, 251255.
Loftus, E. F. (1975). Leading questions and the eyewitness report. Cognitive Psychology, 7, 560
572.
Loftus, E. F. (2000). Remembering what never happened. In E. Tulving (Ed.), Memory, consciousness and the brain: The Tallinn conference (pp. 106118). Philadelphia, PA: Taylor & Francis.
392
REFERENCES
Loftus, E. F., & Marburger, W. (1983). Since the eruption of Mt. St. Helens, has anyone beaten you
up? Improving the accuracy of retrospective reports with landmark events. Memory & Cognition, 11, 114120.
Loftus, E. F., & Palmer, S. J. (1974). Reconstruction of automobile destruction. Journal of Verbal
Learning and Verbal Behavior, 2, 467471.
Logan, G. L. (1988). Toward an instance theory of automatization. Psychological Review, 95,
492527.
Loken, B. A. (1984). Attitude processing strategies. Journal of Experimental Social Psychology, 20,
272296.
Loken, B. A., & Wyer, R. S. (1983). Effects of reporting beliefs in syllogistically related propositions on the recognition of unmentioned propositions. Journal of Personality and Social Psychology, 45, 306322.
Lombardi, W. J., Higgins, E. T., & Bargh, J. A. (1987). The role of consciousness in priming effects
on categorization. Personality and Social Psychology Bulletin, 13, 411429.
Long, D. L., & Graesser, A. C. (1988). Wit and humor in discourse processing. Discourse Processes,
11, 3560.
Loomes, G., & Sugden, R. (1986). Disappointment and dynamic consistency in choice under uncertainty. Review of Economic Studies, 53, 271282.
Lynch, J. G., Chakravarti, D., & Mitra, A. (1991). Contrast effects in consumer judgments: Changes
in mental representations or in the anchoring of rating scales? Journal of Consumer Research,
18, 284297.
Mackie, D. M., & Worth, L. T. (1989). Cognitive deficits and the mediation of positive affect in persuasion. Journal of Personality and Social Psychology, 57, 2740.
Macrae, C. N., Bodenhausen, G. V., Milne, A. B., & Jetten, J. (1994). Out of mind but back in sight:
Stereotypes on the rebound. Journal of Personality and Social Psychology, 67, 808817.
Magliano, J. P., Dijkstra, K., & Zwaan, R. A. (1996). Generating predictive inferences while viewing
a movie. Discourse Processes, 22, 199224.
Maheswaran, D., Mackie, D. M., & Chaiken, S. (1992). Brand name as a heuristic cue: The effects of
task importance and expectancy confirmation on consumer judgments. Journal of Consumer
Mandler, J. (1979). Categorical and schematic organization in memory. In C. R. Puff (Ed.), Memory, organization and structure. New York: Academic Press.
Mandler, J., & Johnson, N. S. (1977). Remembrance of things parsed: Story structure and recall.
Cognitive Psychology, 9, 111151.
Mankowski, E., & Rappaport, J. (1995). Stories, identity, and the psychological sense of community. In R. S. Wyer (Ed.), Advances in social cognition (Vol. 8, pp. 211226). Hillsdale, NJ: Lawrence Erlbaum Associates.
Mankowski, E. S., & Wyer, R. S. (1996). Cognitive processes in perceptions of social support. Personality and Social Psychology Bulletin, 22, 894905.
Markus, H., & Kitayama, S. (1991). Culture and the self: Implications for cognition, emotion and
motivation. Psychological Review, 98, 224253.
Martin, L. L., Achee, J. W., Ward, D. W., & Wyer, R. S. (1993). Mood as input: People have to interpret the motivational implications of their moods. Journal of Personality and Social Psychology, 64, 317326.
Martin, L. L., Seta, J. J., & Crelia, R. A. (1990). Assimilation and contrast as a function of peoples
willingness and ability to expend effort in forming an impression. Journal of Personality and
Martin, L. L., & Tesser, A. (1996). Some ruminative thoughts. In R. S. Wyer (Ed.), Ruminative
thoughts: Advances in social cognition (Vol. 9, pp. 148). Mahwah, NJ: Lawrence Erlbaum Associates.
McAdams, D. (1988). Power, intimacy and the life story: Personalogical inquiries into identity. New
York: Guilford Press.
REFERENCES
393
McClelland, D. C., Atkinson, J. W., Clark, R. A., & Lowell, E. L. (1953). The achievement motive. New
York: Appleton-Century-Crofts.
McGhee, P., & Goldstein, J. (Eds.). (1983). Handbook of humor research: Basic issues. New York:
Springer-Verlag.
McGuire, W. J. (1960). A syllogistic analysis of cognitive relationships. In M. J. Rosenberg & C. I.
Hovland (Eds.), Attitude organization and change (pp. 140162). New Haven, CT: Yale University Press.
McGuire, W. J. (1972). Attitude change: An information processing paradigm. In C. G. McClintock
(Ed.), Experimental social psychology (pp. 108141). New York: Holt, Rinehart, and Winston.
McGuire, W. J. (1981). The probabilogical model of cognitive structure and attitude change. In
R. E. Petty, T. M. Ostrom, & T. C. Brock (Eds.), Cognitive responses in persuasion (pp. 291307).
Hillsdale, NJ: Lawrence Erlbaum Associates.
McGuire, W. J. (1999). Constructing social psychology: Creative and critical processes. New York:
McGuire, W. J., & McGuire, C. V. (1988). Content and process in the experience of self. In L.
Berkowitz (Ed.), Advances in experimental social psychology (Vol. 21, pp. 97144). San Diego,
CA: Academic Press.
McGuire, W. J., & McGuire, C. V. (1991). The content, structure and operation of thought systems.
In R. S. Wyer & T. K. Srull (Eds.), Advances in social cognition (Vol. 4, pp. 178). Hillsdale, NJ:
Mellers, B., Schwartz, A., & Ritov, I. (1999). Emotion-based choice. Journal of Experimental Psychology: General, 128, 332345.
Menon, G., & Raghubir, P. (1998). When automatic accessibility meets conscious content: Implications for judgment formation. Unpublished manuscript, New York University.
Miller, D. T. (1977). Altruism and threat to a belief in a just world. Journal of Experimental Social
Miller, L. C., & Read, S. R. (1991). On the coherence of mental models of persons and relationships: A knowledge structure approach. In G. J. O. Fletcher & F. D. Fincham (Eds.), Cognition
in close relationships (pp. 6999). Hillsdale, NJ: Lawrence Erlbaum Associates.
Miller, P. J. (1994). Narrative practices: Their role in socialization and self-construction. In U.
Neisser & R. Fivush (Eds.), The remembering self: Construction and accuracy in the self-narrative. Emory symposia in cognition (Vol. 6, pp. 158179). New York: Cambridge University Press.
Miller, P. J. (1995). Personal storytelling in everyday life: Social and cultural perspectives. In R. S.
Wyer (Ed.), Advances in social cognition: Knowledge and memory (Vol. 8, pp. 177184). Mahwah, NJ: Lawrence Erlbaum Associates.
Miller, P. J., Fung, H., & Mintz, J. (1996). Self-construction through narrative practices: A Chinese
and American comparison of early socialization. Ethos, 24, 237280.
Miller, P. J., Sandel, T. L., Liang, C.-H., & Fung, H. (2001). Narrating transgressions in Longwood:
The discourses, meanings and paradoxes of an American socializing practice. Ethos, 29,
159186.
Miller, P. J., Wiley, A. R., Fung, H., & Liang, C. H. (1997). Personal storytelling as a medium of socialization in Chinese and American families. Child Development, 68, 557568.
Mindess, H., Miller, C., Turek, J., Bender, A., & Corbin, S. (1985). The Antioch Humor test: Making
sense of humor. New York: Avon Books.
Miniard, P. W., Bhatla, S., Lord, K. R., Dickson, P. R., & Unnava, H. R. (1991). Picture-based persuasion processes and the moderating role of involvement. Journal of Consumer Research, 18,
82107.
Miniard, P. W., & Cohen, J. B. (1981). An examination of the Fishbein-Ajzen behavioral intentions
models concepts and measures. Journal of Experimental Social Psychology, 17, 309339.
Morris, M. W., & Peng, K. (1994). Culture and cause: American and Chinese attributions for social
and physical events. Journal of Personality and Social Psychology, 67, 949971.
394
REFERENCES
Moskowitz, G. B., & Skurnik, I. W. (1999). Contrast effects as determined by the type of prime:
Trait versus exemplar primes initiate processing strategies that differ in how accessible constructs are used. Journal of Personality and Social Psychology, 76, 911927.
Mueller, J. H., Grove, T. R., & Thompson, W. B. (1991). Mood-dependent retrieval and mood
awareness. Cognition and Emotion, 5, 331349.
Murray, N., Sujan, H., Hirt, E. R., & Sujan, M. (1990). The influence of mood on categorization: A
cognitive flexibility interpretation. Journal of Personality and Social Psychology, 59, 411425.
Murray, S. L., & Holmes, J. G. (1996). The construction of relationship realities. In G. J. O. Fletcher
& J. Fitness (Eds.), Knowledge structures in close relationships: A social psychological approach
(pp. 91120). Mahwah, NJ: Lawrence Erlbaum Associates.
Mussweiler, T. (2003). Comparison processes in social judgment: Mechanisms and consequences. Psychological Review,
Mussweiler, T., & Strack, F. (1999a). Hypothesis-consistent testing and semantic priming in the
anchoring paradigm: A selective accessibility model. Journal of Experimental Social Psychology, 35, 136164.
Mussweiler, T., & Strack, F. (1999b). Comparing is believing: A selective accessibility model of
judgmental anchoring. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 10, pp. 135167). Chichester, England: Wiley.
Mussweiler, T., & Strack, F. (2000a). The use of category and exemplar knowledge in the solution
of anchoring tasks. Journal of Personality and Social Psychology, 78, 10381052.
Mussweiler, T., & Strack, F. (2000b). The relative self: Informational and judgmental consequences of comparative self-evaluation. Journal of Personality and Social Psychology, 79,
2338.
Nassi, A. J., Abramowitz, S. I., & Youmans, J. E. (1983). Moral development and politics a decade
later: A replication and extension. Journal of Personality and Social Psychology, 45, 11271135.
Nehrhardt, G. (1976). Incongruity and funniness: Toward a new descriptive model. In A. J. Chapman & H. C. Foot (Eds.), Humor and laughter: Theory, research and application (pp. 5562). New
York: Wiley.
Nelson, K. (Ed.). (1989). Stories from the crib. Cambridge, MA: Harvard University Press.
Nelson, K. (1993). The psychological and social origins of autobiographical memory. Psychological Science, 4, 714.
Nelson, K., & Greundel, J. (1981). Generalized event representations: Building blocks of cognitive
development. In M. Lamb & A. Brown (Eds.), Advances in developmental psychology (Vol. 1,
Newtson, D. A. (1973). Attribution and the unit of perception of ongoing behavior. Journal of Personality and Social Psychology, 28, 2838.
Newtson, D. A. (1976). Foundations of attribution: The perception of ongoing behavior. In J.
Harvey, W. Ickes, & R. Kidd (Eds.), New directions in attribution research (Vol. 1, pp. 223247).
Niedenthal, P. M., Halberstadt, J. B., & Setterlund, M. B. (1997). Being happy and seeing happy:
Emotional state mediates visual word recognition. Cognition and Emotion, 11, 403432.
Niedenthal, P. M., & Setterlund, M. B. (1994). Emotion congruence in perception. Personality and
Social Psychology Bulletin, 20, 401411.
Nielsen, A. C. (1995). Nielsen report on television. Northbrook, IL: A.C. Nielsen.
Nisbett, R. E., & Ross, L. (1980). Human inference: Strategies and shortcomings of social judgments.
Englewood Cliffs, NJ: Prentice Hall.
Norenzayan, A., & Kim, B. (2000). Unpublished data.
Norenzayan, A., & Schwarz, N. (1999). Telling what they want to know: Participants tailor causal
attributions to researchers interests. European Journal of Social Psychology, 29, 10111020.
Norman, D. A., & Bobrow, D. G. (1979). Descriptions: An intermediate stage in memory retrieval.
Cognitive Psychology, 11, 107123.
REFERENCES
395
Nottenburg, G., & Shoben, E. J. (1980). Scripts as linear orders. Journal of Experimental Social Psychology, 16, 329347.
OGuinn, T. C., & Shrum, L. J. (1997). The role of television in the construction of social reality.
Oishi, S., Wyer, R. S., & Colcombe, S. (2000). Cultural variation in the use of current life satisfaction to predict the future. Journal of Personality and Social Psychology, 78, 434445.
Ortony, A., Clore, G. L., & Collins, A. (1988). The cognitive structure of emotions. New York: Cambridge University Press.
Ostrom, T. M., & Upshaw, H. S. (1968). Psychological perspective and attitude change. In A. G.
Greenwald, T. M. Ostrom, & T. C. Brock (Eds.), Psychological foundations of attitudes. New
York: Academic Press.
Ottati, V. C., & Isbell, L. M. (1996). Effects of mood during exposure to target information on subsequently reported judgments: An on-line model of misattribution and correction. Journal of
Personality and Social Psychology, 71, 3953.
Ottati, V. C., & Wyer, R., S. (1993). Affect and political judgment. In S. Iyengar & W. J. McGuire
(Eds.), Current approaches to political psychology (pp. 296320). Durham, NC: Duke University
Press.
Parducci, A. (1965). Category judgment: A range-frequency model. Psychological Review, 72,
407418.
Park, J. W., & Wyer, R. S. (1993). The cognitive organization of product information: Effects of attribute category set size on information recall. Journal of Consumer Psychology, 2, 329357.
Park, J. W., Yoon, S. O., Kim, K. H., & Wyer, R. S. (2001). Effects of priming a bipolar attribute concept on dimension versus concept-specific accessibility of semantic memory. Journal of Personality and Social Psychology, 81, 405420.
Parrott, G., & Sabini, J. (1990). Mood and memory under natural conditions: Evidence for mood
and incongruent recall. Journal of Personality and Social Psychology, 59, 321336.
Pennington, N., & Hastie, R. (1986). Evidence evaluation in complex decision making. Journal of
Pennington, N., & Hastie, R. (1988). Explanation-based decision making: Effects of memory structure on judgment. Journal of Experimental Psychology: Learning, Memory and Cognition, 14,
521533.
Pennington, N., & Hastie, R. (1992). Explaining the evidence: Tests of the Story Model for juror
decision making. Journal of Personality and Social Psychology, 62, 189206.
Pennington, N., & Hastie, R. (1993). A theory of explanation-based decision making. In G. Klein, J.
Orasanu, R. Calderwood, & C. E. Zsambok (Eds.), Decision making in action: Models and methods (pp. 188204). Norwood, NJ: Ablex.
Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion: Central and peripheral routes
to attitude change. New York: Springer-Verlag.
Pham, M. T. (1998). Representativeness, relevance and the use of feelings in decision making.
Pham, M. T., Cohen, J. B., Pracejus, J., & Hughes, G. D. (2001). Affect monitoring and the primacy
of feelings in judgment. Journal of Consumer Research, 28, 167188.
Picek, J. S., Sherman, S. J., & Shiffrin, R. M. (1975). Cognitive organization and coding of social
structures. Journal of Personality and Social Psychology, 31, 758768.
Prinz, W. (1990). On dynamic pertinence models. In H. G. Geissler & M. H. Miller (Eds.), Psychophysical explorations of mental structures (pp. 411421). Kirkland, WA: Hofgrefe & Huber.
Pylyshyn, Z. W. (1973). What the minds eye tells the minds brain: A critique of mental imagery.
Psychological Bulletin, 80, 124.
Pyszczynski, T. A., & Greenberg, J. (1987). Toward an integration of cognitive and motivational
perspectives on social inference: A biased hypothesis-testing model. In L. Berkowitz (Ed.),
Advances in experimental social psychology (Vol. 20, pp. 297340). San Diego, CA: Academic
Press.
396
REFERENCES
Radvansky, G. A., Gerard, L. D., Zacks, R. T., & Hasher, L. (1990). Younger and older adults use of
mental models as representations for text material. Psychology and Aging, 5, 209214.
Radvansky, G. A., Wyer, R. S., Curiel, J. M., & Lutz, M. F. (1997). Mental models and abstract relations. Journal of Experimental Psychology: Learning, Memory, and Cognition, 23, 12331246.
Radvansky, G. A., & Zacks, R. T. (1991). Mental models and the fan effect. Journal of Experimental
Psychology: Learning, Memory and Cognition, 17, 940953.
Ratcliff, R. (1978). A theory of memory retrieval. Psychological Review, 85, 59108.
Read, D. (1985). Determinants of relative mutability. Unpublished research, University of British
Columbia, Vancouver, Canada.
Regan, D., & Totten, J. (1975). Empathy and attribution: Turning observers into actors. Journal of
Reid, L. D., & Wyer, R. S. (1998). Stereotype suppression in a social context. In R. S. Wyer (Ed.),
Advances in social cognition (Vol. 11, pp. 177192). Mahwah, NJ: Lawrence Erlbaum Associates.
Reyes, R. M., Thompson, W. C., & Bower, G. H. (1980). Judgmental biases resulting from differing
availabilities of arguments. Journal of Personality and Social Psychology, 39, 212.
Rhee, E., Uleman, J., & Lee, H. K. (1996). Variations in collectivism and individualism by ingroup
and culture: Confirmatory factor analysis. Journal of Personality and Social Psychology, 71,
10371054.
Roselli, F. (1995, August). Positive affect and heuristic cues in persuasion. Poster presented at the
American Psychological Society Convention, New York.
Rosen, N. A., & Wyer, R. S. (1972). Some further evidence for the Socratic effect using a subjective probability model of cognitive organization. Journal of Personality and Social Psychology,
24, 420424.
Ross, L. D., Greene, D., & House, P. (1977). The false consensus effect: An egocentric bias in social perception and attribution processes. Journal of Experimental Social Psychology, 13,
279301.
Ross, L., Lepper, M. R., & Hubbard, M. (1975). Perseverance in self-perception and social perception: Biased attributional processes in the debriefing paradigm. Journal of Personality and Social Psychology, 32, 880892.
Ross, L., Lepper, M. R., Strack, F., & Steinmetz, J. (1977). Social explanation and social expectation: Effects of real and hypothetical explanations on subjective likelihood. Journal of Personality and Social Psychology, 35, 817829.
Ross, M. (1989). Relation of implicit theories to the construction of personal histories. Psychological Review, 96, 341357.
Ross, M., McFarland, C., & Fletcher, G. J. O. (1981). The effect of attitude on recall of past histories. Journal of Personality and Social Psychology, 10, 627634.
Rubin, D. C. (1995). Stories about stories. In R. S. Wyer (Ed.), Advances in social cognition (Vol. 8,
Rusting, C. L., & DeHart, T. (2000). Retrieving positive memories to regulate negative mood:
Consequences for mood-congruent memory. Journal of Personality and Social Psychology, 78,
737752.
Sanitioso, R., Kunda, Z., & Fong, G. T. (1990). Motivated recruitment of autobiographical memories. Journal of Personality and Social Psychology, 59, 229241.
Sanna, L. J., & Schwarz, N. (2003). Debiasing the hindsight bias: The role of accessibility experiences and (mis)attributions. Journal of Experimental Social Psychology.
Sanna, L. J., Schwarz, N., & Small, E. M. (2003). Accessibility experiences and the hindsight bias: I
knew it all along versus it could never have happened. Memory & Cognition.
Sanna, L. J., Schwarz, N., & Stocker, S. L. (2002). What debiasing backfires: Accessible content
and accessibility experiences in debiasing hindsight. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 28, 497502.
REFERENCES
397
Schachter, S., & Singer, J. (1962). Cognitive, social and physiological determinants of emotional
state. Psychological Review, 69, 379399.
Schaeken, W., Johnson-Laird, P. N., & dYdewalle, G. (1996). Mental models and temporal reasoning. Cognition, 60, 205234.
Schank, R. C., & Abelson, R. P. (1977). Scripts, plans, goals and understanding. Hillsdale, NJ: Lawrence Erlbaum Associates.
Schank, R. C., & Abelson, R. P. (1995). Knowledge and memory: The real story. In R. S. Wyer (Ed.),
Advances in social cognition (Vol. 8, pp. 185). Hillsdale, NJ: Lawrence Erlbaum Associates.
Schick, C., McGlynn, R. P., & Woolam, D. (1972). Perception of cartoon humor as a function of familiarity and anxiety level. Journal of Personality and Social Psychology, 24, 2225.
Schneider, W., & Shiffrin, R. M. (1977). Controlled and automatic human information processing:
I. Detection, research and attention. Psychological Review, 84, 166.
Schooler, J. W., & Engstler-Schooler, T. Y. (1990). Verbal overshadowing of visual memories:
Some things are better left unsaid. Cognitive Psychology, 22, 3671.
Schwarz, N. (1990). Feelings as information: Informational and motivational functions of affective states. In E. T. Higgins & R. M. Sorrentino (Eds.), Handbook of motivation and cognition:
Foundations of social behavior (Vol. 2, pp. 527561). New York: Guilford Press.
Schwarz, N. (1998). Accessible content and accessibility experiences: The interplay of declarative and experiential information in judgment. Personality and Social Psychology Review, 2,
8799.
Schwarz, N. (1994). Judgment in a social context: Biases, shortcomings, and the logic of conversation. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 24, pp. 123162).
San Diego, CA: Academic Press.
Schwarz, N., & Bless, H. (1991). Emotion and social judgments. In J. Forgas (Ed.), Emotion and social judgments (pp. 5571). Elmsford, NJ: Pergamon Press.
Schwarz, N., Bless, H., & Bohner, G. (1991). Mood and persuasion: Affective states influence the
processing of persuasive communications. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 24, pp. 161199). San Diego, CA: Academic Press.
Schwarz, N., Bless, H., Strack, F., Klumpp, G., Rittenauer-Schatka, H., & Simons, A. (1991). Ease of
retrieval as information: Another look at the availability heuristic. Journal of Personality and
Schwarz, N., & Clore, G. L. (1983). Mood, misattribution, and judgments of well-being: Informative
and directive functions of affective states. Journal of Personality and Social Psychology, 45,
513523.
Schwarz, N., & Clore, G. L. (1988). How do I feel about it? Informative functions of affective states.
In K. Fiedler & J. Forgas (Eds.), Affect, cognition, and social behavior (pp. 4462). Toronto:
Hofgrefe International.
Schwarz, N., & Clore, G. L. (1996). Feelings and phenomenal experiences. In E. T. Higgins & A.
Kruglanski (Eds.), Social psychology: A handbook of basic principles (pp. 433465). New York:
Guilford Press.
Schwarz, N., Strack, F., Hilton, D., & Naderer, G. (1991). Base rates, representativeness, and the
logic of conversation: The contextual relevance of irrelevant information. Social Cognition,
9, 6784.
Schwarz, N., & Wyer, R. S. (1985). Effects of rank ordering stimuli on magnitude ratings of these
and other stimuli. Journal of Experimental Social Psychology, 21, 3046.
Scott, C. K., Fuhrman, R. W., & Wyer, R. S. (1991). Information processing in close relationships. In
G. J. O. Fletcher & F. D. Fincham (Eds.), Cognition in close relationships (pp. 3768). Hillsdale,
Sentis, K. P., & Burnstein, E. (1979). Remembering schema-consistent information: Effects of a
balance schema on recognition memory. Journal of Personality and Social Psychology, 37,
22002212.
398
REFERENCES
Sheehan, P. W. (1967). A shortened form of Betts Questionnaire on Mental Imagery. Journal of

Clinical Psychology, 23, 386389.
Shepard, R. N., & Metzler, J. (1971). Mental rotation of three-dimensional objects. Science, 171,
701703.
Sherman, B. R., & Kunda, Z. (1989). Motivated evaluation of scientific evidence. Paper presented at
the American Psychological Society convention, Arlington.
Sherman, S. J., & Corty, E. (1984). Cognitive heuristics. In R. S. Wyer & T. K. Srull (Eds.), Handbook
of social cognition (Vol. 1, pp. 189286). Hillsdale, NJ: Lawrence Erlbaum Associates.
Sherman, J. W., & Klein, S. B. (1994). Development and representation of personality impressions. Journal of Personality and Social Psychology, 67, 972983.
Sherman, S. J., Ahlm, K., Berman, L., & Lynn, S. (1978). Contrast effects and the relationship to
subsequent behavior. Journal of Experimental Social Psychology, 14, 340350.
Sherman, S. J., Skov, R. B., Hervitz, E. F., & Stock, C. B. (1981). The effects of explaining hypothetical future events: From possibility to probability to actuality and beyond. Journal of Experimental Social Psychology, 17, 142158.
Shiv, B., & Fedorikhin, A. (1999). Heart and mind in conflict: The interplay of affect and cognition
in consumer decision making. Journal of Consumer Research, 26, 278292.
Shrum, L. J. (1999). Television and persuasion: Effects of the programs between the ads. Psychology & Marketing, 16, 119140.
Shrum, L. J., OGuinn, T. C., Semenik, R. J., & Faber, R. J. (1998). Process and effects in the construction of normative consumer beliefs: The role of television. In R. H. Holman & M. R. Solomon (Eds.), Advances in consumer research (Vol. 18, pp. 755763). Provo, UT: Association for
Consumer Research.
Shrum, L. J., Wyer, R. S., & OGuinn, T. (1998). The effects of watching television on perceptions of
social reality. Journal of Consumer Research, 24, 447458.
Shurcliff, A. (1968). Judged humor, arousal, and the relief theory. Journal of Personality and Social
Sinclair, R. S., & Mark, M. M. (1992). The influence of mood state on judgment and action: Effects
on persuasion, categorization, social justice, person perception and judgmental accuracy. In
A. Tesser & L. L. Martin (Eds.), The construction of social judgments (pp. 165194). Hillsdale, NJ:
Sinclair, R. C., Mark, M. M., & Clore, G. L. (1994). Mood-related persuasion depends on (mis)attributions. Social Cognition, 12, 309326.
Skowronski, J. J., & Carlston, D. (1989). Negativity and extremity biases in impression formation:
A review of explanations. Psychological Bulletin, 195, 131142.
Smith, E. R. (1984). Model of social inference processes. Psychological Review, 91, 392413.
Smith, E. R. (1990). Content and process specificity in the effects of prior experiences. In T. K.
Srull & R. S. Wyer (Eds.), Advances in social cognition (Vol. 3, pp. 159). Hillsdale, NJ: Lawrence
Erlbaum Associates.
Smith, E. R. (1994). Procedural knowledge and processing strategies in social cognition. In R. S.
Wyer & T. K. Srull (Eds.), Handbook of social cognition (2nd ed., Vol. 1, pp. 99151). Hillsdale,
Smith, E. R. (1998). Mental representation and memory. In D. T. Gilbert, S. T. Fiske, & G. Lindzey
(Eds.), Handbook of social psychology (4th ed., Vol. 2, pp. 391445). New York: McGraw Hill.
Smith, E. R., & DeCoster, J. (1998). Knowledge acquisition, accessibility, and use in person perception and stereotyping: Simulation with a recurrent connectionist network. Journal of Personality and Social Psychology, 74, 2135.
Smith, E. R., & DeCoster, J. (1999). Associative and rule-based processing: A connectionist interpretation of dual-processing models. In S. Chaiken & Y. Trope (Eds.), Dual-process theories in
social psychology (pp. 323336). New York: Guilford Press.
Snyder, M., & Cantor, N. (1979). Testing hypotheses about other people: The use of historical
knowledge. Journal of Experimental Social Psychology, 15, 330342.
REFERENCES
399
Snyder, M. L., Stephan, W. G., & Rosenfield, D. (1976). Egotism and attribution. Journal of Personality and Social Psychology, 33, 435441.
Sperber, D., & Wilson, D. (1986). Relevance: Communication and cognition. Oxford, England: Basil
Blackwell.
Sperling, G. (1960). The information available in brief visual presentations. Psychological Monographs, 74, 129.
Spiro, R. J. (1977). Remembering information from text: The state of schema approach. In R. C.
Anderson, R. J. Spiro, & W. E. Montague (Eds.), Schooling and the acquisition of knowledge.
Srull, T. K. (1981). Person memory: Some tests of associative storage and retrieval models. Journal of Experimental Psychology: Human Learning and Memory, 7, 440463.
Srull, T. K., & Brand, J. F. (1983). Memory for information about persons: The effect of encoding
operations on subsequent retrieval. Journal of Verbal Learning and Verbal Behavior, 22,
219230.
Srull, T. K., Lichtenstein, M., & Rothbart, M. (1985). Associative storage and retrieval processes
in person memory. Journal of Experimental Psychology: Learning, Memory and Cognition, 11,
316345.
Srull, T. K., & Wyer, R. S. (1979). The role of category accessibility in the interpretation of information about persons: Some determinants and implications. Journal of Personality and Social
Srull, T. K., & Wyer, R. S. (1980). Category accessibility and social perception: Some implications
for the study of person memory and interpersonal judgments. Journal of Personality and Social Psychology, 38, 841856.
Srull, T. K., & Wyer, R. S. (1983). The role of control processes and structural constraints in models of memory and social judgment. Journal of Experimental Social Psychology, 19, 497521.
Srull, T. K., & Wyer, R. S. (1989). Person memory and judgment. Psychological Review, 96, 5863.
Stapel, D. A., & Koomen, W. (2000). How far do we go beyond the information given? The impact
of knowledge activation on interpretation and inference. Journal of Personality and Social Psychology, 78, 1937.
Stapel, D. A., Koomen, W., & van der Plight, J. (1997). Categories of category accessibility: The impact of trait concept versus exemplar priming on person judgments. Journal of Experimental
Stein, N. L. (1982). The definition of a story. Journal of Pragmatics, 6, 487507.
Stein, N. L., & Glenn, C. G. (1979). An analysis of story comprehension in elementary school children. In R. O. Freedle (Ed.), New directions in discourse processing. Norwood, NJ: Ablex.
Storms, M. (1973). Videotape and the attribution process: Reversing actors and observers point
of view. Journal of Personality and Social Psychology, 27, 165175.
Strack, F. (1994). Response processes in social judgment. In R. S. Wyer & T. K. Srull (Eds.), Handbook of social cognition (2nd ed., Vol. 1, pp. 287322). Hillsdale, NJ: Lawrence Erlbaum Associates.
Strack, F., & Deutsch, R. (2002). Reflective and impulsive determinants of social behavior. Unpublished manuscript, University of Wrzburg, Germany.
Strack, F., Martin, L. L., & Stepper, S. (1988). Inhibiting and facilitating conditions of the human
smile: A nonobtrusive test of the facial feedback hypothesis. Journal of Personality and Social
Strack, F., Schwarz, N., & Gschneidinger, E. E. (1985). Happiness and reminiscing: The role of time
perspective, affect and mode of thinking. Journal of Personality and Social Psychology, 49,
14601469.
Strack, F., Schwarz, N., Bless, H., & Kuebler, A. 1993). Awareness of the influence as a determinant of assimilation versus contrast. European Journal of Social Psychology, 23, 5362.
Suls, J. M. (1972). Two-stage model for the appreciation of jokes and cartoons: Informationprocessing analysis. In J. H. Goldstein & P. E. McGhee (Eds.), The psychology of humor (pp.
81100). San Diego, CA: Academic Press.
400
REFERENCES
Suls, J. M. (1977). Cognitive processes in humor appreciation. In P. E. McGhee & J. H. Goldstein

(Eds.), Its a funny thing, humor (pp. 4145). Elmsford, NY: Pergamon Press.
Suls, J. M. (1983). Cognitive processes in humor appreciation. In P. E. McGhee & J. H. Goldstein
(Eds.), Handbook of humor research: Basic issues (Vol. 1, pp. 3957). New York: SpringerVerlag.
Tannen, D. (1990). You just dont understand: Women and men in conversation. New York: William
Morrow.
Taylor, H. A., & Tversky, B. (1992). Spatial situation models derived from survey and route descriptions. Journal of Memory and Language, 31, 261292.
Taylor, S. E., & Fiske, S. T. (1978). Salience, attention and attribution: Top of the head phenomena. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 11, pp. 249288).
New York: Academic Press.
Tesser, A. (1978). Self-generated attitude change. In L. Berkowitz (Ed.), Advances in experimental
social psychology (Vol. 11, pp. 289338). New York: Academic Press.
Thurstone, L. L. (1959). The measurement of values. Chicago: University of Chicago Press.
Trafimow, D. A., & Wyer, R. S. (1993). Cognitive representation of mundane social events. Journal
of Personality and Social Psychology, 64, 365376.
Triandis, H. C. (1995). Individualism and collectivism. Boulder, CO: Westview.
Triandis, H. C., & Gelfand, M. J. (1998). Converging measurement of horizontal and vertical individualism and collectivism. Journal of Personality and Social Psychology, 74, 118128.
Trope, Y. (1986). Identification and inferential processes in dispositional attribution. Psychological Review, 93, 239257.
Troutman, M. C., & Shanteau, J. A. (1976). Do consumers evaluate products by adding or averaging attribute information? Journal of Consumer Research, 3, 101106.
Tversky, A., & Kahneman, D. (1973). Availability: A heuristic for judging frequency and probability. Cognitive Psychology, 5, 207232.
Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science,
185, 11241131.
Tversky, A., & Kahneman, D. (1982). Causal schemas in judgments under uncertainty. In D.
Kahneman, P. Slovic, & A. Tversky (Eds.), Judgment under uncertainty: Heuristics and biases
(pp. 117128). New York: Cambridge University Press.
Upshaw, H. S. (1969). The personal reference scale: An approach to social judgment. In L.
Berkowitz (Ed.), Advances in experimental social psychology (Vol. 4). New York: Academic
Press.
Velton, E. (1968). A laboratory task for the induction of mood states. Behavior Research and Therapy, 6, 473482.
Walster, E. (1996). Assignment of responsibility for an accident. Journal of Personality and Social
Wnke, M., Bless, H., & Biller, B. (1996). Subjective experience versus content of information in
the construction of attitude judgments. Personality and Social Psychology Bulletin, 11, 1105
1113.
Wnke, M., & Wyer, R. S. (1996). Individual differences in person memory: The role of sociopolitical ideology and in-group versus out-group membership in responses to socially relevant behavior. Personality and Social Psychology Bulletin, 22, 742754.
Watzlawick, P., Beavin, J. H., & Jackson, D. D. (1967). Pragmatics of human communication. New
York: Norton.
Weber, E., Hsee, C., & Sokolowska, J. (1998). What folklore tells us about risk and risk taking:
Cross-cultural comparisons of American, German and Chinese proverbs. Organizational Behavior and Human Decision Processes, 75, 170186.
Wegener, D. T., Petty, R. E., & Smith, S. M. (1995). Positive mood can increase or decrease message scrutiny: The hedonic contingency view of mood and message processing. Journal of
REFERENCES
401
Wegner, D. M. (1994). Ironic processes of mental control. Psychological Review, 101, 3452.
White, R. W. (1959). Motivation reconsidered: the concept of competence. Psychological Review,
66, 297333.
Wills, T. A. (1981). Downward comparison principles in social psychology. Psychological Bulletin,
90, 245271.
Wolfson, M. R., & Salancik, G. R. (1977). Observer orientation and actor-observer differences in
attributions for failure. Journal of Experimental Social Psychology, 13, 441451.
Wyer, R. S. (1970). The quantitative prediction of belief and opinion change: A further test of a
subjective probability model. Journal of Personality and Social Psychology, 16, 559571.
Wyer, R. S. (1973). Category ratings as subjective expected values: Implications for attitude formation and change. Psychological Review, 80, 446467.
Wyer, R. S. (1974). Cognitive organization and change: An information-processing approach. Hillsdale, NJ: Lawrence Erlbaum Associates.
Wyer, R. S. (1977). The role of logical and nonlogical factors in making inferences about category
membership. Journal of Experimental Social Psychology, 13, 577595.
Wyer, R. S., & Adaval, R. (2003). Message reception skills in social communication. In J. O. Greene
& B. Burleson (Eds.), Handbook of communication and social interaction skills (pp. 291354).
Mahwah, NJ: Lawrence Erlbaum Associates.
Wyer, R. S., Adaval, R., & Colcombe, S. J. (2002). Narrative-based representations of social knowledge: Their construction and use in comprehension, memory and judgment. In M. P. Zanna
(Ed.), Advances in experimental social psychology (Vol. 34, pp. 131197). San Diego, CA: Academic Press.
Wyer, R. S., & Bodenhausen, G. V. (1985). Event memory: The effects of processing objectives
and time delay on memory for action sequences. Journal of Personality and Social Psychology,
49, 304316.
Wyer, R. S., Bodenhausen, G. V., & Gorman, T. F. (1985). Cognitive mediators of reactions to rape.
Wyer, R. S., Bodenhausen, G. V., & Srull, T. K. (1984). The cognitive representation of persons and
groups and its effect on recall and recognition memory. Journal of Experimental Social Psychology, 20, 445469.
Wyer, R. S., & Budesheim, T. L. (1987). Person memory and judgment: The impact of information
that one is told to disregard. Journal of Personality and Social Psychology, 53, 1429.
Wyer, R. S., Budesheim, T. L., & Lambert, A. J. (1990). Person memory and judgment: The cognitive representation of informal conversations. Journal of Personality and Social Psychology, 58,
218238.
Wyer, R. S., Budesheim, T. L., Lambert, A. J., & Martin, L. L. (1989). Person memory: The priorities
that govern the cognitive activities involved in person impression formation. Unpublished manuscript, University of Illinois at Urbana-Champaign.
Wyer, R. S., Budesheim, T. L., Lambert, A. J., & Swan, S. (1994). Person memory and judgment:
Pragmatic influences on impressions formed in a social context. Journal of Personality and Social Psychology, 66, 254267.
Wyer, R. S., Budesheim, T. L., Shavitt, S., Riggle, E. J., Melton, R. J., & Kuklinski, J. H. (1991). Image,
issues and ideology: The processing of information about political candidates. Journal of Personality and Social Psychology, 61, 533545.
Wyer, R. S., & Carlston, D. E. (1979). Social cognition, inference and attribution. Hillsdale, NJ: Lawrence Erlbaum Associates.
Wyer, R. S., & Carlston, D. E. (1994). The cognitive representation of persons and events. In R. S.
Wyer & T. K. Srull (Eds.), Handbook of social cognition (2nd ed., Vol. 1, pp. 4198). Hillsdale, NJ:
Wyer, R. S., Clore, G. L., & Isbell, L. M. (1999). Affect and information processing. In M. P. Zanna
(Ed.), Advances in experimental social psychology (vol. 31, pp. 177). San Diego: Academic
Press.
402
REFERENCES
Wyer, R. S., & Collins, J. E. (1992). A theory of humor elicitation. Psychological Review, 99, 663688.
Wyer, R. S., & Goldberg, L. (1970). A probabilistic analysis of the relationships among beliefs and
attitudes. Psychological Review, 77, 100120.
Wyer, R. S., & Gordon, S. E. (1982). The recall of information about persons and groups. Journal of
Experimental Social Psychology, 18, 128164.
Wyer, R. S., & Gruenfeld, D. H. (1995). Information processing in social contexts: Implications for
social memory and judgment. In M. P. Zanna (Ed.), Advances in experimental social psychology
(Vol. 27, pp. 4991). San Diego, CA: Academic Press.
Wyer, R. S., & Hartwick, J. (1980). The role of information retrieval and conditional inference
processes in belief formation and change. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 13, pp. 241284). New York: Academic Press.
Wyer, R. S., & Hartwick, J. (1984). The recall and use of belief statements as bases for judgments:
Some determinants and implications. Journal of Experimental Social Psychology, 20, 6585.
Wyer, R. S., Lambert, A. J., Budesheim, T. L., & Gruenfeld, D. (1992). Theory and research on person impression formation: A look to the future. In L. Martin & A. Tesser (Eds.), The construction of social judgment (pp. 336). Hillsdale, NJ: Lawrence Erlbaum Associates.
Wyer, R. S., & Martin, L. L. (1986). Person memory: The role of traits, group stereotypes, and specific behaviors in the cognitive representation of persons. Journal of Personality and Social
Wyer, R. S., & Radvansky, G. A. (1999). The comprehension and validation of social information.
Psychological Review, 106, 89118.
Wyer, R. S., Shoben, E. J., Fuhrman, R. F., & Bodenhausen, G. V. (1985). Event memory: The cognitive representation of social action sequences. Journal of Personality and Social Psychology,
49, 857877.
Wyer, R. S., & Srull, T. K. (1989). Memory and cognition in its social context. Hillsdale, NJ: Lawrence
Erlbaum Associates.
Wyer, R. S., Swan, S., & Gruenfeld, D. H. (1995). Impression formation in informal conversations.
Social Cognition, 13, 243272.
Wyer, R. S., & Unverzagt, W. H. (1985). The effects of instructions to disregard information on its
subsequent recall and use in making judgments. Journal of Personality and Social Psychology,
48, 533549.
Yamauchi, H. (1988). Effects of actors and observers roles on causal attributions by Japanese
subjects for success and failure in competitive situations. Psychological Reports, 63, 619626.
Yeung, C. W. M. (2003). Unpublished data.
Yeung, C. W. M., & Wyer, R. S. (in press). Affect, appraisal, and consumer judgment. Journal of
Consumer Research.
Zajonc, R. B. (1968). Attitudinal effects of mere exposure. Journal of Personality and Social Psychology, 9, 127.
Zajonc, R. B. (1980). Feeling and thinking. Preferences need no inferences. American Psychologist,
35, 151175.
Zajonc, R. B., Murray, S. T., & Inglehart, M. (1989). Feeling and facial efference: Implication of the
vascular theory of emotion. Psychological Review, 96, 395416.
Zanna, M. P., & Cooper, J. (1976). Dissonance and the attribution process. In J. Harvey, W. Ickes,
& R. Kidd (Eds.), New directions in attribution research (Vol. 1, pp. 199217). Hillsdale, NJ: Lawrence Erlbaum Associates.
Zanna, M., & Rempel, J. K. (1988). Attitudes: a new look at an old concept. In D. Bar-Tal & A.
Kruglanski (Eds.), The social psychology of knowledge (pp. 315334). Cambridge, England:
Zigler, E., Levine, J., & Gould, L. (1967). Cognitive challenge as a factor in childrens humor appreciation. Journal of Personality and Social Psychology, 6, 332336.
Zillman, D. (1983). Connections between sex and aggression. Hillsdale, NJ: Lawrence Erlbaum Associates.
REFERENCES
403
Zillmann, D., & Bryant, J. (1982). Pornography, sexual callousness, and the trivialization of rape.
Journal of Communication, 32, 1021.
Zillman, D., & Cantor, J. R. (1976). A disposition theory of humor and mirth. In A. J. Chapman &
H. C. Foot (Eds.), Humor and laughter: Theory, research and application (pp. 93116). New York:
Wiley.
Zinn, H. (1980). A peoples history of the United States. New York: Harper-Collins.
Zuckerman, M. (1975). Belief in a just world and altruistic behavior. Journal of Personality and Social Psychology, 31, 972976.
Zwaan, R. A., Magliano, J. P., & Graesser, A. C. (1995). Dimensions of situation model construction
in narrative comprehension. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21, 386397.
Zwaan, R. A., & Radvansky, G. A. (1998). Situation models in language comprehension and memory. Psychological Bulletin, 123, 162185.
Author Index
A
Aaker, J. L., 70, 148, 302, 303, 380
Abramowitz, S. I., 164, 394
Achee, J. W., 312, 345-349, 363, 392
Abelson, R. P., 3, 16, 18, 21, 28, 116,
122, 126, 146, 149, 168, 293, 300,
307, 309, 340, 380, 391, 397
Adaval, R., 29, 34, 47, 63, 86-88, 115,
139, 142-144, 151, 236, 239-241,
243-250, 274, 307, 309, 333, 335,
342, 351, 386, 401
Ajzen, I., 358, 359, 363, 386
Albarracin, D., 62, 121, 277, 307,
338, 360, 362, 363, 380
Albrecht, J. E., 39. 381
Anderson, A., 28, 381
Anderson, J. R., 6, 9, 39, 49, 73, 128,
133
Anderson, N. H., 35, 165, 235, 236,
334, 342, 381, 391.
Anderson, R. C., 51, 381
Anthony, S., 208, 387
Apter, M. J., 191, 193, 195, 196, 198,
200, 381
Arkin, R. M., 66, 264, 265, 278, 292,
324, 381
Ashbrook, P. W., 344, 385
Atkinson, J. W., 204, 210, 393
B
Banaji, M., 13, 387
Banks, W. P., 114, 381
Barclay, J. R.10, 29, 84, 151, 381
Bargh, J. A., 7, 13, 23, 41, 47, 50-53,
55, 57, 62, 69, 73-75, 101, 136, 199,
309, 314, 315, 381, 383-385, 389,
392
Baron, R. A., 311, 381

Barsalou, L. W., 130, 248, 249, 381
Baumeister, R. F., 66, 382
Beavin, J. H., 160, 186, 400
Bell, D. E., 303, 382
Bem, D. J., 35, 62, 275, 276, 382
Bender, A., 191, 194, 393
Benet-Martinez, V., 47, 291, 299,
389
Bergson, H., 193, 382
Berlyne, D. E., 192, 193, 382
Bettman, J. R.,334, 382
Bhatla, S., 236, 397
Birnbaum, M. H., 240, 382
Black, J. B., 18, 19, 85, 126, 259, 382
Blaney, P.H. 306, 316, 382
Bless, H., 55, 282, 283, 285, 307, 339,
340, 342, 343, 351, 356, 382, 397,
399
Bobrow, D. G., 42, 394
Bodenhausen, G. V., 10, 28, 31, 32,
47, 56, 67-69, 110,. 113, 114, 168,
169, 186, 187, 269 , 303, 304, 307,
331, 351, 382, 387, 392, 401, 402
Bohner, G., 307, 342, 243, 356, 382,
397
Bond, R. N., 51, 52, 381
Boorstin, D., 149, 382
Bower, G. H., 9, 18, 29, 85, 126, 128,
255, 259, 306, 311, 312, 315, 316,
318-320, 381-383, 386, 396
Brand, J. F., 128, 399
Bransford, J. D., 29, 84, 108, 151, 383
Brewer, M., 26, 383
Brewer, W. F., 3, 19, 383
Briley, D. A., 47, 69-72, 148, 290,
291, 302, 303, 383
Brock, T. C., 101, 387
Brown, J., 54, 280, 390
Brown, N. R., 113, 383
405
406
AUTHOR INDEX
Brown, R., 304, 391

Brucks, M., 146, 236, 383, 384
Bryant, J., 282, 402
Budesheim, T. L., 13, 28, 32,163,
167-169, 172, 176, 178, 251, 253, 401,
402
Burnstein, E., 122, 397
Burrows, L., 47, 73-75, 104, 381
Byrne, D., 311, 381, 383, 384
C
Cacioppo, J. T., 55, 346, 383, 395
Cantor, J. R., 191, 193, 201, 203, 218,
403
Cantor, N., 331, 398
Capon, N., 334, 382
Carlsmith, M., 179, 385
Carlston, D. E., 8, 10, 12, 21, 22, 24,
26, 28, 39, 63, 122, 166, 240, 271,
311, 334, 383, 398, 401
Ceci, S. J., 146, 383
Chaiken, S., 36, 48, 66, 138, 188,
309, 314, 315, 367, 381, 383-385,
387
Chakravarti, D., 35, 392
Chapman, A. J., 191, 384
Chartrand, T. L., 23, 384
Chen, M., 47, 73-75,104, 381
Chernev, A., 331, 384
Chiu, C., 47, 287, 291, 299, 384, 385,
389
Chodorow, N., 183, 384
Choi, I., 291, 294, 384
Clark, M. S., 311, 390
Clark, R. A., 204, 210, 393
Clore, G. L., 8, 9, 306-309, 311, 312,
314, 322, 325, 328, 338-340, 342,
345, 347, 349-351, 353, 354, 356,
382, 384, 387, 389, 395, 397, 398, 401
Cohen, J. B., 307, 358, 393, 395
Colcombe, S. J., 4,19, 29, 74, 75, 8688, 104, 115-117, 124, 129, 130,
132-136, 292, 293, 384, 385, 401
Collins, A. M., 24, 39, 308, 384, 395
Collins, J. E., 161, 198, 201-203, 205207, 212, 215, 216, 218, 221, 223,
224, 390, 401
Conway, M., 274, 308, 328, 384
Cooper, J., 66, 402
Corbin, S., 191, 194, 393
Corty, E., 280, 398
Costley, C. L., 236, 384
Craik, F. I. M., 162, 282, 384
Crelia, R. A., 55, 392
Curiel, J. M., 17, 109-111, 396
Cutrona, C., 289, 384
D
Danilovics, P., 163, 386
Daubman, L. A., 307, 348, 349,
390
Davis, K. E., 35, 187, 390
DeCoster, J., 39, 398
DeHart, T., 317, 397
DePaulo, B. M., 179, 384
Deutsch, R., 8, 399
Devine, P. G., 304, 385
Dickson, P. R., 236, 393
Dienes, B. P. A., 340, 385
Dijksterhuis, A., 47, 385
Dijkstra, K., 81, 392
Duckworth, K. L., 314, 385
Dweck, C. S., 4, 286, 287, 299, 384,
385, 389
d'Ydewalle, G., 81, 397
E
Eagly, A. H., 48, 188, 309, 383-385
Ebbesen, E. B., 103, 385
Edell, J. A., 235, 385
Einstein, G. O., 340, 389
Ellis, H. C., 344, 385
Ellsworth, P. C., 179, 385
Emler, N., 164, 385
Englis, B. G., 241, 385
Engstler-Schooler, T. Y., 138, 397
Erdley, C. A., 286, 385
407
AUTHOR INDEX
F
Faber, R. J., 280, 398
Fazio, R. H., 83, 314, 315, 358, 385
Fedorikhin, A., 66, 327, 338, 398
Fernandez, G., 163, 386
Festinger, L., 66, 276, 386
Fiedler, K., 386
Fischoff, B., 285, 386
Fishbein, M., 235, 322, 358, 363, 386
Fishkin, J., 164, 386
Fiske, S. T., 36, 47, 217, 303, 307,
309, 380, 386, 400
Fletcher, G. J. O., 163, 279, 386, 396
Fong, C. P. S., 302, 386
Fong, G. T., 279, 396
Foot, H. C., 191, 384
Forgas, J. P., 295, 308, 312, 314-316,
318-320, 322, 357, 358, 386
Franks, J. J., 29, 84, 151, 383
Freud, S., 191, 193, 204, 218, 386
Friedman, H. S., 179, 384
Fry, P. S., 292, 387
Fu, J. H., 287, 384
Fuhrman, R. W., 28, 113, 114, 132,
391, 387, 402
Fung, H., 145-147, 385
G
Gaelick, L., 28, 186, 187, 387
Galambos, J. A., 19, 382
Garcia, M., 314, 385
Garnham, A., 29, 84, 387
Garrod, S. C., 28, 381
Gasper, K., 338, 387
Gelfand, M. J., 291, 400
Gerard, L. D., 29, 396
Gerbner, G., 281, 387
Gerrod, S. C., 28, 381
Ghosh, R., 292, 387
Gick, M. L., 130, 387
Gifford, R. K., 282, 388
Gilbert, D. T., 102, 387
Gilligan, C., 183, 387
Gilligan, S. G., 316, 318, 320, 382
Giner-Sorolla, R., 66, 387

Gleason, J. M., 66, 264, 278, 292,
324, 381
Glenberg, A. M., 81, 85, 387
Glenn, C. G., 4, 19, 399
Goethals, G. R., 275, 276, 387
Goetz, F. T., 51, 381
Gohm, C. L., 295, 387
Goldberg, L., 60, 270, 271, 401
Goldstein, D., 54, 101, 208, 280, 388
Goldstein, J. H., 191, 387, 393
Golisano, V., 339, 340, 342, 350, 382
Gordon, S. E., 27, 126, 127, 167, 169,
387, 402
Gorman, T. F., 67-69, 269, 401
Gould, L., 205, 402
Govender, R., 309, 314, 315, 381
Graesser, A. C., 4, 18, 28, 81, 117,
124, 126, 127, 149, 191, 194, 387,
392, 403
Green, G. M., 153, 387
Green, M. C., 101, 387
Greenberg, J., 66, 395
Greene, D., 133,396
Greenwald, A. G., 11, 13, 352, 387,
388
Greundel, J., 145, 394
Grice, H. P., 79, 153, 260, 388
Gross, L., 281, 387
Grove, T. R., 316, 318, 319, 394
Gruenfeld, D. H., 28, 153, 157-159,
163, 176, 179-181, 183, 185, 186,
388, 402
Gschneidinger, E., 324, 354, 399
H
Haddad, J., 191. 193. 391
Halberstadt, J. B., 312, 313, 317, 318,
394
Hamilton, D. L., 10, 282, 303, 304,
388
Harkins, S. G., 55, 388
Hartwick, J., 60, 61, 118, 120, 162,
271, 359, 402
Harvey, J. H., 264, 292, 388
Hasher, L., 29, 54, 101, 280, 388, 396
408
AUTHOR INDEX
Hashtroudi, S., 280, 390

Hastie, R., 4, 27, 163, 166, 168, 232235, 285, 350, 388, 395
Hawkins, S. A., 285, 388
Heider, F., 122, 388
Heine, S. J., 290, 292, 388
Henninger, M., 60, 119, 272, 388
Herr, P. M., 62, 388
Hervitz, E. F., 256, 398
Higgins, E. T., 10, 12, 41, 47, 49, 50,
51, 52, 55, 57, 66, 69, 144, 148,
153, 199, 260, 301, 321, 388, 389,
392
Hilgard, E. R., 142, 389
Hilton, D., 262, 263, 282, 283, 397
Hintzman, D. L., 40, 126, 389
Hirt, E. R., 341, 347, 394
Ho, D. Y. F., 291, 389
Hofstede, G. H., 290, 389
Holmberg, D., 295, 389
Holmes, J. G., 4, 66, 269, 295, 389,
391, 394
Holyoak, K. J., 130, 387
Hong, Y., 47, 287, 291, 299, 384, 385,
389
House, P., 133, 396
Hsee, C., 70, 400
Hubbard, M., 257, 285, 396
Hughes, G. D., 307, 395
Humphreys, M. S., 39, 389
Hunt, R. R., 340, 389
Hunter, R., 235, 386
Hyman, I. E., 146, 389
Hymes, C., 314, 381
I
Inglehart, M., 308, 324, 402
Innes-Ker, A., 320, 389
Isbell, L. M., 9, 161, 203, 212, 215,
221, 223, 224, 236, 241, 243, 245,
246, 249, 250, 307-309, 314, 322,
328, 335-337, 343, 345, 349-351,
353, 380,389, 390,395,401
Isen, A. M., 307, 311, 316, 347-349,
390
J
Jackson, D. D., 160, 186, 400
Jacoby, L. L., 54, 280, 390
Janis, I. L., 362, 390
Jasechko, J., 54, 280, 390
Jetten, J., 56, 392
Johnson, M., 307, 390
Johnson, M. K., 84, 280, 383, 390
Johnson, N. S., 4, 19, 392
Johnson-Laird, P. N., 16, 80, 81, 390,
397
Johnston, S., 66, 264, 265, 278, 292,
324, 381
Jones, C. R., 47, 49, 52, 321, 389
Jones, E. E., 35, 187, 390
K
Kahneman, D., 35, 64, 257, 258, 260,
263, 264, 280, 390, 400
Kane, T. R., 191, 300
Kardes, F. R., 314, 315, 385
Karp, L., 311, 390
Kashima, Y.,39, 389
Katz, L. B., 10, 388
Kelley, C. M.,54, 280, 390
Kelley, H. H., 187, 390
Keniston, K., 164, 386
Kihlstrom, J. F., 318, 390
Kim, B., 294, 295, 394
Kim, K. H., 24, 58, 59, 395
Kinder, D. R., 307, 309, 380
King, B. T., 362, 390
Kintsch, W., 16, 39, 81, 390
Kirkendol, S. E., 385
Kitayama, S., 290-292, 388, 392
Klein, S. B., 38, 132, 352, 391, 398
Klinger, E., 344, 391
Klumpp, G., 282, 397
Koestler, A., 193, 196, 391
Kohlberg, L., 164, 391
Koomen, W., 52, 63, 321, 399
Kosslyn, S. M., 9, 81, 391
Kramer, G. P., 307, 351, 382
Krantz, S., 320, 386
409
AUTHOR INDEX
Kruley, P., 81, 387
Kuebler, A., 55, 399
Kuklinski, J. H., 251, 253, 401
Kumar, P.A., 27, 163, 166, 168, 388
Kunda, Z., 123, 265, 278, 279, 391,
396, 398
L
LaFave, L., 191, 193, 391
Lambert, A. J., 28, 163, 168, 169,
172, 176, 178, 401, 402
Lampel, A. K., 236, 391
Langer, E. J., 300, 391
Langston, W. E., 81, 387
LaPiere, R. T., 304, 391
Latane, B., 55, 388
Lazarus, R. S., 82, 154, 328, 391
LeDoux, J. E., 310, 391
Lee, A. Y., 70, 148, 302, 303, 380
Lee, H. K., 291, 393
Lefcourt, H. M., 191, 391
Leggett, E. L., 4, 286, 299, 385
Lehman, D. R., 290, 292, 388
Leirer, V. O., 10, 388
Lepore, L., 304, 391
Lepper, M. R., 102, 256, 257, 285,
396
Lerner, M. J., 66, 269, 391
Levine, J., 205, 402
Levine, S. R., 307, 325-327, 355, 391
Liang, C. H., 146, 147, 393
Liberman, A., 48, 383
Lichtenstein, E. H., 19, 383
Lichtenstein, M., 168, 399
Lindem, K., 85, 387
Lindsay, D. S., 54, 280, 390
Locke,H. L., 296, 391
Lockhart, R. S., 162, 282, 384
Loftus, E. F., 10, 24, 39, 113, 138,
146, 384, 389, 391, 392
Loftus, J., 38, 132, 352, 391
Logan, G. L., 36, 40, 392
Loken, B. A., 11, 122, 392
Lombardi, W., 41, 50-52, 55, 69, 199,
381, 389, 392
Long, D. L., 191, 392

Loomes, G., 302, 392
Lord, K. R., 236, 393
Lowell, E. L., 204, 210, 393
Lurie, L., 12, 389
Lutz, M. F., 17, 109-111, 396
Lutz, R. J., 334, 382
Lynch, J. G., 35, 392
M
Mackie, D. M., 307, 342, 343, 356,
392
MacKinnon, C., 164, 386
Macrae, C. N., 31, 47, 56, 303, 304,
382, 392
Maeson, W. A., 191, 193, 391
Magliano, J. P., 28, 81, 392, 403
Malone, B., 164, 385
Mandler, J., 4, 19, 392
Mankowski, E., 4, 144, 148, 288-290,
392
Marburger, W., 113, 392
Mark, M. M., 356, 357, 398
Markus, H., 290-292, 388, 392
Martin, L. L., 55, 166, 168, 169, 308,
312, 324, 344-349, 363, 392, 399,
401, 402
Martin, R. A., 191, 391
Mayer, J. D., 318, 383
McAdams, D. 4, 144, 149, 392
McClelland, D. C., 204, 210, 343
McConnell, H. K., 276, 382
McFarland, C., 279, 396
McGhee, P., 191, 387, 393
McGlynn, R. P., 207, 397
McGuire, C. V., 118, 120, 270, 352,
362, 393
McGuire, W. J., 60, 118, 119, 272,
294, 295, 352, 362, 367, 370, 373,
393
Mellers, B., 302, 393
Melton, R. J. 251, 253, 401
Menon, G., 283, 284, 393
Mertz, E., 307, 390
Metzler, J., 81, 398
410
AUTHOR INDEX
Meyer, M., 85, 387

Miller, C., 191, 194, 393
Miller, D. T., 66, 258, 260, 269, 299,
390, 391, 393
Miller, L. C., 295, 393
Miller, P. J., 4, 145-147, 290, 393
Milne, A. B., 56, 392
Mindess, H., 191, 194, 393
Miniard, P. W., 236, 358, 393
Mintz, J., 145, 146, 393
Mitra, A., 35, 392
Monroe, K. B., 47, 63, 380
Monteiro, K. P., 316, 318, 320, 382
Morgan, M., 281, 387
Morris, M. W., 47, 69, 70, 72, 148,
291, 299, 302, 303, 383, 389, 393
Moskowitz, G. B., 63, 394
Mueller, J. H., 316, 318, 319, 394
Murray, N., 341, 347, 394
Murray, S. L., 4, 295, 389, 394
Murray, S. T., 308, 324, 402
Mussweiler, T., 64, 65, 394
Myers, J. L., 39, 381
N
Naderer, G., 262, 263, 282, 283,
397
Nakamura, G. V., 117, 124, 126,
387
Nassi, A. J., 164, 394
Nehrhardt, G., 197, 394
Nelson, K., 4, 145, 290, 394
Newman, L. S., 66, 382
Newtson, D. A., 103, 115, 394
Niedenthal, P. M., 306, 312, 313,
317, 318, 320, 335, 343, 389, 394
Nielsen, A. C., 280, 394
Nisbett, R. E., 255, 280, 291, 294,
307, 384, 394
Norenzayan, A., 262, 291, 294, 295,
384, 394
Norman, D. A., 42, 394
Nottenburg, G., 18, 114, 395
Nowicki, G. P., 307, 349, 390
O
OGuinn, T. C., 54, 101, 280, 281,
395, 398
Oishi, S., 4, 292, 293, 385
Ortony, A., 308, 395
Ostrom, T. M., 35, 65, 395
Ottati, V. C., 307, 337, 395
P
Palmer, S. J., 138, 392
Parducci, A., 35, 65, 395
Park, J. W., 24, 58, 59, 167, 395
Parrott, G., 316, 317, 395
Pavelchak, M.A., 36, 309, 386
Peng, K., 291, 393
Pennington, N., 4, 232-234, 255,
388, 395
Peters, M. D., 307, 309, 380
Peterson, D., 163, 386
Petty, R. E., 55, 307, 344, 346, 357,
383, 395, 400
Pham, M. T., 307, 327, 395
Picek, J. S., 122, 395
Pietromonaco, P., 53, 57, 62, 381
Powell, M. C., 314, 315, 385
Pracejus, J., 307, 395
Pratkanis, A. R., 352, 308
Pratto, F., 309, 314, 315, 381
Prinz, W., 75, 395
Pylyshyn, Z. W., 9, 81, 395
Pyszczynski, T. A., 66, 395
R
Rabe, C., 339, 340, 342, 351, 382
Radvansky, G. A., 17, 29, 81, 95, 98,
100, 108-111, 136, 396, 402, 403
Raghubir, P., 283, 284, 393
Rappoport, J., 4, 144, 148, 392
Ratcliff, R., 39, 396
Raymond, P., 314, 381
Read, D., 396
Read, S., 19, 259, 295, 382, 393
Reckman, R. F., 275, 276, 387
AUTHOR INDEX
Reeder, G. D., 163, 386
Regan, D., 261, 396
Reich, C. M., 21, 122, 380
Reid, L. D., 304, 396
Rempel, J. K., 309, 325, 360, 402
Renwick, S., 164, 385
Reyes, R. M., 29, 255, 396
Reynolds, R. E., 51, 381
Rhee, E., 291, 396
Rholes, W. S., 47, 49, 52, 144, 321,
389
Riggle, E. J., 251, 253, 401
Rips, L. J., 113, 383
Ritov, I., 302, 393
Rittenauer-Schatka, H., 282, 397
Robinson, G., 307, 390
Roselli, F., 356, 396
Rosen, N. A., 272, 295, 396
Rosenfield, D., 294, 399
Ross, L. 102, 133, 255-257, 280, 285,
394, 396
Ross, M., 4, 274, 275, 279, 384, 396
Rothbart, M., 168, 399
Rubin, D. C., 3, 396
Russell, D., 289, 384
Rusting, C. L., 317, 397
S
Sabini, J., 316, 317, 395
Sacks, R., 287, 389
Salancik, G. R., 261, 401
Sanbonmatsu, D. M., 314, 315, 387
Sandel, T. L., 146, 393
Sanford, A. J., 28, 381
Sanna, L. J., 285, 396
Santioso, R., 279, 396
Sawyer, J. D., 126, 127, 387
Schachter, S., 310, 397
Schaeken, W., 81, 397
Schallert, D. L., 51, 381
Schank, R. C., 3, 18, 28, 116, 126,
146, 149, 293, 340, 397
Schick, C., 267, 397
Schneider, W., 7, 397
Schooler, J. W., 138, 397
411
Scott, C. K., 28, 397
Schwartz, A., 302, 393
Schwarz, N., 8, 35, 55, 64, 262, 263,
280, 282, 283, 285, 304, 307, 308,
311, 312, 322-328, 338-340, 342,
343, 347, 350, 351, 354-356, 382,
384, 391, 394, 396, 397,399
Semenik, R. J., 280, 398
Sentis, K. P., 122, 397
Seta, J. J., 55, 392
Setterlund, M. B., 306, 312, 313, 317,
318, 335, 343, 394
Shalker, T. E., 311, 390
Shanteau, J. A., 334, 400
Shavitt, S., 251, 253, 401
Sheehan, P. W., 250, 398
Shepard, R. N., 81, 398
Sheppard, L. A., 307, 351, 382
Sherman, B. R., 279, 398
Sherman, J. W., 132, 303, 304, 388,
398
Sherman, S. J., 122, 256, 280, 395,
398
Shevell, S. K., 113, 393
Shiffrin, R. M., 7, 122, 395, 397
Shiv, B., 66, 327, 338, 398
Shoben, E. J., 18, 113, 114, 395, 402
Shurcliff, A., 201, 398
Shrum, L. J., 54, 101, 280, 281, 395,
398
Signorielli, N., 281, 387
Simmons, C. H., 66, 269, 391
Simons, A., 282, 397
Simonson, I., 69, 70, 72, 148, 302,
303, 383
Sinclair, R. S., 356, 357, 398
Singer, J., 310, 397
Singer, M., 28, 81, 231, 387
Skowronski, J. J., 240, 398
Skov, R. B., 256, 398
Skurnik, I. W., 63, 394
Slovic, P., 280, 390
Small, E. M., 284, 396
Smith, E. R., 6, 22, 26, 39, 166, 235,
307, 339, 383, 398
Smith, S. M., 307, 344, 357, 400
412
AUTHOR INDEX
Snyder, M., 331, 398

Snyder, M. L., 294, 331, 399
Sokolowska, J., 70, 400
Sperber, D., 153, 217, 260, 399
Sperling, G., 29, 399
Spiro, R. J., 231, 269, 270, 399
Srull, T. K., 5, 8-10,12, 26-29, 31, 32,
34, 36, 37, 40-43, 47, 50, 52, 57, 61,
62, 65, 69, 89, 91, 103, 117, 128,
163, 166-169, 173, 280, 317, 320,
350, 366, 367, 399, 401, 402
Staelin, R., 236, 385
Stapel, D. A., 52, 63, 321, 399
Stein, B. S., 108, 383
Stein, N. L., 4, 19, 383, 399
Steinmetz, J., 102, 256, 396
Stephan, W. G., 294, 399
Stepper, S., 308, 324, 399
Stock, C. B., 256, 398
Stocker, S. L., 285, 396
Storms, M., 261, 399
Strack, F., 8, 35, 55, 64, 65, 102, 262,
263, 282, 283, 307, 308, 324, 342,
343, 354, 382, 394, 396, 397, 399
Sugden, R., 302, 392
Sujan, H., 341, 347, 394
Sujan, M., 341, 347, 394
Suls, J. M., 191, 193-195, 200, 204,
208, 210, 387, 390, 399, 400
Swan, S., 153, 175, 178, 181, 183,
185, 196, 401, 402
T
Tannen, D., 183, 400
Taylor, H. A., 81, 400
Taylor, S. E., 47, 217, 400
Tedeschi, J. T., 191, 390
Tesser, A., 342, 344, 392, 400
Thompson, W. B., 316, 318, 319, 394
Thompson, W. C., 29, 255, 396
Thurstone, L. L., 322, 358, 400
Tong, J. Y., 287, 384
Toppin, T., 54, 101, 280, 388
Tota, M. E., 51, 52, 381
Totten, J., 261, 396
Trabasso, T., 28, 81, 231, 387

Trafimow, D. A., 18, 126-130, 135,
199, 400
Trafton, J. G., 132, 391
Triandis, H. C., 290, 291, 400
Trope, Y., 36, 52, 367, 384, 400
Troutman, M. C., 334, 400
Turek, J., 191, 194, 393
Turner, T. J., 18, 85, 126, 259, 382
Tversky, A., 35, 64, 257, 258, 263,
264, 280, 390, 400
Tversky, B., 81, 400
U
Uleman, J., 291, 396
Unnava, H. R., 236, 393
Unverzagt, W. H., 13, 32, 167, 169,
402
Upshaw, H. S., 35, 65, 395, 400
V
Van der Plight, J., 63, 399
Van Knippenberg, A., 47, 385
Velton, E., 319, 400
W
Wnke, M., 28, 163, 165, 283, 285,
400
Wallace, K., 296, 391
Walster, E., 66, 400
Ward, D. W., 312, 345-349, 363, 392
Watzlawick, P., 160, 186, 400
Weary, G., 264, 292, 388
Weber, E., 70, 400
Wegener, D. T., 307, 344, 357, 400
Wegner, D. M., 31, 56, 400
White, R. W., 205, 400
Wiley, A. R., 146, 393
Williams, K., 55, 388
Wills, T. A., 218, 401
Wilson, D., 153, 217, 260, 399
Woelke, M., 339, 340, 34 2, 351, 382
Wolfson, M. R., 261, 401
413
AUTHOR INDEX
Wood, W., 48, 384
Woolam, D., 207, 396
Worth, L. T., 307, 342, 343, 356, 392
Wyer, R. S., 4, 5, 8-10, 12, 13, 16-19,
21, 22, 24, 26-29, 31, 32, 34-37, 3943, 47, 50, 52, 54, 57-62, 64, 65, 6772, 74, 75, 86-89, 91, 95, 98, 100,
101, 103, 104, 109, 111, 113-122,
124, 126-130, 132-136, 139, 142144, 151, 153, 157-159, 161-163,
165-167, 172, 173, 176, 178-181,
183, 185-187, 198, 199, 201-203,
205-207, 212, 215, 218, 221, 223,
224, 236, 239-241, 243-251, 253,
269-274, 277, 280, 281, 288-293,
295, 302, 304, 307-309, 311, 312,
314, 317, 320, 322, 325-329, 331,
332, 335-338, 345-351, 353, 355,
359, 360, 362, 363, 366, 367, 380,
382-384, 386-392, 395-401
Y
Yamauchi, H., 292, 294, 402
Yeung, C. W. M., 328, 329, 331, 332,
350, 402
Yoon, S. O., 24, 58, 59, 395
Youmans, J. E., 164, 394
Z
Zacks, R. T., 29, 108-110, 396,
Zajonc, R. B., 210, 308, 324, 327, 402
Zanna, M. P., 66, 309, 325, 360, 402
Zigler, E., 205, 402
Zillman, D., 191, 193, 201, 203, 218,
282, 402, 403
Zinn, H., 148, 403
Zuckerman, M., 264, 292, 403
Zwaan, R. A., 28, 81, 392, 403
Subject Index
A
Accessibility of knowledge in
memory
determinants of, 41, 48-49
frequency of use, 49-52
recency of use, 49-52
effects on,
of group membership
salience, 69-72
of prior judgments, 63-64
of thought suppression, 56-57
effects of,
on activation of bipolar
concepts, 57-60
on assimilation and contrast,
57-60, 62-63, 65
on avoidance of negative
outcomes, 69-72
on belief formation, 60-62
on beliefs in a just world, 6669
on comprehension time, 97-98
on goals and motives, 65-72
on interpretation of information, 57-60
on own behavior, 72-75
on perceptions of social reality,
280-282
on reactions to rape, 66-69
on selective attention, 57-60
on use of stereotypes, 70-72
on verification, 98-100
generalizeability of effects, 52-53
nonconscious vs. conscious
effects of, 53-56
Adjustment for bias, 55-56, 336-338
Affect
adjustments for influences of,
336-338
automatic influence of,

on use of categorical criteria,
339-343
definition of, 307-312
vs. cognitions, 309-310
vs. emotions, 308-309
vs. evaluations, 309
vs. descriptive bases for judgment, 325-332
effects of,
on appraisals, 328-332
on interpretation of ambiguous
information, 320-321, 332-337
on political judgment, 335-336
on recall of previously acquired
knowledge, 315-318
on selective attention and
encoding, 318-319
on semantic concept accessibility, 312-314
influence of,
on attitudes and behavior, 359363
on behavior in nonlaboratory
situations, 357-358
on communication and
persuasion, 356
on creativity, 348-349
on impression formation, 349351
on life satisfaction, 352-354
on self-evaluation, 352, 354-356
on use of stereotypes, 351-352
motivational influences of, 342-344
performance-feedback model of,
344-348
and procedural knowledge, 311
as reactions to cognitions, 310
role of narratives in, 324-325
use as informational basis for
judgment
415
416
SUBJECT INDEX
of competence and self-esteem,

325-327
of consumer products, 327-332
Affect-confirmation processes, 332335
Amusement; see Humor elicitation
Assimilation and contrast
and activation of bipolar
concepts, 57-60
and comparative judgment, 63-65
in interpretation of information,
57-60
and use of comparative standards,
62-65
Associative processes
in construction of event representations, 116-123
Attribution processes,
effects of perspective on, 260-261
Automatic evaluation effect, 314-315
Automaticity; see Conscious vs.
nonconscious processes
Awareness; see also Conscious vs.
nonconscious processes
effect on avoidance of bias, 55-56
of group identity, 70-72
B
Belief formation and change
formal models of, 60-63
role of knowledge accessibility
on, 60-63
C
Communication goals
effects on memory, 137-143
pre- vs. post-information, 137-143
Comparative judgments
effects of, on absolute judgments,
63-65
Completion principle of spontaneous inference, 268-270
Comprehender, functions of, 89-92,
106-110
Comprehension; see also Episode

models
effect of social context on, 151-188
pragmatic influences on, 151-188
priming effects on, 97-98
of observed events, 106-116
role in humor elicitation, 161162, 198-209
role of visual imagery in, 81, 83-88
of situation specific events, 106-116
spontaneous,
theoretical formulation of, 88-95
of verbal statements, 92-97
and validation, 98-103
Conscious vs. nonconscious
processes
conceptualization of, 36-37
in knowledge accessibility, 53-56
Conversational norm violations
effects of,
on attitudes, 158-159
on belief formation, 155-158
on construction of narratives,
260
in communication and
persuasion, 159
on humor elicitation, 161-162
on impression formation, 178186
on information seeking, 162165, 184-186
on interpretation of emotions,
186-188
on liking for the communicator,
178-186
on memory, 175-178
on responses to news, 157-159
in informal conversation, 165-186
normative influences on, 178-186
in social interaction, 178-188
spontaneous identification of,
153-157
types of,
conversational style, 178-186
informativeness, 155-160
modesty, 175-178
417
SUBJECT INDEX
politeness, 175-178
truthfulness, 160-165
Conversational style, effects of
on information seeking, 184-186
on liking for the speaker, 180-186
on perception of message implications, 179-180
Cultural identity, effects of
on avoidance of negative outcomes, 72
Cultural norms and values
activation of, 66-67, 72
effect of,
on avoidance of negative outcomes, 72
on implicit theories, 290-295
on representation of self in
memory, 144-147
on prediction of future outcomes, 292-294
on regulatory focus, 72
on self-identity, 144-149
on socialization practices, 144149
on theories of personal causality, 292-294
D
Dual-processing models, 35-36
E
Ease of retrieval, role of
in hindsight bias, 285
in perception of self and others,
282-284
in perception of social reality,
280-282
Emotions; see also Affect
communication of, in close relationships, 186-188
anticipation of, in decisionmaking, 301-303
Episode models; see also Situation
models
construction of, 106-116

effects of familiarity on, 115-116
effect of thematic relatedness
on, 110-112
vs. generalized event representations, 116-123
number formed of, 106-112
temporal coding of, 112-115
Event prototypes; see Generalized
event representations
Event representations; see
Generalized event representations; see Episode models
Expectation, deviations from; see
also Conversational norm
violations
role in humor elicitation, 161162, 198-211
Explanation and prediction
contextual influences on, 161-263
motivational influences on, 263265
role of narratives in, 255-258,
261-265
G
Generalized entity representations
construction of, 23-25, 166-168
Generalized event representations
construction of, 116-123, 130-133
socialization influences on,
144-149
effect of,
on comprehension of others
experiences, 125-130, 135-137
on comprehension of own
experiences, 133-135
on memory for specific experiences, 137-144
Group membership salience
effects on behavioral decisions,
69-72
effects on regulatory focus, 69-72
418
SUBJECT INDEX
H
Humor elicitation
comprehension processes in, 198204
comprehension difficulty,
effects of, 204-207, 218-220
situational and individual
differences in, 207-208
diminishment, role in, 195-198
effects of disparagement on, 218227
effects of distraction on, 210-211
effects of repetition on, 210-211
elaboration processes in,
and responses to sexist humor,
215-217
situational influences on, 212215, 220-227
theoretical effects of, 209-211,
220-222
role of stereotypes in, 218-227
by sexist jokes, 222-227
source effects on, 218-227
attributions of motivation and,
223-224
theories of, 192-209
Apters reversal theory, 195-199
arousal reduction theories,
192-193
comprehension-elaboration
theory, 198-209
incongruity resolution
theories, 193-194
superiority and disparagement theories, 193
I
Imagery; see Visual imagery
Images of politicians
influence of,
on information processing,
240-250
on interpretation of issue
stands, 251-250
Implicational molecules
construction of, 121-122
examples of, 267-270
Implicit theories
cultural differences in, 290-295
effects of, on causal attributions, 291-292
effects of, on predictions of the
future, 292-294
of ease of retrieval, 280-285
effects of perspective on, 288-290
incremental vs. entity, 286-288
motivational influences on, 278279
of personal achievement, 286-287
of personal causality, 292-294
role of,
in anticipated emotional
reactions, 301-303
in behavioral decision making,
301-303
in helping decisions, 299-301
in hindsight, 285
in marital satisfaction, 295-299
in perception of social reality,
280-282
in perception of social support,
288-290
in personal achievement, 286288
in prevention and promotion
focus, 301-303
in reconstructing the past, 274275
in self perception, 275-276
in stereotype-based behavior,
303-305
Impression formation
of communicators, 168-175
effect of social context on, 165174
in get-acquainted conversations,
180-186
in nonsocial situations, 166-167
of politicians, 248-254
419
SUBJECT INDEX
role of affect in, 349-351
role of conversational norms in,
165-186
role of visual imagery in, 140-254
Information processing models
conceptualization of
comprehension, 89-92
conscious vs. nonconscious
processes, 36-37
dual processing, 35-36
free flow of thought, 43-45
knowledge accessibility, 41
reminding, 41-45
storage and retrieval processes, 37-41
Wyer-Srull (1989) theory, 26-37
limitations of, 26-29
modifications of, 22, 89-92
role of consciousness in, 36-37
Information processing strategies
piecemeal vs. holistic, 234-253
effect of presentation format on,
234-235
effect of visual images on, 236253
Informativeness principle
in emotional communication,
186-188
in responses to news, 157-159
in social communication, 155160, 180-196
Innuendo effects, 157-159
J
Just world, belief in
as an implicational molecule,
122, 268
role, in reactions to rape, 66-69
M
Marital relationships
communication of emotions in,
186-188
role of implicit theories in, 295-299
satisfaction with, 295-299

Memory; see also Reminding
effect of communication on, 137143
effect of story-telling on, 144-149
for observed experience, 137-143
retrieval processes, 39-41
goal-directed, 42
resonance metaphor, 39-40
storage of information in, 37-39
N
Narratives; see also Episode Models
construction of,
framing effects on, 258-259
informational influences on,
258-260
motivational influences on,
263-265
perspective effects on, 260-263
situational influences on, 144149, 260-263
socialization influences on,
144-149
ease of constructing, 257-258
effects of,
on social identity, 144-150
vs. implicational molecules, 121123
representation in memory of,
116-123, 144-149
associative processes in, 116123
cultural influences on, 144-149
social interaction processes in,
144-149
role in affect elicitation, 324
role in daily life, 3-4
types of,
histories, 20
stories, 19-20
implicational molecules, 20-21,
121-123
implicit theories, 20
use of,
420
SUBJECT INDEX
in explanation and prediction,

255-258, 261-265
in jury decision making, 232-234
in political impression formation,
240-251
in social judgment, 231-266
Narrative vs. list-based processing,
232-251
P
Parent-child interaction
effects on narrative construction,
144-149
Perception of social reality, 54, 280282
Perspective, effects of
on attribution of responsibility,
260-261
Pragmatic influences on comprehension; see Conversational
norms
Probabilistic inference,
formal model of, 270-274
Procedural knowledge,
in affect-based information
processing, 311-312
R
Regulatory focus,
chronic differences in, 69-70
cultural influences on, 69-70, 7273
effects of, on anticipated decision
outcomes, 301-303
effects of group membership
salience on, 70-72
situational influences on, 70-72
Reminding,
effects on later memory, 142-143
goal directed, 42-43
spontaneous, 43-45
Resonance metaphor,
in retrieval of information, 39-40
S
Self and self-knowledge
cultural differences in, 144-149
representation in memory of,
133-136
and social identity, 144-149
socialization influences on, 144149
and story-telling, 144-145
Self-perception,
theory of, 275-276
role of affect in, 352-356
Self-referent statements,
effects on person impressions,
175-178
Self-serving bias,
in implicit theories, 178-179
Situation models; see also Episode
models
content and structure of, 80-83
of events, 81-83
role of visual imagery in, 81, 83-88
Social knowledge
approach to investigating, 14-16
generalized entity representations of, 22-24
generalized event representations of, 17-21, 116-123
histories, 20
stories, 19-20
implicational molecules, 20-21,
121-123
implicit theories, 20
independent storage of, in
memory, 37-39
modalities of, 9
representation in memory of, 1014
referents of, 7-8
situation models of, 16-17
types of, 6-7
validation of, 98-103
Social support, perceptions of, 288290
421
SUBJECT INDEX
Socialization, influence of
on content of self-narratives, 144149
on social identity, 144-145, 148149
Stereotypes,
nonconscious activation of, 7375
role in behavioral decision
making, 303-305
role in humor elicitation, 218-227
suppression of, 56-57
Sufficiency principle,
Role in comprehension and
judgment, 48
T
Thought suppression, effects of, 5657
Truthfulness, perception of, 98-104,
153-154
Truthfulness principle, effects of
violating
on humor elicitation, 161-162
on information seeking, 163-165
Validation of information, 98-103

Visual imagery
effects of,
on information processing
strategies, 251-254
on interpretation of information, 251-254
on processing of new information, 251-253
on verbal information processing, 236-244
individual and situational
differences in, 249-251
interfering vs. facilitative effects
of, 236-244, 246-249
role of,
in construction of situation
models, 81
in narrative-based vs. listbased inferences, 235-244,
246-251
in text comprehension, 84-85
W
Witticisms, comprehension of, 161163

Social Comrehension and Judement

Transféré par

Informations du document

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Social Comrehension and Judement

Transféré par

Droits d'auteur :

Formats disponibles

Social Comprehension and Judgment

The Role of Situation Models,

Social Comprehension and Judgment

Robert S. Wyer, Jr.

LAWRENCE ERLBAUM ASSOCIATES, PUBLISHERS

This edition published in the Taylor & Francis e-Library, 2008.

Lawrence Erlbaum Associates, Inc., Publishers

Cover design by Kathryn Houghtaling Lacey

Library of Congress Cataloging-in-Publication Data

Books published by Lawrence Erlbaum Associates are printed on acid-free paper,

PART I: INTRODUCTION AND BASIC CONCEPTS

THE ARCHITECTURE OF THE INFORMATION-PROCESSING

DETERMINANTS AND EFFECTS OF KNOWLEDGE

PART II: COMPREHENSION PROCESSES

THE COMPREHENSION AND VALIDATION OF INFORMATION

THE CONSTRUCTION OF EPISODE MODELS AND GENERALIZED

THE IMPACT OF GENERALIZED EVENT REPRESENTATIONS ON

PRAGMATIC INFLUENCES ON THE INTERPRETATION

THE DYNAMICS OF HUMOR ELICITATION: THE EFFECTS

PART III: INFERENCE PROCESSES

CRIMES, VACATIONS, AND POLITICAL CANDIDATES:

THE IMPACT OF IMPLICATIONAL MOLECULES AND IMPLICIT

THE ROLE OF AFFECT IN INFORMATION PROCESSING

EPILOGUE: THE BOOK, THE AUTHOR, AND PHILOSOPHICAL

APPENDIX: SUMMARY OF POSTULATES

This book provides a conceptual integration of this work. It proposes

from previously acquired information about the informations referents

Human memory is a collection of thousands of stories we remember

has not been conducted in controlled laboratory settings. In fact, most of

about specific events. Finally, chapter 11 addresses an issue largely ignored

TYPES OF SOCIAL KNOWLEDGE

are later unable to remember whether actors statements to one another

THE REPRESENTATION OF KNOWLEDGE

much insight into the impact of this information on judgments. Rather,

GENERALIZED EVENT REPRESENTATIONS

The Dimensionality of Event Representations

Abstractness. A narrative representation could be encoded in terms of

can form routinized sequences events that are specific to themselves, or

GENERALIZED ENTITY REPRESENTATIONS

empirically. One set of assumptions, espoused by Hastie and Kumar (1979)

based on information obtained at different times, might also be formed and

A complete theoretical formulation of information processing must specify

ARCHITECTURE OF THE INFORMATION-PROCESSING SYSTEM

The Nonverbal Coding of Social Events

ARCHITECTURE OF THE INFORMATION-PROCESSING SYSTEM

STRUCTURE OF THE PROCESSING SYSTEM

ARCHITECTURE OF THE INFORMATION-PROCESSING SYSTEM

The Original Model. Wyer and Srull conceptualized long-term memory

ARCHITECTURE OF THE INFORMATION-PROCESSING SYSTEM

In the revised theory, therefore, we drop the construct of bins. Instead,

ARCHITECTURE OF THE INFORMATION-PROCESSING SYSTEM

sons likeableness from a pre-existing evaluation of the person to which this

ARCHITECTURE OF THE INFORMATION-PROCESSING SYSTEM

STORAGE AND RETRIEVAL PROCESSES

This assumption has broad implications. For example, it suggests that if

more general, semantic concepts that can be communicated verbally. This

ARCHITECTURE OF THE INFORMATION-PROCESSING SYSTEM

be assigned 10 different strengths, could potentially represent 1010 unique

where n(k) is the number of features in the representation and n( p I k ) is the

Postulate 2.3a is consistent with the assumptions of the independent

ARCHITECTURE OF THE INFORMATION-PROCESSING SYSTEM

Determinants of Knowledge Accessibility

GOAL-DIRECTED VERSUS NON-GOAL-DIRECTED

Goal-Directed Retrieval Processes