Vous êtes sur la page 1sur 416

Consciousness and

Advances in Research

A Continuation Order Plan is available for this series. A continuation order will bring
delivery of each new volume immediately upon publication. Volumes are billed only upon
actual shipment. For further information please contact the publisher.
Consciousness and
Advances in Research


Edited by
Yale University

University of California, Los Angeles


Library of Congress Cataloging in Publication Data
Main entry under title:
Consciousness and self-regulation.
Includes bibliographical references and index.
1. Consciousness. 2. Self-control. I. Schwartz, Gary E., 1944-
II. Shapiro, David, 1924-
BF311.C64 153.8 76-8907
ISBN-13: 978-1-4684-2570-3 e- ISBN -13: 978-1-4684-2568-0
DOl: 10.1007/978-1-4684-2568-0

1976 Plenum Press, New York

Softcover reprint of the hardcover 1st edition 1976
A Division of Plenum Publishing Corporation
227 West 17th Street, New York, N.Y. 10011
All rights reserved
No part of this book may be reproduced, stored in a retrieval system, or transmitted,
in any form or by any means, electronic, mechanical, photocopying, microfJlming,
recording, or otherwise, without written permission from the Publisher
Articles Planned for Future Volumes
, ,
Gyorgy Adam
Interoception, Awareness, and Behavior
Bernard Glueck and Charles Stroebel
Transcendental Meditation: Comparison to EEG Biofeedback
Jerre Levy
Brain and Consciousness: Cerebral Asymmetry
A. R. Luria
Brain and Consciousness: Functional Systems Approach
Wesley Lynch
Biofeedback: Temperature Regulation
F. J. McGuigan
Imagery and Thinking: The Motor System
Martin T. Orne
EEG Biofeedback: Relationship to Anxiety
Robert Ornstein
Dual Modes of Consciousness
Kenneth S. Pope and Jerome Singer
Regulation of the Stream of Thought
Larry Roberts
Biofeedback: Use of Curare
Judith Rodin
Perception and Externality: Obesity
Harold Sackeim and Rubin Gur
Sel/Confrontation, Sel/Deception, and Consciousness
Bernard Tursky and Milton Lodge
Subjective Experience: Psychophysics, Applications to
Assessment of Pain, and Political Opinion
Takami Wananabe
Meditation: Japanese Research
Matisyohu Weisenberg
Sel/Regulation Therapies: Pain
Norman Zinberg
Drugs: Interaction of Set and Setting

THOMAS D. BORKOVEc, Department of Psychology, University of Iowa,

Iowa City, Iowa
MONTE BUCHSBAUM, Unit on Perceptual and Cognitive Studies, Adult
Psychiatry Branch, Division of Clinical and Behavioral Research,
National Institute of Mental Health, Bethesda, Maryland
THOMAS H. BUDZYNSKI, Department of Psychiatry, University of Colo-
rado Medical School and Biofeedback Institute of Denver, Denver,
DAVID B. COHEN, Department of Psychology, University of Texas,
Austin, Texas
DAVID R. ENGSTROM, Department of Psychiatry & Human Behavior
and Student Health Service, University of California, Irvine,
ERNEST R. HILGARD, Department of Psychology, Stanford University,
Stanford, California
E. Roy JOHN, Departments of Psychiatry and Physiology, New York
Medical College, New York, New York
DONALD MEICHENBAUM, Department of Psychology, University of
Waterloo, Waterloo, Ontario, Canada
KARL H. PRmRAM, Department of Psychology, Stanford University,
Stanford, California


The first and foremost concrete fact which every one will affirm to belong to
his inner experience is the fact that consciousness of some sort goes on. I
-William James, 1893

We are witnessing today a mounting interest among behavioral and

biological scientists in problems long recognized as central to our
understanding of human nature, yet until recently considered out of
the bounds of scientific psychology and physiology. Sometimes
thrown into the heading of "altered states of consciousness," this
growing research bears directly upon such time-honored questions as
the nature of conscious experience, the mind-body relationship, and
volition. If one broadly views this research as encompassing the two
interrelated areas of consciousness and self-regulation, one can find
many relevant contemporary examples of creative and experimentally
sophisticated approaches, including research on the regulation of
perception and sensory experience, attention, imagery and thinking,
emotion and pain; hypnosis and meditation; biofeedback and volun-
tary control; hemispheric asymmetry and specialization of brain func-
tion; drug-induced subjective states; and biological rhythms. Because
the material is spread over many different kinds of publications and
disciplines, it is difficult for anyone person to keep fully abreast of the
significant advances. The overall aim of the new Plenum Series in
Consciousness and Self-Regulation: Advances in Research is to provide a
scholarly forum for discussing integration of these diverse areas by
presenting some of the best current research and theory.
It is our hope that these volumes will enable investigators to

I William James, Psychology: Briefer Course (New York: Henry Holt and Company, 1893),


become more well-rounded in related areas of research, as well as

provide advanced students with a ready means of obtaining up-to-
date, state-of-the-art information about relevant problems, theories,
methods, and findings. By selecting significant developments in
theory and research, we also hope that over the years the series can
help legitimate the field as a scientific venture as well as delineate
critical issues for further investigation.
Psychology and biology are going through a reawakening, and
research on the issues to which this series is devoted is helping to
bring these fields closer together. History tells us that Wundt founded
psychology as the science of consciousness, and James expanded it to
encompass "such things as sensations, desires, emotions, cognitions,
reasonings, decisions, volitions and the like."2 But these ideals could
not be achieved, or so it seemed, and psychology turned away from
questions of experience and volition, as well as from biology, and was
replaced with behaviorism. The transformation was arduous, and it
required a certain allowance for inconsistency. For example, Edmund
Jacobson, one of the pioneers in the psychophysiology of higher
mental processes, recalled, "Lashley told me with a chuckle that when
he and Watson would spend an evening together, working out
principles of behaviorism, much of the time would be devoted to
In William James: Unfinished Business (1969), Mandler summarized
the good points, and the bad points, of this era of psychology in his
"Acceptance of Things Past and Present: A Look at the Mind and the
Brain." He aptly noted:
I think the Watsonian behaviorist development was inevitable-I
think it was even healthy-if we learn not to do it again. Watson and the
behaviorists did, once and for all, clean up the problem of the proper data
language for psychology. In that sense, we are all behaviorists. The
behaviorists inveighed against an establishment which imported theoreti-
cal notions and hypotheses into purely descriptive realms of psychology.
They successfully excluded vague notions about the causes of behavior-
the introspective statements-from the facts of psychology. But in the
process the Watsonians felt called upon to do the reverse and to remove
complex and imaginative models from psychology . . . . Behaviorism has
been one of the most antitheoretical movements in science ....

2 Ibid., p. 1.
3 Jacobson, "Electrophysiology of Mental Activities and Introduction to the Psychologi-
cal Process of Thinking." In F. J. McGuigan and R. A. Schoonover (Eds.), The
Psychophysiology of Thinking (New York: Academic Press, 1973), p. 14.

... I submit that it was this anti theoretical stance that prevented any
close attention to physiology . . . . If the mechanisms we postulate are
"like" physiological mechanisms, then we will have heeded James in
modem terms. But if we are, as we were, afraid to postulate complex
mental mechanisms, we will never find the corresponding complex phy-
siological mechanisms.'

This series is dedicated to William James, emphasizing the

integration and patterning of multiple processes, coupled with the
most significant advances in methodology and knowledge. Some of
the chapters will be broad-based and theoretical; others will focus on
specific research problems or applications. Inclusion of material in all
cases is determined by the investigator's focus on or concern with
consciousness and related processes, whether in normal or in abnor-
mal populations. While the editors have a decided bias toward
biologically oriented approaches to consciousness and self-regulation,
papers that deal primarily with cognition or self-report are included
when of particular significance to these topics. Since important find-
ings in this area are often derived from the study of clinical popula-
tions and are of direct relevance to the assessment and treatment of
psychological and psychophysiological disorders, chapters dealing
with basic research are interwoven with chapters of more clinical
concern. In this way it is hoped that the series can provide a fertile
interchange between the basic and applied sides of this area. To help
the reader understand the perspective and rationale for the diverse
selections comprising a given volume, a brief overview of each volume
is presented by the editors.
The impetus for and organization of the series grows out of
student response to our interdiSCiplinary seminars at Harvard on the
psychophysiology of consciousness, emotion, and self-regulation, cou-
pled with the enthusiasm and support of Seymour Weingarten, Senior
Editor of Plenum. Their input, and prodding, is gratefully acknowl-

4 G. Mandler, "Acceptance of Things Past and Present: A Look at the Mind and the
Brain." In R. B. MacLeod (Ed.), William James: Unfinished Business (Washington, D.C.
American Psychological Association, 1969), pp. 13, 14.
Overview of Volume 1

In "A Model of Consciousness," E. R. John presents the thesis that

"'mind,' under which rubric are subsumed such phenomena as
consciousness, subjective experience, the concept of self, and self-
awareness, is an emergent property of sufficiently complex and appro-
priately organized matter." John outlines seven levels of information
processing in the brain that correspond to sensations, perceptions,
consciousness, content of consciousness, subjective experience, self,
and self-awareness. He presents electrophysiological data on both
lower animals and man in support of this classification. Based on
these findings, he postulates the existence of unique "hyperneurons"
in the brain reflecting "complex, three-dimensional volumes of isopo-
tential contours, with a topology encompassing portions of neural
membranes, glial membranes, and extracellular binding sites."
Karl H. Pribram, in his chapter on "Self-Consciousness and
Intentionality ," develops a neuropsychological control-theory model of
self-regulation and self-consciousness. Pribram distinguishes among
attention, emotion, and motivation and specifically reevaluates Jame-
sian theory in light of current findings. He argues, drawing on clinical
examples as well as research in biofeedback, that "the concepts of
feedback and feedforward as they describe closed and open (helical)
loop systems are useful in the formulation of a testable model of this
domain of inquiry in precise, scientifically useful terms."
Beginning with an interest in clinical pain, Monte Buchsbaum
reviews the extensive research on augmentation and reduction of
sensory input in his chapter "Self-Regulation of Stimulus Intensity."
He is particularly interested electrophysiological measures of individ-
ual differences in central nervous system control of sensory experi-
ence. The relationship of the EEG to psychophysical scaling proce-

dures is documented, and Buchsbaum illustrates the similarity of

these findings to the Russian work on "strength of the nervous
system." He introduces the notion of "sensory homeostasis," empha-
sizing that there exists "an optimal level of continuous sensory
stimulation to maintain optimal intellectual functioning or the feeling
of well-being."
In "Neodissociation Theory of Multiple Cognitive Control Sys-
tems," Ernest R. Hilgard notes that "man does more than one thing at
a time--all of the time--but the representation of these actions in
consciousness is never complete." Drawing initially on the early work
in clinical hypnosis and multiple personality, Hilgard presents a
"modern comprehensive theory to account for the multiplicity of
processes that control overt behavior and conscious processes, with
full recognition that something like parallel processing may occur and
that all processed information is not available at anyone time to
consciousness." Research is reviewed on divided attention, recovera-
ble amnesia, state-dependent learning, hemispheric asymmetry, dis-
sociation within sleep, and multiple personalities, with special atten-
tion devoted to new findings on hypnotic analgesia and the recovery
of dissociated experiences.
Hypnosis and individual differences are discussed further in the
chapter by David R. Engstrom on "Hypnotic Susceptibility, EEG
Alpha, and Self-Regulation." Engstrom reviews research on the as-
sessment, stability, and modification of individual responses to hyp-
notic suggestions. Special attention is given to the EEG parameters
associated with hypnosis and such related phenomena as perceptual
or sensory deprivation. Engstrom explores the relationship between
EEG alpha and hypnotic susceptibility in research applying biofeed-
back procedures in regulating not only the EEG but also the skin
temperature. He contends that "biofeedback, hypnosis, meditation,
and other training operations which enhance these abilities (muscle
relaxation, concentration of attention, and reduction of distraction)
should have a similar effect on highly susceptible subjects, reflected in
the EEG."
In "Toward a Cognitive Theory of Self-Contro!," Donald Meichen-
baum sets himself the task cif explaining why "modifying a client's
internal dialogue (i.e., self-statements and images) results in behavior
change." Drawing on the neurological concept of the final common

pathway, Meichenbaum reviews clinical studies and suggests that

"the final common pathway to behavior is the internal dialogues in
which our clients engage." A three-stage process of cognitive self-
control is outlined. He suggests that self-instructions and images affect
behavior through influencing attentional direction, as well as influenc-
ing a person's interpretation and experience of his physiological state.
The interaction of cognitive and physiological processes in a
clinical context is discussed by Thomas D. Borkovec in his chapter on
"Physiological and Cognitive Processes in the Regulation of Anxiety."
Borkovec reviews findings from his research program and outlines a
descriptive multiprocess model of anxiety and its regulation. Accord-
ing to Borkovec, the experience of anxiety is elicited both by external
fear cues and by internal fear cues, the latter consisting of autonomic
arousal, verbal and nonverbal images, and proprioception from overt
behavior. Borkovec comes from a decidedly behavior-therapy orienta-
tion with strong interests in self-control procedures, and his observa-
tions have relevance to an understanding of how different processes
interact and combine to elicit the subjective experience of anxiety.
In "Dreaming: Experimental Investigation of Representational and
Adaptive Properties," David B. Cohen reviews the diverse strategies
employed for the study of mental processes during sleep. Dreaming,
according to Cohen, is a "psychological process (analogous to think-
ing) presumably inherent in the neurophysiological activity of the
sleeping nervous system." The chapter considers problems of dream
recall, the validity of dream reports, and the special adaptive role that
dreaming may play in optimal functioning in the waking state. Using
available data, Cohen speculates that if "problem-oriented dreaming
is an effective vehicle for promoting desirable change in the individ-
ual, would it be possible to encourage such changes by experimental
manipulation of dream content through pre sleep or sleep suggestion?"
In the final chapter, Thomas H. Budzynski considers this question
in "Biofeedback and the Twilight State of Consciousness." Interested
in altered states of consciousness, Budzynski notes that "when pat-
terning of input to the brain from internal and external stimuli is
unusual, out of the ordinary, then the experience may be labeled an
altered state." The particular state of consciousness emphasized in the
chapter is the "transitory condition wherein one is neither fully awake
nor deep asleep," a state once defined by William James as the "fringe

of consciousness." Budzynski reviews the available evidence of the

use of biofeedback procedures to help induce and sustain this state
and considers possible changes in novel mentation and learning,
including sleep learning. Budzynski argues that this learning may
involve the minor hemisphere, with associated problems of memory
retrieval. Case studies are offered as examples of how biofeedback
training of this low-arousal state of awareness, coupled with verbal
input, may be used as a behavior-change procedure in psychotherapy.

1 A Model of Consciousness 1
I. Levels of Information 3
II. A Personal Research Strategy 8
III. EEG Studies 10
A. Changes in Synchrony 10
B. Tracer Technique 11
IV. Average Evoked Potentials 14
A. Appearance of New Components and Increased
Similarity of AERs from Different Brain Regions
during Learning 15
B. Readout to Absent but Expected Events 15
C. Propagation of Readout from Central Structures 17
D. Differential Readout in Differential
Generalization 17
E. AER Correlates of "Meaning" in Human
Perception 21
F. Anatomical Distribution of the "Engram" 24
V. Unit Studies 26
VI. Brain Stimulation Studies 31
A. Rapid Transfer to Direct Electrical Stimulation of the
Brain 32
B. Peripheral-Central Conflict 32
C. Perceptual Integration 33
D. Loci Responsible for Perceptual Integration 33
E. Role of Cortex and Thalamic Reticular Nuclei 35
VII. Theoretical Discussion of Electrophysiological
Evidence 38
References 46

2 Self-Consciousness and Intentionality: A Model Based

on an Experimental Analysis of the Brain Mechanisms
Involved in the Jamesian Theory of Motivation and
Emotion 51

1. A Neurobehavioral Analysis of Brain Mechanisms in

Motivation and Emotion 51
A. Introduction 51
B. Case History 53
C. A Mediobasal Motor System 54
D. The Limbic Systems and Behavior 59
II. The Role of Attention in Motivational and Emotional
Reactions 66
A. Transfer of Training 66
B. Psychophysiological Experiments 68
C. Habituation 69
D. James Reconsidered 73
III. Effort and the Expression of Motivation and Emotion 74
A. Part Behaviors and Their Integration 74
B. The Precentral Motor Cortex and Action 76
C. Effort and Volition 80
D. The Jamesian Theory of Will 81
IV. A Control-Theory Model of Self-Regulation and Self-
Consciousness 83
A. The Model 83
B. Attention Span and Self-Consciousness 88
C. Central Competency 89
D. External Versus Internal Constraint 91
References 95

3 Self-Regulation of Stimulus Intensity: Augmenting!

Reducing and the Average Evoked Response 101

I. Introduction 101
II. Sensory Experience and Augmenting/Reducing 101

A. Petrie and Kinesthetic Figural Aftereffects 101

B. Evoked Responses and Augmenting/Reducing 103
III. Amplitude/Intensity Relationships in Man 105
A. Visual AERs 105
B. Auditory AERs 107
C. Somatosensory AERs 109
D. Summary of Amplitude/Intensity Relationships 110
IV. Augmenting/Reducing Reliability and the Measurement of
the AER 111
V. Genetic Factors in Augmenting/Reducing 115
A. Twin Studies 115
B. Sex and Chromosome Differences 117
VI. Tolerance for High-Intensity Stimulation 117
A. Pain Tolerance 117
B. Noise Tolerance 119
VII. Effects of Arousal, Attention, and Sensory Overload 120
A. AER Decrement over Sessions 120
B. AER Decrement with Mental Arithmetic 120
C. AER Decrement with Loud Noise 122
D. Differential Types of AER Decrement 122
VIII. Individual Differences and Intensity Judgments 123
A. Psychological Magnitude and Power Functions 123
B. Power Function Exponents and Augmenting/
Reducing 123
C. AER and Psychophysical Scaling 124
IX. Sensory Sensitivity and "Strength of the Nervous
System" 125
A. Response to Low-Intensity Stimuli 125
B. "Strength of the Nervous System" and
Reducing 125
C. Determination of Strength 126
X. Self-Regulation and Sensory Homeostasis 127
A. Optimum Levels of Stimulation 127
B. Relationships between Pain Tolerance, Sensory
Homeostasis, and Distraction 127
C. Conclusion 128
References 128

4 Neodissociation Theory of Multiple Cognitive Control

Systems 137

I.Pierre Janet's Theory of Dissociation 138

II.Why a Neodissociation Theory? 141
III.The Hypnotic Model 142
IV. Neodissociation Model of Multiple Cognitive Control
Structures 145
V. Empirical Approaches to Multiple Control Structures and
Divisions of Consciousness 152
VI. The Duality of Responsiveness to Pain as Related to
Neodissociation Theory 157
VII. Conclusion 168
References 169

5 Hypnotic Susceptibility, EEG-Alpha, and

Self-Regulation 173

I. Introduction 173
II. The Assessment of Hypnotic Susceptibility 175
A. Early Objectification 175
B. Modem Hypnotic Susceptibility Scales 176
III. Stability of Hypnotic Susceptibility 180
IV. Modification of Hypnotic Susceptibility 181
V. Hypnotic Susceptibility and Personality 182
A. Age and Development 183
B. Motivation 184
VI. Hypnosis and the EEG 185
VII. EEG and Hypnotic Susceptibility: Indirect
Relationships 187
A. Age 187
B. Perceptual or Sensory Deprivation 188
VIII. EEG and Hypnotic Susceptibility: Direct Evidence 189
A. Base-Rate Alpha Density 189
B. Base-Rate Alpha Amplitude 191
C. EEG Asymmetry 192
D. Evoked Potentials 192
E. Conclusion 192

IX. The Stability of EEG Base Rates 193

X. Increasing Susceptibility by EEG Feedback 195
XI. Changes in EEG during Hypnosis 203
XII. Task-Specific EEG Changes 207
XIII. Conclusions 215
References 217

6 Toward a Cognitive Theory of

Self-Control 223

I. Introduction 223
II. Conclusions from Treatment 224
A. How Shall We Treat Our Clients' Cognitions? 225
B. Cognitions as Final Common Pathways 238
C. Initial, Conceptualization Phase of Therapy 239
III. A Cognitive Theory of Self-Control 243
A. A Three-Stage Process 243
B. How Does Behavior Change through Internal
Dialogue? 248
IV. Summary 253
References 255

7 Physiological and Cognitive Processes in the

Regulation of Anxiety 261

I. A Descriptive Model of Anxiety Process 264

A. Current Stimulus Conditions 264
B. The Immediate Anxiety Reaction 266
C. Subsequent Maintaining and Reducing
Reactions 268
D. Intervention Strategies 272
II. Research Studies on the Maintenance and Reduction of
Anxiety 276
A. The Role of Physiological Arousal and Cognition 279
B. The Role of Individual Differences in Physiological
Arousal and Autonomic Perception 289
III. Summary and Conclusions 305
References 308

8 Dreaming: Experimental Investigation of

Representational and Adaptive Properties 313

I. Dream Recall 313

A. The Role of Repression 313
B. Alternative Factors: Salience and Interference 317
C. Implication for Theory 324
II. Representational Properties of Dreaming 327
A. Validity of Dream Reports 327
B. Two Strategies for Investigating Dreaming 328
III. Functional Properties of Dreaming 345
A. Functions of REM versus NREM Sleep 346
B. REM Psychology versus REM Physiology 349
C. Dream Content and Psychological Change 351
References 355

9 Biofeedback and the Twilight States of

Consciousness 361
I. The Twilight State 362
A. Is a Twilight State the Source of Creative Ideas? 363
B. Biofeedback and Creativity 364
C. Learning in the Twilight State? 367
D. The Production of Low Arousal through
Biofeedback 373
E. A Twilight-State Biofeedback System 374
II. Future Considerations 379
A. Is Twilight Learning Minor-Hemisphere
Learning? 380
B. A Language for the Minor Hemisphere 381
C. Retrieval Difficulties 381
D. Cognitive Balance 382
References 382

Author Index 387

Subject Index 395
1 A Model of Consciousness


In the first textbook of physiological psychology, written by Wilhelm

Wundt (1910) at the end of the 19th century, Wundt defined the task of
physiological psychology as the analysis of the physiological bases of
consciousness and subjective experience. In the textbook of physiologi-
cal psychology which I used when a student, written by Morgan and
Stellar (1950) in the middle of the 20th century, physiological psychol-
ogy was defined as the study of the physiological bases of behavior.
The word consciousness does not even appear in the index of the latter
volume, nor have I encountered it anywhere in the text.
Behaviorism, and "operationism," virtually legislated the prob-
lems of consciousness and subjective experience out of the domain of
the legitimate concerns of "scientific" and especially physiological psy-
chology, whence they remain essentially excluded until this day. Con-
temporary experimental and physiological psychology, in its zeal to
sanitize itself from any taint of its philosophical heritage and to be even
more scientific than the "real" sciences, has preoccupied itself with the
analysis of behaviors as if they were performed by unconscious or
mindless automata. Attention has largely been focused on clarification
of the effects of various schedules of reinforcement on operant re-
sponses or the brain mechanisms mediating conditioned responses,
rather than on the neural bases of cognition.
I have a confession to make. I am not now, nor have I ever been,
interested in behavior as such. The main reason I work in physiological
psychology is because I am interested in the physiological bases of
consciousness and subjective experience. I believe that "mind," under
which rubric are subsumed such phenomena as consciousness, subjec-
tive experience, the concept of self, and self-awareness, is an emergent
property of sufficiently complex and appropriately organized matter.
E. Roy JOHN . Departments of Psychiatry and Physiology, New York Medical College,
New York, New York.


In some fashion, cooperative processes between elements of living

matter which individually possess only rudimentary properties gener-
ate this emergent property for the system, which qualitatively tran-
scends a simple summation of the elementary properties of the constit-
uent parts. One system which possesses this emergent property is the
brain, and the relevant constituent elements are the neurons and the
glial cells. We do not understand the nature of this cooperative process,
the physical and chemical interactions between the elements of matter
which produce mental experience. We do not know how big a neuronal
system must be before it can sustain the critical reactions, nor whether
the critical reactions depend exclusively upon the properties of neurons
or only require a particular organization of energy in matter.
These fascinating and enormously important problems, in my
opinion, should be among the central topics of investigation in physiol-
ogical psychology and neurophysiology. They have been neglected far
too long, while seemingly inexhaustible energy has been lavished on
problems of lesser import. I welcome the signs of a resurgent interest in
consciousness and subjective experience, as evinced by the appearance
of this series of volumes on consciousness and self-regulation. I am
convinced that sufficiently powerful experimental and analytical tools
are now available to permit significant progress to be made in the
understanding of these issues. This chapter provides an opportunity to
examine these problems and to discuss how current research findings
might be relevant.
One becomes painfully aware of the paucity of contemporary
thinking about these issues at the very outset of any attempt to formu-
late meaningful experimental or analytic approaches to the physiologi-
cal processes responsible for consciousness and subjective experience.
A prerequisite for experimental analysis of these problems must be an
adequate definition of what is to be analyzed. Especially because most
experiments requiring manipulation of the brain must be carried out in
animals, except for the small although invaluable body of data slowly
accumulating from the study of "nature's experiments" in cases of
human brain injury or disease, an operational definition of conscious-
ness is absolutely essential. Without an unequivocal definition of con-
sciousness, it is hopeless to attempt to identify the responsible proc-
esses in the brain.
How shall we decide when consciousness is present in an experi-
mental preparation? What constitutes the content of consciousness? Is

there a difference between consciousness and self-awareness? What do

we mean by subjective experience? Is subjective experience sensation, or
the perception of sensations, or the apperception of sensations, or
something more than any of these?


It is inordinately difficult to formulate answers to these questions

which seem at all adequate, evading the pitfalls of triviality on the one
hand and of useless vague generality on the other. I propose the
following definitions as first approximations which provide a basis for
an experimental approach. In a later portion of this article, I will relate
some current experimental results to these definitions.
1. Sensations are the spatiotemporal patterns of information arriv-
ing in the central nervous system because of the excitation of exterocep-
tive and interoceptive organs. They are a product of the irritability of
living matter and constitute first-order information. Such irritability is
manifested throughout the phylogenetic scale and is already present in
protozoans. Sensations can elicit reflex responses, adjusting the orga-
nism to its environment.
2. Perceptions are the interpretation of the meaning of sensations in
the context of stored information about previous experiences. Percep-
tions constitute second-order information resulting from an interaction
between sensations and memories.
Wundt and his contemporaries argued that the presence of con-
sciousness was revealed when behavioral responses to stimuli ceased to
be reflexive and displayed "purposiveness," by which they meant
actions which were adaptive and resulted in the adjustment of the
organism to its environment as a function of the experiential context of
a stimulus rather than to the action of the stimulus alone. For this
reason, they considered identification of the lowest phylogenetic level
showing learning as crucial for the decision as to the lowest level of
organization capable of sustaining consciousness. In this regard, it is
noteworthy that Coming, Dyal, and Willows, in their authoritative
review of invertebrate learning (1973), reached the conclusion that
although the evidence for simple learning or associative conditioning
remains highly controversial, there exists compelling evidence that
protozoans display the ability to learn not to respond, i.e., habituation,

and some evidence for associative learning has been forthcoming. The
capacity for complex learning clearly appears in the phylum Platyhel-
minthes, with the advent of a brain, defined sensory systems, and
complex nerve bundles.
We choose to define perception, as well as sensation, provisionally
as preconscious or unfelt categories of information processing. Sensa-
tions and perceptions are unimodal, referring to the detection and
interpretation of stimuli within individual sensory modalities. These
functions can be performed by machines, which do not possess con-
sciousness. We contend that under ordinary circumstances, fundamen-
tal sensations and much of perception, as defined, do not enter con-
sciousness, although we can make ourselves aware of them by an
analytic process.
3. Consciousness is a process in which information about multiple
individual modalities of sensation and perception is combined into a
unified, multidimensional representation of the state of the system and
its environment and is integrated with information about memories
and the needs of the organism, generating emotional reactions and
programs of behavior to adjust the organism to its environment. Con-
sciousness is third-order information. Many levels of consciousness can
exist, in which these dimensions are present in variable amounts. The
content of consciousness is the momentary constellation of these differ-
ent types of information.
At the same time that consciousness is the product of an integra-
tion of preconscious sensations and perceptions structured in the light
of previous experience and reflecting emotional state, drive level, and
behavioral plans, feedback from consciousness to these more funda-
mental levels must take place. Memories are activated, attention is
focused, perceptions influenced, emotions aroused, drive priorities
altered, and plans of behavior revised as a result of this feedback,
producing a continuous reorganization of basic processes because of
the influence of higher-level integrative and analytical functions.
4. Subjective experience derives from information about the content
of consciousness. It is a process which reorganizes the sequential series
of events into a single experiential episode, which merges sequential
constellations of multisensory perceptions, memories, emotions, and
actions into a unified and apparently continuous event, or "experi-
ence," which has a beginning and end. Two critical transformations
occur as a result of the process which generates this fourth-order infor-

mation. First, although the information impinging upon the neuronal

populations mediating each of the different dimensions of conscious-
ness is represented by the same mechanism (spatiotemporal patterns of
neural discharge) in every such population, the fourth-order informa-
tion about each different dimension of consciousness is qualitatively
distinct. Subjective experience consists of diverse colors, shapes,
sounds, textures, smells, tastes, emotions, plans, movements, and
thoughts, rather than a uniformly encoded description of these disparate
facets of experience. Somehow, qualitative diversity at this higher level
of information is constructed out of representational uniformity at
lower levels. At the same time, in spite of these qualitative distinctions
between the different facets of consciousness and the capability to
decompose experience into its constituent components, subjective ex-
perience merges these facets into an apparently simultaneous and con-
tinuous multidimensional unity.
As this unified subjective experience begins to take shape from a
related series of episodes, memories relevant to this holistic event are
activated, many of them in modalities not involved in the episodes
taking place. Some of these memories are of rudimentary or fragmen-
tary sensations, while some are of prior subjective experiences (see
5. The self: Second, as subjective experience extends through time
and an individual history is accumulated, memory of the sequence of
episodes is constructed. This personal history, the accumulated memo-
ries of sets of fourth-order information, constitutes the basis for what
we call the self. The concept of the self arises as a result of long-
term memories constituting the record of an individual's subjective
experiences. This individual historical record constitutes fifth-order
6. Self-awareness: If we consider subjective experiences as "higher-
order sensations," then "self-awareness" is analogous to the perception
of those sensations. By this is meant the interpretation of subjective
experience in terms of the previous history of life experiences of the
individual. Self-awareness is the interpretation of present subjective
experience in the context of the salient features, especially the more
invariant features, of the pattern of previous subjective experiences.
Self-awareness constitutes sixth-order information.
As the momentary content of consciousness is interpreted in the
light of past experience, feedback to lower levels occurs which is
6 E. Roy JOHN

probably more powerful than any described thus far. This feedback can
be expected to activate trains of memories of other relevant life experi-
ences, with a high probability that important occurrences (high drive
level, high-emotion events) will be followed by systematic or "rational"
memory searches. The relatively global feedback resulting from the
integration of lower-level information as it enters consciousness is
modulated and made far more selective and better-focused. Among the
consequences envisaged as resulting from this highest level of informa-
tion are systematic evaluation of a flood of memories, identification of
appropriate perceptions and rejection of more inappropriate percep-
tions which arose earlier in the experience, selection of the most appro-
priate emotional response, adjustment of drive levels to correspond to
the exigencies and possibilities of the moment, and rational construc-
tion of the optimal program of behavior. These processes are far more
deliberate and analytical than those previously described.
A characteristic of self-awareness is the capacity for cognitive proc-
esses. By cognition or thought we mean the ability to have subjective
experience vicariously, by activating stored memories about percep-
tions and prior experiential episodes in a fashion which may be arbi-
trarily organized rather than occurring according to a previously estab-
lished sequence. Because of this ability to manipulate, recombine, and
reorganize the accumulated store of memories, the self is continuously
in the process of modification and of analysis of its own experience.
A cognitive process is the representation of an experience in an
abstract symbolic fashion, whether or not that experience actually oc-
curred in that form in the personal history of the individual. The
distinction between the memory of a rudimentary sensation postulated
as essential for perception and the memory of a subjective experience is
the amount and diversity of the stored information. The basic neuro-
physiological mechanisms may be quite the same, and even the ana-
tomicalloci may be shared. I see no compelling need to separate those
mechanisms conceptually. Under some circumstances, particularly
when cross-modal stimulation is utilized, generalization affords evi-
dence for the presence of cognitive processes. In generalization, an
organism interprets the meaning of a sensory stimulus as equivalent to
some other sensation because of similarities in the abstract properties
common to both stimuli. If the two stimuli are in different sensory
modalities, it is clear that some nonsensory specific abstraction has
been performed. If the stimuli are in the same senso'ry modality,

interpretation becomes more equivocal because of the possibility that

similar receptors were activated.
Observational learning seems to constitute a more unequivocal
type of evidence for the presence of cognitive processes. We and others
(Chesler, 1969; John, Chesler, Bartlett, and Victor, 1968; Grinberg-
Zylberbaum, Carranza, Cepeda, Vale, and Steinberg, 1974) have shown
that naive animals can learn complex discriminative behaviors simply
by observing the performance of trained animals. Since the observing
animals do not directly experience the reinforcing stimuli, their acqui-
sition of the discrimination must be attributed to their interpretation of



1 1


1 J

.-----l EMOTIONS .~

f---- - -1 PR~m~I~R OF t: . ~
t ,(.



FIGURE 1. Flow scheme for the levels of information involved in consciousness and self-
8 E. Roy JOHN

what they observe by referring it to memories of previous experiential

episodes which they did experience directly. Observational learning
already requires consciousness and probably requires self-awareness.
I have found it useful to postulate a series of different levels of
information processing, each dependent upon all the levels below
(feedforward) and each influenced by the levels above (feedback), in
order to define sensation, perception, consciousness, the content of
consciousness, subjective experience, the emergence of a self-concept,
and self-awareness. The proposed definitions treat each of these proc-
esses as fundamentally similar to all the others in that they are all
representations of information, presumably in a common neuronal
code. They are all different in that they constitute successively higher
derivations extracted from the information representing the lower deri-
vations. These ideas are illustrated in Figure 1, which has been limited
to two sensory modalities to simplify the diagram.
If we accept these formulations as working definitions, the task of
experimental analysis of these processes may become easier. The proc-
esses representing information at the lower levels must be analyzed
first. As we gain insights into the representation of lower-level infor-
mation, it becomes possible to seek invariances across the representa-
tion of multiple items on the same level, which share a common
informational feature. Such invariances constitute the representation of
information on a higher level. In this "bootstrap" fashion, it would
appear possible to progress in a systematic development from initial
studies of sensory mechanisms to eventual investigations into the
neurophysiological basis of self-awareness.


When I began to do research, the physiological process by which

neuronal activity became transmuted into subjective experience was
the problem of greatest interest to me. The difficulty of objective
definition of the momentary content of a spontaneous stream of con-
sciousness seemed insuperable, especially in animal experiments. The
study of memory offered what appeared to be a unique solution to this
dilemma. If one could succeed in understanding how the information
about a specific experience was encoded, stored, and retrieved, then
one could identify the physiological processes corresponding to a spe-

cific memory. When that memory was remembered, the corresponding

physiological process would appear.
Appearance of that process would constitute an objective indica-
tion that a specific past experience was the content of consciousness at
that moment. Examination and analysis of the features of such a repre-
sentational process would provide a description of the physiological
mechanism which generated or corresponded to a specific thought.
While this description might not explain how the physiological proc-
esses engendered the subjective experience, it would tell us what the
relevant processes might be.
Ideally the experimental situation would be so devised that per-
formance of some particular behavior became much more likely when a
specific past experience was remembered, providing objective reassur-
ance that the subjective experience did in fact take place when pre-
dicted. For more than 20 years, I have been pursuing this strategy,
constantly trying to improve the resolution of my measurements
of neurophysiological processes and the design of the experimental
These experiments have primarily been aimed at obtaining a de-
tailed description of the electrical activity of different brain regions in
unrestrained animals as they acquired and performed differentiated
conditioned responses to discriminative stimuli in each of several sen-
sory modalities. At first, these studies evaluated changes in ongoing
electroencephalographic (EEG) activity during conditioning. With the
advent of average response computers, our attention shifted to the
details of the evoked potentials elicited by the discriminanda. As the
results of evoked potential studies provided a relatively clear and
consistent picture of the slow-wave phenomena related to information
coding and memory retrievel, we used the higher resolution afforded
by microelectrode techniques to investigate the behavior of single
neurons and small neuronal ensembles under comparable experimental
conditions. Finally, when experimental observations permitted formu-
lation of a tentative theory about the salient features of the process by
which a past experience was represented, we made an attempt to test
the theory directly by using electrical stimulation of the brain to repro-
duce those hypothetically crucial features, observing behavior to infer
what subjective experience had ensued.
During this period, other workers carried out a large number of
related experiments. In the next section of this article I will briefly
10 E.RoYJoHN

survey this body of research. I will rely mostly upon the work of my
own laboratory, because phenomena observed by me personally have
had the most impact upon my thinking.
Electrophysiological methods have provided unique insights into
the details of physiological processes within various anatomical regions
and the dynamic transactions between as well as within those regions
which take place during learning and which occur when memories are
activated. These insights allow us to construct a description of how
experiences build an anatomically distributed mediational system in
which different parts of the brain cooperate in the representation and
procession of information. Detailed reviews of the voluminous evi-
dence on which this description is based are available elsewhere (John,
1961, 1967b, 1971, 1972, 1974; John and Thatcher, 1976; Morrell, 1961b;
Thompson, Patterson, and Teyler, 1972). In this article, we have
ignored the problem of the chemical processes involved in information
storage, which we have discussed elsewhere in detail (John, 1967b).


A. Changes in Synchrony
Since the discovery that tiny electrical voltage fluctuations could be
recorded from the scalp, the EEG correlates of conditioning have been
studied by numerous workers. The general features of the EEG changes
observed in such studies are that when a conditioned stimulus (CS), to
which the subject has previously been habituated, is initially paired
with an unconditioned stimulus (US), widespread changes from rela-
tively low-frequency high-voltage activity (synchronization) to higher-
frequency low-voltage activity (activation) occurs in the scalp EEG. As
training proceeds, this activation or desynchronization pattern be-
comes limited to only a few "relevant" regions, for example, over the
motor cortex if the conditioned response (CR) requires a movement,
over the visual cortex if the CS is a visual signal. Usually changes in the
EEG occur prior to the appearance of the first behavioral CRs. During
extinction, learning-induced EEG changes persist beyond the disap-
pearance of CRs, with a gradual reversal of the changes seen during

Such findings led to the conclusion that during conditioning there

was initial widespread "irradiation" of information over the cortex
(adduced as evidence of involvement of the mesencephalic reticular
formation early in learning), followed by "consolidation" or more dif-
ferentiated and localized mediation of performance of well-learned
responses (interpreted to indicate a later shift to a dominant role for the
thalamic reticular formation, the intralaminar nuclei of the diffuse
projection system). Studies of the habituation of the "arousal" re-
sponse, i.e., gradual disappearance of the activation pattern caused by
repeated presentation of a novel stimulus (often considered as a primi-
tive type of perceptual learning), led to analogous concepts of adjust-
ment of iterated inconsequential events, initially involving a phasic
diminution of response in the thalamic reticular formation followed by
tonic adaptation in the mesencephalic reticular formation, eventually
leading to complete suppression of the desynchronization response.

B. Tracer Technique

Since its introduction in the Soviet Union by Livanov and Poliakov

(1945) and in the United States by John and Killam (1959), tracer
technique has been the most useful method for distinguishing between
the electrical activity of the brain related to information processing
about the learned experience (which we will define as "signal") and the
other ongoing business of the brain (which we arbitrarily refer to as
"noise" because of our primary concern with brain mechanisms in-
volved in learning and memory). Like many crucial methodological
innovations, the idea underlying tracer technique is very simple. The
signal of "tracer-conditioned stimulus" (TCS) for the learned behavior
under study is presented intermittently at a characteristic rate of repeti-
tion. Electrical rhythms which appear in different brain regions at the
frequency of the TCS are considered to be "labeled responses" reflect-
ing processing of information about the stimulus.

1. Participation of Many Brain Regions in Learned Behavior

The first findings provided by tracer technique showed that during

learning, widespread changes take place in the anatomical distribution
12 E. Roy JOHN

of the brain's responses to the CS. Although the same phenomena of

irradiation and consolidation described in prior studies of the desyn-
chronization of the EEG were also observed with labeled rhythms, the
decrease in the anatomical extensiveness of the responsive system with
well-learned behaviors was only observed in simple CRs, where the
animal merely needed to detect the CS. When differential responses
required discrimination between different signals, the labeled re-
sponses stabilized throughout a widespread anatomical system.

2. Display of Similar Electrical Activity by Many Brain Regions

In the original studies of Livanov and Poliakov and of John and

Killam, it was noted with surprise that a number of brain regions which
showed markedly different electrical responses to the TCS before condi-
tioning acquired striking similarities in electrical activity during acqui-
sition and subsequent performance of a new behavioral response to
that stimulus. Many other workers have commented upon the same
phenomenon (Yoshii, Pruvot, Gastaut, 1957; John and Killam, 1959,
1960; Liberson and Ellen, 1960; Galambos and Sheatz, 1962; Glivenko,
Korol'kova, and Kuznetsova, 1962; Livanov, 1962, 1965; John, Ruchkin,
and Villegas, 1963, 1964; Dumenko, 1967; Knipst, 1967; Korol'kova and
Shvets, 1967).
These findings indicate that during learning a representational
system is established which involves many different anatomical regions
in a cooperative, similar mode of activity.

3. "Assimilation" of the Rhythm of the TCS

The most intriguing phenomenon observed in the early studies

with tracer technique, and since confirmed in many different species
and experimental situations, was named assimilation by Livanov and
Poliakov. This term referred to the fact that while a CR was being
established to a TCS, the spontaneous EEG during the intertrial inter-
vals became dominated by electrical rhythms at the frequency of the
absent stimulus. Such rhythms were absent in the home cage but
appeared as soon as the animal entered the familiar training environ-
ment (Yoshii and Ogura, 1960). It was as if the animal were rehearsing
the experience of the meaningful signal which had previously appeared

in that situation and was again anticipated. Recently, assimilated

rhythms with remarkable precision of frequency have been found in
the firing patterns of small groups of cortical cells (Ramos and Schwartz,
1976b). The functional significance of these assimilated rhythms is
clearly established by the fact that they appear only on the trained side
of a split-brain cat (Majkowski, 1967).
These phenomena show that the representational system built
during an experience with a rhythmic stimulus can produce an electri-
cal facsimile of that rhythm in the absence of the TCS.

4. Exogenous and Endogenous Components of EEG Rhythms

Appearance of assimilated rhythms often precedes spontaneous

performance of the CR (Yoshii, 1962). Another line of evidence further
suggests the functional significance of endogenous electrical patterns,
i.e., temporal patterns of activity originating within the brain. As
mentioned previously, many different brain regions come to display
similar electrical rhythms during conditioning with a TCS. However,
when animals who have been highly overtrained in the performance of
differential CRs to discriminated stimuli at two different frequencies
commit errors, certain brain regions often display electrical rhythms
inappropriate to the actual stimulus. Instead, these rhythms correspond
to the frequency of the absent stimulus which would be the appropriate
cue for the behavior which was performed (John and Killam, 1960;
John, 1963, 1967a, 1967b, 1972; John, Leiman, and Sachs, 1961;
Lindsley, Carpenter, Killam, and Killam, 1968; Majkowski, 1966). This
phenomenon is illustrated in Figure 2.
Thus the similar electrical patterns observed in different brain
regions of trained animals come from two origins. One, which we term
exogenous, reflects afferent input due to external reality. The other,
which we term endogenous, reflects the release of previously stored
electrical patterns from some internal representational system. The
appearance of two different electrical patterns in the brain of an animal
as it commits a behavioral error, indicative of misinterpretation of a
signal, suggests that the exogenous activity caused by the actual envi-
ronment somehow activated inappropriate endogenous activity reflect-
ing the significance normally attributed to a different signal. A mis-
match has taken place between stimulus input reflecting reality and the
retrieval from memory necessary to interpret that reality.
14 E. Roy JOHN

L MOT ,.-..}o-_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ...}--I-._..,.,..._ _......._ _ _ _...-_......_ _..-_ _

R MOT""",,_ .'PII~.,-
... - - - - - - - - - - - - - - -
RAUD ..~.;o.vJ~,~At-..J.,.~
LVPL~"~ ...~~,....,..W\"......~~~~
L GL ,~"'.:;M,"trWtr.,.i'i"",-y.y,...,~.- ~~'A.'t'..I."PfWrMrl'~l~
R GL"J'I\N""Y....",......v..~oJ/V' ~"'."v."'N~"~Wwr~
L GM~ _ _--............""""-""""""""""'.......""",..,.....,.,..,.,.,.."""__....
RGM""W"'-~'"".y..".,...._ _~
L MRF',.,..Jtv~""-'\~~,...,.J-r...,.'HV.....JI"'.f'.......,..~~

FIGURE 2. Comparison of electrophysiological activity elicited by the same visual signal

interpreted in two different ways. Records on the left were obtained as the differentially
trained cat responded correctly to a negative CS (707-Hz flicker) by performing a condi-
tioned avoidance response. Records on the right were obtained on the next trial, when the
cat erroneously responded to the same flicker frequency by performing the conditioned
approach response appropriate to the positive stimulus (301-Hz flicker). (MOT, motor
cortex; AUD, auditory cortex; VIS, visual cortex; VPL, ventroposterolateral nucleus; GL,
lateral geniculate nucleus; GM, medial geniculate nucleus; MRF, mesencephalic reticular
formation.) All records bipolar. (Data from John, 1972.)


With the advent of special and general-purpose minicomputers

capable of averaging many evoked potentials while an experiment was
in progress ("on-line" average-response computation), it became possi-
ble to analyze the data obtained in conditioning studies using tracer
technique by computation of the average evoked response (AER) to the
TCS. This enabled examination of the actual waveshape of the response
to the stimulus in each brain region and largely replaced the previous
preoccupation with the frequency and amplitude of labeled rhythms.
AER studies confirmed and extended the conclusions reached in
the earlier EEG studies of conditioning. During learning, the anatomi-
cal distribution of evoked responses to the TCS becomes more wide-
spread. While responses continue to be displayed by sensory-specific
structures of the modality of the signal, new responses appear in
regions which showed little or no response to the CS before training.

These new responses are particularly striking in the mesencephalic

reticular formation, the intralaminar nuclei of the thalamus, and in
various portions of the limbic system, especially the hippocampus.

A. Appearance of New Components and Increased Similarity of

AERs from Different Brain Regions during Learning
As conditioning proceeds, a new late process with an onset latency
about 60 milliseconds after the CS appears in the AER recorded from
many brain regions (Asratyan, 1965; Begleiter and Platz, 1969; Galam-
bos and Sheatz, 1962; John, 1963, 1967a, 1967b; John and Killam, 1959;
John and Morgades, 1969a; Killam and Hance, 1965; Leiman, 1962;
Lindsley, Carpenter, Killam, and Killam, 1968; Sakhuilina and Mer-
zhanova, 1966). Different brain regions display markedly disparate
AERs at the onset of training and acquire similarities in AER waves-
hape as conditioning proceeds. Further, when the signal fails to elicit
CR performance, the new late process often is absent from the AER in
many regions. These observations are illustrated in Figure 3.

B. Readout to Absent but Expected Events

A body of evidence has accumulated which shows that certain

aspects of the evoked potential (EP) may reflect previous experience
rather than responses to afferent input and are thus of endogenous
rather than exogenous origin. One important line of such evidence
comes from studies primarily carried out on human subjects and is
particularly important in the assessing of the likelihood that these
released electrical patterns actually correspond to the activation of spe-
cific memories, because it has been possible to establish unequivocally
that there is a subjective correlate to the appearance of these released
potentials. These studies show that when an expected event does not
occur, a cerebral potential appears at a latency similar to that of poten-
tials usually evoked by the expected stimulus. EPs elicited in man by
absent events have been reported (Barlow, Morrell, and Morrell, 1967;
Klinke, Fruhstorfer, and Finkenzeller, 1968; Picton, Hillyard, and Gal-
ambos, 1973; Riggs and Whittle, 1967; Rusinov, 1959; Sutton, Tueting,
16 E. Roy JOHN



VISUAL CORTEX -" ,r---. ~ V'-- ~

-vv ~ V'- y--
J'v. ~
v---- -v-:--,--
NUe. RETICULARIS ,~ \/./'------ ""\ v r--- ~
NUC. CENTRAlIS LATERALlS ~' --yv---" ~. r----- \f\t/
DORSAL HIPPOCAMPUS ........./'\...rJ' ~- .~ ~ ~ r
MEDIAl. FOREBRAIN BUNDLE ~"'~ v ' /---"-
_ "\.
v.~ "- t
PREPYRIFORM CORTEX -' _~ /"- ' _~....... YI '\J\., ~
V" t
\;' .-,, - . ., " ~ -~
SUBSTANTIA NIGRA 'v-~.. " .J'" ... \ :. \......... Ay.- "'-
v 12' .. I \ / t
FIGURE 3. Evolution of visual evoked responses. Control, average responses evoked in
different brain regions of a naive cat by presentation of a novel flicker stimulus. Several
regions show little or no response. Early CAR, responses to the same stimulus shortly
after elaboration of a simple conditioned avoidance response (CAR). A definite response
with similar features can now be discerned in most regions. Differential CAR, changes in
the response evoked by the flicker CS shortly after establishment of differential approach-
avoidance responses to flicker at two different frequencies. As usual, discrimination
training has greatly enhanced the response amplitude, and the similarity between re-
sponses in different structures has become more marked. Overtrained CAR, after many
months of overtraining on the differentiation task, the waveshapes undergo further
changes. The arrows point to a component usually absent or markedly smaller in behav-
ioral trials on which this animal failed to perform (nuc. reticularis, nucleus reticularis; nuc.
subthalamus, nucleus subthalamus) . (Data from John, 1972.)

Zubin, and John, 1967; Weinberg, Walter, and Crow, 1970; Weinberg,
Walter, Cooper, and Aldridge, 1974). Similar findings in the cat were
reported by John (1963). These cerebral events, termed readout or
emitted potentials, have been interpreted by Weinberg et al. to reflect
the generation of processes corresponding to the memory of past or
imaginary stimuli.

C. Propagation of Readout from Central Structures

When generalization occurs upon presentation of a novel test stim-

ulus, the AER in the lateral geniculate body and in many other struc-
tures closely resembles the waveshape usually evoked by the visual CS.
However, if generalization fails to occur, the response to the test
stimulus differs radically from the typical AER to the CS, lacking the
late components (Ruchkin and John, 1966). The same phenomenon has
been found in the firing patterns of neuronal ensembles in the lateral
geniculate during generalization (John and Morgades, 1969b). This
phenomenon is illustrated in Figure 4.
By subtraction of AERs from trials in which no behavioral response
was elicited by presentation of the test stimulus from AERs computed
during trials resulting in generalization, it was possible to construct the
difference waveshapes, showing the forms and latency of the readout
process released in different brain regions during generalization (John,
Ruchkin, Leiman, Sachs, and Ahn, 1965). Readout processes were
found in most brain regions studied and displayed a general similarity
of waveshape with marked latency differences from region to region.
The readout process seems to arise in a central corticoreticular system,
from which it propagates to involve other brain regions in a systematic
sequence, appearing last in the lateral geniculate body (when a visual
cue is used). The fact that the thalamic "relay" nucleus for visual
information is so dramatically influenced by this centrifugal process
gives some insight into the compelling influence of experience upon
perception. This finding is illustrated in Figure 5.

D. Differential Readout in Differential Generalization

The readout process is not merely a nonspecific indicator that

memory retrieval is in progress. The shape of the readout process
depends upon which memory is activated. This has been established
by a technique called differential generalization. Cats are trained to
perform two different CRs (CRI and C~) to two discriminated visual
stimuli, consisting of flicker at two different repetition rates (VI and V2 ).
After thorough overtraining, a test stimulus (Va) is occasionally intro-
duced into random sequences of VI and V2 The frequency of Va is
midway between VI and V2
18 E. Roy JOHN



TO 7.7/SEC

TO 7.7/SEC





,, '


Sometimes the cat treats V3 as equivalent to VI and CR1 is per-

formed (V3CR1 ). On other trials the cat treats V3 as if it were V2 and C~
is elicited (V3C~), AER waveshapes during V3CR1 trials are signifi-
cantly different from those found during V3C~ trials. When V3 presen-
tation results in CR1 performance, the AER elicited by V3 closely resem-
bles the usual evoked response to V2 Conversely, when V3 presentation
results in C~ performance, the AER to V3 is like that usually caused by
V2 (John, Shimokochi, and Bartlett, 1969). These findings are illustrated
in Figure 6.
This phenomenon has been analyzed in great detail, by the use of
visual, auditory, and electrical stimuli delivered directly to brain struc-
tures. A wide variety of different instrumental tasks have been utilized
and many controls introduced to rule out possible unspecific causes for
this phenomenon. Methods of computer-pattern recognition have been
developed to permit classification of single evoked response wave-
shapes. These further studies have shown that readout processes
during differential generalization can be found in most brain regions,
are demonstrable under all stimUlus-response contingencies thus far
explored, and cannot be attributed to unspecific origins. Repeated
presentation of the repetitive test stimulus elicits a variable sequence
of EP waveshapes or modes. The CR subsequently performed to the

FIGURE 4. (A) Computations of average responses obtained from the lateral geniculate
nucleus and nucleus reticularis of the cat under various conditions during the same
experimental session. First row of averages is based upon 100; second and third rows are
based upon 42 repetitions of the same stimulus applied during a number of behavioral
trials. Analysis epoch was 90 msec. First row: Average responses evoked in structures by
the 10-Hz CS (flicker) actually used in training, during repeated correct behavioral
performances. Second row: Average responses evoked by a novel 7.7-Hz CS, during
repeated generalization behavior. Test trials with the 7.7-Hz stimulus were interspersed
among trials with the actual 10-Hz CS and were never reinforced. Third row: Average
responses evoked by the 7.7-Hz flicker on presentations when no generalization behavior
was elicited. The waveshape elicited by the actual CS is similar to the response evoked by
the novel stimulus during generalization behavior. Notice the absence of the second
positive component in the EP when generalization behavior failed to occur. (Data from
Ruchkin and John, 1966.) (B) (Top) Records of AER's and PSH's obtained during 18 trials
that resulted in CR to the 2-Hz CS (dotted curves) and during 32 trials that resulted in
behavioral generalization in response to a I-Hz flicker used as a test stimulus (solid
curves). The test stimuli were randomly interspersed between presentations of 2-Hz
(dotted curves) and 8-Hz flickers in a long experimental session. (Bottom) Records of
AER's and PSH's obtained during 17 trials that resulted in failure to elicit generalization
behavior in response to the test stimulus. Note change in late components. Analysis
epoch, 100 msec. (Data from John and Morgades, 1969a.)
20 E. Roy JOHN

test stimulus is consistently related to the predominant EP mode

identified by the pattern recognition procedure (John, Bartlett, Shimo-
kochi, and Kleinman, 1973).
Thus an ambiguous stimulus activates a variety of readout proc-
esses identifiable by the different features of late portions of the EP. The
behavior eventually displayed seems to depend upon the particular
readout mode which becomes dominant in the representational system.
These findings indicate that in many brain regions the waveshape of
AER elicited by a stimulus is not determined solely by its phYSical

RPOST "A::t~



R.N.VENT 7\r


R.N.LAT p~

FIGURE 5. Difference waveshapes obtained for a

number of regions by subtraction of averaged re-
R.N.RETIC~ sponses, evoked by a 7.7-Hz test stimulus during
nonperformance from averaged responses elicited
when generalization occurs. All averages in these
R.ANTLG~ computations were based upon 200 EPs distributed
among a number of behavioral trials in each cate-
gory, with a 62.5 msec analysis epoch. (Data from
John, Ruchkin, Leiman, Sachs, and Ahn, 1965.)

FIGURE 6. Waveshapes of average responses re-

corded from the LG (bipolar) under various
stimulus-response contingencies: V,CR, during
trials resulting in. correct performance of an ap-
proach response (CR) to a 3.1-Hz flicker CS;
V,CAR, during trials in which a conditioned
P (.01
avoidance response (CAR) was correctly per-
formed in response to a 7.7-Hz CS; V3 CR, during V3 CAR
generalization trials in which a neutral 5-Hz test
stimulus elicited CR behavior; V3 CAR, during
generalization trials in which the same 5-Hz test
stimulus elicited CAR behavior. The interrupted
line between V3 CR and V3 CAR indicates time 100 MSEC
intervals during which V3 CR and V3 CAR were significantly different at better than the P
= .01 level. The numbers at the right indicate the correlation coefficients between the
corresponding bracketed waveshapes. (Data from John, 1972.)

parameters but is strongly influenced by the meaning attributed to it in

the context of memories about previous similar experiences.

E. AER Correlates of "Meaning" in Human Perception

Jacobo Grinberg-Zylberbaum and I recently carried out an electro-

physiological experiment on human subjects which showed that the
shape of the AER in some cortical regions depends upon the meaning
attributed to a visual stimulus, rather than upon its form. The experi-
ment consisted of two portions. In the first part, subjects seated before a
tachistoscope viewed brief presentations of a vertical line followed by
presentation of the number 2. This stimulus sequence was repeated 100
times at intervals of 400 msec, while evoked responses to the vertical
line were recorded from occipital (01 and O 2 ), parietal (P3 , P4 , and Pz ),
and temporal (Ts and Ts) derivations by use of a linked earlobe refer-
ence. 1 The subject then viewed 100 presentations of the same vertical
line, but now followed by the letter K. Evoked responses to the vertical
line were again recorded during this second stimulus sequence.
During the first sequence, in which the vertical line was followed
by the number 2, it was perceived as the number 1. During the second
sequence, when the vertical line was followed by the letter K, it was
perceived as the letter I. Thus the same vertical line (sensation) acti-
vated two different perceptions.
Using a PDP 12 computer, we computed the AERs and the stand-

1 Letters refer to electrode position in the International 10/20 System.

22 E. Roy JOHN




0.01 LEVEL - I1iW. jIl.

t. TEST i.L.~_ [J)N\.
S.J. S.J. S.J.

P4 T5

rJ\; ~~


0.01 LEVEL -+ ~f0 ..
_1'.1.1: ~J.C.

I 396 msee. I

FIGURE 7. (Top) Examples of averaged EPs to a vertical line

presented in a context of numbers (Line 1) and in a context of
letters (Line 2). The difference wave obtained by the subtraction
of Line 2 from Line 1 is shown in Line 3. Line 4 shows the value
of the t test at each point along this analysis epoch. Statistical
significant differences were obtained in parietal and temporal
derivations in the EP components located between 150 and 200
msec of latency. Each average EP was computed from 100
samples. Average responses, variances, difference wa'les, and
the t test were computed with a PDP-12 computer. (Bottom) Same
data from a second subject.

ard deviations from each deviation for the vertical line in the two
different sequences. The AER to the vertical line perceived as a number
was then subtracted from the AER to the vertical line perceived as a
letter. The significance of the resulting difference wave was assessed at
many points along the wave, each representing successive latency

increments of 2 msec, by use of the t test. The results from typical

subjects are illustrated in Figure 7.
Figure 7 shows that no significant differences were found between
the AERs to the vertical line under the two different perceptual sets in
the primary visual receiving areas (01 and ~). That is, the sensation
caused by the vertical line was essentially the same in both stimulus
sequences. However, significant differences did occur in the parietal
and temporal derivation. Essentially, the same procedure was used by
Johnston and Chesney (1974), who obtained results comparable to ours.
Differences between the two perceptual sets were found in frontal but

DiU. !\ ................ 0',& .. .0 '.rn=p . .



P3 01
Diff. V"'"" 0", "'V Q ....,.,... 6, V., ..."'-"

eT~M-l. ~ &~ r%"F\>.

B.K. 1 396111S1c I

FIGURE 8. (Top) The difference wave (Line 1) and the t test (Line 2) obtained
by comparison of average EPs (100 samples) elicited by a big A and a little Q.
Only the occipital location shows a statistically significant difference. (Bottom)
The same calculations, but now of EPs elicited by a capital A and capital E. All
the locations show highly significant differences between these EPs.
24 E. Roy JOHN

not occipital regions. No data were obtained from parietal or temporal

derivations in that study.
In the second experiment, a four-stimulus sequence was tachisto-
scopically presented, consisting of a large A, a small a, a large E, and a
small e. This sequence was repeated 100 times. AERs and standard
deviations were again computed for the response to each stimulus
from every derivation. The results are shown in Figure 8.
When the AERs elicited by small and large versions of the same
letter were compared, significant differences were found in the occipital
derivations. That is, large and small letters produce different sensa-
tions. However, no significant differences were found in temporal or
parietal derivations. Large and small versions of the same letter activate
the same perception, denoting a particular symbol in the alphabet.
Finally, when AERs elicited by As and Es of the same size were
compared, significant differences were found in all derivations. Both
the sensations and the perceptions elicited by two different letters are

F. Anatomical Distribution of the "Engram"

If these endogenous or readout processes represent the activation

of specific different memories, the anatomical distribution of these
endogenous electrical patterns provides information about the locus of
the neural representational system which mediates the storage and
retrieval of a memory. Traditionally this representational system has
been referred to as the engram.
By appropriate computer manipulations of AERs from different
stimulus-response combinations, the waveshapes of residuals reflecting
only exogenous or only endogenous processes can be constructed. By
these methods, it is possible to obtain a quantitative estimate of the
relative contributions of exogenous and endogenous processes to the
AERs obtained from any brain region. Such quantitative estimates have
been computed for a wide sample of brain regions in many cats, by the
use of data from thousands of behavioral trials of differential generali-
zation to auditory or visual stimuli (Bartlett and John, 1973). For each
brain region, the contribution of exogenous processes to the AER was
plotted versus the contribution of endogenous processes. The results
are illustrated in Figure 9.


. ~, L
.... MOTOR,
z.... .15
z S~~~~Y
0 .09 SYSTEM
.......Ja: .08
a: .07
z .06
'":l .05

.00 .10 .20 .30 .40 .50 .60
FIGURE 9. Plot of mean correlation coefficients between exogenous residuals versus
endogenous residuals for different neural systems and for different cue modalities.
Closed circles: Flicker frequencies as stimuli. Auditory system: N = 305; aud. cx. (16
cats), med. genic. (16), brach. info coil. (1). Limbic system: N = 303; hippocampus (16),
dentate (5), cingulate (5), septum (5), prepyriform (6), med. forebrain bundle (6), mamm.
bodies (5), hypothalamus (7). Mesencephalic nonspecific: N = 158; retic. form. (18), cent.
gray (1), cent. teg. tract (1). Motor system: N = 146; motor cs. (4), subs. nigra (10), nuc.
ruber (4), nuc. vent. (9), subthal. (5). Other sensory: N = 54; sensorimotor CX. (4), nuc.
post. lat. (1), nuc. vent. post. lat. (5), nuc. vent. post. med. (1) Thalamic nonspecific: N =
139; cent. lat. (13), nuc. retic. (6), nuc. reuniens (1), med, dors. (5), pulvinar (1). Visual
system: N = 394, visual cX. (18), lat. genic. (18), sup. coil. (2).
Open circles: Click frequencies as stimuli. Auditory system: N = 48; aud. cx. (5 cats),
med. genic. (5). Limbic system: N = 69; hippocampus (5), dentate (3), cingulate (3),
septum (3), prepyriform (2), med. forebrain bundle (3), mamm. bodies (3), hypothalamus
(2). Motor system: N = 37; motor cx. (1), subs. nigra (4), nuc. ruber (1), nuc. vent. ant. (5),
subthal. (2). Nonspecific system: N = 50, mesen. retic. form. (6), cent. gray (1), cent. teg.
tract (1), cent. lat. (3), nuc. retic. (3). Visual system: N = 55; visual cX. (6), lat. genic. (6),
sup. coli. (1).
N denotes the number of independent measurements within the designated system.
Data from monopolar and bipolar derivations were combined. Replications varied across
cats and structures. (Data from Bartlett et a/., 1975.)

Figure 9 reveals a systematic relationship between the contribution

of exogenous and endogenous processes to the AER: The amount of
endogenous process in any brain region is logarithmically proportional to the
amount of the exogenous process in that region. Both exogenous and
endogenous processes vary greatly from region to region. All regions
are not equivalent with respect to signal to noise ratio nor participation
in the representation of experience. In spite of these large quantitative
differences, the qualitative effects of a sensory stimulus are distributed
throughout the brain, and all brain regions participate in the represen-
tation of experience to an extent proportional to that afferent input.


The EEG and AER observations thus far described suggest that
neurons in widespread regions of the brain are involved in the media-
tion of learning and memory. A large body of evidence supports this
conclusion and has been thoroughly reviewed (John, 1967a,b, 1971,
1972; John and Thatcher, 1976). Abundant documentation has estab-
lished the existence of strong correlations between spontaneous or
evoked activity on one hand and neural discharge on the other. In
view of the anatomically extensive changes in slow waves during
learning, these correlations lead one to expect that changes in single
unit activity during learning should be detected with ease. That such
is indeed the case can be seen from studies of single neurons during
conditioning, which consistently report that a large proportion of cells
(10%-60%) alter their response to the CS during conditioning (Jasper,
Ricci, and Doane, 1960; Morrell, 1961a; Morrell, 1967; Olds and Olds,
1961; Bure!!, 1965; Bure~ and Bure!!ova, 1965, 1970; Kamikawa, Mc-
nwain, and Adey, 1964; Buchwald, Halas, and Schramm, 1965; Hori
and Yoshii, 1965; Adam, Adey, and Porter, 1966; Yoshii and Ogura,
1960; Woody, Vassilevsky, and Engel, 1970; Travis and Sparks, 1967,
1968; Travis, Hooten, and Sparks, 1968; Travis, Sparks, and Hooten,
1968; O'Brien and Fox, 1969a, 1969b; Olds and Hirano, 1969; Olds,
Disterhoft, Segal, Kornblith, and Hirsch, 1972; Leiman and Cristian,
The fact that such a large percentage of neurons change their
behavior during even the simplest learning task constitutes a strong
argument against connectionistic formulations which localize learning
as synaptic alterations causing facilitation of the firing of cells in se-
lected pathways, with such firing constituting the activation of memory

about the learned experience. If mere neuronal firing per se constituted

information about present or prior experience, such representations
would be hopelessly ambiguous since so many cells alter their reactivity
in a single learning session and since neurons also fire spontaneously,
display variable responses to identical stimuli, and discharge in re-
sponse to a wide variety of events. Such reasoning, discussed else-
where in great detail (John, 1967b, 1971, 1972; John and Thatcher
1976), has led to the formulation of a statistical theory of neuronal
representation of memory, which proposes that information is repre-
sented by the time sequence of deviations from random or baseline activity
in extensive ensembles of neurons. In this formulation, learning is a
cooperative process and the activity of any single neuron is information-
ally relevant only insofar as it contributes to the statistical features of the
behavior of the ensemble.
With the use of multiple chronically implanted movable microelec-
trodes, neuronal responses to discriminated stimuli have been studied
(John and Morgades, 1969a, 1969b, 1969c). As the movable microelec-
trodes traversed across extensive anatomical domains, it was found that
both the waveshape of the AER and the firing pattern of neuronal
ensembles (average poststimulus histograms) were basically similar at
different electrode positions. Analysis of variance was carried out com-
paring the variance within the responses to the two different signals at
each region to the variance between responses to the same signal in
many electrode positions. The difference in response to the two dis-
criminated signals at any location was greater than the variability in
response to a particular signal at different locations. This finding is
illustrated in Figure 10.
In other words, not only AER waveshapes but the firing patterns of
neuronal groups in different brain regions displayed a characteristic
temporal pattern whenever a particular learned signal was presented.
Different anatomical regions were qualitatively homogeneous in their
representation of a particular learned stimulus. However, signals with
different significance elicited different patterns of ensemble responses.
Finally, for three years my colleagues and I have been studying
unit behavior in discrimination learning (Ramos, Schwartz, and John,
1974, 1976a-d; John, 1974; Ramos and Schwartz, 1976a, 1976b;
Schwartz, Ramos, and John, 1976). In those studies we have utilized
chronically implanted, movable microelectrodes to examine the re-
sponses of small groups of neurons and, more recently, well-isolated
single neurons during correct responses and errors in tasks requiring

2 liz 8 tlz 2 Hz minus 8 tlz

(Grand average)
evoked e,
P value
o 125 msec 0 0 125 msec
(Grand average)
Post stimulus 6
histograms PSH
P value

o 125 msec o 125 msec 0 125 msec
FIGURE 10. (Top left) The top curve shows the grand average of the AERs elicited
by the 2-Hz es across all electrode positions in the mapped region, while the
lower curve shows the standard deviation (S.D.) of the group of AERs. (Bottom
left) The top curve shows the grand average of the poststimulus histograms
(PSHs) elicited by the 2-Hz es across the same electrode positions and the lower
curve shows the standard deviation. (Top center) The grand average of the AERs
elicited by the 8-Hz es and the corresponding S.D. (Bottom center) The grand
average PSH elicited by the 8-Hz es and its S.D. (Top right) The top curve shows
the difference waveshape resulting from the subtraction of the grand average AER
elicited by the 8-Hz es from the grand average AER elicited by the 2-Hz es. The
lower curve shows the p value, as computed by the t test, for each point of the
difference wave. (Bottom right) The top curve shows the difference waveshape
resulting from the subtraction of the grand average PSH elicited by the 8-Hz es
from the grand average PSH elicited by the 2-Hz es. The lower curve shows the p
value for each point of the difference. (Data from John and Morgades, 1969b.)

differentiated behavioral responses to discriminated visual stimuli and

also during differential generalization to ambiguous stimuli delivered
to differentially trained cats.
EPs and unit responses were simultaneously recorded from both
cortical and subcortical microelectrodes in these cats. Cortical electrodes
were located both in specific projection and in association areas. Com-
puter-pattem-recognition techniques (John, Bartlett, Shimokochi, and
Kleinman, 1973) and multidimensional scaling methods (Ramos,
Schwartz, and John, 1976a-d) were used to classify single EPs from
trials resulting in correct versus erroneous performance to the same CS

or from differential generalization trials in which two different behav-

iors were elicited by the same novel test stimulus. This classification
procedure identified the EPs as belonging to one or another of the
"readout modes" which reflected the activation of memories about
different stimulus-response contingencies. Particular readout modes
were found to be differentially correlated, at extremely high signifi-
cance levels, with the subsequent behavioral performance. Therefore
occurrence of a particular readout mode can be interpreted as evidence
that the stimulus eliciting that electrophysiological and behavioral
response was perceived as a signal with a particular significance.
When the pattern-recognition procedure had identified the readout
mode activated by each stimulus in the behavioral trial, the firing
patterns of the simultaneously recorded unit activity corresponding to
each readout mode were separately analyzed. We found two types of
neurons in these analysis. One type showed an invariant average
pattern of response to a given stimulus, no matter how it was perceived
(Le., no matter what behavioral response ensued). These units might
be described as "stimulus-bound," responding to the signal in the
same way independent of perception. The second type of neuron
showed one average temporal pattern of response during one readout
mode and a different average temporal pattern of response during
another readout mode. These units might be described as "gnostic"
units, with a response pattern related to the perception rather than
determined by the sensation. Such units showed great variability in
their responses during a specified readout mode but displayed a char-
acteristic and specific average response in each mode. Different units of
this type in the same anatomical region showed closely similar average
responses during the same mode, although their responses to single
stimuli were poorly synchronized. These findings suggested an "er-
godic" hypothesis, that is, the response to a single stimulus presentation
eliciting a specific readout mode averaged across the set of units of this type
corresponded to the response of any unit of this type averaged across a set of
stimulus presentations eliciting that specific readout mode. In view of these
data, it would seem that the perception of a stimulus is mediated by the
averaged temporal firing pattern of an ensemble or set of units of this
type. The activity of any single unit is important only insofar as it
contributes to the statistical behavior of the ensemble (Ramos,
Schwartz, and John, 1976a, 1976b, 1976c, 1976d).
Further, firing of any unit per se seems to have no unique informa-
30 E. Roy JOHN

AEP 30jlV


2Qijms 20I~60jlV

:. AEP
AvJVv ~30)lV


FIGURE 11. On the left of this figure is shown the way in which single
evoked responses obtained from the ends of several differential generaliza-
tion trials were classified by use of a multidimensional scaling technique.
Open circles represent EPs from trials resulting in eN.. The open and
closed circles fall into domains which are clearly different, showing that a
consistent waveshape or mode difference existed between the two sets of
EPs referred as Mode 1 and Mode 2. In the middle of the figure are shown
the average of the EPs in the two portions of the signal space. On top is the
average of the EPs classified as Mode 1, while the average of the EPs
classified as Mode 2 is on the bottom. These two averages display clear
differences in a late component. On the right of the figure are shown two
PSHs of the firing pattern of two different cells separated by computer
spike-height discriminators during each mode. The two PSHs on the top
show the firing pattern of these two cells during Mode 1 corresponding to
the readout of memory related to CR,. The two bottom PSHs show that one
of these cells displayed a completely different firing pattern during Mode 2
corresponding to the readout of memory relating to eN..

tional value. All units thus far observed not only fire "spontaneously"
and show great variability in response to any stimulus but show a
differential response consisting of differences in graded temporal patterns,
rather than an all-or-none behavior.
As yet, we have not observed a single neuron which fired only
during one memory readout mode and not during the other. An exam-
ple of these findings, showing the different AER readout modes and the
corresponding firing patterns of well-resolved units isolated by elec-
tronic discriminators, is presented in Figure 11.


The electrophysiological observations summarized above provided

the basis for a tentative picture of how memories were stored in the
brain. The data suggested that many regions of the brain were directly
or indirectly affected by presentation of the CS. Rapid informational
transactions between these regions resulted in the establishment of a
common mode of neural activity, reflected by the emergence of similar
AER waveshapes in different brain regions as learning progressed.
Presumably these similar electrophysiological processes represented
not only the influence of the stimulus but acquired information about
the stimulus-response contingencies as well as the emotional reactions
and motivational state of the animal. Stimulus-bound features of this
common mode of activity were represented by relatively short-latency,
exogenous processes in the AER, while reactive features related to the
meaning attributed to the stimulus were represented by longer-latency,
endogenous processes. As the learned behavior became well consoli-
dated, this anatomically extensive representational system acquired
the capacity to reproduce an accurate facsimile of the common mode of
activity in the absence of the conditioned stimulus, when situational
factors or generalization stimuli activated the corresponding memory.
These facsimiles of the effects of previous meaningful experiences
were apparently generated in a cortico-thalamic-mesencephalic sys-
tem, whence they propagated centrifugally to other regions of the
Investigation of the neuronal activity correlated with these slow-
wave and EP phenomena revealed that extensive neuronal populations
displayed an average temporal firing pattern corresponding to the
different memory readout modes. Single units displayed highly varia-
ble firing patterns to individual stimulus presentations but converged
to the average ensemble behavior as their long-term behavior was
observed. It appeared that the activity of any single neuron was infor-
mationally significant only insofar as it contributed to the ensemble
statistics. No evidence was obtained for the existence of neurons whose
firing uniquely denoted the identification of a specific CS, as would be
required by theories attributing learning to the establishment of new
connections in specific neuronal pathways.
Electrical stimulation of the brain with appropriate temporal pat-
terns of current provided a method of testing of these inferences to
ascertain whether this model was sufficiently accurate to permit direct
32 E. Roy JOHN

control of decision-making behavior by the manipulation of the average

firing pattern of neuronal ensembles in differentially trained animals. A
series of such experiments have been performed.

A. Rapid Transfer to Direct Electrical Stimulation of the Brain

A group of cats was trained to discriminate between two different
repetition rates of either visual (Vl and V2) or auditory (Al and Az)
stimuli. After prolonged overtraining, these animals were subjected to
electrical stimulation of the mesencephalic reticular formation (RF) with
high-frequency current pulses at the same two repetition rates (RFl and
RF2). Immediate discriminated response at high levels was displayed by
most of the animals, and all animals in the study achieved criterion
rapidly, indicating a very large transfer of training effect (John and
Kleinman, 1975). Subsequent stimulation of the visual cortex (VIS),
lateral geniculate (LG), medial geniculate (MG), and intralaminar nuclei
of the thalamus (INT) showed less consistent results, but marked trans-
fer was displayed by at least one animal for each of these structures.
Electrical stimulation of this sort cannot conceivably activate particular
pathways within the stimulated domain in a selective way, correspond-
ing to the hypothetical circuitry postulated by theories which attribute
learning to the establishment of new connections or facilitation of
unique pathways. It seems necessary to concede that these gross stimuli
can only accomplish massive coordinated or coherent firing patterns in
the stimulated neuronal population.

B. Peripheral-Central Conflict
After transfer of the discriminated behaviors from the peripheral
modalities to direct electrical stimulation, "conflict" studies were car-
ried out in which RF stimuli at either frequency were pitted against
concurrent discordant visual or auditory stimuli of the other frequency
(RFl vs. V2, RF2 vs. Vl t RFl vs. Az, RF2 vs. Al). Parametric tests were
performed that explored the effects of ascending and descending series
of RF current values. In every case, values of RF stimuli were found
such that central stimulation at either frequency completely or substan-
tially controlled the behavioral outcome, preempting the decision and
effectively contradicting the concurrent visual or auditory cue. These
results are illustrated in Figure 12.

In contrast, conflict studies using LG stimulation revealed that LG

stimuli could control behavior only when contradicting visual cues and
then only when the visual cues were at the slower repetition rate
(Kleinman and John, 1975). These results suggested that RF stimuli
provided uniquely intimate access to the decision-making system,
effectively contradicting events in either peripheral sensory modality,
while LG stimuli seemed rather to simulate visual sensations.

C. Perceptual Integration
Processes of perceptual integration were investigated in the same
animals. The cats were trained using a 4-Hz signal and a 2-Hz signal as
the discriminated cues. After stimuli at this repetition rate had been
established as effective signals whether delivered via vision, audition,
or electrical stimulation of RF, VIS, LG, MG, or INT, temporal summation
studies were carried out. Modalities of input were studied two at a
time, in all possible combinations. In each study, simple stimuli at 4 Hz
and 2 Hz were delivered in random order via the two selected modali-
ties in random sequence. This series of simple stimuli served as control
measures. Interspersed within this series were two kinds of compound
stimuli. "In-phase" compound stimuli consisted of 2 Hz signals deliv-
ered simultaneously via the two modalities. These signals served as a
control for total current and modality interaction effects. "Out-of-
phase" compound stimuli consisted of 2-Hz signals delivered via the
same two modalities but occurring 250 milliseconds out of phase. These
signals served to test whether the animal would behave as if the two
separate signals were perceived as individual 2-Hz cues, or whether
they would be perceived as a unified 4-Hz cue. Although some pairs of
modalities showed a much higher capability for cross-stimulus integra-
tion than did others, highly significant levels of integration were found
for most of the combinations studied (John and Kleinman, 1975). Thus
it appeared that informational events in many different brain regions
could be integrated into a perceptual whole.

D. Loci Responsible for Perceptual Integration

In order to identify the brain regions mediating this cross-modal

integration, EEG and AER measurements were obtained during such
34 E. Roy JOHN

Cat No.2 (++) Cat No.6 (--)

N = 125 ti = 123
o 80
os 60

E 40

~ 150 200 250 300 350 40 50 60 70 80
,., Cat No. 1 (--) Cat No. 5 (--)
N = 114 N = 112
g 100 100

80 80
60 60
c: , .....' ..
20 20
-' ,
30 50 70 90 110 25 30 35 40 45
Stimulus (u.a) V, _ _
V2 ~--..

Cat NO.3 (++) Cat NO.6 (--)

N = 273 ti = 74
100 ...----- 100

~E :: /~;""""".
40 , . 40 /
III 20
60 80 100 120 140 50 60 70 80 90

Cat No.2 (++) Cat No.4 (++)

ti = 164
e ti 143

- 100 100

c: 80 80

o ,
60 60
'" 40 40
~ /
'" 20
.' 20
c.. .'
---150 200 250 150 200 250 300
Stimulus (J.ta)

summation experiments. During out-of-phase stimulation at 2 Hz via

two modalities which resulted in behavior as if a 4-Hz stimulus had
been perceived, most brain regions displayed labeled responses and
AER waveshapes showing only a 2-Hz component. The only exceptions
to this were the VIS and the INT. These structures showed a marked 4-
Hz component when cross-modal integration occurred. This was partic-
ularly marked and consistent in the INT, no matter what pair of modalities
was being integrated. These findings are illustrated in Figure 13A-E.
These findings suggest that a system including the visual cortex
and the INT reticular formation plays a particularly important role in
cross-modal perceptual integration.

E. Role of Cortex and Thalamic Reticular Nuclei

Other brain stimulation results confirm the apparent importance of
the sensory cortex and the INT in perceptual processes. We have
observed (John, 1963) that it is possible to interrupt stimulus-con-
trolled behaviors by electrical stimulation of a wide variety of brain
regions concurrent with the presentation of a sensory cue. At a
sufficiently high current, the brain stimulus blocks performance to the
CS (occlusion). In many cases, this occlusion persists for many
seconds or even a few minutes after termination of the electrical
stimulation. Recording during this "poststimulus absence" reveals
high-voltage spindle waves in the INT independent of the locus of the
electrical stimulation which produced the absence. Seizurelike after-
discharges can be produced in a variety of regions, especially in the

FIGURE 12. Each graph shows the effectiveness with which stimulation of the mesence-
phalic RF at either of two frequencies (RFl and RF2 ) contradicted simultaneously pre-
sented visual stimuli (V2 and Vl' top) or auditory stimuli (Ao and At< bottom), plotted as a
function of increasing current intensity. For cats 1, 3, and 6, frequency 1 was 4/sec and
frequency 2 was 2/sec. For cats 2, 4, and 5, frequency 1 was 5/sec and frequency 2 was 1.81
sec. Solid lines show the outcomes when peripheral stimulation at the higher frequency
(Vl in top graphs, Al in bottom graphs) was pitted against RF stimulation at the lower
frequency (RF2 ), while the dotted lines show the outcomes when the higher-frequency
stimulus was delivered to the RF. Cats 1, 5, and 6 were trained to perform an avoidanc~
avoidance discrimination (- -), while cats 2, 3, and 4 were trained to perform approach-
approach discrimination (+ + ). N refers to the total number of conflict trials carried out in
each cat, accumulated in three sessions for cats 2, 5, and 6 and four sessions for cat 1
(visual-RF conflict), and in three sessions for cat 2, four for cat 6, five for cat 4, and seven
for cat 3 (auditory-RF conflict). (Data from Kleinman and John, 1975.)
36 E. Roy JOHN

CR as 2
GENICULATE _----\'----".--.....J._____'..... ~I_____'.I___'.~\____.\____.~.~.~.~I_____



,.,.: ~~~{1\{4Vl)RM.
2+2 ~ 2

FLICKER - - - - 4 - -........""""'-.................................-......-------
f- 1 SEC -lj

FLICKER ., \ \ \ I I I I \ I \ \ I \ \ I I

STIMULUS " , \ I , .

o .v'J.'1~1 ~~"'1~"MV;-."

~1 SEC~

CAR LEFT (2+2 4)

FIGURE 13. Data from a cat CAR RIGHT (2+2=2)
trained to press a, lever on the 7 TRIALS N-187
left side of a work panel to avoid
foot shock when a 4/sec es was AVERAGE A
presented in any of a number of EWlCED
modalities, including electrical RESPONSE
stimulation of the brain, and to
press a lever on the right side of
the work panel to avoid foot
shock when a 2/sec es was pre-
sented in any of the same mod- L2~ IIS.J...2~ IoJSoI
110 R. IIRF 110 R.IlAF
alities. 9O)lA 6O"A 60.. A
(A-D) Each of these figures
shows the activity recorded from
the intralaminar-midline thala- DIFFERENCE
mus under two conditions: WAVE
(Top) Presentation of a 2/sec es (CAR LEFT -CAR RIGHT)
in one modality (MI) plus a 2/sec
es in a second modality (M2)
delayed by Z50ms with respect to t _TEST
p Ol--l/'r'),.,~
MI resulting in performance of E
the behavior appropriate to a 4/
sec es: (Bottom) Presentation of the same compound stimuli resulting in behavior
appropriate to a 21sec es. A-D differ in the modalities of es presentation. (A) MI = LG
es; M2 = VIS es. (B) MI = peripheral flicker es; M2 = VIS es. (C) MI = peripheral
flicker es; M2 == mesencephalic RF es. (D) MI = peripheral flicker es; M2 =
mesencephalic RF es. Note that when the 2/sec ess in the two modalities were
effectively merged, a 4/sec rhythm is prominent in the intralaminar record. If the two
signals were not effectively combined, a 2/sec rhythm dominates the record. This is
particularly striking during behavioral vacillation (D).
(E) stimulation as above with MI = medialis dorsalis and M2 = right mesencephalic
RF. All data recorded from right VIS, bipolar derivation. (Top left) AER when effective
merging of the two 2/sec ess occurs as indicated by behavioral performance of the
animal appropriate to a 4/sec es. Note that at the cortex, essentially four afferent volleys
per second are occurring. Below is the variance. (Top right) Stimulation identical as top
left, however, the two signals were not effectively combined as indicated by the
behavioral performance of the animal appropriate to a 21sec es. Note that at the cortex
essentially two afferent volleys per second are occurring, the EP to the right mesence-
phalic (RF) stimulus is barely discernible. Below is the variance. (Bottom) Difference
wave obtained by subtraction of the average EP of trials in which summation failed to-
occur (top'right) from the average EP of trials in which summation occurred (top left).
The I-test wave at the bottom shows that the difference between the two average
EPs is statistically significant.
38 E.RoYJoHN

limbic system, by such a procedure. However, by careful adjustment

of current values and recording from the stimulated region as well as
the INT, it is possible to confirm that such absences can occur when
there is no indication of electrical seizure in the stimulated regions.
Presentation of the es during the period of intralaminar spindles may
elicit an orientation reflex, but the cat shows no sign of comprehend-
ing the signal. As the spindle waves vanish, the animal often gives a
startle response and looks about in an agitated fashion, as if awaken-
ing. Thus, interference with the activity of the INT serves to disrupt
Finally, we have used electrical stimulation to explore the infor-
mational significance of early and late components in the AER from
visual cortex, by phase-locking high-frequency current pulses to the
es so as to coincide with the short-latency exogenous or longer-
latency endogenous processes (John, 1967b). Electrical stimulation of
sensory cortex during the exogenous components produced no disrup-
tion of discriminative responses, while stimulation with identical
current parameters timed to coincide with endogenous processes (80-
110 milliseconds) totally abolished eRs. During such stimulation, the
cat would orient toward the es but would behave as though it
possessed no cue value. These findings show that the late components
of the cortical AER, generally considered to reflect the influence of the
nonsensory-specific mesencephalic and thalamic reticular systems, are
essential for identification of the meaning of a stimulus. It is not clear
whether these results should be interpreted as indicating that the
essential functions are performed at the cortical level, or whether the
cortical stimulation precludes the corticothalamic outflow necessary to
establish an interactive transaction between these sensory-specific and
nonsensory-specific domains of the brain.



The evidence which has been summarized suggests that informa-

tion, past or present, is represented in the brain by a statistical process,
the average spatiotemporal pattern of activity in anatomically extensive
neuronal populations. The activity of the single neuron is not informa-
tionally significant except insofar as it contributes to the activity of the
ensemble. The same information can be represented in diverse regions,
with a varying signal-to-noise ratio (SIN). In any region, some cells

appear to be stimulus-bound, displaying the same average firing pat-

tern to stimuli independent of how they are perceived, although they
may display different response patterns to different stimuli. Such cells
would appear to be relatively reliable reporters of sensation, in terms of
their ability to construct reproducible average firing patterns character-
istic for each different stimulus, in spite of the short-term variability of
their responses. Other cells in the same regions display average pat-
terns of response to the same stimulus which are more reactive, de-
pending upon the meaning attributed to the signal. These latter cells
seem to be involved in perceptual processes and in the storage of
memories about the stimulus-response contingencies. The mixture of
these two types of cells varies from region to region, producing variable
SIN for both exogenous and endogenous processes, that is, making
different relative contributions to sensation and to perception.
Consciousness, subjective experience, the concept of self, and self-
awareness, while representing successively higher orders of informa-
tion, must nonetheless also be mediated in the same statistical fashion.
There is no compelling evidence or logical argument to suggest that
these higher levels of information are represented by qualitatively
different neuronal processes. The content of consciousness is the sum of
all informational processes in all the various functional systems of the
brain. The information in each area comprises a coherent temporal
pattern. The outflow of this coherent pattern to other brain regions
constitutes afferent input elsewhere. The result of these rapid multidi-
rectional transactions of information between different regions estab-
lishes a common mode of activity shared by many anatomical regions,
with the relative contribution (SIN) of each type of activity varying from
region to region as a function of its afferent connectivities.
These regional messages, each with its characteristic pattern and 51
N, converge upon the cortical association areas and via collateral path-
ways and corticofugal pathways upon the INT, the mesencephalic RF,
and the limbic system. In man and other mammals, consciousness
depends upon integrity of the thalamic and mesencephalic reticular
systems. Lesion of these systems produces long-lasting or permanent
coma (Moruzzi and Magoun, 1949). Yet the fact that recovery from such
coma sometimes ensues or that multistage lesions of these regions fail
to produce a comatose animal or to interfere with information process-
ing (Adametz, 1959; Chow and Randall, 1964) suggests that the process
is distributed and can be effectively mediated by other brain regions
under appropriate circumstances.
40 E.RoY]OHN

Similarly, although voluminous data from neurological clinics at-

tests to the catastrophic effects of brain damage in certain regions upon
specific perceptions or other higher intellectual functions because of
head trauma or cerebrovascular accidents, yet the literature also
abounds with evidence of functional compensation for a good part of
such damage with time although the damaged tissue was irreversibly
destroyed. A substantial body of evidence indicates that retention of
preoperatively learned tasks often occurs in mammals when brain areas
relevant to the task are removed serially in multiple-stage operations,
although identical lesions made in a single-stage operation may abolish
performance and prevent reacquisition. Interestingly, if animals sub-
jected to multiple-stage lesions of the VIS are deprived of visual experi-
ence between surgeries, the resulting visual deficit is comparable to the
effects of a one-stage ablation. Recently it was demonstrated that rats
permitted unrestrained movement in a patterned visual environment
during the interval between two-stage lesions of the VIS can rapidly
relearn a pattern discrimination established prior to surgery, while rats
passively transported through the same environment fail to do so (Dru,
Walker, and Walker, 1975). These findings suggest that functional
reorganization must occur during a multiple-stage procedure, that sen-
sory stimulation in the damaged modality is crucial for such reorganiza-
tion, and that cross-modal transfer may facilitate this process.
Phenomena such as compensation for brain damage and absence of
functional deficits after multiple-stage lesions constitute strong evi-
dence that the brain possesses alternative methods for performing
many functions which are mediated by some particular structure under
ordinary circumstances.
Perhaps it is possible to reconcile the evidence that information
about particular kinds of experience appears to be localized-because
local lesions can cause such discrete dysfunctions as alexic agnosias or
loss of previously learned discriminations-with the evidence adduced
above that information is distributed throughout many different ana-
tomical regions. The electrophysiological data indicate that the endo-
genous processes reflecting particular perceptual and cognitive func-
tions have a very widespread distribution. The fact that the values
found for these processes span a range of 1,000 indicates great quanti-
tative differences between anatomical regions in the density or inten-
sity of representation of a memory. Perhaps under normal circumstan-
ces, the brain of an individual requires some threshold value for the SIN
to be exceeded in order for information represented by the correspond-

ing neural activity to be functionally useful. If that threshold value is

surpassed by only one particular brain region, damage to that region
will produce impairment of that function, while such symptoms should
not result from damage of any other region. Nonetheless, the relevant
information is available in many other places. Perhaps if the threshold
value for the SIN could be lowered, restoration of the impaired function
might be achieved even though irreversible damage had been sus-
tained by the region which previously achieved the highest SIN.
This reasoning seems particularly plausible if we consider that the
greater informational reliability of a high SIN, as life experiences accu-
mulated, would tend to establish functional dependence upon the
region displaying the highest SIN and a learned threshold setting
which would reject information from regions with a lower SIN. Thus a
learned functional inhibition might even be established which pre-
vented such alternate regions from resuming functional utility in the
event that the region usually mediating that function were damaged.
These speculations offer a way to reconcile the large body of
evidence about specific agnosias and other deficits, which seemingly
involve memory and consciousness and result from localized brain
damage, with the apparently contradictory electrophysiological find-
ings in our studies. For example, perhaps the SIN for activity related to
the perception of letters and numbers is highest in a particular cortical
region and the threshold in the normal brain is usually set to reject
lower SINs for that activity. Such lower SINs might exist in other
regions. Thus although information relevant to the perception of letters
and numbers is available in those regions, damage there will not result
in alexic agnosia, nor can those regions sustain such perception alone if
the salient cortical field is damaged. While I have no evidence at present
that these speculations are correct, they provide an attractive working
hypothesis-attractive not only because thus no contradiction need
exist between two bodies of data (lesion and electrophysiological), both
of which reflect real aspects of brain function, but also because were
this hypothesis correct, much functional deficit due to brain damage
which we now consider irreversible might be reversed by procedures
which lower the relevant thresholds or block the learned functional
Perhaps the apparent dependence of consciousness upon an intact
reticular formation can be similarly explained. The convergence of
information from every sensory region and many other functional
systems upon the reticular formation create here a uniquely favorable
42 E. Roy JOHN

anatomical design for integration to take place. Anatomy also favors the
effective and widespread distribution of activation from the RF
throughout a host of other brain regions. These factors may well give
the RF of the mesencephalon and thalamus a unique ability to achieve a
high SIN both for convergent afferent input and divergent efferent
output. The brain may well come to rely upon this high SIN, setting
thresholds as a result of experience such that it comes to depend upon
these regions both for integrative processes and for the maintenance of
consciousness, in the sense of organizing excitability in the system so
that sufficiently high levels of coherence may be achieved. Single-stage
lesions in this system, as attested by a huge volume of literature, do
result in long-lasting or permanent loss of consciousness. Perhaps the
ability of the brain to maintain consciousness and integrative activity
when this sytem is destroyed in multiple stages is due to systematic
increases in the relative SIN of other regions participating in the same
functions, increased absolute coherence in those regions as the func-
tional inhibition is extinguished, and lowering of the threshold for
acceptable SIN as the series of lesions is inflicted.
Thus, as with many other functions, it may well be fruitless to ask
whether any brain region is uniquely responsible for consciousness
and subjective experience. These functions are probably distributed
over a widespread anatomical substrate, every region of which makes
a contribution to the overall process and many of which may be capable
of sustaining the process if damage occurs elsewhere in the system.
Is this an evasion of the issue? Is the whole system conscious?
Under normal circumstances, whether or not "backup" systems exist, is
there some circumscribable system which mediates consciousness of
the fluid patterns of information and the continuity of subjective expe-
rience? What is the nature of the process constituting the intimate basis
of the emergent property of subjective experience, which transcends
the activity of the constituent elements of the system?
The electrophysiological data reviewed above indicate that sensa-
tions are encoded as organized spatiotemporal patterns of average activ-
ity in stimulus-bound neurons whose density varies from region to
region of the brain, with a concomitant variation in SIN. Perceptions are
similarly encoded as average spatiotemporal patterns, but in ensembles
of neurons capable of responding to the arrival of information from the
stimulus-bound ensemble with a firing pattern creating a facsimile of
responses previously displayed to other events. These intermingled
firing patterns, rapidly evolving into a common mode which represents

sensations and perceptions in many modalities, as well as information

about drive levels from the hypothalamus and about affective state from
the limbic lobe, are distributed throughout widespread brain regions
but converge most intensively upon the "centrencephalic" thalamic and
mesencephalic reticular systems. Outflow from these systems feeds
back upon cortical and thalamic regions which contributed high SIN to
the afferent barrage upon the centrencephalic system, further enhanc-
ing the SIN of the reverberating cortico-thalamic-centrencephalic pat-
tern which emerges.
As a result of this reverberation, unusually high levels of local
coherence are achieved in the participating ensembles. These ensem-
bles do not become fully synchronized but do achieve higher coherence
levels and higher SIN than could occur without this feedback process.
As these coordinated temporal patterns of firing occur in the
densely packed cells of the centrencephalic system, the membranes of
the participating cells are depolarized and ionic shifts occur. Potassium
concentration increases in the extracellular space, and ionic binding
probably occurs to mucopolysaccharide filaments and on the surfaces of
glial cells. Complex gradients of charge are thereby established, with
distributions which depend upon the spatiotemporal coherence pat-
terns in the neuronal ensemble. One can envisage a complex, three-
dimensional volume of isopotential contours, with a topology encom-
passing portions of neuronal membranes, glial membranes, and extra-
cellular binding sites. Let us call this set of isopotential contours or
convoluted charge surfaces a hyperneuron.
Every representational system has a corresponding particular dis-
tribution of energy, a unique hyperneuron. The special features of a
particular hyperneuron will be determined by the statistical processes
in local ensembles which established the set of coherent spatiotemporal
patterns within this volume of neural tissue. The contribution of any
individual cell to a hyperneuron will be insignificant. Ensembles of
neurons in regenerative circuits will contribute a stable component to
all or most hyperneurons, while ensembles with lower positive feed-
back will make more variable contributions. Thus one can envisage
sequences of hyperneurons which would display stable, invariant
features as well as modulated, reactive features.
We postulate that the property of consciousness emerges from the
cooperative interaction of neuronal populations, resulting in the estab-
lishment of hyperneurons whose characteristics transcend the features
of the cellular constituents of the ensemble, which serve as responsive,
44 E. Roy JOHN

charge-distributing elements. The content of subjective experience is

the momentary contour of the hypemeuron. As the contour of the
hypemeuron fluctuates in time, the content of consciousness varies, so
that the invariant features of the hypemeuron constitute the "1" of self-
awareness, while the variable features are the fleeting "here-now" of
the momentary subjective experience.
It may be that only the centrencephalic system and perhaps the
limbic and cortical neuronal masses can sustain the hyperstructure
required for a hypemeuron to develop. On the other hand, the brain
may contain an extended hypemeuron with many lobules located in
different anatomical regions and all interconnected with each other and
modulating the centrencephalic-cortical hypemeuron. A certain critical
mass and critical density of elements may be a prerequisite for tissue to
be capable of sustaining a hypemeuron, or this cooperative process
may be a property of any mass of neural tissue.
It is conceivable that a "rudimentary hypemeuron" can exist in any
form of living matter, with a complexity of experience limited by the
number of energy states which the matter can attain. A priori, there
seems no compelling reason to insist that this cooperative process is
restricted only to certain types of brains or to certain types of tissue. We
simply do not know enough about the essential features upon which
this emergent property depends to be arbitrary about which organisms
can and cannot possess it.
Finally, we have postulated that mental experiences are produced
by and consist of cooperative electrochemical phenomena which arise
within volumes of neural tissue. Yet the crucial features of neuronal
masses for the production of subjective experience may not depend
upon the neurons themselves. Were the hypemeuron postulate accu-
rate, it would not be clear whether the property of mental or subjective
experience arose from the action of the charge contours upon the
neurons present within that complex field or whether the subjective
experience were an intimate consequence of the energy distribution
itself. Neurons may not possess any inherent quality essential for this
transformation but may be uniquely well suited for the production of a
wide variety of improbable distributions of energy. Were it possible to
achieve comparable distributions of energy without neurons-in other
words, to simulate' a "neuron-free" hypemeuron-perhaps quite the
same subjective experience would arise. Subjective experience may
actually be a property of a certain level of organization in matter.
This article began with a series of questions about the nature of

subjective experience. A body of experimental evidence has been re-

viewed which provides some insight into the brain mechanisms which
mediate information representation, memory retrieval, and decision
making and suggests possible answers to those questions. In view of
the distributed statistical nature of the representational processes re-
vealed by such experimental studies, it seemed necessary to postulate
that subjective experience is the product of a cooperative process
involving both cellular and extracellular constituents of neural tissue,
most probably in the centrencephalic system, to which we assigned the
label of hyperneuron.
This postulate can be subjected to test. If it is correct, then there
must exist physical parameters of the energy distribution which will
alter the content of subjective experience when manipulated. A major
experimental task for us must be to ascertain what features of organized
energy in neural tissue produce subjective experience. As these essen-
tial features become more apparent, it will be possible to develop a
better-informed basis for evaluating whether this emergent property is
necessarily limited to brains with certain architectonic specifications,
exists in any neuronal systems, is a general property of living matter, or
might arise in a sufficiently organized system of energy.
Although these questions will be extremely difficult to answer, I
am confident that answers will be provided, and relatively soon. A
great deal has been learned about informational processes in the brain,
and additional information is steadily accumulating. Something very
much like the postulated hyperneuron must exist, and it is only a
question of time until we understand it.
There is one aspect of this set of issues, however, which I still find
totally baffling. Reality is a continuously fluctuating distribution of
physical energy in different frequency domains, located in various
regions of space. This energy, impinging upon the receptors of the
body, causes the firing of neurons in afferent pathways and ultimately,
if our postulate is correct, produces a modulation in the contours of a
hyperneuron. Subjective reality, produced by this hyperneuron, is a
constellation of vivid colors, shapes, sounds, and images synthesized
from these neuronal patterns, reflecting the energy spectrum of reality
in a reproducible but not literal fashion. Reality is not our experience of
reality. How is this transformation accomplished? What aspect of the
hyperneuron's dimensionality might produce the rich, diverse qualities
of this abstraction from reality? Perhaps this is the fundamental ques-
tion, and as yet I fail to discern the faintest glimmering of an answer.
46 E. Roy JOHN


This work has been supported in part by Grant #MH20059 from

the National Institutes of Health and grant #BMS 7502819 from the
National Science Foundation. Dr. John is a Career Scientist of the
Health Research .council of the City of New York under Grant #1-752.


ADAM, G., ADEY, W. R., AND PORTER, R. W. Interoceptive conditional response in cortical
neurones. Nature, 1966,209,920-921.
ADAME1Z, J. H. Rate of recovery of functioning in cats with rostral reticular lesions.
Journal of Neurosurgery, 1959,16,85-97.
ASRATYAN, E. A. Changes in the functional state and pattern of electrical activity in
cortical areas involved in the establishment of conditioned connection. Proceedings of
XXIII International Congress of Physiological Sciences (Tokyo), 1965,4, 629-636.
BARLOW, J. S., MORRELL, L.,AND MORRELL, F. Some observations on evoked responses in
relation to temporal conditioning to paired stimuli in man. Proceedings of Interna-
tional Colloquium on Mechanisms of Orienting Reactions in Man (Bratislava-Smolenice,
Czechoslovakia), 1967.
BARTLETT, F., AND JOHN, E. R. Equipotentiality quantified: The anatomical distribution of
the engram. Science, 1973, 181, 764-767.
BARTLETT, F., JOHN, E. R., SHIMOKOCHI AND KLEINMAN, D. Electrophysiological signs of
readout from memory. II. Computer classification of single evoked potential
waveshapes. Behavioral Biology, 1975, 14, 409-449.
BEGLEITER, H., AND PLA1Z, P. Modifications in evoked potentials by classical condition-
ing. Science, 1969, 166, 769-771.
BUCHWALD, J. S., HALAS, E. S., AND SCHRAMM, S. Progressive changes in efferent unit
responses to repeated cutaneous stimulation in spinal cats. Journal of Neurophysiol-
ogy, 1965,28, 200-215.
BURES, J. Discussion. In D. P. KIMBLE (Ed.), Anatomy of memory. Palo Alto: Science
Behavior Books, 1965, pp. 49-50.
BUREs,J., AND BURESOVA, O. Plasticity at the single neuron level. Proceedings of the XXIII
International Congress of Physiological Sciences (Tokyo), 1965,4, 359--364.
BURE~, J., AND BURESOVA, O. Plastic changes of unit activity based on reinforcing
properties of extracellular stimulation of single neurons. Journal of Neurophysiology,
BURE~, J., AND BURESOVA, O. Plasticity in single neurons and neural populations. In G.
Hom and R. A. Hinde (Eds.), Short-term changes in neural activity and behavior.
London and New York: Cambridge University Press, 1970, pp. 9-35.
CHESLER, P. Maternal influence in learning by observation in kittens. Science, 1969, 166,
CHOW, K. L., AND RANDALL, W. Learning and retention in cats with lesions in reticular
formation. Psychonomic Science, 1964, 1, 259--260.
CORNING, W. c., DYAL, J. A., AND WILLOWS, A. O. D. Invertebrate learning (Vol. 1). New
York: Plenum Press, 1973.
DRU, D., WALKER, J. P., AND WALKER, J. B. Self-produced locomotion restores visual
capacity after striate lesions. Science, 1975,187,265-266.
DUMENKO, V. N. The electrographic study of relationships between various cortical areas

in dogs during the elaboration of a conditioned reflex stereotype. In l. N. KNIPST

(Ed.), Contemporary problems of electrophysiology of the central nervous system. Mos-
cow: Academy of Science, 1967, pp. 104-112.
GALAMBOS, R, AND SHEAtz, G. C. An electroencephalograph study of classical condition-
ing. American Journal of Physiology, 1962, 203, 173-184.
GLlVENKO, E. V., KOROL'KOVA, T. A., AND KUZNETSOVA, G. D. Investigation of the spatial
correlation between the cortical potentials of the rabbit during formation of a condi-
tioned defensive reflex. Fizicheskii Zhurnal SSSR Sechenova, 1962, 48, 1026.
N. N. Caudate nucleus stimulation impairs the processes of perceptual integration.
Physiology and Behavior 1974, 12; 913-918.
HORI, Y., AND YOSHII, N. Conditioned change in discharge pattern for single neurons of
medial thalamic nuclei of cat. Psychological Report, 1965,16, 241.
JASPER, H. H., RICCI, G., AND DOANE, B. Microelectrode analysis of cortical cell discharge
during avoidance conditioning in the monkey. Electroencephalography and Clinical
Neurophysiology Supplement, 1960,13, 137-155.
JOHN, E. R Higher nervous functions: Brain functions and learning. Annual Review of
Physiology, 1961, 23, 451.
JOHN, E. R. Neural mechanisms of decision making. In W. S. FIELDS AND W. ABBOT (Eds.),
Information storage and neural contra!. Springfield: Thomas, 1963, pp. 243-282.
JOHN, E. R. Electrophysiological studies of conditioning. In G. C. QUARTON, T. MELNE-
CHUK, AND F. O. SCHMITT (Eds.), The neurosciences: A study program. New York:
Rockefeller University Press, 1967a, pp. 690-704.
JOHN, E. R. Mechanisms of memory. New York: Academic, 1967b.
JOHN, E. R. Brain mechanisms of memory. In J. McGAUGH (Ed.), Psychobiology. New
York: Academic, 1971, pp. 199-283.
JOHN, E. R. Switchboard versus statistical theories of learning and memory. Science, 1972,
177, 850-864.
JOHN, E. R. Cellular mechanisms in conditioning. Paper presented at International Union
of Physiological Sciences, New Delhi, 1974.
JOHN, E. R. BARTLETT, F., SHlMOKOCHI, M., AND KLEINMAN, D. Neural readout from
memory. Journal of Neurophysiology, 1973,36, 893-924.
JOHN, E. R., CHESLER, P., BARTLETT, F., AND VICTOR, l. Observation learning in cats.
Science, 1968, 159, 1489-1491.
JOHN, E. R., AND KILLAM, K. F. Electrophysiological correlates of avoidance conditioning
in the cat. Journal of Tharmacological and Experimental Therapeutics, 1959,125,252.
JOHN, E. R., AND KILLAM, K. F. Electrophysiological correlates of differential approach-
avoidance conditioning in the cat. Journal of Nervous and Mental Diseases, 1960,131,
JOHN, E. R., AND KLEINMAN, D. Stimulus generalization between differentiated visual,
auditory and central stimuli. Journal of Neurophysiology, 1975,38, 1015-1034.
JOHN, E. R., LEIMAN, A. L., AND SACHS, E. An exploration of the functional relationship
between electroencephalographic potentials and differential inhibition. Annals of
New York Academy of Sciences, 1961,92, 1160-1182.
JOHN, E. R, AND MORGADES, P. P. Neural correlates of conditioned responses studied
with multiple chronically implanted moving microelectrodes. Experimental Neurol-
ogy, 1969a, 23, 412-425.
JOHN, E. R., AND MORGADES, P. P. The pattern and anatomical distribution of evoked
potentials and multiple unit activity elicited by conditioned stimuli in trained cats.
Communications in Behavioral Biology, 1969b, 3, 181-207.
JOHN, E. R., & MORGADES, P. P. A technique for the chronic implantation of multiple
48 E. Roy JOHN

movable micro-electrodes. Electroencephalography and Clinical Neurophysiology,

1969c, 27, 205-208.
JOHN, E. R., RUCHKIN, D. S., LEIMAN, A., SACHS, E., AND AHN, H. Electrophysiological
studies of generalization using both peripheral and central conditioned stimuli.
Proceedings of the XXIII International Congress of Physiological Sciences (Tokyo), 1965,
JOHN, E. R., RUCHKIN, D. S., AND VILLEGAS, J. Signal analysis of evoked potentials
recorded from cats during conditioning. Science, 1963,141,429-431.
JOHN, E. R., RUCHKIN, D. S., AND VILLEGAS, J. Signal analysis and behavioral correlates of
evoked potential configuration in cats. Annals of New York Academy of Sciences, 1964,
112, 362-420.
JOHN, E. R., SHIMOKOCHI, M., AND BARTLElT, F. Neural readout from memory during
generalization. Science, 1969,164, 1534-1536.
JOHN, E, R., AND THATCHER, R. W. Integrative neuroscience. New Jersey: Lawrence
Erlbaum Associates, 1976, in press.
JOHNSTON, V. 1., AND CHESNEY, G. 1. Electrophysiological correlates of meaning. Science,
1974, 186, 944-946.
KAMIKAWA, K., MciLWAIN, J. T., AND ADEY, W. R. Response patterns of thalamic neurons
during classical conditioning. Electroencephalography and Clinical Neurophysiology,
1964, 17, 485-496.
KILLAM, K. R., AND HANCE, A. J. Analysis of electrographic correlates of conditional
responses to positive reinforcement: I. Correlates of acquisition and performance.
Proceedings of the XXIII International Congress of Physiological Sciences (Tokyo), 1965,
4, 1125.
KLEINMAN, D., AND JOHN, E. R. Contradiction of auditory and visual information by
brain stimulation. Science, 1975, 187, 271-272.
KLINKE, R., FRUHSTORFER, H., AND FINKENZELLER, P. Evoked responses as a function of
external and stored information. Electroencephalography and Clinical Neurophysiology,
1968, 26, 216-219.
KNIPST, I. N. (Ed.). Spatial synchronization of bioelectrical activity in the cortex and some
subcortical structures in rabbit's brain during conditioning. In Contemporary prob-
lems of electrophysiology of the central nervous system. Moscow: Academy of Science,
1967, pp. 127-137.
KOROL'KOVA, T. A., AND SHVETS, T. B. Interrelation between distant synchronization and
steady potential shifts in the cerebral cortex. In I. N. KNIPST (Ed.), Contemporary
problems of electrophysiology of the central nervous system. Moscow: Academy of
Science, 1967, pp. 16~167.
LEIMAN, A. 1. Electrophysiological studies of conditioned responses established to central
electrical stimulation. Doctoral thesis, University of Rochester, Rochester, New York,
LEIMAN, A. 1., AND CRISTIAN, C. N. Electrophysiological analyses of learning and
memory. In. J. A. DEUTSCH (Ed.), The physiological basis of memory. New York:
Academic Press, 1973, pp. 125-173.
LIBERSON, W. T., AND ELLEN, P. Conditioning of the driven brain wave rhythm in the
cortex and the hippocampus of the rat. In J. WORTIS (Ed.), Recent advances in biological
psychiatry. (Vol. 2). New York: GRUNE & STRATTON, 1960.
behavior in the cat: I. Pattern discrimination and its alteration by atropine and LSD-
25. Electroencephalography and Clinical Neu~ophysiology, 1968,24, 497-513.
LIVANOV, M. N. Spatial analysis of the bioelectric activity of the brain. Proceedings of the
XXII International Congress of Physiological Sciences (Leiden), 1962, 899-907.

LIVANOV, M. N. The significance of the distant brain potential synchronization for

realization of temporal connections. Proceedings of the XXIII International Congress of
Physiological Sciences (Tokyo), 1965,4, 600-612.
LIVANOV, M. N., AND POLIAKOV, K. L. The electrical reactions of the cerebral cortex of a
rabbit during the formation of a conditioned defense reflex by means of rhythmic
stimulation. Izvestiya Akademiya Nauk. USSR Series Biology, 1945,3,286.
MAJKOWSKI, J. Electrophysiological studies of learning in split-brain cats. Presented at
Electroencephalography Meeting (California), 1966.
MAJKOWSKI, J. Electrophysiological studies of learning in split-brain cats. Electroencephal-
ography and Clinical Neurophysiology, 1967,23, 521-53l.
MORGAN, C. T., AND STELLAR, E. Physiological psychology. New York: McGraw-Hill,
MORRELL, F. Effect of anodal polarization on the firing pattern of single cortical cells.
Annals of New York Academy of Science, 1961a, 92, 86~76.
MORRELL, F. Electrophysiological contributions to the neural basis of learning. Physiologi-
cal Review, 1961b, 41, 443.
MORRELL, F. Electrical signs of sensory coding. In G. C. QUARTON, T. MELNECHUK, AND F.
O. SCHMITT (Eds.), The neurosciences: A study program. New York: Rockefeller Uni-
versity Press, 1967, pp. 452-469.
MORUZZI, G., AND MAGOUN, H. W. Brain stem reticular formation and activation of the
EEG. Electroencephalography and Clinical Neurophysiology, 1949,1, 455-473.
O'BRIEN, J. H., AND Fox, S. S. Single-cell activity in cat motor cortex: I. Modifications
during classical conditioning procedures. Journal of Neurophysiology, 1969a, 32, 267-
O'BRIEN, J. H., AND Fox, S. S. Single-cell activity in cat motor cortex: II. Functional
characteristics of the cell related to conditioning changes. Journal of Neurophysiology,
1969b,32, 285-296.
OLDS, J., DISTERHOFf, J. F., SEGAL, M., KORNBLITH, c., AND HIRSCH, R. Learning centers
of the rat brain mapped by measuring latencies of conditioned unit response. Journal
of Neurophysiology, 1972,35, 202-219.
OLDS, J., AND HIRANO, T. Conditioned responses of hippocampal and other neurons.
Electroencephalography and Clinical Neurophysiology, 1969,26, 159-166.
OLOS, J., AND OLOS, M. E. Interference and learning in paleocortical systems. In J. F.
ium on brain mechanisms and learning. Oxford: Blackwell, 1961.
PICTON, T. W., AND HILLYARD, S. A. Human auditory evoked potentials: II. Effects of
attention. Electroencephalography and Clinical Neurophysiology, 1974,36, 191-199.
PICTON, T. W., HILLYARD, S. A., AND GALAMBOS, R. Cortical evoked responses to
omitted stimuli. In M. N. Livanov (Ed.), Major problems of brain electrophysiology.
Moscow: U.S.S.R. Academy of Sciences, 1973.
RAMOS, A., AND SCHWARTZ, E. Observation of assimilation of the rhythm at the unit
level in behaving cats. Physiology and Behavior, 1976a, in press.
RAMOS, A., AND SCHWARTZ, E. Observation of frequency specific discharges at the unit
level in conditioned cats. Physiology and Behavior, 1976b, in press.
RAMOS, A., SCHWARTZ, E., AND JOHN, E. R. Unit activity and evoked potentials during
readout from memory. Presented at XXVI International Congress of Physiological
Sciences (New Delhi), 1974.
RAMOS, A., SCHWARTZ, E., AND JOHN, E. R. Evoked potential-unit relationships in
behaving cats. Brain Research Bulletin, 1976a, in press.
RAMOS, A., SCHWARTZ, E., AND JOHN, E. R. Cluster analysis of evoked potentials from
behaving cats. Behavioral Biology, 1976b, in press.

RAMOS, A., SCHWAR1Z, E., AND JOHN, E. R. An examination of the participation of

neurons in readout from memory. Brain Research Bulletin, 1976c, in press.
RAMOS, A., SCHWAR1Z, E., AND JOHN, E. R. Differential neuronal responses during
conditioning in cats. Science, 1976d, in press.
RIGGS,1. A., AND WHlTILE, P. Human occipital and retinal potentials evoked by subjec-
tively faded visual stimuli. Vision Research, 1967, 7, 441-451.
RUCHKIN, D. S., AND JOHN, E. R. Evoked potential correlates of generalization. Science,
1966, 153, 209-211.
RUSlNov, V. S. Electroencephalographic studies in conditioned reflex formation in man.
In M. A. B. BRAZIER (Ed.), The central nervous system and behavior, New York: JOSIAH
MAClIl, JR. Foundation, 1959.
SAKHUILINA, G. T., AND MERZHANOVA, G. K. Stable changes in the pattern of the
recruiting response associated with a well-established conditioned reflex. Electroen-
cephalography and Clinical Neurophysiology, 1966,20, 50-58.
SCHWAR1Z, E., RAMOS, A., AND JOHN, E. R. Single cell activity in chronic unit recording: A
quantitative study of the unit amplitude spectrum, Brain Research Bulletin, 1976, in
SUTTON, S., TUETING, P., ZUUIN, J., AND JOHN, E. R. Information delivery and the sensory
evoked potential. Science, 1967, 155, 1436-1439.
THOMPSON, R. F., PATTERSON, M. M., ANDTEYLER, T. J. The neurophysiology of learning.
Annual Review of Psychology, 1972,23, 73-104.
TRAVIS, R. P., JR., AND SPARKS, D. L. Unitary responses and discrimination learning in
the squirrel monkey. Physiological Behavior, 1968,3, 187-196.
TRAVIS, R. P., JR., AND SPARKS, D. L. Changes in unit activity during stimuli associated
with food and shock reinforcement. Physiology and Behavior, 1967,2,171-177.
TRAVIS, R. P., JR., HOOTEN, T. F., AND SPARKS, D. 1. Single unit activity related to
behavior motivated by food reward. Physiology and Behavior, 1968,3, 309-318.
TRAVIS, R. P., JR., SPARKS, D. 1., AND HOOTEN, T. F. Single unit response related to
sequences of food motivated behavior. Brain Research, 1968,7, 455-458.
WEINBERG, H., WALTER, W. G., COOPER, R., AND ALDRIDGE, V. J. Emitted cerebral events.
Electroencephalography and Clinical Neurophysiology, 1974,36,449-456.
WEINBERG, H., WALTER, W. G., AND CROW, H. H. Intracerebral events in humans related
to real and imaginary stimuli. Electroencephalography and Clinical Neurophysiology,
1970,29, 1-9.
WOODY, C. D., VASSILEVSKY, N. N., AND ENGEL, Conditioned eye blink-unit: Activity at
coronal precruciate of cat. Journal of Neurophysiology, 1970,33, 838.
WUNDT, W. Principles of physiological psychology. (Originally published in 1901.) New
York: MACMILLAN, 1910.
YOSHII, N. Electroencephalographic study on experimental neurosis, a conditioned partly
awake state. Proceedings of the XXII International Congress of Physiological Sciences
(Leiden), 1962, 1088.
YOSHII, N., AND OGUBA, H. Studies on the unit discharge of brain stem reticular
formation in the cat: I. Changes of reticular unit discharges following conditioning
procedure. Medical Journal of Osaka University, 1960,11, l.
YOSHII, N., PRUVOT, P., AND GASTAUT, H. Electroencephalographic activity of the mesen-
cephalic reticular formation during conditioning in the cat. Electroencephalography
and Clinical Neurophysiology. 1957,9, 595.
2 Self-Consciousness and
A Model Based on an Experimental Analysis of the
Brain Mechanisms Involved in the Jamesian Theory of
Motivation and Emotion




A. Introduction

The recent revolution in psychology has readmitted cognition and

consciousness as legitimate areas of scientific investigation. The study
of cognitive processes has made rapid strides by taking as its model
brain mechanisms assumed to be similar to those of the digital com-
puter (Miller, Galanter, and Pribram, 1960) and by utilizing reaction-
time data investigations of memory for verbally coded materials. The
currently projected volumes on consciousness and self-regulation pre-
suppose that equally effective strides can be made in our research on,
and understanding of, consciousness. The title of the series, in fact,
suggests that data on self-regulation, utilizing biofeedback procedures,
will provide the substance upon which such strides will be based.
But if understanding comparable to that attained for cognition is to
be achieved, an experimentally based model of the brain processes
operative in consciousness must also be made available. The purpose of
this paper is to provide the outlines of such a model.
KARL H. PRIBRAM . Department of Psychology, Stanford University, Stanford,


There are many meanings attributed to the term consciousness.

Webster's dictionary covers a dozen. There have been articles written
on the consciousness of cells; Eastern mystics speak of the conscious-
ness of stones; Jungians deal with universal human consciousness (see
Ornstein, 1972, 1973 for review). I have, in another manuscript (Pri-
bram, 1976b), taken a somewhat more direct and perhaps practical
approach to definition. These largely definitional issues need not con-
cern us here since self-consciousness can be clearly distinguished from
other forms of consciousness.
Self-consciousness is said to occur when an observer is able to
describe both the observed and the observing. Philosophers (Husserl,
1928) have called this ability intentionality-thus the subtitle of the
present manuscript. The term is derived from intention, an aim of an
action which mayor may not be realized. The separateness of intent
and outcome of an action was generalized by Brentano (1925, 1960) to
the objects of perceptual "acts." This generalization has proved to be
prescient. Recent evidence from brain research (see Pribram, 1971,
1974, 1976a) has shown that the same parts of the brain (the basal
ganglia) that control motor function also control sensory input. These
controls operate by changing the set point of receptors (see below) in
muscles or sense organs and are therefore ideally suited to function as
in ten tionali ty mechanisms.
Thus the outcomes of actions and the objects of perception come to
form one universe-the realized universe of existence-while the inten-
tional universe is dispositional and may even be unrealizable (the
awkward term intentional inexistence was meant to convey this nonreal-
ity). The difficulty with such terminolgy is, of course, that other philos-
ophers can counter with the proposition that the phenomenal experi-
ence of dispositions exists just as much as the outcomes of actions and
the objects of experience and that, to some, these dispositions are the
existential reality. For us here, the distinction, not the argument, is the
important concern: In studies, of self-regulation both disposition and
outcome are realized. After all, the instrumentation that allows the
externalization (objectively demonstrating the separate existence) of
dispositions is the innovation that makes the scientific study of self-
consciousness now possible.
Behavioral psychologists have ordinarily deSignated dispositions
by the terms emotion and motivation. The intentionality of motivation is
relatively obvious, although Miller, Galanter, and Pribram (1960) dis-

tinguish between motive and intent as follows:

Jones hires Smith to kill someone. Smith commits the murder, but he is
caught and confesses that he was hired to do it. Question: Is Smith Guilty? If
we consider only the motives involved, the employer is guilty because he
was motivated to kill, but the gunman is not guilty because his motive was
merely to earn money (which is certainly a commendable motive in a
capitalistic society). But if we consider their intentions, then both parties are
equally guilty, for both of them knowingly undertook to execute a Plan
culminating in murder. (p. 61)

Motive in ordinary and legal language apparently refers more to the

feelings involved, while intent refers to the aims of actions. A similar
distinction can be made in the case of emotions: The feelings of emo-
tional elation or upset can be separated from their "aim" or target-
e.g., accomplishing rapport with someone whom one is in love with.
The fact that motivational and emotional feelings (dispositions) can be
distinguished from their referents makes intentionality possible. The
purpose of studies of self-consciousness is to enhance intentionality by
providing independent controls (self-regulation) on motivational and
emotional dispositions. Stated in this fashion, it becomes clear that the
terms intentionality and volition have a good deal in common. In
popular parlance, of course, to "intend" something is to "will" it. The
issue of self-consciousness is therefore also the issue of voluntary
control, and any proposed brain model must take this into account.
Interestingly, William James (1950) dealt with these related issues
in a most sophisticated manner. I want here, therefore, to develop and
evaluate by both positive and negative comment the Jamesian model,
critically but not polemically. Rather the presentation will review a
series of clinical observations and laboratory experiments specifically
designed to test aspects of the model with the aim of providing a
modification based on currently available data.

B. Case History

The observations and experiments were begun within the frame-

work of a James-Lange view of the problem, a view that William James
(1950) proposed as follows: "Bodily changes follow directly the percep-
tion of the exciting fact and-our feeling of the same changes as they
occur is the emotion" (Vol. II, p. 449). As did most investigators at the
time, and perhaps even now, I took this to mean that emotional feelings

result when visceroautonomic mechanisms become activated. This as-

pect of the theory is attributed by William James to Carl Lange, who
had suggested that emotional feelings were due to changes in vascular-
ity and other visceral processes. True, the work of Walter Cannon (1927)
had made it mandatory to replace peripheral with central mechanisms,
but these were still thought of in terms of visceroautonomic processes.
It was, after all, the "head ganglion" of the autonomic nervous system
that concerned Cannon.
My entry into the problem was due to a patient, a Greek woman in
her early fifties, who suffered from Jacksonian epileptic seizures always
initiated and almost always limited to the left part of her face. Charac-
teristically, even before any muscular twitching, there would be a
profuse outpouring of perspiration sharply restricted to the left side of
the face and neck as if by a line drawn to separate the two sides. To
make a long story short, Paul Bucy and I (Bucy and Pribram, 1943)
diagnosed a brain tumor and found and removed a circumscribed
oliogodendroglioma located in the right precentral motor cortex. The
patient recovered completely with no residual paralysis and with elimi-
nation of the seizures.
The localized sweating shown by this woman was caused by
irritation of the precentral motor cortex and thus called into question
the idea then held that it was the hypothalamus which was the "head
ganglion" of the autonomic nervous system. Obviously, cortical mecha-
nisms played some role in the regulation of visceroautonomic activity.

C. A Mediobasal Motor System

After publication of this patient's story, it became clear to me that

visceroautonomic auras were not altogether rare in epileptic patients.
However, the great majority of such auras could be referred to pathol-
ogy in and around the Island of Reil and the pole of the temporal lobe. I
therefore began a program of research to map the cortical sites in
nonhuman primates from which visceroautonomic responses could be
obtained by electrical stimulations. The initial experiments were per-
formed at the Yerkes laboratory with a Harvard inductorium and pro-
duced equivocal and unreliable results. I heard, however, that at Yale a
new method of cortical stimulation had been developed-a thyrotron
stimulator which put out square waves instead of sine waves-and that

pulse duration as well as voltage and frequency could be controlled.

With this stimulator, Arthur Ward had been able to produce visceroau-
tonomic effects from excitation of the anterior part of the cingulate
gyrus (Ward, 1948), and Robert Livingston had succeeded in showing
similar effects from the posterior orbital surface of the frontal lobe
(Livingston, Fulton, Delgado, Sachs, Brendler, and Davis, 1948). Be-
cause of the connections of these portions of the frontal cortex via the
uncinate fasciculus, which had been demonstrated not only anatomi-
cally but with strychnine neuronography by McCulloch, Bailey, and
von Bonin (Bailey, von Bonin, and McCulloch, 1950)-experiments I
had had an opportunity to observe-I decided to go to Yale to extend
the stimulation experiments to the temporal pole. There I found Birger
Kaada, who had just begun his thesis with the aim of analyzing not
only the visceroautonomic but also the "suppressor" effects of cingu-
late gyrus stimulation. Working in adjacent laboratories, obtaining
identical effects from stimulation of the temporal pole and the cingu-
late cortex, late one night we joined forces and mapped the entire area
from above the corpus callosum to below it, and by turning the
monkey on his back and letting the frontal lobe hang away from the
base of the skull, we traced the entire region effective in producing
visceroautonomic responses: cingulate, subcallosal, medial and poste-
rior orbital, anterior insular, periamygdaloid, and temporal polar
cortex. This was made especially easy once the Sylvian fissure was
opened by gentle retraction and temporarily packed with cotonoid
patties. In short, we mapped (Kaada, Pribram, and Epstein, 1949;
Kaada, 1951) a continuous region of cortex lying on the edge, the
limbus of the anterior portion of the cerebral hemisphere, which,
when stimulated, produced respiratory arrest, a drop in blood pres-
sure, changes in heart rate, eye movements, turning of the head, and
under proper circumstances, suppression (or occasionally enhance-
ment) of spinal reflexes. We had mapped a mediobasal motor cortex.
What then of the visceroautonomic seizures in the patient with the
precentral oliogodendroglioma? In another series of experiments Pat-
rick Wall and I (Wall and Pribram, 1950) mapped the lateral surface of
the cortex and, again to make a long story short, found that such
responses could be obtained from the classical precentral motor cortex.
Despite a whole series of attempts, we were unable to specify the
difference between these responses and those obtained from the me-
diobasal cortex.

FIGURE 1. The medial-basal motor cortex. Black dots indicate areas

for which electrical stimulation produces changes in blood pressure,
heart rate, respiratory rate, eye movements, and gross bodily move-
ments. (A) Lateral surface: (B) Medial-basal surface. One accom-
plishes this view by making a slit in the lateral part of the
hemisphere and bringing the basal surface in line with the medial

It should not have been altogether surprising that visceroauton-

omic responses are obtained from stimulations that also produce soma-
tomotor responses. Even stimulation of the hypothalamus, the head
ganglion of the autonomic nervous system, produces somatomotor as

FIGURE 2. Points of stimulation in the somatosensory motor cortex of the lateral extent
of the hemisphere giving rise to changes in blood pressure, heart rate, respiratory rate,
and discrete movement.

well as visceral responses. In fact, William James (1950) had stated the
issue clearly:
If the neural process underlying emotional consciousness be what I
have now sought to prove it, the physiology of the brain becomes a simpler
matter than has been hitherto supposed. Sensational, associational, and
motor elements are all that the organ need contain. The physiologists who,
during the past few years, have been so industriously exploring the brain's
functions, have limited their explanations to its cognitive and volitional
performances. Dividing the brain into sensory and motor centres, they have
found their division to be exactly paralleled by the analysis made by
empirical psychology of the perceptive and volitional parts of the mind into
their simplest elements. But the emotions have been so ignored in all these
researches that one is tempted to suppose that if these investigators were
asked for a theory of them in brain-terms, they would have to reply, either
that they had as yet bestowed no thought upon the subject, or that they had
found it so difficult to make distinct hypotheses that the matter lay among
the problems of the future, only to be taken up after the simpler ones of the
present should have been definitely solved.

And yet it is even now certain that of two things concerning the
emotions, one must be true. Either separate and special centres, affected to
them alone, are their brainseat, or else they correspond to processes occur-
ring in the motor and sensory centres already assigned, or in others like
them, not yet known. If the former be the case, we must deny the view that
is current, and hold the cortex to be something more than the surface of
"projection" for every sensitive spot and every muscle in the body. If the
latter be the case, we must ask whether the emotional process in the sensory
or motor centre be an altogether peculiar one, or whether it resembles the
ordinary perceptive processes of which those centres are already recognized
to be the seat. Now if the theory I have defended be true, the latter
alternative is all that it demands. Supposing the cortex to contain parts,
liable to be excited by changes in each special sense-organ, in each portion
of the skin, in each muscle, each joint, and each viscus, and to contain
absolutely nothing else, we still have a scheme capable of representing the
process of the emotions. An object falls on a sense-organ, affects a cortical
part, and is perceived; or else the latter, excited inwardly, gives rise to an
idea of the same object. Quick as a flash, the reflex currents pass down
through their preordained channels, alter the condition of muscle, skin, and
viscus; and these alterations, perceived, like the original object, in as many
portions of the cortex, combine with it in consciousness and transform it
from an object-simply-apprehended into an object-emotionally-felt. No new
principles have to be invoked, nothing postulated beyond the ordinary
reflex circuits, and the local centres admitted in one shape or another by all
to exist. (Vo!. II, pp. 472-474)

Note that James emphasizes the sensory aspects of these "reflex

currents." We shall return to this point presently. But at the time of the
discovery of the mediobasal motor mechanism I was surprised and, in a
way, a little disappointed-we had not been able to confirm our hy-
pothesis that some part of the cerebral mantle dealt exclusively with
visceral mechanisms and could thus be thought of as a "visceral
brain"-a substrate for a Langian conception of "emotion." I might add
that everyone did not share this disappointment-Paul MacLean, my
office mate and collaborator in experiments mapping by electrical stim-
ulation and strychnine neuronography the organization of mediobasal
cortex (Pribram, Lennox, and Dunsmore, 1950; Pribram and MacLean,
1953; MacLean and Pribram, 1953), was more convinced by our reports
of visceroautonomic regulations by mediobasal cortex than by their
invariable concomitance with somatomotor effects (MacLean, 1949).
But for me the disappointment was real and led to puzzlement as to just
what could be the meaning of this juxtaposition of visceroautonomic
and somatomotor mechanisms to the brain's role in emotion and moti-
vation. I therefore turned to other techniques to help resolve these

D. The Limbic Systems and Behavior

First, it was necessary to obtain evidence that the limbic medio-

basal motor mechanisms are in fact critically involved in motivational
and emotional processes. To rephrase the question in experimentally
testable terms, evidence had to be obtained to show that behavior
ordinarily considered to he representative of motivational and emo-
tional processes is disrupted by resections or excitations of the limbic
mediobasal mechanism. As it turned out, the results of the experi-
ments undertaken took us a long way into reformulating the problem
of what constitutes such behavior.
A series of studies designed to analyze the syndrome described by
Heinrich Kluver and Paul Bucy (KlUver and Buey, 1937) to follow total
temporal lobectomy provided the evidence. Kluver and Buey had in-
cluded taming, increased oral and sexual behavior, and visual agnosia

DAve 1
Self-Assured. Feared
~.::::"",,, RIVA3
A9\lfturve. Ac:1rve


PlilCid, un,aggrf$$ive

Submiui"'e, Cowering, SHORTY 7
Frequenlly Atl,ac:ktd ARNIE 6

~~ Submfutve to Others, Aggreuive

Toward L.etry
AJert,. Active food Getter

FIGURE 3(A) Dominance hierarchy of a colony of eight preadolescent male rhesus

monkeys before any surgical intervention.

ZeKE 1
Oarinv. Competes With Zeke


AnKk1. O;we
Oomln.alIH .. nd

1 "-

Comploto'v s..bm;"'v F.. "u'


FIGURE 3(B) Same as (A) after bilateral amygdalectomy had been

performed on Dave. Note his drop to the bottom of the hierarchy.

in their syndrome. Our studies (Blum, Chow, and Pribram, 1950;

Chow, 1951; Mishkin and Pribram, 1954; and Pribram, 1954) showed
the agnosia to be due to resection of the lateral portions of the temporal
lobe; however, these results make up a body of evidence which, though
related to the issues being examined here, constitute a sufficiently
separate domain to warrant presentation on another occasion (see, for
example, Pribram, 1969, 1975) . The remaining part of the syndrome
was obtained full-blown when lesions were restricted to the anterior
limbic portions of the lobe-those comprising the temporal lobe
portions of the mediobasal motor mechanism (Pribram and Bagshaw,
1953). Subsequent studies showed the entire mediobasal motor system
to be involved (Pribram and Weiskrantz, 1957).
Specifically, tests were performed to measure feeding, fleeing
(avoidance), fighting (dominance), mating, and maternal (nesting) be-
havior (see reviews by Pribram, 1958, 1960). The fairly gross changes in

~ ~

Oomm.ant, Not Th, .. tantd

<, .." JNI
by Othtfs




SubmluiwlII 10 Others,
Intarminenlly Aljlgu!'UI'I.
TOWlrd 0.."".

oo -

DAve 8
Ctingtor. Avoids Int.rlKlion

FIGURE 3(C) Same as (A) and (B) except that both Dave and Zeke have
received bilateral amygdalectomies.

these behaviors following lesions of the limbic motor systems are so

well known by now that I want here to present more quantitative data.
The effects of these lesions on fighting were observed in social
situations (Rosvold, Mirsky, and Pribram, 1954). A group of eight
preadolescent male monkeys were housed together until a more or less
straight-line dominance hierarchy became stably established. Domi-
nance was initially rated on the basis of order of obtaining food pellets
inserted into the colony space, one by one through a metal tube. The
dominance rating obtained in this way was then checked against quan-
titative observations of threatening gestures, actual fighting contacts,
grooming contacts, and position displacements. Such observations
were made not only in the colony when the group was together as a
whole but also for all possible dyads. Then the clearly dominant mon-
key was submitted to psychosurgery. As expected, he fell to the bottom
of the hierarchy. Interestingly, however, this drop was effected over a

~ . . . . ,I!' More Oomin.nt.

Unpri!dictablV Aw ressiye
and Vicioul


Contin .... n In,ermi11enlJy Aggressh,"
Tow.rd O.....e

Out<asc FI ... I'om All 717'~
~ ~

FIGURE 3(0) Final social hierarchy after Dave, Zeke and Riva have all had
bilateral amygdalectomies. Note that Riva fails to fall in the hierarchy. Minimal
differences in extent of locus of the resections do not correlate with differences in
the behavioral results. The disparity has been shown in subsequent experiments
to be due to Herby's nonaggressive "personality" in the second position of the

48-hour period during which interactions with the other monkeys

gradually lowered the "status" of the previously dominant monkey.
This experience was essentially replicated when we performed
surgery on the formerly Number 2 and now dominant animal. When,
however, we attempted to repeat the procedure for a third time, the
expected effects did not occur. In fact, the original Number 3 animal
became, if anything, more aggressive and dominant. My colleagues in
the study, Hal Rosvold and Alan Mirsky, of course, blamed inadequate
surgery for this development, but histological verification failed to
confirm their suspicions. In fact, the lesion of this last monkey extended
further than that of one of the others, and all lesions encompassed the
same anatomical structures.
What, then, could account for our results? Briefly, examination of
our data, especially of the 48 initial postoperative hours in the colony,

and observations of other colonies of monkeys made us believe that lack

of interaction between the operated subject and the original Number 4
monkey was responsible for the original Number 3 monkey's failure to
fall in dominance. Many subsequent observations in dyadic situations
confirmed this belief: Postoperative monkeys were especially sensitive
to the way they were treated by their cage mates and handled by their
caretakers. The immediate postoperative taming could be prolonged for
years by gentling procedures, whereas ordinary neglect and occasional
rougher treatment would produce either an excessively fearful or an
unpredictably aggressive monkey. These results make it unlikely that
some fundamental mechanism responsible for aggression had been
excised; rather some brain process sensitive to the social environment
seems to have been tapped.



0-10 5

FIGURE 4(A) Graph of the percentage of time spent by various groups of animals in the
dark (previously shocked) compartment of a shuttle box during postoperative extinction
of a preoperatively acquired conditioned avoidance. The scores for the first extinction
day are recorded in lO-trial blocks; subsequent extinction trials are plotted in 50-trial
blocks (one test day). Vertical bars indicate range of performance.


~ 50
~ 40
f= 20



FIGURE 4(B) Graph of the percentage of time spent by the various groups of animals in
the dark (previously shocked) compartment of a shuttle box during postoperative
reextinction of postoperatively reacquired conditioned avoidance. Trial blocks divided
as in 4(A). Bars indicate range of performance.

Similar results were obtained when fleeing, (avoidance) was tested

by Weiskrantz and myself (Pribram and Weiskrantz, 1957) in a classical
shuttle box. Escape proved unaffected by limbic lesions, but learning,
extinction, and relearning of conditional avoidance behavior were af-
fected. It should be noted in passing that limbic and not lateral fore-
brain lesions (except for frontal in the case of extinction) produced such
results; however, neither classical sensory and motor resections nor
basal ganglion removals were included in these studies.
Perhaps the clearest indication of what type of regulation is accom-
plished by the mediobasal motor mechanism came from our studies of
feeding. Postoperatively, monkeys with lesions in these limbic areas
often failed to eat for a time (usually not more than a week). Once they
recovered, however, they might eat twice as much per day as preopera-

tively, and this increased food intake usually lasted for months (Pri-
bram and Bagshaw, 1953).
Further analysis almost immediately uncovered a paradox. Despite
this marked increase in feeding in an ad libitum situation, Schwartz-
baum and I (Schwartzbaum, 1961) found that the monkeys with lesions
were less sensitive to food deprivation or satiation when made to work
for their food. They appeared "hungrier" when food was available but
less "hungry" when they were deprived and food could be obtained
only by the pressing of a lever to obtain pellets (on a variable interval
schedule). But this was not all; the loss of sensitivity was not limited to
variations in the internal state produced by the deprivation but ex-
tended as well to variations in the external characteristics of the food,



II:: 130
- Normals 175-480
~ 00 Amygdalas 274-375
FIGURE 5(A) Effect of food dep- gloo
rivation on number of responses 2 3 4 5
emitted by normal and bilater- SESSION
ally amygdalectomized monkeys
in a fixed-interval operant-con-
ditioning situation. Note that ~ B LARGE REWARD
the percentage change in total z
~ 140
~ ~o
number of responses is plotted u
on the graph. Absolute values ~ ..
are indicated in the right lower o 120 o 0
comer. 5(B) The effects of the ~
changing of reward size on bi- II:: 100
laterally amygdalectomized and
~ -Normals
normal monkeys on response ~ 80 o 0 Amygdalas
rate in a fixed-interval situation. ffl
Note that normal animals satiate II::

rapidly in a session when the 2 3 4 5 6

pellet size is increased. 10- MINUTE INTERVALS

such as the size of the pellets received as a consequence of lever

pressing (Schwartzbaum, 1960a, 1960b).
These results paralleled those that were being reported from analy-
ses of feeding disturbances produced by ventromedial and far-lateral
hypothalamic lesions. Miller, Bailey, and Stevenson (1950) found that
rats who became obese in ad libitum feeding situations might starve if
required to make the effort to cross a barrier to obtain their food. And
Teitelbaum (1955) showed that rats who would starve in ad libitum
situations could be induced to attend and eat if the sensory attractive-
ness of the food was sufficiently enhanced.
Two major hypotheses derive from these observations, one with
respect to the sensory, the other with respect to the motor processes
regulated by the limbic (and hypothalamic) mechanisms:
1. Attention (Le., reaction) to external as well as to internal
stimulation is involved in motivational and emotional feeling.
2. Effort, not drive (e.g., as defined by ad libitum feeding and
lever pressing), is the critical variable determining motiva-
tional and emotional expression in behavioral responses.
Let us examine the evidence related to each of these hypotheses in



A. Transfer of Training

In order to bring the altered reactions to external stimulation of the

lesioned monkeys into sharp focus, I decided to test them in a series of
tasks which were as minimally related to motivation and emotion as I
could find and yet might provide some indication of function. The
point of departure for selecting these tasks was the dramatic change in
dominance displayed in the social colony experiment. The lesioned
monkeys behaved postoperatively as if they had never experienced the
colony structure-they seemed to have to learn anew the repertoire of
aggressive interactions that established their place in the hierarchy.
They appeared not to transfer their prior experience to the postopera-
tive situation.

Thus a series of transfer-of-training tasks was devised. The first,

undertaken with Schwartzbaum (Schwartzbaum and Pribram, 1960),
used a transposition paradigm in which the monkeys were initially
trained to choose the lighter of two gray panels and were then tested on
a pair of panels of which the formerly lighter one was now the darker of
the test pair. All control subjects continued to choose the lighter panel,
but the lesioned animals behaved oddly. They hesitated and then chose
randomly between the two panels. It appeared as if they perceived the
test situation to be novel and proceeded accordingly.
The results of a second experiment performed with Muriel Bag-
shaw (Bagshaw and Pribram, 1965) supported the findings of the first.
In this experiment Heinrich Khiver's equivalence test (Kluver, 1933)
was used. The monkeys were trained to choose the larger of two
moderately sized squares and tested on a pair of smaller ones. Again
the control subjects chose the larger panel throughout while the
lesioned monkeys behaved as if the test panels presented them with
an entirely novel situation.
Both the transposition and the equivalence results could be attrib-
uted to an altered gradient of generalization of input following the
lesion. Eliot Hearst and I (Hearst and Pribram, 1964a, 1964b) put this
possibility to test and found generalization unimpaired. In neither
positive nor negative reinforcing situations were the monkeys' general-
ization gradients changed by the lesion. Thus transfer of training
appears to be dissociable from sensory generalization-and perhaps
might be more appropriately thought of in terms of "motor generaliza-
tion"-a view consonant with the fact that lesions of the mediobasal
motor cortex were the responsible agent in producing the changes.

Number of Transposed Responses Made on Transposition Tests

Normals Amygdalectomized

439 441 443 447 Mdn. 397 405 438 442 Mdn.

1 6 5 6 6 2 5 2 4
2 5 5 5 6 3 6 2 2
Total 11 10 11 12 11.0 5 11 4 6 5.5

The effects of amygdalectomy on transfer of training to a new but related task. Note that
the amygdalectomized monkeys treat the task as completely novel. whereas their
normal controls transpose their responses on the basis of their earlier experience.

B. Psychophysiological Experiments
The observations on response to novelty made in these neurobe-
havioral experiments were considerably enhanced by some obtained in
psychophysiological studies. One possibility, consonant with the Jame-
sian hypothesis, would be that motivation involved the somatomotor
system while emotion invoked visceroautonomic processes. Thus, de-
spite the juxtaposition of their central mechanism, peripheral differ-
ences could account for the behavioral distinction. This possibility was
put to direct test in a series of experiments which assayed the effects on
visceroautonomic indicators of resections rather than stimulations of
portions of the mediobasal, limbic motor mechanism. Such experi-
ments allowed observations to be made under physiologically normal
conditions for many months and even years in a variety of environmen-
tal circumstances that ordinarily produce visceroautonomic reactions.
The question was asked as to which of these circumstances produced
altered reactions or absence of reaction in the lesioned monkeys.
To summarize a decade of experiments, Muriel Bagshaw, Daniel
Kimble, and I (Bagshaw, Kimble, and Pribram, 1965; Kimble, Bagshaw,
and Pribram, 1965) found that the forebrain lesions had, as might be
expected, no effect on peripheral, reflexly produced visceroautonomic
reactions. Galvanic skin responses (GSR), heart, and respiratory
changes occurred in normal amount and frequency when the monkeys
moved or when gentle electric shock was applied to the soles of their
feet. We found (Bagshaw and J. Pribram, 1968), if anything, that the
threshold for obtaining such reactions was lower than in normal sub-
jects. By contrast, however, when response to novel stimulation or to
conditioning was tested, visceroautonomic reactivity was grossly defi-
cient. The visceroautonomic components of orienting and conditioning
were markedly attenuated or completely eliminated by the lesions
(Bagshaw and Benzies, 1968).
Analysis showed that the deficit in conditioning was to some
considerable extent due to a restriction in anticipatory reactions to the
unconditional stimulus which occurred in control subjects (Bagshaw
and Coppock, 1968). Thus the situations in which the deficit was
manifest were those that demanded reactions to recurring events, not
reactions to the events themselves. The fact that limbic forebrain struc-
tures, the mediobasal motor mechanisms, are critically involved in the
reaction to novelty as well as in motivational and emotional reactions

- -..... N
80 x, 'X H
00 IT
x ..

\~ ..........

\,\ ........

\, \ ..",...

.~..::>\:;:::.... ~ "
u \\
...... - -
'!- ::.~: Ji............ .:!.

"..., ,., ...
\,. ............... 0


o 5
FIGURE 6. Curves for overall percentage of GSR responses to the first 50
presentations of an irregularly presented 2-sec tone. Amygdalectomized
(AM), hippocampectomized (H), control (IT), and unoperated (N) groups.

suggests a major modification of Jamesian theory: Emotional and motiva-

tional feeling comes about not by any direct bodily reaction to perceived
events, but by a change in the sets, expectations, and anticipations produced
by such events.

C. Habituation
The simplest expression of such sets or expectations is habituation.
And, in fact, habituation of a locomotor response in a repetitive situa-




;. v/- - - - . \.
::: 60 / . . ....~
............ \
.... '..

&.20. ~
,:' '.
o .1-4 41 I I 1-4 4 .1 o
60 B ...... AM
... -N
'"z ,...........
...'"240 .:

.IA .4-.1 .11.2 1.21.6 1.6-2.0
FIGURE 7(A) Curves of percentage GSR generated by three runs
of stimuli of ascending and descending intensity (in map) by the
amygdalectomized (AM) and control (N) groups. 7(B) A finer
breakdown of stimulus values from .1 to 1.0 mamp, pooled
ascending and descending values.

tion has been repeatedly shown to be impaired by lesions of the

mediobasal motor mechanism (Ruch and Shenkin, 1943). This failure to
habituate has often been termed hyperactivity, though it was soon
established that it was more truly a hyperreactivity (Mettler and Mc-
Lardy, 1948). The hyperreactivity is, however, not so much an in-
creased initial reaction (though there is some of this as seen in the GSR

threshold to shock experiment) as it is a persistence of reactivity long

after controls have become habituated.
But here we meet a paradox. Behaviorally, the monkeys with
lesions of the mediobasal motor system fail to habituate, i.e., they
continue to orient long after control subjects react to a recurring
situation as familiar (Schwartzbaum, Wilson, and Morrissette, 1961).
As already described, however, when we looked to visceral-autonomic
responses to indicate orienting, such responses could hardly be found
(Bagshaw, Kimble, and Pribram, 1965). The evidence thus suggests
that visceroautonomic responses are integral to habituation.
Can visceroautonomic responses be integral to habituation and
also be the determinants of emotional feeling? In man habituation
precludes awareness. We are not ordinarily aware of wearing clothes
that have become familiar, of movements that have become habitual, of
digestive functions or heart beats that recur more or less regularly. Only
when dishabituation takes place do we notice such objects and events.
The feeling is therefore attendant on dishabituation-disruption of the
current set or state. Bodily changes may accompany the disruption, and
the ensuing visceral and somatosensory input may in fact contribute to
the general emotional or motivational feeling. But, as I have reviewed
elsewhere (Pribram, 1967a, 1967b, 1971, Chapter 11), the attribution of
specific feelings to the change in state is as much a function of the
situation in which the change occurs as it is of the visceral and somatic
changes per se. As pointed out in the introduction, the feeling of upset
can readily be distinguished from the disposition (e.g., being in love)
from which the upset takes its origin.
Habituation poses another problem. If we habituated in every
recurring situation we would never be able to deal with such situations;
we would never learn, would never be able to attend to now this, now
that aspect of a situation. The organism must possess a mechanism
which overrides habituation. In a recent review, Diane McGuinness
and I (Pribram and McGuinness, 1975) spelled out the details of this
system, which appears to depend on greater involvement of the soma-
tomotor rather than of the visceroautonomic system. The data suggest
that three separate but interacting neural systems govern the reaction to
novelty and its habituation. One system controls arousal, which is
defined in terms of phasic physiological responses to input. The arousal
control circuits center on the amygdala, a core structure in the mediob-
asal motor system. A second system controls activation, which is de-
SEC tv
o 5 :::::: :: :::::: :~::: ::: :' 5 5 '0 '5
o. ~
I ............ ....
............ ....
... . ... ... .. I.o.....................
.. ..
.... ....
.... o.........
o... ...... ..... .....
.... ............ .. .. .
.......... .... .. o. .......... .. .. ..

A 6~t A

TRIALS GRP 5-10 SEC ON 10-15 SEC ON 5-10 SEC OFF 10-15 SEC OFF
FIRST 40 NORM 3.7 7. Om': 3.9 7.0
AMX 3.2 3.3 3.9 6.3

SECOND 40 NORM 5. 7~'c* 8.8n 6.2 4.5

AMX 2.7 4.8 3.5 4.3
-AI I 80 NORM 9.3 14. 5~': 10.3 7.0 ~

AHX 5.8 8.2

7.3 6.3 ~

FIGURE 8. Mean number of GSR occurring in 10-sec period of light on just preceding light offset (CS) in the first 40 and in the second 40 ~
trials for each group. Conditioning paradigm presented above table. E::

fined in terms of tonic physiological readiness to respond. The readi-

ness circuits center on the basal ganglia of the forebrain. A third system
was discerned which coordinates arousal and activation. This coordi-
nating activity apparently demands effort. Its circuitry centers on the
Even at the hypothalamic level the distinction between an arousal
and a readiness mechanism exists. In the feeding mechanism, for
instance, the ventromedial region (the one involved in making obese
rats in the ad libitum situation) has been shown to monitor the utiliza-
tion of blood sugar during satiety, while the far lateral hypothalamic
region functions reciprocally to intiate feeding when utilization has
come to a halt. The satiety mechanism stops behavior; the feeding
mechanism makes behavior go. And recent evidence (Ungerstedt, 1974;
Fibiger, Phillips, and Clouston, 1973) has shown these far lateral hypo-
thalamic effects to be due to disruption of basal ganglia circuits. Else-
where (Pribram, 1971, Chapter 10) I have suggested that emotional
arousal becomes organized around "stop" mechanisms and that moti-
vational readiness is an elaboration of "go" mechanisms. As detailed in
the review, there is ample evidence that the limbic forebrain (e.g., the
amygdala and the hippocampus) participates in such processes. Here it
is important to recall that the evidence also shows that this participation
takes place by way of the organism's reactivity to novelty and familiar-
ity. The evidence is thus consonant with that from the clinic where
epileptic auras of deja and jamais vu preceding psychomotor seizures are
considered pathognomonic of disturbances of the limbic mediobasal
motor formations.

D. James Reconsidered

In his opening paragraph on emotions, William James suggested

that "emotional reaction usually terminates in the subject's own body
whilst the instinctive (motivational) reaction is apt to go farther and
enter into practical relations with the exciting object" (1950 Vol. II, p.
442). This distinction rather than the one more commonly attributed to
James-that emotion is essentially viscerally determined-is borne out
by our review of current data. James was overly impressed with Lange's
argument because "reactions that terminate in one's own body" (Le.,
self-regulatory reactions) tend to display a larger visceral component

than do "reactions" which "enter into practical relations with the

exciting object." Entering into practical relations demands somatomo-
tor activity. Yet emotional arousal also involves the somatomotor sys-
tem, and somatomotor activity is accompanied by visceroautonomic
changes. In short, we have a considerable amount of evidence which
demands a modification of the James-Lange position that "bodily
changes follow directly the perception of the exciting fact and our
feeling of these same changes is the emotion." Feelings of familiarity, of
elation and depression, of assertion and aggression, and of sleepiness
and alertness have been shown to depend on brain processes (see
Pribram, 1971, Chapters 10 and 15; Pribram and McGuinness, 1975; or
Marshall and Teitelbaum, 1974, for a review of the evidence that relates
neurochemical brain systems to these dispositional states). Bodily
changes are initiated by these brain processes, but not, as James
thought, by the processes that directly perceive. Rather, the bodily
changes are induced by mechanisms which monitor the familiarity and
novelty of situations. Bodily changes, both visceroautonomic and so-
matomotor, do appear to be integral to emotional and motivational
expression, however, in that they do in fact help to distinguish their
mechanisms of operation. But even here the distinction does not rest on
which peripheral mechanism becomes activated, but rather on how
they both are used. If the brain processes regulating bodily changes
lead the organism into doing something about a situation, i.e., "enter-
ing into practical relations" with it, motivational mechanisms become
active; when these brain processes result in reactions "terminating
within the subject's own body," emotional mechanisms are set into
operation. This, then, is the essential distinction between motivational
and emotional expression: What mechanism sets one rather than the
other process in motion?



A. Part Behaviors and Their Integration

In attempting to answer this question, we need to examine the

impact of the finding that effort, rather than drive, is the critical
variable determining the behavioral response. Presumably, therefore, it

is effort that resolves whether the organism will react emotionally (i.e.,
by attempted self-regulation) or motivationally (i.e., by entering into
practical action). Effort is a measure of the resistance which must be
overcome in order to do a certain amount of work in a specified time. It
is analogous to force in physical systems and an expression of the
capacity for exerting power, the rate of doing work. And work is a
measure of the energy required to change the state of a system (see
McFarland, 1971, p. 4, for the derivation of these definitions).
The critical question is, therefore, What are the variables which
constrain a system to resist a change in state? A homeostatic system, by
definition, is one that resists change by virtue of its negative feedback.
But additional constraints develop (hyperstability) when several such
systems interact (Ashby, 1960). Thus a drop in basal temperature may
result in shivering, in motor activity, in sleeping, or in eating. Motor
activity and eating have been shown to be reciprocally related over
short time periods-they appear constrained by the basal temperature
variable. It therefore takes effort to attempt to eat during or immediately
after exercising and vice versa (see Brobeck, 1963, for a thought-provok-
ing and thorough review of these data).
There is considerable evidence as to the neural organization that
invokes such constraints. Electrical stimulations in the hypothalamic
region, when carried out with small electrodes, give rise to only parts of
behavioral acts, such as jaw or tongue movements, swallowing, pilo- or
genital erection, head thrusts, etc. Further, these part behaviors appear
to be more or less randomly interspersed with one another. Adjacent
stimulations do not produce a completed behavior pattern (Roberts,
When larger electrodes are used, a different pattern emerges. Now
chewing, drinking, sexual, or aggressive behaviors are elicited full-
blown. But interestingly, which behavior is elicited depends to some
considerable extent on the environmental situation in which the stimu-
lation occurs. Thus a rat, when initially stimulated, may drink every
time the electrical current is applied to his brain. He is now left
overnight in a cage with no opportunity to drink but with several pieces
of wood to chew on. The brain stimulation is kept up intermittently all
night. The next morning the rat is provided with the opportunity either
to drink or to chew. Now, brain stimulation from the identical site will
as often elicit chewing as drinking (Valenstein, Cox, and Kakolewski,
1969; Valenstein, 1970).

B. The Precentral Motor Cortex and Action

These results have led to a controversy similar to that which for

many years centered on the functions of the classical precentral motor
cortex. The issue concerned the nature of the motor representation: Is it
punctate, representing discrete muscles or even parts of muscles, or are
movements, sets of muscle contractions, flexibly represented? I have
elsewhere (Pribram, 1971, Chapters 12 and 13) pointed out that neu-
roanatomically the representation is indeed punctate, that neurophys-


iologically, i.e., with fairly gross electrical stimulations of awake ani-

mals and humans, movements rather than discrete muscle contractions
are obtained, and that which movement is elicited depends on body
and limb position, prior stimulation, etc. But I also have shown (Pri-
bram, Kruger, Robinson, and Berman, 195~56) that neither of these
views is sufficient to explain the results of neurobehavioral experi-
ments. These show that resection of the classical motor cortex fails to
interfere with any muscular contraction, or even with any set of muscu-
lar contractions. All movements can be shown to remain intact when


C 6
4 -
2 -

FIGURE 9(A) Subject in black costume with white tape. (Reprinted with permission
from N. Bernstein, 1967.) 9(B) Cinematograph of walking. Movement is from left to
right. The frequency is about 20 exposures per second. (Reprinted with permission from
N. Bernstein, The Co-ordination and Regulation of Movements. Pergamon Press Ltd.,
1967.) 9(C) Force curves at the center of gravity of the thigh in normal walking.
(Above) vertical components. (Below) Horizontal components. (Reprinted with permis-
sion from N. Bernstein, The Co-ordination and Regulation of Movements. Pergamon
Press, Ltd., 1967.)

they are examined in a sufficient range of situations. Yet the monkeys

were defective in solving latch-box problems, and the deficiency was
not due to any overt difficulty in sequencing the movements. I therefore
came to the conclusion that the essential representation in motor cortex
was neither of individual muscles nor of movements, but of actions,
defined as environmental consequences of movements.
Subsequent reports (Bernstein, 1967) have clarified the possible
mechanisms by which a representation of environmental consequences
could come about. For instance, cinematographic records are made of
people performing relatively repetitious tasks, such as hammering a
nail or running over rough terrain when dressed in black with white
markings on their limbs. Such records display continuous wave forms
which can be analyzed as if they were modulated sine waves. The use of
such a Fourier analysis allows accurate prediction to be made of the
extent of the next movement in the series, the next hammer blow or
step. If this can be done by an investigator in this fashion, it is not far-
fetched to believe that it can be done in a similar way by the subject's
motor system. The essential representation would therefore be the
equivalent of the mathematical operation of Fourier analysis, a consid-
erable saving in storage over representing each movement and se-
quence of movements that might ever be utilized. This program, or a
similar stored set of mathematical rules, could readily assemble the
more or less randomly dispersed part-functions which have been
demonstrated with discrete stimulations, much as these rules have
been used to make a computer-driven dot display that is interpreted
by the observer as a running or dancing figure (Johansson, 1973).
Other experiments (Evarts, 1967) have shown with microelectrodes
that muscle length is not the relevant variable to which motor cortex
neurons respond. These experiments were performed on fully awake

FIGURE 10(A) View of the lateral surface of the cerebral cortex showing the distribution
of potentials evoked by the stimulation of a cutaneous or a muscular branch of an arm
nerve. Plus (+) indicates a response of 100 microvolts or more; triangle (Il) indicates a
response of from 50 to 100 microvolts; and open circles indicate response of from 0 to 50
microvolts. 10(B) Cortical responses evoked by sciatic nerve stimulation before resec-
tion of postcentral cortex and cerebellum. (1) Upper trace, postcentral; lower trace,
precentral. Time: 10 msec. (2) Same immediately after resection of both cerebellar
hemispheres. (3) Same immediately after resection of both cerebellar hemispheres. (3)
Same following additional resection of anterior lobe of cerebellum. (4) Same after
additional resection of both postcentral gyri. Note that postcentral record how registers
only white matter response.

monkeys taught to pull an adjustable counterweighted lever. The re-

sponse of motor cortex units did not vary with the length of the
excursion of the lever but with the effort necessary to move it-the force
that had to be applied to overcome the resistance to movement due to
the weight.

C. Effort and Volition

Thus both in the hypothalamic experiments and in the experiments

on classical motor cortex, situational stimulus variables are seen to be
critically involved (this accounts for the findings of Teitelbaum, 1955,
that enhanced attractiveness of food helps overcome the resistance to
eating shown by the animals with lesions in the lateral hypothalamus).
Both of these parts of the brain do, of course, receive rather direct
inputs from exteroceptors. In the case of the classical motor cortex, our
discovery (Malis, Pribram, and Kruger, 1953) of these paradoxical in-
puts to a "motor" region provided the original impetus for this line of
investigation. In the case of the hypothalamus and mediobasal motor
cortex, the existence of these direct inputs from exteroceptive receptors
is just now beginning to be established with microelectrode and new
neuroanatomical autoradiographic techniques (see, e.g., Cowan, Ad-
amson and Powell, 1961; Cowan, Gottlieb, Hendrickson, Price, and
Woolsey, 1972). However, some early neurophysiological results exist
showing changes in electrical activity evoked by peripheral stimulation
(Bailey and Sweet, 1940; Dell, 1952; Pribram and MacLean in Fulton,
1951, p. 57).
The distributed representation of part behaviors is the neural sub-
strate upon which effort variables (as induced by deprivation, for
instance) critically operate to determine whether the expressed behav-
ior is to be emotional or motivated. We have seen that even at the
neural unit level, neurons in motor systems are directly sensitive to
effort variables-i.e., they respond according to the constraints of the
moment. The constraints that need to be overcome by effort have thus
been shown to be externally as well as internally determined. This was
especially clear in experiments on the classical motor cortex: Effort is
what correlated with neural unit activity, not changes in muscle length.
Similar correlations need to be established at the unit level in the
mediobasal motor systems, but, as noted above, the indications from

neurobehavioral data are that some sort of anticipatory mechanism

based on the constraints developed by repetition (familiarity, habitua-
tion) rather than "bodily change" per se is involved.
The upshot of these results is that motivation and emotion reflect
the effort involved in changing bodily systems, not the changes them-
selves. Effort is a brain process (involving the hippocampal circuit-see
Pribram and McGuinness, 1975, for a review of the evidence) that
apppears to be critical in determining whether a reaction is to be
motivated or emotional.

D. The Jamesian Theory of Will

We need, therefore, to take a look at another domain of Jamesian

theory. We have already noted the fact that some sort of appraisal of
familiarity rather than a direct perception of a situation initiates the
motivational-emotional process. We also reviewed the evidence that
the distinction between emotional and motivational behavior was best
stated by the Jamesian view that emotional reactions "terminate within
the body" (i.e., are self-regulatory), while motivational reactions "enter
into practical relations with the exciting object." The Langian portion of
the James-Lange theory-that emotions are the feelings generated by
visceral reactions-we found untenable. And finally we found that we
must invoke "effort" as the critical variable which determines whether
a reaction is to be emotional or motivated. Effort is discussed by James
under the rubric of will. His definition of what leads to voluntary
actions reads much as we have stated it here, if we interpret the words
anticipatory image to mean the resultant of the mathematical operation
from which the next movement in a series can be predicted:
An anticipatory image, then, of the sensorial consequences of a movement,
plus (on certain occasions) the fiat that these consequences shall become
actual, is the only psychic state which introspection lets us discern as the
forerunner of our voluntary acts. (1950, Vol. II, p. 501)

But again, James can be interpreted as taking a dual stance. He

quoted at length from Ferrier, who attempted to show that input from
muscular contraction (usually the holding of one's breath when other
evidences of muscular contraction are missing) is necessary for the
experience of effort. (Ferrier was making the argument against Wundt,
whose views were that efferent rather than the afferent neural activity

was perceived as effortful). James heartily endorsed Ferrier's views:

[the experiments reviewed] prove conclusively that the consciousness of
muscular exertion, being impossible without movement effected somewhere,
must be an afferent and not an efferent sensation; a consequence, and not an
antecedent, of the movement itself. An idea of the amount of muscular
exertion requisite to perform a certain movement can consequently be
nothing other than an anticipatory image of the movement's sensible effects.
(Vol. II, p. 505)

Note carefully here what James was saying. Superficial reading

makes the statement sound like another version of the James-Lange
theory: James-Lange for emotion; James-Ferrier for motivation and
will. But here James clearly stated that "an anticipatory image of the
movement's sensible effects" is involved. Such an "image" must be a
brain, not a peripheral, process. A careful reading of this passage makes
one wonder whether the Jamesian theory of emotions, interpreted as
peripheralist, has not been grossly misinterpreted as well. James im-
plicitly and explicitly always had a brain process in mind whenever
discussing mind (thoughts, feelings, consciousness, attention, etc.). In
summarizing his chapter on emotions, James was discussing brain
processes, not peripheral ones:
To sum up, we see the reason for a few emotional reactions; for others a
possible species of reason may be guessed; but others remain for which no
plausible reason can even be conceived. These may be reactions which are
purely mechanical results of the way in which our nervous centres are
framed, reactions which, although permanent in us now, may be called
accidental as far as their origin goes. In fact, in an organism as complex as
the nervous system there must be many such reactions, incidental to others
evolved for utility's sake, but which would never themselves have been
evolved independently, for any utility they might possess. Sea-sickness, the
love of music, of the various intoxicants, nay, the entire aesthetic life of
man, we have already traced to this accidental origin. It would be foolish to
suppose that none of the reactions called emotional could have arisen in this
quasi-accidental way. (Vol. II, p. 484)

We note, therefore, that the contemporary view of the theory of

motivation and emotion proposed by William James is in one respect
grossly misleading. While James wrote that emotional feeling was
based on visceral sensations, he also wrote that such feeling was
coordinate with a brain process resulting from the visceral sensation.
This central aspect of Jamesian theory becomes even more clearly stated
with respect to motivation and has been little appreciated by James's

On the other hand, James was in error in suggesting that emotion

depended on immediate visceral sensation (or that motivation de-
pended on immediate sensations derived from the somatic muscula-
ture). Cannon's classic experimental demonstrations that an organism
is capable of emotional responses despite visceral deafferation have
been the source of the major rebuttal to James's position, although
exceptions to the validity of Cannon's claims have also been voiced
(e.g., Beebe-Center, 1971; Schachter, 1967; Mandler, 1967).
Reviewed here have been additional experiments that make it
necessary to modify Jamesian theory of emotion. James clearly wrote
of "an anticipatory image" when dealing with motivation. On the
basis of the experimental results on limbic system function-which
have shown that visceral responsiveness follows the appreciation of
novelty and the appraisal of changes in sets, expectations, and
anticipations, not directly on perceived events per se--it is now
mandatory to think in like fashion about emotion.
However, we also reviewed an aspect of Jamesian theory which is
acceptable today and accounts for James's overemphasis on immediate
bodily sensations while at the same time providing a useful distinc-
tion between emotion and motivation: Emotional expression tends to
terminate within the organism while motivations enter into practical
relations with the exciting event. Entering into practical relations often
involves effort or will-thus the distinction still used in the neurologi-
cal clinic of the apposition of emotional to voluntary behavior.



A. The Model

In the introduction I suggested that the scientific study of self-

consciousness would show rapid progress provided a technique and a
brain model were made available. The technique of biofeedback lead-
ing to the self-regulation of dispositional states appears to fill part of
this need. The present manuscript has attempted to show that Jamesian
theory might be useful in launching the necessary brain model.

What sort of model, then, can be constructed from these elementary

observations? What form of discourse is available to describe the
model? What type of "in vitro" simulations can aid our understanding
of the processes and mechanisms involved?
I want to propose that contemporary control theory can and does
provide the model, the language, and initial understanding of the brain
processess and mechanisms involved in the interrelated domains of
motivation and emotion, of effort and will, and of self-regulation and
self-consciousness. Specifically, the concepts of feedback and feedfor-
ward as they describe closed and open (helical) loop systems are useful
in the formulation of a testable model of this domain of inquiry in
precise, scientifically useful terms.
Control theory is no foreigner to biological and neurological expla-
nation. The term regulation is as often used as the term control, but
biological principles are almost universally regulatory principles-i.e.,
principles invoking mechanisms of control. With regard to the issues
under consideration, homeostasis and homeorhesis are thoroughly
tested conceptions and biofeedback has become a household word. And
to view the nervous system as an information-processing mechanism
is now standard practice among neurophysiologists.
The proposal derived directly from Jamesian theory states simply
that emotion is essentially based on closed-loop feedbacks, while moti-
vations go beyond these and "enter into practical relations" by way of
information-processing, open-loop, feedforward mechanisms. The
maintenance of "practical relations" demands repeated changes (bias-
ing) in the constraints (the feedbacks) operating on the system; thus
voluntary effort is a necessary concomitant of open-loop processes.
The most generally known innovation in control theory has been
the formal description of the concept of feedback (e.g., Miller, Galanter,
and Pribram, 1960), a circular process initiated by a test, a matching of
two settings. When there is mismatch, one of the settings becomes
fixed, while the other triggers an operation which continues until a
match is produced. Thus a test-operate-test-exit sequence, a TOTE,
characterizes the feedback: For example, if the setting of a thermostat
and that of room temperature are incongruent (mismatch) a furnace is
turned either on or off until congruence is established.
More recently another, equally useful conception-feedforward
(e.g., see MacKay, 1969; Mittelstaedt, 1968; Pribram, 1971)-has been
found important. In feedforward control, an operation procedes to a


Control Motor Extraocular

muscle Eyeball
centre outflow control system


B Control Motor Limb muscle Limb
centre outflow control system


FIGURE l1(A) Eyeball position control system: (open-loop, feedforward) and (B) limb
position control system: (closed-loop, feedback). (Reprinted with permission from D. J.
McFarland, Feedback Mechanisms in Animal Behavior. Academic Press, 1971.)

predetermined end point. For example, in most apartments, the furnace

continues to operate for fixed periods, irrespective of local temperature
The distinction between feedback and feedforward has been ex-
tremely useful in the analysis of engineering and biological systems,
which ordinarily are composed of complex combinations of feedback
and feedforward processes. Two types of combinations have been
extensively studied. In one, feedback processes become associated or
multiply linked with each other, producing an extremely stable system
resistant to change (i.e., they exhibit equilibrium and inertia). An
engineering example of such a system is the multilinking of power
plants in the northeastern United States, which guards against frequent
local disruptions, though it is vulnerable to occasional massive failure.
Biologically, physiological drive systems have been found to display
this type of organization. Thus food intake, muscular activity, tempera-
ture regulation, and water metabolism are interdependent regulatory
mechanisms which, as a rule, operate to maintain basal temperature

constant. The associative links which make up this and similar systems
have been studied extensively (e.g., see review by Brobeck, 1963) and
their operating characteristics thoroughly analyzed (Ashby, 1960).
Combinations based primarily on feedforward processes are ubiq-
uitous; they constitute our computer technology. For the most part,
such combinations contain feedbacks as well. Biologically, combina-
tions of feedforwards occur in parallel, processing signals simultane-
ously by virtue of overlapping neighborhood interactions, and consti-
tute one class of cognitive processes (see Neisser, 1967; Eccles, 1967;
Pribram, 1971). When feedback loops are included, hierarchical se-
quential arrangements called plans or programs are constituted (Miller,
Galanter, and Pribram, 1960). Parallel and hierarchical processing
mechanisms provide the foundations of contemporary cognitive
In biology, homeostatic processes, oscillating phenomena such as
biological rhythms and clocks, and load-adjusting mechanisms such as
those regulating muscular contraction have all been shown dependent
upon feedback organization (Pribram, 1971). The essential characteris-
tic of such systems is that they depend upon a match between two
settings. A mismatch produces an error signal which controls the
operation of the system until equilibrium-match-is reestablished.
This homeostatic conservation of equilibrium is akin to that described
by the first law of thermodynamics, which states that the conservation
of energy is maintained because change elicits an "equal and opposite
reaction." Energy concepts are therefore appropriate to a description
and an understanding of feedback organizations and, in fact, are regu-
larly used, as for example in the description of the "effort" or "work"
involved in load-adjusting mechanisms.
Information concepts, by contrast, have often been linked to the
second law, and in fact, information has often been termed (e.g., by
Brillouin, 1962) neg-entropy (see also von Foerster, 1965). Confusion
has arisen because there has been a tendency to label "error" (the
mismatch signal) information. But "error" has nothing in common with
this type of "information": The amount of information contained in a
message does not depend on the processing of its errors. Ashby (1963)
details the distinction in terms of the constraints (limits on the
independence of the functioning parts) operating on the processing
system, the constraints on variety. Information is a measure of variety;
redundancy (repetition), a measure of constraints. (For a comprehen-

'OPU'-rl Tcst


I reedback
I reedf'orward


I reedbaC}.J I I

l_1 Bia5

FIGURE 12. The TOTE servomechanism modified to

include feedforward. Note the parallel processing fea-
ture of the TOTE.

sive discussion of the nature of constraint in physics and biology, see

Pattee, 1971). Information thus refers to the content of a communica-
tion, while redundancy reflects the context or code in which informa-
tion is communicated. Feedback organizations constrain systems to
equilibrium. Thus error becomes a term denoting redundancy or lack
thereof, not information. When a long-distance conversation is inter-
rupted by a periodic whoosh, the constraint (the context in which the
information is relayed) becomes disturbed-the conversation becomes
unintelligible and a mismatch is conveyed to the sender, who then
repeats the same information more slowly with greater emphasis and
perhaps several times, changing the structure of the constraints
operating during the conversation without altering its content (the
amount of information).
From this it follows that the invoking of biofeedback procedures
accomplishes its purpose by providing an external bias on the internal
feedbacks that maintain the ordinary homeostases operating in the
system. The bias, maintained with effort, produces conscious voluntary
control on the system, which now is an information-processing, feed-
forward, open-loop, helical mechanism rather than just an unconscious
error-processing, feedback, closed-loop system.
The outline of a model for self-regulation and self-consciousness

thus appears to be relatively easy to discern. Even the neural mecha-

nisms involved are to some extent becoming known. With regard to the
effort involved in the coordination of internal control with external
demand, the evidence of the critical role of hippocampal circuit has
been reviewed elsewhere (see Pribram, 1971, Chapter 15; Pribram and
McGuinness, 1975). With respect to the coordinations involved in t~e
maintenance of "practical relations" with the exciting event, the cere-
bellar circuit appears critical (Pribram, 1971, Chapter 13). Both hippo-
campal and cerebellar mechanisms, based on somewhat comparable
anatomical structure, can be thought to perform rapid calculations of
probable future states from extrapolations of present and immediately
past circumstances. The remainder of the system can thus change its
operations to achieve or preclude that particular estimated future state's
occurring. New calculations then take place and the process is repeated,
monitoring and extrapolating continuously the changes, or lack thereof,
which result.

B. Attention Span and Self-Consciousness

While the proposal of a plausible model of self-regulation and self-

consciousness is thus feasible, understanding the genesis of self-con-
sciousness poses greater difficulties. Two points are clear: A change
from feedback to feedforward organization effected through biofeed-
back procedures leads to conscious, voluntary control; we exercise this
control by "paying" attention. Thus the key to understanding the
genesis of self-consciousness is attention, and specifically the set of
problems psychologists deal with under the rubrics of attention span
and central capacity.
James, in discussing the span of attention, reviewed (1950, Vol. I,
pp. 427-435) the reaction-time experiments of Wundt, Exner, and Mun-
sterberg. He concluded that the results indicate that shorter times are
elicited by the following mechanisms:
(1) The accommodation or adjustment of the sensory organs; and (2) The
anticipatory preparation from within of the ideational centres concerned
with the object to which the attention is paid ... The two processes of
sensorial adjustment and ideational preparation probably coexist in all our
concrete attentive acts. (Vol. I, p. 434).

Again, attention consists of:

a collection of activities physiologically in no essential way different from

the overt acts themselves. If we divide all possible physiological acts into
adjustments and executions, the nuclear self would be the adjustments
collectively considered. (Vol. I, p. 302)

We note here the recurring theme which differentiates "adjustments"

that end within the organism's body and "executions," which go
beyond into practical actions. Further, a "nuclear self" can be ascer-
tained by consideration of the collection of adjustments which shorten
reaction time.
Today, reaction-time experiments have once more raised the issue
of attention span and its dependence on some sort of nuclear self or
competency to adjust central capacity. The issue is handled in the
Pribram-McGuinness review (1975) in terms of control theoretic con-
cepts and is worth repeating here because of its relevance to the
problem of the mechanism which brings about self-consciousness.

C. Central Competency

In living systems, an arousing stimulus often increases the uncer-

tainty of the organism by its novelty. This effect of input information is
contrary to that obtained in nonliving communication systems, where
information conveyed always reduces uncertainty. The difference be-
tween living and nonliving systems can be conceptualized in terms of
the channel over which the communication takes place. In nonliving
communication systems the channel is akin to a sensorimotor channel
which is fixed in capacity and does not alter with the communication.
Living systems (and also computers) have the capability of memory,
which alters the competence with which they process information (Pri-
bram, 1971, Chapters 14 and 16). This is produced by the alteration of
channel redundancy and superficially resembles a change in the num-
ber of channels with fixed capacity. The increase in competence is the
result of an increase in the complexity of the neuronal model, an
encoding process described as "chunking" the information (Miller,
1956; Simon, 1974). This and similar mechanisms in human informa-
tion-processing effect a change in central processing very different from
that produced by a simple increase in the number of fixed-capacity
channels available.
The evidence that information-processing competency can be
changed in living organisms comes from a variety of problem-solving
situations. Kahneman (1973), in reviewing several such studies from

the psychophysiological literature, suggested that" arousal" is in fact an

indicator of a change in capacity-"the allocation of spare capacity"-
much as this is changed in nonliving systems by an increase in the
number of channels available. He also goes on to equate "arousal" and
"capacity" with" effort" and" attention" in a global fashion. As noted,
however, emotional arousal is indicative of but one sort of attention,
and effort is involved only when the situation demands the regulation
of arousal and motivational readiness to produce a change in informa-
tion-processing competency.
The way in which competency is controlled by brain systems in the
living primate is demonstrated by the finding that removal of the area
of the brain usually called sensory or posterior intrinsic or "association"
cortex reduces the sampling of novel alternatives. The opposite effect is
obtained when the lateral frontal cortex is resected. Removal of this
same frontal cortex leads to an increase in behavioral orienting and an
abolition of the viscerautonomic components of orienting. There thus
appear to be opposite effects (posterior and frontal) on the number of
alternatives sampled in a situation. This was interpreted to indicate a
dual control mechanism determining the ability to sample (Pribram,
Supportive behavioral evidence came from an experiment by
Butter (1968, 1969), in which he investigated the number of features
usually attended by monkeys while discriminating between two cues.
He did this by eliminating each feature in turn in various combina-
tions. He found that resection of the same brain region (the posterior
cortex) that produced a restriction in the number of alternatives
sampled also produced a restriction in the number of features used to
make the discrimination.
Electrophysiological evidence has been obtained that the posterior
and frontal cortex contribute opposing controls on sensory channels.
This evidence is based on changes produced in the recovery cycles of
the system (the speed with which the system recovers to its full capacity
after a sudden, intense stimulus) and the alterations produced in the
shape of visual receptive fields (Spinelli and Pribram, 1966, 1967).
These changes in sensory channels were, however, not attributed
to a simple change in the number of channels of fixed capacity, as the
effects of surgical resection have shown that as little as a few percent of
an anatomically defined sensory channel is sufficient for ordinary dis-
crimination learning, performance, and transfer (Lashley, 1929; Galam-

bos, Norton, and Frommer, 1967; Chow, 1970). The remainder of any
input channel appears to be redundant, spare channel capacity, under
most circumstances. The results on the control of input channels by
posterior and frontal cortex were therefore interpreted (Pribram, 1967c)
as influencing redundancy, not sensory capacity in the usual information
theoretic sense. Specifically, it was suggested that the input systems
acted as channels in which spatial and temporal multiplexing could
occur, a suggestion similar to that put forward by Lindsay (1970).
On the basis of the data reviewed above, Kahneman's (1973) con-
cept that arousal involves an increase in the number of sensory chan-
nels available can be generalized to include constraints involving the
redundancy characteristics (the competency) of that capacity. Kahne-
man's discussion approached such a generalization when he spoke of
changes in "structural connections between components." In technical
language, such changes in competency would be reflected in changes in
the equivocation of the channel (defined as the sum of noise and
redundancy). Competency is the reciprocal of equivocation. Effort can
then be defined as the measure of the attention "paid" to increase
or maintain efficiency by reducing equivocation, i.e., enhancing

D. External Versus Internal Constraint

Gamer (1962) in his analysis of the structure of redundancy has

shown that the total amount of constraint operating in any system of
variables can be divided into internal and external components. Internal
constraints refer to the relationships among the system of variables
under consideration, while external constraints refer to the relationship
between these variables and some external referent system of variables.
In our neurophysiological experiments we considered the constraints
that describe the central operation of the channel as internal and the
constraints that refer to operations on the environmental situation
which control its sensory input as external. In addition, it was found
important to distinguish between temporal (repetition of the use of the
channel or variable over time) and spatial (replication of the variable
over space) redundancy for each of Gamer's categories.
Specifically, it was suggested (Pribram, 1967c) that when the fron-
tal system becomes involved in the orienting reaction, the internal

redundancy in the input channel is increased so that all of the informa-

tion being simultaneously processed becomes uchunked" into one unit.
By contrast, when the posterior cortex becomes involved in the atten-
tional process, internal redundancy in the input channels is decreased,
separating the bits of information in each channel from each other. This
is concomitant with enhancement of external redundancy, which ac-
cording to Gamer's findings enhances the ability to make discrimina-
tions, i.e., to categorize input.
In short, the controls on emotion and motivation operate on the
mechanisms of redundancy, on the constraints operating within and
between channels, rather than on the uinformation" being processed.
These constraints involve a neuronal model and may be conceived of
as operating on memory rather than on input information. Another
way of stating this is to say that the controls operate on the representa-
tional context in which the informational content is processed.
A good deal of additional evidence can be cited to show that
competency rather than sensory channel capacity per se is controlled by
the attentional systems discussed here. For instance, the studies of
Anderson and Fitts (1958) cited by Gamer (1962) show that as much as
17 bits of sensory information can be simultaneously processed. The
work of Lindsay (1970), which demonstrated the relationship between
sensory discriminability (difficulty in distinguishing between inputs)
and central processing competency, has already been mentioned. Pri-
bram, Lim, Poppen, and Bagshaw (1966) and Mishkin and Pribram
(1955), using various forms of the delayed alternation tasks, attributed
the differential effects obtained after resections of two reciprocal
frontoamygdala systems as due to selective alterations in the structure
of internal redundancy (spatial and temporal, respectively) of the re-
maining processing competency. Further, Pribram and Tubbs (1967)
have shown that when the delayed alternation task, the nemesis of
monkeys with frontal-lobe resections, is externally parsed or chunked
as a result of making the intertrial intervals asymmetric, the deficit is
completely overcome. Similarly Wilson (1968) analyzed the trade-off
between tasks involving external temporal and spatial redundancy in
reciprocal mechanisms (anterior and posterior inferotemporal cortex)
which have been delineated within the posterior system.
Thus, both Kahneman's (1973, pp. 8, 9, 15) and this analysis
attribute the control of attention to alternations in information-pro-
cessing channels, not the direct control on information and uncertainty

per se. We differ in that Kahneman focused on the problem of

increasing the number of channels of fixed capacity-the" allocation of
spare capacity"-while this analysis emphasizes the broader issue of
competency, defined by any constraints operating on the structure of
channel redundancy. We also differ in separating readiness from
arousal and in that we do not identify attention, arousal, readiness,
and effort as different names for the same process. Finally, the model
put forward here specifies that effort accompanies only those atten-
tional processes which result in a change in the representational
organization of the information-processing mechanism. Part of the
mechanism detailing how and when effort is expended during atten-
tion has been revealed by studies measuring peripheral autonomic
and somatic changes.
A way to picture this somewhat technical account of the model is as
follows. Most psychologists today view the limitation on central pro-
cessing to be due to a fixed frame (i.e., "frames of consciousness"),
which limits the momentary capacity of a channel, much as does the
exoskeleton of a crustacean. The model proposed here is that the limits,
(the constraints) are not exoskeletal but endoskeletal-they operate by
virtue of the internal structure of the channel, not by some outer shell,
or "frame," that encases it. Further, the evidence from brain research as
well as from behavioral research indicates that the internal skeleton is
flexible: It can be reorganized into a variety of configurations. Organi-
zation involves "paying" attention and comes about in two ways:
through purely mnemonic internal emotional "adjustments/l (control of
internal redundancy) or through the motivational "execution of practi-
cal relations" with external events (control of external redundancy).
A good deal remains to be explained. Do these observations deal-
ing with overall central capacity and competency also apply to how the
attentional mechanism becomes intentional? That is, what brain pro-
cesses allow the act and actor, percept and perceiver, to be simultane-
ously attended? Does self-consciousness accrue simply as a dividend
from the fact that central competence as a whole fluctuates around the
"magical number 7/1 (Miller, 1956), or is a higher-order constraint
necessary to its genesis? Is, as suggested here, the change from a
homeostatic, error-processing feedback mechanism to the parallel
processing of information in an open-loop mechanism sufficient
explanation, or is the change from a holonomically constrained system
(described by integrable differential equations) to a nonholonomic

system the essential development? What role do the limbic forebrain

structures (involved in psychomotor seizures) have in making parallel
or nonholonomic processes possible? And are either parallel or nonho-
lonomic attentional mechanisms the essence of such typically human
information-processing abilities as practical skills and linguistic com-
munication and the memory mechanisms associated with such abili-
This is as far as the outlines of the model can take us today. Further
neuropsychological and neurophysiological research and even more
precise formulation of the brain mechanisms of intentionality, atten-
tion, emotion, and motivation in terms of control theory are needed.
But such formulations need not begin de novo. A beginning was made
by William James, as we have seen. In discussing certain clinical
observations, he clearly foreshadowed the endoskeletal model of com-
petency developed here and its relationship to self-consciousness:
If we speculate on the brain-condition during all these different perversions
of personality, we see that it must be supposed capable of successively
changing all modes of action, and abandoning the use for the time being of
whole sets of well organized association-paths. In no other way can we
explain the loss of memory in passing from one alternating condition to
another. And not only this, but we must admit that organized systems of
paths can be thrown out of gear with others, so that the processes in one
system give rise to one consciousness, and those of another system to
another simultaneously existing consciousness. Thus only can we understand
the facts of automatic writing, etc., whilst the patient is out of trance, and
the false anaesthesias and amnesias of the hysteric type ... Each of the
selves is due to a system of cerebral paths acting by itself. If the brain acted
normally, and the dissociated systems came together again, we should get a
new affection of consciousness in the form of a third "Self" different from
the other two, but knowing their objects together, as the result ...
Some peculiarities in the lower automatic performances suggest that
the systems thrown out of gear with each other are contained one in the
right and the other in the left hemisphere. The subjects, e.g., often write
backwards, or they transpose letters, or they write mirror-script. All these
are symptoms of agraphic disease. The left hand, if left to its natural
impulse, will in most people write mirror-script more easily than natural
script ... On Hughlings Jackson'S principles, the left hemisphere, being
the more evolved organ, at ordinary times inhibits the activity of the right
one; but Mr. Myers suggests that during the automatic performances the
usual inhibition may be removed and the right hemisphere set free to act by
itself. This is very likely to some extent to be the case. But the crude
explanation of "two" selves by "two" hemispheres is of course far from Mr.
Myer's thought. The selves may be more than two, and the brain systems
severally used for each must be conceived as interpenetrating each other in very
minute ways. [Italics mine) (1950, Vol. I, pp. 399-400)


The research involved was supported by NIMH Grant # MH12970-

09 and NIMH Career Award #MH15214-13 to the author.


ANDERSON, N. S., AND FITTS, P. M. Amount of information gained during brief exposures
of numerals and colors. Journal of Experimental Psychology, 1958,56, 362-369.
ASHBY, W. R. Design for a brain: The origin of adaptive behavior (2nd ed.). New York: Wiley,
ASHBY, W. R. An introduction to cybernetics. New York: Wiley, 1963.
BAGSHAW, M. H., AND BENZIES, S. Multiple measures of the orienting reaction and their
dissociation after amygalectomy in monkeys. Experimental Neurology, 1968,20,175-
BAGSHAW, M. H., AND COPPOCK, H. W. Galvanic skin response conditioning deficit in
amygdalectomized monkeys. Experimental Neurology, 1968,20, 188-196.
BAGSHAW, M. H., KIMBLE, D. P., AND PRIBRAM, K. H The GSR on monkeys during
orienting and habituation and after ablation of the amygdala, hippocampus and
inferotemporal cortex. Neuropsychologia, 1965,3, 111-119.
BAGSHAW, M. H., AND PRIBRAM, J. D. Effect of amygdalectomy on stimulus threshold of
the monkey. Experimental Neurology, 1968,20, 197-202.
BAGSHAW, M. H., AND PRIBRAM, K. H. Effect of amygdalectomy on transfer of training in
monkeys. Journal of Comparative and Physiological Psychology, 1965,59, 118-121.
BAILEY, P., AND SWEET, W. H. Effects on respiration, blood pressure and gastric motility
of stimulation of orbital surface of frontal lobe. Journal of Neurophysiology, 1940,3,
BAILEY, P., VON BONIN, G., AND MCCULLOCH, W. S. The isocortex of the chimpanzee.
Urbana: The University of Illinois Press, 1950.
BEEBE-CENTER, J. G. The psychology of feeling. In Encyclopedia Brittanica, (Volume 9).
Chicago: Benton, 1971 edition.
BERNSTEIN, N. The co-ordination and regulation of movements. New York: Pergamon, 1967.
BLUM, J. S., CHOW, K. L., AND PRIBRAM, K. H. A behavioral analysis of the organization of
the parieto-temporo-preoccipital cortex. Journal of Comparative Neurology, 1950,93,
BRENTANO, F. Psychologie vom empirischen Stanpunkt (3rd ed.). Leipzig: 1925.
BRETANO, F. The distinction between mental and physical phenomena. In R. M.
CHISHOLM (Ed.), Realism and the background of phenomenology. New York: Free Press,
1960, pp. 39-61.
BRILLOUIN, L. Science and information theory (2nd ed.). New York: Academic Press, 1962.
BROBECK, J. R. Review and synthesis. In M. A. B. BRAZIER (Ed.), Brain and behavior (Vol.
2). Washington: American Institute of Biological Sciences, 1963, pp. 389-409.
BuCY, P. c., AND PRIBRAM, K. H. Localized sweating as part of a localized convulsive
seizure. Archives of Neurology and Psychiatry, 1943,50, 456-461.
BUTTER, C. M. The effect of discrimination training on pattern equivalence in monkeys
with inferotemporal and lateral striate lesions. Neuropsychologia, 1968, 6, 27-40.

BUTTER, C. M. Impairments in selective attention to visual stimuli in monkeys with

infero-temporal and lateral striate lesions. Brain Research, 1969,12, 374--383.
CANNON, W. B. The James-Lange theory of emotions: A critical examination and an
alternative theory. American Journal of Psychology, 1927,34, 106-124.
CHOW, K. L. Effects of partial extirpations of the posterior association cortex on visually
mediated behavior. Comparative Psychology Monographs, 1951,20, 187-217.
CHOW, K. L. Integrative functions of the thalamocortical visual system of cat. In K. H.
PRIBRAM AND D. BROADBENT (Eds.), Biology of memory. New York: Academic Press,
1970, pp. 273-292. .
COWAN, W. M., ADAMSON, L., AND POWELL, T. P. S. An experimental study of the avian
visual system. Journal of Anatomy (London!, 1961,95, 545-563.
The autoradiographic deminstration of axonal connection in the central nervous
system. Brain Research, 1972,37, 21-51.
DELL, P. J. Correlations entre Ie systeme vegetatif et Ie systeme de la vie de relation;
mesencephale, diencephale et cortex cerebral. Journal de Physiologie (Paris), 1952, 44,
ECCLES, J. c. Postsynaptic inhibition in the central nervous system. In G. C. QUARTON, T.
MELNECHUK, AND F. O. SCHMITT (Eds.)' The neurosciences. New York: Rockefeller
University Press, 1967, pp. 408-427.
EVARTS, E. V. Representation of movements and muscles by pyramidal tract neurons of
the precentral motor cortex. In M. D. YAHR AND D. P. PURPURA (Eds.), Neurophysiol-
ogical basis of normal and abnormal motor activities. Hewlett, N.Y.: Raven Press, 1967,
pp. 215-254.
FIBIGER, H. c., PHILLIPS, A. G., AND CLOUSTON, R. A. Regulatory deficits after unilateral
electrolytic or 6-0HDA lesions of the substantia nigra. American Journal of Physiology,
1973,225(6), 1282-1287.
FULTON, J. F. Frontal labotomy and affective behavior: A neurophysiological analysis. New
York: Norton, 1951.
GALAMBOS, R., NORTON, T. T., AND FROMMER, C. P. Optic tract lesions sparing pattern
vision in cats. Experimental Neurology, 1967,18, 8-25.
GARNER, W. R. Uncertainty and structure as psychological concepts. New York: Wiley, 1962.
HEARST, E., AND PRIBRAM, K. H. Appetitive and aversive generalization gradients in
amygdalectomized monkeys. Journal of Comparative Physiological Psychology, 1964a,
58, 296-298.
HEARST, E., AND PRIBRAM, K. H. Facilitation of avoidance behavior by unavoidable
shocks in normal and amygdalectomized monkeys. Psychological Reports, 1964b, 14,
HUSSERL, E. Logische Untersuchungen (4th ed.). Halle, 1928.
JAMES, W. Principles of psychology (Vols. 1 and 2 complete). New York: Dover, 1950.
JOHANSSON, G. Visual perception of biological motion and a model for its analysis.
Perception and Psychophysics, 1973,14(2), 201-211.
KAADA, B. R. Somato-motor, autonomic and electro-corticographic responses to electrical
stimulation of "rhinencephalic" and other structures in primates, cat and dog. Acta
Physiologica Scandinavica, 1951, 23(Suppl. 83), 285.
KAADA, B. R., PRIBRAM, K. H., AND EpSTEIN, J. A. Respiratory and vascular responses in
monkeys from temporal pole, insula, orbital surface and singulate gyrus: A prelimi-
nary report. Journal of Neurophysiology, 1949,12, 347-356.
KAHNEMAN, D. Attention and effort. Englewood Cliffs, N.J: Prentice-Hall, 1973,
KIMBLE, D. P., BAGSHAW, M. H., AND PRIBRAM, K. H. The GSR of monkeys during

orienting and habituation after selective partial ablations of the cingulate and frontal
cortex. Neuropsychologia, 1965,3, 121-128.
KLUvER, H. Behavior mechanisms in monkeys. Chicago: University of Chicago Press, 1933.
KLUvER, H., AND BuCY, P. C. "Psychic blindness" and other symptoms following bilateral
temporal lobectomy in rhesus monkeys. American Journal of Physiology, 1937, 119,
LASHLEY, K. S. Brain mechanisms and intelligence. Chicago: University of Chicago Press,
LINDSAY, P. H. Multichannel processing in perception. In D. I. MOSTOFSKY (Ed.), Atten-
tion: Contemporary theory and analysis. New York: Appleton-Century-Crofts, 1970,
pp. 149-17l.
DAVIS, G. Stimulation and regional ablation of orbital surface of frontal lobe. Research
Publications Association for Research in Nervous and Mental Disease, 1948,27, 405-420.
MACKAY, D. M. Information, mechanism and meaning. Cambridge, Mass.: MIT Press, 1969.
MACLEAN, P. D. Psychosomatic disease and the "visceral brain": Recent developments
bearing on the Papez theory of emotion. Psychosomatic Medicine, 1949, 11, 33~353.
MACLEAN, P. D., AND PRmRAM, K. H. A neuronographic analysis of the medial and
basal cerebral cortex: I. Cat. Journal of Neurophysiology, 1953,16, 312-323.
MALIS, L. I., PRIBRAM, K. H., AND KRUGER, L. Action potentials in "motor" cortex evoked
by peripheral nerve stimulation. Journal of Neurophysiology, 1953,16, 161-167.
MANDLER, G. The conditions for emotional behavior. In D. C. GLASS (Ed.), Neurophysiol-
ogy and emotion. New York: Rockefeller University Press and Russell Sage Founda-
tion, 1967, pp. 96-102.
MARSHALL, J. F., AND TEITELBAUM, P. Further analysis of sensory inattention following
lateral hypothalamic damage in rats. Journal of Comparative and Physiological Psychol-
ogy, 1974,86(3), 375-395.
McFARLAND, D. J. Feedback mechanisms in animal behavior. New York: Academic Press,
METTLER, A., AND McLARDY, T. Posterior cuts in prefrontal leucotomy: A clinical-
pathological study. Journal Mental Science, 1948,94, 555-564.
MILLER, G. A. The magical number seven, plus or minus two, or, some limits on our
capacity for processing information. Psychological Review, 1956,63(2),81-97.
MILLER, G. A., GALANTER, E. H., AND PRIBRAM, K. H. Plans and the structure of behavior.
New York: Holt, 1960.
MILLER, N. E., BAILEY, C. J., AND STEVENSON, J. A. Decreased "hunger" but increased
food intake resulting from hypothalamic lesions: Science, 1950, 112, 256-259.
MISHKIN, M., AND PRIBRAM, K. H. Visual discrimination performance following partial
ablations of the temporal lobe: I. Ventral vs. lateral. Journal of Comparative Physiologi-
cal Psychology, 1954,47, 14-20.
MISHKIN, M., AND PRmRAM, K. H. Analysis of the effects of frontal lesions in monkey: I.
Variations of delayed alternation. Journal of Comparative and Physiological Psychology,
1955,48, 492-495.
MITTELSTAEDT, H. Discussion. In D. P. KIMBLE (Ed.), Experience and capacity. New York:
The New York Academy of Sciences, Interdisciplinary Communications Program,
1968, pp. 46-49.
NEISSER, U. Cognitive psychology. New York: Appleton-Century-Crofts, 1967.
ORNSTEIN, R. E. The psychology of consciousness. San Francisco: Freeman, 1972.
ORNSTEIN, R. E. The nature of human consciousness: A book of readings. San Francisco:
Freeman, 1973.

PATTEE, H. H. Physical theories of biological coordination. Quarterly Review of Biophysics,

1971,4,2, & 3, 255-276.
PRIBRAM, K. H. Toward a science of neurophysiology: (Method and data). In R. A.
PATTON (Ed.), Current trends in psychology and the behavioral sciences. Pittsburgh: The
University of Pittsburgh Press, 1954, pp. 115-142.
PRIBRAM, K. H. Comparative neurology and the evolution of behavior. In A. ROE AND G.
G. SrMPSON (Eds.), Behavior and evolution. New Haven, Conn: Yale University Press,
1958, pp. 140-164.
PRIBRAM, K. H. A review of theory in physiological psychology. In Annual Review of
Psychology (Vol. 11). Palo Alto, Calif.: Annual Reviews, Inc., 1960, pp. 1-40.
PRIBRAM, K. H. Emotion: Steps toward a neuropsychological theory. In D. C. GLASS Ed.),
Neurophysiology and emotion. New York: Rockefeller University Press and the Russell
Sage Foundation, 1967a, pp. 3-39.
PRIBRAM, K. H. Memory and the organization of attention. In D. B. LINDSLEY AND A. A.
LUMSDAINE (Eds.), Brain Function (Vol. 4). Berkeley: University of California Press,
1967b, pp. 79-122.
PRIBRAM, K. H. The new neurology and the biology of emotion. American Psychologist,
1967c, 10, 830-838.
PRIBRAM, K. H. The amnestic syndromes: Disturbances in coding? In G. A. TALLAND AND
N. C. WAUGH (Eds.), Pathology of memory. New York: Academic Press, 1969, pp. 127-
PRIBRAM, K. H. Languages of the brain: Experimental paradoxes and principles in neuropsy-
chology. Englewood Cliffs, N.J.: Prentice-Hall, 1971.
PRIBRAM, K. H. How is it that sensing so much we can do so little? In F. O. SCHMITT AND
F. G. WORDEN (Eds.), The Neurosciences Study Program, III. Cambridge, Mass.: MIT
Press, 1974, pp. 249-261.
PRIBRAM, K. H. The isocortex. In D. A. HAMBURG AND H. K. H. BRODIE (Eds.), American
handbook of psychiatry (Vol. 6). New York: Basic Books, 1975, pp. 107-129.
PRIBRAM, K. H. Holonomy and structure in the organization of perception. In Proceed-
ings of the Conference on Images, Perception and Knowledge, University of Western
Ontario, 1976a.
PRIBRAM, K. H. Problems concerning the structure of consciousness. In G. GLOBUS, G.
MAXWELL AND 1. SAVODNIK (Eds.), Consciousness and the Brain. New York: Plenum
Press, 1976b.
PRIBRAM, K. H., AND BAGSHAW, M. H. Further analysis of the temporal lobe syndrome
utilizing frontotemporal ablations in monkeys. Journal of Comparative Neurology,
1953, 99, 347-375.
PRIBRAM, K. H., KRUGER, L., ROBINSON, R., AND BERMAN, A. J. The effects of precentral
lesions on the behavior of monkeys. Yale Journal of Biology and Medicine, 1955-56,28,
PRIBRAM, K. H., LENNOX, M. A., AND DUNSMORE, R. H. Some connections of the orbito-
fronto-temporal, limbic and hippocampal areas of Macaca mulatta. Journal of Neuro-
physiology, 1950,13, 127-135.
PRIBRAM, K. H., LIM, H., POPPEN, R., AND BAGSHAW, M. H. Limbic lesions and the
temporal structure of redundancy. Journal of Comparative and Physiological Psychol-
ogy, 1966,61, 365-373.
PRIBRAM, K. H., AND MAcLEAN, P. D. A neuronographic analysis of the medial and basal
cerebral cortex: II. Monkey. Journal of Neurophysiology, 1953, 16, 324-340.
PRIBRAM, K. H., AND MCGUINNESS, D. Arousal, activation and effort in the control of
attention. Psychological Review, 1975,82,116-149.

PRIBRAM, K. H., AND TUBBS, W. E. Short-term memory, parsing, and the primate frontal
cortex. Science, 1967,156, 1765-1767.
PRIBRAM, K. H., AND WEISKRANIZ, 1. A comparison of the effects of medial and lateral
cerebral resections an conditioned avoidance behavior of monkeys. Journal of Com-
parative and Physiological Psychology, 1957,50, 74-80.
ROBERTS, W. W. Are hypothalamic motivational mechanisms functionally and anatomi-
cally specific? Brain, Behavior and Evolution, 1969,2, 317-342.
ROSVOLD, H. E., MIRSKY, A. F., AND PRIBRAM, K. H. Influence of amygdalectomy on social
interaction in a monkey group. Journal of Comparative and Physiological Psychology,
1954, 47, 173--178.
RucH, T. c., AND SHENKIN, H. A. The relation of area 13 on orbital surface of frontal lobes
to hyperactivity and hyperphagia in monkeys. Journal of Neurophysiology, 1943, 6,
SCHACHTER, S. Cognitive effects on bodily functioning: Studies of obesity and eating. In
D. GLASS (Ed.), Neurophysiology and emotion. New York: Rockefeller University Press
and Russell Sage Foundation, 1967, pp. 117-144.
SCHWAR1ZBAUM, J. S. Changes in reinforcing properties of stimuli following ablation of
the amygdaloid complex in monkeys. Journal of Comparative Physiological Psychology,
1960a, 53, 388--395.
SCHWAR1ZBAUM, J. S. Response to changes in reinforcing conditions of bar-pressing after
ablation of the amygdaloid complex in monkeys. Psychological Reports, 1960b, 6,215-
SCHWAR1ZBAUM, J. S. Some characteristics of amygdaloid hyperphagia in monkeys.
American Journal of Psychology, 1961,74, 252-259.
SCHWAR1ZBAUM, J. S., AND PRIBRAM, K. H. The effects of amygdalectomy in monkeys on
transposition along a brightness continuum. Journal of Comparative and Physiological
Psychology, 1960,53, 396-399.
amygdalectomy on locomotor activity in monkeys. Journal of Comparative and Physiol-
ogical Psychology, 1961, 54(3), 334-336.
SIMON, H. A. How big is a chunk? Science, 1974,183, 482-488.
SPINELLI, D. N., AND PRIBRAM, K. H. Changes in visual recovery functions produced by
temporal lobe stimulation in monkeys. Electroencephalography and Clinical Neuro-
physiology, 1966,20,44-49.
SPINELLI, D. N., AND PRIBRAM, K. H. Changes in visual recovery functions and unit
activity produced by frontal and temporal cortex stimulation. Electroencephalography
and Clinical Neurophysiology, 1967,22, 143--149.
TEITELBAUM, P. Sensory control of hypothalamic hyperphagia. Journal of Comparative and
Physiological Psychology, 1955,48, 156-163.
UNGERSTEDT, U. Brain dopamine neurons and behavior. In F. G. SCHMITT AND F. O.
WORDER (Eds.), The Neurosciences Study Program, III. Cambridge, Mass.: MIT Press,
1974, pp. 695-704.
VALENSTEIN, E. S. Stability and plasticity of motivation systems. In G. C. QUARTON, T.
MELNECHUK, AND G. ADELMAN (Eds.), The neurosciences. New York: Rockefeller
University Press, 1970, pp. 207-217.
VALENSTEIN, E. S., Cox, V. c., AND KAKOLEWSKI, J. W. The hypothalamus and motivated
behavior. In. J. T. TAPP (Ed.), Reinforcement and behavior. New York: Academic Press,
1969, pp. 242-285.
VON FORESTER, H. Memory without record. In D. P. KIMBLE (Ed.), The anatomy of memory.
Palo Alto, CaU.: Science and Behavior Books, 1965, pp. 388-433.

WALL, P. D., AND PRmRAM, K. H., Trigeminal neurotomy and blood pressure responses
from stimulation of lateral cerebral cortex of Macaca mulatta. Journal of Neurophysiol-
ogy, 1950,13, 409--412.
WARD, A. A., JR. The cingular gyrus: Area 24. Journal of Neurophysiology, 1948,11, 13-23.
WILSON, M. Inferotemporal cortex and the processing of visual information in monkeys.
Neuropsychologia, 1968, 6, 135-140.
3 Self-Regulation of
Stimulus Intensity:
Augmenting/Reducing and the
Average Evoked Response



Clinical practice in an emergency room quickly dramatizes individual

differences in pain tolerance. I remember a Swedish carpenter who,
declining analgesia, stoically allowed me to dig out a splinter from
under his fingernail with a scalpel as he gaily discussed baseball. Holy
men rest on their beds of nails; Lesch-Nyhan patients mutilate them-
selves; rock groups blast listeners with sound above normal auditory
pain threshold-all of which raises the question, how do combinations
of experimental and neurophysiological mechanisms work together to
produce these variations in tolerance of extreme intensities of sensory
input? This article will review individual difference research in sensory
overload with especial interest in the possible electrophysiological cor-
relates of these differences.


A. Petrie and Kinesthetic Figural Aftereffects

Petrie (1960), following studies of individual differences in pain

tolerance, classified subjects into "augmenters" and "reducers" based
on their performance on a kinesthetic figural aftereffects task (KF A).
Augmenters were those individuals who typically judged the magni-

MONTE BUCHSBAUM . Unit on Perceptual and Cognitive Studies, Adult Psychiatry

Branch, Division of Clinical and Behavioral Research, Intramural Research Program,
National Institute of Mental Health, U.S. Department of Health, Education and Welfare,
Bethesda, Maryland.


tude of a standard stimulus as larger after kinesthetic stimulation;

reducers judged the standard as markedly reduced. Augmenters were
found to evidence a special sensitivity to pain. In contrast, reducers had
a notable tolerance for pain (Petrie, Collins, and Soloman, 1958; Petrie,
1960). Several studies have tended to support this concept (Blitz, Din-
nerstein, and Lowenthal, 1966; Sweeney, 1966; Ryan and Foster, 1967).
Petrie (1960) also suggested that if the tolerance of reducers for pain is
partially due to their tendency to diminish the intensity of stimulation
they receive, they should be at a disadvantage in sensory deprivation,
where they would further reduce the already limited stimulation avail-
able. She found that augmenters could tolerate more hours in a tank-
type respirator than reducers, who could not tolerate this stress. Aug-
menters could be pushed in the reducing direction by bombardment
with continuous loud noise (Petrie, Holland, and Wolk, 1963). Since
auditory stimulation could cause kinesthetic augmentation, she con-
cluded that some central nervous system (CNS) mechanism regulated
levels of sensory input.
Petrie's KFA procedure was adapted from an apparatus used by
Kohler and Dinnerstein (1947) to study effects of what they termed
satiation-the tendency for persons to judge a stimulus as less intense
after they had been stimulated by a more intense stimulus. Petrie
compared size judgments of the width of a standard block of wood
before and after subjects stimulated their fingers by rubbing either a
wider or a narrower block of wood. Finding that the reducer would tend
to judge the standard block of wood as smaller not only after rubbing
the wide block of wood but after rubbing a narrow block of wood as
well, she concluded that satiation theory alone could not explain the
observed patterns of individual differences, and the augmenter/reducer
construct was developed. Reducers thus were defined as consistently
underestimating the size of the standard after stimulating their hands
with any size block, and augmenters were defined as consistently
overestimating the size of the standard.
Careful consideration of the Petrie task from today's vantage point
of research on individual differences reveals a potential both for unreli-
ability and for measuring several types of individual differences at
once. First, initial size or width judgments themselves, before stimula-
tion, usually show individual tendencies to over- or underestimate.
These size-estimation individual differences show moderate reliability
across sessions (r = 0.76, Platt, Holzman, and Larson, 1971), but differ-

ences between values of such reliability will have much lower reliability
(Platt et al., 1971). Second, individual differences in contrast effects
(judging the standard as smaller after rubbing the large block and larger
after rubbing the small block) are probably much more prominent than
Petrie credited (1967). Hilgard, Morgan, and Prytulak (1968) found
large-group contrast effects, not augmenting/reducing, predominating
in their sample. Broadhurst and Millard (1969) and Schooler and Silver-
man(1971) found nonsignificant correlations between large- and small-
block augmenting/reducing scores. Since individuals with large con-
trast effects would show negative correlations and individuals with
small contrast effects would show positive ones, these negligible corre-
lations are expected. In our own studies, individual consistencies in
large-blocklsmall-block contrast effects and in ascending/descending
starting-position effects were much more reliable than the augmenting/
reducing (prestimulation minus poststimulation) measure. Silverman's
(1964) KFA procedure was similar to the Petrie in this respect. Contrast
effects also seem to operate across days (Hilgard et al., 1968), further
interfering with large-blocklsmall-block differences. While it might be
argued that contrast effects are balanced out from the augmenting/
reducing score by the opposite directions of contrast in the use of the
small and large blocks, the assumptions that contrast effects are entirely
linear, equivalent in both size directions, and independent of augment-
ing/reducing are probably unjustified.
Thus since size estimation and contrast effects were tapped as well
as the hypothetical augmenting/reducing dimension, the Petrie proce-
dure might give very different scores or reliabilities across patient
groups or experimental treatment depending upon the relationships of
these perceptual styles. These problems may help to explain the failure
of Morgan, Lezard, Prytulak, and Hilgard (1970) to confirm the pain
tolerance of reducers and the failure of Peters, Benjamin, Helvey,
and Albright (1963) to confirm the sensory deprivation tolerance of

B. Evoked Responses and Augmenting/Reducing

The concept of augmenting/reducing still remained as an intriguing

source of predictions about individual differences in neurophysiologi-
cal data on sensory response to stimuli at different intensities. Average

FIGURE 1. Average evoked responses (AERs) to four intensities of flash: top to bottom,
dim to bright. The individual on the left shows "augmenting"-increasing amplitude
with increasing stimulus intensity for component PI (PI00)-Nl (N140). The individual on
the right shows "reducing"-decreasing amplitude with increasing stimulus intensity for
PI-Nl (from Buchsbaum and Pfefferbaum, 1971).

evoked response (AER) amplitudes showed striking differences in their

behavior at high intensities, certain individuals showing marked drops
as intensity increased (Figure 1). Was this AER reducing in any way
analogous to Petrie's concept of a CNS mechanism regulating sensory
input? Early studies tended to show correlations between AER intensity
effects and KFA (Buchsbaum and Silverman, 1968; Spilker and Calla-
way, 1969; Borge, 1973), and AER measures of augmenting/reducing
have been used by Soskis and Shagass (1974), Hall, Rappaport, Hop-
kins, and Griffin (1973b), and Zuckerman, Murtaugh, and Siegel (1974).
Studies in cats (Hall, Rappaport, Hopkins, Griffin, and Silverman,
1970) and monkeys (Redmond, Borge, Buchsbaum, and Maas, 1975)
have been possible using AER techniques. Parallels between Petrie's
KFA results and AER results also appeared. AER reducers were more
tolerant of electric shock (see Section VI.A). Women were more aug-
menting than men (Petrie, 1960; Buchsbaum and Pfefferbaum, 1971).

Schizophrenics were reducers (Petrie et a/., 1963; Buchsbaum and Sil-

verman, 1968; Landau, Buchsbaum, Carpenter, Strauss, and Sachs,
1975). These and other studies detailed in the following sections sug-
gest some predictive utility in the augmenter/reducer dimension. Re-
cently, even Petrie herself (1974) has hailed the use of AER techniques
for measuring augmenting/reducing. We now tum to focus on the AER
as a possible indicator of the profound differences in sensory response
and its regulation.


A. Visual AERs

The amplitude of the visual AER tends to increase with increasing

stimulus intensity up to some intermediate intensity. Further increases
in stimulus intensity may yield no further increase in amplitude or
actual decreases in amplitude. This tendency for amplitude of visual
evoked potentials to saturate or reduce was reported as early as 1937 by
Cruikshank in a subject whose responses were big enough to observe
without averaging. In one of the first systematic studies of the proper-
ties of visual AERs, Tepas and Armington (1962) found that "maximum
responses were not obtained with the strongest stimuli" for large
stimulus fields. Similarly, reducing was observed by many authors
(Tepas, Armington, and Kropfl, 1962; Rietveld, 1963; Wicke, Donchin,
and Lindsley, 1964; Rietveld and Tordoir, 1965; Vaughan and Hull,
1965; Vaughan, 1966; Buchsbaum and Silverman, 1968) and in the AERs
of some authors themselves (Shipley, Jones, and Fry, 1966). Prominent
individual differences were noted by Rietveld (1963) and first systemat-
ically studied by Shagass, Schwartz, and Krishnamoorti (1965) who
found higher amplitude/intensity slopes in psychiatric patients.
Decreases in amplitude with increasing stimulus intensity appear
to be related to central factors rather than merely reflecting poor stimu-
lus control or peripheral adjustment mechanisms. Since subjects can
show quite linear amplitude increases in the electroretinogram while
showing reducing in the AER (Armington, 1964a, 1964b), we know that
the intensity information can register at the retina and still yield nonlin-
ear AER amplitude/intensity functions. Similarly, DeVoe, Ripps, and
Vaughan (1968) reported linear latency decreases for PlOD (PI), together

with reducing in the amplitude of the P100-N140 (P1-N2) component,

and other authors have also noted this latency/amplitude difference
(e.g., Wooten, 1972; Clynes, Kohn, and Lifshitz, 1964; Vaughan and
Hull, 1965; Figure 6, P100 (P1)-N140 (N2) in Vaughan, 1966).
Peripheral factors such as pupillary diameter do not seem to ex-
plain reducing since the vertex AER is uninfluenced by pupillary diam-
eter in humans (Kooi and Bagchi, 1964; Richey, Kooi, and Waggoner,
1966). Visual AER-reducing is unchanged by pilocarpine instilled in the
eye to fix pupillary diameter (Soskis and Shagass, 1974), and in animals
such as the pigeon, no AER changes with pupillary diameter were
observed (Samson and Young, 1973). Studies in our laboratory of
abeltalipoproteinemia, which causes retinal dysfunction, also failed to
show vertex AER changes either on baseline or on treatment with
vitamin A. Maneuvers such as averting or covering the eyes do not
seem to be involved. The finding of AER reducing from vertex but not
from occipital leads and in PlOD more than in P200 (Buchsbaum and
Pfefferbaum, 1971), as well as the high correlation between amplitude/
intensity slopes obtained with randomized and blocked presentation of
stimulus intensities (Buchsbaum, Landau, Murphy, and Goodwin,
1973), all tend to mitigate against the influence of such peripheral
factors. Even under the rigid stimulus control of flaxedil administered
to cats wearing corneal contact lenses with artificial pupils, single-unit
recordings from the lateral geniculate showed some units decreasing
with increasing stimulus intensity (Hamasaki and Winters, 1973).
AER differences between augmenters and reducers do not appear
to result from differences in habituation rates or interstimulus interval
effects. Reducers do not habituate their AERs faster to high-intensity
stimuli; if anything they habituate more slowly. Reducing the intersti-
mulus interval also does not diminish augmenter/reducer differences
(Buchsbaum and Pfefferbaum, 1971).
As with discrete stimuli, when rapid sinusoidal modulation of
light is used, individual differences in amplitude/intensity or ampli-
tude/depth of modulation relationships are prominent (Kitajima, 1967;
Pfefferbaum and Buchsbaum, 1971). The cortical response may decrease
as the intensity of stimulation increases (Il'yanok, 1961; van der Tweel
and Verduyn Lunel, 1965; Montagu, 1967; Regan and Beverley, 1973).
This reducing does not seem to be due to poor stimulus control. Regan
and Beverely (1973) presented their stimuli in a maxwellian-view opti-

cal system and still found saturation for first harmonic components and
marked reducing for second harmonic components.

B. Auditory AERs

As was found for visual AERs, auditory AERs generally increase in

amplitude with increasing stimulus intensity (e.g., Abe, 1953; Keidel
and Spreng, 1965; Suzuki and Taguchi, 1965). There are, however, great
individual differences in the rate of increase (Davis, Mast, Yoshie, and
Zerlin, 1966; Rapin, Schimmel, Tourk, Krasnegor, and Pollak, 1966;
Davis and Zerlin, 1966; Davis, Bowers, and Hirsh, 1968; Rothman,
1970). Many investigators have reported that some individuals show a
maximum response at around 70-75 dB with further increase in stimu-
lus intensity causing a decrease in AER amplitude (Davis and Zerlin,
1966; Beagley and Knight, 1967; Moore and Rose, 1969; Picton, Good-
. man, and Bryce, 1970; Kollar, 1971; Marco, 1972; Khechinashvili, Kev-
anishvili, and Kajaia, 1973) or a leveling in amplitude (Butler, Keidel,
and Spreng, 1969). Rose and Ruhm (1966) even found some amplitude
decreases at levels as low as 40 dB. Picton et al. (1970) recorded stape-
dius muscle reflexes and concluded that this paradoxical reduction in
AER amplitude did not result from peripheral factors but that the most
probable explanation seemed to involve either the ipsilateral cochlear
efferent, the uncrossed efferent olivocochlear, or the central descending
auditory inhibitory system. Khechinashvili et al. (1973) also argued that
middle-ear muscle contraction was not important in AER amplitude
decrease at high intensities since the decrease became most marked at
higher tone frequencies (e.g., 4000 Hz), whereas contractions of the
stapedius muscle predominantly affect the conduction of lower fre-
quencies. The absence of any significant difference in amplitude/inten-
sity slope between serial and randomly presented tone intensities
(Henry and Teas, 1968) also argues against a peripheral mechanism.
Reducing appears with the good stimulus control available with ear-
phones (e.g., Khechinashvili et al., 1973) or bone conduction (Liebman
and Graham, 1967). Even in anesthetized cats, single units in the
cochlear nucleus were found where the "greatest excitation was pro-
vided by tones of moderate intensity and further increases in intensity
resulted in a marked decline in the response" (Lavine, 1971).

Stage 3
z 2
~ Stage 4
Z Awake
:!: Rem
" . ; : Stage 2
Stage 1

50 60 70 80

FIGURE 2. Mean amplitude (deviatiori in microvolts) for P200
(168-248 ms) for four click intensities for awake, REM sleep, and
Stages 1-4.

The great decrease of reducing that we observed (Buchsbaum,

Gillin, and Pfefferbaum, 1975) during Stages 3 and 4 sleep is of interest
because of the well-known loss of inhibitory mechanisms during deep
sleep. Click AERs to four intensities were recorded during all-night
sleep recordings. AERs collected over I-minute epochs were sorted by
sleep stage in nine normal subjects. Figure 2 illustrates that awake,
Stage 1, REM, and Stage 2 showed little increase in AER amplitude with
increasing stimulus intensity, whereas Stages 3 and 4 showed a marked
Tepas, Boxerman, and Anch (1972) suggested that the "failure" of
some studies to produce increasing linear relationships between AER
amplitude and stimulus intensity results from the failure to control
stimulus conditions and parameters and not from individual differ-
ences. However, even in their own report of carefully collected data for
the B-C component for bipolar lead Fz-Pz (similar latency to P100-
N140) one of their three subjects showed a correlation between ampli-
tude and stimulus intensity of + .87 (augmenting) and one of -.64

(reducing). They further suggested that lead Cz-Oz produces the most
linear amplitude/intensity functions, but this was not borne out in
another report from the same laboratory (Klingaman and Anch, 1972).
Data pooled across three subjects showed a 15% drop in AER B-C
amplitude (about P80-N110) as stimulus intensity increased from 75 to
85 dB. Similarly, in a recent report (Schweitzer and Tepas, 1974), two of
the three subjects showed trivial (.21 and .10) correlations between
stimulus intensity and AER amplitude for component B-C, for lead

C. Somatosensory AERs

As with visual and auditory AERs, reducing or saturation has been

observed in both human and animal recordings. Since no mechanical
peripheral adjustment exists for the somatosensory system analogous to
the pupil or the stapedius muscle, and since electrical stimuli are more
easily delivered with great precision, the presence of reducing should
provide strong evidence for central inhibitory processes. Mark and
Steiner (1958) recorded from cat somatosensory receiving areas and the
superficial radial nerve: they found that the cortical response reached a
ceiling at only 40% of the radial nerve volley and suggested that "some
peculiarity of the complex cortical responding mechanism may be in-
volved." Early human studies by Uttal and Cook (1964) also showed
AER amplitude reaching maximal levels with relatively weak stimuli.
Rosner and Goff (1967) reported that the amplitude/intensity relation-
ship for somatosensory AERs was best fitted by two straight-line func-
tions and that the slope of the second line was nearly zero for certain
components (perhaps analogous to our time bands 76-112 and 168-248
ms). Beck and Rosner (1968) later reanalyzed these data and concluded
that a single-power function with a threshold correction was a more
parsimonious explanation. However, prominent individual differences
in amplitude/intensity slopes, with some subjects showing reducing,
were reported by Uttal and Cook (1964), Shagass and Schwartz (1964),
Debecker and Desmedt (1964), Ikuta (1968), Shagass, Overton, Barto-
lucci, and Straumanis (1971), and Mushin and Levy (1974).
Saturation or reducing phenomena also appear in single-unit re-
cordings in the dorsal hom of the spinal cord (Wall, 1967; Hillman and
Wall, 1969; Lundberg and Oscarsson, 1961), where the response to

higher intensities is apparently reduced by descending inhibitory im-

pulses originating in the brain. Such descending impulses may origi-
nate in the cerebral cortex (Hagbarth and Kerr, 1954; Anderson, Eccles,
and Sears, 1964), the diencephalon (Calma, 1966), or the reticular forma-
tion (Melzack, 1973) and may act upon thalamic relay nuclei as well as
spinal levels (Mountcastle, 1974). Such descending inhibitory mecha-
nisms would be capable of modulating nociceptive afferent input be-
fore it reaches the cortex and are a part of the well-known gate-control
theory of pain advanced by Melzack and Wall (1970).

D. Summary of Amplitude/Intensity Relationships

Unfortunately the plethora of recording and stimulating techniques

in the studies discussed above and the absence of cross-modal studies
within the same individual make a comprehensive summary very diffi-
cult. Nevertheless in a general way for visual, auditory, and somatosen-
sory stimuli, AER amplitudes initially increase with increasing inten-
sity; further increases in intensity often bring about a decrease in
amplitude in many individuals. These decreases seem not to reflect
poor stimulus control, peripheral adjustments, or transducer response,
but rather result from underlying CNS mechanisms. The P100-N140
component frequently reveals these paradoxical intensity effects. Paral-
lel nonlinearities are recordable from single units in the lateral genicu-
late, cochlear nuclei, and dorsal horn of the spinal cord. Three types of
neural pathways may be involved: descending inhibitory, nonspecific
arousal, and cortical-cortical. Descending inhibitory pathways have
been implicated in the somatosensory system and are certainly active in
the auditory system. However, if very early signals from primary
sensory areas do not show reducing but later components do, then
descending inhibitory pathways seem less likely to be the major cause.
With auditory signals there is some evidence that an almost linear
function may exist for a Pll-N21 ms component (Madell and Goldstein,
1972) with far less linearity for later components. Early tactile stimulus
AERs also apear to produce quite log-linear amplitude/intensity func-
tions (Franzen and Offenloch, 1969) but electrical stimuli (Mushin and
Levy, 1974) seem clearly to show reducing. Thus descending inhibitory
systems appear to be involved in somatosensory AER-reducing and are
possibly less prominent in auditory AERs.

Nonspecific arousal systems, including subcortical structures, were

suggested as a possible mechanism in AER augmenting/reducing by
Zuckerman, Murtaugh and Siegel (1974). They pointed out that fibers
originating in the cortex may regulate brainstem arousal systems
(Koella and Ferry, 1963; Dell, Bonvallet, and Hugelin, 1961). These
systems could then modulate AER amplitude in the cortex (Knispel and
Siegel, 1972, 1973).
Cortical-cortical systems as a hypothetical cause of AER reducing
also cannot be ruled out. The many experiments in humans showing
the influence of complex logical tasks on AER amplitude certainly
suggest the operation of such systems. Descending inhibitory, nonspe-
cific arousal, and cortical-cortical systems may work together to modu-
late sensory input levels depending on the vigilance, arousal, and other
needs of the organism.



The amplitude/intensity slope appears to be a relatively stable

individual characteristic. Exactly how reliable it is depends on the
method of AER measurement. Most authors have relied upon the visual
identification of a triphasic wave in the vertex AER to visual stimuli; a
positive peak at 70-110 ms (P100 or P1), a negative peak at 110-160
(N140 or N1), and a positive peak at 180-250 (P200 or P2). The N140-
P200 component in our experience is usually the most unambiguous,
with the P100 often split into two peaks, or appearing as a point of
inflection between some early (30-50 ms) component and N140. Since
we have generally had one or more sets of AERs collected at four
intensities, we have tried to identify P100-N140-P200 as components
most reliably appearing in the largest number of AER curves. The
latency of these components is then used to decide peak identification
for the most ambiguous curves.
Using such a strategy, we examined AER amplitude/intensity slope
reliability across two sessions about 2 weeks apart in various groups.
Test-retest correlations for peak-to-trough measured P100-N240 were
.67 in a group of psychiatriC patients (Buchsbaum, Goodwin, Murphy,
and Borge, 1971) and .52 in a group of 120 normal adult twins (Buchs-
baum, 1974).

Somewhat higher reliabilities (.72-.91) were reported by Stark and

Norton (1974), who, after comparing reliabilities of latency, amplitude,
and amplitude/intensity slope, concluded that slope was "clearly the
most reliable AER parameter employed." Soskis and Shagass (1974)
found within-session reliability of .66 and concluded that eye move-
ment, blinks, or pupillary diameter had little influence on AER aug-
Despite these reasonable reliabilities, however, AER component
identification is troubled by missing or aberrant components, espe-
cially when large groups of subjects are studied and analogous compo-
nents must be identified in four or more AERs within a person. Hall,
Rappaport, Hopkins, and Griffin (1973a) suggested a helpful computer
algorithim for peak identification but were still troubled by peaks
disappearing on one or more of a set of AERs. They noted that a point of
inflection was frequently present at exactly the latency that a peak Was
clearly present for a stimulus of greater or lesser intensity. This suggests
that numerical techniques (not based on visual peak identification)
developed by Shagass (1972, area integration) and Dustman and Beck
(1965, perimeter) might be useful.
For studies of individual differences in perceptual style, a measure
should be reliable over time. Because of our interest in finding possible
genetic markers of vulnerability to psychiatric illness, we also wanted
measures showing similarities across genetically identical individu-
als-i.e., monozygotic twins. A variety of AER measurement tech-
niques were assessed in a group of 64 normal adult twin pairs-32
monozygotic (Mz) and 32 dizygotic (Dz). Each subject was tested on the
four-intensity AER procedure on two separate occasions, generally 2
weeks apart. This consisted of presenting four intensities of light in a
randomized order behind a translucent plastic screen. The stimuli were
generated by fluorescent tubes under computer control and had a rise
time of 3 ms with a duration of 500 ms so that only "on" responses were
studied. AER peaks were identified visually and by use of several
numerical techniques. We calculated the amplitude/intensity slope by
fitting a straight line to the amplitude measures (however obtained)
using least-squares technique. The visually identified peaks yielded
either peak-to-trough measures, P100-N140 and N140-P200, or presti-
mulus baseline-to-peak measures, P100, N140, and P200. For the nu-
merical techniques, we chose three time intervals centered on the three
peak:; identified on visual inspection: P100 (76-112 ms), N140 (116-152
1 z
Amplitude/Intensity Slope Test-Retest and Mz-Twin Pair Correlations o>,j
P100 N140 P200
76-112 ms 116-156 ms 168-248 ms

Test-retest MZ twins Test-retest MZ twins Test-retest MZ twins

Visual inspection"
Peak-to-trough .52 .51 .53 .56
Baseline-to-peak .28 .29 .25 .05 .47 .28
Absolute area relative to mean .63 .63 .37 -.01 .65 .60
Signed area relative to pre stimulus .46 .45 .32 -.10 .58 .55
RMS .61 .52 .36 .62 .65 .53
Perimeter .18 .35 .54 .60 .55 .36
Point at 100, 140, and 200 ms .43 .71 .37 .69 .63 .72

a Peak-la-trough PIOO-N140 and N140-P200.


ms), and P200 (168-248 ms). They correspond fairly well to time inter-
vals chosen by Lewis, Dustman, and Beck (1972), and the critical P100
band is similar to that used by So skis and Shagass (75-150 ms).
Two techniques of areal integration were compared. For the first,
termed "absolute area relative to mean," the mean value of each AER,
used as a baseline, was subtracted from each AER coordinate; within
each time band, the mean of the absolute values of successive coordi-

!z 0.6
t;J 8 0.4
I- ~ 0.2
a:: -p<0.05

; 3
a:: I-
u 0 2
:i ~
~ 0
LU ....J
;; '" 0
~ -I ~-I
:E -2
FIGURE 3. Analysis of AER amplitude and slope for consecutive 4 ms time intervals. The
actual AER value for the highest-intensity AER and the slope for each coordinate (calcu-
lated across four intensities) were obtained for each of two AER sessions on 128 subjects.
Data are shown for 300 ms following the stimulus mean across the 128 subjects (lower),
and test-retest correlations across the two sessions (upper) are presented for amplitude
data (left) and slope data (right). Thus the curve in the lower left is the mean AER for the
group. Circled points are the mean latencies (across intensities) for P100, N140, and P200
for the entire group as determined by visual inspection. Note that P100 and P200 fall on
maxima for test-retest reliability (above) both for raw amplitude (left) and slope (right).
P100 falls on the minimum for the slope function (lower right) and P200 on the maximum;
note that the P100 slope value is actually negative as well as reliable. (Data from
Buchsbaum, 1974.)

nates was calculated. For the second, "signed area relative to prestimu-
Ius baseline," the pre stimulus baseline was subtracted from each AER
coordinate and the mean of the signed values of successive coordinates
was calculated. A root-mean-square measure was also calculated. The
line-length measure was the sum of the absolute values of point-to-
point differences. In addition, the individual AER points (after removal
of AER mean) were used. Table 1 compares the test-retest reliability
and Mz-twin similarity for the seven techniques.
All techniques showed statistically significant test-retest correla-
tions for PIOO and P200 (for r > .16, p < .05, 1 tailed). Positive peaks
PIOO and P200 appear to have more reliable amplitude/intensity slopes
than negative peak N140. Measurements referred to the mean level of
the AER appear to be more reliable than those referred to the prestimu-
Ius baseline. Generally reliability and Mz-twin similarity are in high
agreement. Overall, the absolute area relative to the mean technique
(closest to that of Shagass, 1972) seems to have the highest reliability
and heritability, and we have adopted this technique.
The data from the point-by-point analysis on the total group of 128
subjects indicate that considerable information is carried in the single
AER coordinates (Figure 3). The circled dots are the mean latencies (as
visually identified) for PIOO, N140, and P200. For both amplitude and
slope, the pattern of higher test-retest correlations for positive peaks is
seen. The plot for the mean slope illustrates that the minimum slope on
the entire curve coincides exactly with PIOO--and is actually negative,
indicating that AER values were lower for high-intensity flashes than
for low ones. The high positive peak at P200 is consistent with our
reports and those of others (see Section IILA) that P200 shows far higher
slopes than PIOO.


A. Twin Studies

For the visual AER, amplitude/intensity slopes appear to be par-

tially set genetically as we have seen above and as is illustrated in
Figure 4. Heritability estimates for the AER amplitude/intensity slope
on a group of 30 Mz and 30 Dz normal adult twins ranged from .52 for
the PIOO--Nl40 component measured visually to .68 for certain quantita-


FIGURE 4. Visual AERs in two pairs of Mz twins. Illustrated are AERs to four intensities
of light, as in Figure 1. For each pair, the AER is shown as a solid line for one twin and
as a dotted line for the co-twin. Note similarity in latency of peaks, waveform, and
changes with stimulus intensity. For Pair 1 on the left, component P100-N140 increases
markedly with stimulus intensity (augmenting) and for Pair 2 the component decreases
with stimulus intensity (reducing) (Buchsbaum, 1974).

tive measures (Buchsbaum, 1974). These heritabilities reflect relatively

high intraclass correlations in Mz twins and often absent or negligible
correlations in Dz twins. Other investigators studying amplitude meas-
ure in the visual AER have reported similar results (Lewis, Dustman,
and Beck, 1972). Osborne (1970) also found very low (.08 to .12) intra-
class correlations in Dz twins for certain AER measures, together with
much higher correlations in Mz twins. Similarly for EEG frequency
spectra, Lykken, Tellegen, and Thorkelson (1974) found intraclass corre-
lations of about .8 for Mz twins and near zero for Dz twins. These
findings tend to suggest that the augmenting/reducing similarities be-
tween Mz twins are not primarily due to additive genetic or dominance
effects since such effects would create similarities between Dz twins as

well. Epistasis or dominance interactions involving several loci or dis-

assortative mating could be candidates for such effects. Alternatively,
common environmental factors might act to increase similarity in Mz
twins and decrease similarity in Dz twins (Lykken et al., 1974).

B. Sex and Chromosome Differences

We have generally found that women have greater amplitude/
intensity slopes for P100-N140 than men for visual AER (Buchsbaum
and Pfefferbaum, 1971). Our normal twin sample also showed signifi-
cantly greater amplitude intensity slopes in females than in males for
P100-N140i no difference was found for N140-P200. Patients with
chromatin-negative gonadal dysgenesis (Turner's syndrome, XO)
showed even greater amplitude/intensity slopes than XX normal fe-
males (Buchsbaum, Henkin, and Christiansen, 1974). Again the sex-
difference effects appeared for P100-N140, not N140-P200.


Petrie's original conception that reduction was an adaptive, protec-

tive mechanism for withstanding sensory overload was only weakly
borne out by studies using the KFA test, as we have earlier seen.
However, two current AER studies suggest that AER reducers are
indeed pain- and noise-tolerant.

A. Pain Tolerance

Individual differences in pain response were studied by the use of

brief electrical stimuli (constant current 1 ms square waves) adminis-
tered to the dorsum of the left forearm with the Tursky electrode
(Tursky and Watson, 1964).
Stimuli of four different intensities (2, 9, 16, and 23 mA) were
delivered 1 s apart with 64 randomized presentations of each intensity,
and AERs to these stimuli were collected. Subjective pain ratings were
also obtained in a separate run in which stimuli (1 to 31 mA in 1-mA
steps) were presented three times each in random sequence. Subjects

rated each stimulus on a 4-point scale (1 = just noticeable, 2 = distinct,

3 = unpleasant and 4 = painful) using a special keyboard. EEG was
recorded from vertex (Cz, right ear), and AER amplitude was
measured using the "absolute area relative to mean" technique (see
Section IV) over two time bands centered on PlOD (76-112 ms) and
P200 (168-248 ms).
Shock-response rating data were analyzed according to simple
nonparametric analogues of signal detection parameters Cx and d'.
Since subjects differed widely in the shape, symmetry, and size of the
dispersion of their judgments of the four categories, the usual mathe-
matical assumptions underlying signal detection analysis did not seem
well met. A measure of response criterion (Cx) was obtained for the
distinct/unpleasant categories by location of the stimulus strength for
which the least overlap between the categories occurred. A measure of
sensitivity (d') was the percentage of total responses in error by use of
the above criterion. Mean ratings for the 6-12, 13-19, and 20-26 rnA
ranges were also computed, each value being based on 21 judgments.
Figure 5 shows the somatosensory evoked-response amplitudes of nor-
mal college students divided into pain-tolerant (n = 18) and pain-
intolerant (n = 18) groups on the basis of their being above or below the
median on their mean rating for the 20-26 rnA range stimuli. As
anticipated, individuals with reducing or relatively low amplitude/
intensity slopes were pain-tolerant, and augmenters were pain-intoler-
ant. These slope differences were confirmed statistically by use of a
two-way analysis of variance (p < .05, group by intensity interaction).
Individual criterion measures (Cx) for the distinct/unpleasant division
were correlated -.48 with the amplitude/intensity slope for PlOD. The d'
measure was not significantly correlated. The criterion measure has
been previously linked to placebo and suggestion effects (Clark and
Goodman, 1974), and our neurophysiological correlates may be related
to these phenomena.
This linkage between placebo/suggestion effects and PlOD ampli-
tude/intensity slope was further borne out in studies of audioanalgesia
(Lavine, Buchsbaum, and Poney, in press). Subjects who heard music
with the suggestion that it would reduce pain had lower mean slopes of
their somatosensory/intensity function than subjects who were given
neither music nor suggestion. This effect was most prominent for PlOD
at vertex and over the right postcentral gyrus.

76-112 MS 168-248 MS
0 6 PAIN
0 ,.. ~

~ 2 .- ~
.- / PAIN
~~~~~---. TOLERANT
:>w ,-~~...,


2 9 16 23 2 9 16 23
FIGURE 5. Mean somatosensory AER amplitude for four intensities of electric shock for
the 76-112 ms and 168-248 ms time bands (generally equivalent to PIOO and P200).
Subjects were divided into pain-tolerant and -intolerant groups on the basis of their
subjective ratings of shock unpleasantness. Note that the pain-tolerant subjects are
relative reducers--have lower rates of increase in AER amplitude with increasing stimu-
lus intensity. Group differences are greatest at the highest intensities.

B. Noise Tolerance

As with the pain experiment, we expected reducers to be noise-

tolerant. We assessed noise tolerance by measuring the rate at which
subjects pressed a key to decrease noise while using a teaching machine
(Molino, 1974).
In a separate session, auditory AERs were collected in a manner
similar to that used for somatosensory AERs in the pain tolerance
study; i.e., random presentation of four intensities of noise bursts.
Individuals showed wide variation in tolerated noise-56-112 dB.
Individuals who were relatively tolerant of noise on the key-pressing
task were reducers on the auditory AER (Figure 5), whereas noise-
intolerant individuals were augmenters. Again, this was statistically
confirmed by two-way analysis of variance on the two groups and by
correlations between the noise tolerance score (in dB) and the slope of
the AER amplitude/intensity function for PlOD. These effects were most

evident for the highest-octave-frequency band of noise used (8000 Hz),

which is of interest in view of the findings of Khechinashvili et al.
(1973), who found reducing greatest for high-frequency tones.



Since attention and/or arousal is known to affect AER amplitude,

could individual differences in augmenting/reducing be explained as
differences in these factors? Clearly this would require a differential
effect with stimulus intensity; otherwise the AER slope would be
unaffected. Difficulties with designing experimental conditions which
really distinguish "arousal" and "attention" also make answering this
question problematical. Avoiding getting too aroused about these
global definitions and attending only to the precise experimental condi-
tions used may be helpful at this point.

A. AER Decrement over Sessions

Overall AER amplitude diminution across time or with reduced

"arousal" has been widely reported (e.g., Eason, Aiken, White, and
Lichtenstein, 1964; Eason, Harter, and White, 1969; Roth, 1973; Hartley,
1970; Landau and Buchsbaum, 1973). Our sample of 128 adult twins had
two successive AER sessions in our laboratory about 2 weeks apart.
They were initially naIve to EEG recording procedures and received no
instructions other than to look forward at the visual stimuli-four
intensities of light flashes presented in random order. From the initial
to the second session, AER amplitude dropped fairly evenly across all
intensities (Figure 6, left). This could reflect diminished "arousal"
during the second session in our intimidating laboratory-or dimin-
ished attention to the visual stimuli.

B. AER Decrement with Mental Arithmetic

Experiments in which attentional factors were manipulated within

counterbalanced sessions yielded somewhat different results, however.
8.0 3.5
~ oz
Light intensity o."
judgement til
7.0 3.0 No tone


6.0 2.5 ~
>< Tone

2.0 L'_ - - ' -_ _ _

--0._ _
2 3 4 4
FIGURE 6. Effect of experimental conditions on visual AER amplitudelintensity slopes. (Left) AERs from 128 adult twin subjects tested in
two sessions about 2 weeks apart (described in Buchsbaum, 1974). The major effect is a decrease in AER amplitude from the first to the
second session, about equal at all intensities. (Middle) AERs from 24 normal subjects (redrawn from Schechter and Buchsbaum, 1973)
while making judgments of light intensity and while doing mental arithmetic-the major effect of the mental arithmetic distraction is a
decrease in AER at low intensity. (Right) Visual AERs from 40 normal subjects (see text) while exposed to a 95-dB continuous auditory
tone and compared with no tone. Major effect is a high-intensity AER-amplitude decrease.

In another experiment (Schechter and Buchsbaum, 1973) using the

same stimuli and recording techniques, subjects either "attended"
(counted the appearance of specific-intensity pairs in the sequence) or
were "distracted" (performed mental arithmetic while watching the
lights). Attending produced AERs of high and equal amplitude across
intensities; distraction especially diminished low intensities (Figure 6,
middle). Thus what happens across sessions seems to be different
from the distraction effect. Consistent with our findings, both Kopell,
Wittner, and Warrick (1969) and Chapman and Bragdon (1964) found
more enhancement of AER amplitude with selective attention for low-
intensity stimuli than for high-intensity stimuli.

C. AER Decrement with Loud Noise

In a third experiment (Buchsbaum, Landau, and Morgan, 1972) we

presented blocks of 10 flashes of the intensity (intensities 1 and 4 using
recording techniques in Buchsbaum and Pfefferbaum, 1971) in a ran-
domized order to 40 normal volunteers. Between the blocks of flashes
was a 4-s pause. About 2 s before the onset of half of the blocks of
flashes (chosen randomly) we presented a 94-dB 1000-Hz tone which
continued throughout the block-a quite noxious sound. For the flashes
presented in the blocks without the tone, AER amplitude increased
with increasing intensity. For flashes presented during the sounding of
the 94-dB tone, however, AER amplitude decreased-with the effect
only for the highest-intensity flash (Figure 6, right). A similar effect on
the high intensity only was seen in our experiments with audioanalge-
sia (Lavine, Buchsbaum, and Poncy, in press).

D. Differential Types of AER Decrement

Three different effects were seen: (1) an across-session, "arousal"

effect which acted on all intensities fairly evenly, affecting slope mini-
mally; (2) an "attention" effect which operated primarily to enhance
amplitude for low-intensity stimuli and thus lower AER slope; and (3) a
"sensory overload" effect which operated primarily to reduce AER
amplitUde at high intensities. Thus reducing appears to be linked to the
active phenomena of paying attention and protection from too-intense

levels of sensory input. Habitual tendencies to attend to sensory stimuli

may be reflected in the AER amplitude/intensity slope as well as habit-
ual tendencies to inhibit sensory input.


A. Psychological Magnitude and Power Functions

The responses people make when asked, "How intense is that

stimulus?" are variable. Psychological magnitude can be reported with
magnitude estimates, rations, or cross-modal matching, and these pro-
cedures produce results that roughly fit a power function

where '" is the psychological magnitude and cf> the stimulus intensity
(Stevens, 1963, 1971). This relationship holds fairly well when data are
averaged across subjects, but the exponent f3 varies widely from
subject to subject (e.g., Stevens and Guirao, 1964)-suggesting that
the subjective intensity of a stimulus may increase more or less rapidly
with increasing stimulus intensity. These individual differences in
power-function exponent seem quite reliable across session and mod-
alities (Jones and Marcus, 1963; Rule, 1966; Ekman, Hosman, Lind-
man, Ljungberg, and Akesson, 1968; Reason, 1972; Wanschura and
Dawson, 1974). Luce (1972) cogently reviewed the problems of com-
paring psychophysical responses and physical units.

B. Power Function Exponents and Augmenting/Reducing

Much debate has raged over the source of these differences. Sug-
gested factors include ratio concepts (Stevens, 1971), response bias
(Kunnapas, Hallsten, and Soderberg, 1973), regression effects (Wan-
schura and Dawson, 1974), and memory factors (Engeland and Dawson,
1974). Recently a correlation was found between the amount of aug-
menting on the Petrie KFA and the slope of the magnitude-estimation
function for loudness measured by use of Stevens' procedure (Sales and

Throop, 1972). This is supported by Cavonius, Hilz, and Chapman

(1974), who similarly report KFAlbrightness-estimation slope correla-
tions, although the Kohler and Dinnerstein KFA procedure was used.
Reason (1968, 1972) also found correlations between an individual's
susceptibility to the spiral aftereffect and the slope of the loudness-
estimation function. His explanation of slope differences as individual
differences in "reactivity" is analogous to the augmenting/reducing
concept. Steeper loudness functions were also found to be correlated
with noise-annoyance susceptibility (Moreira and Bryan, 1972), a find-
ing consistent both with the reducer concept and with our own AER
study of noise tolerance. Subjects with higher-magnitude judgment
slopes also evidenced more anxiety (Stephens, 1970) and more excitabil-
ity (de Barbenza, Bryan, and Tempest, 1970) on personality tests, fur-
ther suggesting the importance of complex inhibitory processes.

C. AER and Psychophysical Scaling

In view of the vast variety of individual differences in amplitude/

intensity functions for the AER, these individual differences in psycho-
physical functions are not at all surprising. It would be attractive to
believe that a neurophysiological measure might avoid the methodo-
logical problems inherent in asking subjects to attach numbers to the
loudness of tones. We know, however, that cortical centers where AERs
originate are removed several synapses from the sensory transducer
and have both inhibitory and excitatory input. Attempts to fit power
functions to AER parameters have had only spotty success, especially
when AER amplitude is used; the commonly observed pattern of AER
saturation or reducing cannot, of course, be fitted with a power func-
tion over its whole range (e.g., Butler, Keidel, and Spreng, 1969).
Extensive experiments in which ratings of individual stimuli are
made and AERs collected simultaneously are needed to relate subjective
reports and particular AER components. It seems unlikely, however,
that reducing will be paralleled with reports of diminished intensity.
AER amplitude for the P100--P200 components seems more linked to the
process of making such a report or of suppressing such a process in
situations in which it is redundant, unpleasant, or inconvenient to



A. Response to Low-Intensity Stimuli

Buchsbaum and Silverman (1968) hypothesized that reducers were
hypersensitive to low-level sensory stimuli and thus required some
compensatory process to protect them from sensory inundation at high
intensities. Indeed, AER reducers were found to have lower visual
thresholds (when the method of limits was used) than AER augmenters
(Silverman, Buchsbaum, and Henkin, 1969). And at low to moderate
light levels, reducers had larger amplitude AERs than augmenters. This
suggests that we might expect reducers to tolerate sensory deprivation
better than augmenters-the opposite of the Petrie prediction. The
finding that pain-tolerant individuals tolerate sensory isolation better
(Peters et al., 1963; Zubek, 1963) is also consistent with the Buchsbaum-
Silverman stimulus-intensity control model. Petrie viewed reduction as
operating evenly on all intensities, whereas we conceptualized it as
operating primarily at high intensities in individuals who were unu-
sually responsive at low levels. In this form, the augmenting/reducing
dimension and the Pavlov-Teplov "strength of the nervous system"
dimension are similar. Pavlov's typology and the subsequent develop-
ment of this construct are well presented in Gray (1964) and Nebylitsyn
and Gray (1972) and are briefly summarized below.

B. "Strength of the Nervous System" and Reducing

Soviet psychophysiology since Pavlov has paid close attention to

the individual. Subjects are characteristically studied in great detail and
attempts are made to explain the variety of responses observed. Pav-
lov's theories of nervous system types grew out of such an effort. Three
continuous dimensions-strength, equilibrium of excitation, and inhi-
bition and mobility-were postulated and animated discussion of dog
personality enlivened the Wednesday conferences of Pavlov's group
(see Teplov, 1964). Two dogs studied in 1926-1928 were seen as exam-
ples of strong and weak nervous systems. Vot-te Chort showed condi-
tioned reflexes which increased with increasing tone intensity; Zhurka

showed no increase past moderate intensities (reducing, in our

terminology) .
Pavlov's "law of strength" implied that response should increase
evenly with increasing stimulus intensity-and Zhurka broke the law at
a lower stimulus intensity than Vot-te Chort. Some animals showed a
subsequent decrease in response as intensity was further increased.
The level at which the increasing relationship between intensity and
response magnitude ceased was termed the "threshold of transmarginal
inhibition." Animals with "strong" nervous systems were viewed as
having higher thresholds and maintaining linear amplitude/intensity
relationships at higher levels than animals with "weak" nervous sys-
tems. Transmarginal inhibition was viewed as a special "protective"
type of inhibition. Individuals with weak systems were seen as lacking
ordinary "internal" inhibition: "the weaker the nervous system the
more intense is the excitatory process which is set up by a given
physical stimulus" (Gray, 1964). Thus weak nervous systems should
have lower sensory thresholds and show decreasing responses with
increasing stimulus intensity at the high end of the intensity contin-
uum-like reducers. Strong nervous systems would have high thresh-
olds but show linear amplitude/intensity relationships-like augmen-
ters. Both augmenting/reducing and strength dimensions thus may
tend to imply an inverted U function relating response to stimulus
intensity with the curves of augmenters/"strongs" shifted toward the
high-intensity end.

C. Determination of Strength

Strength of the nervous system like augmenting/reducing has been

measured in a number of ways, including "absolute thresholds," reac-
tion time or conditioned reflexes to stimuli of varying intensities, and
effects of distracting stimuli on psychophysical task performance. Many
of these intriguing paradigms have their experimental techniques pre-
sented only in sketchy outline and without detailed statistical analysis
(Gray, 1964). The one measure which has been used both in Western
and in Soviet psychophysiology is the slope of the reaction-time/stimu-
Ius-intensity function. Strong nervous systems (augmenters) should
have large increases in reaction rate (decreases in reaction time) with
increasing stimulus intensity; weak nervous systems (reducers) should
show little change with increasing stimulus intensity. Sales and Throop

(1972) found that individuals with low sensory thresholds, indicative of

weak nervous systems, indeed had low reaction-rate intensity slopes,
and Sales and Throop's subjects were also reducers on the Petrie KFA.
Reason (1972) found correlations between the slope of the reaction-rate
function and the slope of loudness-magnitude functions. Interestingly,
negative reaction/stimulus intensity funct~ons were found by Tizard
and Venables (1956) in schizophrenics-who have been reported to be
extreme reducers for both the Petrie apparatus (Petrie et al., 1963) and
the AER (Landau et al., 1975).


A. Optimum Levels of Stimulation

The concept of an optimum level of continuous sensory stimulation
to maintain optimal intellectual function or the feeling of well-being has
been advanced on a neurological and several psychological levels. This
concept was recently reviewed by Zuckerman (1974). Silverman (1967),
Ludwig (1971), Sales (1971), and Zuckerman (1974) have suggested that
the augmenting/reducing stimulus control mechanism may be one of
the methods utilized by individuals to optimize the level of incoming
sensory stimulation. Defects in this mechanism could reduce the flexi-
bility of the organism in meeting the challenge of sensory overload (see
Miller, 1960; Ludwig, 1972; Lipowski, 1973).

B. Relationships between Pain Tolerance, Sensory

Homeostasis, and Distraction
The focusing of attention away from the source of painful stimula-
tion appears to be an important source of variation in pain tolerance
(e.g., Melzack and Wall, 1970). The ability to reduce, implicated in
pain tolerance, is viewed as a dimension of attention by Silverman
(1967). The perception-pain-control mechanism of Lykken, Macinoe,
and Tellegen (1972) requires paying attention to the warning stimulus.
The finding of an interaction between stimulus intensity and attention
in three AER studies (see Section VII.B) further links attentional
effects to pain and/or homeostatic sensory-control mechanisms. Since
the AER technique using stimuli of varying intensities makes it
possible to examine correlates of pain response in a subject "attend-

ing" to other tasks than making pain judgments, it may be especially

valuable for research separating these effects.

C. Conclusion
As progress is made in the neuroanatomicallocalization of specific
AER components, increasingly specific neural models will be possible
for the important stimulus-intensity-control mechanisms. The general
importance of such processes can be recognized in the diverse sources
of such concepts as Petrie's "augmenting/reducing," Pavlov's
"strength," Freud's "stimulus barrier," and others. An understanding
of such mechanisms may provide important clues about psychiatric
dysfunction: AER tools for diagnostic evaluation, drug selection, and
identification of genetic vulnerability are all potential applications
(Buchsbaum, 1975; Silverman, 1972).

The author wishes to thank Dr. Robert Lavine for helpful com-
ments, Sherry Buchsbaum for editorial assistance, and Cathy King and
Arlene Ammerman for technical and secretarial aid.


ABE, M. Electrical responses of the human brain to acoustic stimuli. The Tohoku Journal of
Experimental Medicine, 1953,60, 47-58.
ANDERSON, P., ECCLES, J. c., AND SEARS, T. A. Cortically evoked depolarization of
primary afferent fibers in the spinal cord. Journal of Neurophysiology, 1964,27,63-77.
ARMINGTON, J. C. Adaptational changes in the human electroretinogram and occipital
response. Vision Research, 1964a,4, 179--192.
ARMINGTON, J. c. Relations between electroretinograms and occipital potentials elicited
by flickering stimuli. Documenta Ophthalmogica, 1964b,18, 194-206.
BEAGLEY, H. A., AND KNIGHT, J. J. Changes in auditory evoked response with intensity.
Journal of Laryngology and Otology, 1967,81, 861--873.
BECK, C., AND ROSNER, B. S. Magnitude scales and somatic evoked potentials to percuta-
neous electrical stimulation. Physiology and Behavior, 1968,3, 947-953.
BLfIZ, B., DINNERSTEIN, A. J., AND LOWENTHAL, M. Relationship between pain tolerance
and kinesthetic size judgment. Perceptual and Motor Skills, 1966, 22, 463-469.
BORGE, G. F. Perceptual modulation and variability in psychiatric patients. Archives of
General Psychiatry, 1973,29, 760-763.
BROADHURST, A., AND MILLARD, D. W. Augmenters and reducers: A note on a replication
failure. Acta Psychologica, 1969,29, 290-296.
BUCHSBAUM, M. Average evoked response and stimulus intensity in identical and frater-
nal twins. Physiological Psychology, 1974,2, 365-370.
BUCHSBAUM, M. Average evoked response augmenting/reducing in schizophrenia and

affective disorders. In D. X. FREEDMAN (Ed.), The biology of the major psychoses: A

comparative analysis. New York: Raven Press, 1975.
BUCHSBAUM, M., GILLIN, J. c., AND PFEFFERBAUM, A. Effect of sleep stage and stimulus
intensity on auditory average evoked responses. Psychophysiology, 1975,12(6), 707-
BUCHSBAUM, M., GOODWIN, F., MURPHY, D., AND BORGE, G. AER in affective disorders.
American Journal of Psychiatry, 1971,128, 19-25.
BUCHSBAUM, M., HENKIN, R. I., AND CHRISTIANSEN, R. L. Age and sex differences in
averaged evoked responses in a normal population with observations on patients
with gonadal dysgenesis. Electroencephalography and Clinical Neurophysiology, 1974,
37, 137-144.
BUCHSBAUM, M., LANDAU, S., AND MORGAN, C. Arousal and attention: Evoked response
stimulus intensity modifiers. Presented at the 80th annual convention, American
Psychological Association, Honolulu, Hawaii, 1972.
BUCHSBAUM, M., LANDAU, S., MuRPHY, D. L., AND GOODWIN, F. Average evoked
response in bipolar and unipolar affective disorders: Relationship to sex, age of
onset and monoamine oxidase. Biological Psychiatry, 1973,7, 199-212.
BUCHSBAUM, M., AND PFEFFERBAUM, A. Individual differences in stimulus intensity
response. Psychophysiology, 1971,8, 60()...{j1l.
BUCHSBAUM, M., AND SILVERMAN J. Stimulus intensity control and the cortical evoked
response. Psychosomatic Medicine, 1968,30, 12-22.
BUTLER, R. A., KEIDEL, W. D., AND SPRENG, M. An investigation of the human cortical
evoked potential under conditions of monaural and binaural stimulation. Acta Oto-
laryngologica, 1969, 68, 317-326.
CALMA, I. Presynaptic inhibition of the terminals of cutaneous nerve fibers by stimula-
tion of the ventral thalamo-diencephalic region. Journal of Physiology (London), 1966,
CAVONIUS, C. R., Hnz, R., AND CHAPMAN, R. M. A possible basis for individual
differences in magnitude-estimation behavior. British Journal of Psychology, 1974,65,
CHAPMAN, R. M., AND BRAGDON, H. R. Evoked responses to numerical and non-numeri-
cal visual stimuli while problem solving. Nature (London), 1964,203, 1155-1157.
CLARK, W. c., AND GOODMAN, J. Effects of suggestion of d' and Cx for pain detection and
pain tolerance. Journal of Abnormal Psychology, 1974,83, 364-372.
CLYNES, M., KOHN, M., AND LIFSHIlZ, K. Dynamics and spatial behavior of light evoked
potentials, their modification under hypnosis, and on-line correlation in relation to
rhythmic components. Annals of New York Academy of Science, 1964,112,468-509.
CRUIKSHANK, R. M. Human occipital brain potentials as affected by intensity-duration
variables of visual stimulation. Journal of Experimental Psychology, 1937,21, 625-64l.
DAVIS, H., BOWERS, c., AND HIRSH, S. K. Relations of the human vertex potential to
acoustic input: Loudness and masking. Journal of the Acoustical Society of America,
1968,43, 431-438.
DAVIS, H., MAST, T., YOSHIE, N., AND ZERLIN, S. The slow response of the human cortex
to auditory stimuli: Recovery process. Electroencephalography and Clinical Neurophys-
iology, 1966,21, 105-113.
DAVIS, H., AND ZERLIN, S. Acoustic relations of the human vertex potential. Journal of the
Acoustical Society of America, 1966,39, 109-116.
DE BARBENZA, C. M., BRYAN, M. E., AND TEMPEST, W. Individual loudness functions.
Journal of Sound Vibration, 1970, 11, 399-410.
DEBECKER, J., AND DESMEDT, J. E. Les potentiels evoques cerebraux et les potentiels de
nerf sensible chez l'homme. Acta Neurologica Belgica, 1964,64, 1212-1248.

DELL, P., BONVALLET, M., AND HUGELIN, A. Mechanisms of reticular deactivation. In G. E.

W. WOLSTENHOLME AND M. O'CONNOR (Eds.), The nature of sleep. London: Churchill,
DEVOE, R. G., RIPPS, H., AND VAUGHAN, H. G. Cortical responses to stimulation of the
human fovea. Vision Research, 1968,8, 135-147.
DUSTMAN, R. E., AND BECK, E. The visually evoked potential in twins. Electroencephalog-
raphy and Clinical Neurophysiology, 1965,19, 580-575.
EASON, R. G., AIKEN, L. R., WHITE, C. T., AND LICHTENSTEIN, M. Activation and
behavior: II. Visually evoked cortical potentials in man as indicants of activation
level. Perceptual and Motor Skills, 1964,19, 875-895.
EASON, R. G., HARTER, M. R., AND WHITE, C. T. Effects of attention and arousal on
visually evoked cortical potentials and reaction time in man. Physiology and Behavior,
1969,4, 283-289.
dual differences in scaling performance. Perceptual and Motor Skills, 1968,26,815-823.
ENGELAND, W., AND DAWSON, W. E. Individual differences in power functions for a 1
week intersession interval. Perception and Psychophysics, 1974,15, 349-352.
FRANZEN, 0., AND OFFENLOCH, K. Evoked response correlates of psychophysical magni-
tude estimates for tactile stimulation in man. Experimental Brain Research, 1969,8, 1-
GRAY, J. A. (Ed.). Pavlov's typology. New York: Macmillan, 1964.
HAGBARTH, K. E., AND KERR, D. I. B. Central influences on spinal afferent conduction.
Journal of Neurophysiology, 1954,17, 295.
HALL; R. A., RAPPAPORT, M., HOPKINS, H. K., AND GRIFFIN, R. B. Peak identification in
visual evoked potentials. Psychophysiology, 1973a, 10, 52--{;0.
HALL, R. A., RAPPAPORT, M., HOPKINS, H. K., AND GRIFFIN, R. B. Tobacco and evoked
potential. Science, 1973b, 180, 212-214.
response and behavior in cats. Science, 1970,170, 998--1000.
HAMASAKI, D. I., AND WINTERS, R. W. Intensity-response functions of visually deprived
LGn neurons of cats. Vision Research, 1973, 13, 925-936.
HARTLEY, L. R. The effect of stimulus relevance of the cortical evoked potentials. Journal of
Experimental Psychology, 1970,22, 531-546.
HENRY, G. B., AND TEAS, D. Averaged evoked response and loudness: Analysis of
response estimates. Journal of Speech and Hearing Research, 1968,11, 334-342.
HILGARD, E. R., MORGAN, A. H., AND PRYTULAK, S. The psychophysics of the kinesthetic
aftereffect in the Petrie block experiment. Perception and Psychophysics, 1968,4, 129-
HILLMAN, P., AND WALL, P. D. Inhibitory and excitatory factors influencing the receptive
fields on lamina 5 spinal cord cells. Experimental Brain Research, 1969,9, 284-306.
lKUTA, T. Influences of different stimulus intensities on the human somatosensory evoked
response (A proposal for the data processing). Folia Psychiatrica et Neurologica Japan-
ica, 1968,22, 121-141.
IL'YANOK, V. A. Effect of intensity and depth of pulsation of flickering light on the
electrical activity of the human brain. Biofizika, 1961, 6, 72-82.
JACOBSON, J. H., KAWASAKI, K., AND HIROSE, T. The human electroretinogram and
occipital potential in response to focal illumination of the retina. Investigative Oph-
thalmology, 1969,8, 545-556.
JONES, F. N., AND MARCUS, J. J. The subject effect in judgments of subjective magnitude.
Journal of Experimental Psychology, 1963,61, 40-44.

KEIDEL, W. D., AND SPRENG, M. Neurophysiological evidence for the Stevens' power
function in man. Journal of the Acoustical Society of America, 1965,38, 191-195.
KHECHINASHVILl, S. N., KEVANISHVILI, Z., AND KAJAIA, O. A. Amplitude and latency
studies of the averaged auditory evoked responses to tones of different intensities.
Acta Otolaryngologica, 1973, 76, 395-401.
KITAJIMA, H. On the cerebral evoked response in man as a function of the intensity of
flicker stimulation. Electroencephalography and Clinical Neurophysiology, 1967, 22,
KLINGAMAN, R. 1., AND ANCH, A. M. Human auditory evoked response intensity
functions from monopolar and bipolar scalp recordings. Psychonomic Science, 1972,
KNISPEL, J. D., AND SmGEL, J. Tegmental stimulation: Aversive effects on behavior and
modulation of visual evoked potentials. Brain Research, 1972,37, 317-321.
KNISPEL, J. D., AND SmGEL, J. Habituation of aversive reticular stimulation effects on
evoked potentials. Brain Research, 1973,56, 340-344.
KOELLA, W. P., AND PERRY, A. Cortico-subcortical homeostasis in the cat's brain. Science,
1963, 142, 586--589.
KOHLER, W., AND DINNERSTEIN, D. Figural aftereffects in kinesthesis. In A. MICHOTTE
(Ed.), Miscellanea Psychologica. Paris: Vrin., 1947.
KOLLAR, A. Die Beziehungen zwischen Lautstarke, Latenzzeit und Amplitude in der
EEG-Audiometrie. Mschr. Ohrenheilk (Wien), 1971,105, 49-59.
KOOI, K. A., AND BAGCHI, B. K. Observations on early components of the visual evoked
response and occipital rhythms. Electroencephalography and Clinical Neurophysiology,
1964, 17, 63~3.
KOPELL, B., WITTNER, W. AND WARRICK, G. The effects of stimulus differences, light
intensity, and selective attention on the amplitude of the average evoked potential
in man. Electroencephalography and Clinical Neurophysiology, 1969,26, 619-622.
KUNNAPAS, T., HALLSTEN, L., AND SODERBERG, G. Interindividual differences in homo-
modal and heteromodal scaling. Acta Psychologica, 1973,37, 31--42.
LANDAU, S. G., AND BUCHSBAUM, M. Average evoked response and muscle tension.
Physiological Psychology, 1973,1, 56--60.
nia and stimulus intensity control. Archives of General Psychiatry, 1975,32(8), 1239-
LAVINE, R. A. Phase-locking in response of single neurons in cochlear nuclear complex of
the cat to low-frequency tonal stimuli. Journal of Neurophysiology, 1971,34,467--483.
LAVINE, R. A., BUCHSBAUM, M. AND PONCY, M. Auditory analgesia: Somatosensory
evoked response and subjective pain rating assessment. Psychophysiology, in press.
LEWIS, E. G., DUSTMAN, R. E., AND BECK, E. C. Evoked response similarity in monozy-
gotic, dizygotic and unrelated individuals: A comparative study. Electroencephalogra-
phy and Clinical Neurophysiology, 1972, 32, 309-316.
LmBMAN, J., AND GRAHAM, J. T. Frequency and intensity effects of bone conduction
signals on averaged evoked auditory potentials. Journal of Auditory Research, 1967,7,
LIPOWSKI, A. J. Affluence, information inputs and health. Social Science and Medicine,
1973, 7, 517-529.
LuCE, R. D. What sort of measurement is psychophysical measurement. American Psy-
chologist, 1972,27, 96--106.
LUDWIG, A. M. Self-regulation of the sensory environment. Archives of General Psychiatry,

LUDWIG, A. M. "Psychedelic" effects produced by sensory overload. American Journal of

Psychiatry, 1972,10, 114--117.
LUNDERBERG, A., AND OSCARSSON, O. Three ascending spinal pathways in the dorsal part
of the lateral funiculus. Acta Physiologia Scandinavia, 1961,51, 1-16.
LYKKEN, D. T., MACINOE, 1., AND TELLEGEN, A. Preception: Autonomic response to shock
as a function of predictability in time and locus. Psychophysiology, 1972,9, 318-333.
LYKKEN, D. T., TELLEGEN, A., AND THORKELSON, K. Genetic determination of EEG
frequency spectra. Biological Psychology, 1974,1, 245-259.
MADELL, J. R., AND GOLDSTEIN, R. Relation between loudness and the amplitude of the
early components of the averaged electroencephalic response. Journal of Speech and
Hearing Research, 1972,15, 134--141.
MARCO, J. Cortical audiometry. Acta Otolaryngologica, 1972,73, 197-202.
MARK, R. F., AND STEINER, J. Cortical projection of impulses in myelinated cutaneous
afferent nerve fibers of the cat. Journal of Physiology, 1958,142, 544--562.
MEIZACK, R. The puzzle of pain. New York: Basic Books, 1973.
MEIZACK, R., AND WALL, P. D. Psychophysiology of pain. International Anesthesiology
Clinics, 1970,8, 3-34.
MILLER, J. C. Information input overload and psychopathology. American Journal of
Psychiatry, 1960,116,695-704.
MOLINO, J. Equal aversion levels for pure tones and 1/3 octave band of noise. Journal of the
Acoustical Society of America, 1974,55, 1285-1289.
MONTAGU, J. D. The relationship between the intensity of repetitive photic stimulation
and the cerebral response. electroencephalography and Clinical Neurophysiology, 1967,
23, 152-161.
MOORE, E. J., AND ROSE, D. E. Variability of latency and amplitude of acoustically evoked
responses to pure tones of moderate to high intensity. International Audiology, 1969,
8, 172-181.
MOREIRA, N. M., AND BRYAN, M. E. Noise annoyance susceptibility. Journal of Sound and
Vibration, 1972,21, 449-462.
MORGAN, A. H., LEZARD, F., PRYTULAK, S., AND HILGARD, E. R. Augmenters, reducers,
and their reaction to cold pressor pain in waking and suggested hypnotic analgesia.
Journal of Personality and Social Psychology, 1970,16, 5-11.
MOUNTCASTLE, V. B. (Ed.), Medical physiology (13 ed.), (Vol. 1). Chapter 10: Neural
Mechanisms in Somesthesia. St. Louis: C. V. Mosby, 1974.
MUSHIN, J., AND LEVY, R. Averaged evoked response in patients with psychogenic pain.
Psychological Medicine, 1974,4, 19-27.
NEBYLITSYN, V. D., AND GRAY, J. A. Biological bases of individual behavior. New York:
Academic Press, 1972.
OSBORNE, R. T. Heritability estimates for the visual evoked response. Life Science, 1970,9,
PETERS, J., BENJAMIN, F. B., HELVEY, W. M., AND ALBRIGHT, C. A. Study of sensory
deprivation, pain and personality relationships for space travel. Aerospace Medicine,
1963,34, 830-837.
PETRIE, A. Some psychological aspects of pain and the relief of suggering. Annals of the
New York Academy of Science, 1960,86, 13-27.
PETRIE, A. Individuality in pain and suffering. Chicago: University of Chicago Press, 1967.
PETRIE, A. Reduction or augmentation? Why we need two "planks" before deciding.
Perceptual and Motor Skills, 1974,39, 460.
PETRIE, A., COLLINS, W., AND SOLOMAN, P. Pain sensitivity, sensory deprivation and
susceptibility to satiation. Science, 1958,128, 1431-1433.

PETRIE, A., HOLLAND, T., AND WOLK, I. Sensory stimulation causing subdued experience:
Audio-analgesia and perceptual augmentation and reduction. Journal of Nervous and
Mental Disease, 1963, 137, 312-321.
PFEFFERBAUM, A., AND BUCHSBAUM, M. Handedness and cortical hemisphere effects in
sine wave stimulated evoked responses. Neuropsychologia, 1971,9, 237-240.
PICTON, T. W., GOODMAN, W. S., AND BRYCE, D. P. Amplitude of evoked responses to
tones of high intensity. Acta Otolaryngologica, 1970,70, 77-82.
PLATT, D., HOlZMAN, P. S., AND LARSON, D. Individual consistencies in kinesthetic
figural aftereffects. Perceptual and Motor Skills, 1971,32, 787-795.
responses to clicks and tones of varying intensity in waking adults. Electroencephal-
ography and Clinical Neurophysiology, 1966,21, 335-344.
REASON, J. T. Relations between motion sickness susceptibility, the spiral aftereffect and
loudness estimation. British Journal of Psychology, 1968,59,385-393.
REASON, J. T. Some correlates of the loudness function. Journal of Sound and Vibration,
1972,20, 305-309.
REDMOND, D. E., BORGE, G. F., BUCHSBAUM, M., AND MAAS, J. W. Evoked potential
studies of brain catecholamine alterations in monkeys. Journal of Psychiatric Research,
REGAN, D., AND BEVERLEY, K. I. Relation between the magnitude of flicker sensation and
evoked potential amplitude in man. Perception, 1973,2, 61-65.
RICHEY, E. T., KOOI, K. A., AND WAGGONER, R. W. Visually evoked responses in
migraine. Electroencephalography and Clinical Neurophysiology, 1966,21, 23-27.
RIETVELD, W. J. The occipitocortical response to light flashes in man. Acta Physiologica et
Pharmacologia Neerlandica, 1963,12, 373-407.
RIETVELD, W. J., AND TORDOIR, W. E. M. The influence of flash intensity upon the visual
evoked response in the human cortex. Acta Physiologia et Pharmacologia Neerlandica,
1965, 13, 160-170.
ROSE, D. E., AND RUHM, H. B. Some characteristics of the peak latency and amplitude of
the acoustically evoked response. Journal of Speech and Hearing Research, 1966,9,412-
ROSNER, B. S., AND GOFF, W. R. Electrical responses of the nervous system and subjective
scales of stimulus intensity. In W. D. NEFF (Ed.), Contributions to sensory physiology
(Vol. 2). New York: Academic Press, 1967.
ROTH, W. T. Auditory evoked responses to unpredictable stimuli. Psychophysiology, 1973,
10, 125-138.
ROTHMAN, H. H. Effects of high frequencies and intersubject variability on the auditory-
evoked cortical response. The Journal of the Acoustic Society of America, 1970,47,569-
RULE, S. J. Subject differences in exponents of psychophysical power functions.
Perceptual and Motor Skills, 1966,23, 1125-1126.
RYAN, E. c., AND FOSTER R. Athletic participation and perceptual augmentation and
reduction. Journal of Personality and Social Psychology, 1967,6,472-476.
SALES, S. M. Need for stimulation as a factor in social behavior. Journal of Personality and
Social Psychology, 1971, 19, 124--134.
SALES, S. M., AND THROOP, W. F. Relationship between kinesthetic aftereffects and
"strength of the nervous system." Psychophysiology, 1972,9, 492-497.
SAMSON, H. H., AND YOUNG, M. L. The relation of flash intensity and background
illumination to the photic evoked potential in the pigeon's optic tectum. Vision
Research, 1973,13, 253-261.

SCHECHrER, G., AND BUCHSBAUM, M. The effects of attention, stimulus intensity and
individual differences on the average evoked response. Psychophysiology, 1973,10.
SCHOOLER, c., AND SILVERMAN, J. Differences between correlates of perceptual styles and
Petrie task performance in chronic and acute schizophrenics. Perceptual and Motor
Skills, 1971,32, 595-601.
SCHWEI'IZER, P. K., AND TEPAS, D. I. Intensity effects of the auditory evoked brain
response to stimulus onset and cessation. Perception and Psychophysics, 1974, 16,
SHAGASS, C. Evoked potentials in psychiatry. New York: Plenum Press, 1972.
SHAGASS, c., OVERTON, D. A., BARTOLUCCI, G., AND STRAUMANIS, J. J. Effect of attention
modification by television viewing on somatosensory evoked responses and recovery
function. Journal of Nervous and Mental Disease, 1971, 152, 53-62.
SHAGASS, c., AND SCHWARTZ, M. Evoked potential studies in psychiatric patients. Annals
of the New York Academy of Science, 1964,112, 526-541.
SHAGASS, c., SCHWARTZ, M., AND KRISHNAMOORTI, S. Some psychologic correlates of
cerebral response evoked by light flash. Journal of Psychosomatic Research, 1965, 9,
SHIPLEY, T., JONES, R. W., AND FRY, A. Intensity and the evoked occipitogram in man.
Vision Research, 1966, 6, 657-667.
SILVERMAN, J. Perceptual control of stimulus intensity in paranoid and nonparanoid
schizophrenia. Journal of Nervous and Mental Disease, 1964,6, 545-549.
SILVERMAN, J. Variations in cognitive control and psychophysiological defense in the
schizophrenias. Psychosomatic Medicine, 1967,29, 225-251.
SILVERMAN, J. Stimulus intensity modulation and psychological dis-ease. Psychopharma-
cologia, 1972,24, 42-80.
SILVERMAN, J., BUCHSBAUM, M., AND HENKIN, R. Stimulus sensitivity and stimulus
intensity control. Perceptual and Motor Skills, 1969,28, 71-78.
SOSKIS, D. A., AND SHAGASS, C. Evoked potential tests of augmenting-reducing. Psycho-
physiology, 1974,11, 175-190.
SPILKER, B., AND CALLAWAY, E. "Augmenting" and "reducing" in averaged visual
responses to sine wave light. Psychophysiology, 1969, 6, 49-57.
STARK, 1. H., AND NORTON, J. c. The relative reliability of average evoked response
parameters. Psychophysiology, 1974,11, 600-602.
STEPHENS, S. D. G. Personality and the slope of loudness function. Journal of Experimental
Psychology, 1970,22, 9-13.
STEVENS, S. S. The psychophysics of sensory function. In W. A. ROSENBLITH (Ed.), Sensory
communication. Boston: MIT Press, 1963.
STEVENS, S. S. Issues in psychophysical measurement. Psychological Review, 1971,78,426-
STEVENS, J. C. AND GUIRAO, M. Individual loudness functions. J. Acoustical Soc. Amer.,
1964,36 (11), 221!.>-2213.
SUZUKI, T., AND TAGUCHI, K. Cerebral evoked response to auditory stimuli in waking
man. Annals of Otology, Rhinology and Laryngology, 1965,74, 128-139.
SWEENEY, D. R. Pain reactivity and kinesthetic aftereffect. Perceptual and Motor Skills,
1966,22, 763-769.
TEPAS, D. I., AND ARMINGTON, J. c. Properties of evoked visual potentials. Vision
Research, 1962,2, 449-461.
TEPAS, D. I., ARMINGTON, J. c., AND KROPFL, W. J. Evoked potentials in the human visual
system. In E. E. BERNARD AND M. R. KARE (Eds.), Biological prototypes and synthetic
systems. New York: Plenum Press, 1962.

TEPAS, D. I., BOXERMAN, 1. A., AND ANCH, A. M. Auditory evoked brain response:
. Intensity functions from bipolar human scalp recordings. Perception and Psychophys-
ics, 1972,11, 217-22l.
TEPLOV, B. M. Problems in the study of general types of higher nervous activity in man
and animals. In J. A. GRAY (Ed.), Pavlov's typology. New York: Macmillan, 1964.
TIZARD; J., AND VENABLES, P. H. Reaction time responses by schizophrenics, mental
defectives and normal adults. American Journal of Psychiatry, 1956, 112, 8OH07.
TURSKY, B., AND WATSON, P. D. Controlled physical and subjective intensities of electric
shock. Psychophysiology, 1964,1, 151-162.
UTIAL, W. R., AND COOK, 1. Systematics of the evoked somatosensory cortical potential:
A psychophysical-electrophysiological comparison. Annals of the New York Academy
of Science, 1964, 112, 60-80.
VAN DER TWEEL, 1. H., AND VERDUYN LUNEL, H. F. E. Human visual responses to
sinusoidally modulated light. Electroencephalography and Clinical Neurophysiology,
1965, 18, 587-598.
VAUGHAN, H. G. The perceptual and physiologic significance of visual evoked responses
recorded from the scalp in man. In Clinical Electroretinography Suppl. Vision Research,
1966, 20~223.
VAUGHAN, H. G., AND HULL, R. C. Functional relation between stimulus intensity and
photically evoked cerebral response in man. Nature, 1965,206, 720-722.
WALL, P. D. The laminar organization of the dorsal hom and effects of descending
impulses. Journal of Physiology (London), 1967,188, 403-423.
WANSCHURA, R. G., AND DAWSON, W. E. Regression effect and individual power func-
tions over sessions. Journal of Experimental Psychology, 1974,102, 80'Hl12.
WICKE, J. D., DONCHIN, E., AND LINDSLEY, D. B. Visual evoked potentials as a function of
flash luminance and duration. Science, 1964, 146, 8~.
WOOTEN, B. R. Photopic and scotopic contributions to the human visually evoked cortical
potential. Vision Research, 1972, 12, 1647-1660.
ZUBEK, J. P. Pain sensitivity as a measure of perceptual deprivation tolerance. Perceptual
and Motor Skills, 1963, 17, 641-642.
ZUCKERMAN, M. The sensation seeking motive. Progress in Experimental Personality Re-
search, 1974,7, 80-148.
ZUCKERMAN, M., MURTAUGH, T., AND SIEGEL, J. Sensation seeking and cortical augment-
ing-reducing. Psychophysiology, 1974,11, 535-542.
4 Neodissociation Theory of
Multiple Cognitive
Control Systems


Man does more than one thing at a time-all of the time-but the
representation of these actions in consciousness is never complete. On
occasion he becomes conscious of much that happens within his body
and of much that is happening currently in the external world, as well
as of remembered or imagined events. His awareness can shift from one
to another of these happenings, and there is some question about how
much he can comprehend at once.
All of this is familiar, but problems arise when a scientific account
is attempted. The total problem is usually seen as too large, so that the
investigator approaches it topically, that is, by selecting some manage-
able features that can be appropriately labeled and then studying them.
The recent upsurge of interest in locus of control represents an approach
taken largely by those committed to the study of personality and social
psychology; selective attention in its various manifestations becomes
the focus of experimental psychologists, extending into various studies
of task interferences. Those with psychophysiological interest become
concerned with hemispheric laterality or with state-dependent learn-
ing. Psychopathologists are again studying multiple personalities. I am
proposing that it is time to take a look at the whole problem and to see if
some unifying framework may not bring together the various lines of
evidence into a general theory of consciousness and of cognitive control
The study of hypnosis moves headlong into this set of problems
because hypnotic procedures change the balance between voluntary
ERNEST R. HILGARD Department of Psychology, Stanford University, Stanford,


and involuntary processes, interfere with normal information-process-

ing and memory retrieval, and produce a variety of distortions of
awareness and splits in consciousness. The phenomena are baffling and
are in many ways troublesome to investigate properly. If, however,
hypnotic investigators are successful in what they are doing, they
should be able to tell us not only about hypnosis but about human
functioning more generally and so contribute to the understanding of
normal consciousness and the control systems affecting it.


Pierre Janet (1859-1947) was one of the first to propose a general

theory of the organization of consciousness derived in large measure
from observations made with the help of hypnosis. He was the first to
introduce the term subconscious to refer to a level of cognitive function-
ing out of awareness that could on occasion be brought to conscious-
ness. The term unconscious was already familiar through the writings of
von Hartmann, but Janet preferred to substitute his term to avoid the
romantic excesses that were already centered in the term unconscious.
Later Morton Prince introduced the term coconscious for essentially the
same reason. Janet's explanation for the constellation of ideas not
available to the main consciousness was that they were dissociated. The
term derived, of course, from the prevalent doctrine of association. If
memories are brought to consciousness by way of the association of
ideas, then those not available to association must be "dis-associated."
Janet offered a simple diagram to show how a system of ideas, coherent
in itself, might be separated off from the primary personal conscious-
ness (Figure 1). The diagram was used to illustrate the case of Irene,
who in her somnambulistic state repeatedly rehearsed the death of her
mother, which she had experienced under traumatic circumstances. In
her normal condition she not only forgot what she had dramatized in
her somnambulism, but she forgot the events themselves. "I know very
well my mother must be dead," she is reported to have said, "since I
have been told so several times, since I see her no more, and since I am
in mourning; but I really feel astonished at it. When did she die?"
The case of Irene represented what Janet called monoideic som-
nambulism. He had :::ome to his point of view earlier through the study

FIGURE 1. Janet's diagrammatic

representation of the dissocia- s ~--~
tion in the case of Irene. The po-
lygon represents the ideas re-
lated to the death of her mother,
separated from the main person- ~~------+-~ M
ality (P). S stands for the sight of
the face of the dead mother, V for
the sound of her voice, M for the
feeling of movements in carrying
her body, and so on. This iso-
lated, yet integrated, set of ideas
and memories is responsible for
Irene's strange behavior in the
somnambulistic state (Janet,
1907, page 41).

of alternating personalities manifested by the subjects he studied. In

the thesis on Psychological Automatism that he presented to the Faculty
of Letters, before he went on to the MD a few years later, he distin-
guished between the total automatism of the catatonic patient and the
partial automatisms of the hysterical patients whom he had studied
(Janet, 1889). Two of these, Leonie and Lucie, may serve to illustrate the
kinds of phenomena with which he dealt.
Leonie was his first-reported and most thoroughly experimented-
upon case. She came to Janet's attention early in his career, when he
was more interested in experimenting upon her than in attempting to
resolve her problems. She had apparently had natural attacks of som-
nambulism since the age of 3, and had been repeatedly hypnotized by
all sorts of people from the age of 16 on; she was 45 when Janet was still
studying her. She had spent her childhood as a peasant in country
surroundings, but the rest of her life had been spent in "drawing rooms
and doctors' offices," as Janet put it. In her normal state she was
serious, timid, mild, and a little sad; when hypnotized she became
vivacious and noisy, with a tendency to irony and jesting with respect
to the strangers who had come to witness her hypnotic behavior. Janet

at first performed some dramatic but poorly controlled studies of hyp-

notic influence at a distance, but he repudiated any parapsychological
influences and attempted to give a purely naturalistic account of what
he observed. Leonie eventually turned up with three personalities
uncovered with the aid of hypnosis, on occasion called Leonie I, II, and
III, sometimes given the names by which the first two referred to
themselves--Leonie and Leontine-and, for the third, Leonore, a name
given by the magnetizer who had first discovered the third personality.
Leontine appeared when hypnotized by Janet, as she had for other
hypnotists before him. Later on, when Leontine was herself hypno-
tized, a third personality (Leonore) made her appearance. It was only
after studying Leonie for some time that Janet found out that she had
been treated hypnotically years before by some of the "magnetizers" of
that period and that the "new" personality had already been elicited
and christened 20 years earlier. In a much-studied and much-hypno-
tized case of this kind doubts always arise as to the role of the hypnotist
in consolidating personalities out of amnesic material: Janet was not
totally unaware of the problem. He recognized the role of defining the
secondary personality by naming it: "Once baptized, the unconscious
personality is clearer and more definite; it shows its psychological traits
more clearly" Oanet, 1889, p. 318). The three personalities of Leonie
showed signs of their origins. The first was appropriate to her upbring-
ing as a simple country girl and housewife, now placed in a sophisti-
cated urban setting. She had had her first child while hypnotized and
spontaneously fell into the hypnotic state when her other children were
born; it is not very surprising that the hypnotized personality (Leonie
II) claimed the children as her own, while assigning the husband to
Leonie I, who accepted both the husband and the children. Leonore
(Leonie III), doubtless a product of the hypnotic manipulations, might
have been made use of to reintegrate the personality, for she was aware
of the others, although considering Leonie I to be stupid and Leonie II
to be disturbed. The amnesic barriers that persisted made Leonie I
know only herself; Leonie II knew Leonie I as well as herself; and
Leonie III knew them all.
At the age of 19 Lucie came to Janet's attention because she was
seized with fits of terror without motivation: "I am afraid and I don't
know why." Through the use of automatic writing, a favorite 19th-
century method of getting at hidden aspects of personality (deriving in

the first place from mediumistic seances), Janet uncovered the roots of
her terror. She had been severely frightened at the age of 7 by two men
hiding behind a curtain and trying to playa joke on her. A second
personality, Adrienne, relived this experience during her fits of terror.
It should be noted that the dissociation was not complete; the emotion
broke through to the normal consciousness, although the events
that were the occasion of the emotion lay concealed behind a veil of
amnesia. Janet relieved her of her symptoms through a combina-
tion of hypnosis and automatic writing and the second personality
Janet's theory was supported for a time in America by Boris Sidis
and Morton Prince, but it gradually died out, in part through being
superseded by psychoanalysis, in part by failures in experiments to
show complete dissociation between events conscious and subcon-
scious. The epitaph was written by White and Shevach (1942), who felt
that the concept was no longer useful. Little has been heard of it since,
except for occasional purely descriptive references to some of the dis-
junctions between events and their conscious representation.


In proposing a kind of revival of dissociation theory, I am calling

attention to the phenomena that gave rise to dissociation theory in the
first place. Somnambulisms, amnesias, fugues, and multiple personali-
ties are phenomena of nature, occurring spontaneously and independ-
ent of hypnosis. Even if they are rare, they may tell us much about
human personality and consciousness. A simple revival of the dissocia-
tion concept would, however, be misleading. Such a revival would
carry with it Janet's insistence that these and related phenomena are
found only among those with hysterical personalities, and a revival
would be a defense of many observations of doubtful value. Further-
more, it would reinstate a doctrine of sharp separation between disso-
ciated activities. If the concept of neodissociation is used, many of the
controversial questions can be restated in contemporary form, so that
while the historical roots of the concept are given full credit, there is no
need for adherence to the views of those who propounded the original
doctrine of dissociation. Even the associationistic flavor does not have

to be preserved, if modem information-processing language proves to

be more useful. In summary, neodissociation theory proposes a fresh
start in the attack on the kinds of problems that gave rise to the classical
The aim is to produce a modem comprehensive theory to account
for the multiplicity of processes that control overt behavior and con-
scious processes, with full recognition that something like parallel
processing may occur and that all processed information is not available
at anyone time to consciousness. This is a large order, and the thoughts
presented in this chapter are therefore incomplete. It is hoped, how-
ever, that they may call attention to an enterprise in which others may
wish to share.


One of the earliest discovered aspects of hypnosis was posthyp-

notic amnesia for the events taking place within hypnosis. That this
was a true concealment, and not ordinary forgetting, was illustrated by
the fact that the memories lost in hypnotic amnesia could be recovered.
Amnesia was so consistently a part of hypnotic responsiveness that it
was the original defining characteristic of hypnosis, reflected in the
term somnambulist assigned to the highly hypnotizable person. While
the term somnambulist is still used to describe the hypnotic virtuoso, it
has lost its original meaning, which derived from the fact that the
sleepwalker is usually amnesic for his sleepwalking activities.
Amnesia provides a paradigm for dissociation because it demon-
strates that fully registered ideas, ordinarily available to memory, can
become temporarily "dissociated" and unavailable. Although amnesia
sounds like Freudian repression, the repressed ideas in posthypnotic
amnesia need not have any emotional significance for the subject; they
are not necessarily "primary-process" thoughts any more than "sec-
ondary-process" ones. This difference from repression is worth noting,
because psychoanalytic thinking is somewhat more familiar than the
concept of dissociation. The distinction is diagramed in simplified form
in Figure 2. It may be noted that there is no necessary quarrel between
the amnesic dissociation and a dissociation caused by repression, only
that they are in some respects different. Under other circumstances they

f Amnesic Barrier

z Not
o Cs Available
~ and to
Consciousness to
Pcs Consciousness
u ~======~======~ ____ Repression
~ Barrier
~ Available to Consciousness
~ Ucs Only Indirectly

FIGURE 2. A distinction between the divisions of consciousness in dissociation and in
psychoanalytic theory. In psychoanalytic theory (simplified for this purpose) the available
memories lie in the conscious (Cs) and the preconscious (Pes), while the hidden ones are
concealed under a repression barrier and lie in the unconscious (Ues). The unavailable
ideas are largely those bound up with affect and impulse, and they enter consciousness
only indirectly. In a dissociation through amnesia the split is among the usually available
memories, and the unavailable memories need have no special affective or impulsive

may overlap, as when, in a partial amnesia, it is shown that the targets

of the amnesia may very well be those predicted by repression theory
(Clemes, 1964).
Amnesia lies at the heart of fugues and alternating personalities. If
it were not for the amnesia, there would be less problem of integration.
We all play different roles under different circumstances, but because
our memories are intact we do not thereby suffer a sense of alternating
There is a special problem not raised by ordinary posthypnotic
amnesia, a problem common also in repression theory. The problem is
this: Can an experience be registered, so that information regarding it is
being processed, and at the same time that experience be deflected from
consciousness before it has ever been conscious? This deflection in ad-
vance of awareness is something other than an afterexpulsion and, if it
occurs, raises additional theoretical problems.

This deflection into amnesia before the event is conscious appears

to take place in experiments on hypnotic analgesia. Subjects who are
analgesic report no pain or suffering whatever; it is not as though they
have forgotten a pain after experiencing it (as is found in some cases of
barbiturate sedation), but the pain is simply not felt at all. It can be
shown, however, by an automatic talking technique (related to the
automatic writing that Janet used) that the subject did indeed feel the
pain and processed it at nearly its full intensity (Hilgard, 1973; Knox,
Morgan, and Hilgard, 1974; Hilgard, Morgan, and Macdonald, 1975).
These results indicate that it is possible to have an experience regis-
tered, processed, and stored in memory in a recoverable form, even
though it has never been consciously felt or reported. Of course there may
be some arguments about the possibility of a progressive amnesia or
intermittent amnesia, and these will have to be faced when explana-
tions are considered, but the facts are as stated: recoverable amnesia is
evidenced for an experience of which the subject has never been aware.
Although this kind of dissociation is dramatic in experiments on
pain, it is by no means limited to pain. It has been known informally for
a long time that part of the hypnotized person processed information
more accurately than the hypnotized consciousness knew. William
James devoted several pages to an account of gaps in consciousness,
with evidence that the mind is active even when the person afterwards
ignores the fact (James, 1890, Vol. 1, pp. 201-213). He pointed to
experiments of Janet and of Binet, showing that hysterics with anes-
thesia could be shown under another condition to be sensitive. An
anesthetic person, ordinarily unable to distinguish between the two
points of a compass as in experiments on the two-point threshold, is
able to discriminate as accurately as anyone else by way of automatic
writing. James recorded his own experiment: "In a perfectly healthy
young man who can write with a planchette, I lately found the hand to
be entirely anesthetic during the writing act; I could prick it severely
without the Subject knowing the fact. The writing on the planchette,
however, accused me in strong terms of hurting the hand" (p. 208). We
have been able to demonstrate in our own laboratory that hypnotic
blindness and hypnotic deafness, as well as positive hallucinations, can
all be penetrated by automatic talking. That is, the "hidden observer,"
as we metaphorically describe the part out of normal awareness, tells us
the actual physical situation, the numbers that were not seen, the

sentences that were not heard, the nothingness that was seen as a
playful dog.
The hypnotic experience does not stop with this double conscious-
ness, one of distorted reality as perceived within hypnosis, the other of
undistorted reality as described in automatic talking. There is a third
split that can best be described as an observing consciousness, so far as
its conscious role is concerned. By inference it is also an active monitor-
ing and controlling agent. In its phenomenal role the observing con-
sciousness is present during hypnosis, knows whether the person feels
hypnotized or not, is sometimes surprised by how effective the hypno-
sis is, is disappointed when the hypnotic suggestions do not work.
Subjects often describe the situation as if the hypnotized self were
taking the center of the stage, performing the acts called for by the
hypnotist, while the observing part is in the wings, watching what is
going on. This observing part does not have access directly to the
hidden observer and hence is separated from it by an amnesic block.
At the same time it may be inferred that this observing, monitoring,
and controlling part is not powerless; when the hypnotist desires access
to the hidden observer, there must be some central controlling part that
allows this. In self-hypnosis, it must be the observing-monitoring part
that gives the self-suggestions that the hypnotized part then carries out.
There are puzzling metaphors in this kind of commentary on what goes
on, but the puzzles cannot be resolved satisfactorily by shortcuts such
as labeling everything as role behavior or by assuming that everything
is confabulated.
The hypnotic model, as I have described it, is not an explanatory
one. It serves to dramatize the phenomenal or inferred splits that a
theory has to account for in more general terms.



With the aid of certain basic assumptions we can construct a

generalized model of cognitive control systems, holding the evidence
from hypnosis in view but moving beyond that evidence. The first
assumption is that there are many subordinate cognitive structures,
each with a degree of unity, persistence, and autonomy of functioning.

The concept of unity of the total consciousness is an attractive one, but

it does not hold up under examination; there are too many shifts, as, for
example, between the waking consciousness and the dream conscious-
ness. There are also degrees of automatization achieved with practice,
so that well-learned habits-such as playing a musical instrument,
driving an automobile, or saying the alphabet-can go on with a
minimum of conscious control once the activity is begun. It is impor-
tant to note, however, that all these available activities do not go on all
of the time or all at once; hence there must be some method of inhibit-
ing them, on the one hand, and of facilitating them, on the other. The
second assumption is that there is some sort of hierarchical control that
manages the competition between these structures; if not, there would
be a veritable deluge of thoughts and actions going on all of the time.
This hierarchy must have a controlling or regulatory mechanism, how-
ever, or the hierarchies could not shift appropriately to the demands
being made upon the person. Hence in addition to multiple structures
in some sort of hierarchy, it is convenient to make a third assumption of
some sort of central monitoring and controlling structure. In my first
attempt to call attention to these essential features I presented the
diagram of Figure 3.
This highly generalized diagram was designed to convey the idea
of multiple structures (of which only three are shown), arranged in
hierarchical order, as suggested by their positions in the chart, each
with appropriate inputs and outputs and with multiple feedbacks
among them. At the top is an executive ego or central control structure
that has the planning, monitoring, and managing functions that are
required for appropriate thought and action. So that the appearance of
favoring excessive freedom on the part of this executive could be
avoided, another box was added to indicate that the executive system
has constraints upon it. The assumption was made that suggestions
from the hypnotist might influence the executive function and perhaps
change the hierarchical arrangements within the subsystems.
Before criticizing and amending this diagram, I wish to call atten-
tion to the many other psychologists who have found it desirable to
consider subordinate cognitive structures of one kind or another. With-
out being exhaustive, I shall mention six kinds of concepts closely
related to the cognitive control structures of Fig. 3. My purpose is to
show that this part of my analysis is closely allied to other develop-
ments within general psychology.

Constro i nts Executive Ego;

on Ego Autonomy
(Including Hypnosis) Central Control Structure

FIGURE 3. Subordinate cognitive control systems, in a hierarchical order subject to

change, under a dominant executive ego or central control structure. (From Hilgard, 1973,
p. 405.)

1. The term cognitive structure was made familiar by Kurt Lewin

(1935) and Edward Tolman (1932). Such a structure has much in com-
mon with my substructures; it may be pervasive, but a single structure
is never all-embracing, and there are problems of communication be-
tween them.
2. One of Clark Hull's central concepts was that of habit-family
hierarchy, in which a number of habits (each of which may be consid-
ered a small substructure) permit the organism to achieve its goals in a
given situation, and these small substructures are organized in a prefer-
ential or hierarchical system, so that if one is blocked the next is
activated (Hull, 1934). Berlyne (1965) later translated Piaget's theory into
these terms.
3. Donald Hebb's conception of cell assemblies serves to provide a
physiological substratum for cognitive structures (Hebb, 1949), and in a
talk before the Canadian Psychological Association in 1974 he noted a
possible coordination between his proposals and the existence of the
"hidden observer" as I have described it (Hebb, 1975).
4. The images and plans of Miller, Galanter, and Pribram (1960) are

their selection of levels of analysis that would provide for control of

thought and action; these, too, may be considered to be substructures.
5. The subordinate ego-structures of Gill and Brenman (1959), with a
dominant ego, represent an interpretation similar to mine, at least
within hypnosis. In a larger setting, the various ego-apparatuses of
Hartmann (1958), especially those in his conflict-free ego sphere, are
readily assimilated.
6. Sarbin's roles can also be considered to be cognitive substruc-
tures (e.g., Sarbin and Coe, 1972). The important problem of the pres-
ent theory is, however, to provide for the hierarchical management of
roles and for their inhibitions and facilitations.
The nature of the central control structure is an important but
troublesome problem. The extreme possibilities are that there is a very
powerful central control (replacing the older concept of a strong will), or
that there is none at all, that the hierarchy is determined by a competi-
tion among the parts for the control of the final common path. For many
years psychologists evaded the problem of a planning self, so that, in
essence, the second of these alternatives was accepted. To the extent
that man is controlled by external stimuli, and the habits conditioned to
them, what he does will be the compromise behavior that adapts to the
totality of these forces upon him. It was not only the stimulus-response
psychologists, however, who evaded the problems of a central mecha-
nism. William James, whose pluralism led him to make a number of
forays into problems of control, not always self-consistent, at one point
assigned the role of the thinker to the passing thought. He was given to
some excesses, especially when demolishing the views of those he
opposed, in this case the transcendental ego of the post-Kantians: "Our
'thought'-a cognitive phenomenal event in time--is, if it exists at all,
itself the only thinker that the facts require" (James, 1890, Vol. 1, p.
369). He went on to say a little later: "We may sum up by saying that the
personality implies the incessant presence of two elements, an objective
person, known by a passing subjective Thought and recognized as
continuing in time. Hereafter let us use the words Me and I for the
empirical person and the judging Thought" (Vol. 1, p. 371). All of this is
in the chapter on the consciousness of self, in which he also gave
favorable accounts of Janet's case of Leonie and of a fugue state with
which he himself had dealt, that of the Reverend Ansel Bourne. It is
hard to see how these larger unities could be accepted solely because of

persistence in time and the passing thought as the thinker, but this
extreme possibility cannot be ignored. In different words, this view
asserts that all behavior is merely a modification of the entering behav-
ior, and that way of putting it sounds more modem than James.
The idea of a control system, however, also finds contemporary
expression. If one examines the concept of planning, it appears that a
sort of planner must be inferred. Even a simple matter such as making
an appointment for luncheon next week is negotiated with those in-
volved, is written down, and is then acted upon at a later time. This
planfulness controls the possible behavior on that future date quite
effectively. The many competing thoughts that occur the morning of the
luncheon do not appear to be as determinative of what is going to
happen in the midday as the plans that were made by some responsible
part of the cognitive apparatus during the prior week. Appointments of
this kind are kept with a very high probability-perhaps as high as 90%
of the time--so that the planning function must be taken seriously. The
event may be trivial, but its implications are not.
The support for a special executive function has come into the open
from an unlikely source--the computer. Heuristic computer programs
commonly have an executive program that monitors the computer's
attempts to solve problems (e.g., Newell and Simon, 1972). If one
direction goes on too long without reaching a solution, the executive
calls a halt and a new direction is taken. This close analogy to what a
thinker does makes the idea of an executive ego a plausible one.
The diagram of Figure 3 is therefore all right as a suggestion of
some of the features of cognitive control systems, but it has to be
developed in far greater detail in order to lead to decisive experimenta-
tion. For one thing, the indication of separate inputs and outputs for
each system does not take into account that the organism has a limited
set of receptors and effectors and that the subsystems will be in compe-
tition for information sources through these receptors. There is compe-
tition also for the final common path leading to action and for cognitive
capacities that are limited. To make this more concrete, we may exam-
ine the kind of thing that happens when a person is immersed in one
activity, controlled by one structure, and a strong invitation to another
activity intrudes. To what extent will the conflict be resolved at the level
of these structures themselves, and to what extent will a higher-order
control process intervene?

Suppose that someone is playing a well-practiced piece on the

piano, without an audience (so that the social pressure to continue is
reduced), when the selection is interrupted by a ringing telephone. He
can hear the phone ringing without interrupting his play and can begin
a conversation with himself, along the lines: "If I weren't home, I
wouldn't answer; if it's important they will call again. I shall go on
playing." The phone keeps on ringing intermittently, and he decides to
interrupt his playing and to answer. How do we analyze this behavior,
in terms of the cognitive systems involved and the central control
For convenience, we shall consider Cognitive Control Structure 1 to
be the behavior related to playing the particular learned selection on the
piano and Cognitive Control Structure 2 to be the habit of answering
ringing telephones. These come into conflict, and the question arises
whether the conflict can be resolved according to the relative strengths
of these two systems, or whether some additional decision-mechanism
is needed. A diagram of these relationships, which may be considered
to be a further specification of Figure 3, is given in Figure 4.
Playing the piano has now come so high in the hierarchy that it
dominates the behavior of the piano player. Whose decision was it to
play the piano at this time? Presumably there was some tendency to
expression inherent in the background of piano playing, so that the
performance was ready to be emitted; still, it had to be "permitted" to
play against other competing possibilities. Here a central mechanism,
involved in managing the daily activities, may have exerted an influ-
ence. Once the piano playing gets under way the central monitor steps
aside and is a relatively passive observer, except for vigilance regarding
other plans, for example, that would call for terminating the piano
playing at a set time, even though the "piano playing part" might favor
continuing. Now the ringing telephone intrudes. There is an arrow
pointing from this new input to the piano playing input because this is
an uninvited interruption. The new input arouses the second cognitive
system that has habits regarding telephone answering, and a seIf-
conversation may ensue with little interruption of the piano playing if
that is well automatized. The alternatives diagramed are either continu-
ing to play or interrupting in order to answer the phone. The outputs
are incompatible; there is a competition for the gross musculature that
will either keep the person at the piano or move him to the telephone.
The issue now is whether the conflict can be settled at the level of these

Central Monitor

Input Cogn. Control * I Planned Output A

Visual- Play Piano

Musica I Score;
Keyboard II Playing Piano
Planned Output B
Kinesthetic Stop Ploying

Planned Output A

Auditory - Do not Answer

Ringing Phone, Answering
Repetition of Telephone
Planned Output B
Alternation Answer Phone

FIGURE 4. Resolution of a conflict between two cognitive systems: piano playing and
telephone answering. At issue is the question whether the conflict between the two
systems can be resolved by their relative strengths, or whether some intervention is
needed by the central monitor and control.

two competing cognitive control systems, or whether the central control

mechanism becomes involved. 1 The planning boxes have been added
merely to call attention to the idea that some decision time is involved
during which reflection upon the alternatives is possible. In the case of
the telephone this is fairly short, because the caller may hang up and
the phone stop ringing. In other cases the monologue may be more
protracted. In a related case that came to my attention last summer, an
organist became physically tired after having played most of the day in
which a college baccalaureate and a commencement program occurred
on the same day, with an organ recital between them. Part of the fatigue
was in his legs, which he had to hold up much of the time in order to
play on the foot levers of the organ. In the college hymn he foresaw
some difficulty in moving his right leg and foot at a climactic part in

1 More complex motivations may enter, such as a preference for self-initiated activities as
against those externally imposed, perhaps along with some unconscious ones based on
the individual life history. For the purposes of illustration, the conflict is treated at a
simplified level.

which a move from the left extreme to the right had to be made. He told
me that he talked to himself a lonb time about this, while playing right
along, trying to decide whether or not he could do it accurately and
what would happen if he omitted it. He decided to omit it, lest he make
an error, and so far as I know nobody detected his omission. Does the
central control have to be involved? This is somewhat problematical
when everything is in the open, as in these illustrations, but many
aspects of memory and of plans unrelated specifically to telephones and
piano playing enter the picture. The ultimate weight of factors influenc-
ing the decision may well go beyond the two structures primarily
involved. One way of putting it is that if there were an identifiable
central agency, the chance of intrusion would be great. Here again the
hypnotic model is of some pertinence: in a practiced hypnotic subject
the monitoring part could simply assure not listening to the telephone
by a self-suggestion not to hear it. This favors the piano player, but it
does not seem to be something on which the piano-playing system
alone would take the initiative.
I have deliberately chosen an illustration in which the question of
consciousness is secondary to the question of control. Once the notion
of multiple controls is familiar, the possibility of inhibited cognitive
structures becomes more plausible, and then one is ready to consider
splits in consciousness.
Figure 4 supplements Figure 3 by making more explicit the manner
in which cognitive control structures compete, both in attention to
stimulus inputs and for the final common path in outputs, and it shows
circumstances in which the central mechanism might conceivably step
in to influence a decision by throwing its weight (presumably via
information from other sources) to resolve the conflict. This begins to
get close enough to concrete situations that one senses the empirical
problems that may be amenable to study, but it is not yet specific
enough to define a program of research.



As noted earlier it is somewhat easier to approach these large

problems by breaking them down into topics that lend themselves to

experimentation, even though some of the issues may be bypassed

For example, the study of divided attention is very relevant to the
issues being discussed here. One theoretical issue is whether or not
when one of two channels of information is suppressed the suppressed
channel is filtered out before registration or perhaps processed but the
information not perceived. The former point of view has been made
familiar in the filter theory of Broadbent (1958). Although Treisman
(1969) permitted the information to move somewhat deeper into the
processing mechanism, her views are not very different from Broad-
bent's on the issue of recognition and conscious perception. The ex-
treme position was taken by Deutsch and Deutsch (1963), who said
essentially that the information is processed in some kind of recogni-
tion system but diverted prior to conscious perception. For my purpose
it is sufficient to call attention to the large literature bearing on these
problems (e.g., Norman, 1969; Moray, 1970; Kahneman, 1973) and to
note that at least one of the theories, that of the Deutsches, hypothes-
izes the very kinds of processes that our experiments on neodissocia-
tion have revealed: information that is incorporated within a cognitive
system but does not reach the conscious level until special procedures
are used to bring it to subsequent awareness.
Investigations of attention characteristically study the division of
attention between tasks, because a central problem is that of the parallel
versus the serial processing of simultaneous inputs. Kahneman (1973)
has pointed out that there has been some neglect of capacity models,
which should explain how, when there is enough capacity, more than
one cognitive activity can be engaged in, although if some task requires
too much cognitive effort it excludes others. Such considerations are
obviously relevant to the hypnotic dissociation experiments in which
simultaneous tasks have to meet the special requirements of hypnotic
suggestions, e.g., performing one task out of awareness while the other
task is performed with full awareness. Kahneman's discussion of task
interference does not consider hypnotic behavior, but it is relevant to
the demands made upon subjects in hypnotic experiments because
these often strain their cognitive capacities as they attempt to comply
with the demands made upon them.
Studies that bear on recoverable amnesia, whatever the source of
the amnesia, are pertinent to neodissociation theory. The literature of

psychopathology furnishes many clinical illustrations, and these must

be incorporated into psychological science. There are also experimen-
tally induced amnesias, outside hypnosis, that deserve consideration.
Some of these have been studied in the context of state-dependent
learning. If something learned in one state, as while under the influence
of a drug, is forgotten in the nondrugged state but recalled again in the
drug state, that is an experimental illustration of a recoverable amnesia.
This arrangement is, of course, the paradigm of state-dependent learn-
ing. The literature has been reviewed by Overton, who is also one of
the leading investigators in the field (Overton, 1972, 1973).
An alternative method of studying the differential effects of the
drugs has been widely used, especially with animal subjects. For exam-
ple, a rat can be taught to tum right in a T-maze under the influence of a
drug and to the left in the normal, nondrugged state. According to
Overton (1973) drug discrimination can be established more rapidly
than most sensory discriminations, and it is doubtful that the discrimi-
nations involve the sensory consequences of the drug action. Instead he
believes that the differential responding may be based "on the dissocia-
tive barrier which impairs a transfer of training between the drug and
no-drug conditions." The concept of dissociation employed here is
consonant with that which I am using, that is, two types of behavior
control through information isolated from each other.
There are a number of experiments with human subjects which
indicate the possibilities. Several anesthetics are "dissociative" in that
they permit the patient to be analgesic but conscious and perhaps to
remain amnesic for the operation after the experiment is over. Ketam-
ine is such an anesthetic in common clinical use; patients are not
unconscious, their cardiac functions and respiration are not depressed,
and they can reply to the surgeon's questions. They may have some
delirium, possibly as an accompaniment of the memory disorder; after
the drug wears off the patient is amnesic for what has happened.
Unfortunately there are few studies in which there has been any serious
effort made to see if the memories are recoverable.
In one study in which I participated, in which the drug thiopental
(trade name, Sodium Pentothal) was used with volunteer normal
subjects not undergoing surgery, the subject while intoxicated with
the drug was able to learn simple paired associates and to demonstrate
the capacity for picture recognition, but after a few hours' rest during

which the drug wore off he was commonly amnesic for all that had
happened and showed no retention of the material learned. However,
when the subject was then hypnotized and asked to recall, there was
some slight recovery of memory, especially in the more hypnotizable
subjects (Osborn, Bunker, Cooper, Frank, and Hilgard, 1967). State-
dependent learning was not demonstrated, in that under the drug
there was no forgetting of material learned in the prior waking state,
and when the drug was readministered there was no recall of the
material learned in the earlier session and later forgotten. The experi-
ment should be repeated, especially with selected highly hypnotizable
subjects, because some pretest subjects, who were more hypnotizable
than those used in the experiment, did show more dramatic recovery
of their memories for things learned in the drugged state. There have
indeed been some reports of the recall of things said by the surgeon in
ordinary general chemoanesthesia, when later tested by hypnotic
methods (Cheek, 1966; Levinson, 1965). In these experiments the
action of the drugs was the primary consideration; hypnosis was used
merely as a method of inquiry to determine whether or not the
amnesia produced was recoverable. It makes a difference whether the
absent memories have simply not made their way into the permanent
memory store or whether they are present and in some sense disso-
ciated and unavailable to recall by ordinary methods.
One would like to know the physiology behind these states, but
the physiology of consciousness is elusive in any case. One lead that
has proved interesting is the differential function of the two hemispheres,
which, at least in the split-brain studies, represents a massive dissocia-
tion of two brains in one head (Sperry, 1968; Gazzaniga and Sperry,
1966; Gazzaniga, 1970). As these studies are carried over into the study
of normal laterality of function, it is quite possible that additional
aspects related to dissociation will emerge. For example, it has been
shown that in well-Iateralized right-handed males, a preference for
right-hemisphere function is associated with hypnotizability (Gur and
Gur, 1974). The distinction has to be made between specialization of
function and preference, for those who consistently do "right-hemi-
sphere tasks" with evidence that they are using the right hemisphere,
and "left-hemisphere tasks" with evidence that they are using the left
hemisphere, do not by these signs indicate hypnotizability. Given both
kinds of tasks, the hypnotizable right-hander tends to use his right

hemisphere more frequently than the nonhypnotizable right-hander,

that is, he shows a preference for using that hemisphere. This is all
somewhat puzzling, and there may be some intermediaries to provide
the explanation, such as a preference for the use of visual imagery when
presented with a problem.
The dissociations within sleep provide another source of information
that does not rely on extreme pathology. For example, the studies on
sleepwalking and sleep talking (somniloquy) illustrate states in which
considerable control may be manifested, including interactions with
the geographical and social environment, yet forgotten when the per-
son has returned to normal sleep (Arkin, Toth, Baker, and Hastey,
1970). The dreams recalled from the night are not usually connected
with these episodes. Of course night dreams themselves remain inter-
esting as dissociated activities.
The evidence from psychopathology ought to be included in any
account of dissociative phenomena. Although fugues and multiple per-
sonalities are rather infrequent, now that clinicians are again on the
lookout for them, more are being found. After a dearth of cases of
multiple personality, several have been reported since the case reported
as the Three Faces of Eve several years ago (Thigpen and Cleckley, 1957;
the new ones include Ludwig, Brandsma, Wilbur, Bendfeldt, and Jame-
son, 1972; Schreiber, 1973; Stoller, 1973). Others, unpublished, have
come to my attention. Long ago Hart (1929), who had himself done
much to popularize psychoanalytic concepts, felt that psychoanalysts
had failed to take note of dissociation and double personality and that
they should overcome this neglect. The dearth of cases during the
height of psychoanalytic popularity could have resulted, if Hart was
correct, because a neurotic who might have demonstrated these symp-
toms was most likely to be treated by an analyst. If at the first sign of a
split the analyst began to seek an integration of the parts (perhaps quite
properly from a therapeutic standpoint), the extent of the dissociation
might not have become fully revealed.
The most interesting and puzzling feature of multiple personalities
is nearly always the distribution of memories among the components.
In Janet's case of Leonie, for example, Leonie I knew only herself,
Leonie II knew Leonie I as well as herself, and Leonie III knew them all.
In the more recent case of Jonah, the primary personality (Jonah) was
not aware of the three others, though they were all aware of Jonah and
dimly aware of each other (Ludwig et al., 1972).



The consciousness of pain can serve as an illustration of specific

problems that a neodissociation theory is designed to deal with. An
advantage of this area of investigation is that the neuroanatomical
problems, if not all resolved, are at least well identified because of the
many efforts to relieve pain by surgery and drugs and by non drug
procedures, of which hypnosis is one.
The basic facts of supra threshold pain can be summarized as
1. Felt pain has two components related to a source of stimulation,
of which one is the sensory-information aspect, which makes pain very
much like other sense-perception experiences. The sensory pain can
usually be localized, quite precisely, as in cutaneous pains, or at least
regionally and often laterally, as in toothache, headache, or abdominal
pains. Sensory pain increases with the intensity of stimulation and can
be studied by standard psychophysical methods. The second compo-
nent is variously described as the reactive component (e.g., Beecher,
1959) or the motivational-emotional component (Melzack and Casey,
1968). The second component is also referred to as anguish or suffering,
to distinguish between this aspect and sensory pain. We may temporar-
ily neglect a third component, the central or cognitive control of pain.
There are some issues here reminiscent of the James-Lange theory
of emotion. The reactive interpretation assumes that the sensory pain
comes first in time and the suffering follows upon it, just as in the
James-Lange theory the visceral or motor responses were said to come
first and the emotion to come as a repercussion from them ("1 am afraid
because I run"). The Cannon-Bard criticism was that both the re-
sponses and the felt emotion might be released at once by way of
different systems, so that the two aspects could be viewed as concurrent
rather than as successive. The same issue arises in whether pain and
suffering are concurrent or successive; before we decide that issue, it is
preferable to note the distinction between the two aspects and to await
the evidence from experiments.
2. The two components have an anatomical basis in separate nerve
pathways in the spinal cord and higher regions, with differing central
connections. There are still some ambiguities here, but in general the
sensory conduction system is by way of neospinothalamic fibers to the

thalamus and the somatosensory cortex, and the motivational-affective

conduction system is by way of medially coursing fibers comprising a
paramedial ascending system reaching the reticular formation, the me-
dial intralaminar thalamus, and the limbic system (e.g., Melzack and
Casey, 1968; Casey, 1973). If these two systems are recognized, then the
possibility of the simultaneous occurrence of pain and suffering has to
be taken into consideration before commitment to a totally reactive
position with respect to suffering. Some difficulties are encountered in
the clearing up of all of the issues because the sensory conduction
system is indeed the more rapid one, and even though the motiva-
tional-emotional system has been activated concurrently with the sen-
sory system, it is quite likely that the suffering is in some way moder-
ated by feedback from the prior sensory information. Hence the
motivational-emotional system may be partly aroused in parallel while
also being reactive to the sensory component.
3. The interactions between the two components may be either
integrated or dissociated. The possibilities of integration have been
well explored in the gate-control theory of pain (Melzack and Wall,
1965). This theory proposes that an interaction between the two sys-
tems occurs at the spinal cord level, in the substantia gelatinosa of the
dorsal hom. The anatomical basis is provided by collaterals from the
large fibers of the sensory-information system and the small fibers of
the motivational-affective system entering through the dorsal roots of
the cord. A further control is exercised by central processes that operate
centrifugally, sending impulses down to the gate from above. If large
fibers are stimulated simultaneously with small ones, the theory pro-
poses that the gate should be closed and suffering reduced; this impli-
cation found immediate practical application in the control of deep pain
by superficial stimulation (e.g., Wall and Sweet, 1967). It may be that
this is one of the routes by which acupuncture is effective; it also offers
some support for the folk practices of relieving deep pain by liniments
and mustard plasters that produce surface stimulation of the large fibers
in the regions in which small-fiber stimulation is occurring.
The two systems may also be separately affected. One of the
observations made when lobectomy was used in the attempted relief of
intractable pain was that patients often reported that they still felt the
pain but that it no longer bothered them (Freeman and Watts, 1950).
The cutting of the paths in the prefrontal area was probably responsible

for destroying some of the responsiveness of the motivational-affective

pain system while leaving the innocuous sensory-information process-
ing intact. The separateness of the two systems allows for psychological
dissociati~n because nonsurgical procedures may modify one system
more than the other.
4. Pain mechanisms are incompletely described by any explana-
tion that relies exclusively on a segmental interpretation of impulse
transmission and local signs. The notion of dermatones related to
specific segments of the spinal cord is so firmly ingrained that well-
established alternative distributions of pain are treated as anomalous, if
not ignored altogether. The first of these anomalies is referred pain, that
is, pain that is felt in some part of the body not the source of the
irritation. A referred pain violates the principle of local signs that
usually locate the source of stimulation. Referred pains are well known,
and diagrams can be found in most neurology texts. Often they are
referred within the same segment, as when the pain of a diseased
kidney is felt as a pain in the normal one or anginal pain is felt in the
arm. But the pain may be in a different segment, as when a lower-back
injury may be felt in the eye region or a knee injury in the wall of the
chest. An interesting set of observations concerns so-called trigger
points, that is, points which when stimulated give rise to referred pains
at a distance. If these points are anesthetized-or in some cases hyper-
stimulated-the referred pain may be reduced (Travell and Rinzler,
Rather than being viewed as anomalies, these anatomical complex-
ities should be incorporated into our understanding of the total neuro-
physiology of pain perception. Some of these nonsegmental indications
are very dramatic, as in the case of a patient with a transected cord
whose right leg is totally analgesic, yet who feels a pain on the left side
in the abdomen when a particular area of the analgesic right leg is
stimulated (Nathan, 1956). The autonomic system may be involved
when the referred pains do not follow usual segmental somatic pat-
terns. The nonsegmental aspect of hysterical anesthesias, such as glove
anesthesias, or corresponding nonsegmental changes induced by hyp-
nosis become more acceptable when we recognize the many anomalies
of pain in varied contexts.
5. Another fact of the pain experience is the persistence of pains,
whether referred or not, after the source of irritation no longer exists.

Phantom limb pains fall into this category, that is, the pains that are felt
in an amputated limb that no longer exists. The possibility that this is
merely an illustration of the "law of specific energies," with pain
produced through stimulation of the cut ends of nerves formerly serv-
ing the limb, is not a sufficient explanation, for the pain may persist
even when the stump is anesthetized. Melzack (1973) points out that the
pain is seldom felt unless there was pain in the bodily member prior to
its loss. Hence there is perhaps a memory component that assigns the
pain to the phantom. The persistence of pain reduction, as in the
continuing effects of superficial stimulation of large fibers after the
stimulation ceases, also has to be accounted for by explanations that go
beyond any immediate sensory process of stimulation and response.
The explanation again must be in terms of some sort of persisting
process, whether peripheral or central.
There is much more that might be said about pain; for example,
many social and psychological factors are known to influence it. There
is the athlete who completes the game without being aware of his injury
until the game is over, or the wounded soldier who does not express
pain until stuck by a hypodermic needle (Beecher, 1956). The modem
"pain clinic" takes advantage of these facts and has numerous methods
by which to ease the pain of patients without the use of drugs or
surgery. Sternbach (1970, 1974) has listed a number of these such as
desensitization, progressive relaxation, biofeedback and conditioning
methods, and strategems of distracting attention, as well as hypnosis.
Our earlier experiments on laboratory pain did not distinguish
between pain and suffering, confusing the issue slightly by using an
aversive component in defining the pain scale as one on which 10
would be a pain so severe that the subject would wish to terminate it.
This did not matter for the original purposes, but it became important
when the experiments were extended to the study of the distinction
between pain and suffering.
The first published experiment from our laboratory which clearly
used the distinction was that of Knox, Morgan, and Hilgard (1974). The
experiment employed two reporting scales, one for sensory pain, one
for suffering. In the ordinary nonanalgesic hypnotic condition, the
subjects were asked for separate reports of pain and suffering at 2-
minute intervals. They found it easy to make the distinction; both pain
and suffering rose with time of ischemia, the suffering mounting more
slowly than the pain, as shown in Figure 5. To determine whether there



Critical Level


c 6
Sensory Pain
0--0 Hypnotic Verbal (Open)
...-- Automatic Talking (Hidden)
2 Suffering
0- -{) Hypnotic Verbal (Open)
... -e Automatic Talking (Hidden)

o 2 4 6 8
Minutes of Ischemia

FIGURE 5. Reports of pain and suffering in ischemia in the hypnotic nonanalgesic state.
Reports of pain and suffering were obtained every 2 minutes in the usual way and plotted
as Open reports. Immediately following each open report, a report was obtained by the
method of automatic talking and plotted as Hidden reports. In the hypnotic nonanalgesia
condition the two sets of reports are essentially alike, indicating no amnesic effect on the
open reports. (From Knox, Morgan, and Hilgard, 1974, p. 844.)

might be some amnesiac component even in the nonanalgesic report by

these highly hypnotizable subjects, a report by the "automatic talking"
method used for studying experiences concealed under amnesia was
used; the reports were practically identical, as shown by the overlap-
ping curves in Figure 5, so that there was no evidence for an amnesic
component under the nonanalgesic condition.
When the corresponding reports were called for in hypnotic anal-
gesia, the verbal report obtained in the usual way in hypnosis was of no
pain and no suffering throughout; the hidden report, of the experience

concealed under amnesia, was of almost full pain and suffering. This
was puzzling because there were no overt signs of suffering within
hypnotic analgesia, and retrospective accounts following hypnotic anal-
gesia with other subjects had indicated that the concealed pain was felt
normally in hypnotic analgesia but that it was not aversive. Some
doubts were expressed in the report about the findings. One possibility
noted there was that the hypnotic nonanalgesia may have itself resulted
in a quiescent subject, so that although the pain was normal, the
suffering was not very severe to begin with. Perhaps the report of
suffering was inferred to be the amount of suffering that ought to be felt
for that amount of pain, in view of the definition of both scales as
converging at a pain state of 10, for at that point the subject would find
the pain (hence the suffering) sufficiently aversive as to want the
experiment terminated. Owing to the repeated call for concealed reports
while the stimulation was occurring, the subject might have adopted a
practice of "sampling" the sensory pain by momentarily turning off the
analgesia and then turning it back on, inferring what suffering would
accompany that amount of pain if the analgesia were broken for a
longer time. In any case, it seemed desirable to repeat the experiment in
some other manner. A few pilot subjects have shown what the results
are likely to be; the new experiments are still in progress and a full
account is not yet available, but the direction of the results differs
enough from the reported experiment to justify some comments.
The pilot subjects of the new experiment have been run first in a
normal waking condition, rather than a nonanalgesic hypnotic one,
with reports of pain and suffering at 30-second intervals. They have
responded as the earlier subjects did, by reporting pain and suffering
both increasing with time in ischemia, with the suffering less than the
concurrent pain. Then, in another session, the ischemia was continued
with hypnotic analgesia for a sufficient time after exercising for the pain
and suffering to have become quite aversive, according to the reports
obtained without analgesia. The earlier findings were repeated: no pain
and no suffering were reported. After the ischemic arm was restored to
normal, the subject, while still hypnotized, was questioned in the
manner that provides the hidden report, breaking the amnesia for the
experience. The consistent finding with two pilot subjects has been that
the hidden sensory pain was essentially normal, reaching the level of
the non analgesic day, but the suffering was entirely absent. This contra-
dicts the results of Knox et al. (1974) but is more coherent with other

observations made in the laboratory. The discrepancies point to the

hazards of interpretation in this kind of experiment, but presumably
further experiments, with the same subjects serving under various
aITangements, will eventually resolve the discrepancies. At least for
these pilot subjects, we have some complex results according to which
to test the theory of duality of pain and the role of hypnosis in relation
to it.
The situation is characterized in Figure 6. The normal waking
nonanalgesia pattern of consciousness is shown on the left, with the
central control permitting the full experience of pain, and the central
monitor then noting the pain and suffering felt and reporting it ver-
bally. "Permitting" pain is said advisedly, because these subjects need
not have felt pain and suffering unless they had consented to. The
diagram on the right shows what is happening in a concealed way
while the central monitor is reporting no pain and no suffering. The
central control system has been responsible for accepting the hypnotic
suggestions for pain reduction, which includes setting up the amnesia
barrier against reporting what the hidden consciousness is experienc-
ing. Only when the amnesia barrier is broken by the "automatic-
talking" procedure does the report of the hidden material find its verbal
expression. Separating the central control from the central monitor may
seem arbitrary, but phenomenally this is what takes place. Some aspect
of the person (the central control) permits the ascendance of the hyp-
notic responsiveness, and once it is operative, the monitor rather
passively reports the experiences that are being felt.
The lower boxes indicate that the involuntary physiological compo-
nents, such as the cardiovascular changes, reflect the physical stresses
much as in the waking state, but the voluntary responses of fist clench-
ing, squirming, and grimacing are absent in hypnotic analgesia. The
absence of the voluntary components may be due to less felt suffering,
or the reduction in responses may be responsible for reducing
the suffering; present data do not permit a choice between these
The bearing on the pain-suffering dichotomy may be formulated in
three assertions:
1. Hypnotic analgesia, on the part of highly responsive hypnotic
subjects, eliminated both sensory pain and suffering and hence did not
discriminate between these two components. The findings are clear
evidence, however, for central control of felt pain and suffering. This is


CENTRAL r------i Centro I Central

ASPECTS Control Monitor


SUFFERING Suffering Felt


FIGURE 6. The dissociation between the usual verbal reports in hypnotic analgesia and
those revealed through automatic key-pressing or automatic talking, interpreted accord-
ing to an amnesia barrier. The diagram on the left represents the situation in normally
experienced pain, with the central control permitting the pain and suffering; the central
monitor is aware of all that is going on. The diagram on the right represents the situation

in itself a dissociation between stressful stimulation and consciousness

of what is happening.
2. The hidden report, made after the amnesia barrier is broken,
shows that sensory pain is less reduced at this level than suffering. This
differentiation represents a pain-suffering dichotomy and is in that



Central Central
Control Monitor

L- - ---'1
Pain Felt JI
Suffering Reduced J
or not Felt

Components E


in hypnotic analgesia, the central control having set up the amnesia barrier, making the
experiences of pain and suffering unavailable to the central monitor. Only when the
central control releases the amnesia barrier (through the automatic methods) will the
experience of pain and reduced suffering be available to the central monitor.

way coherent with the physiological and anatomical evidence and with
one aspect of the gate-control theory. There is a dissociation here
between the normally simultaneous sensory pain and suffering.
3. Because the reduction in suffering is accompanied by a reduc-
tion in the voluntary components of the total reaction, those who favor

a reaction theory of suffering can assimilate this finding to their posi-

tion. That is, when sensory pain is registered but there are no voluntary
responses to its aversive character, the sensory pain may be assumed
not to lead to reported suffering. This interpretation could account for a
separation between pain and suffering even if there were only a single
afferent system involved. However, the data can also be interpreted on
the assumption that the afferent suffering system, relatively independ-
ent of the sensory pain system, may itself be responsible for evoking
the voluntary indications of suffering.
Now to summarize the bearing of these assertions on neodissocia-
tion theory. With respect to the sensory pain aspect, the bearing on
dissociation is clear, for here is a sensory experience registered within
some cognitive system, yet denied by another. The essence appears to
lie in an amnesialike process that is responsible for making unavail-
able the information about either pain or suffering in the usual hyp-
notic analgesia report. If suffering is reported in neither the usual report
nor the report of the concealed experience, then the amnesic aspect is
not relevant for suffering; now, however, the dissociation between pain
and suffering has to be accounted for on other grounds. The dissocia-
tion is still important because pain and suffering are usually found
together, and the fact of different physiological systems does not ac-
count for the hypnotic dissociation; the separate systems hint only at a
possible physiological locus for the effect of hypnotic analgesia.
To illustrate how the phenomena in this kind of experiment and
this kind of theory may be used to test specific hypotheses about pain
and pain controC I wish to mention a hypothesis suggested by Dr.
Avram Goldstein, which he and I are now engaged in testing. 2 As
indicated above, the dissociation of the pain and suffering systems that
shows in the hypnosis experiments may have a basis in the separate
conduction systems. What would be a possible mechanism by which
this might come about? It could come about if hypnotic analgesia were
to modify the central neural activity occurring in the medial portions of
the thalamus that are related to the suffering system but not to the
sensory pain system. It is known that destroying the medial thalamus
surgically may reduce the affective component of intractable pain with-

2 I am grateful to Dr. Goldstein for proposing the hypothesis, for participating in

carrying out the relevant collaborative experiment, and for permitting this reference to
it prior to its publication. (Reference added in proof: Goldstein and Hilgard, 1975.)

out destroying the sensory discriminative capacity (Mark, Ervin, and

Yakolev, 1963). Morphine injected into this area may have the same
effect. Interestingly enough, electrical stimulation in the same region
may also reduce pain. This was demonstrated by Reynolds (1969), who
operated on rats without their showing signs of pain while the electrical
stimulation continued; the rats then showed signs of pain when the
stimulation ceased. Corresponding findings in the cat have been re-
ported by Liebeskind, Guilbaud, Besson, and Oliveras (1973). Is there
any connection between morphine and electrical stimulation? From all
we know about neurohumoral transmission, there is a strong possibil-
ity that the electrical inhibition of pain depends upon a chemical
intermediary; perhaps this chemical intermediary has some pharmaco-
logical similarities to morphine.
We now have a background for Dr. Goldstein's hypothesis. Sup-
pose that the pain-reducing intermediary for pain reduction in medial
thalamic stimulation is morphine-related. How can this be found out?
One method is to use a known morphine antagonist to counteract
whatever agent is responsible for the pain reduction. When electrical
stimulation produces the analgesia, naloxone, a morphine antagonist
that is alone neutral with respect to pain, does in fact counteract the
pain reduction and hence gives support to the hypothesis (Mayer and
Hayes, 1973). Suppose, now, that hypnotic analgesia were to work in
the same way. Then it might be responsible for the release of a
morphinelike substance somewhere in the medial system; sensory
pain would not be reduced but suffering would, which is the result
reported in hypnosis for the hidden reactions to normally painful
stress. The hypothesis can then be advanced that naloxone might
counteract the hypnotic analgesia, as it counteracts the analgesias of
morphine and electrical stimulation.
The hypothesis offered by Dr. Goldstein was tested in a double-
blind experiment in which pilot subjects known to be responsive to
hypnotic analgesia were sometimes injected intramuscularly with nor-
mal saline solution as a placebo and sometimes with naloxone. If the
hypothesis were confirmed, no change in the amount of reduction of
pain and suffering under hypnotic analgesia as usually reported would
be required, because the underlying changes are masked by some sort
of amnesic process. After the release of the amnesia, by the methods of
exploring the hidden effects, a change with naloxone would be ex-

pected. The suffering, normally absent, would be expected to return if

naloxone neutralized a morphinelike substance responsible for the
reduction of suffering in the first place. The preliminary results have
failed to confirm the hypothesis; that is, there was no increase in
reported suffering after the amnesia was lifted, whether naloxone or the
placebo had been administered. Of course negative results are some-
what uncertain at this stage of the investigation, but at least no positive
effects of naloxone in reversing the suffering have thus far been found.
An important point to be made is that the ordinary procedures of
hypnotic analgesia would not have been critical for this experiment,
because even if the experiment had in other ways shown positive
results for the effect of naloxone, there would have been no effect on the
reduction of pain and suffering in the usual reports of hypnotic analge-
sia. The theoretical reason for the persistence of the analgesic effect is
that this usual report has an amnesic component in it, so that no pain
or suffering would be reported regardless of what was happening at
some other level of information processing. It was only under circum-
stances in which the dissociation of pain and suffering could be shown
that the experimental test became critical. As a result, a series of
experiments related to neodissociation theory opened up the possibility
of testing a neuropharmacological hypothesis. The experimental design
has permitted a crucial test; that the results thus far have been negative
is a matter of empirical fact. After all, the purpose of an experiment is to
find out what is so.


A new theory is valuable when its formulation calls attention to

problems that have been neglected and shows ways in which, with the
aid of the theory, the problems can be converted to experimentable
form in a manner that would be unlikely without the theory. In this
introduction to neodissociation theory I have sketched its historical
background and the kinds of general observations on which it is based.
The example of hypnosis in the investigation of pain and suffering was
elaborated in order to show in concrete detail the empirical conse-
quences of exploring phenomena in the context of neodissociation
theory. While hypnosis dramatizes the existence of dissociative phe-
nomena, neodissociation theory serves to call attention to many empiri-

cal problems on nonhypnotic topics in which dissociative aspects are

also prominent. The ultimate aim of neodissociation theory is to
achieve a coherent understanding of these diverse phenomena. 3


The preparation of this chapter and the conducting of the experi-

ments reported here were aided by Grant #MH-03859 from the Na-
tional Institute of Mental Health.


ARKIN, A. M., TOTH, M., BAKER, J., AND HASTEY, J. M. The degree of concordance
between the content of sleep-talking and mentation recalled in wakefulness. Journal
of Nervous and Mental Disease, 1970,151, 375--393.
BEECHER, H. K. Significance of wound to pain experienced. Journal of the American
Medical Association, 1956,161, 1609-1613.
BEECHER, H. K. Measurement of subjective responses. New York: Oxford University Press,
BERLYNE, D. E. Structure and direction in thinking. New York: Wiley, 1965.
BROADBENT, D. Perception and communication. London: Pergamon, 1958.
CASEY, K. L. Pain: A current view of neural mechanisms. American Scientist, 1973,61, 194-
CHEEK, D. The meaning of continued hearing sense under general chemo-anesthesia: A
progress report and report of a case. American Journal of Clinical Hypnosis, 1966,8,
CLEMES, S. R. Repression and hypnotic amnesia. Journal of Abnormal and Social Psychol-
ogy, 1964,69, 62-69.
DEUTSCH, J. A., AND DEUTSCH, D. Attention: Some theoretical considerations. Psychologi-
cal Review, 1963,70, 80-90.
FREEMAN, W., AND WATTS, J. W. Psychosurgery in the treatment of mental disorders and
intractable pain. Springfield, Ill.: Thomas, 1950.
GAZZANIGA, M. S. The bisected brain. New York: Appleton-Century-Crofts, 1970.
GAZZANIGA, M., AND SPERRY, R. Simultaneous double discrimination response following
brain bisection. Psychonomic Science, 1966,4, 261-262.
GILL, M. M., AND BRENMAN, M. Hypnosis and related states: Psychoanalytic studies in
regression. New York: International Universities Press, 1959.
GOLDSTEIN, A. AND HILGARD, E. R. Failure of the opiate antagonist naloxone to modify
hypnotic analgesia. Proc. Nat. Acad. Sci. USA, 1975, 72, 2041-2043.
GUR, R. E., AND GUR, R. C. Handedness, sex, and eyedness as moderating variables in the
relationship between hypnotic susceptibility and functional brain asymmetry. Jour-
nal of Abnormal Psychology, 1974,83,635--643.
HART, B. Psychopathology (2nd ed.). Cambridge, England: Cambridge University Press,

3 The sketch of neodissociation theory in this chapter is elaborated in Hilgard (in

preparation) .

HARTMANN, H. Ego psychology and the problem of adaptation. New York: International
Universities Press, 1958.
HEBB, D. O. The organization of behavior. New York: Wiley, 1949.
HEBB, D.O., Science and the world of imagination. Canadian Psychological Review, 1975,
16, 4-11.
HILGARD, E. R. A neodissociation interpretation of pain reduction in hypnosis. Psychol-
ogical Review, 1973,80, 396-411.
HILGARD, E. R. Divided consciousness. New York: Wiley-Interscience (in preparation).
HILGARD, E. R., MORGAN, A. H., AND MACDONALD, H. Pain and dissociation in the cold
pressor test: A study of hypnotic analgesia with "hidden reports" through auto-
matic key-pressing and automatic talking. Journal of Abnomzal Psychology, 1975,84,
HULL, C. 1. The concept of the habit-family hierarchy and maze learning. Psychological
Review, 1934,41, 33-52, 134-152.
JAMES, W. Principles of psychology 2 vols. New York: Holt, 1890.
JANET, P. L'Automatisme psychologique. Paris: A1can, 1889.
JANET, P. The major symptoms of hysteria. New York: Macmillan, 1907.
KAHNEMAN, D. Attention and effort. Englewood Cliffs, N. J.: Prentice-Hall, 1973.
KNOX, V. J., MORGAN, A. H., AND HILGARD, E. R. Pain and suffering in ischemia: The
paradox of hypnotically suggested anesthesia as contradicted by reports from "the
hidden observer." Archives of General Psychiatry, 1974,30, 840-847.
LEVINSON, B. W. States of awareness during general anesthesia-preliminary communi-
cation. British Journal of Anesthesiology, 1965,37, 544-546.
LEWIN, K. A dynamic theory of personality. New York: McGraw-Hill, 1935.
electrical stimulation of the periaqueductal gray matter in the cat: Behavioral obser-
vations and inhibitory effects on spinal cord interneurons. Brain Research, 1973,50,
objective study of a multiple personality. Archives of General Psychiatry, 1972, 26,
MARK, V. H., ERVIN, F. R., AND YAKOVLEV, P. I. Stereotactic thalamotomy. Archives of
Neurology, 1963,8, 52&-538.
MAYER, D. J. AND HAYES, R. L. Stimulation-produced analgesia: Development of toleran""
and cross-tolerance to morphine. Science, 1973, 188, 941-942.
MEIZACK, R. The puzzle of pain. New York: Basic Books, 1973.
MEIZACK, R., AND CASEY, K. 1. Sensory, motivational, and central control determinants of
pain: A new conceptual model. In D. Kenshalo (Ed.), The skin senses. Springfield, Ill.:
Thomas, 1968.
MEIZACK, R., AND WALL, P. D. Pain mechanisms: A new theory. Science, 1965,150,971-
MILLER, G. A., GALANTER, E., AND PRIBRAM, K. H. Plans and the structure of behavior. New
York: Holt, 1960.
MORAY, N. Attention: Selective processes in vision and hearing. New York: Academic Press,
NATHAN, P. W. Reference of sensation at the spinal level. Journal of Neurology, Neurosur-
gery, and Psychiatry, 1956,19,8&-100.
NEWELL, A., AND SIMON, H. A. Human problem solving. Englewood Cliffs, N.J.: Prentice-
Hall, 1972.
NORMAN, D. A. Memory and attention: An introduction to human information processing.
New York: Wiley, 1969.

OSBORN, A. G., BUNKER, J. P., COOPER, L. M., FRANK, G. S., AND HILGARD, E. R. Effects of
thiopental sedation on learning and memory. Science, 1967,157, 574--576.
OVERTON, D. A. State-dependent learning produced by alcohol and its relevance to
alcoholism. In B. Kissin and H. Begleiter (Eds.), The biology of alcoholism Vol. 2. New
York: Plenum Press, 1972.
OVERTON, D: A. State-dependent learning produced by addictive drugs. In S. Fisher and
A. M. Freedman (Eds.), Opiate addiction: Origins and treatment. Washington, D. c.:
Winston, 1973.
REYNOLDS, D. V. Surgery in the rat during electrical analgesia induced by focal brain
stimulation. Science, 1969,164, 444-445.
SARBIN, T. R., AND COE, W. C. Hypnosis: A social psychological analysis of influence
communication. New York: Holt, Rinehart and Winston, 1972.
SCHREIBER, F. R. Sybil. Chicago: Regnery, 1973.
SPERRY, R. W. Hemisphere disconnection and unity in conscious awareness. American
Psychologist, 1968,23, 723-733.
STERNBACH, R. A. Strategies and tactics in the treatment of patients with pain. In B. L.
Crue (Ed.), Pain and suffering: Selected aspects. Springfield, III.: Thomas, 1970, pp.
STERNBACH, R. A. Pain patients: Traits and treatment. New York: Academic Press, 1974.
STOLLER, R. J. Splitting: A case of female masculinity. New York: Quadrangle Books, 1973.
THIGPEN, C. H., AND CLECKLEY, H. M. The three faces of Eve. New York: McGraw-HilI,
TOLMAN, E. C. Purposive behavior in animals and men. New York: Century, 1932.
TRAVELL, J., AND RINZLER, S. H. The myofascial genesis of pain. Postgraduate Medicine,
1952, 11, 425-434.
TREISMAN, A. M. Strategies and models of selective attention. Psychological Review, 1969,
76, 282-299.
WALL, P. D., AND SWEET, W. H. Temporary abolition of pain in man. Science, 1967,155,
WHITE, R. W., AND SHEVACH, B. J. Hypnosis and the concept of dissociation. Journal of
Abnormal and Social Psychology, 1942,37, 309-328.
5 Hypnotic Susceptibility,
EEG-Alpha, and Self-



Hypnosis is probably more popularly known, revered, feared, and

mystified than any other informational control known to man. Its
methods, motives, and outcomes have been bent out of shape by
hundreds of old movies and old wives' tales. The mass media have
formed popular knowledge of hypnosis largely from improbable
clinical applications and naively effective coercive control by bad guys.
Because of this kind of popularity, and despite its responsible
professional use in the treatment of certain behavior disorders, hypno-
sis has never gained much respectability outside of its own praction-
ers and fans. This is almost surely because until the last several
decades its application has not been systematically investigated. And
it has not been widely used because it has not been systematically
understood, especially in terms of its physiological and psychological
Despite its long history, it is difficult to be very specific about
hypnosis. There are several good reasons for this.
First, the hypnotic state has been hard to define by any rigorous
behavioral, subjective, or physiological boundaries. The same is true
for most other psychological states, which are similarly influenced by
the subjectivity of observer and participant. Most internal states are
too unstable within the same person and inconsistent between people
for much generality at all.
DAVID R. ENGSTROM . Department of Psychiatry & Human Behavior and Student
Health Service, University of California, Irvine, California.


Secondly, hypnosis can mean two very different things. It is

popularly known as a distinct psychological state, with unique attri-
butes that separate it from other states. But it is also an operation, or a
label for a set of operations that constitute the hypnotic situation, or
Uwhat hypnotists do." So the word can mean either a state or an
instructional set. And, in fact, the most basic theoretical difference
among scientific investigators of hypnosis centers around whether
hypnosis is best defined as the private, internal responses of the
hypnotized subject or by external characteristics implicit in the hyp-
notic situation.
The definition of hypnosis has always been impure, and it is by
virtue of this impurity that hypnosis is an appealing model for
eliciting the effects of words on behavior, of instructions on action,
and of cognitive control on performance. As an operation, it serves to
magnify the subtleties of human informational control, thereby expos-
ing them for systematic manipulation as variables. In the process it
becomes a caricature of all communication, which makes it seem more
funny than scientific. At times, it is certainly both.
Hypnosis is most often characterized as a unique state of con-
sciousness, either concurrent with or antecedent to specific behaviors.
It can be arrived at either arbitrarily in life situations of through
formal induction techniques. Certainly the "trance state" has been the
central explanatory concept in most analyses of hypnotic phenomena
that cannot be attributed to observable input, or to more objective
role-theory or motivational accounts. But this position has often been
challenged, both in theoretical argument and in objective fact. The
number of behaviors that can be called unique to the hypnotized
subject have been diminished by these challenges. Many behaviors
considered "hypnotic" have been exactly replicated by unhypnotized
subjects who are either encouraged to pretend to be hypnotized or are
simply given a pep talk about what is expected of them in careful
laboratory studies. Experienced hypnotists cannot always tell the
difference. And the phenomenal reports of subjects who have been
hypnotized are often misleading because of their demonstrated close
relationship to the demands upon and expectations of the subject in
the hypnotic situation.
To date, no one has adequately defined the conditions that are
both necessary and sufficient for hypnosis to occur. Under some
circumstances, some subjects will enter hypnosis spontaneously, with-

out formal induction procedures, while others never become hypno-

tized despite exhaustive formal induction techniques.
The most important advance in the scientific study of hypnosis
has been the development of valid, reliable scales to measure individ-
ual differences in response to standardized hypnotic procedures. The
question "What is hypnosis?" is replaced with the more answerable
"What responses are brought about among subjects exposed to hyp-
notic operations?"
Defined as the degree to which an individual is able to enter into
hypnosis and become involved in its characteristic behaviors and
experiences under standardized conditions, hypnotic susceptibility is
an ability, in which people vary. Most measures to susceptibility are
direct and standardized. They are designed to assess the behavioral
and subjective limits of hypnotic involvement of different individuals
during and following the same hypnotic induction and instruction


A. Early Objectification
The oldest problem for systematic investigators of hypnosis is
response specification, that is, the identification of a set of criterion
behaviors elicited following hypnotic instructions. Many clinical prac-
ticioners of hypnosis have developed observational systems for the
assessment of responsiveness to or of depth of hypnosis. Braid (1843)
used the criterion of spontaneous amnesia for events occurring during
trance to differentia~e highly susceptible subjects. From this early
specification of a single criterion as a measure of response or state,
there developed a large number of more complex systems, specifying
discrete kinds of hypnotic states. Charcot (1882) and his co-workers
described three discrete types of hypnotic state: catalepsy, lethargy,
and somnambulism. Each was characterized by unique behavioral
attributes including waxy flexibility of the limbs, passive unrespon-
siveness, and automatic responsiveness to verbal suggestion, respec-
tively. From the time of Charcot through the end of the 19th century,
methods of scaling hypnotic susceptibility were qualitative and
evolved in two ways. First, there was the addition of many more

discrete states to earlier criteria, still with the assumption that each
state was unique and separate from the others. And second, there is
evidence of beginnings of the idea that hypnosis can be measured
unidimensionally and continually, that subjects who respond to a
given degree of depth would show all of the symptoms of lesser
degrees on the same scale (Bernheim, 1888; Liebeault, 1889).
Hilgard (1965) has suggested that, in a sense, these continuous
categories have the properties of a Guttman-type scale, in that re-
sponse at one level indicates that a subject will probably be able to
perform all of the behaviors at lower levels. Another important
characteristic of this scale is the emergence of a continuous ability
factor that varies from person to person.
Fortunately records of subjects were frequently kept by investiga-
tors in the late 19th century. Hilgard, Weitzenhoffer, Landes, and
Moore (1961) have combined these records and standardized subject
classification on a four-point scale of responsiveness. The distribution
they obtained for 14 early investigators, including 19,534 subjects in
these categories, was (1) refractory or nonsusceptible, 9%; (2) drowsy-
light, 29%; (3) hypotaxy-moderate, 36%; and (4) somnambulistic-
deep, 26%. Hilgard (1965) cautioned that there are several uncontrolla-
ble variables in such a study that would be crucial for definitive
results, including individual variations in criteria for specifying differ-
ent states, differences in induction procedures, and semantic-opera-
tional differences among hypnotists.

B. Modern Hypnotic Susceptibility Scales

Since the beginning of the 20th century, two important predeces-

sors of modem assessment instruments have been developed. Davis
and Husband (1931) constructed a scale in which hypnotic suggestions
were scored in clustered groups, rather than on an individual basis.
Subjects were classified according to a five-point scale, with each
point associated with a group of items. The scale did not have a
standardized induction technique, however, and norms and standards
for scoring have never been made adequate. In 1938, Friedlander and
Sarbin introduced a four-part scale with a standardized verbatim

induction procedure, objective criteria for scoring each item, and self-
scoring by the experimenter. Because of these features, the scale has
higher interrater reliability than its predecessors.
The development and refinement of the modem hypnotic suscep-
tibility scales used in current research has built upon the progressively
more systematic methods of earlier scales. This effort has resulted in a
number of scales for different purposes which are thoroughly stand-
ardized and easily administered and have good psychometric proper-
ties. The bulk of this work has been done at the Laboratory of
Hypnosis Research at Stanford University under the direction of
Ernest R. Hilgard.
The Stanford Hypnotic Susceptibility Scales, Forms A and B
(SHSS:A and B) were the original scales, developed by Weitzenhoffer
and Hilgard (1959) and are parallel alternate forms for repeated
measurements, when necessary. They are partly derived from earlier
scales, but with the addition of easier items to normalize the distribu-
tion of obtained scores and to avoid the piling up of scores at the
lower end of the range. The scales have 12 items, all scored on a pass-
fail basis and all given equal weights. The items included in Forms A
and B are described in Table 1. They include a "waking suggestion"
for postural sway, before eye closure, in which the hypnotist stands
behind the subject and repeatedly suggests that he will sway back-
ward into the hypnotist's arms. This item is described by the experi-
menter as a nonhypnotic suggestion, to prepare the subject for future
suggestions. The hypnotist then gives eye-closure suggestions, fol-
lowed by "deepening suggestions," intended to increase hypnotic
responsiveness once the eyes have closed. The items are presented in
a mixed sequence of difficulty, so that the subject does not reach a
plateau and meet with repeated failures beyond that point. The scale,
which has served as a prototype for others, is constructed like a good
aptitude test.
In describing the factorial composition of the SHSS:A items,
Hilgard (1965) reported that three unrotated factors account for 69% of
its total variance. These are (1) the "challenge" or "negative sugges-
tion" items, involving loss of voluntary control over bodily muscula-
ture, which accounts for 53% of the variance; (2) the direct suggestion
motor items, in which there is a muscular response to direct sugges-
tion (10% of the variance); and (3) the cognitive items, represented

Items on the Stanford Hypnotic Susceptibility Scales, Forms A and B"

Item FormA Form B Criterion for passing

1. Postural sway Backward Backward Falls without forcing

2. Eye closure Form A induction Form B induction Eyes close without forcing
3. Hand lowering Left Right Lowers at least 6 inches
4. Arm immobilization Right arm Left arm Arm rises < 1 inch in 10 seconds
5. Finger lock Before chest Over head Incomplete separation of fingers after 10
6. Arm rigidity Left arm Right arm < 2 inches of arm bending in 10 seconds
7. Moving hands Together Apart Hands move 6 inches
8. Verbal inhibition Name Home town Name unspoken in 10 seconds
9. Hallucination Fly Mosquito Behavioral acknowledgement
10. Eye catalepsy Both eyes closed Both eyes closed Eyes remain closed 10 seconds
11. Posthypnotic suggestion Change chairs Rise, stretch Any movement response
12. Amnesia Recall items 3-11 Recall items 3-11 Recall of 3 or fewer items t:l
a From Weitzenhoffer and Hilgard, 1959. ~


by hallucination, amnesia, and posthypnotic suggestion (6% of the

The SHSS:A and B have test-retest reliabilities of r = .83 for
the original standardization sample (N = 124), and a later sample
(N = 96) yielded a retest reliability of r = .90. Reliabilities of the
individual items on retest are high, and the scales show a high in-
ternal consistency.
A third form of the same scales, the Stanford Hypnotic Suscepti-
bility Scale, Form C (Weitzenhoffer and Hilgard, 1962) was developed
within the same format to sample a wider variety of hypnotic
experiences that SHSS:A and B, particularly of the cognitive type
(hallucination, dream, age-regression). The scale is administered in
increasing item difficulty and has score distributions, reliabilities, and
internal consistency similar to SHSS:A and B. The correlation between
SHSS:C and SHSS:A and B is r = .72.
Because of their excellent psychometric properties, the format and
content of the Stanford scales have been a basis for the development of
other useful susceptibility scales. This parallel development has kept
data closely comparable.
The Harvard Group Scale of Hypnotic Susceptibility (Shor and
Orne, 1962) was derived from SHSS:A to measure hypnotic respon-
siveness in groups. Items are somewhat modified to permit self-
scoring in booklets given to subjects so that they can rate their
behavior and experiences as an observer would. Group self-scoring
norms are very similar to those of individual-observer scoring, al-
though average scores tend to be slightly higher when the subject
rates himself (Shor and Orne, 1963).
The Childrens Hypnotic Susceptibility Scale (London, 1962) is
another scale which closely parallels the content and style of the
Stanford scales but was developed to test susceptibility in children.
Both qualitative and quantitative observer scoring-systems are used.
Two scales that use different formats are the "diagnostic scale"
developed by Orne and O'Connell (1967) and the Barber Suggestibility
Scale (Barber and Glass, 1962). The "diagnostic scale" utilizes both the
subject's observed responses and the experienced hypnotist's qualita-
tive impressions, and instead of a standardized-item format, tech-
niques and items are freely selected which are best tailored to the
individual subject. This scale gives the experienced hypnotist freedom
to choose the items that might best assess maximum susceptibility in

subjects. The Barber Suggestibility Scale is a scale which presents

items without the usual hypnotic induction. The items are presented
as tests of imagination, and "task-motivating instructions" are used
instead of hypnotic instructions. Norms indicate that average scores
for subjects who are administered the Barber scale without prior
induction are slightly lower than those for subjects who received prior
hypnotic induction (Barber and Calverley, 1963).
The development of standardized, face-valid, and reliable hyp-
notic susceptibility scales has led to the accumulation of distributions
of susceptibility, as it exists as an ability in all people. From these
basic data, considerable research has been done on the various
parameters of hypnotic susceptibility, including situational influences,
personality variables, and physiological responses.


An individual's ability to enter into hypnosis and become in-

volved in its characteristic objective behaviors and subjective experi-
ences under standardized conditions has long been considered a very
stable trait or subject variable, incapable of significant modification.
An enormous amount of data reveals that the susceptibility scores of
individual subjects obtained at two different times under standard
conditions are usually highly correlated. Hilgard (1965) has reported
correlations, based on retest reliability, in the .80-.90 range, when
subjects are retested after several days or weeks. More recently,
Morgan, Johnson, and Hilgard (1974) retested 85 subjects on the
SHSS:A, following a retest interval of between 8 and 12 years (mean
interval = 10.5 years). There were no overall changes in susceptibility
scores, and the correlation between total scores of the two testings was
r = .60. Interestingly, intratest item changes over time were much
more apparent, with motor "challenge" items, such as arm immobili-
zation, arm ridigity, and verbal inhibition, all improving even in the
adult years and cognitive-type items, such as amnesia, posthypnotic
suggestion, and hallucinations, all decreasing with age. These changes
in opposite directions tend to balance out and mask themselves in the
stability of the whole test score. Normally, without special interven-
tion, susceptibility scores reflect little change over time.


Although it is widely believed that hypnotizability is easily

modified with practice, this idea probably came about because of the
increased speed in which the hypnotic state can be brought on with
practice or rehearsal. Especially in the clinical setting, as the subjeCt
undergoes repeated induction with a hypnotist, the cuing process for
bringing about hypnosis can be reduced to a preselected signal,
greatly increasing induction speed, while having no effect on ultimate
depth of hypnosis or ability to respond to hypnotic suggestion.
Some investigators have attempted to modify it within subjects,
typically through simple practice or repeated hypnotic induction. The
majority of these attempts have met with little or no success (As,
Hilgard, and Weitzenhoffer, 1963; Barber and Calverly, 1966; Cooper,
Banford, Schubot, and Tart, 1967; Gill and Brenman, 1959; Sachs and
Anderson, 1967; Shor, Orne, and O'Connell, 1962).
In one such experiment (As, Hilgard, and Weitzenhoffer, 1963), a
wide variety of manipulations were used flexibly, including practice
induction techniques and psychotherapeutic attitude discussions, re-
sulting in "only very slight gains" on before-and-after susceptibility
measures. Repeated similar results led London (1967) to conclude that
"scale results are so consistent from session to session as to suggest
that susceptibility is a very stable personality trait, extremely resistant
to change" (p. 72).
Although there were early reports of success in increasing suscep-
tibility by intensive practice and deepening procedures (Blum, 1963;
Wiseman and Reyher, 1962), these results are both sparse and restric-
tive in the number of subjects used, making them hard to generalize.
From this evidence, a conservative conclusion would be that without
special intervention susceptibility is fairly stable, and even with
intervention it is not very easy to change.
Despite these early findings and conclusions, later investigations
have demonstrated significant increases in susceptibility-scale scores
following diverse training procedures within the experimental situa-
tion. Both operant-conditioning techniques (Sachs and Anderson,
1967) and observational-learning procedures (Diamond, 1972) have
been used successfully to raise susceptibility scores. In addition to
traditional learning procedures, a number of other techniques have

recently been used to increase susceptibility, including sensory depri-

vation (Sanders and Reyher, 1969; Wickramasekera, 1970), EEG alpha
biofeedback training (Engstrom, London, and Hart, 1970), EMG bio-
feedback training (Wickram, 1972; Engstrom, 1972), and LSD-25 and
dextroamphetamine (Ulett, Akpinar, and Hil, 1972). Tart (1970) and
Shapiro and Diamond (1972) found that nonlaboratory related "per-
sonal growth experience" such as residential programs and encounter
groups tend to increase the hypnotizability of participants. Of course,
there are large differences among these studies in the selection of
subjects, the criterion instruments used to measure hypnotic suscepti-
bility, and the initial level of susceptibility of the subjects. And, most
importantly, many of these results have not yet been replicated.
When combined, however, these studies do indicate that suscep-
tibility can be modified for at least some more susceptible subjects by
assorted experimental techniques, including behavioral shaping, bio-
feedback, sensory restriction, and pharmacological and other nonspe-
cific methods. The very fact that this ability is not completely fluid and
changeable on the one hand and not totally stable on the other
supports the assumption that there may be two interacting parts of
susceptibility: a state-specific skill component, reflected in the sub-
ject's ability to alter his internal state, and a less stable situational-
attitudinal component, dictated by the subject's assessment of and
response to the hypnotic situation. Trying to tease out responses that
are attributable to internal changes is very difficult. The hypnotizable
subject's capacity to resist hypnosis is always present, as is the
unhypnotizable subject's ability to simulate hypnotic performance.
Both conditions can alter obtained scores on susceptibility scales,
since all scales rely on objective behavior and subjective report for


Currently there is little understanding of individual differences in

responsiveness to hypnosis as they relate to measured personality
traits. If susceptibility to hypnosis is more than a situational response
and is, as many experimental findings imply, a structural trait, then it

should have ongoing correlates in physiological, cognitive, and social

response patterns. But if they are there, they are hard to find, and
efforts to relate susceptibility to other stable aspects of personality
have been extremely unsuccessful. According to Hilgard's review
(1967), the search for personality correlates of hypnotizability has
produced rather sparse results in light of the considerable amount of
effort expanded. Studies relating personality variables to hypnotic
susceptibility have generally yielded few and often contradictory
results. The variables studied have included sex (Cooper and London,
1966; Hilgard, Weitzenhoffer, and Gough, 1958; Weitzenhoffer and
Weitzenhoffer, 1958); intelligence (Hilgard, 1967; Weitzenhoffer,
1953); projective measures of personality (Sarbin and Madow, 1942;
Schafer, 1947); objective personality inventories (Faw and Wilcox,
1958; Schulman and London, 1963); attitudes toward hypnosis (Melei
and Hilgard, 1964; Rosenhan and Tomkins, 1964); and psychiatric
diagnosis (Gill and Brenman, 1959; Kramer and Brennan, 1964; Lon-
don, Cooper, and Johnson, 1962).
These studies have all assessed hypnotizability in a standardized,
reliable manner and have all yielded results that are highly equivocal.
Either the variable measured was uncorrelated with hypnotic suscepti-
bility at all, or it yielded a significant correlation of low predictive
power and with poor results on replication. The relationship between
most measured personality variables and susceptibility is, at best, a
very subtle one. Still, more is known about the trait dimensions of
susceptibility than its state properties.

A. Age and Development

On the positive side, a number of studies (Stukat, 1958; Barber

and Calverley, 1963; Cooper and London, 1966; London and Cooper,
1969; Morgan and Hilgard, 1973) have found a generally higher level of
susceptibility in children than in adults. The average maximum level
of susceptibility is reached between ages 9 and 14, with a slow decline
to stable adult levels between ages 14 and 16. In a detailed analysis of
age differences in susceptibility, Moore and Lauer (1963) found that
children tend to pass hallucination and amnesia items much more
frequently than motoric items.

Recently Josephine Hilgard (1970) has shown, through very care-

ful and extensive interview methods, that high susceptibility is
correlated with certain aspects of childhood development and present
personality. The correlation was especially high (in decreasing order of
magnitude) for imaginative involvements, severity of punishment in
childhood, ease of communication, motivation for hypnosis, and
temperament similarity to the opposite-sex parent, as well as normal,
outgoing personality attributes. From these results it is clear that
hypnotic susceptibility does have longitudinal, ongoing correlates not
easily measured by personality tests. Some people, through certain
developmental and life experiences, are predisposed to be more
hypnotizable than others. Apparently part of the ability to respond
to hypnotic operations is age and experience-related and changes

B. Motivation

Motivation seems to be another potentially important determinant

of hypnotic susceptibility. Shor (1959) observed that low-susceptible
subjects volunteered to tolerate higher-intensity levels of shock than
did high-susceptibles in the same experiment, and Orne (1963) found
that highly susceptible subjects tended not to return to the laboratory
for subsequent appointments, while un susceptible subjects were more
reliable and punctual.
London and his co-workers, using hypnotic susceptibility as the
primary criterion of subject selection, unexpectedly found that low-
susceptible subjects tried harder and performed better in a laboratory
situation on tasks involving strength, endurance, and motoric stabil-
ity. Since these were base-rate tasks, administered before the subjects
were hypnotized, the results seem to have more to do with differences
in motivational set that high- and low-susceptible subjects brought
into the experimental situation than with any set elicited by experi-
mental manipulation. Further, a number of different experiments have
attempted to examine these differences between high- and low-
susceptible subjects across different parameters of instruction, learn-
ing and performance tasks, and experimental conditions (hypnosis,

base rate, exhortative, emotionally involving, and relaxing} on physi-

cal performance measures of maximum exertion and motor coordina-
tion (London and Fuhrer, 1961; Rosenhan and London, 1963). Low-
susceptible subjects also perform better on rote verbal-learning tasks
under all instruction conditions, including relaxation, muscle tension,
exhortation, and hypnosis, and in two of four experiments did better
under base-rate instructions as well (London and Rochman, 1967;
5chaefler and London, 1968). These same differences between groups
hold up under experimental conditions of severe environmental stress
(London and McDevitt, 1967) and even when the subjects have no idea
of the nature of the experiment or that it is to involve hypnosis
(Rosenhan and London, 1963; London and Rochman, 1967).
From this series of experiments, it is apparent that low-suscepti-
ble subjects "try harder" than high-susceptibles under almost all
experimental conditions and that this difference in motivational set is,
to some extent, brought into the experimental situation and not just
stimulated by it. In fact, Rosenhan and London (1963) have concluded
that low susceptibility is a positively motivating trait.


Another set of variables whose relationship to hypnosis has been

superficially investigated are the physiological ones, with particular
emphasis on events in the central nervous system and specifically the
EEG. For many decades behavioral scientists have attempted to find a
criterion for hypnosis that is more convincing than the subject's self-
report. In hypnosis the search for a physiological indicator of trance
has continued unfailingly.
There is inherent complexity in attempts to define any state using
the EEG as a criterion measure. For one thing, the brain-wave
recordings of subjects who are in light sleep, deep hypnosis, relaxed,
or fully aroused look more like than different from each other on
examination. These records all typically consist of mixtures of slower,
high-amplitude alpha waves (8-13 Hz, 30-90 fLV) and low-voltage fast
activity, in apparently random patterns. So there is a problem in the

clear discrimination of state variables based solely upon EEG patterns.

Also, as Evans (1972) has pointed out, the alpha rhythm is paradoxical
in its appearance. It desynchronizes and disappears in the relaxed,
waking subject as he becomes either drowsy and increasingly sleepy
on the one hand, or increasingly aroused and preparing for more
complex cognition on the other. Evans suggests that alpha is related to
attention in a V-shaped manner, in the sense that it disappears at
either extreme of arousal and attention.
Most of the early investigations relating EEG and hypnosis have
not used the sophisticated computerized techniques now available but
have relied on visual scanning of analogue records, as well as hand
counting and scoring. They have been of a purely correlational nature
and have failed to demonstrate EEG differences between waking and
hypnotic states (Barber, 1970; True and Stephenson, 1963; Weitzenhof-
fer, 1953). Early studies in this area strongly suggested that waking
EEG patterns persist during passage into and throughout the hypnotic
state (Chertok and Kramarz, 1959; Dynes, 1947; Loomis, Harvey, and
Hobart, 1936). Similarly Edmonston (1967) reported failure to reliably
demonstrate changes in autonomic functioning as a result of hypnosis
alone. These results, and others like them, have led Barber (1970) to
conclude that "physiological variables vary in hypnotic subjects in the
same way as in normal individuals, that is, in accordance with
whatever activity they are engaged in" (p. 159).
Several studies have, however, shown evidence of a relationship
between hypnosis and specific EEG output. Barker and Burgwin (1949)
reported the induction of EEG patterns resembling those of sleep by
utilization of hypnotic suggestion. Through hypnotic suggestion to
relax, Ford and Yeager (1948) found that persistent EEG alpha rhythms
could be established in subjects who had previously exhibited little or
no alpha.
The vast majority of these early studies have tried to identify
hypnotic state differences, but little or no attention was paid to more
basic individual differences in susceptibility between subjects. Either
the investigators were random and unselective in their choice of
subjects by susceptibility, or they chose only extremely hypnotizable
subjects by nonstandardized measures. Although reliable instruments
for the assessment of this trait have existed for many years, most early
investigators have failed to score subjects on susceptibility, allowing

no adequate means for evaluating results in terms of individual

differences in responsiveness to hypnotic procedures.



Reliable changes in both EEG activity and hypnotic susceptibility

have been noted by different investigators, concurrent with both
developmental patterns and environmental manipulation. One area
includes changes in EEG and susceptibility as a function of age, the
other as a result of perceptual or sensory deprivation.

A. Age
The relationship between age and hypnotic susceptibility, de-
scribed earlier in this chapter, has been explored by a number of
different investigators (Barber and Calverley, 1963; London, 1965;
Moore and Lauer, 1963; Stukat, 1958). Among 240 children, London
found marked changes in responsiveness to hypnosis which were age-
related, with the maximum susceptibility scores attained between ages
9 and 14, falling slightly to stable adult levels between 14 and 16 years
of age.
A similar age-related pattern has been reported for changes in
slow-wave EEG activity. Stevens, Sachdev, and Milstein (1968) found
that the EEG in infancy and early childhood is slower and higher in
amplitude and shows less regional differentiation of wave forms than
the EEG of older children and adults. Maturation of the EEG is
gradual, and 4-7 Hz theta predominates early from all regions. By
mid-childhood (6-8 years old) a strong 8--12 Hz alpha component
appears posteriorly, and average frequency increases through the
alpha range throughout early adolescence.
Figure 1 shows fitted curves representing the developmental
patterns of both susceptibility and brain-wave changes for ages 4
through 16. Hypnotic susceptibility scores are combined means re-
ported in two studies using a standardized scale with hypnotic
induction procedures (London, 1965; Stukat, 1958). The scores from

12 12

11 11

til ~

~ 10 10

9 9
8 8
~ 7 7 '"~
e 6 6
5 Susceptibility - - - - 5 '"
4 Alpha Rate 4
3 3
til 2 2

1 1
0 0
4 5 6 8 9 10 11 12 13 14 15 16

FIGURE 1. Changes in hypnotic susceptibility and average alpha rate with

age. (Susceptibility data: London, 1962; Stukat, 1958. Age data: Lindsley and
Wicke, 1974)

the different scales used are made comparable by the fitting of a

standard 1-12 scale to both. The closeness of fit of these two variables
suggests that they may undergo parallel developmental influences
which change them in a similar way.

B. Perceptual or Sensory Deprivation

The experimental restriction of sensory or perceptual experience

has also produced similar effects on brain waves and susceptibility,
measured independently, in different situations. The slowing of EEG
activity during exposure to deprivation has been reported in numer-
ous experiments (see Zubek, 1973). A decrease in the mean frequency
of the occipital alpha rhythm is usually greater with perceptual
deprivation than with sensory deprivation (Zubek and Welch, 1963).
Perceptual deprivation is a condition characterized by greater percep-

tual-motor and cognitive impairment. Also, the recovery period after

deprivation is slow and gradual, with the mean EEG frequency slowly
returning to preexperimental levels. The magnitude of this alpha-
frequency slowing is dependent upon deprivation conditions, but the
result is very reliable (Zubek, 1973).
Taken together, these findings are not definitive but barely
suggestive of a relationship between hypnotic susceptibility and EEG
changes. They do suggest a variable that is basic to hypnosis: the
restriction of sensory experience. If lower levels of cortical arousal
represent a greater predisposition in certain subjects to restrict sensory
experience, this should relate to the skills involved in becoming
hypnotized. These interrelationships and the speculation generated by
them converge on the direct evidence of a relationship between
hypnotic susceptibility and the EEG.


A. Base-Rate Alpha Density

The EEG alpha rhythm is a recurring 8-13 Hz brain-wave pattern

associated with a relaxed, waking condition with the eyes closed. The
alpha pattern disappears in response to most visual, auditory, and
tactile stimuli and is replaced with the higher-frequency fast EEG
activity associated with alertness and vigilance. Most of the existing
correlational studies of EEG and hypnotic susceptibility have meas-
ured the alpha rhythm recorded from a single occipital site. Since
Kamiya's (1969) claims of operant control of these brain rhythms, there
has been a renewed interest in alpha brain waves and their meaning,
and the alpha rhythm has been studied more extensively as a well-
defined part of consciousness.
London, Hart, and Leibovitz (1968) originally found evidence that
highly susceptible subjects produced more waking alpha than nonsus-
ceptibles, in a sample of 125 subjects. Subjects were grouped by score
on the HGSHS:A, and subsequently, the number of seconds per
minute of alpha were recorded during waking and eyes-closed rest
periods. Those subjects scoring between 0 and 4 on susceptibility had
a mean alpha density of 37%; those from 4 to 7 in susceptibility, 56%;
from 7 to 11, 42%; and 12, 70%. At the extremes, all subjects scoring

12 on the HGSHS:A produced base-rate alpha for an average of 42.3

seconds per minute, while 25 subjects scoring 4 or less produced alpha
for 24.0 seconds per minute (p < .005).
From an evaluation of his own unpublished data, Evans (1972)
compared the resting 2-minute alpha base-rate measures of 139 volun-
teer subjects with their scores on the HGSHS:A, the SHSS:C, and
clinical diagnostic ratings of hypnotizability. Although the three
subjects who initially scored 12 on the HGSHS:A generated an average
of 90% alpha, 40 low-susceptible subjects (scores 0--4) produced 65%,
considerably more than the 37% obtained by London, Hart, and
Leibovitz (1968). The addition of the individual Stanford Scale and
diagnostic rating also reduced the correlation. Using the same suscep-
tibility categories as London, Hart, and Leibovitz, Evans obtained
insignificant correlations (p > 1.0) between all susceptibility measures
and frequency, amplitude and density of alpha output.
A study by Engstrom, London, and Hart (1970) partially sup-
ported the correlation, using a sample of 30 subjects preselected for
medium and low hypnotic susceptibility (average score of 7 or less on
both HGSHS:A and SHSS:A) and less than 50% alpha recorded during
a subsequent 4-minute base-rate period (2 minutes eyes open, 2
minutes eyes closed). For these 30 subjects with low to medium
susceptibility and low alpha output, a correlation of .56 (p < .01)
between alpha density and susceptibility was obtained.
Using similar selection criteria, Evans (1972) obtained a correla-
tion between HGSHS:A and alpha of .26 (p < .05) for a group of 48
subjects low in susceptibility and alpha output, drawn from data of
prior experiments, although the correlation between HGSHS:A and
alpha for unselected subjects was still nonsignificant (- .02).
Nowlis and Rhead (1968) found a correlation of .70 (p < .001)
between the sum of HGSHS:A and SHSS:C and resting, eyes-closed
alpha for a group of 21 unselected subjects. Evans (1972) has calculated
separate correlations according to susceptibility and obtained a corre-
lation of .16 (nonsignificant) for the 12 high-susceptible subjects and
.79 (p < .05) for the 9 low-susceptibles.
Hartnett, Nowlis, and Svorad (1969) obtained a rank-order corre-
lation of .69 between alpha and SHSS:C for a subgroup of 14 subjects
selected from 28 subjects by unreported criteria. For all 28, the
correlation was nonsignificant (- .27).
Two groups of investigators (Galbraith, London, Leibovitz,

Cooper, and Hart, 1970; Ulett, Akpinar, and Hil, 1972) have related
hypnotic susceptibility to quantitative digital computer methods of
cross-spectral EEG-frequency-analysis data collection. Galbraith et al.
gave the HGSHS:A to 80 subjects and 2 weeks later selected 59 of these
to participate further in an ostensibly independent "study of brain-
waves." Quantitative EEG data were analyzed through stepwise
multiple linear regression to isolate the EEG frequencies most closely
related to susceptibility, but alpha variables did not significantly
predict susceptibility scores. Conversely, Ulett et al. found decreased
slow-wave activity (4-7 Hz) and increased alpha and very-high-
frequency superimposed beta activity (40-50 Hz) correlated with
hypnotizability, defined by the Barber Suggestibility Scale.
These results are confusing. They include significant positive
correlations and nonsignificant correlations, as well as no correlation
at all between alpha and susceptibility. Additionally there are consid-
erable data suggesting a correlation limited to especially selected
subgroups of subjects (Hartnett, Nowlis, and Svorad, 1969), particu-
larly to low susceptibles (Engstrom et al., 1970; Evans, 1972; and
Nowlis and Rhead, 1968).
Evans (1972) has correctly noted that in the only two studies in
which subjects were not told of the connection between susceptibility
and EEG measures (Evans, 1972; Galbraith et al., 1970), no correlation
was obtained.

B. Base-Rate Alpha Amplitude

There are some data which relate EEG alpha-amplitude differ-

ences to susceptibility. Engstrom (1971) and Evans (1972) have both
reported nonsignificant correlations between mean alpha amplitude
and score on HGSHS:A (r = .12; r = .04). Engstrom found a similar
nonsignificant correlation between amplitude and SHSS:A (.09) ad-
ministered to the same subjects while Evans further reported no
correlation between amplitude and SHSS:C (.03) and a diagnostic
rating of hypnotizability (- .02) among the same subjects.
In contrast to these findings, Morgan and MacDonald (1973) found
that among 26 subjects, high-susceptibles produced higher amplitudes
of alpha. The relationship between alpha amplitude and susceptibility
is unclear from the present data.

C. EEG Asymmetry
The proposition that the two hemispheres of the brain operate
somewhat autonomously and perform different functions has resulted
in several relevant studies involving EEG differences between hemi-
spheres. Lateral asymmetry is the measured difference in electrical
activity between the same part of the two hemispheres of the cerebral
cortex, and hemispheric dominance has been shown to be related to
subjective reports of thought process and type of task.
Morgan, McDonald, and MacDonald (1971) recorded EEG alpha
activity bilaterally in a sample of 10 high-susceptible and 10 low-
susceptible subjects under task conditions designed to activate first the
left and then the right hemisphere. More alpha was recorded from the
right than the left in both conditions, and there was significantly less
alpha in the right hemisphere when it was engaged in a task.
Morgan and MacDonald (1973) found significant asymmetry in
lateral alpha between analytic and spatial tasks, and between eyes-
open baseline and eyes-open measurement during hypnotic amnesia,
in 26 right-handed subjects, but no differences were found between
low- and high-hypnotizables in the percentage difference (laterality)
measure. Results suggest a relationship between EEG asymmetry and
type of task, but not susceptibility.

D. Evoked Potentials
The sudden, high-amplitude spike responses in the EEG tracing
that typically follow presentation of an unexpected stimulus, called
evoked potentials, have been reported to be alterable by hypnotic
suggestion (Clynes, Kohn, and Lifshitz, 1963). These almost reflexive
patterns, which apparently reflect the attentional process at some
level, have enormous appeal to hypnosis researchers. Unfortunately an
abundance of recent evidence (Beck and Barolin, 1965; Beck, Dustman,
and Beier, 1966; Halliday and Mason, 1964) has failed to find a
significant relation between hypnosis and evoked potentials.

E. Conclusion
In reviewing the experimental findings on hypnotic susceptibility
and alpha, Evans (1972) has stated that "in spite of conflicting results

it is concluded that waking alpha frequency, density and amplitude

are probably not correlated with susceptibility to hypnosis" (p. 59).
The evidence is equivocal and it is not possible to conclude that
there is a relationship between hypnotic susceptibility and waking
EEG activity among un selected subjects. Nor is it possible to conclude
that there is not a relationship.
Before a conclusion is possible, the search for a systematic
relationship between EEG and either the ability to be hypnotized or
the result of hypnotic induction must be extended and evaluated in
more detail.
First it is imperative to determine the stability of an EEG base-rate
measure over time. Physiological base rates can change, sometimes
drastically, as a result of a subject's accommodation to the laboratory
setting. Evans (1972) found that base-rate differences in skin potential
between susceptible and non susceptible subjects which exist in initial
sessions disappear with subsequent sessions. This suggests
that differences in base rates are an artifact of relaxation, since re-
peated measures in subsequent sessions weakened the initial group


To try to relate a fairly stable variable like hypnotic susceptibility

to a potentially unstable one may be foolish to begin with. EEG
measures in humans, which change from moment to moment, might
change so much from session to session that any correlation between
them is spurious. Whether base-rate EEG changes as a result of special
interventions or life experiences is only a meaningful question after it
is established that it does not change on repeated measurement in the
same setting, without special interventions.
Since repeated measurement in the same setting requires an
ongoing baseline measure without experimental intervention, 18 sub-
jects with no prior experience in experiments were brought into the
laboratory for "a series of physiological measurements." The base-rate
EEG was recorded during eight sessions, within a period of 35 days.
All recordings were taken while the subjects were instructed to remain
in a relaxed, waking state for a total of 6 minutes, 3 minutes eyes-

.... Eyes Closed - - - -
u 60f
50 Eyes Open __ _


30 : -
--------, ...----...... "
i:;] " ............ "
1 2 3 4 5 6 7 8

FIGURE 2. Mean eyes-closed and eyes-open alpha density over
eight base-rate recordings.

closed and 3 minutes eyes-open. EEG recordings were taken from a

monopolar occipital (02 ) to right mastoid site and recorded on a Grass
Model 7 polygraph with output to filter cutoffs at 8 and 13 Hz. Output
to an EPUT meter permitted all alpha activity between 8- and 13-Hz
frequency and 30- and 70-p,v amplitude to be counted in seconds per
minute. The subjects were given instructions to relax and close their
eyes for 3 minutes and then remain relaxed with eyes open for 3
minutes. Following these periods, all subjects were instructed to
perform a variety of routine card-sorting tasks. Eight repeated base-
rate measures of alpha output were taken for all subjects and the
results are shown in Figure 2.
When both eyes-closed and eyes-open averages were examined
for all eight sessions and 18 subjects, there were increasing trends in
early sessions, which suggest relaxation and accommodation, and a
slight decrease in both conditions toward the last session, perhaps
caused by drowsiness or boredom. But the deviations from original
baseline for both eyes-closed condition (+ 1, - 3) and eyes-open (+ 3,
-6) are still small. With an analysis of variance repeated-measures
design, none of the differences are significant for eyes closed (F =
0.13, P < .20) or for eyes open (F = 0.11, P < .20). The difference in
eyes-closed alpha density from Session 1 to Session 8 was not
significant (t = 0.31, P < .10) and was moderately Significant for eyes-
open (t = 1.93, P < .05). The overall correlations for all subjects
between sessions were R = .61 (eyes-closed) and R = .37 (eyes-open).

These results indicate that base-rate alpha is at least fairly stable,

and eyes-closed base-line measures are slightly more stable over
sessions than those taken with eyes open. Relaxation, boredom, and
other situational consequences of repeated laboratory visits would
appear to have a minimal effect on alpha output. This finding is
especially important since the waking EEG is such a variable measure
when examined from moment to moment. In the same setting, over
time, average alpha density appears to be fairly stable.


In independent investigations, Hart (1967) and Kamiya (1969)

found that EEG alpha can be brought under voluntary control by the
placement of the subject in an electronic feedback loop in which EEG
alpha activity produces an audible tone. The subjects who produced
high alpha levels described a state of "passive alertness" and "selec-
tive or focused attention," very similar to reports about the hypnotic
experience by highly susceptible subjects (Engstrom, 1970).
Since high susceptibility may be reflected in long durations of
base-rate alpha output and since it is known that alpha output can be
increased by feedback methods, it follows that a manipulation which
increases alpha production might also serve to raise susceptibility.
This was demonstrated experimentally by Engstrom, London, and
Hart (1970).
In a screening of 180 volunteers subjects of low to moderate
susceptibility and low alpha base rate were selected. The HGSHS was
administered to them by tape recording, and those subjects who
scored 7 or less were asked to return for EEG recordings and individ-
ual testing on the SHSS:A. Subjects with an average score on the
HGSHS and the SHSS:S of 7 or less and an alpha density of 25% or
less in a total of 4 minutes of eyes-open and eyes-closed recordings
were retained to participate further.
Of the 30 subjects who met these requirements and were selected,
20 were assigned to an experimental condition (contingent feedback
training) and 10 to a control condition (pseudofeedback training). An
attempt was made to assign subjects to the conditions on a random
Subsequent to the publication of Engstrom et al. (1970), it was

found that 4 of the subjects in the experiment had undergone previous

EEG biofeedback training and were included in the results reported. 1
For this revision, 1 experimental subject and 3 control subjects who
had had previous training were eliminated, leaving 19 experimental
and 7 control subjects. The groups were very similar with regard to
age and sex.
The EEG recording and training apparatus consisted of a Grass
Model 7 polygraph; Krone-Hite Model 330-A bandpass filters set at 8
and 13 Hz; an amplitude-sensitive trigger set to close for unattenuated
alpha signals; a Berkeley Model 554 EPUT meter for digital timing of
alpha density; a Mighty-Light photoelectric strobe activated by unad-
justed-amplitude EEG signals; a Heath EUW27 audio oscillator acti-
vated by adjusted-amplitude EEG signals; an Ampex SP300 tape
recorder; and a Roberts 1670 internally modified tape recorder. Out-
puts from the trigger activated the EPUT meter, and a signal marker
on the polygraph permitted visual checking of the trigger accuracy.
Outputs from the trigger and three channels of the polygraph were
connected to the Ampex tape recorder. Output from the Roberts tape
recorder connected to a trigger input. This system provided tracking
feedback via the strobe light and the on-off target feedback via the
speaker when the trigger was operated in the feedback mode. The
apparatus was located in a polygraph room, which was separate from
the subject room except for the feedback light and speaker. A two-way
intercom and a two-way tone signal device connected the two rooms.
For both screening and training, Grass silver disk electrodes were
placed according to the international 10-20 system. Channell (FP2 -
Lear) was selected for on-line monitoring and training. The frontal
electrode site was chosen to set a stringent criterion for the alteration
of alpha levels.
Once having qualified for the study, each subject received six
training sessions spread over 1-2 weeks. Each training session con-
sisted of a 4-minute base-rate EEG measurement period and six blocks
of 5-minute feedback periods in which all frontal brain waves pro-
duced synchronous flashes, and alpha waves produced the tone as
well. The six blocks consisted of alternating 5-minute "free" se-
quences, in which the subjects were told to experiment with different
mental states to see what effects they had, and "test" sequences, in
which the subjects were told to try to keep the tone on as much as
1 The recalculation of data was done by Leslie M. Cooper.

possible. During each 30-minute training session, a tape recording

was made of the feedback signals. The feedback tone for the control
subjects was not their own but was activated for all of them by a
prerecorded EEG tape of a single experimental subject, who had
previously demonstrated progressive increases in alpha density dur-
ing feedback. After each control subject finished the experiment, he
was informed that he had been in a control group and was told about
the nature of the feedback he had received. He was then given the
opportunity to receive real feedback in two postexperimental sessions.
At the end of each session, the subjects were asked to describe their
experiences during alpha production or hypnosis by rating each of 48
adjectives describing different mental states. At the end of the sixth
session, a final EEG base-rate measurement was recorded, and the
SHSS: B was administered by an experimenter who had not previously
tested the subjects.
The two groups did not differ significantly on their pretraining
susceptibility scores (SHSS:A) nor on their pretraining base-rate alpha
density. The experimental group had a mean susceptibility score of
3.16 (SD = 2.27) and the control group, a mean of 4.00 (SD = 2.00).
The experimental group had a mean of 27.26 seconds of alpha per
240 seconds (SD = 16.78), and the control group, a mean of 33.00
(SD = 35.33). Neither of these differences in means was statistically
Table 2 presents the mean number of seconds of base-rate alpha
duration per 240 seconds of both groups for the base-rate measure, the
six training sessions, and the administration of the SHSS: B. For the

Mean Seconds of Alpha per 240 Seconds of EEG: Raw Scores

Training sessions
Base Posttraining:
Group rate 1 2 3 4 5 6 SHSS:B

X 27.26 46.47 62.26 74.16 71.79 87.32 104.79 109.95
SD 16.78 35.50 45.96 44.73 43.37 35.09 35.92 39.99
X 33.00 45.14 41.14 51.00 48.14 57.14 74.00 79.28
SD 35.53 26.59 20.76 22.77 23.48 30.69 46.52 54.71

six training sessions these figures represent the mean alpha output
during the initial 4-minute base-rate phase of each training session
prior to the feedback training phase. Thus they reflect the effect of the
learning from the previous session.
It had been anticipated that there would be a positive relation
between base-rate alpha output and initial hypnotic susceptibility.
The findings clearly supported this expectation. For the experimental
group, the correlation between pretraining susceptibility and pretrain-
ing base-rate alpha was .80 (p < .001); and for the control group, the
correlation was .88 (p < .05). These correlations did not differ
significantly from each other.
Both the experimental (contingent-feedback) group and the con-
trol (pseudofeedback) group increased their base-rate alpha produc-
tion over the six sessions; however, the experimental group increased
significantly more. The experimental mean was 104.79 seconds (SD =
35.92) at the beginning of the sixth training session, while the control
mean was only 74.00 seconds (SD = 46.52) at that point (Table 2). This
indicates that the training procedure was successful in increasing
alpha output.
A cursory examination of the means and variances of the two
groups over all sessions suggested that the two are related. A more
formal analysis revealed that the means and variances correlated .55 (p
< .05). Consequently, a square-root transformation (Winer, 1971) was
performed to obtain more independent sample variances. The trans-
formed means and standard deviations are presented in Table 3 (from
London, Cooper, and Engstrom, 1974).
It will be noted that the control-group means for Sessions 4 and 5
are slightly larger than the corresponding means of the experimental
group. As would be expected, however, the experimental means for
Session 6 are higher than the control means. In all training sessions,
moreover, the raw experimental means exceeded the raw control
An analysis of covariance was performed between the alpha-
duration measures of the control and experimental groups for Session
6 using the base-rate measure as the covariate. The means for the two
groups differed significantly from one another (F = 4.88, dt = 1/23, p <
These findings are further supported by the alpha production
measured during the administration of the SHSS: B, which occurred


Mean Seconds of Alpha per 240 Seconds of EEG: Transformed Scores gJ
Training sessions
Group Base rate 1 2 3 4 5 6 SHSS:B j
X 9.9976 16.0839 17.6393 19.3613 18.1485 19.1286 21.6227 20.6672 ~
SD 3.4202 3.7621 5.0845 4.3688 5.0984 4.4831 3.8810 3.9326 ~
Control ~
X 10.3268 16.7366 22.07?3 18.7929 20.5162 21.1919 19.5416 17.0718
SD 5.6471 6.0159 4.0437 4.1985 3.9075 1.7465 3.0782 5.6838



Means and Standard Deviations of Number of Seconds of Alpha Production
per 240 Seconds during Training Phase and Correlation with Base-Rate Alpha


Statistic 1 2 3 4 5 6

Experimental group
X 69.11 82.26 100.00 90.21 98.05 119.89
SD 29.96 39.91 39.94 46.59 42.35 40.63
r with base-rate alpha .66" .83" .53" .82" .72" .75"
Control group
X 79.29 128.00 98.00 11D.43 115.71 97.00
SD 53.89 45.21 47.19 40.82 18.69 32.07
r with base-rate alpha -.14 .59 .58 .52 .48 .33

.p < .05.

after the sixth session. The raw mean number of seconds of alpha per
240 seconds was 109.95 (SD = 39.99) for the experimental group, while
for the control group it was only 79.28 (SD = 54.71).
An analysis of covariance was performed between the correspond-
ing transformed means of the alpha measures (obtained during the
administration of the SHSS: B) by use of the base-rate measure as the
covariate. These differed significantly from one another (F = 4.45, df =
1/23, P < .05).
A more striking perspective of the differential effects of the
training procedures on the two groups can be noted in the alpha
production during the actual training itself, rather than during the 4-
minute base rate of each session. The latter was thought to be the
more conservative measure of the effects of feedback and, as indi-
cated, did yield different results for the experimental and control
Table 4 presents the mean alpha production per 240 seconds of
both groups during the actual feedback and the correlation for each
group between its feedback alpha production and its base-rate alpha
production in each session. There is a fairly steady monotonic
increase in the mean alpha production of the experimental group over
the six sessions, but there are large and erratic variations for the control
group. The variations of the control group are certainly not typical of

what we would expect in a learning situation. Furthermore, large

correlations between training and base-rate alpha production over the
six sessions are all significant for the experimental group, but not one
of the correlations for the control group differs significantly from zero.
A significant increase in hypnotic susceptibility as measured by
the SHSS:B was found for the experimental group only. Its mean score
increased from the pretraining score of 3.16 (SO = 2.27) to 7.42 (SO =
2.69) (t = 6.67, df = 17, P < .01). For the control group, the score
increased from 4.00 (SO = 2.00) to 5.14 (SO = 3.10), an insignificant
difference. An analysis of covariance of the posttraining susceptibility
scores between the two groups, with the pretraining score used as the
covariate, yielded a significant difference (F = 4.97, df = 1123, P < .05).
Thus, the gain for the experimental group was significantly greater
than the gain for the control group.
The most important finding of the original study by Engstrom et
al. (1970) was that the experimental group rose in hypnotic suscepti-
bility significantly more than the control group did as a result of alpha
feedback training. The present analysis reconfirms that finding, even
after the contaminating influence of four subjects who had had
previous experience with alpha training is removed. In comparison
with their base-rate performances, the subjects in the experimental
group improved significantly more, in both alpha production and
susceptibility, that did the subjects in the control group, though the
latter also improved significantly in alpha production.
The present analysis produced two other important findings.
First, the higher rate of alpha production obtained at the beginning of
the sixth training session persisted through the administration of the
SHSS: B at the end of that session for both contingent and pseudofeed-
back groups. Second, alpha production during the training phase of
each session correlated significantly with pretraining base-rate alpha
for the experimental (contingent) group but not for the control (Pseu-
dofeedback) group.
The fact that alpha output during the feedback part of each
training session correlates significantly with pretraining base-rate
alpha production for the experimental group but not for the controls is
important because it indicates that learning occurred more consist-
ently in the former than in the latter group. Some learning did take
place among the pseudofeedback subjects as well, though, because
their overall alpha production did increase significantly from the

beginning to the end of the experiment. The lack of correlation

between alpha production during training and in the pretraining base
rate indicates that this learning must have been far more sporadic than
in the contingent feedback group.
Although something changed systematically in the experimental
group and unsystematically in the control, these results may not be
direct effects of alpha training. Perhaps relaxation or accommodation
to the experimental procedures was indirectly facilitated by real
feedback and/or inhibited by false feedback.
The significant increase in alpha production for the control group
is of interest and requires comment. This increase may have been due
to the fact that some portion of the feedback for this group may have
occurred when alpha was actually being produced. The feedback
mechanism for this group consisted of the actual alpha record of a
subject in the experimental group. Whenever he received feedback, as
timed from the beginning of the session, the control subject received
feedback. Some portion of the feedback tone or flashing light probably
coincided with the production of alpha by the control subject, al-
though the exact amount is unknown. To assure that the control
subjects were receiving false feedback at all times would require cross-
filtering of alpha production of the control subjects in such a way that
the tone would have sounded only when there was no alpha being
produced. This was not done in the present study. A future study
might actually measure different degrees of pseudofeedback and test
for differential effects on alpha production.
Paskewitz and Orne (1973) have presented data that support the
idea that alpha production is largely under situational control. They
argue that attempts to increase alpha by feedback often mask the
impact on the subject of the experimental situation and do not take
into account the initial suppression of alpha to levels below base rate.
Instead of learning to increase alpha beyond base rate during subse-
quent feedback training, the subject is simply returning to baseline
from an initially suppressed level. So, according to Paskewitz and
Orne, observed alpha increases are more attributable to the subject's
acquisition of the ability to disregard the situational stimuli which
depress alpha below base rate than they are to significant increases
over a relaxed base-rate measure. This argument is one of perspective,
and it makes the increases observed in the control group more
understandable. In any case, the experimental group increased in

base-rate alpha production to a significantly greater degree than the

control group in the present study.


In a recent review of this area, Evans (1972) noted that most

studies trying to find EEG changes concurrent with hypnosis have
been anecdotal. He concluded that a definitive study should demon-
strate relevant EEG alpha changes in hypnotized subjects, without
similar changes in unhypnotizable subjects appropriately motivated to
experience hypnosis as well as they can (simulators).
Evans described preliminary data on separate groups of 12 hypno-
tized and 12 simulating subjects tested by an experimenter who was
"blind" as to which subjects were "real" and which were simulators.
Alpha frequency, amplitude, and density (percentage) were recorded
for all subjects, with eyes closed, during normal waking and deeply
hypnotized (or simulating) conditions, administered identically to
each subject by tape recording. These were scored blind. Simulating
subjects produced significantly more alpha than hypnotized subjects
in both waking (52% versus 27%) and hypnosis (or simulating) (45%
versus 17%) conditions (p < .001). All subjects produced a smaller
percentage of alpha density during the hypnosis (or simulating)
condition than in the waking condition (p < .01).
These are both striking results. The first finding is puzzling,
especially in view of repeated results from different laboratories
indicating an increase in alpha density during hypnosis, especially
among subjects who are highly susceptible (Engstrom, 1970; Morgan
and MacDonald, 1973; Ulett, Akpinar, and ltiI, 1972). One investiga-
tion (Ulett et al.) further reported an increase in amplitude of alpha
during trance induction and hypnosis. But in all cases, no comparison
was made with physiological changes in either simulating subjects or
highly motivated unsusceptible subjects.
Evans's second finding, that Simulating subjects produced more
alpha than those in the hypnosis group, is difficult to interpret as it
stands. Hypnotic susceptibility scores are not reported for either
group, nor is there an indication if these data were collected. As a
result, the effect of the real versus the simulator conditions is unclear,
since base-line differences between the two groups are not fully

known. The importance of this preliminary finding depends largely on

the differences in hypnotic susceptibility between groups.
Evans's study does raise a fundamental question implicit in any
physiological investigation of hypnosis. It has been shown by Orne
(1959) that unhypnotized subjects who are urged to simulate the role
of a hypnotized subject in an experiment exhibit many behaviors
previously attributable to hypnosis. These behaviors are often a result
of the exigencies of the experimental situation. The same question
may be asked about the EEG of hypnotized subjects. As an alternative
method, extremely high- and low-susceptible subjects might be com-
pared after the unsusceptibles learn to play the role of hypnotized
Diamond (1972) has recently demonstrated that susceptibility can
be significantly increased in unsusceptible subjects through exposure
to observationally presented information about hypnosis. Further-
more, there appear to be certain kinds of verbal information that are
more effective than others, and in fact the presentation of some kinds
of behavioral information about hypnosis leads to slight decreases in
susceptibility. The investigation suggests an informational, primarily
verbal technique for reliably enhancing the susceptibility scores of
previously unsusceptible subjects.
If susceptibility is increased by observationally presented verbal
information, is this increase accompanied by physiological changes
such as increased alpha density or reduced muscle tension? Or are the
effects of attitudinal and informational manipulation physiologically
In order to maximize the disparity in hypnotic susceptibility and
to expose the effects of experimental conditions on the EEG of high-
and low-susceptible subjects, a study was devised (Engstrom, 1973a)
to teach low-susceptible subjects to respond to hypnosis as high-
susceptible subjects do by specialized practice and to compare their
subsequent EEG alpha output during hypnotic induction with that of
initially high-susceptible subjects.
From a group of 67 students 30 volunteer subjects were selected
who were administered a tape-recorded version of the HGSHS. The
subjects represented the extremes of the distribution of obtained
susceptibility scores with one group of 10 scoring 9 and above, and
two groups of 10 scoring 3 and below on the Harvard scale. All
subjects were later administered the SHSS:A individually by a blind

experimenter. The means of the combined scale scores were 10.7, 2.6,
and 2.3 for each group of 10 subjects. Each of the 30 subjects was
scheduled for two laboratory sessions, spaced 1-4 days apart.
At the first session, on the day after individual susceptibility
screening, base-rate EEG and EMG measures were taken for each
Both EEG and EMG signals were input to a Grass Model 7
polygraph. Amplified output was displayed on a strip-chart recorder
for on-line monitoring, and amplified EEG and EMG signals were put
into two EPUT meters which were set to indicate the percentage of
time of brain-wave activity in the alpha range and the EMG activity in
microvolts, peak-to-peak, for each 5-minute trial. The subjects were
told to relax, sit still, and close their eyes, while 5 minutes of occipital
(02 ) EEG and frontalis EMG records were taken simultaneously.
On the second visit each subject was told that he would be given
approximately 45 minutes to become acquainted with the laboratory
surroundings. During this time 10 unsusceptible subjects were each
exposed to 45 minutes of observational information and coaching in
hypnosis. Two identical 45-minute presentations were videotaped
with a male and a female model presented to same-sex subjects. The
information presented was divided into the following general cate-
gories: (1) verbal modeling cues, consisting of both disinhibitory
information and facilitative information, 18 minutes (adapted from
Diamond, 1972); (2) motivational information, 7 minutes; (3) coaching
in the role skills involved in hypnosis and the behavior expected from
a hypnotized subject, 20 minutes. The other 20 subjects (10 susceptible
and 10 un susceptible) were left alone to read magazines or study in
the laboratory during the time. Although a monitor was present in the
room during the time, the subjects were instructed not to interact with
him. At the end of this time EEG and EMG electrodes were again
attached to each subject and a blind experimenter administered a
standardized hypnotic trance induction adapted from the SHSS:A,
with additional deepening instructions. During this 20-minute induc-
tion, 5-minute EEG and EMG records were taken at 5- and IS-minute
intervals. After trance induction and a 5-minute rest period, all
subjects were given the SHSS: B.
The results are presented in Table 5, Base-rate alpha was signifi-
cantly higher for the susceptible group than for either group of
unsusceptibles (p < .01), duplicating earlier findings that base-rate

Changes in Mean Susceptibility Scores before and after Treatment and in Alpha Density prior to and during Hypnotic

Mean pretreatment Alpha

susceptibility Significance of during Significance of
(HGSHS+ change (two- Base-rate hypnosis change (two-
Group SHSS:A) SHSS, B Change tailed) alpha (%) (%) Change (%) tailed)

Susceptible 10.7 9.8 -0.9 ns 39 47 8 .05

(No information)
(N = 10)
Unsusceptible 2.6 2.8 +0.2 ns 28 23 -5 ns
(No information)
(N = 10)
Unsusceptible 2.3 5.7 +3.4 .01 24 28 4 ns
(N = 10)


alpha and hypnotic susceptibility are related. The unsusceptible group

exposed to observational information was the only group to increase
hypnotic susceptibility scores significantly (p < .01), supporting the
results obtained by Diamond. This group did not significantly in-
crease alpha output during this time (p < 1), however, and only the
susceptible group increased significantly in alpha density during
trance induction testing (p < .05). This finding supports the observa-
tions of Ulett et al. (1972) that, at least among susceptible subjects,
there is a significant increase in alpha density during hypnotic trance
Base-rate EMG was not found to be significantly related to
hypnotic susceptibility or alpha.
Both the no-treatment un susceptible and the susceptible control
groups maintained fairly stable susceptibility scores, but the suscepti-
ble group showed significant increases in alpha during trance testing.
The unsusceptible controls showed no change in either alpha or EMG.
These results suggest that while observationally presented infor-
mation can significantly increase susceptibility, there are at least some
physiological differences during trance induction between subjects
who are initially susceptible and those whose susceptibility is
increased by the learning of the overt behaviors associated with
As a result of situational demands and external activities, physiol-
ogical functions may vary in subjects who are simulating hypnosis.
Playing the role of a hypnotized subject may simply produce changes
as a result of relaxation or other intervening variables. These results
must be cautiously interpreted, but they at least suggest the possibil-
ity that there can be physiological differences among highly hypnotiz-
able subjects.


Almost all of the previous studies have measured the EEG of

subjects either in a relaxed waking state or while performing hypnotic
behaviors. An alternative way of exploring susceptibility, frequently
used in studies of the relationship of susceptibility to performance and
personality measures, is to examine EEG differences among high- and

low-susceptible subjects in nonhypnotic tasks as well as those involv-

ing hypnosis.
Morgan and MacDonald (1973) recorded occipital alpha bilaterally
in 26 subjects during analytic (verbal and numerical), spatial (im-
agery), and music tasks, as well as during SHSS:A administration.
Although no lateral differences were found, highly hypnotizable
subjects produced more alpha activity, both outside of hypnosis and
within it, except during eyes-open base-line and eyes-open amnesia
conditions. The authors concluded that the fact that high-susceptible
subjects showed more alpha density and amplitude suggests that the
overall production of alpha may be positively related to the particular
cognitive mode that characterizes the subject who is able to experience
hypnotic phenomena, since few task-related EEG differences were
found. The high-susceptible subjects produced more alpha on most
tasks. This finding fits the conceptualization of hypnotic susceptibility
as a trait and suggests, as have other observed base-rate correlations,
that alpha output covaries with the trait of susceptibility.
In a recent study in our own laboratory (Engstrom, 1973b), alpha
output changes as a function of task, operation, and hypnotizability
were observed. Specifically, a comparison was made between alpha
output during biofeedback and hypnotic alteration of peripheral skin
Body-temperature regulation is one of the most stable, predict-
able, and automatic physiological functions in humans. Normally the
only circumstances under which it is not are when external environ-
mental conditions are extreme or when pathological internal condi-
tions are present, and in both cases the changes are involuntary.
Recently, however, several studies have demonstrated that this basic
regulatory body function can be brought under voluntary control by
two techniques that are vastly different from an operational point of
view: hypnosis and biofeedback.
The hypnosis literature includes several informal accounts of skin-
temperature alterations in one or a few subjects either assumed or
shown to be highly susceptible to hypnosis, but these are reported
anecdotally and lack any control subjects. One recent exception is a
study by Maslach, Marshall, and Zimbardo (1972), in which a group of
three subjects trained extensively in hypnosis were successful in
changing the temperature of their two hands in opposite directions,

while a group of six waking control subjects with no hypnosis training

were not, even when all the subjects received the same motivating
instructions and verbal suggestions. In spite of the small number of
subjects in the hypnosis group for definitive statistical purposes, the
difference between bilateral temperature changes of hypnotic and
control subjects is highly significant (p < .001). No measure of
hypnotic susceptibility was reported, and it would seem crucial to
know whether the subjects in each group were equivalent in their
ability to be hypnotized.
Similar results were obtained by Roberts, Kewman, and Mac-
Donald (1972) using 5 and 9 hours of both hypnosis and skin-
temperature feedback to train six subjects to raise the temperature of
one hand relative to the other. All the subjects attained significant
skin-temperature changes in the appropriate direction.
Although the authors of both of these studies parenthetically
noted that the feedback seemed either to inhibit or to have no effect
on performance when given alone, there are several examples of the
successful application of skin-temperature feedback in the literature.
Studies by Green, Green, and Walters (1970), as well as Taub and
Emurian (1972), describe the use of feedback to establish rapid operant
control over hand temperature, typically of the same differential
magnitude as the hypnosis studies. Feedback was provided by an
electronic differential thermometer that represented the temperature
differential between any two sites on the surface of the skin. A dial,
centered at zero, moved in one direction when one site was warmer
than the other and reversed direction when it was cooler.
None of these hypnosis or biofeedback studies present data on
the long-term retention of skin-temperature control, and nothing is
reported about the subsequent performance of subjects when the
external supports of trance induction and/or feedback information are
In each of these studies the subjective strategies of subjects were
idiosyncratic and diverse. Subjects used unilateral warming, unilateral
cooling, holding one hand constant while changing the other, or
changing the temperature of both hands. To do this, they reported
using realistic imagery, symbolic imagery, and imageless commands.
Clearly there was no meaningful pattern of cognitive strategy used by
successful subjects. The same is not true for the context of successful

control. The only similarity in set among the reports of successful

subjects was the concept of "passive volition." Subjects frequently
found that "trying too hard" proved futile in learning the task.
So peripheral skin temperature is an objective, easily measured
variable that it is possible to learn by several superficially different
operations. Its relevance to the EEG and susceptibility is derived from
the similarity of reports of subjects learning to control it. The phenom-
enon of "passive volition" described by Green et al. (1970) bears a
striking subjective similarity to the observations of Kamiya (1969) and
Engstrom (1970) of subjects trying to increase alpha output. In both
cases, when the subjects "tried too hard" they failed to increase the
desired response. And it was only after the subject reported increases
in subjective states called "passive," "relaxed," and "not trying" that
any progress was made.
Since alpha and skin-temperature changes apparantly must be
brought under voluntary control within the same context, or learning
set, perhaps they are related in some basic way. Our study sought to
compare the separate effects of the acquisition of bilateral hand-
temperature control by biofeedback and hypnosis training among
high- and low-susceptible subjects on EEG alpha output.
Of 102 student volunteers who were initially given the HGSHS via
tape recording, 42 who scored 10 or more or 3 or less were individually
administered the SHSS: C; 12 subjects who scored 10 or more on both
scales were randomly split into two groups of 6, and 12 subjects
scoring 3 or less on both scales were similarly divided into two
One group of 6 high-susceptibles and one group of 6 low-
susceptibles then received 11/2 hours of hypnosis training in a group
format. The training consisted of standardized deepening instructions,
verbal suggestions stressing the subjects' ability to achieve self-
hypnosis independently, and several depth-criterion tests. The other
two groups of 6 subjects spent the same amount of time in a group in
which they received general information about biofeedback, indivi-
dualized demonstration on a device irrelevant to the experimental task
(EMG biofeedback), and verbal information about the potential ability
to transfer skills learned via biofeedback to real-life situations. Every
effort was made to present parallel motivational sets and information
about hypnosis and biofeedback to the four groups, without specific
reference to the task to be subsequently performed.

All the subjects were individually tested in a temperature-con-

trolled room, relaxed in a comfortable chair with feet elevated. A
thermistor was taped to the center palm of each hand, and bilateral
occipital EEG leads were connected at 0 1 and O2 , Temperature
information was input to a digital feedback thermometer, and EEG
leads were input to a modified filtered dual trigger which provided
information on average alpha density from both leads to an EPUT
All subjects were told to relax with their eyes closed while a 2-
minute base-rate recording of peripheral skin temperature from both
hands and averaged bilateral occipital EEG measures were taken. The
task was explained to all the subjects in the same way. They were told
to try to increase the temperature of the right hand and decrease the
temperature of the left hand at the same time. They were further told
to remain still and relaxed and to keep their eyes colsed, with palms
facing upward.
The test procedure consisted of three lO-minute trials, each
ending with a 2-minute test. All the subjects were told to try to
maximize the temperature differential between the hands as much as
possible during the 2-minute tests.
The subjects in the hypnosis groups were given a standardized 8-
minute hypnotic induction procedure before the trials, while the
subjects in the biofeedback groups were not. The subjects in the
biofeedback groups got two types of feedback: continual auditory
feedback from a speaker that changed pitch with changes in tempera-
ture, and verbal feedback every 15 seconds about the disparity of the
temperature of their hands in degrees Fahrenheit. The hypnosis
subjects got no feedback during the test sequence. During the 2-
minute test at the end of each trial, temperature and EEG data were
recorded. All 24 subjects were rescheduled for return visits at 2- and 8-
week intervals, and on retest all the subjects were given the same task
and test sequence. The only difference was that no hypnotic induction
was administered to the hypnotic subjects and no feedback given
to the other groups-the task was presented in a straightforward
Mean temperature and EEG changes are summarized in Table 6.
Both the high-susceptible biofeedback and hypnosis groups and the
low-susceptible biofeedback group learned the task, and in fact all of
the subjects in both biofeedback groups and the high-susceptible

Mean Maximum Bilateral Temperature Differences and Mean Changes in EEG Alpha Density during Test Periods for All
Subjects and Conditions

difference during test Alpha during test Significance of change
Group (F') Base-rate alpha (%) period (%) Change (%) (one-tailed)

Unsusceptible 1.4 31 47 16 .05

Unsusceptible 4.2 29 60 31 .01
Susceptible 6.3 46 76 30 .01
Susceptible 6.0 39 71 32 .005

subjects in the hypnosis group demonstrated the ability to produce

bilateral changes in skin temperature. Mean temperature differences
between base rate and test scores were significant for three groups
(susceptible, hypnosis t = 6.17, P < .005; susceptible, biofeedback t =
5.91, P < .01; unsusceptible, biofeedback t = 4.79, P < .05) but
nonsignificant for the unsusceptible hypnosis group t = 1.99, P > .10).
Within this last group, 3 of the 6 subjects were able to control
temperature significantly.
Alpha output increased from base line during task performance
for all four groups (susceptible, hypnosis t = 7.06,p < .01; susceptible,
biofeedback t = 7.19, P < .005; unsusceptible, biofeedback t = 6.93, P
< .01; unsusceptible, hypnosis t = 3.87, P < .05), but the unsuscepti-
ble hypnosis group was significantly lower in alpha change than the
other three groups (p < .01).
The mean algebraic sum of temperature differences and alpha
density changes for all groups are shown in Figure 3. The pattern of
learning the temperature task was somewhat different for hypnotic
subjects, particularly high-susceptibles, whose performance across the
test sequences was characterized by rapid increases followed by stable
levels of response. The subjects in the biofeedback groups, by con-
trast, learned the response gradually, with slow, steady progress. The
only group to show no appreciable change was the unsusceptible
hypnosis group.
The mean change in alpha production for the four groups was
striking. Alpha output showed an overall correlation of .72 (p < .005)
with temperature changes for all test periods and groups. The unsus-
ceptible hypnosis group did not significantly change temperature and
showed lower overall changes in alpha density. Even the three
unsusceptible hypnosis subjects who were able to do the task to a
limited extent did not significantly change alpha output.
After 2 weeks, retention of temperature control was just above
70% for the hypnosis groups combined but only 61 % for the biofeed-
back groups. And after 8 weeks, the hypnotic subjects could still
produce nearly 70% of the original rate attained after hypnotic
induction, but the biofeedback groups dropped to 56%. In all, the
hypnotic subjects retained the original performance levels better than
the biofeedback subjects.
In this study, highly susceptible subjects produced more alpha
when performing the same kind of task, regardless of whether they

7 100 100


75 I
I ,
I ,

,, , 50

-------- 25 25
1 1

a -+--------10 a a
a b
1 2 3 1 2 3

7 100 7



I ""
...... ... I
25 25
1 1

O~----------------~ 0-11--------------10

c d
1 2 3 1 2 3
FIGURE 3. Mean bilateral skin temperature and alpha density scores for all subjects over
three test periods: (a) unsusceptible/hypnosis group, (b) Susceptible/hypnosis group, (c)
unsusceptible/biofeedback group, (d) susceptible/biofeedback group. - - Tempera-
ture, - - - - alpha density.

were hypnotized or not, from an operational point of view. One

explanation for these results is that what the subjects were doing in
both hypnosis and biofeedback conditions was identical. In other
words, although they were operationally different means to the same
end, the end was still the same. All the successful subjects learned to
assume a particular cognitive set-one that focused attention on the
specific body response and eliminated distracting stimuli. Now if the
nature of the task is a more important variable than susceptibility,
hypnotic susceptibility should not be critical. The subjects should all
be able to perform it and should produce more alpha when performing
it, regardless of hypnotizability. These data suggest, first, that among
high-susceptible subjects, biofeedback and hypnosis can be used to
alter skin temperature with about the same efficiency and that hyp-
notic training seems to lead to a more permanent skill in this regard.
But the more important implication is that there appears to be a
similar cognitive set, reflected in alpha output, that accompanies
successful performance of this task, which seems to transcend the
apparent operational differences between hypnosis and biofeedback.
This set seems much more directly related to hypnotizability and the
nature of the task and may suggest a more general state related to it.


The origins of the EEG in terms of its neural mechanisms or

behavioral determinants are still unknown. And whether increased
EEG alpha output represents a controlled increase over base rate or,
conversely, a return to base rate following its initial suppression is an
equivocal issue. What is clear, however, is that alpha output in a
standardized situation is a variable ability among subjects, and
whatever the attentional skills required to produce high alpha rates,
some subjects have them and some do not.
Enough evidence has already accumulated to support a moderate
relationship between alpha output and hypnotic susceptibility. From
the present evidence it is further apparent that alpha density is fairly
stable over time and that its sustained increase, by different methods,
can result in increased susceptibility scores. Also some means of
increasing measured hypnotic susceptibility, such as role-skills learn-
ing, can increase behavioral responsiveness to hypnosis without

increasing alpha density. If this finding is confirmed, it implies

that there are several variables within susceptibility, that the EEG is
critical in their separation, and that role theory alone cannot explain
This finding magnifies some of the definitional differences re-
ferred to earlier. Hypnosis is both a state and an operation, and the
state cha:racteristics are partially define by the EEG.
Researchers and reviewers in the area are often too quick to
ignore the physiological study of hypnosis, citing shaky and often
contradictory experimental results. But the results are really very
decent, in light of the sparse and crude effort expended. With more
specificity the EEG could become a refined criterion variable, espe-
cially valuable to the study of hypnosis and other self-regulatory
There are many fruitful areas still largely unexplored. For norma-
tive purposes, alpha base-rates might be compared between suscepti-
ble and un susceptible subjects during specific task performance,
instead of in the relaxed, waking state, which often makes them too
nosey anyway. Also very few data are now available for item analysis
regarding EEG changes during the administration of hypnotic suscepti-
bility scales.
More specificity is also possible through the laterality measure of
the EEG. More attention should be paid to the relationship between
laterality and susceptibility and other personality variables.
Although many recent studies demonstrate that hypnotic suscep-
tibility can be increased, the areas of sensory deprivation and drug
effects on both susceptibility and the EEG are still largely unexplored.
In these cases physiological changes may accompany psychological
ones, leading to greater understanding of how the two interrelate.
And much needs to be added to knowledge of the relationships
between different operations, such as hypnotic induction, verbal
motivational procedures, and biofeedback, with regard to task per-
formance and EEG criteria. Only by these increasingly specific tests
will the effects of hypnosis become clear.
Hypnosis is one operation that permits muscular relaxation,
enhancement of concentration of attention on the relevant task, and
the complementary removal of distracting stimuli from attention.
Highly susceptible subjects are better at performing some tasks that

require this ability following different operations including but not

limited to hypnosis.
In this sense, biofeedback, hypnosis, meditation, and other train-
ing operations which enhance these abilities should have a similar
effect on highly susceptible subjects, reflected in the EEG. They all
permit control of behavioral and physiological processes by altering
consciousness in such a way that verbal and symbolic modalities can
be transformed into specific responses.
Hypnosis is just one operation by which an individual can change
his perception of his own control of internal and external responses, or
his attributional system. And susceptibility to hypnosis is an impor-
tant variable in determining the extent of this change.


As, A., HILGARD, E. R., AND WEITZENHOFFER, A. M. An attempt at experimental

modification of hypnotizability through repeated individualized hypnotic experi-
ence. Scandinavian Journal of Psychology, 1963,4, 81-89.
BARBER, T. X. Hypnosis: A scientific approach. New York: Van Nostrand, 1969.
BARBER, T. X. LSD, marihuana, yoga, and hypnosis. Chicago: Aldine, 1970.
BARBER, T. X., AND CALVERLEY, D. S. Hypnotic-like suggestibility in children and
adults. Journal of Abnormal and Social Psychology, 1963,66, 589-97.
BARBER, T. X., AND CALVERLEY, D. S. Toward a theory of hypnotic behavior: Experimen-
tal evaluation of Hull's postulate that hypnotic susceptibility is a habit phenome-
non. Journal of Personality, 1966,34, 416--433.
BARBER, T. X. AND GLASS, L. B. Significant factors in hypnotic behavior. Journal of
Abnormal and Social Psychology, 1962,64, 222-228.
BARKER, W., AND BURGWIN, S. Brain wave patterns during hypnosis, hypnotic sleep and
normal sleep. Archives of Neurology and Psychiatry, 1949,62, 412-420.
BECK, E. c., AND BAROLIN, G. S. Effect of hypnotic suggestions on evoked potentials.
Journal of Nervous and Mental Disease, 1965,140, 154-16l.
BECK, E. c., DUSTMAN, R. E., AND BEIER, E. G. Hypnotic suggestions and visually
evoked potentials. Electroencephalography and Clinical Neurophysiology, 1966, 20,
BERNHEIM, H. Hypnosis and suggestion in psychotherapy. Reprinted, New Hyde Park,
N.Y.: University Books (1888), 1964.
BLUM, G. S. Programming people to simulate machines. In S. S. TOMKINS AND S.
MESSICK (Eds.), Computer simulation of pt!rsonality. New York: Wiley, 1963.
BRAID, J. Neurypnology: Or the rationale of nervous sleep considered in relation to animal
magnetism. London: Churchill, 1843.
CHARCOT, J. M. Physiologie pathologique: Sur les divers etats nerveux determines pour
l'hypnotization chez les hysteriques. CR Academy of Science, 1882, 94, 403-405.
CHERTOK, L., AND KRAMARZ, P. Hypnosis, sleep and electroencephalography. Journal of
Nervous and Mental Disease, 1959, 128, 227-238.

CLYNES, M., KOHN, M., AND LIFSHnz, K. Dynamics and spatial behavior of light evoked
potentials, their modification under hypnosis, and on-line correlation in relation to
rhythmic components. Annals of the New York Academy of Science, 1963, 112, 468-
COOPER, L. M., BANDORD, S. A., SCHUBOT, E., AND TART, C. T. A further attempt to
modify hypnotic susceptibility through repeated individualized experience. Journal
of Clinical and Experimental Hypnosis, 1967,15, 118-124.
COOPER, L. M., AND LONDON, P. Sex and hypnotic susceptibility in children. Journal of
Clinical and Experimental Hypnosis, 1966, 14, 55-60.
DAVIS, L. W., AND HUSBAND, R W. A study of hypnotic susceptibility in relation to
personality traits. Journal of Abnormal and Social Psychology, 1931,26, 175-182.
DIAMOND, M. J. The use of observationally presented information to modify hypnotic
susceptibility. Journal of Abnormal Psychology, 1972,79, 174-180.
DYNES, J. B. Objective method for distinguishing sleep from the head hypnotic trance.
Archives of Psychiatry and Neurology, 1947,57, 84-93.
EDMONSTON, W. E. Stimulus-response theory of hypnosis. In J. E. GORDON (Ed.),
Handbook of clinical and experimental hypnosis. New York: Macmillan, 1967.
ENGSTROM, D. R. The enhancement of EEG alpha production and its effects on hypnotic
susceptibility. Unpublished doctoral dissertation, University of Southern Califor-
nia, 1970.
ENGSTROM, D. R. Previous exposure to feedback: A subject variable in psychophysiolog-
ical hypnosis research. Paper presented at meeting of American Psychological
Association, Washington, D.C., September, 1971.
ENGSTROM, D. R. Interactional effects of muscle tension and EEG alpha production on
hypnotic susceptibility. Paper presented at meeting of American Psychological
Association, Honolulu, September, 1972.
ENGSTROM, D. R. Effects of observationally presented information on hypnotizability:
Physiology of enhanced susceptibility. Proceedings, Annual Convention, American
Psychological Association, 1973a.
ENGSTROM, D. R. Task-specific EEG output among highly hypnotizable subjects. Paper
presented at meeting of Society Clinical and Experimental Hypnosis, Newport
Beach, California, December, 1973b.
ENGSTROM, D. R, LONDON, P., AND HART, J. T. Hypnotic susceptibility increased by
EEG alpha training. Nature, 1970,227, 1261-1262.
EVANS, F. J. Hypnosis and sleep: Techniques for exploring cognitive activity during
sleep. In E. FROMM AND R. E. SHOR (Eds.), Hypnosis: Research developments and
perspectives. Chicago: Aldine-Atherton, 1972.
FAW, V., AND WILCOX, W. W. Personality characteristics of susceptible and unsuscepti-
ble hypnotic subjects. Journal of Clinical and Experimental Hypnosis, 1958,6, 83-94.
FORD, W. L., AND YEAGER, C. L. Changes in the electroencephalogram in subjects under
hypnosis. Diseases of the Nervous System, 1948,9, 190-192.
FRIEDLANDER, J. W., AND SARBIN, T. R The depth of hypnosis. Journal of Abnormal and
Social Psychology, 1938,33, 453-475.
and hypnotic susceptibility. Journal of Comparative and Physiological Psychology,
1970, 72, 125-13l.
GILL, M. M., AND BRENMAN, M. Hypnosis and related states: Psychoanalytic studies in
regression. New York: International Universities Press, 1959.
GREEN, E. E., GREEN, A. M., AND WALTERS, E. D. Self-regulation of internal states. In J.
ROSE (Ed.), Progress of cybernetics: Proceedings of the International Congress of
Cybernetics, London, 1969. London: Gordon and Breach, 1970.

HALLIDAY, A. M., AND MASON, A. A. Cortical evoked potentials during hypnotic

anesthesia. Electroencephalography and Clinical Neurophysiology, 1964,16, 314.
HART, J. T. Autocontrol of EEG alpha. Paper presented at meeting of the Society for
Psychophysiological Research, San Diego, October, 1967.
HARTNETT, J., NowLIs, D., AND SVORAD, D. Hypnotic susceptibility and EEG alpha:
Three correlations. Hawthorne House Research Memorandum, No. 97, 1969.
HILGARD, E. R. Hypnotic susceptibility. New York: Harcourt, Brace and World, 1965.
HILGARD, E. R. Individual differences in hypnotizability. In J. E. GORDON (Ed.),
Handbook of clinical and experimental hypnosis. New York: Macmillan, 1967.
HILGARD, E. R., WEfIZENHOFFER, A. M., AND GOUGH, P. Individual differences in
susceptibility to hypnosis. Proceedings of the National Academy of Science, 1958, 44,
HILGARD, E. R., WEIlZENHOFFER, A. M., LANDES, J., AND MOORE, R. K. The distribution
of susceptibility to hypnosis in a student population: A study using the Stanford
Hypnotic Susceptibility Scale. Psychological Monographs, 1961, 75, 1-22.
HILGARD, J. R. Personality and hypnosis: A study of imaginative involvement. Chicago:
University of Chicago Press, 1970.
KAMIYA, J. Operant control of the EEG alpha rhythm and some of its reported effects on
consciousness. In C. T. TART (Ed.), Altered states of consciousness. New York: Wiley,
KRAMER, E., AND BRENNAN, E. P. Hypnotic susceptibility of schizophrenic patients.
Journal of Abnormal and Social Psychology, 1964,69, 657-659.
LnlBEAULT, A. A. Le sommeil provoque et les etats analogues. Paris: Doin, 1889.
LINDSLEY, D. B., AND WICKE, J. D. The electroencephalogram: Autonomous electrical
activity in man and animals. In R. F. THOMPSON AND M. M. PATTERSON (Eds.),
Bioelectric Recording Techniques-Part B. New York: Academic Press, 1974.
LONDON, P. Childrens Hypnotic Susceptibility Scale. Palo Alto, California: Consulting
Psycholgists Press, 1962.
LONDON, P. Developmental experiments in hypnosis. Journal of Projective Techniques and
Personality Assessment, 1965,29, 189-199.
LONDON, P. The induction of hypnosis. In J. E. GORDON (Ed.), Handbook of clinical and
experimental hypnosis. New York: Macmillan, 1967.
LONDON, P., AND COOPER, L. M. Norms of hypnotic susceptibility in children.
Developmental Psychology, 1969,1, lU-124.
LONDON, P., COOPER, L. M., AND ENGSTROM, D. R. Increasing hypnotic susceptibility by
brain wave feedback. Journal of Abnormal Psychology, 1974, 83, 554-560.
LONDON, P., COOPER, L. M., AND JOHNSON, H. Subject characteristics in hypnosis
research: II. Attitudes toward hypnosis, volunteer status and personality measures;
III. Some correlates of hypnotic susceptibility. Journal of Clinical and Experimental
Hypnosis, 1962,10, 13--21.
LONDON, P., AND FUHRER, M. Hypnosis, motivation and performance. Journal of
Personality, 1961, 29, 321-333.
LONDON, P., HART, J. T., AND LEIBOVIlZ, M. P. EEG alpha rhythms and susceptibility to
hypnosis. Nature, 1968,219, 71-72.
LONDON, P., AND McDEVITT, R. A. AMRL-TR-67-142 (W-P AF Base, Ohio: Aerospace
Medical Research Laboratories, 1967).
LONDON, P., AND ROCHMAN, G. Untitled mimeographed paper, Department of Psychol-
ogy, University of Southern California, 1967.
LOOMIS, A. L., HARVEY, E. N., AND HOBART, G. Brain potentials during hypnosis.
Science, 1936,83, 239-241.

MASLACH, c., MARSHALL, G., AND ZIMBARDO, P. Hypnotic control of peripheral skin
temperature. Psychophysiology, 1972, 9, 60~05.
MELEI, J., AND HILGARD, E. R. Attitude toward hypnosis, self-predictions and hypnotic
susceptibility. Journal of Clinical and Experimental Hypnosis, 1964, 12, 99-108.
MOORE, R. K., AND LAUllR, L. W. Hypnotic susceptibility in middle childhood. Journal of
Clinical and Experimental Hypnosis, 1963,11, 167-174.
MORGAN, A. H., AND HILGARD, E. R. Age differences in susceptibility to hypnosis.
Journal of Clinical and Experimental Hypnosis, 1973,21, 78-85.
MORGAN, A. H., JOHNSON, D. L., AND HILGARD, E. R. The stability of hypnotic
susceptibility: A longitudinal study. Journal of Clinical and Experimental Hypnosis,
1974,22, 249-257.
MORGAN, A. H., AND MACDONALD, H. EEG alpha: Lateral asymmetry related to type of
task, task difficulty, and hypnotizability. Paper presented at meeting of the Society
for Clinical and Experimental Hypnosis, Newport Beach, California, December,
MORGAN, A. H., McDONALD, P. J., AND MACDONALD, H. Differences in bilateral alpha
activity as a function of experimental task, with a note on lateral eye movements
and hypnotizability. Neuropsychologia, 1971,9,459-469.
NOWLIS, D., AND RHEAD, J. c. Relation of eyes-closed resting EEG alpha activity to
hypnotic susceptibility. Perceptual and Motor Skills, 1968,27, 1047-1050.
ORNE, M. T. The nature of hypnosis: Artifact and essence. Journal of Abnormal and Social
Psychology, 1959,58,277-299.
ORNE, M. T., AND EVANS, F. J. Inadvertent termination of hypnosis with hypnotized and
simulating subjects. Journal of Clinical and Experimental Hypnosis, 1966, 14, 61-78.
ORNE, M. T., AND O'CONNELL, D. N. Diagnostic rating of hypnotizability. Journal of
Clinical and Experimental Hypnosis, 1967, 15, 125-133.
PASKEWITZ, D. A., AND ORNE, M. T. Visual effects of alpha feedback training. Science,
1973,181, 360-363.
ROBERTS, A. H., KEWMAN, D. G., AND MACDONALD, H. Voluntary control of skin
temperature: Unilateral changes using hypnosis and auditory feedback. Paper
presented at meeting of Biofeedback Society, 1972.
ROSENHAN, D., AND LONDON, P. Hypnosis: Expectation, susceptibility and performance.
Journal of Abnormal and Social Psychology, 1963,66, 77-81.
ROSENHAN, D., AND TOMKINS, S. S., On preference for hypnosis and hypnotizability.
Journal of Clinical and Experimental Hypnosis, 1964, 12, 109-114.
SACHS, L. B., AND ANDERSON, W. L. The modification of hypnotic susceptibility. Journal
of Abnormal Psychology, 1967,15, 172-180.
SANDERS, R. S., AND REYHER, J. Sensory deprivation and the enhancement of hypnotic
susceptibility. Journal of Anbnormal Psychology, 1969,74, 375-381.
SARBIN, T. R., AND COE, W. C. Hypnosis: A social psychological analysis of influence
communication. New York: Holt, Rinehart and Winston, 1972.
SARBIN, T. R., AND MADOW, L. W. Predicting the depth of hypnosis by means of the
Rorschach test. American Journal of Orthopsychiatry, 1942,12, 268-27l.
SCHAEFLER, K., AND LONDON, P. Untitled mimeographed paper, Department of Psychol-
ogy, University of Southern California, 1968.
SCHAFER, R. A study of personality characteristic related to hypnotizability. Unpub-
lished masters thesis, Department of Psychology, University of Kansas, 1947.
SCHULMAN, R. E., AND LONDON, P. Hypnotic susceptibility and MMPI profiles. Journal
of Consulting Psychology, 1963,27, 157-160.
SHAPIRO, J., AND DIAMOND, M. J. Increases in hypnotizability as a function of encounter

training: Som,e confirming evidence. Journal of Abnormal Psychology, 1972, 79, 112-
SHOR, R E. Hypnosis and the concept of the generalized reality orientation. American
Journal of Psychotherapy, 1959,13, 582-602.
SHOR, R E., AND COBB, J. c. An exploratory study of hypnotic training using the
concept of plateau responsiveness as a referent. American Journal of Clinical
Hypnosis, 1968,10, 178-193.
SHOR, R E., AND ORNE, E. C. Harvard Group Scale of Hypnotic Susceptibility. Palo Alto,
Calif.: Consulting Psychologists Press, 1962.
SHOR, R. E., ORNE, M. T., AND O'CONNELL, D. N. Validation and cross-validation of a
scale of self-reported personal experiences which predicts hypnotizability. Journal of
Psychology, 1962,53, 55-75.
STEVENS, J. R, SACHDEV, K., AND MILSTEIN, V. Behavior disorders of childhood and the
electroencephalogram. Archives of Neurology, 1968,18, 160-177.
STUKAT, K. G. Suggestibility: A factorial and experimental analysis. Stockholm: Almqvist
and Wiksall, 1958.
TART, C. T. Increases in hypnotizability resulting from a prolonged program for
enhancing personal growth. Journal of Abnormal Psychology, 1970,75, 260-266.
TAUB, E., AND EMURIAN, C. Autoregulation of skin temperature using a variable
intensity feedback light. Paper presented at meeting of the Biofeedback Research
Society, 1972.
TRUE, R. M., AND STEPHENSON, C. W. Controlled experiments correlating electroenceph-
alogram, pulse and plantar reflexes with hypnotic age regression and induced
emotional states. In M. V. KLINE (Ed.), Clinical correlations of experimental hypnosis.
Springfield, Ill.: Thomas, 1963.
ULETT, G. A., AKPINAR, S., AND ITIL, T. M. Hypnosis: Physiological, pharmacological
reality. American Journal of Psychiatry, 1972,128, 799-805.
WEITZENHOFFER, A. M. Hypnotism: An objective study in suggestibility. New York: Wiley,
WEITZENHOFFER, A. M., AND HILGARD, E. R. Stanford Hypnotic Susceptibility Scale, Forms
A and B. Palo Alto, Calif.: Consulting Psychologists Press, 1959.
WEITZENHOFFER, A. M., AND HILGARD, E. R. Stanford Hypnotic Susceptibility Scale, Form
C. Palo Alto, Calif.: Consulting Psychologists Press, 1962.
WEITZENHOFFER, A. M., AND WEITZENHOFFER, G. B. Sex, transference and susceptibility
to hypnosis. American Journal of Clinical Hypnosis, 1958,1, 15-24.
WICKRAM, 1. Effects of EMG feedback training on hypnotic susceptibility: More
preliminary observations. Paper presented at meeting of the American Psychologi-
cal Association, Honolulu, September, 1972.
WICKRAMASEKERA, 1. Effects of sensory restriction on susceptibility to hypnosis: A
hypothesis and more preliminary data. Journal of Abnormal Psychology, 1970,76, 69-
WINER, B. J. Statistical principles in experimental design (2nd ed.). New York: McGraw-
Hill, 1971.
WISEMAN, R J., AND REYHER, J. A. A procedure utilizing dreams for deepening the
hypnotic trance. American Journal of Clinical Hypnosis, 1962,5, 105-110.
ZUBEK, J. P. Behavioral and physiological effects of prolonged sensory and perceptual
deprivation: A review. In J. RASMUSSEN (Ed.), Man in isolation and confinement.
Chicago: Aldine, 1973.
ZUBEK, J. P., AND WELCH, G. Electroencephalographic changes after prolonged sensory
and perceptual deprivation. Science, 1963,139, 1209-1210.
6 Toward a Cognitive
Theory of Self-Control



For the last 10 years we have been conducting research designed to

bring together the clinical concerns of semantic, or cognitive, thera-
pists (e.g., Aaron Beck, Albert Ellis, Jerome Frank, George Kelley) and
the technology of behavior therapy (e.g., procedures such as operant
and aversive conditioning, desensitization, modeling, and behavioral
and imagery rehearsal). This marriage of somewhat strange bedfellows
has bred a set of therapy procedures that we have come to call
cognitive-behavior modification. At one time we tended to call the
procedures self-instructional training, but this title was too delimiting,
not permitting ample recognition of imagery- and fantasy-based
factors in the change process. This program of research has been
described elsewhere (Meichenbaum, 1973, 1975b; Meichenbaum and
Cameron, 1974). These studies have indicated the promising outcome,
in terms of generalization and persistence of treatment effects, that
follow from the alteration of "standard" behavior-therapy procedures
to include self-instructional and imagery processes. For example, the
efficacy of behavior-therapy procedures such as modeling (Meichen-
baum, 1971), desensitization (Meichenbaum, 1972), operant condition-
ing (Meichenbaum and Goodman, 1971), and aversive conditioning
(Steffy, Meichenbaum, and Best, 1970) was enhanced by the focusing
of treatment of the client's cognitive processes. See Mahoney's (1974)
recent book for a review of the cognitive-behavior modification
The purpose of the present paper is (1) to share some general
conclusions that derive from our treatment research and (2) to offer a

DONALD MEICHENBAUM . Department of Psychology, University of Waterloo,

Waterloo, Ontario.


cognitive theory of self-control based on these conclusions. Specifi-

cally, an attempt will be made to explain why modifying the client's
internal dialogue (i.e., self-statements and images) results in behavior


Let me first state the general conclusions that have issued from
our treatment research; then we can explore each of them in more
detail. First, it has become increasing clear that there are a host of
different ways to view our clients' cognitions but that at present we
have few or no data to determine the relative merit or long-term
effectiveness of these different approaches. A related point is that no
matter how one views his client's cognitions, the distinction between
a purely behavioral versus a cognitive intervention program is mis-
leading and mistaken. Perhaps we can help put to rest the false
distinction between behavioral and cognitive therapies by an interac-
tional model, in which the behavioral and the cognitive processes that
underlie change are interdependent.
Another general conclusion is that therapeutic change comes
about by means of a sequential, mediating process, in which (1) the
client becomes aware of his maladaptive intra- and interpersonal
behaviors; (2) this self-recognition is the occasion for the client to emit
a set of incompatible images and self-statements and incompatible
behaviors; (3) finally, what the client says to himself (i.e., his
appraisals, attributions, self-statements, and images), following the
emission of the new behavioral act and its accompanying conse-
quences, will influence the nature and stability of the change. As far
as we, as therapists, can anticipate and subsume the content of the
client's internal dialogue in our treatment package, we will be that
much more effective. By analogy, the neurological concept of final
common pathway explains how behavior change follows from diverse
therapy procedures. It is suggested that clients who see therapists of
wholly different persuasions go through similar psychological proc-
esses in achieving behavioral change. The final common pathway to
behavior change is the alteration in the internal dialogues in which
our clients engage.
The final and perhaps the most important conclusion is that we as

researchers have not paid sufficient attention to what happens in

therapy prior to the point of implementation of the behavior-modifica-
tion procedures. Little has been written about the initial, conceptuali-
zation phase of therapy.
In summary, the conclusions I wish to emphasize concern first,
the variety of ways therapists conceptualize their clients' thinking
processes and the interactional role of cognitive and behavioral proc-
esses; second, the suggestion that behavioral change is mediated by
means of changes in our clients' internal dialogues; and finally, the
significant role of the initial, conceptualization phase in the therapy

A. How Shall We Treat Our Clients' Cognitions?

When I see a client for assessment and perform a situational

analysis of his presenting problem (a la Peterson, 1968, or Kanfer and
Saslow, 1969), I also ask him to share with me the feelings and
thoughts he has that precede, accompany, and follow the presenting
problem. I try to secure his description, his perception, and the
meaning that he has fabricated to explain his behavior. I wish to
discern from his point of view what is going on, what led up to his
present difficulties, and what he thinks should be done to help him.
Up to this point I think my clinical behavior is not atypical of
clinicians, no matter of what particular persuasion or orientation. My
behavior-therapy and psychoanalytic colleagues are likely to have
similar concerns. But it is how we conceptualize our client's answers
to these questions that will illuminate the variety of clinical orienta-
tions. More specifically, the role and the significance attributed to the
client's cognitions seem to be the fulcrum that truly distinguishes the
various clinical approaches. I would like to describe seven different
ways in which therapists have viewed their clients' cognitions. Where
appropriate, I will illustrate from our own research program the
particular conceptualization.

1. Cognitions as Behaviors
How shall we treat our client's cognitions? Shall we view the
client's thoughts that precede, accompany, and follow the maladaptive

behaviors as "behaviors," similar in nature to other nonverbal behav-

iors that he emits and subject to the same "laws of learning" and
social learning principles, such as reinforcement and contingency
manipulations? There is a long tradition in making such a continuity
assumption between overt and covert events, going back to Dollard
and Miller (1950) and Skinner (1953). Homme (1965) has even offered
the term coverant (that is, covert-operant) to describe our client's
thinking processes. If we view the client's cognitions in this manner,
then it suggests that we can affect the frequency and strength of those
thoughts by systematically pairing them with the onset and offset of
various consequences. Indeed, there is a host of therapeutic proce-
dures, including covert sensitization and anxiety-relief conditioning,
that are based upon this notion. However, recent treatment studies by
Ashem and Donner (1968), Marshall, Polgrin, and Boutilier (1974),
Marks, Boulougouris, and Marset (1971), Meichenbaun and Cameron
(1973a), and Sachs and Ingram (1972) have seriously questioned the
continuity assumption. When the contingency variable in procedures
such as covert sensitization or anxiety-relief conditioning was inverted
or made noncontingent, treatment efficacy did not deteriorate. These
studies question the validity of viewing and treating our client's
cognitions in the same manner as overt behaviors. (See Mahoney,
1974, and Meichenbaum, 1974a, for a more detailed review of this

2. Cognitions as Part of the Response Chain

Perhaps instead we should view the client's cognitions as in-
stances of automatic thoughts (i.e., images and self-statements), which
are only part of the maladaptive response chain. According to this
conception, the task of therapy is to have the client become aware of
the role such thoughts play in the behavioral sequence. A number of
theorists, such as Premack (1970) and Bergin (1967), have emphasized
the therapeutic value of having the client interrupt the maladaptive
response chain by controlling automatic thoughts and producing
incompatible self-instructions and images.
This viewpoint maintains that target behaviors that are habitual
in nature (Le., not premeditated) should first be returned to a
"de automatized" condition, in which the habitual maladaptive behav-
iors come to be preceded by cognitive activity occurring within the

client's awareness. Such "forced mediation" increases the separation

between stimuli and responses and thereby provides an additional
opportunity for interrupting the behavioral sequence. In this way one
can impose an interruption of the response chain, thus increasing the
likelihood of termination of the sequence at an earlier stage and
production of incompatible thoughts, images, and behaviors. Illustra-
tive of this approach is a self-instructional training program developed
by Meichenbaum and Goodman (1971) to treat hyperactive, impulsive
The impetus for self-instructional training program came from the
work of the Soviet psychologists Luria (1961) and Vygotsky (1962).
Luria suggested that the child goes through three stages in developing
internalized control of behavior. His performance is first controlled by
the verbal instructions and reactions of external agents (e.g., parents).
The child then begins to regulate some of his own actions through
audible self-talk. Finally, these self-statements become covert (i.e., go
"underground," to use Vygotsky's term) and expand their extensive
regulatory influence. The Meichenbaum-Goodman self-instructional
training program followed, in abbreviated form, such a developmental
progression. In order to achieve covert self-instructional control of
behavior in hyperactive children, the training regimen was as follows:

1. An adult model performed a task while talking to himself out loud

(cognitive modeling),
2. The child performed the same task under the directions of the
model's instructions (overt, external guidance),
3. The child performed the task while instructing himself aloud (overt
4. The child whispered the instructions to himself as he went through
the task (faded, overt self-guidance),
5. And finally, the child performed the task while guiding his per-
formance via private speech (covert self-instruction).

Over a number of training sessions the package of self-statements

modeled by the experimenter and rehearsed by the child (initially
aloud and then covertly) was enlarged by means of response chaining
and successive approximation procedures. For example, in a task that
required the copying of line patterns, the examiner performed the task
while cognitively modeling as follows:

Okay, what is it I have to do? You want me to copy the picture with
the different lines. I have to go slowly and carefully. Okay, draw the line
down, down, good; then to the right, that's it; now down some more and
to the left. Good, I'm doing fine so far. Remember, go slowly. Now back
up again. No, I was supposed to go down. That's okay. Just erase the line
carefully . . . Good. Even if I make an error I can go on slowly and
carefully. I have to go down now. Finished. I did it!
In the thinking-out-Ioud phase the model displays several per-
formance-relevant skills: (1) problem definition ("What is it I have to
do?"); (2) focusing attention plus response guidance ("Be care-
ful ... draw the line down."); (3) self-reinforcement ("Good, I'm
doing fine."); and (4) self-evaluative coping skills plus error-correcting
options ("That's okay, even if I make an error I can go slowly.").
Such training, provided over a number of different tasks, was
successful in causing hyperactive children to learn to think before they
act, to employ mediational processes, and to develop verbal control of
behavior (Meichenbaum and Goodman, 1971). A number of other
investigators have also successfully trained children to bring their
behavior under self-instructional and imagery control (d. Bem, 1967;
Palkes, Stewart, and Kahana, 1968; Blackwood, 1970; Ridberg, Parke,
and Hetherington, 1971; Monahan and O'Leary, 1971; Palkes, Stewart,
and Freedman, 1972; Denny, 1972; Hartig and Kanfer, 1973; Mischel,
1974; Schneider, 1974). In each of these studies self-control was
enhanced as the involuntary act was made voluntary. This was
accomplished as the child's behavior was brought under his own
cognitive control through the emission of deliberate self-statements
and images. Then, with the development of task proficiency, or what
Kimble and Perlmutter (1970) call the "automatization of voluntary
acts," the child's private speech became more abrupt, incomplete, and
whispered and then completely vanished. This process of abbreviation
and interiorization of private speech also applies to adults as they
acquire skills. For example, one can imagine a similar sequence in the
learning of a new motor skill such as driving a car. As Henry Murray
(1938) noted some years ago;
When one is learning to drive an automobile, one is, at first, aware of
every accessory intention and subsequent motor movement, but later,
when proficiency has been attained, the details of the activity are seldom
in consciousness. (p. 51)

In other words, early in the mastery of a voluntary act, speech serves

a useful supporting function. With practice, these verbalizations

Thus the therapist can assess the role of his client's cognitions as
part of the response chain and help his client produce intentional,
adaptive cognitions (Le., self-instructions and images) that will inter-
rupt the maladaptive behavioral chain and foster incompatible, adap-
tive behavior. It should be noted that the client must learn to use his
own behaviors, feelings, and thoughts as well as the behavior of
others as cues or signals to engage in this newly acquired, internal
dialogue. The importance of self-observation in the change process
will be more fully discussed below.

3. Cognitions as Instances of Irrational Thinking Styles

We have viewed our client's cognitions as instances of behavior

per se, or as part of the response chain. There are alternatives. Aaron
Beck has been most helpful in noting some of the stylistic qualities of
client's cognitions, especially depressed clients (Braff and Beck, 1974).
Beck (1970a) attempts to have clients become aware of the distortions
in their thought patterns. These distortions include (1) arbitrary
inference--the drawing of a conclusion when evidence is lacking or is
actually contrary to the conclusion; (2) magnification-exaggeration of
the meaning of an event; (3) cognitive deficiency-disregard for an
important aspect of a life situation; (4) dichotomous reasoning-overly
simplified and rigid perception of events as good or bad, right or
wrong; and (5) overgeneralization-taking a single incident such as
failure as a sign of total personal incompetence, and in this way
generating a fallacious rule. Such cognitive distortions result in the
dient's selectively attending to and inaccurately anticipating conse-
quences and in his making logical errors. By means of pinpointing
such stylistic qualities, the client is brought to understand that his
affective experiences and maladaptive behaviors are a result of his
faulty thinking processes-thinking processes that the client is capable
of changing and controlling. Note that the focus has shifted from
treating the client's cognitions as an instance of behavior to the
stylistic qualities of the client's thinking processes.
It is important to underscore Beck's (1970b, 1971) observation that
the client's faulty cognition may often take a pictorial form instead of,
or in addition to, the verbal form. For example, Beck reported that a
woman with a fear of walking alone found that her spells of anxiety
followed images of her having a heart attack and being left helpless; a

college student discovered that her anxiety of leaving the dormitory

was triggered by visual fantasies of being attacked. As Goldfried,
Decenteceo, and Weinberg (1974) have indicated, because of the
habitual nature of one's expectations or beliefs it is likely that such
thinking processes and images become automatic and seemingly
involuntary, like most overlearned acts. The client's faulty cogni-
tions-negative and anxiety-engendering self-statements and im-
ages-become habitual and in many ways are similar to the automati-
zation of thought that accompanies the proficiency of a motor skill
such as driving a car. However, the therapist can make the client
aware of such thinking processes and increase the likelihood that such
an awareness will be the trigger that produces incompatible thoughts
and behaviors. (See Meichenbaum, 1975b, for a description of how
the clinician can achieve this process.)

4. Cognitions as Instances of Irrational Belief Systems

Let us continue the list of alternative conceptualizations available

to the clinician as he sits in his chair, listening to his client. Whereas
Beck emphasized the process and style of our client's thinking, Albert
Ellis (1961) emphasized the so-called underlying premises that contrib-
ute to our client's faulty thinking, emotional disturbance, and mala-
daptive behavior. Ellis (1961) proposed that a major core of emotional
disturbances has to do with the client's preoccupation with what
others think of him and the mistaken belief that an individual's self-
worth is determined by others. Ellis encouraged the clinician to note
the themes, the irrational premises, that underlie our patients's self-
statements, images, and cognitions. This view, that psychological
problems arise from misperceptions and mistaken cognitions about
what a client perceives was most SUCcinctly summarized by the stoic
philosopher Epictetus (60 A.D.), who said, "Men are disturbed not by
things, but the views they take of them." Therefore Ellis attempted to
have clients examine the irrational ideas and beliefs, such as the
following, that give rise to misperceptions:
1. I must be loved or approved of by practically every significant
person in my life, and if I'm not it's awful.
2. I must not make errors or do poorly, and if I do it's terrible.
3. People and events should always be the way I want them to be.

In order to counteract such irrational beliefs, the rational-emotive

therapist encourages, goads, challenges, and educates by means of a
Socratic dialogue; provides information; conducts rational analyses;
assigns behavioral homework assignments; and so on, in order to
have the client entertain the notion that his maladaptive behavior and
emotional disturbance are a reflection of a commitment to irrational
As a result of such therapeutic interventions it is hoped that the
client will replace the irrational beliefs described above with the
1. It's definitely nice to have people's love and approval, but even
without it, I can still accept and enjoy myself.
2. Doing things well is satisfying, but it's human to make mistakes.
3. People are going to act the way they want to, not the way I want
them to.
The complexity of such cognitive restructuring therapy is illus-
trated in research conducted by Meichenbaum (1972, 1974c) and
Meichenbaum, Gilmore, and Fedoravicius (1971). Included in
the cognitive restructuring treatment regimen were the following
1. Didactic presentation and guided self-discovery of the role of self-
statements in subjective distress and performance inadequacies.
2. Training in the fundamentals of problem solving (e.g., problem
definition and anticipation of consequences).
3. Training in the discrimination and systematic observation of self-
4. Graduated performance assignments.
5. Structured modeling of both overt and cognitive skills in the form
of self-statements and images.
6. Modeling and encouragement of positive self-evaluation and of
coping and attentional focusing skills.
7. Depending on the treatment package employed, the use of behav-
ior-therapy procedures such as relaxation training, coping imagery
training, and behavioral rehearsal.
Thus a complex, multifaceted training procedure was employed to
change the client's irrational beliefs, self-statements, images, and
maladaptive behaviors. Although therapeutic procedures such as El-

lis's rational-emotive therapy (RET) have been available and profes-

sionally visible for well over a decade, there is a sparsity of controlled
experimental data bearing on their efficacy. A few encouraging studies
of the efficacy of RET have been offered (Baker, 1966; DiLoreto, 1971;
Karst and Trexler, 1970; Meichenbaum et al., 1971; Trexler and Karst,
1972). However, after reviewing the outcome literature for RET and
cognitive restructuring therapy in general, Mahoney (1974) con-
cluded-and my own assessment of the literature is in full accord: "the
clinical efficacy of Ellis' rational-emotive therapy has yet to be ade-
quately demonstrated" (p. 182). A similar assessment could be made
of the cognitive-therapy position of Beck.
The cognitive restructuring procedures as conducted by Beck,
Ellis, and Meichenbaum vary in several respects, most notably in
terms of the relative emphasis placed on formal logical analysis (i.e.,
isolation and evaluation of premises), the directiveness with which
the therapeutic rationale and procedures are presented, and the
adjunctive use of behavior-therapy procedures. Future research is
necessary to determine the Significance of such differences.

5. Cognitions as Instances of Problem-Solving Ability

Whereas the Ellis's approach sensitizes the therapist to listen for

the presence of maladaptive, self-defeating, anxiety-engendering cog-
nitions, the therapist with a problem-solving orientation listens for
the absence of specific, adaptive cognitive skills and responses.
Illustrative of this approach are D'Zurilla and Goldfried (1971) and
Goldfried (1975). They suggest that our client's cognitions reflect a
deficit in systematic, problem-solving skills. Treatment is designed to
have clients learn how to specify problems, generate alternative
solutions, tentatively select a solution, and then test and verify that
solution. The clinical potential of such a problem-solving approach is
illustrated in the treatment research of Spivack and Shure (1974) with
disruptive preschool children. In previous research, Spivack and his
colleagues (Shure and Spivack, 1972; Shure, Spivack, and Jaeger, 1971)
found that children exhibiting maladaptive behavior are often less
capable of employing means-ends thinking and frequently limit their
problem solutions to impulsive and aggressive methods. By means of
problem-solving training Spivack and Shure (1974) were able to train
disruptive children to consider alternatives and to engage in cause-

effect thinking. Such training seemed to reduce superfluous and

irrelevant thinking and to help children explore nonforceful possibili-
ties, resulting in significant behavioral change. These results take on
particular significance when it is noted that adolescent and adult
psychiatric patients have also been noted to manifest problem-solving
deficits, most notably the absence of foresight in considering the
possible consequences of various actions (e.g., see studies by Platt and
Spivack, 1972a,b; Spivak and Levine, 1963).
Several investigators have explored the therapeutic potential of
problem-solving training. Their applications have included (1) the use
of imagery and thematic fantasy play with culturally deprived and
behavior-problem children (Freyberg, 1973; Saltz and Johnson, 1974;
Schneider, 1974); (2) the use of problem solving in crisis clinics
(McGuire and Sifneos, 1970); (3) the use of problem solving in
assisting adolescents to handle various conflict situations (Kifer,
Lewis, Green, and Phillips, 1973) and ex-drug addicts to remain drug-
free (Copeman, 1973); and (4) the potential application of self-instruc-
tional, problem-solving skills with the aged, in the form of a "cogni-
tive prosthesis," to overcome any age-related deficits which may
appear (Meichenbaum, 1974b).
A recent, encouraging application of a self-instructional problem-
solving-training approach was offered by Hanel (1974), as cited by
Heckhausen (1974). Hanel was working with fourth-graders who were
selected for a marked fear of failure, in addition to poor academic
records. Using the self-instructional training procedure developed by
Meichenbaum and Goodman (1971), Hanel was successful in teaching
these children to talk to themselves differently, to problem-solve, in
order to change their motivational style and academic performance.
The experimenter cognitively modeled for the children how to set
standards, plan actions, calculate effort output, monitor performance,
evaluate performance outcome, weigh causal attributions, and admin-
ister self-reward. Then the students took turns in performing tasks
while emitting similar cognitions (initially aloud and then covertly).
The result of the children's adopting the modeled cognitive processes
was improved academic performance and changes on laboratory
measures such as level of aspiration, attributional style, and patterns
of self-reinforcement. The Hanel study, as well as those mentioned
earlier, indicates the therapeutic promise in viewing our client's
cognitions as instances of problem-solving ability.

Common to each of these problem-solving approaches is an

attempt to teach the client to engage in covert problem-solving
entailing symbolic stimulus-transformation, cognitive rehearsal, and
tests of alternate solutions. An illustration of the role such covert
problem-solving rehearsal plays is offered by Singer and McCraven
(1961), who, in a questionnaire study of daydreaming behavior in a
normal adult population, found that 96% of their subjects engaged
daily in some form of daydreaming. Their daydreams took the form of
fairly clear images of people and events. Daydreams dealing with
planning for future actions, and particularly interpersonnal contacts,
were in high frequency, with the largest percentage of daydreams
involving fairly practical immediate concerns. For most of the re-
spondents, daydreaming was not a matter of wish-fulfillment ideation
but rather an attempt to explore the future through positing a variety
of alternatives, not specifically involving satisfactory outcomes.

6. Cognitions as Instances of Coping Skills

Closely akin to a problem-solving approach to our client's cogni-

tions is a skills-oriented training approach. Whereas the problem-
solving approach emphasizes the client's learning to stand back and
systematically analyze a problem situation in the absence of any acute
stress, the coping-skills approach concentrates on what the client must
do when immediately confronted with an acute stress-situation. In-
deed, the problem-solving process may include rehearsal of coping
skills as clients fantasize dealing with stressful events.
The increasing clinical attention given to a skills-oriented treat-
ment approach has been noted by Mahoney (1974), who commented
that a shift in behavior treatment research is underway: from a focus
on discrete, situation-specific responses and problem-specific proce-
dures to a coping-skills model, which can be applied across situations
and problems. This model views the client's cognitions as instances of
cognitive skills that he can employ in confronting stressors.
As with the other therapeutic approaches reviewed, the support-
ing evidence for the effectiveness of mediationally based, coping-skills
training is very sparse, but initial investigations are encouraging.
Some examples of how we can teach our clients cognitive coping skills
are offered. Meichenbaum and Cameron (1973b) developed a coping-
skills-training package, which they described as "stress-inoculation"

training. The treatment regimen drew heavily on Meichenbaum's self-

instructional research, and included (1) a discussion of stress reactions
(with emphasis on labeling, attribution, and arousal-inducing self-
statements); (2) relaxation training (presented as an active, coping
skill); (3) instructed practice in the emission of coping self-statements
(cognitive self-monitoring, preparation for stress, self-reinforcement);
and (4) supervised practice in utilizing the coping skills in an actual,
stressful situation (e.g., an unpredictable shock situation). Meichen-
baum and Cameron (1973b) found that such a stress-inoculation-
training procedure was effective in modifying the fearful behavior of
mutiphobic clients. Turk (1975) has recently sucessfully applied the
stress-inoculation training to subjects who received experimentally
induced pain, and Novaco (1974) to clients with extreme anger.
Interestingly, Turk employed a number of cognitive training proce-
dures, including imagery and self-instructional rehearsal, to teach
pain tolerance. The imagery manipulations included changes and
transformations of the pain. These cognitive coping procedures were
presented in a "cafeteria" style and subjects could pick and choose
those that worked best for them.
A number of other investigations (Goldfried, 1971; Goldfried and
Trier, 1974; Langer, Janis, and Wolfer, 1973; Suinn and Richardson,
1971; Tori and Worrell, 1973) have also successfully applied a coping-
skills approach. In each case the client's cognitions playa major role in
the change process. The clinical potential of coping training was
further illustrated in the cognitive modeling research of Kazdin (1973,
1974a,b). In a series of studies, Kazdin demonstrated that covert
modeling (i.e., mental rehearsal by the subject of modeled behavior)
was effective in reducing phobic behavior and engendering assertive
behavior. Such symbolic rehearsal, especially when it included coping
self-statements and behaviors, proved to be a most effective therapeu-
tic intervention. Such cognitive rehearsal in preparation for a stressor
is similar to the cognitive process of the "work of worrying" which
Marmor (1958) and Janis (1958) have described. The cognitive-behav-
ior modification approach suggests that clients can be explicitly taught
how to worry in such a constructive fashion. The "work of worrying"
can now be translated into sets of self-statements and images, which
can be modeled by the therapist and rehearsed by the client. Sarbin
(1972) has viewed such imagery rehearsal as a form of muted role-

A research area that may prove most heuristically valuable for the
behavior modifier who is interested in such cognitive manipulations
is the work on mental practice. Richardson (1967a,b) has summarized
a considerable body of evidence that indicates that in a variety of
different physical tasks, subjects improve their performance after
spending varying amounts of time in "thinking about" or imagining
themselves in the act of performing. Increased motivation as a result of
mental practice and increased task sophistication, analogous to test
sophistication, might account for improvement. The importance of
internalizing a very clear model of what a good performance of the
task is like is indicated by the fact that the more familiar a task has
become the greater the relative gain that can be expected from mental
practice. Thus an examination of such factors as the degree of task
familiarity, accuracy of anticipated outcomes, clarity and control of
visual and kinesthetic imagery, degree of proficiency on the task,
length of time provided for imagery, and the alternation of mental and
physical practice, which have been found to be important in the
mental practice area, is likely to be of importance to the cognitive-
behavior therapist.

7. Cognitions as Instances of Defense Mechanisms

A long tradition derives from psychoanalytic theory, in which the

client's cognitions are viewed in terms of their defensive aspects. The
client's cognitions are viewed as manifest signs of underlying con-
flicts, many of which the client will be unaware of. Illustrative of this
approach is the work of Shapiro (1965), who has noted client's
neurotic styles. An evaluation of this conception of our client's
cognitions is beyond the scope of the present review.
In summary, we can view our clients' cognitions as behaviors,
automatic thoughts and thus part of the response chain, reflections of
cognitive styles and faulty belief systems, inadequate problem solving
and coping skills, or defense mechanisms. Indeed, the task for the
research clinician is to match the most useful conceptualization with
the specific client's problem and the goals of treatment (Le., an
adaptive treatment approach).
But what difference does it make how we view the client's
cognitions? "I am a behavior therapist." It is behavior that I'm after.
My approach is to have the client emit the incompatible adaptive

behavior and the cognitions will take their natural course. Haven't you
heard the old adage, "It is easier to act your way into a new way of
thinking, than to think your way into a new way of acting"? Can
behavior therapy be viewed as simply as the adage suggests? Indeed
not! It is suggested that cognitions play a substantial role in the
change process. It is proposed that there is an interactional process
between cognitions and behaviors, one preceding and following the
other. Each time we attempt to modify some aspect of the client's
behavior (e.g., teaching him to become more interpersonally assertive
or to exert self-control in areas of eating or smoking), the client is also
changing his internal dialogue or what he says to himself. The
evidence as reviewed by Mahoney (1974) and Meichenbaum (1973)
suggests that treatment efficacy is enhanced when the client's internal
dialogue is incorporated into the treatment regimen. Even in such
behavioral therapy procedures as operant conditioning, the client's
perceptions or attributions or what he says to himself about the
dispensed reinforcement influence the outcome. (For example, see
work of de Charms, 1968, Deci, 1971, and Steiner, 1970.) Bandura
(1974), in his APA presidential address, also questioned the automat-
icity of reinforcing consequences in the absence of mediating cogni-
tive processes:
So-called conditioned reactions are largely self-activated on the basis of
learned expectations rather than automatically evoked. The critical factor,
therefore, is'not that events occur together in time, but that people learn to
predict them and to summon up appropriate anticipatory reactions. (p. 2)

As Bergin (1970) has suggested, "There may be highly specific

interventions which have a behavioral or cognitive focus, but these
are always embedded in a multidimensional context or have multiple
consequences" (p. 208). Our clients present problems that require
changes in motoric, affective, and cognitive domains. The focus on
behavior or cognition would thus appear to be misguided and short-
In the same way that psychologists have been seduced into
arguing the either-or position of heredity versus environment, trait
versus situationism, we have been seduced into arguing behavioral
versus cognitive change. Our job is to find out how cognitive and
behavioral processes interact in leading to change. Perhaps it is time
to consider changing the title of the journal Behavior Therapy to
Cognitive-Behavior Therapy, with most emphasis on the,hyphen.

B. Cognitions as Final Common Pathways

In order to illustrate the role cognitions play in the change
process, let me offer a quote from Jerome Singer's (1974) recent book
on imagery and daydreaming methods in psychotherapy and behavior
modification. In describing the successful treatment of psychoanalysis,
he suggests that the following process of self-awareness and change
A patient experiences a sudden sense of unrest or annoyance upon
entering a room. Under some past conditions he might have hastily left the
room or perhaps talked rudely in response to questions raised. His analytic
experience now alerts him to the fact that this sudden unease is occasioned
by an irrational anticipation or transference in the situation. He replays in
his mind the thoughts just previous to entering the room or what he was
thinking about immediately prior to this situation. On this mental screen,
he "instant replays" the thoughts and perceptions that occurred and
suddenly is aware that he had been thinking about some obligation to one
of his parents and that on entering the room he noticed across the wayan
elderly gentleman who rather resembled his father. He now perceives that
his distress is a combination of anticipatory image plus the scene occurring
in the room and generally is freed of his anxiety and certainly is less likely
to engage in an irrational and self-defeating bit of behavior in this new
situation. (p. 64)

The Singer quote nicely illustrates two points that I would like to
make. First, in order to bring about change, the client must recognize
some behavior that he emits (e.g., a set of thoughts, images, and
physiological and motoric responses) or the interpersonal response of
someone else. This recognition is the necessary but not sufficient
condition for change. This recognition, this self-awareness, acts as the
cue, the bell ringer, the discriminative stimulus for producing a set of
incompatible thoughts and behavior. Following therapy the client no
longer responds impulsively, in a stimulus-response manner, to
externally or internally generated events. Instead, a mediational proc-
ess is elicited by stimuli and such internal processes now precede the
emission of the overt response. Insofar as stimuli or situations elicit
the same mediational processes or internal dialogue, the treatment
effects will generalize. It should be noted that generalization is
engineered into the treatment package. For now the client's own
maladaptive behavior is always the reminder to use the coping skills
that were taught in therapy.
What incompatible thoughts and behavior the client emits at this

point vary with the orientation of the therapy and the nature of the
conceptualization that has evolved between patient and therapist. The
client's internal dialogue may be in terms of pscyhoanalytic interpreta-
tions as in the Singer example, or learned response habits a la Wolpe,
or faulty belief systems a la Ellis. Indeed it is suggested that our clients
have sufficient life experiences to provide data consistent with any
one of these therapy conceptualizations, whether psychoanalytic,
Jungian, Rogerian, Gestalt, semantic, or behavioral. The human life
condition provides sufficient experience to maintain the employment
of a host of therapists of widely different persuasions. The important
point is that our clients have a need to fabricate a meaning, some
understanding, a conceptualization about what is happening to them
and what can be done to help them to change. What becomes essential
for the cognitive-behavior therapist is how to have the client adopt a
conceptualization of his problem that will lead to specific behavioral
and cognitive changes that can be transferred to real-life situations.
This leads me to the important role of the initial, conceptualization
phase in therapy.

C. Initial, Conceptualization Phase of Therapy

The role of the conceptualization process in therapy has not

received much attention by behavior-modification researchers or prac-
titioners. It is usually subsumed under such terms as nonspecific
therapy factors or included as those aspects that go "beyond" behavior
therapy (e.g., Lazarus, 1971). What goes on in therapy before desensi-
tization, or some other behavior-therapy procedure, is implemented,
is rarely discussed. Few therapy studies mention the rationale that has
been offered to the patient prior to treatment.
There are a variety of ways to have the client and the therapist
evolve a common conceptualization. Some therapists are very directive
and didactic and seem to force upon the client a particular conceptuali-
zation by power of their personalities, jargon, or positions. In some
cases such a "hard-sell" approach may prove successful. But the
therapist must be concerned about the client's self-statements and
attributions about the therapy process, as well as those concerning his
presenting problems.

An alternative way to proceed is to have the client and therapist

evolve a common conceptualization, so that the client feels that he is
an active participant and contributor. The manner in which the
therapist discusses the presenting problem, the kinds of questions he
asks, the type of assessment procedures employed, the content of the
therapy rationale, and the kinds of homework assignments given are
all used to evolve a common client-therapist conceptualization. Once
the client accepts a certain conceptualization of his problem then he
readily engages in the treatment assignments. As Jerome Frank (1961)
pointed out, the shared conceptual system between therapist and
client is important in the change process. Insofar as the client adopts a
particular paradigm, or comes to view his behavior from a given
perspective, thus far are the client's "assumptive world" (to use
Frank's term) and behavior open to change. Such a conceptualization
evolves over the course of treatment as the therapist cognitively
models particular beliefs and encourages the client to engage in an
active self-examination. The role that conceptualization plays in the
change process is illustrated in both our laboratory research and our
clinical work.
In our laboratory research most emphasis has been placed on this
initial, conceptualization phase. In treating phobics or interpersonally
anxious clients we have conceptualized their presenting symptoms of
arousal (e.g., muscular tension, pounding hearts, sweaty palms, and
heavy breathing) and accompanying task-irrelevant, anxiety-engen-
dering thoughts in terms of Schachter and Singer's (1962) theory of
emotion. Thus treatment could be directed naturally toward (1) help-
ing the client control his physiological arousal by means of relaxation
and (2) substituting positive, coping self-statements for the anxiety-
engendering self-statements that habitually occupy the client's mind
under stressful conditions. In the treatment of pain patients we
conceptualized the subjects' pain in terms of Melzack and Wall's (1962)
gate-control theory of pain (Meichenbaum, Turk, and Burstein, 1975).
It should be noted that the scientific validity of a given concep-
tualization is less important than the aura of its plausibility. The aim
of the therapist is not primarily to impart precise, scientific informa-
tion, but rather to provide the client with a conceptualization that will
facilitate therapy of making its rationale comprehensible. Similar
examples could be offered from our work with low-creativity subjects,
test-anxiety subjects, and others (see Meichenbaum, 1975b). The gen-
eral treatment strategy is to share with the client, in terms that he can

clearly understand, the rationale that led to the present treatment

More specifically, the goal of the conceptualization phase is to
have the client talk to himself differently about his presenting prob-
lem. An attempt is made to have the client change his perceptions,
attributions, sense of control, and sense of helplessness about his
presenting problem-in short, to alter the client's internal dialogue
regarding his appraisal of his maladaptive behaviors and emotional
disturbance. For example, in treating multiphobic patients the thera-
pist helped the client change his perception of how he behaved in a
fearful situation. Instead of viewing his response as a massive panic
reaction, the therapist suggested that the phobic client's response
seemed to include several stages, namely, preparing for, then con-
fronting or handling the stressor, possibly being overwhelmed, and
finally, reinforcing himself for having coped. In this way the phobic
patient no longer had a "massive phobic reactions," but rather a series
of responses that went through four stages, for each of which the
client could be trained to employ adaptive coping responses. For once
the client views his problem from a given perspective, then the
acquisition of a number of specific skills makes sense and they are
actively rehearsed.
One could offer a number of clinical cases to illustrate how the
initial phase of therapy is designed to have the client talk to himself
differently about his presenting problem. A translation process occurs.
Initially the client describes what's bothering him, often with a sense
that he is losing control and feeling helpless and hopeless. The
therapist, with skill, has the client come to view his problem from a
different perspective, to fabricate a new meaning or explanation for
the etiology and maintenance of the client's maladaptive behavior.
Whereas prior to therapy the client may view his compulsion to wash
as a sign of his "losing his wits," being depressed, etc., as a result of
therapy he may come to view his washing as a communication
problem, or as a manifestation of deep-seated conflict about guilt, or
as a behavioral repertoire maintained because of the secondary gains
(reinforcers) that accrue, etc. Many other conceptualizations could be
offered. Indeed the patient may provide enough data to support each
of these conceptualizations. Most patients do! The exact conceptualiza-
tion adopted in therapy will vary with the therapist's orientation, the
patient's expectations and the goals of therapy.
From the present perspective, the important theoretical problem is

"Why does altering the client's conceptualization-how he views his

problem and what he says to himself about it-result in behavior
It should be apparent that the clinician has a host of alternative
ways to view his client's cognitions. At present we have little
empirical evidence (Mahoney, 1974) to guide us in determining which
conceptualization will prove to be most efficacious. In fact, we may
come to learn, by means of systematic investigation, which concep-
tualization works best with which clients.
As Dember (1974) has stated, "Psychology has gone cognitive"
(p. 161), and it is time for behavior modification to do likewise. In order
to engender the shift to cognition, the following list of conclusions and
implications is offered.
1. First, we can no longer compare the effectiveness of specific
behavior-therapy procedures such as desensitization with an "in-
sight" or "semantic" therapy. The uniformity myth with respect to
treatment procedures that Kiesler (1966) described can no longer be
applied to semantic or cognitive therapies. Instead we must encourage
comparisons between different cognitive approaches in order to iden-
tify the parameters that underlie cognitive restructuring. What are the
relative therapeutic merits of viewing our clients' cognitions from such
perspectives as those of Ellis, Beck, Meichenbaum, D'Zurilla, and
2. When reading a therapy study one must carefully attend to the
details of the therapist's manual, especially those phases in which an
initial conceptualization is offered. The conceptualization phase must
be seen as an active ingredient of the therapy process and not
something beyond the researchable interests of behavior therapists.
3. As clinicians we must become more sensitive to the thoughts
and feelings our clients have in the criterion situations. What would
you like your client to say to himself in order to cope more ade-
quately? Can we not teach our clients to use their own maladaptive
behaviors as cues for using coping skills? Indeed we can use the
technology of behavior therapy to influence our clients' internal
dialogues (see Meichenbaum and Cameron, 1974).
4. Finally, behavior therapists may wish to consider how they
would alter such therapy techniques as operant-conditioning pro-
grams with parents, self-control training with obese patients, desensi-

tization with phobics, aversive conditioning with pedophiles, asser-

tive training with college students, biofeedback with headache
patients, and so on, in order to include self-instructional and imagery
processes. It is suggested that as we become more cognitive in our
orientation, we will become more effective in our practice.
The need for a cognitive-hyphen-behavior therapy approach is
now! In order to explain further the nature of the hyphen, a cognitive
theory of self-control is offered.


As already mentioned, the therapist has a variety of ways to view

his client's cognitions. Each conceptualization leads to different thera-
peutic interventions. However, it has been suggested that even
though the therapy procedures differ in emphasis and techniques,
clients go through a similar cognitive process in achieving behavioral
change. The research conclusions that have been outlined can now be
integrated into a more general and coherent theory of a three-stage
process accounting for therapeutic change.

A. A Three-Stage Process

1. Stage 1: Self-Observation

The first step in the change process is the client's becoming an

observer of his own behavior. Through heightened awareness and
deliberate attention, the client monitors, with increased sensitivity,
his thoughts, feelings, and/or interpersonal behaviors. In the second
stage of the change process, this self-observation of the client's
inappropriate actions will, upon the occurrence of a maladaptive
behavioral event, serve as a signal, or cue, to produce thoughts and
behaviors incompatible with the continuation of the inappropriate
cognitions and behaviors. The process of self-observation is a neces-
sary but not a sufficient condition for change.
On which behaviors the client focuses depends upon the concep-
tualization process that evolves during therapy. The important role of

this process needs once again to be underscored. The client's motiva-

tion and skill in acting as a self-observer are fostered as a result of the
conceptualization process. During the course of therapy a translation
process occurs, through which the client's definition, description, and
fabricated meaning concerning his problems are subjected to examina-
tion and change. The exact terms of the translation vary with each
therapist's orientation and each client's characteristics (e.g., client
expectations and background). The important point is that the client
and the therapist go through a series of steps, such as interviewing,
testing, and homework assignments, whereby the client comes to
entertain a different view of his maladaptive behaviors and emotional
disturbances. The client may come to view his problems from a
psychoanalytic, a social learning, or a semantic viewpoint. Whereas
the client enters therapy with one description or language system,
with particular referent terms and explanatory concepts, as a result of
therapy he comes to view his complaints and dissatisfactions in
different terms.
The therapist uses a host of clinical tools, such as reflections,
explanation, interpretation, information giving, and cognitive model-
ing, to achieve this translation process. The goal of each of these
therapeutic techniques is to have the client view his behavior differ-
ently. The end result is that the client becomes an observer of his own
One of the by-products of the translation process and the in-
creased self-awareness is that the client gains a sense of control of his
emotional state and behavior; he feels that he is an active contributor
to his own experience and not a helpless victim of his thoughts and
feelings and the reactions of others. A sense of hopefulness and
"faith" are aroused. As Strupp (1970) pointed out, a major component
of effective psychotherapy is the client's experience of having in-
creased his control over his own emotions and overt behavior. This
translation process, in the form of a new conceptualization, results in a
cognitive restructuring or new internal dialogue by the client.
Terms such as sense of control, hope, faith, expectancy, and cogni-
tive restructuring have been offered by many theorists to explain the
therapy process. For purposes of the present explication, each of these
concepts is viewed as part of the client's internal dialogue. Our clients
think; they engage in an emission of thoughts, images, and self-
statements. We, as psychologists, are interested in finding out the

most theoretically useful and heuristically productive conceptual sys-

tem to explain our clients' internal dialogues. The goal is a kind of
ethological description of our clients' thinking processes. At present,
the "marketplace" is filled with competing concepts and schemes to
explain our clients' thoughts and images. In the same way that we try
to fashion an explanation or fabricate a meaning for our clients'
thinking processes, our clients attempt to develop an understanding of
their own thoughts, feelings, and behaviors.
It is postulated that both therapist and the client-and man in
general-have a need to believe, to understand, to impose an explana-
tion on events. Evidence that such a need exists may be offered from
such diverse areas as the psychology of superstition (Jahoda, 1969), the
universal role of religious belief (Allport, 1950), and the wide appeal of
scientific investigation, and it is perhaps most interestingly docu-
mented by Pitkin's book A Short Introduction to the History of Human
Stupidity (1935). The translation process in therapy serves to create a
conceptual framework, which provides a basis for the client to monitor
his cognitions and behavioral productions effectively in terms that will
then serve as springboards for therapeutic change.

2. Stage 2: Incompatible Thoughts and Behaviors

Once the client has become an observer of his behavior and these
self-observations have been reinforced by, and in tum, reinforce, the
conceptualization process, the second stage in the change process
occurs. The process of self-observation becomes the occasion or
stimulus for the client to emit different cognitions and behaviors. This
point was illustrated before, with the quote from Singer's book. The
content of what the client now says to himself will vary with the
conceptualization that emerged in therapy. If the client's behavior is to
change, then what he now says to himself, and/or imagines, must
initiate a new behavioral chain, one which is incompatible with his
maladaptive behaviors.

3. Stage 3: Cognitions Concerning Change

The third step in the change process, what the client says to
himself about his newly acquired behaviors, determines whether the
behavioral change will be maintained and will generalize. As the

client attempts to behave differently, he will often elicit different

reactions from significant others. What the client says to himself and
imagines about these reactions and his own behavior change will
influence the stability and generalizability of the treatment. As stated
before, insofar as we as therapists can anticipate and incorporate the
client's internal dialogue into the therapy process, we will increase our
Once again, the situation is that our client is emitting a set of
thoughts and images, and the question is how best to describe such
mediational events. Social psychologists are prone to characterize such
internal dialogue in terms of appraisals, attributions, perceived free-
doms, psychological reactance, etc. Some therapists may conduct a
hierarchical analysis of the client's internal dialogue in terms of self-
statements, attitudes, beliefs, values, etc. Others may engage in a
cluster analysis of the client's cognitions in terms of self-concept. Our
own preference is to engage in a situational analysis of the specific
self-statements and images the client emits, noting their similarities
and differences. It is hypothesized that consistency of behavior across
situations or treatment generalization is a function of the degree to
which the individual emits a set of similar self-statements andior
images across situations. Parenthetically it may be noted that what is
being offered is a concept-formation view of personality. Insofar as the
same mediators (i.e., appraisals, attributions, self-statements, and
images) are elicited across situations, one will observe behavioral
similarities. Once description of the mediating events has been
recently offered by Mischel (1974). From this viewpoint, it is interest-
ing to observe that rarely, if ever, do we ask our clients, or for that
matter, even less frequently our experimental subjects, how they
would characterize their own thoughts and feelings. What are the
summary terms and concepts that our clients employ to describe their
own self-statements and images.

4. Summary

In summary, a three-stage theory of behavioral change is offered.

The client must first become an observer of his thoughts, feelings, and
behaviors by means of heightened awareness. This process is facili-
tated by means of a conceptualization or translation process that

evolves over the course of therapy. The process of self-observation lays

the foundation for the client to emit incompatible thoughts and
behaviors, which constitute the second stage of change. The third
stage, which determines the persistence and generalization of treat-
ment effects, involves the nature and content of the client's internal
dialogue and images about the behavior change. Although these three
stages may be viewed as occurring sequentially, they often overlap in a
continual process of change.
The reader may be concerned that one of the things that already
characterizes some clients prior to therapy is a heightened awareness,
a self-preoccupation, and general egocentrism, e.g., in the case of the
obsessive. Such a description of clients may be accurate, and indeed,
the therapist would incorporate such behaviors into the conceptualiza-
tion process. But what the client attends to and says to himself about
his behavior prior to therapy are qualitatively and quantitatively
different than what he will observe and how he will appraise his
behavior following therapy. Prior to therapy the obsessive's self-
perception is likely to be delimited and repetitive, eliciting a sense of
helplessness and despair. As a result of therapy, the client will come to
view his obsessive ideation differently, with the consequence of an
increased sense of resourcefulness, of control of his own behavior.
Whether he views his obsessions as manifest symptoms of underlying
conflicts, or instrumental acts to control anxiety, or interpersonal ploys
to influence others, etc., will depend on his therapist's orientation.
Behavior is sufficiently multi determined that it is likely that there will
be aspects of each of the above explanations contributing to the
obsessional style. Moreover, both client and therapist have the capa-
bility of entertaining anyone of these conceptualizations that can be
offered to explain the client's obsessive style. The important ingredi-
ent is that the translation process provides the stimulus for the other
stages of change to occur.
The theory thus far provides an overview of the change process.
The therapeutic picture that is offered is that one can treat a client's
verbal utterances that occur between the thought and the act and by
doing so reach backward to change the thoughts and reach forward to
modify the behavior. Through monitoring and modifying his thinking
(i.e., self-statements and images), a client can effectively change his

B. How Does Behavior Change through Internal Dialogue?

But how exactly does changing the client's internal dialogue lead
to behavior change? The answer to this question is surely complex.
Indeed, there are likely to be a number of answers, given the
heterogeneity of private speech. In part, if will depend upon which
popUlation and which behaviors one is trying to change. The psychol-
ogical mechanisms involved in altering the internal dialogue of hyper-
active children, versus adult psychotics, versus adult neurotics, may
be quite different. We may find value in developing a number of
minitheories of verbal control of behavior.

1. From Related Investigations

One place to begin is exploring interpersonal instructions in order
to ascertain if they bear any applicability to intrapersonal self-instruc-
tions. Several investigators (e.g., Gagne, 1964; Marlatt, 1972; Simkins,
1963; Sutcliffe, 1972) have speculated about the role of interpersonal
instructions in controlling behavior. They emphasized both the insti-
gational and directive functions of such instructions in controlling
behavior: instructions both initiate or facilitate performance in a
general sense and direct attention to stimulus conditions and to
specified performance requirements in a task. Moreover, instructions
control extraneous behaviors by directing subjects not to engage in
certain responses.
Gagne (1964), working within a problem-solving framework, has
viewed instructions as serving the following functions: (1) motivating
the subject by eliciting an achievement set; (2) helping him identify
the criterion performance and the salient parts of the stimulus situa-
tion; (3) aiding recall of relevant subordinate performance capabilities
necessary to the task; and (4) channeling thinking in terms of task-
relevant hypotheses and controlling extraneous thoughts and behav-
iors. In this way, instructions provide the subject with a rule of
principle by which he can mediate his behavior.
In describing the role of overt self-verbalizations, or self-instruc-
tions, in a problem-solving task, McKinney (1973) offered the follow-
ing list of functions: the overt self-instructions (1) increase the distinc-
tiveness of the stimulus attributes; (2) direct the subject's attention to

the relevant dimensions; (3) assist the subject in formulating a series

of hypotheses; and (4) maintain information in short-term memory.
The similarity between the Gagne and McKinney lists of psychol-
ogical functions for inter- and intrapersonal instructions is notewor-
thy. This leads to the first, rather obvious, hypothesis, that self-
instructions operate in a similar fashion to interpersonal instructions.
As noted earlier, Vygotsky (1962) and Luria (1961) have theorized that
developmentally the child comes to exercise verbal control of his
behavior by incorporating adults' instructions. To quote Vygotsky (as
cited by Zaporozhets and Elkonin, 1971):
Apparently, egocentric speech, besides having a purely expressive
function and a function of discharge, besides merely accompanying the
child's activity, very readily becomes a means of thinking in its own sense,
i.e., it begins to fulfill the function of formulating a plan for the solution of
a problem emerging in the course of behavior. (p. 124)

Thus it is posited that one aspect of the functioning of self-

instructions depends upon their being analogous to interpersonal
instructions and serving many of the same purposes.
From a different vantage point, Janis (1968) discussed the role of
appraisal, or what we would call internal dialogue (i.e., self-statements
and images), in the handling of stress. Janis suggested that such
cognitive appraisal includes the subject's (1) making plans for coping
with a number of different contingencies; (2) attempting to reassure
himself; (3) warding off disturbing thoughts, and (4) noting the level
of his shortcomings that becomes a cue for actions. Janis suggested
that such appraisal responses also raise the client's concern for others
and help him identify with a group. In some general sense the "work
of worrying," or emitting self-statements, increases the likelihood that
the subject will be an observer of his situation and of his ongoing
behavior. By thus self-instructing, the client becomes less egocentric
and develops the ability to take the point of view of others or, in
Piagetian terms, the ability to decenter perceptions (Looft, 1972). Thus
by training a client to talk to himself, one reduces the likelihood that
he will see events from only his own perspective.
Common to both Gagne's and Janis's formulations is the role of
self-instructions in directing the subject'S attention to either specific
aspects of the task (i.e., in the problem-solving situation) or to the
viewpoints of others (i.e., in the stress situation). Another postulate of

the present theory is, then, that changing the client's internal dialogue
effects behavior change specifically by causing differential attentional
The literature on the effect of instructional sets of autonomic
functioning indicates that changing the client's style of self-instruc-
tions can have directive physiological effects (Barber, 1965; May and
Johnson, 1973; Platonov, 1959; Schwartz, 1971; Sternbach, 1964; Zim-
bardo, 1969). Cognitive activity has been suggested as a mediational
factor (Le., facilitator or inhibitor) in operant, autonomic condition-
ing. Katkin and Murray (1968) proposed that an internal source of
stimulation, rather than the external, experimenter-controlled reinfor-
cers, may be controlling the autonomic responses. The subject may be
involved in arousing or inhibiting subvocal activity (thinking), which
produces a previously conditioned autonomic response. An illustra-
tion of the role of cognitive set is the work on emotion by Schachter
(1966), who provided evidence for the important role that the client's
restructuring of a situation plays in mediating behavior. In our own
research the clients, following cognitive-behavior modification treat-
ment, came to label their physiological arousal as facilitative rather
than debilitative (Meichenbaum, 1972; Wine, 1970). Sweaty palms,
increased heart and respiratory rates, and muscular tension now
became allies, bell ringers, cues to use the coping techniques for
which they had been trained. The physiological arousal that the
client had previously labeled as totally debilitating anxiety and fear,
the harbinger of further behavior deterioration leading to feelings of
helplessness, was now relabeled as eagerness to demonstrate his
competence, as a desire to get on with a task, and as a sign to cope.
The result was a change to a sense of "learned resourcefulness,"
replacing a sense of "learned helplessness." In other words, the client
learns to respond to the same physiological cues when they do arise
with different cognitions: originally he entertained cognitions that
mediated further autonomic arousal (e.g., "I'm really nervous; I'm
sweating; others will see it; I can't handle this"); after treatment, his
cognitions have a coping orientation and move the focus away from
his arousal toward response alternatives. This shift in cognitions in
itself may mediate a shift in autonomic functioning.
The present theory postulates that it is not the physiological
arousal per se that is debilitating, but rather what the client says to
himself about that arousal that determines his eventual reactions. The

distinction between thought and feeling becomes obscure, for each

thought has an affective component and each feeling has a cognitive
component. Thus the arbitrary distinction between "feeling" and
"cognitive" therapies is misguided. "Touchy-feely" therapy, to use
the cliche, has as much cognitive restructuring occurring-that is, if it
leads to behavioral change--as semantic therapies have affective
Our original question was "How does changing the client's
internal dialogue lead to behavior change?" In order to approach our
question we have briefly covered such diverse research areas as
problem solving, appraisal or stress, and autonomic conditioning.
From these vantage points we have been able to discern several of the
factors contributing to the effect that cognitive restructuring has on a
client's behavior: self-instructions playa direct role, analogous to that
served by interpersonal instructions; self-instructions and images
affect behavior through influencing attentional direction; and they
influence a client's interpretation and experience of physiological
Perhaps another useful approach that should not be overlooked
is to be found in an examination of the conditions under which a
client's self-instructions do not enhance self-control nor lead to be-
havioral change and autonomic conditioning. (Consider New Year's
resolutions. )

2. When Self-Instructions Fail

"Tous les jours, a tous points de vue, je vais de mieux en mieux."

In English, "Day by day, in every way, I'm getting better and better."
This is what Emil Coue, the Fren<1h psychiatrist, enjoined his patient
to say to themselves in order to improve themselves (Coue, 1922). The
banner of Coue's approach of fostering the power of positive thinking
has been take~ up by a number of individuals writing for the general
public, including Norman Vincent Peale (1960), Dale Carnegie (1948),
and W. Clement Stone (1962). The use of a formula such as Coue's to
be repeated daily, or in times of need, has found widespread appeal in
the lay literature.
It has been argued that such a "general formula leaves every mind
free to unfold and develop in the manner most natural to itself"
(Brook, 1922). However encouraging, the use of a formula, or "psy-

chological litany," tends to lead to rote repetition and emotionless

patter that has been found ineffective as a coping tool (Meichenbaum
and Cameron, 1973b). They are, in effect, self-instructions that fail.
Several reasons for this failure have been logically and experimentally
derived. In the first place, the client's self-instructions may be too
general or too broad, not specific enough nor sufficiently individual-
ized. For instance, the client may resolve that he will give up smoking
but not specify when and how or what incompatible responses he will
substitute in its place. Just why such a resolution would be likely to
fail can be seen from the analogous situation of a person attempting to
become a vegetarian without specifically substituting anything for the
meat in his diet. Moreover, the client's general self-instructions are
likely to be insensitive to situational conditions and are unlikely to
incorporate such contingencies or rewards and punishments as would
strengthen appropriate, and weaken inappropriate, responses to his
Instead of the client's making only a general self-statement (a la
Coue), it seems necessary for him to use a more detailed set of self-
instructions, specifying the specific desirable responses he will emit,
under what conditions he will emit them, and the set of reinforcement
contingencies he will employ. A general, self-motivating statement
may enhance the functioning of such a detailed plan, but the detail
itself is likely to lead to more self-control than generalized prohibi-
tions or exhortations, repeated mantra-like.
Two other, related examples illustrate conditions under which
self-instructional rehearsal may not work. First, in our research on
reducing fears by modifying what clients say to themselves (Meichen-
baum and Cameron, 1973a), we repeatedly compared a treatment group
who received only self-instructional rehearsal with a self-instructional
rehearsal group who also received application training in the form of
exposure to electric shock. The results indicated that self-instructional
rehearsal was a necessary but not a sufficient condition for the
elimination of fears. Having clients merely cognitively rehearse the
self-instructions, saying that they could overcome their fears, did not
lead to consistent significant behavioral and affective change. In fact,
when the phobic clients who had received only self-instructional
rehearsal confronted the phobic object following treatment, they
initially reported minimal anxiety, indicating that they were calm and
in control. However, when the demands to handle the phobic objects

increased, their self-reports of anxiety precipitously rose with the

consequence of a rearousal of their fears. The initial bravado which
followed from mere self-instructional training gave way when the task
demands increased. In comparison, those phobic clients who re-
hearsed the self-controlling self-statements and who then had an
opportunity to use them in confronting a stress (e.g., electric shock)
significantly reduced their fears following treatment.
Thus the mere rehearsal of self-instructions without the opportu-
nity for application training on tasks other than the criterion tasks will
be likely to result in those self-instructions' exerting a minimal self-
controlling influence. Saying the "right" things to yourself may not be
a sufficient condition for change. One may have to "try out" these
self-statements gradually in real situations that are similar to the
criterion task.
In addition to our earlier excursion into the literature from related
investigations, this brief review of some self-instructions that fail had
indicated another field of investigation that may be productively
explored in an attempt to understand better the role of alterations of
internal dialogue in effecting behavioral change. Some of the complex-
ity involved in an understanding of the psychological mechanisms
underlying behavioral change is further illustrated when we learn that
there are approximately 15 different ways to alter our clients' imagery
in therapy (Le., see Beck, 1970b; Singer, 1974). We can proceed by
systematically analyzing each of these procedures, or rather proceed
by searching for a set of common principles that underlie these
different approaches. The present cognitive theory of self-control was
designed to help us pursue the latter course.


The alternative ways the therapist may view his clients' cogni-
tions were reviewed. These included cognitions as behaviors, auto-
matic thoughts, and thus part of the response chain, reflections of
cognitive styles and faulty belief systems, inadequate problem-solving
and coping skills, and defense mechanisms. Each conceptualization
leads to different therapeutic interventions. However, it was sug-
gested that even though the therapy procedures differ in emphasis
and techniques, clients go through a similar cognitive process in

achieving behavior change. A cognitive theory of self-control, which

includes a three-stage process, was offered to explain the behavior
change process. Briefly, the theory proposed that the client must first
become an observer of his thoughts, feelings, and behaviors by means
of heightened awareness. This process is facilitated by means of a
conceptualization process, or translation of how the client views his
problem, which occurs over the course of therapy. The language
system of the conceptualization process will be influenced by the
therapist's orientation. The process of self-observation acts as a
stimulus for the client to emit incompatible thoughts and behaviors,
constituting the second stage of the change process. The third stage,
which determines the persistence and generalization of treatment
effects, involves the nature and content of the client's internal dialogue
and images about behavior change.
The chapter also addressed the question, "How does changing the
client's internal dialogue lead to behavior change?" Elements of the
answer came from the literature on problem solving, appraisal and
stress, and antonomic conditioning. It was theorized that (1) self-
instructions playa direct role in changing behavior, analogous to that
served by interpersonal instructions; (2) self-instructions and images
affect behavior through influencing attentional direction; and (3) they
also influence a client's interpretation and experience of his physiolog-
ical state. An examination of the conditions under which self-instruc-
tions fail further elucidated the issues.
A cognitive-behavior modification treatment approach was pro-
posed with most of the emphasis on the psychological mechanisms
that underlie the hyphen.


Portions of this paper were presented at the eighth annual

meeting of the Association for the Advancement of Behavior Therapy,
Chicago, Illinois, November, 1974. The author is grateful to Myles
Genest and Roy Cameron for their helpful editorial comments.
The research reported in the paper was supported by grants from
the Canada Council (#S73-0024-V1) and the Ontario Ministry of

ALLPORT, G. The individual and his religion. New York: MacMillan Company, 1950.
ASHEM, B., AND DONNER, L. Covert sensitization with alcoholics: A controlled replica-
tion. Behaviour Research and Therapy, 1968,6, 7-12.
BAKER, J. Reason versus reinforcement in behavior modification. Unpublished doctoral
dissertation, University of Illinois, 1966.
BANDURA, A. Principles of behavior modification. New York: Holt, Rinehart, & Winston,
BANDURA, A. Behavior theory and models of man. Paper presented at APA meeting in
New Orleans, 1974.
BARBER, T. Physiological effects of "hypnotic suggestions": A critical review of recent
literature (1960-1964). Psychological Bulletin, 1965,63, 201-222.
BECK, A. Cognitive therapy: Nature and relation to behavior therapy. Behavior Therapy,
1970a, 1, 184-200.
BECK, A. Role of fantasies in psychotherapy and psychopathology. Journal of Nervous and
Mental Disease, 1970b, 150, 3-17.
BECK, A. Cognition, affect and psychopathology. Archives of General Psychiatry, 1971,24,
BEM, S. Verbal self-control: The establishment of effective self-instruction. Journal of
Experimental Psychology, 1967,74, 485-491.
BERGIN, A. A self-regulation technique for impulse control disorders. Psychotherapy:
Theory, research and practice, 1967,6, 113-118.
BERGIN, A. Cognitive therapy and behavior therapy: Foci for a multidimensional
approach to treatment. Behavior Therapy, 1970,1, 205-212.
BLACKWOOD, R. The operant conditioning of verbally mediated self-control in the
classroom. Journal of School Psychology, 1970,8, 251-258.
BRAFF, D., AND BECK, A. Thinking disorder in depression. Archives of General Psychia-
try, 1974,31, 45Cr462.
BROOKS, C. H. The practice of autosuggestion by the method of Emile Coue. In E. COUE
AND C. H. BROOKS (Eds.), Better and better every day. London: Unwin, 1960.
CARNEGIE, D. How to stop worrying and start living. New York: Simon & Schuster, 1948.
COPEMAN, C. Aversive counterconditioning and social retraining: A learning theory
approach to drug rehabilitation. Unpublished doctoral dissertation, S.U.N.Y. at
Stony Brook, 1973.
Com" E. The practice of autosuggestion. New York: Doubleday, 1922.
DE CHARMS, R. Personal causation: The internal affective determinants of behavior. New
York: Academic Press, 1968.
DECI, E. The effects of externally mediated rewards on intrinsic motivation. Journal of
Personality and Social Psychology, 1971,18, 105-115.
DEMBER, W. Motivation and the cognitive revolution. American Psychologist, 1974, 29,
DENNY, D. Modeling effects upon conceptual style and cognitive tempo. Child Develop-
ment, 1972,43, 105-419.
DILORETO, A. Comparative psychotherapy: An experimental analysis. Chicago: Aldine-
Atherton, 1971.
DOLLARD, J., AND MILLER, N. Personality and Psychotherapy. New York: McGraw-Hill,
D'ZURILLA, T., AND GOLDFRlED, M., Problem solving and behavior modification. Journal
of Abnormal Psychology, 1971,78, 107-126.

ELLIS, A. Reason and emotion in psychotherapy. New York: Lyle Stuart Press, 1961.
FRANK, J. Persuasion and healing. Baltimore: Johns Hopkins Press, 1961.
FREYBERG, J. Increasing the imaginative play of urban disadvantaged kindergarten
children through systematic training. In J. SINGER (Ed.), The child's world of make
believe. New York: Academic Press, 1973.
GAGNE, R. Problem solving. In A. MELTON (Ed.), Categories of human learning. New
York: Academic Press, 1964.
GOLDFRIED, M. Systematic desensitization as training in self-contro!. Journal of Consult-
ing and Clinical Psychology, 1971,37,228-234.
GOLDFRIED, M., DECENTECEO, E., AND WEINBERG, L. Systematic rational restructuring as
a self-control technique. Behavior Therapy, 1974,5, 247-254.
GOLDFRIED, M., AND GOLDFRIED, A. Cognitive change methods. In F. KANFER AND A.
GOLDSTEIN (Eds.), Helping people change. New York: Pergamon Press, 1975.
GOLDFRIED, M., AND TRIER, C. Effectiveness of relaxation as an active coping skill.
Journal of Abnormal Psychology, 1974,83,348-355.
HANEL, J. Der Einfluss eines Motivanderungsprogramms auf schulleistungschwach-
miszerfolgsmotivierte Grundschuler der 4. Klasse. Psycho!. Institue der Ruhr-Univ-
ersitat: Unveroffentlichte Diplomarbeit, 1974.
HARTIG, M., AND KANFER, F. The role of verbal self-instructions in children's resistance
to temptation. Journal of Personality and Social Psychology, 1973,25, 259-267.
HECKHAUSEN, H. How to improve poor motivation in students. Paper presented at the
81th International Congress of Applied Psychology, Montreal, August, 1974.
HOMME, L. Perspectives in psychology: Control of coverants, the operants of the mind.
Psychological Record, 1965,15, 501-511.
JAHODA, G. The psychology of superstition. Baltimore: Penguin, 1969.
JANIS, I. Psychological stress. New York: Wiley, 1958.
JANIS, I. Human reactions of stress. In E. BORGOTTA AND W. LAMBERT (Eds.), Handbook of
personality theory and research. New York: Rand McNally, 1968.
KANFER, F., AND SASLOW, G. Behavioral diagnosis. In C. FRANKS (Ed.), Behavior therapy:
Appraisal and status. New York: McGraw-Hill, 1969.
KARST, T., AND TREXLER, L. Initial study using fixed-role and rational-emotive therapy
in treating public-speaking anxiety. Journal of Consulting and Clinical Psychology,
1970,34, 360-366.
KATKIN, E., AND MURRAY, E. Instrumental conditioning of autonomically mediated
behavior: Theoretical and methodological issues. Psychological Bulletin, 1968,70, 52-
KAZDIN, A. Covert modeling and the reduction of avoidance behavior. Journal of
Abnormal Psychology, 1973,81,87-95:
KAZDIN, A. Covert modeling, model similarity, and reduction of avoidance behavior.
Behavior Therapy, 1974a, 5, 325-340.
KAZDIN, A. The effect of model identity and fear-relevant similarity on covert modeling.
Behavior Therapy, 1974b,5, 624-636.
KIESLAR, D. Some myths of psychotherapy research and the search for a paradigm.
Psychological Record, 1966, 65, 110-136.
KIFER, R., LEWIS, M., GREEN, D., AND PHILLIPS, E., The S.O.C.S. model: Training pre-
delinquent youths and their parents in negotiation responses to conflict situations.
Paper presented at the American Psychological Association meeting, Montreal,
August, 1973.

KIMBLE, G., AND PERLMUTIER, L. The problem of volition. Psychological Review, 1970,
77, 361-384.
LANGER, E., JANIS, 1., AND WOLFER, J. Effects of cognitive devise and preparatory
information on psychological stress in surgical patients. Unpublished manuscript,
Yale University, 1973.
LAZARUS, A. Behavior therapy and beyond. New York: McGraw-Hill, 1972. The psychology
of preschool children. Cambridge: MIT Press, 1971.
LOOF!", W. Egocentrism and social interaction across the life span. Psychological Bulletin,
1972, 78, 7~9i
LURIA, A. The role of speech in the regulation of normal and abnormal behavior. New York:
Liveright, 1961.
MAHONEY, M. Cognition and behavior modification. Cambridge, Mass.: Ballinger Publish-
ing Co., 1974.
MARKS, 1., BOULOUGOURIS, J., AND MARSET, P. Flooding versus desensitization in the
treatment of phobic patients. British Journal of Psychiatry, 1971, 119, 35~375.
MARLATT, G. Task structure and the experimental modification of verbal behavior.
Psychological Bulletin, 1972, 78, 335--350.
MARMOR, J. The psychodynamics of realistic worry. Psychoanalysis and Social Science,
1958,5, 115--163.
MARSHALL, W., POLGRIN, D., AND BOUTILIER, J. Reinforcement contingencies in covert
reinforcement. Unpublished manuscript, Kingston Psychiatric Hospital, 1974.
MAY, J., AND JOHNSON, H. Physiological activity to internally elicited arousal and
inhibitory thoughts. Journal of Abnormal Psychology, 1973,82, 239-245.
MCGUIRE, M., AND SIFNEOS, P. Problem solving in psychotherapy. Psychiatric Quarterly,
1970,44, 667-673.
McKINNEY, J. A developmental study of the effects of hypothesis verbalizations and
memory load on concept attainment. Unpublished manuscript, University of North
Carolina, Chapel Hill, 1973.
MEICHENBAUM, D. Examination of model characteristics in reducing avoidance behav-
ior. Journal of Personality and Social Psychology, 1971,17,298-307.
MEICHENBAUM, D. Cognitive modification of test anxious college students. Journal of
Consulting and Clinical Psychology, 1972, 39, 370-380.
MEICHENBAUM, D. Cognitive factors in behavior modification: Modifying what clients
say to themselves. In C. FRANKS AND T. WILSON (Eds.), Annual review of behavior
therapy: Theory and practice. New York: Bruner/Maze!, 1973.
MEICHENBAUM, D. Cognitive behavior modification. Morristown, N. J.: General Learning
Press, 1974a.
MEICHENBAUM, D. Self-instructional training: A cognitive prosthesis for the aged.
Human Development, 1974b, 17, 27~280.
MEICHENBAUM, D. Therapist manual for cognitive behavior modification. Unpublished
manuscript, University of Waterloo, 1974c.
MEICHENBAUM, D. A self-instructional approach to stress management: A proposal for
stress inoculation training. In C. SPEILBERGER AND 1. SARASON (Eds.), Stress and
anxiety in modern life. New York: Winston and Sons, 1975a.
MEICHENBAUM, D. Self-instructional methods. In F. KANFER AND A. GOLDSTEIN (Eds.),
Helping people change. New York: Pergamon Press, 1975b.
MEICHENBAUM, D., AND CAMERON, R. An examination of cognitive and contingency
variables in anxiety relief procedures. Unpublished manuscript, University of
Waterloo, 1973a.

MEICHENBAUM, D., AND CAMERON, R. Stress inoculation: A skills training approach to

anxiety management. Unpublished manuscript, University of Waterloo, 1973b.
MEICHENBAUM, D., AND CAMERON, R. The clinical potential of modifying what clients
say to themselves. Psychotherapy: Theory, Research and Practice, 1974,11, 103-117.
MEICHENBAUM, D., GILMORE, B., AND FEDORAVIClUS, A. Group insight vs. group
desensitization in treating speech anxiety. Journal of Consulting and Clinical Psychol-
ogy, 1971,36, 410-421.
MEICHENBAUM, D., AND GOODMAN, J. Training impulsive children to talk to themselves:
A means of developing self-control. Journal of Abnormal Psychology, 1971,77, 115-
MEICHENBAUM, D., TURK, D., AND BURSTEIN, S. The nature of coping with stress. Paper
presented at NATO conference on Stress and Anxiety, Korsor, Denmark, 1975.
MElZACK, R. The puzzle of pain. Ringwood, Australia: Penguin Books, 1973.
MElZACK, R., AND WALL, P. On the nature of cutaneous sensory mechanisms. Brain,
1962,85, 331-356.
MISCHEL, W. Toward a cognitive social learning reconceptualization of personality.
Psychological Review, 1973,80, 252-283.
MISCHEL, W. Processes in delay of gratification. In L. BERKownz (Ed.), Advances in
experimental social psychology (Vol. 7). New York: Academic Press, 1974.
MONAHAN, J., AND O'LEARY, D. Effects of self-instruction on rule-breaking behavior.
Psychological Reports, 1971,79, 1059-1066.
MURRAY, H. Explorations in personality. New York: Oxford Press, 1938.
NOVACO, R. A treatment program for the management of anger through cognitive and
relaxation controls. Unpublished doctoral dissertation, Indiana University, Bloom-
ington, Indiana, 1974.
PALKES, H., STEWART, M., AND FREEDMAN, J. Improvement in maze performance on
hyperactive boys as a function of verbal training procedures. Journal of Special
Education, 1972, 5, 337-342.
PALKES, H., STEWART, M., AND KAHANA, B. Porteus performance of hyperactive boys
after training in self-directed verbal commands. Child Development, 1968,39, 817-
PEALE, N. The power of positive thinking. Englewood Cliffs, No. J.: Prentice-Hall, 1960.
PETERSON, D. The clinical study of social behavior. New York: Appelton-Century-Crofts,
PITKIN, W. A short introduction to the history of human stupidity. London: Roworth & Co.,
PLATONOV, K. The word as a physiological and therapeutic factor. Moscow: Foreign
Languages Publishing House, 1959.
PLATT, J., AND SPIVACK, G. Problem-solving thinking of psychiatric patients. Journal of
Consulting and Clinical Psychology, 1972a, 39, 148-151.
PLATT, J., AND SPIVACK, G. Social competence and effective problem-solving thinking in
psychiatric patients. Journal of Clinical Psychology, 1972b,28, 3-5.
PREMACK, D. Mechanisms of self-control. In W. HUNT (Ed.), Learning and mechanisms of
self-control in smoking. Chicago: Aldine, 1970.
RICHARDSON, A. Mental practice: A review and discussion. Part 1. Research Quarterly,
1967a, 38, 95-107.
RICHARDSON, A. Mental practice: A review and discussion. Part II. Research Quarterly,
1967b, 38, 263-273.
RlDBERG, E., PARKE, R., AND HETHERINGTON, M. Modification of impulsive and reflective

cognitive styles through observation of film mediating models. Developmental

Psychology, 1971,5, 369-377.
SACHS, L., AND INGRAM, G. Covert sensitization as a treatment of weight control.
Psychological Reports, 1972,30, 971-974.
SALlZ, E., AND JOHNSON, J. Training for thematic-fantasy play in culturally disadvan-
taged children. Journal of Educational Psychology, 1974, 66, 62:>-630.
SARBIN, T. Imagining as muted role-taking: A historical-linguistic analysis. In P.
SHEEHAN (Ed.), The function and nature of imagery. New York: Academic Press, 1972.
SCHACIITER, S. The interaction of cognitive and physiological determinants of emotional
state. In C. SPIELBERGER (Ed.), Anxiety and behavior. New York: Academic Press,
SCHACIITER, S., AND SINGER, J. Cognitive, social, and physiological determinants of
emotional state. Psychological Review, 1962,69, 379-399.
SCHNEIDER, M. Turtle technique in the classroom. Teaching Exceptional Children, Fall,
1974, 22-24.
SCHWARlZ, G. Cardiac responses to self-induced thoughts. Psychophysiology, 1971, 8,
SHAPIRO, D. Neurotic styles. New York: Basic Books, 1965.
SHURE, M., AND SPIVACK, G. Means-ends thinking, adjustment and social class among
elementary school-aged children. Journal of Consulting and Clinical Psychology, 1972,
38, 348--353.
SHURE, M., SPIVACK, G., AND JAEGER, M. Problem-solving thinking and adjustment
among disadvantaged preschool children. Child Development, 1971,42, 1791-1803.
SIMKINS, L. Instructions as discriminative stimuli in verbal conditioning and awareness.
Journal of Abnormal and Social Psychology, 1963,66, 21:>-219.
SINGER, J. Imagery and daydream methods in psychotherapy. New York: Academic Press,
SINGER, J., AND MCCRAVEN, V. Some characteristics of adult daydreaming. Journal of
Psychology, 1961,51, 151-164.
SKINNER, B. F. Science and human behavior. New York: MacMillan, 1953.
SPIVACK, G., AND LEVINE, M. Self-regulation in acting-out and normal adolescents.
Report M-4351. Washington, D.C. National Institutes of Health, 1963.
SPIVACK, G., AND SHURE, M. Social adjustment of young children: A cognitive approach to
solving real-life problems. San Franscisco: Jossey-Bass, 1974.
STEFFY, R., MEICHENBAUM, D., AND BEST, A. Aversive and cognitive factors in the
modification of smoking behavior. Behaviour Research and Therapy, 1970, 8, 115-
STEINER, 1. Perceived freedom. In L. BERKOWITZ (Ed.), Advances in experimental social
psychology. New York: Academic Press, 1970.
STERNBACH, R. The effects of instructional sets on autonomic activity. Psychophysiology,
1964, 1, 67-72.
STONE, W. C. The Success System that Never Fails. Englewood Cliffs, N. J.: Prentice Hall,
STRUPP, H. Specific vs. nonspecific factors in psychology and the problem of control.
Archives of General Psychology, 1970,23, 393-401.
SUINN, R., AND RICHARDSON, F. Anxiety management training. A nonspecific behavior
therapy program for anxiety control. Behavior Therapy, 1971,2,498-510.
SUTCLIFFE, J. On the role of "instructions to the subject" in psychological experiments.
American Psychologist, 1972,27, 755-758.

TORI, c., AND WORRELL, 1. Reduction of human avoidant behavior: A comparison of

counterconditioning, expectancy, and cognitive information approaches. Journal of
Consulting and Clinical Psychology, 1973,41, 269-278.
TREXLER, 1., AND KARST, T. Rational-emotive therapy, placebo, and no treatment effects
on public-speaking anxiety. Journal of Abnormal Psychology, 1972,79, 6(k;7.
TURK, D. Cognitive control of pain: A skills training approach. Unpublished manu-
script, University of Waterloo, 1975.
VYGOTSKY,1. Thought and Language. Cambridge, Mass.: MIT Press, 1962.
WINE, J. Investigations of an attentional interpretation of test anxiety. Unpublished
doctoral dissertation, University of Waterloo, Waterloo, Ontario, 1970.
ZAPOROZHETS, A., AND ELKONIN, D. The psychology of preschool children. Cambridge: MIT
Press, 1971.
ZIMBARDO, P. The Cognitive Control of Motivation. Glenview, Ill.: Scott Foresman, 1969.
7 Physiological and
Cognitive Processes in the
Regulation of Anxiety


For the past five years we have been engaged in a program of research
whose ultimate goal has been the development and evaluation of
therapeutic methods for reducing anxiety. A basic assumption under-
lying our work has been that the successful evolution of such strate-
gies will be facilitated by advances in our knowledge about the nature
of anxiety itself. Consequently the majority of the research has
attempted to identify basic conditions (environmental and subject)
that serve to maintain or reduce the anxiety response.
In the present chapter I would like to share with you a descriptive
model of anxiety process that has guided our research and studies
from our program that relate to portions of that model. As will become
clear later on, the central focus emerging from our work has been the
role of physiological arousal, cognitive processes, their interaction,
and the importance of individual differences in those variables in
determining the maintenance and the reduction of anxiety.
An early, behavioristic account of anxiety suggested an essentially
physiological model based on simple Pavlovian conditioning. Re-
peated pairings of a conditioned stimulus (CS) and an unconditioned
stimulus (UCS) ultimately result in the elicitation of a conditioned
response (CR) by the CS alone. In the aversive conditioning situation,
that CR involves both autonomic and skeletal responses. Maintenance
of the CS-CR relationship is viewed as a function of the frequency and
intensity of CS-UCS presentations. Extinction is a simple matter of
repetitious CS presentation in the absence of the UCS. The model was
THOMAS D. BORKOVEC Department of Psychology, University of Iowa, Iowa City,


historically exemplified by Watson and Rayner's (1920) classic demon-

stration of conditioned fear acquisition and jones's (1924) program for
the elimination of acquired fear.
Although Skinner foreshadowed a two-factor theory of learning in
his distinction between Type I and Type II learning, an operant view
of anxiety is essentially a behavioral model and focuses on escape and
avoidance behavior and the discriminative role of stimuli for those
instrumental responses. Thus certain stimuli come to signal the
imminent occurrence of aversive stimuli. Those stimuli also set the
occasion for escape and avoidance responses which are negatively
reinforced by the termination of those stimuli and the preclusion of
the aversive stimulus. Avoidance responses occur with shorter latency
and to more remote discriminative stimuli because of negative rein-
forcement and stimulus generalization. Since extinction requires non-
occurrence of the aversive stimulus in the presence of the discrimina-
tive stimuli, avoidance maintains unless the organism is somehow
prevented from making the avoidance response.
Attacking the monistic theories of Pavlov and Watson, Mowrer
(1947) offered a two-factor theory of learning by suggesting that
neither the principle of association nor the law of effect could by
themselves provide a unified theory of learning. Rather,

it seems necessary to assume that there are two basic learning processes: The
process whereby the solutions to problems, i.e., ordinary "habits," are
acquired; and the process whereby emotionalleaming, or "conditioning,"
takes place. (p. 114)

The origins of anxious behavior from this model involved a two-

process sequence: (1) eS/noxious ues pairings result in the establish-
ment of the es as a danger signal which elicits fear, an intervening
variable; and (2) fear serves as an acquired drive motivating subse-
quent escape and avoidance behavior and as a reinforcer via the drive-
reducing consequence of the instrumental behavior. Anxiety main-
tains subsequent to acquisition since repeated es exposure is pre-
cluded by efficient avoidance responses to more remote cues.
Mowrer's (1947) theory has stimulated an enormous quantity of
research. While subsequent investigations have been equivocal in
supporting his mediational hypothesis (Rescorla and Solomon, 1967),
Mowrer's model has come to serve as the framework for many

behavior therapy researchers and has been the point of departure for
our own hypothesis. Specifically, we are concerned with the response-
produced feedback arising from the CR and stressed by Mowrer in
establishing the drive properties of anxiety which motivate skeletal
coping responses. Four additional considerations have contributed to
our developing model, however. First, the important work of Schach-
ter (1964) has indicated that specific emotional experience and behav-
ior may be mediated by the subject's interpretation of his own
physiological arousal and the extant environmental cues. The subse-
quent influence of his theory and research has been broadly felt, with
numerous writers such as Beck, Epstein, Lazarus, Mandler, and
Spielberger (d. Spielberger, 1972) emphasizing a central role for
cognitive factors in current anxiety theory. We feel that such a
cognitive model offers important dimensions to the role of CR feed-
back in human anxiety. Second, all of the previous learning models
assume that extinction of anxious behavior will take place as long as
feared stimuli are presented, avoidance precluded, and the UCS
absent. In clinical work, however, it is clear that the anxious client
often confronts the feared situation repeatedly in his daily life and yet
anxiety maintains despite the absence of skill deficits, overt coping
responses, or clear aversive consequences from the environment.
Among several alternatives there are two positions regarding these
observations that have interested us. Mowrer's model might be
extended to allow the substitution of cognitive avoidance behavior for
overt instrumental behavior. Alternately there may be conditions of
nonreinforced CS exposure which do not lead to extinction. Very little
direct data from anxious humans have been collected relevant to these
hypotheses. Third, while the physiological component has long been
central in behavioral definitions of anxiety, the investigation of
physiological variables has been relatively ignored. Renewed empha-
sis on its functional role in theories of anxiety appears to be appropri-
ate. Finally, although individual differences have entered into various
general theories of anxiety (e.g., Spielberger's A-trait; Taylor'S Mani-
fest Anxiety Scale), little attention has been devoted to the functional
importance of those differences in the behavior therapy literature. The
absence of that consideration has contributed to a great deal of
ambiguity regarding the theoretical mechanisms both of treatment
techniques and of the anxiety response itself.


The focus of the descriptive model emerging from these consider-

ations is on maintenance and reduction processes. Issues regarding
the origins of anxiety are relatively ignored. The variables included in
the framework are schematically represented in Table 1 and described
by the series of working assumptions presented below. The emphasis
of the model is on (1) the sequence of events that may be elicited at
any given moment upon the presentation of an anxiety-eliciting
stimulus; (2) the maintaining or reducing effects that any particular
sequence of events has upon response to the subsequent recurrence of
the feared stimulus; and (3) individual differences in the functional
relevance of any particular component of the sequence.

A. Current Stimulus Conditions

In terms of current eliciting conditions, measured anxiety is

considered to be a function of external and/or internal cues. External
cues include aspects of the environment which, because of the past
history of learning, produce fear responses. Internal cues include
verbal and nonverbal images, physiological activity, and propriocep-
tive or perceptual feedback from skeletal behavior. Each of these
internal cues may, depending on its history, serve conditional or
discriminative functions (or both) for subsequent fear components.
While either external or internal cues may be separately capable of
eliciting anxious behavior, observed anxiety reactions are most often a
function of an interaction of both sets of cues: if a fear stimulus is
presented, internal cues may be elicited; if diffuse arousal occurs, the
organism may search its environment for external cues. Although the
influence of future events on the development of the anxiety response
sequence has traditionally focused on the external cues (e.g., addi-
tional, periodic ues presentations and stimulus-generalization proc-
esses), our concern has been primarily with the effect of history on the
internal cue component of this interaction. Finally, individuals are
assumed to differ, because of differences in learning history or genetic
consitution, in terms of the relative functional importance of external
and internal cues and their interaction in determining responses to
feared stimuli.

TABLE 1 no
A Descriptive Model of Anxiety Process ~
Subsequent maintaining and reducing
Current stimulus conditions Immediate anxiety reaction reactions

External fear cues ~

1. '"gJ
Past - {
history + Cognition ~ ~b. Cognition
.!I----.--t2. '"z
3. Overt behavior----.J Lc. Overt behavior
Internal fear cues m
Autonomic arousal gr
Verbal and nonverbal images Intervention strategies
Proprioception from overt behavior




The origins of a fear response to either set of cues may reside in

classically conditioned autonomic arousal; observational learning of
autonomic, self-report, or avoidance behavior; cultural role-learning
and direct reinforcement of reports and behavioral display of fear; or a
combination of these factors. As mentioned earlier, the issue of origin
has not been our focus. It is important only to suggest that the
recurrence of these various events may playa role in strengthening the
functional effects of external and internal fear cues.

B. The Immediate Anxiety Reaction

Our working definition of anxiety is in partial agreement with

Spielberger (1966) and Paul (1969b) in suggesting that anxiety may be
conceptualized as a label denoting a complex pattern of responses
characterized by subjective feelings of apprehension and tension and
the occurrence of physiological arousal. However, we would add
behavioral manifestations of arousal (e.g., facial grimace and trem-
bling hands) and avoidance behavior to that definition. While there
would be little argument regarding behavioral arousal effects, few
definitions since Watson would include avoidance behavior. Such
coping reactions are reasonably considered to be anxiety-elicited and a
consequence, therefore, of anxiety itself. While this is probably the
frequent case, it is important to allow for a notion of skeletal
avoidance behavior not mediated by the other presumed components
of anxiety. For example, it may be possible that an "avoidance"
behavior is learned via direct positive reinforcement. The individual
may subsequently learn to label that behavior as fear, even though no
clear CR is elicited by the stimulus situation. One might argue that
such behavior should not be considered within the realm of fear
research. We would disagree, if only because many of the fear
behaviors investigated in the behavior-therapy literature may be a
function of exactly these conditions. Second, while various investiga-
tors have attempted to distinguish between fear and anxiety at both
the theoretical and empirical level, we are in agreement with Spielber-
ger that the distinction is meaningless unless the response patterns of
the two emotions differ and that little attention has yet been devoted

to the identification of those differential patterns (Spielberger, 1972).

Thus the terms will be used interchangeably in the present context.
Given this working definition of the construct, anxiety will be
operationally defined by the multiple measurement of three general
response components: cognitive, overt behavioral, and physiological.
Notice that these are gross categories which may contain several
subsets of response measures potentially reflecting several responses
within a given category. There has been considerable concern with the
empirical fact that these three measurement groups are not highly
interrelated. Two reactions are often offered to this state of affairs: the
low correlations are a consequence of dealing with subjects at the
extreme and restricted end of the anxiety distribution, and improved
measurement of each response group will ultimately reduce error
variance in the instruments and therefore produce higher intercorrela-
tions. Both reactions fail to recognize an extremely important point
that is at the core of our conceptualization. In agreement with Lang
(1968), anxiety must be considered to involve any or all of three
separate but interacting response components. Cognitive behavior,
motor behavior, and physiological reactions may be separately influ-
enced by different environmental conditions at different points in time
and may even obey different learning principles. Yet because of their
potential interaction, changes in one response component due to the
direct manipulation of its conditions may ultimately affect subsequent
changes in the response of one or both of the remaining components.
Most importantly, individuals differ in terms of the learning history
associated with each response component, resulting in individual
differences in the intensity and/or functional importance of the re-
sponse from each component in reaction to a particular feared stimu-
lus. Some individuals, for example, will report intense distress and
display rapid avoidance when confronted with feared situations, but
no evidence of increases in physiological arousal can be detected.
Others may show such autonomic increases but differ in the degree to
which they are aware of the arousal, the degree of avoidance behavior,
the level of reported discomfort, etc. The separateness of responses
considered to be evidence of anxiety (either by the researcher or by
the anxious person) allows for such important individual differences,
and their interaction will be hypothesized to account for maintaining
reactions and for the development or reduction of anxiety over time.

Unfortunately such individual differences have been largely ig-

nored in the study of anxiety and its modification. It is our conviction
that theoretical anxiety research and the development of self-regula-
tory techniques must focus on those differences. The importance of
this assumption is magnified when we consider interactions among
the response components and the impact of intervention or reduction
strategies on the anxiety reaction.

C. Subsequent Maintaining and Reducing Reactions

It is assumed that the three response components interact. That is,

the occurrence of an immediate anxiety reaction in one or more
components in response to external and/or internal cues may result in
the occurrence of subsequent reactions in one or more of the compo-
nents. Such reactions may be a function of subject characteristics of
the individual (constitutional or learned behavior) or a function of
environmental manipulations during the immediate anxiety response.
The effect of any interaction may be either the maintenance of anxiety
or its reduction upon subsequent exposure to the feared situation.
Assuming three response components potentially involved in the
immediate and subsequent reactions, there are nine possible relation-
ships. These relationships are offered as hypotheses regarding the
sequential interactions which may occur in any given case of anxiety.
None of the relationships are considered -necessary or sufficient for
anxiety maintenance or reduction. The contribution of anyone rela-
tionship to the anxiety process and the effects such a relationship may
have on the anxiety process are hypothetical. Only future research can
support the reasonableness of these relationships and their effects.
While the assumptions have some basis in the theories and research of
various investigators, the statement of the assumptions will not
represent an extensive review of the evidence in support of those
assumptions. In addition, the assumptions are stated in terms of
maintenance conditions only. Reduction conditions are implied either
in the sense that the maintaining reactions do not occur or that other
reactions antagonistic to anxiety occur. The latter conditions will be
dicussed more fully in the subsequent intervention section.

la. The Occurrence of a Physiological Reaction to Fear Cues May Result

in Subsequent Physiological Reactions That Contribute to the
Maintenance of Anxiety.

The notion of anxiety maintenance subsequent to a single expo-

sure is similar to Lader and Mathews's (1968) suggestion of a feedback
loop among highly anxious individuals. Immediate arousal cues, as
response-produced stimuli, may serve as conditional stimuli for other
physiological reactions. Alternately, physiological reactions may be
maintained even by the occurrence of nonfeared stimuli once arousal
level has been elevated by initial fear cues. The possible influence of
physiological reactions resulting in maintenance over repeated expo-
sures is similar to Eysenck's (1968) theory of incubation. The auton-
omic CR may hypothetically be viewed as a noxious stimulus which
under certain conditions serves to strengthen the CS-CR relationship
despite the absence of a UCS.

lb. Immediate Physiological Reactions to Feared Stimuli May Set the

Occasion for Subsequent Cognitive Behaviors Which Serve to Maintain
the Anxiety Response.

This relationship focuses most directly on the role of physiological

cues in maintaining human anxiety. Mowrer (1947) hinted at this
notion in referring to the "cognitive restructuring" that occurs as the
subject undergoing aversive conditioning learns to fear not only the
CS but also the entire experimental situation. The more recent and
relevant conceptualization of this process is provided by Schachter
(1964). The occurrence of diffuse arousal sets the occasion for search
for environmental cues to explain the arousal. The resulting cognitive
interpretation strongly influences the emotional reaction.

lc. Physiological Reactions to Fear Cues May Set the Occasion for
Subsequent Overt Behavior Which Maintains the Anxiety Response.

The state-dependent learning literature provides clear, general

support for this interaction. This relationship, of course, was the focus
of Mowrer's original two-process theory implicating a mediational role
for the CR in controlling instrumental avoidance. Rescorla and Solo-

mon's (1967) review found weak support for the mediational hypothe-
sis. But their conclusion was only that CRs are not required for
avoidance behavior. There is evidence that CRs may serve such a
mediational role, and these authors readily admitted that avoidance
may be "mediated by a complex of CRs, both autonomic and skeletal;
no one of these may be necessary for operant behavior, but each
contributes to that behavior" (p. 163). It is sufficient for our purposes
that there is some evidence for CR mediation (e.g., Gantt and
Dykman, 1957; Black, 1959), not as a necessary condition but as a
potentially sufficient contributor to avoidance behavior.

2a. Presentation of Fear Cues May Immediately Result in Thoughts and

Images Which Elicit Subsequent Physiological Responses Contributing
to Anxiety Maintenance.

There is ample evidence that imagery is capable of such CR

elicitation (d. Mathews, 1971). As long as cognitive events of an
anxiety-eliciting nature continue to occur, physiological arousal may
be maintained at high levels.

2b. Thoughts and Images in Immediate Response to the Feared Situation

May Result in Further Chains of Catastrophizing Self-Verbalizations
Regarding the Feared Stimulus and the Individual's Ability to Cope
with It.

Ellis's (1958) theory of neurotic behavior emphasizes this relation-

ship, while Meichenbaum (this volume) has presented empirical
evidence supporting the role of self-verbalizations in anxiety.

2c. Cognitive Response to Fear Cues Can Influence Subsequent Overt


Again, Meichenbaum's studies suggest that the presence or ab-

sence of self-instructions importantly contributes to problem behavior,
while recent research on covert rehearsal indicates that cognitive
instructions and imagery may be as powerful in improving perform-
ance as overt rehearsal (e.g., Cautela, Flannery, and Hanley, 1974;
McFall and Twentyman, 1973).

3a. Immediately Elicited Avoidance Behavior, Inept Coping Behavior, etc.,

May Serve as Conditional Cues for Increased Arousal.

The James-Lange theory of emotion represents an early, similar

version of the relationship. In support of that relationship, Black (1959)
found maximum heart-rate reaction to occur following the avoidance
response and to maintain for some time. This finding agrees with
clinical observations that some phobic individuals do not report
intense physiological reactions until after they have removed them-
selves from the feared situation and have found safety, at which time
large (and aversive) reactions occur.

3b. Behavioral Reactions to a Feared Situation May Serve to Elicit

Negative, Self-Evaluative Cognitions.

This general relationship is best represented currently by Bem

(1972) who postulates that:
Individuals come to "know" their own attitudes, emotions, and other
internal states partially by inferring them from observations of their own
overt behavior andJor the circumstances in which this behavior oc-
curs .... To the extent that internal cues are weak, ambiguous or uninter-
pretable, the individual is functionally in the same position as an outside
observer, an observer who must necessarily rely upon those same external
cues to infer the individual's inner states. (p. 2)

The Bem hypothesis may be viewed in our context as a comple-

ment to Schachter's consideration of internal states and their contribu-
tion to the subsequent cognitive interpretations mediating anxiety.

3c. Immediate Fear Behavior May Set the Occasion for Continued Fear

Again, Mowrer's (1947) focus on CR proprioception providing

response-produced stimuli for avoidance serves as an early example of
this relationship. While few data have been collected on the develop-
ment of behavioral response chains in the fear situation, Patterson
(1975) has recently suggested that most behaviors appear to occur in
"bursts," implying a facilitation effect of one response on the proba-
bility of occurrence of immediately subsequent responses of the same

The above assumptions regarding response-component interac-

tions only suggest that each component may serve conditional or
discriminative functions for other components and that those interac-
tions will influence the immediate anxiety reaction upon subsequent
exposure to the feared stimulus. While the added consideration of
individual differences in those interactions results in a highly complex
model, its basic assumption is simple: Stimuli elicit one or more
immediate anxiety-response components, which in tum may elicit
further responses; some individuals, upon exposure to the feared
situation, respond with an immediate image or thought which elicits
physiological arousal, further catastrophizing images and thoughts,
and!or avoidance behavior; others respond immediately with a phy-
siological increase, leading to cognitive apprehension, avoidance, and!
or further arousal; etc. The implications of this assumption relate
importantly to both the maintenance and the reduction of any given
case of anxiety. If, for example, the cognitive component of the
immediate anxiety reaction is primarily or solely relevant to an
individual's anxiety reaction and to the mediation of maintaining
reactions in other components, then the second three relationships
(2a-2c) are primarily or solely relevant to the maintenance of his
anxiety. By the same token, the cognitive component and its interac-
tive relationships with subsequent responses are therefore primarily
or solely relevant to strategies for reducing his anxiety. This model,
which considers such individual differences in combination with the
interactive relationships of response groups, provides a framework for
theoretical research on anxiety maintenance and reduction as well as
for analyzing the effects found in the current behavior-therapy litera-
ture and suggesting a guideline for the development of appropriate
self-regulation procedures. The latter analysis will be briefly pre-
sented first, while the former will be later exemplified by studies from
our research program.

D. Intervention Strategies

The presented description of the anxiety process allows us to view

the current behavior-therapy literature from two different perspec-
tives. The first refers to the assumed point of therapy impact on the

anxiety process. The traditional behavioral approach to conceptualiz-

ing therapeutic intervention suggests that techniques aimed at reduc-
ing maladaptive behavior and increasing adaptive response to feared
situations focus on the relationships between fear cues and immediate
anxiety reactions. A more recent view exemplified by anxiety re-
searchers in the area of cognitive behavior therapy and self-control
suggests that it may be therapeutically wise to apply intervention
strategies between immediate anxiety reactions and the subsequent
maintaining reactions. Thus, for example, while systematic desensiti-
zation is often considered to influence the relationships between fear
cues and the immediate anxiety reactions, recent developments sug-
gest that the therapist should train the client to relax himself when
confronted by the feared situation and to modify his self-verbaliza-
tions to promote more effective coping strategies (cf. Meichenbaum,
this volume). The focus is thus shifted from the permanent treatment
of a previously established stimulus-response relationship to the
teaching of a generalizable coping skill that can be used when the
client confronts any future, fear-producing situation. The client is
therefore taught to administer his own interventions between the
immediate anxiety responses and the subsequent maintaining re-
sponses. In terms of the descriptive model, such stress-inoculation
training assumes that the physiological and cognitive components are
relevant responses maintaining the anxiety reaction and that self-
regulatory techniques attacking both of those responses would be the
treatment of choice. By logical extension of the model, the therapeutic
process would involve (1) identification of the relevant response
components comprising the immediate anxiety reaction; (2) identifica-
tion of subsequent maintaining reactions; and (3) application of the
most efficient and efficacious techniques to eliminate the immediate
reactions and training the client in self-control skills that preclude the
occurrence of maintaining reactions. Research in behavior therapy is
only beginning to develop such techniques and skills. Hopefully
research on the conditions of anxiety maintenance and reduction
within this descriptive framework will facilitate those efforts.
Second, the role of individual differences in immediate response-
group strength allows us to make some sense out of a current literature
which seems to indicate that a variety of independent manipulations
can effectively modify fear behavior. Recent reviews of animal condi-

tioning (Wilson and Davison, 1971), human psychophysiology (Ma-

thews, 1971), and the role of cognitive factors in desensitization
(Wilkins, 1971) have arrived at a very similar conclusion: repeated CS
exposure appears to be the only necessary condition for the ultimate
elimination of fear. Beyond this, there is little agreement. Procedur-
ally, there are many ways to present the CS. Theoretically several
underlying processes have been offered to explain the effectiveness of
various CS presentation techniques. Empirically many techniques
have been found effective in reducing fear. However, the existing
literature addressing procedural, theoretical, or even basic effective-
ness questions has almost completely ignored the role of individual
differences in anxiety-response components, despite Paul's (1969a)
warning that the question of technique effectiveness cannot be di-
vorced from subject characteristic considerations. Such neglect has
given rise not only to invalid claims of effectiveness (Bernstein and
Paul, 1971) but has resulted in erroneous conclusions regarding the
underlying theories of the techniques as well as of the anxiety process
If, for example, the theory of a technique implies that the
physiological component of fear is (1) important in the definition of
fear and (2) the focus of that therapy technique, then tests of the
technique are valid only if subjects displaying a strong physiological
response make up the treatment sample and only if assessment of
physiological responding is included in the outcome measures. If the
theory of another technique implies that the behavioral component is
important and the specific target of the treatment, then valid conclu-
sions are possible only if subjects display strong avoidance and are
measured primarily by behavioral scales. While this may appear to be
intuitively obvious, disregard for this basic notion is common and
stems from the inaccurate assumption that a technique generally
influences "anxiety" and that any measure of "anxiety" is sufficient to
draw conclusions relevant to that technique and its theory. Selection
of subjects and measures which fit the theory never occurs to the
investigator who accepts that assumption.
We may categorize several common manipulations in terms of
which response components are their focus.
1. Physiological Reactions: relaxation therapies, systematic desen-
sitization, implosive therapy, biofeedback.

2. Cognitions: rational emotive therapy, semantic conditioning,

self-instructions, modified self-verbalizations, thought stop-
ping, placebo and expectancy manipulations.
3. Overt Behaviors: reinforcement of approach behavior, response
prevention, modeling, contact desensitization.
The theory and/or procedures involved in each manipulation
appear to focus on a particular anxiety-response component. This is
not to say that the other components are conceptually ignored or that
other components are not influenced. Rather, the focus of the tech-
niques is at least on the specified response component, while addi-
tional focus on other components and the interactive relationships
among components allows for multiple effects by any technique.
However, each technique involves to a greater or lesser degree
repeated, nonreinforced CS exposure. Thus each technique, if admin-
istered for a long enough period of time, will, hypothetically, effect a
reduction of anxiety, however measured. But this is sufficient for the
experimental analysis of neither the technique nor the anxiety process.
The current desensitization literature is an excellent example of a
body of therapy research which has ignored subject characteristics and
as a consequence has produced endless debates regarding active
mechanisms of effect. A variety of such mechanisms has been offered,
each with associated empirical support. Thus systematic desensitiza-
tion effects have been said to be due to: counterconditioning, extinc-
tion, expectancy, therapist reinforcement of nonfearful behavior, in-
formation feedback, controlled attention shifts, exposure to
contingencies of nonavoidance behavior, modified cognitions via attri-
bution manipulations, covert rehearsal, self-controlleaming, semantic
associations, modifed self-sentences, simple demand and placebo
effects, etc. The majority of the relevant studies have involved phobic
college students selected on the basis of self-reported fear and behav-
ioral avoidance on a pretest. The conclusions of the studies are
ordinarily based on change in the behavioral component of anxiety
after four to eight treatment sessions.
As I have previously argued (Borkovec, 1973b), almost all therapy-
outcome studies employ selection criteria and dependent measures
which do not directly tap a response group considered of major
importance in the definition of anxiety and desensitization processes
(i.e., the physiological component). It is of little wonder that manipu-

lations such as reinforcement, modeling, demand, placebo, etc., pro-

duce significant change in approach behavior and self-report if one
assumes that avoidance behavior and reports of fear can be currently
maintained by reinforcement, modeling, demand, suggestions, etc.
Without matching theory, measures, and subject differences in rele-
vant response groups, erroneous conclusions regarding any anxiety-
reduction techniques will continue. As long as researchers continue to
select phobic subjects on the basis of only one or two non stringent
criteria, they will continue to find that scores of manipulations, some
of which are inherent in desensitization procedure, can be separately
effective in changing avoidance behavior. Such demonstrations are
indeed important. They are relevant, however, only to anxious behav-
ior that is not mediated by a physiological response component and
are irrelevant to the basic theoretical issues of desensitization.



The presented descriptive framework of anxiety process predicts

that everything influences everything else, making it useless as a
theory. The model does, however, delineate some of the variables
considered to be of importance in anxiety. The task of the researcher
operating from such a model is to identify the conditions (environ-
mental and subject) under which interrelationships among the varia-
bles do and do not occur and what effects those interrelationships
have on behavior over time. This has been the goal of our program.
The majority of our research has focused on two of the response
components: physiological and cognitive processes. Since the model
addresses anxiety maintenance/reduction, our aim has been to identify
the conditions under which interrelationships of those two variables
lead to the maintenance or reduction of anxiety. Since individual
differences in response components are at the core of the model, our
research strategy has been to assess the effects of manipulations on
subjects differing in response-component strength, principally in level
of physiological arousal and autonomic awareness.
The specific experimental strategy we have adopted has involved
repeated exposure of subjects to feared stimuli, with intervening
cognitive and/or physiological manipulations. Two main target behav-

iors have been employed: snake phobias and social anxiety. In the
course of our research, it has been found that these two groups are
quite different in two important respects. First, the behavioral compo-
nent of the fear reaction of snake phobics, as typically selected in
outcome research, is very susceptible to modification by demandi
suggestion (Borkovec, 1973a; Bernstein, 1973). Simply ask the subject
(implicitly or explicitly) to show behavioral improvement and suggest
that he will be less anxious, and reduced behavioral fear will be
observed. The fear behavior of socially anxious subjects has not been
so influenced (Borkovec, Stone, O'Brien, and Kaloupek, 1974). Second,
on rare occasions when physiological measurement has been obtained
during snake avoidance tests (e.g., Borkovec, 1973a; Craighead, 1973),
maximal average heart rate has been no greater than 96 beats per
minute. In studies of speech and social anxiety, similar measurements
have revealed increases in arousal typically between 113 and 118 beats
per minute (Borkovec, Wall, and Stone, 1974; Borkovec, Stone,
O'Brien, and Kaloupek, 1974; Singerman, 1974). On the basis of the
demand studies, it is clear that the use of snake phobics requires
controls for demand and suggestion effects. What often appears to be
fear reduction in therapy outcome studies with this target behavior
may often simply reflect the influence of extratherapeutic variables on
avoidance behavior. On the basis of the heart-rate data, it is equally
clear that the physiological response will oridinarily be a relevant fear
component in studies involving social anxiety and an irrelevant
component in those employing snake phobias. Since we have used
both targets, we have had an opportunity to observe the effects of
physiological and cognitive manipulations on two anxiety problems
differing in the extent to which the physiological response is function-
ally relevant. In the following presentation of research studies, there-
fore, I will refer to investigations with low physiological reactors
(snake phobics) and high physiological reactors (social- and speech"
anxious subjects).
These two characteristics (demand/suggestion susceptibility and
level of physiological arousal) appear to be related to each other and
give rise to one of the central theses of our work. To the extent that the
immediate anxiety reaction involves a weak physiological component,
simple manipulations of the cognitive and behavioral components of
fear (such as demandisuggestion) will be effective in changing those
components. To the extent that the immediate anxiety reaction in-

volves a strong physiological component, such manipulations will be

ineffective and will be effective only after the autonomic component is
reduced. By logical extension throughout the anxiety model, to the
extent that any response component is strong and immediately elicited
by the feared stimulus, that component must be modified directly if
efficient fear elimination is to be accomplished.
The origin of this assumption stems from my dissertation (Borko-
vec, 1970, 1972) and a brief summary of that study will serve to set the
stage for a description of our subsequent research program.
In a rather typical outcome study, snake phobic college females
who failed to touch a live snake on pretest were randomly assigned to
one of four treatment conditions: systematic desensitization, implo-
sive therapy, avoidance-response placebo, and no treatment. In order
to assess the contribution of client expectancy for improvement, half of
the subjects in each condition received therapeutic instructions re-
garding the purpose of the procedures, while half were presented
non therapeutic instructions. The theories underlying desensitization
(Wolpe, 1958) and implosive therapy (Stampfl and Levis, 1967) both
suggest that phobic behavior is maintained because of negatively
reinforced avoidance of feared stimuli. This common assumption
directly suggested a model-relevant placebo condition. Subjects visu-
alized the same hierarchy items presented to the two therapy condi-
tions. However, as soon as the subject signaled the presence of
anxiety, a standardized avoidance response was visualized. The
placebo procedure thus analogized the phobic's behavior assumed
to occur in real-life situations and served as a theoretically inert
Posttest and four-week follow-up assessments were made subse-
quent to four weekly therapy sessions. Several important findings
emerged and suggested the major variables to be explored in our
developing program: (1) Expectancy instructions had a dramatic effect
on the behavioral component of fear, suggesting a potentially impor-
tant role for cognitive variables in fear reduction. (2) The model-
relevant placebo condition showed behavioral improvement equal to
that displayed by the therapy conditions. A serious problem was
apparently raised for the theories underlying desensitization and
implosion. (3) Desensitization and implosion both reduced the phy-
siological component of fear at posttest regardless of whether subjects
were given positive or neutral expectation for improvement. Subse-

quent avoidance behavior for these subjects was influenced, however,

only when an expectation for improvement was instructionally estab-
lished. The role of physiological cues and their interaction with the
subject's interpretation of the procedures and the effects of the
procedures were thus implicated.

A. The Role of Physiological Arousal and Cognition

In our subsequent attempts to identify maintaining/reducing

conditions for anxiety, we have focused primarily on the role of
physiological arousal and cognition. In terms of the model, we have so
far been concerned with internal cues, immediate physiological and
cognitive reactions, interrelationships between those reactions and
subsequent physiological and cognitive responses, and the effects that
the relationships have on subsequent anxiety. Subject characteristics
enter into these relationships through investigations of individual
differences in magnitude of the physiological response and in the
perception of that response. Studies relating to physiological reactions
and cognition will be presented first, followed by investigations of the
contribution of subject characteristics to such relationships.

1. The Effect of Cognitive Avoidance Behavior in Maintaining the Anxiety

of Low Physiological Reactors

The theoretical models underlying both systematic desensitization

and implosive therapy suggest that avoidance behavior precludes CS
exposure and hence extinction. The ideal control condition in a
theoretical investigation of either technique would involve a group of
subjects who visualize the same phobic hierarchy but visualize per-
forming an avoidance response upon elicitation of the CR. As men-
tioned earlier, such a group was included in the outcome study of
desensitization, implosion, and expectancy effects on snake phobic
behavior (Borkovec, 1972). While basal skin conductance data collected
during therapy revealed that all three treated groups displayed reduc-
tions in arousal over hierarchy-item presentations, a review of the
psychophysiological research suggested that heart-rate measures pro-
vided data during imagery procedures that were more in line with
theoretical predictions than were skin-conductance measures (Mat-

hews, 1971). Subsequent analyses of samples from continuously

monitored heart rate indicate that the avoidance-response group
remained at high levels throughout imaginal exposure to the CS
scenes, while both desensitization and implosion groups displayed
significant decreases in heart rate (Borkovec, 1974). Thus despite
repeated CS exposure frequently demonstrated to produce autonomic
decreases (Mathews, 1971), the addition of imagined avoidance behav-
ior in response to the CS resulted in maintenance of physiological
arousal. In addition, outcome pulse-rate of the avoidance-response
group failed to decrease relative to nontreated control subjects, while
both desensitization and implosion groups displayed significantly
greater reductions than no treatment. The combined process and
outcome results, which occurred irrespective of whether the subject
expected to improve or not because of the therapy administration,
directly support Mowrer's two-factor theory of anxiety maintenance in
humans for a physiological response and provide evidence for a
physiological-cognitive interaction in determining that maintenance.
In terms of the descriptive model, external cues (therapist verbaliza-
tion of the CS hierarchy scene) and internal cues (subject visualization
of the scene) elicited a conditioned fear response. Upon the subject's
awareness of an anxiety response, a cognitive avoidance response was
made. The effect of this relationship was the maintenance of the
physiological component of fear as evidenced by both process and
outcome heart-rate measures. In contrast, scene presentations in both
desensitization and implosion involved repeated, nonreinforced CS
exposures with a relative preclusion of cognitive avoidance, and both
techniques resulted in reductions of autonomic process and outcome
measures. The fact that these processes occurred in fearful subjects for
whom the physiological component is not very strong is quite encour-
aging. The results suggest that a simple cognitive response can
maintain anxiety despite repeated CS exposure.

2. Cognitive Expectancy Related to Self-Reported Outcome and

Physiological Process

Research by Goldstein and others (e.g., Friedman, 1963; Gold-

stein, 1960; Goldstein and Shipman, 1961; Piper and Wogan, 1970) has
suggested that client expectancy for improvement is powerfully related
to therapeutic outcome. Unfortunately the majority of the research

from which this conclusion derives is based on correlational analyses

of self-report data. More recent investigations have attempted to
manipulate expectancy via the presence or absence of therapeutic
instructions during therapy administration. Half of these studies have
found such a manipulation to be unrelated to therapy outcome, while
the other half (including the above study from our laboratory) demon-
strated greater behavioral improvement under positive (therapeutic)
than neutral (nontherapeutic) expectancy. Reviews of this literature
have indicated that the differences between the two sets of studies
appear to reside in important methodological considerations. Wilkins
(1973), for example, found that therapists and/or observers were not
"blind" to the condition status of subjects in studies demonstrating a
significant expectancy effect. One aspect of positive expectancy in-
structions, however, is that high demand is placed on the subject for
displaying less avoidance behavior on the posttest. He knows that the
experimenter expects improved behavior and responds accordingly. A
review of these studies has noted that investigations finding a
significant difference between positive and neutral expectancy condi-
tions (i.e., high and low demand) employed nonstringent selection
criteria, while studies failing to find an expectancy effect used severe
selection criteria (cf. Borkovec, 1973b). If it is assumed that the
strength of the physiological component increases as fear selection-
criteria become more stringent (Bernstein and Paul, 1971), then the
outcomes of these studies support the hypothesis that expectancy
manipulations are effective in changing the behavioral component
only to the extent that the physiological component is absent. In the
only clinical trial of an expectancy manipulation where a strong
physiological response is likely (Gelder, Bancroft, Gath, Johnston,
Mathews, and Shaw, 1973), high- versus low-expectancy instructions
had no effect on the outcome of neurotic patients treated by desensiti-
zation or implosive therapy.
Despite the apparently weak contribution of expectancy to out-
come change, the importance of client expectation for improvement,
born in clinical impressions and Goldstein's research, remains a firmly
entrenched assumption in clinical folklore. Our research has obtained
two bits of further information regarding this construct. First, in three
studies within a sleep-disturbance program (Borkovec, Kaloupek, and
Slama, 1975; Slama, 1975; Steinmark and Borkovec, 1974) correlations
obtained between expectancy ratings and outcome were generally

nonsignificant. Second, however, analysis of the heart-rate data dur-

ing desensitization, implosion, and avoidance response placebo con-
ditions (Borkovec, 1974) revealed a significant main effect of expect-
ancy during the process of treatment. This effect provides an example
of a cognitive-physiological relationship in anxiety process. Subjects
who were given therapeutic instructions regarding the purpose of the
treatment procedures displayed lower heart rate throughout the ther-
apy sessions than subjects given nontherapeutic instructions, relating
expectancy manipulations to an objectively measured and potentially
important therapy process variable. If low physiological arousal is
facilitative of desensitization effects as Lader and Mathews (1968)
have suggested, then positive expectancy instructions may contribute
to fear extinction via its arousal-reducing effects during repeated CS
exposure. If generation of intense CR is an essential part of implosive
therapy process, as Stampfl and Levis (1967) have indicated, then
positive expectancy may initially mitigate implosion outcome.

3. Manipulation of Physiological Cues among Low Physiological Reactors

The anxiety outcome study mentioned above exemplified a strat-

egy of monitoring naturally occurring physiological arousal during CS
exposure and observing what anxiety effects occur as a function of
manipulating responses subsequent to the immediate anxiety reac-
tion. A second approach for investigating the role of physiological
cues in fear involves the manipulation of false physiological feedback.
This strategy directly addresses the physiological-cognitive relation-
ship in that the subject's perception of the autonomic cues is of
primary importance in determining the effects of those cues. The
earliest study of this nature was conducted by Valins and Ray (1967) in
an attempt to provide evidence for a cognitive interpretation of the
effects of systematic desensitization. Valins and Ray presented alter-
nating slides of the word shock and feared stimuli to snake phobic
subjects. Experimental subjects were informed that sounds presented
during the slide presentations were their heartbeats, while control
subjects were told that the sounds were extraneous noise. No change
in feedback heart-rate occurred during presentation of the feared
slides, while feedback heart-rate increased upon presentation of the
word shock. The experimental group displayed significantly less
avoidance to a live snake than the control group. The authors argued

that the subjects' interpretation of the physiological feedback resulted

in a modification of cognitions regarding the feared object, i.e., "That
stimulus no longer affects me internally." It is this revised cognition
that presumably mediates changes in overt behavior toward the fear
object. Several attempted replications (Gaupp, Stem, and Galbraith,
1972; Kent, Wilson, and Nelson, 1972; Sushinsky and Bootzin, 1970)
failed to produce this phenomenon. These studies differed from the
original Valins and Ray investigation in one or more procedural
aspects. Some studies included a pretest to ensure the selection of
truly phobic subjects. Such an in vivo experience with the feared
object, however, may mitigate the effects of symbolic presentation of
the feared stimulus in association with the bogus feedback. Second,
Valins and Ray employed a high-demand posttest (a money incentive
was offered), while some of the replications essentially involved low-
demand posttesting. Because of the susceptibility of avoidance behav-
ior to demand!suggestion influences and without control for the types
of demand inherent in the conditions, it would perhaps be desirable
to ensure high posttest demand in order to control for uncontrolled
demand influences. Differential avoidance improvement among treat-
ment conditions would more clearly reflect level of anxiety than
situational determinants. Borkovec and Glasgow (1973) replicated the
Valins and Ray study in a Solomon four-groups design. Experimental
subjects heard false heart-rate feedback during the presentation of
slides of snakes. The feedback occurred to a hierarchy of slides similar
to hierarchies constructed in systematic desensitization treatment.
Heart rate was heard to increase to initial presentations of items and to
decrease across repetitions of the same slide and across hierarchy
items. Control subjects observed the same slides and heard the same
sounds but were not informed that the sounds were representative of
their heart-rate response to the pictures. Half of the subjects in each of
the conditions were pretested in a behavioral approach test, while half
were not pretested. All subjects were informed that the effects of the
procedures would be to reduce their fear of snakes and that it was
important for the success of the experiment that they approach as close
as possible. Thus the posttest instructions were high in demand!
suggestion, offering a valid design despite the use of low-fearful
snake-phobic subjects. As predicted, the experimental group showed
significantly less avoidance than the control group, but only under
nonpretested conditions. In addition, continuous heart-rate monitor-

ing during the slide presentation task indicated that pretested subjects
displayed significantly greater heart-rate reactivity to the snake slides
than did nonpretested subjects regardless of feedback conditions. The
original predictions of the study were based on the notion that fear
occurring during an in vivo pretest would mitigate the effects of
feedback manipulations during symbolic CS presentations. Under
those conditions the only modified cognition that might result would
be, "I am afraid of the actual snake, but I am not afraid of slides-
of snakes." In addition, however, the heart-rate data during slide
presentations suggested that in vivo exposure to the feared situ-
ation sensitized the subject to symbolic presentations of that feared
object and resulted in maintained autonomic arousal during those
In terms of the descriptive model, symbolic CS presentations were
paired with a tape-recorded sound. Depending on whether the sounds
were labeled as noise or physiological arousal, the subject's subse-
quent cognitive interpretation either was irrelevant to his fear and
therefore to its modification or was relevant and resulted in reduced
fear. This effect importantly occurred only on the behavioral compo-
nent of anxiety and only in the absence of a pretest exposure. Such an
outcome suggests that cognitive manipulations regarding physiologi-
cal cues will be effective only if subjects are low physiological reactors.
The fact that pretest exposure both increased actual physiological
arousal and mitigated the effects of the cognitive manipulation sug-
gests that in vivo presentations can lead to arousal maintenance to
both symbolic CS exposures and subsequent in vivo exposures, even
among subjects who ordinarily display little physiological reactivity to
the feared stimulus. Thus while false feedback experiments may
supply evidence for the role of physiological cues in maintenance or
modification of fear, actual physiological arousal may produce an
overriding effect.

4. Manipulation of Physiological Cues among High Physiological Reactors

Unfortunately previous studies of false feedback effects on fear

employed feedback tasks intervening between pre- and posttest in
vivo exposures. The critical issue regarding the role of physiological
cues centers on the occurrence of feedback during the subject's
confrontation with the actual feared situation. During the conduct of

two studies investigating the effects of demand instructions on snake

phobic avoidance (Borkovec, 1973a), increasing versus decreasing false
heart-rate feedback was presented to two groups of subjects during
the postlest snake exposure. No main effects of feedback occurred on
any of the dependent measures. Yet the Valins and Ray (1967) and
Borkovec and Glasgow (1973) results suggested that avoidance behav-
ior may be modified upon exposures subsequent to feedback experi-
ence. Consequently a study was designed in which false physiological
feedback was presented during the actual exposure to feared situation,
and its effects were assessed on a subsequent exposure (Borkovec,
Wall, and Stone, 1974). Speech-anxious subjects prepared and pre-
sented three consecutive speeches. During the second speech, one of
five feedback conditions was presented: heart rate increasing, heart
rate decreasing, heart rate no change, and two control conditions.
Subjects in the heart-rate-increasing condition heard what was re-
ported to be their own heart rate during the last minute of the
preparation period of the second speech and throughout the presenta-
tion of the second speech. Feedback indicated fluctuating but increas-
ing rate in heartbeats throughout the task. Similarly subjects in the
heart-rate-decreasing condition heard gradual decreases in heart rate
throughout the speech. Subjects in the heart-rate-no-change condition
were given heart-rate feedback indicating relatively little fluctuation at
a relatively low level throughout the second speech. Subjects in a
feedback control condition heard the same heart-rate tape that was
employed in the heart-rate-increasing condition but were told that the
sounds were extraneous noise. Subjects in a no-feedback control
condition wore the headphones but heard no sound during the
speech. Pronounced heart-rate increases did in fact occur during all
three speeches. There was, however, no effect of feedback conditions
on actual heart rate. As predicted, no significant differences were
found during the feedback speech among any of the conditions. Also
as hypothesized, the heart-rate-decreasing and no change conditions
resulted in significantly less self-reported and behavioral manifesta-
tions of anxiety on the posttest (third) speech than did the heart-rate-
increasing condition. The combination of high physiological arousal
which occurred in all groups and increasing false feedback in the
heart-rate-increasing condition resul~ed in the maintenance or facilita-
tion of fear behavior upon exposure to the feared situation subsequent
to the feedback experience. Despite repetitious presentation of the

feared situation, fear behavior maintained. Given the Valins and Ray
and the Borkovec and Glasgow studies demonstrating posttest behav-
ioral differences subsequent to feedback tasks and the above study
demonstrating a lag in false feedback effects during repeated expo-
sures to the feared situation, some process must occur between the
feedback experience and the subsequent exposure to the feared
situation to result in maintenance of behavioral anxiety. If phYSiologi-
cal feedback is considered to be a response-produced stimulus and if,
in agreement with Eysenck (1968), such conditioned responses may
serve as aversive stimuli, then it can be hypothesized that awareness
of the physiological component of the CR can in itself serve to
maintain fear behavior during nonreinforced CS presentations. False
physiological feedback may then facilitate or maintain fear because of
its similarity to actual history of experience with real feedback,
assumed to contain aversive properties. An alternative hypothesis in
line with Valins and Ray's theory regarding systematic desensitization
is that internal cuing sets the occasion for verbal mediators generated
by the subject in response to perceived autonomic functioning. In the
case of the speech-anxiety study, subjects in the heart-rate-increasing
condition had an opportunity during the preparation period prior to
the third speech and subsequent to the feedback speech to verbalize to
themselves interpretation and/or evaluation of their arousal as repre-
sented by the false feedback. Focusing of attention on such cognitions
stimulated by the false feedback experience may have disrupted the
preparation of a subsequent speech, or the cognitions themselves
elicited additional arousal which interfered with subsequent speech
performance and increased reported anxiety.
It is important to note that self-report and behavioral-anxiety
measures of the heart-rate-decreasing and no-change conditions did
not show significantly greater reductions relative to the control condi-
tions. In contrast to snake-phobic samples, in which cognitive manip-
ulations may, under limited circumstances, result in reduced avoid-
ance behavior relative to controls, it appears necessary to modify the
strong physiological component present in speech anxiety before
cognitive interpretations can result in decreases in fear behavior. The
facilitated behavioral and self-reported fear in the heart-rate-increas-
ing condition, however, indicates that fear behavior can be main-
tained under conditions of both high actual arousal and a cognitive
interpretation of high arousal.

5. Cognitive Manipulations on Naturally Occurring Arousal in High

Physiological Reactors

Recently research from the Schachter tradition has resulted in the

development of an attribution therapy, whereby cognitive manipula-
tions attempt to minimize aversive or dysfunctional emotional states
by leading the individual to attribute his arousal to some nonemo-
tional source. While a variety of studies have found significant effects
(e.g., Dienstbier and Munter, 1971; Nisbett and Schachter, 1966; Ross,
Rodin, and Zimbardo, 1969; Storms and Nisbett, 1970), our suspicion
that simple cognitive manipulations are likely to be ineffective when
there is a strong physiological component to the fear reaction led to
the conduct of an attribution study of speech anxiety (Singerman,
1974). In this study 60 speech-anxious subjects (30 moderately fearful
and 30 highly fearful) presented two consecutive speeches. Meaning-
less noise, used as the misattribution stimulus, was presented from
the time the subject entered the laboratory to the end of the first
speech but was absent during the second speech. Prior to the first
speech subjects were randomly assigned to one of three attribution
conditions. Subjects in the arousal condition were told that the noise
would increase physiological activity. Attribution theory would pre-
dict that these subjects should be less anxious during the stressful
situation, since they would attribute their naturally occurring arousal
cues to the noise rather than to the speech situation. Subjects in a
sedation condition were given a placebo statement indicating that the
noise would have a decremental effect on physiological activity.
Control subjects were given no expectation regarding the likely effects
of the noise. No evidence was found for the reverse placebo effect
predicted by attribution theory. In fact, among highly fearful subjects
the arousal condition produced greater behavioral anxiety than the
sedation and control conditions, and this difference maintained to the
second speech. The nature of the target behavior and its associated
physiological activity appears to have been critical in the failure of the
cognitive manipulation. If subjects have an extensive past history of
association between a setting and strong physiological responses,
successful cognitive misattribution appears unlikely. In terms of the
descriptive model and similar to the conclusions of the false feedback
study on speech anxiety, the combination of high physiological
arousal and a cognitive set indicating that an extraneous stimulus

would produce increases in arousal (the arousal condition) appears to

have an additive effect resulting in increased saliency of arousal cues
and therefore increased fear behavior.

6. The Development of a Physiological Fear Component in Nonanxious

Subjects and Possible Cognitive Contributions

In the study of social-anxiety measurement (Borkovec, Stone,

O'Brien, and Kaloupek, 1974), high- and low-anxious males had
undergone pre- and posttest exposures to a laboratory interaction with
a female research assistant. Half of the subjects received high-demand!
suggestion posttest instructions at posttest, while half were given low-
demand instructions. The demand manipulation failed to produce
improvement on any measure. There were two surprising outcomes,
however, which may relate to the origins of an anxiety response and
the role of cognition in facilitating anxiety over time: (1) low anxious
subjects displayed a significant increase in heart rate from pre- to
posttest and (2) low anxious subjects in the low-demand condition
showed that increase during the anticipatory phase prior to the
beginning of the interaction, while subjects in the high-demand
condition evidenced that increase only during the interaction phase
It is reasonable to conclude that confronting a nonreceptive female
is functionally an aversive experience for ordinarily nonanxious males,
an experience which can result in increased physiological activity over
repeated exposures. In addition, it may be hypothesized that cognitive
appraisals of that experience between testings may contribute to the
heightened arousal. The significant interaction between demand,
phase of interaction, and testing supports such a cognitive interpreta-
tion. In high demand these subjects were reassured that they would be
less anxious, a suggestion that indeed mitigated anticipatory arousal
but failed to mitigate reactive arousal upon presentation of the
interaction situation. In low demand, on the other hand, the aversive
experience led to facilitated anticipatory arousal. Once the subjects
were interacting, the facilitative effect of the aversive pretest experi-
ence on arousal was absent. Thus given the increased arousal that
occurs in nonanxious subjects as a function of the aversive nature of
the pretest, the presence or absence of an expectation of reduced

posttest fear established via the demand manipulation may influence

the point of inflection of arousal (anticipatory or reactive). These
hypotheses require further research. It would be important to deter-
mine whether the general physiological increase among nonanxious
subjects ultimately results in increased behavioral and self-reported
anxiety and whether cognitive manipulations or point of arousal
inflection contribute to such increased anxiety.

B. The Role of Individual Differences in Physiological Arousal

and Autonomic Perception

Assessment of ongoing physiological activity and the manipula-

tion of cognitive and physiological cues via instructional set and false
feedback have given us several means for investigating the contribu-
tion of physiological and cognitive reactions to fear maintenance and
reduction. An additional method involves the investigation of indi-
vidual differences in physiological reactions and the perception of
those reactions. The potential importance of the physiological fear
component was discussed earlier with regard to response group
patterns. In the present section, studies relating subject characteristic
differences in that component to anxiety and manipulation of its
conditions will be presented.
Since our interest has resided in the functional role of physiologi-
cal feedback as a response-produced stimulus, it appeared that either
direct assessment of an autonomic response or measurement of per-
ceived physiological activity would provide a potentially useful di-
mension of individual differences. Certainly the greater the autonomic
activity, the greater the probability that physiological cues would be
functionally related to subsequent behavior. On the other hand, mild
or moderate arousal may provide sufficient feedback for some individ-
uals to be functionally important in their fear behavior. In addition,
measurement of the perception of autonomic feedback has the
advantage of providing access to a response that relates to both physio-
logical and cognitive response groups of the individual. The majority
of our studies, therefore, has focused on autonomic perception

1. The Autonomic Perception Questionnaire

Reports by Mandler and his associates in the late 1950s indicated

the existence of a self-report measure of perceived autonomic func-
tioning and provided initial evidence of its potential utility as a bridge
between physiological activity and self-reports of anxiety. Their Ques-
tionnaire on the Perception of Feeling consisted of 28 items which
requested subjects to indicate the degree to which they noticed
various bodily reactions during two emotional states. The first 21
items (Autonomic Perception Questionnaire (APQ); see Table 4) related
to subjective experiences when the individual was anxious; the last 7
items were rated in terms of experiences when the individual was
happy. Ratings on the original scales were made by the checking of a
point on a continuous line, with the ends of each line representing
absence versus extreme presence of each bodily reaction. Scoring
involved the use of a transparent overlay dividing the line into 10
equal parts. Our program has used the same 21 APQ items, although
the item scales have been rated by a simple circling of the appropriate
number (0-9) representing the rater's experience of that reaction. In
both Mandler's and our own research, high and low perceivers have
been defined in terms of the sum over individual item scores. While
Mandler has employed a general form of the APQ ("When you are
anxious, do you notice ... "), we have variously used both general
and task-specific ("During the experiment, did you notice ... ")
Normative data have not been previously presented. Conse-
quently the general APQ has been administered to two samples of
introductory psychology students (spring and fall semesters, 1971) at
the University of Iowa and to all adult patients seen in the university's
psychological clinic from 1971 to 1973. Table 2 presents the mean
scores on each item of the APQ and its total for females and males
from these three groups.
Females tended to obtain higher total scores than males, although
the difference was significant only in one of the normal samples.
Females scored significantly higher than males in both normal samples
on items reflecting awareness of cold hands, shallow breathing, lump
in throat, upset and sinking stomach, and the bothersomeness of
bodily reactions. Differences in the clinic sample again favored the
females but were restricted to experience of face becoming hot,

perspiring, and headache. When the normal samples were pooled and
compared to the clinic sample, neither female nor male patients were
found to differ significantly from normal subjects of the same sex on
total APQ score. Clinic females did score significantly higher than
their n6rmal counterparts on frequency of awareness of bodily reac-
tions and muscle tension. Clinic males displayed significantly higher
scores than normal males on frequency of awareness, lump in throat,
upset stomach, and bothersomeness of the bodily reactions, while
they scored significantly lower on perception of hot face and perspira-
tion. The greater frequency of awareness of bodily reactions among
the clinic samples suggests the particular relevance of those cues to
clinical problems.

APQ Item Means for Females and Males from Normal and Clinic Samples

Spring sample 1971 Fall sample 1971 Clinic sample 1971-73

Females Males Females Males Females Males

APQ Item (N = 239) (N = 181) (N = 215) (N = 211) (N = 37) (N = 38)

1 4.87 4.82 5.35 5.09 5.49 5.34

2 4.85 5.05 5.16 5.18 5.84" 6.00"
3 4.30 3.95 4.26" 3.74 4.16" 2.55"
4 4.32" 2.75 4.22" 3.21 3.54 2.89
5 5.08 5.36 5.51 5.42 5.49" 4.29"
6 3.56 3.78 4.00 3.89 3.46 3.55
7 4.84 4.79 5.15 4.90 6.32" 5.53
8 3.49 1.99 3.28" 2.11 4.05" 2.53
9 4.54 4.70 4.96 4.78 4.35 4.53
10 4.78 4.92 5.19 4.90 4.65 4.76
11 4.68 4.91 5.09 4.77 4.78 4.55
12 3.25 3.39 3.36 3.44 3.81 3.71
13 3.43 3.45 3.36 3.35 3.95 3.34
14 2.99 3.24 2.90 3.11 3.16 3.21
15 2.77" 2.20 3.0Qb 2.46 3.35 2.92
16 3.01 3.02 3.47" 2.95 3.46 2.63
17 3.97" 3.37 4.46" 3.59 4.35 4.39"
18 4.41" 3.35 4.76" 3.43 4.86 4.87"
19 4.58" 3.80 5.31" 3.99 5.27 4.68
20 3.92 3.63 4.37" 3.79 4.76 4.00
21 3.88" 3.39 4.15" 3.62 4.76 4.79"
Total 85.52 79.86 91.35" 81.73 93.86 85.08

a I-tests (p < .05) for same-sex scores between clinic sample and pooled normal samples.
I-tests (p < .05) for between-sex scores within a sample.

In the first study to employ the APQ (Mandler, Mandler, and

Uviller, 1958) subjects selected for reporting high levels of autonomic
perception displayed significantly greater autonomic reactivity (heart
rate, PGR, and respiration) during stress than low perceivers and
overestimated that feedback relative to the underestimates of the low
perceivers. The general APQ was found to correlate significantly with
Taylor's Manifest Anxiety Scale (MAS) and was related to the number
of physiological cues reported during the stress situation itself. Small
but significant correlations between perceived and actual autonomic
reactivity during stress were found in a second study (Mandler and
Kremen, 1958), which employed an unselected sample of subjects. In
contrast to the earlier study, in which high- and low-perception
groups were selected on the basis of extreme scores, low-perception
subjects overestimated and high perceivers underestimated actual
arousal. The APQ again correlated significantly with the MAS. Per-
formance on a vocabulary test was unrelated to actual autonomic
arousal but negatively related to the perceived arousal. Our fall sample
has also completed the primary items of Ullmann's (1962) facilitation-
inhibition scale, which correlates significantly with Byrne's (1961)
repression-sensitization scale. Correlations between the APQ and F-I
scales were significant for both females (r = .31, P < .01) and males (r
= .37, P < .001). As would be expected, high perceivers are high
facilitators (sensitizers).
Mandler's replicated relationship between perceived autonomic
feedback and both actual physiological measures and a general self-
report anxiety index suggested that the APQ might be a promising
bridge between cognition and physiology. In addition, the perform-
ance differences between low and high perceivers suggested that the
subject characteristic of autonomic perception may be related to
important overt behavior. In addition, recent investigations have
found that APQ is related to performance in heart-rate-control tasks
(Bergman and Johnson, 1971; Blanchard, Young, and McLeod, 1972).
There appeared to be sufficient evidence, then, that the instrument
would provide a meaningful dimension of individual differences
potentially related to fear, its maintenance, and its reduction.

2. Perceived Arousal, Actual Arousal, and Demand/Suggestion

The first attempt to obtain evidence for a relationship between
subject characteristics in arousal, anxiety, and the conditions influenc-

ing anxiety occurred in the two early demand/suggestion studies

(Borkovec, 1973a). Instructional demand and suggestions of improve-
ment were chosen as the external stimuli to be manipulated because of
our concurrent interest in the general issue of demand characteristics
in fear research. Individual differences in physiological cues were
defined by (1) actual pulse-rate response in anticipation of exposure to
the feared object and (2) level of autonomic perception as measured by
the general APQ. Snake-phobic subjects were categorized into high or
low levels on both the reaction and the perception factors. It was
predicted that the presence of strong physiological cues would reduce
the effects of the demand manipulation. In the first of two studies,
reaction and perception were unrelated to any outcome measure
among nonphobic subjects. Phobic subjects who were high in reac-
tion, however, displayed less behavioral improvement than phobic
subjects low in reaction. Contrary to the prediction, high-perception
subjects showed greater behavioral improvement than low-perception
subjects. On the pulse-rate outcome measure, phobic subjects who
were high or low in both reaction and perception displayed signifi-
cant reductions in pulse-rate response to the phobic stimulus from
pre- to posttest, while subjects in the other two groups showed no
change, that is, they showed a maintenance of the physiological fear
The second study involved two manipulations: increasing versus
decreasing heart-rate feedback during the posttest exposure and low-
versus high-demand instructions. Subject-characteristic groupings
were made in the same manner as in the first study. Under low-
demand instructions, no effects of false feedback, perception, or
reaction were found. Under high demand, however, high perception
was again related to greater approach improvement than low percep-
tion. Three of the four low-reaction groups displayed greater improve-
ment than three of the four high-reaction groups. One of the discre-
pant groups (low-perceptionllow-reaction subjects under decreasing
false feedback) showed little change due to ceiling effects on the
avoidance scale. Substantial and unpredicted improvement occurred
in the second discrepant group (high perception/high reaction under
increasing feedback). In the high-demand condition and under in-
creasing feedback, pulse-rate improvement differences among subject-
characteristic groups were almost identical to those occurring in the
first study: accurate perceivers (subjects high or low in both percep-

tion and reaction) displayed reductions over tests, while inaccurate

perceivers showed a maintenance of pulse-rate response.
Several implications follow from these two studies. First, while
demand characteristics were found to exert strong effects on avoidance
behavior, their influence depended on arousal-cue factors. In general,
the greater the actual reaction, the less effective was the demand
manipulation. The principle exception was the striking approach
improvement for subjects high in perception and reaction under the
increasing false-feedback and high-demand condition. This same
feedback and demand condition, however, resulted in significant
pulse-rate improvement among these subjects, providing further sup-
port for the hypothesis that the physiological fear component must be
modified before demand or cognitive manipulations will affect overt
behavior. Second, high-perception subjects were influenced by de-
mand to a greater extent than low-perception subjects, contrary to
predictions. Either autonomic perception does not reflect the func-
tional role of physiological feedback, or its relation to behavior is more
complex than was originally conceptualized. Third, the interaction of
perception and reaction determined the probability of extinction of a
physiological component of fear response. Whether accuracy of percep-
tion or a more basic attention-focusing characteristic underlies this
effect remains unknown.

3. Perceived Arousal and the Incubation Phenomenon

Eysenck (1968) has suggested that under certain conditions, re-

peated es exposure in the absence of ues presentations may result in
increases, rather than decreases, in fear. The noxious character of the
eR itself is highlighted in contributing to this incubation effect. In
addition to studies cited in his review, other investigations involving
repeated or prolonged exposure to feared stimuli (Boulougouris,
Marks, and Marset, 1971; Bresnitz, 1967; Miller and Levis, 1971;
Rankin, Nomikos, Opton, and Lazarus, 1964; Rohrbaugh and Riccio,
1970; Rohrbaugh, Riccio, and Arthur, 1972) have supported the
occurrence of the phenomenon in both animal and human subjects.
Three studies implicated duration of es exposure as an important
variable determining the occurrence of the incubation process. Specifi-
cally, in each study brief es exposure resulted in greater fear relative
to zero or long exposure. Stone and Borkovec (1975) attempted to

replicate the effect and to determine the relationship of autonomic

perception to that effect. Because of Eysenck's stress on the noxious
effects of the CR in facilitating fear increases during CS presentation,
it was reasonable to expect that the incubation process would be
greater among high than among low autonomic perceivers. Phobic
college students observed a snake for 0, 15, or 45 minutes subsequent
to a behavioral pretest. In line with earlier suggestions regarding the
valid use of snake-phobic subjects, posttest was administered under
high demand. In a replication of the earlier studies, the results
revealed less avoidance for the 0- and the 4S-minute conditions
relative to the lS-minute condition. In addition, analysis of continu-
ously monitored heart-rate data indicated reductions in heart-rate
reaction to the phobic stimulus from pre- to posttest among zero and
long-exposure subjects and an increase among brief-exposure sub-
jects. Analyses of high and low autonomic-perception groups revealed
that the incubation effect on avoidance behavior occurred only among
high perceivers. The results of this study supported the existence of an
incubation phenomenon in humans and the