Vous êtes sur la page 1sur 21

Journal of the History of the Behavioral Sciences, Vol.

41(2), 131149 Spring 2005


Published online in Wiley Interscience (www.interscience.wiley.com). DOI 10.1002 /jhbs.20079
2005 Wiley Periodicals, Inc.

OPERATIONISM IN PSYCHOLOGY: WHAT THE DEBATE IS ABOUT, WHAT THE


DEBATE SHOULD BE ABOUT*
ULJANA FEEST

I offer an analysis of operationism in psychology, which is rooted in an historical study of


the investigative practices of two of its early proponents (S. S. Stevens and E. C. Tolman).
According to this analysis, early psychological operationists emphasized the importance of
experimental operations and called for scientists to specify what kinds of operations were
to count as empirical indicators for the referents of their concepts. While such specifications were referred to as definitions, I show that such definitions were not taken to constitute a priori knowledge or be analytically true. Rather, they served the pragmatic function of enabling scientists to do research on a purported phenomenon. I argue that
historical and philosophical discussions of problems with operationism have conflated it,
both conceptually and historically, with positivism, and I raise the question of what are the
real issues behind the debate about operationism. 2005 Wiley Periodicals, Inc.

The term operationism (or operationalism) is commonly associated with the Harvard physicist Percy Bridgman (18821961), who famously claimed that in general, we mean by a concept
nothing more than a set of operations; the concept is synonymous with the corresponding sets of
operations (Bridgman, 1927, p. 5). While the notion of operationism was never very influential
within physics, it gained a fair amount of popularity within psychology and the social sciences
(Smith, 1997, p. 668). Even today, the notion of operationism plays an important role in psychology, with many introductory textbooks on psychological methods devoting a section to operationism. However, over the years there has also been a sporadic but ongoing debate about the nature and tenability of psychological operationism. After varieties of this position were first
formulated and elaborated on by Stevens (1935b, 1935c, 1936, 1939a) and Tolman (1936/1958f,
1937, 1938/1958d), a number of papers appeared in the Psychological Review (Bergmann &
Spence, 1941; Israel, 1945; Israel & Goldstein, 1944; Pennington & Finan, 1940; Waters &
Pennington, 1938; Weber, 1940), critically discussing the vices and virtues of this position. This
debate culminated in a symposium on operationism, organized by the Psychological Review
(Boring et al., 1945). A decade later, another philosophical symposium was devoted to the issue
(Frank, 1956). During the same time period, the notion of operationism underwent certain modifications within psychology (Campbell & Fiske, 1959; Garner, Hake, & Erikson, 1956). The notion of operationism was once again discussed in the early 1980s (Leahey, 1980, 1981, 1983, who
was critical of the position, and Kendler, 1981, 1983, who defended it), and the early 1990s (Green,
1992; Koch, 1992, both of whom were very critical of operationism). Most recently, the journal
Theory and Psychology published a positive analysis of the development of operationism (Grace,
2001), followed by several critical commentaries (Bickhard, 2001; Green, 2001; Rogers, 2001).
*Editors Note: This article, based on a paper delivered at the annual conference of the European Society for the
History of the Human Sciences held in York in the summer of 2003, is the winner of the ESHHS/JHBS Early Career
Award for 2003.

ULJANA FEEST is a research scholar at the Max Planck Institute for the History of Science in Berlin,
where she works on the historical relationship between Gestalt psychology and philosophy of science, and
the epistemology of psychological experiments. She received a degree in psychology at the University of
Frankfurt (Germany), and a Ph.D. at the Department of History and Philosophy of Science, University of
Pittsburgh. This paper is based on parts of her dissertation, entitled Operationism, Experimentation,
and Concept Formation (University of Pittsburgh, 2003).

131

132

ULJANA FEEST

The disagreements between advocates and critics of operationism appear to be largely of


a philosophical nature. However, in discussing problems with operationism, proponents of this
debate (in particular, critics of operationism) have made certain conceptual and historical assumptionsi.e., they have worked with particular notions of what the position of operationism
states and what its historical origins were. In this article, I want to question some of these conceptual and historical assumptions. To this end, I will draw on case studies of two early operationists, Stanley Smith Stevens (19061973) and Edward Chace Tolman (18861959). These
case studies serve a dual purpose. On the one hand, I use them to provide a conceptual analysis of operationism, which is grounded in an analysis of the role operationism played in the experimental research of some of its early proponents. On the other hand, I use them to provide
evidential support for my thesis that operationism was historically not as closely tied to certain
other movements (most importantly, logical positivism) as is assumed by its critics.
ANALYZING METHODOLOGY IN THE CONTEXT OF INVESTIGATIVE PRACTICE
Critics of operationism have proceeded by pointing out that operationism is fatally flawed,
because it has its roots inand/or is identical withflawed philosophical positions (e.g., Green,
1992; Leahey, 1980, 1981, 1983; Rogers, 2001), or arises from misunderstandings of such positions (Green, 1992; Koch, 1992). In this article, I will focus on the assumption that operationism
is an expression of a logical positivist epistemology and a verificationist theory of meaning.1 This
assumption is clearly behind the charge that psychologists have failed to notice that neither of
those positions is still en vogue within philosophy. Varieties of this critique have been formulated
both by philosophers and by historians of psychology. For example, the philosopher Fred Suppe
writes: [I]t seems to be characteristic, but unfortunate, of science to continue holding philosophical positions, long after they are discredited (Suppe, 1977, p. 19). In a similar vein, the historian Thomas Leahey states that [p]sychologists . . . persisted (and still persist) in attempting to
define each theoretical term empirically even after the positivists had given it up for cognitively
significant interpretive systems (Leahey, 1980, pp. 132133). Interestingly, however, the same
historian also asserts that in actual scientific practice psychologists do not provide operational
definitions, despite their claims to the contrary. Instead, he observes that [t]he test maker must
persuade the psychological community that his definition . . . is a good one. . . . We find that
operational definitions are not analytic truths, but are subject to empirical confirmation. This suggests that they are not definitions at all (Leahey, 1980, p. 138).
However, if psychologists do not, as a matter of fact, operationally define their terms along
the model of an outmoded philosophical theory of knowledge and meaning (and I believe that
Leahey is right about this), it is not so clear that a critique of those outmoded philosophical theories has much relevance to operationism as practiced by psychologists. Thus, we need an analysis of what the position really states, such that we can then delineate the grounds on which we
want to attack, or defend, it. In this article, I aim to provide such an analysis of operationism,
grounded in analyses of the scientific contexts that first gave rise to the emergence of this concept. By asking what function this concept played in the investigative practices of its proponents,
I follow a trend in the historiography of psychology (e.g., Danziger, 1990) that does not take
methodological writings at face value, but tries to gain an understanding of methodologies by
asking how these methodologies were applied to, and emerged in the context of, specific re1. Another critique has been to argue that psychologists misunderstood Bridgman (Koch, 1992). While there is some
historical evidence that Bridgman did not play an important role in the emergence of operationism in psychology
(e.g., Hardcastle, 1995), an in-depth treatment of this issue would require an analysis of Bridgmans operationism.
Such an analysis cannot be provided in this article.

OPERATIONISM IN PSYCHOLOGY

133

search questions. Some recent commentators on operationism share this focus on investigative
practice (Rogers, 2001). However, while raising interesting questions about the emergence and
longevity of operationism (see also Rogers, 1989), this author does not offer a detailed analysis
of the position itself. Instead, he criticizes defenders of operationism for believing that grand
metatheoretic problems of the positivist project can be solved by the introduction of a new
method (Rogers, 2001, p. 61). The assumption here is that operationism has the very same
metatheoretic foundations (and, hence, problems) as the positivist project. It should be emphasized that not all critics of operationism assume that it was historically closely linked to logical
positivism. For example, Green (1992) explicitly denies this. Yet, the suggestion that psychology would be better off if operationists had taken notice of the decline of logical positivism is a
pervasive theme in the critical literature. I will now take a closer look at this suggestion.
TWO NOTIONS OF OPERATIONISM
Roughly, we can distinguish between two theses that are frequently attributed to operationism, an epistemological thesis and a semantic thesis. According to the epistemological thesis, all knowledge claims have to be reducible to actual or potential observations (see Salmon,
1985). According to the semantic thesis, the meaning of a concept can be exhaustively defined
by stating particular operations and their observational results. Both of these theses are commonly associated with the philosophy of logical positivism. Therefore, I will dub this construal
of operationism the positivist reading of operationism. In this section, I contrast this reading
with my own methodological reading of operationism.
Operationism: The Methodological Reading
The thesis of this section is that psychological operationism was never intended as a theory of meaning or a theory of knowledge in the philosophical sense. By this, I mean that psychologists did not intend to say, generally, what constitutes the meaning of a scientific term. Nor
did they intend to provide a general account of justification for scientific knowledge. This does
not mean that semantic and epistemological questions were of no concern. Thus, I believe that
in offering operational definitions, scientists were partially and temporarily specifying their
usage of certain concepts by saying which kinds of empirical indicators they took to be indicative of the referents of the concepts.2 For example, in his psychophysical work on attributes of
tonal experience, the operationist Stanley Smith Stevens started with the prior assumption that
discriminatory behavior in response to auditory stimuli could be treated as indicative of auditory experience. This assumption is what was behind his definition of experiencei.e., that
to experience is, for the purpose of science, to react discriminatively (Stevens, 1935c, p. 521).
Given, further, a certain methodology for prompting subjects to discriminate according to particular features of their auditory experience, this prior assumption enabled Stevens to do empirical research on the density and volume of tones and to show how each of these attributes
of experience vary as a function of particular physical stimuli. In a similar vein, in his research
on problem-solving behavior in rats, the operationist Edward Chace Tolman, who believed that
behavior was dependent on cognitions and demands, worked with a prior assumption, according to which the demand for a certain object varies relative to the degree with which the organism has been deprived of that item, and that the vigorousness of searching behavior was, in
turn, a function of this. Given these prior assumptions, Tolman took himself to be able to show
experimentally how particular desires vary as a function of deprivation.
2. The terms operational definition and operationalization were, and are to this day, frequently used synonymously.

134

ULJANA FEEST

I will argue in the next section that these types of definitions did not have the status that
philosophers usually associate with the termi.e., they did not have the status of a priori knowledge or analytical truths. Rather, they were either temporary assumptions about typical empirical indicators of a given subject matter, which allowed researchers to get empirical investigations
off the ground, or they were presentations of the outcomes of experiments, which were assumed to individuate a given phenomenon particularly well. Thus, on my construal, they had a
methodological function. Regarding the question of whether operationism was intended as an
empiricist epistemology, it may be helpful to distinguish between two notions of epistemology.
According to the first notion, the aim of epistemology is to provide a theory of what it would take
to justify existing systems of knowledge. According to the second, the aim of epistemology is to
formulate guidelines for the acquisition of new knowledge. This latter notion of epistemology
may also be referred to as methodology (I take this distinction from Dingler, 1936/1988). While
philosophers are traditionally interested in the former, it is a contention of this article that early
psychological operationists, as practicing experimental scientists, were interested in the latter.
On the methodological reading suggested here, an investigation of operational definitions
in psychology provides a framework for studying the process of concept formation in an experimental context, orto use the terminology of Hans-Jrg Rheinbergerof the coming into
being of epistemic things (Rheinberger, 1997).
Operationism: The Positivist Reading
As is well known, the epistemological and semantic tenets of positivism were, for related
reasons, soon recognized as problematic, and were subject to gradual changes and refinements
until the 1960s (Carnap, 1936/1937; Hempel, 1950, 1952, 1954; Quine, 1951). In a nutshell, it
was recognized that (a) there are statements, which scientists take to be justified, even though it
is impossible to exhaustively rephrase them in terms of observation sentences, and (b) there are
statements about objects, which we intuitively recognize as meaningful, despite the fact that the
concepts that occur in those statements cannot be exhaustively defined in terms of operations
and resulting observations. The epistemological recognition led to the insight that a theoretical
sentence cannot be verified, but at best confirmed, by observations. This implies that the meanings of such statements cannot be reduced to methods of verification. Carnap realized this early
on for statements containing disposition terms, stating that such terms can only partially be defined by observations (Carnap, 1936/1937), and later adding that many terms of a theory are implicitly defined by their place in the theory (Carnap, 1956). In a similar vein, philosophers called
into question the notion of explicit definitions as requiring necessary and sufficient conditions
of application (Hempel, 1965). The recognition that observations can, at best, confirm scientific
statements went hand-in-hand with the introduction of a dichotomy between theory- and observation language, such that sentences in the latter could be used to confirm sentences in the former. This dichotomy, however, was soon called into question by reference to the theory-ladenness of observation (e.g., Hanson, 1958), and because it relied on the distinction between
analytic and synthetic sentences, which was famously and radically attacked by Quine (1951).
With this simplified overview of the development of positivist philosophy in mind, we
can now appreciate that critics of operationism attribute to this doctrine various views that
were held (or are commonly thought to have been held) by philosophers of the positivist tradition.3 First, there is the charge that operational definitions purport to provide necessary and
3. The received view about positivist philosophy of science has recently been challenged (see Friedman, 1999). In
the current article, I will not deal with this aspect of the problem. In other words, I will merely question the assumption that psychological operationists held the beliefs commonly attributed to positivists, not that positivists held
the beliefs commonly attributed to positivists.

OPERATIONISM IN PSYCHOLOGY

135

sufficient conditions of application. This, as was observed in the literature and reiterated by
later critics, would lead to an uncritical multiplication of concepts (Leahey, 1981; Weber,
1940), and fly in the face of the scientific intuition that one and the same concept can apply
in more than one situation, and that it can be applied even in the absence of an empirical condition of application (Hempel, 1954). Related to the worry about a multiplication of concepts
is the charge that operationists are antirealists about the referents of their conceptsi.e., that
one can arbitrarily introduce concepts regardless of whether they refer to something in the
world (e.g., Leahey, 1983; Michell, 1999). Second, there is the charge that operationists buy
into the analytic/synthetic distinction, believing statements that contain operational definitions to constitute a priori knowledge and thereby be analytically true, in which case there
is nothing left to discover (Green, 1992, p. 296), and any empirical evidence in support of a
definition would be circular, thus robbing it of its explanatory value (Leahey, 1981).
OPERATIONISM IN SCIENTIFIC CONTEXT: THE CASES OF STEVENS AND TOLMAN
In this section, I will elaborate on my case studies of Stevenss and Tolmans operationisms,
showing that the positivist reading does not do justice to their operationisms. In the following section, this analysis will be complemented with brief historical outlines of how their operationisms
originated and what was the nature of their contact with positivist philosophers of science.
Stevenss Operational Treatment of Consciousness
Stanley Smith Stevens (19061973), who is mainly known for his work on psychophysics and measurement theory, arrived at Harvard for graduate studies in the fall of
1931. Less than two years later, he defended his dissertation on the perceived attributes of
tones. His advisor was E. G. Boring (18861968). After a few more yearshaving worked as
a researcher in various other departments at Harvardhe accepted an offer to become an instructor of psychology. During the second half of the 1930s, he published his four papers on
operationism (Stevens, 1935b, 1935c, 1936, 1939a). Since these papers are full of references
to his psychophysical work on attributes of tones, I shall begin by outlining this work, as well
as the way Stevens uses the notion of definition there.
As mentioned above, Stevenss psychophysical work was concerned with providing evidence for the existence of certain kinds of conscious auditory experience. While it had long
been known that the subjective experience of loudness and pitch of a tone is a function of the
energy and frequency (respectively) of the physical stimulus (Boring, 1935), Stevens wanted
to show that the subjective experience of density and volume of a tone also varies as a
function of variations in those two physical stimuli, thereby making the case that the experience of density and volume (which had previously only been reported in a qualitative fashion)
are genuine phenomena, which are distinct from other kinds of auditory experiences (Stevens,
1934a, 1934b, 1934c, 1935a). In conducting this research, he presupposed that discriminatory
behavior in response to auditory stimuli could be treated as indicative of auditory experience.
Based on this broad assumption, he then tried to develop an experimental design that would
allow him to provide empirical evidence for particular kinds of experience (volume and
density). Here, I will focus on his research on volume. Stevens came up with an experimental paradigm, whereby subjects were given two tones of different frequencies and asked to regulate the intensity of the lower one until the perceived volumes of both tones were equal. It
turned out that subjects increased the intensity of the stimulus with the lower frequency in
order to make its volume equal to the stimulus with the higher frequency, and that the required
increase in intensity was smaller for pairs of tones with high frequency. Thus, Stevens delin-

136

ULJANA FEEST

eated a unique pattern of responses to particular combinations of stimulus dimensions, which


he treated as evidence for the existence of experienced tonal volume.
In what sense may Stevens be said to be offering definitions? I will begin by looking at his
assertion that to experience is, for the purpose of science, to react discriminately (1935c, p.
521). Did he mean by this that the expression experience has the same meaning as discriminative behavior? Did he mean that the presence of discriminative behavior is always a necessary
and sufficient condition for the correct application of the term experience? Based on his research, I think that this is clearly not what he has in mind. Rather, Stevens presupposed that experience of tonal volume or density is phenomenal (and thereby, presumably, that the phenomenal aspect is an integral part of the meaning of the term). The question, for him, is how to get
at particular kinds of phenomenal experience in an experimental context. His assertion is that
this can only be done via the behavior of the organismi.e., that in an experiment, discriminative behavior is a necessary condition for attributing conscious experience to an organism.
Having devised an experiment that elicits such behavior in a regular fashion, he concluded,
[w]e are justified in saying that volume is a phenomenal dimension of tones (1934a, p. 406).
Now, what about the definitions of tonal density and volume that Stevens offered as a
result of his empirical investigation? While Stevens seemed to think that the criteria offered in
his definition of tonal volume were sufficient conditions for the applicability of the term, I
dont believe that he took them to be necessary conditions.4 This point is related to the question of whether he was an antirealist about the referent of the concept. I believe that the fact
that he conducted research on the neurophysiological basis of the experience of tones (see
Stevens & Davis, 1936, 1938) shows that he believed the concept of tonal volume to be physically realized. This suggests that (a) he took the term consciousness to be more than a logical
construct or a useful fiction (i.e., that he was not an antirealist about its referent) and (b) he
would have been open to the possibility that it might in principle be detectable in more than
one way (i.e., that the operational definition he offered was not intended to state necessary conditions for the applicability of the term). The status of his definition, I would like to suggest,
was that of an empirical finding that was taken to confirm the existence of a phenomenon. This
leads us to our last questioni.e., whether Stevens took either of those two types of definitions to be a priori true or unrevisable. The answer, I believe, is quite explicit in his own writings. He thought of definitions as factual statements that can be changed5:
[Definitions] take into account the state of factual knowledge at a given time. It is for
good reason that the discovery of new related facts may make a revision of the criteria
necessary so that we may include or exclude the new observation from the class denoted
by the original definition. . . . No concept can be defined once and for all: every concept
requires constant purging to keep it operationally healthy. (Stevens, 1935c, pp. 519, 527)
Tolmans Operationalization of Desires
Edward Chace Tolman (18861959) went to graduate school in the joint department of
philosophy and psychology at Harvard University, where he received his PhD in psychology in
4. I will return to epistemological problems with the assumption that the empirical result in question is a sufficient condition of application for the term (i.e., that it is sufficient evidence for the existence of the phenomenon in question).
5. An example of this is provided by Stevens (1935b) when he writes that his research shows Klpes definition of sensory attribute to be wrong. Klpe (1893) had defined this notion in terms of the criterion of independent variability,
according to which a particular attribute of sensation can be varied by varying the associated stimulus-dimension (e.g.,
frequency or pitch). Stevenss research shows that some dimensions of auditory experience can only be varied by varying both stimulus dimensions.

OPERATIONISM IN PSYCHOLOGY

137

1915. He taught at Northwestern University for three years and then assumed a position at the
University of California at Berkeley in 1918, where he stayed for the rest of his life. In 1932,
he published his book, Purposive Behavior in Animals and Men. In the second half of the
1930s, Tolman published three papers in which he outlined his operationism (Tolman,
1936/1958f, 1937, 1938/1958d). Tolmans operationism has to be seen in the context of his research on problem-solving behavior in rats. As is well known, Tolman attributed to rats the
ability to form mental maps of their environments. As is less well known, he also assumed that
the rats behavior is influenced by an internal drive or demand, which is dependent on the state
of biological need the rat is in at the time when the problem solving is required. Given that
Tolman saw himself as a behaviorist, he recognized that it was problematic to posit the existence of these two types of internal states (cognitive maps and drives), which he called intervening variables. His operationism was motivated by his desire to justify the practice of manipulating the state of organismic need by providing evidence for the existence of these states.
In his first explicit paper on operationism, Tolman (1936) characterized this position as asserting the existence of a set of intervening variables, certain functions whereby these intervening variables are related to particular independent variables (external stimuli), and certain functions whereby these intervening variables are related to particular dependent variables (types of
behavior).6 While his presentation of his position is at times confused, I believe that, based on
his scientific work, the following rational reconstruction can be given: Intervening variables are
posited internal entities or phenomena that are assumed to causally intervene between stimuli
and outward behavior. As in the case of Stevens, we find, strictly speaking, two kinds of definitions in the work of Tolman. To illustrate this, let us look at his papers, An Operational
Analysis of Demand and The Determiners of Behavior at a Choice Point (Tolman, 1937,
1938/1958d). As mentioned above, Tolman started out with a general presupposition about demandsi.e., that they can be affected by depriving organisms of particular types of objects (e.g.,
the hunger demand can be affected by depriving an organism of food) and that this in turn has
an impact on behavior. Based on this prior assumption, Tolman conducted what he called a
defining experiment, as a result of which he offered a mathematical equation of how the intensity of food-searching behavior in rats varies as a function of food deprivation.
Once again, we may ask in what sense Tolman can be said to be offering definitions here.
First, it seems highly unlikely that Tolman intended his general notion of demand (as involving some kind of behavior in response to particular kinds of stimuli) to be exhaustive of
the meaning of the term in ordinary language. Just as Stevens did not deny the subjective element of conscious experience, so Tolman did not deny the subjective character of certain mental states (see also Tolman, 1958c, 1927/1958b), but rather denied that this subjective feel could
serve as a basis for a privileged kind of data in psychology. Thus, as in Stevenss case, the point
of the definition was to get an empirical handle on a phenomenon. Furthermore, when operationalizing hunger in his research, the point, frequently, was not to do research on hunger per
se, but to get a grip on this variable, in order to be able to control for this variable when testing hypotheses pertaining to other variables. For example, in his 1932 book (i.e., before the
publication of his operationism papers), Tolman operationalizes hunger in terms of time since
last feeding. This, then, enabled him to conduct experiments about the question of whether
different kinds of food are equally attractive to rats, given the same level of hunger (see
Tolman, 1932a, for many examples of this strategy). When Tolman (1937) later provided what
he called a defining experiment for the notion of hunger, he was attempting to legitimate
the scientific practice he was already engaged in.
6. Some of these ideas are already expressed in Tolman (1935/1958g), albeit without using the term operation(al)ism.

138

ULJANA FEEST

We may now ask what he took the status of this defining experiment to be: Did it provide
a definition in the sense of necessary and sufficient conditions of application? Did he take it
to refer to a real entity or was he an antirealist? Did he take the definitions to be a priori true
and unrevisable? With respect to the first two questions, the verdict is mixed. In Tolmans writings on operationism, we find statements to the effect that a theory is a complex mathematical
function, the terms of which are useful fictions. This is also how Tolman has been interpreted
by MacCorquodale and Meehl (1948). However, I believe that we see here a certain lip service
paid to the philosophy of logical positivism, whose proponents he had met by that time (more
about this below). In his earlier work, it is clear that Tolman believed intervening variables to
be causally efficacious in the production of behavior. This suggests to me that he took them to
be real and that there might, in principle, be more than one way to identify them (i.e., the
defining experiment does not provide a necessary, but at best a sufficient condition of application; see also Amundsen, 1983, in support of my claim that Tolman was not an antirealist
about the referents of his concepts). With respect to the last question, Tolman was less explicit
than Stevens. However, if we look at his writings, we find statements like the following: [This
behavior] will be taken as empirical evidence for, and definition of, immanent expectancy
(Tolman, 1932a, p. 20). This suggests to me that for Tolman (as for Stevens), the line between
definition and empirical fact was not cut in stone, and that his definitions were working assumptions, based on what was known so far.
SOME HISTORICAL ORIGINS OF STEVENSS AND TOLMANS OPERATIONISMS
In the previous section, I have shown that the positivist reading of operationism does
not seem to fit the writings or practices of two operationists investigated here. In this section,
I will complement these accounts with some aspects of the conditions that contributed to the
emergence of these operationisms. The thesis of this section is that both Stevenss and Tolmans
methodological views about operationism co-evolved with their scientific ideas and investigative practices, which preceded their contact with logical positivists.
Stevens and the Psychophysical Tradition
An important factor toward the development of Stevenss operationism was his advisor,
Edwin Boring (18861968). Boring was important both in shaping Stevenss understanding of
the history of psychophysics (through his historical work, e.g., Boring, 1929) as well as his theoretical and methodological views. With respect to the latter, Borings The Physical Dimensions
of Consciousness (1933) is especially important. About this book, Boring later wrote, [t]ucked
away in my little book was this basic faith in operationism (Boring, 1952, p. 44).7 Stevens was
closely involved in the production of the book.8 This book served the dual purpose of articulating a physicalist methodology for psychology and attempting to develop a modern-day variety
of Wundt and Titcheners project of investigating elements of consciousness. Let us take a brief
look at both of these aspects (for a more detailed account, see Feest, 2002).
7. In this context, Boring also mentions that this basic faith in operationism goes back at least to his paper The
Stimulus-Error, published in 1921. Borings 1923 paper, Intelligence as the Test Tests It, should also be mentioned. Both of these articles precede Stevenss operationism by more than a decade. A detailed historical investigation of the relationship between those earlier papers by Boring is needed to assess Borings own claims about
the matter.
8. This becomes apparent both in Borings preface to the book as well as in Borings autobiographical recollections
(Boring, 1952) and Stevenss private notebooks at the time. These notebooks, as well as Stevenss correspondence
with Boring, are at the archives of Harvard University (shelf number HUG (FP)-2.45, Boxes 1 and 2).

OPERATIONISM IN PSYCHOLOGY

139

Borings book had the aim of rejecting psychophysical dualism in its ontological and
methodological variety. Ontologically, I believe we can attribute to Boring some kind of
mind/brain identity theory. By asserting the existence of conscious experience, but stating that
it is identical with brain states, he distanced himself from both traditional substance dualism
and from behaviorism. Here, I will focus on the methodological aspect.9 Boring wanted to reject the notion that psychology required a different methodology from other sciences. In arguing this, he distanced himself from Titcheners introspectivism (see Boring, 1929). Borings
1933 book was, in part, an attempt to reject Titcheners methodological dualism and replace it
with a single, physicalist conception of scientific method. Historically science is physical science. Psychology, if it is to be a science, must be like physics (Boring, 1933, p. 6). For Boring,
this meant that experience was a theoretical entity, to be inferred on the basis of behavior:
There is no way of getting at direct experience, because experience gives itself up to science
indirectly, inferentially, by the experimental method (Boring, 1933, p. 6). Two years later,
Stevens used a very similar formulation in arguing for his view that if experience was to be
a scientific concept, it had to be amenable to the operational procedure, rather than being accessed by some more immediate, introspective means: Operationism requires that we deal
only with the reportable aspects of experience. Operational psychology knows precisely nothing of unreported consciousness (Stevens, 1935b, p. 327).
Having talked about the origins of Stevenss methodological commitments, let us now look
at the origins of his theoretical commitments. Here, I believe we can make out two strands. On
the one hand, Stevenss research was in the psychophysical tradition, going back to Fechners attempts to measure experience as a function of physical stimuli (Fechner, 1860).10 Stevens saw
himself as carrying forth this research, while explicating quite clearly the underlying assumptions about empirical indicators of conscious experience. This need for explication was, I believe, prompted by the heightened methodological awareness of the problematic status of consciousness as an object of scientific study in the aftermath of behaviorist critiques. The other
strand, once again, takes us back to Borings book of 1933, Dimensions of Consciousness. It was
an explicit goal of this book to further develop Titcheners notion of dimensions of consciousness, albeit stripped of Titcheners dualism: My book was a move away from Titchener, but it
also served to make his dimensionalism clear (Boring, 1961, p. 53). What Boring was referring to in the latter part of the sentence here was Titcheners structuralismi.e., the project of
studying the mind by dissecting it into its basic parts (Boring, 1933, also referred to this project
as mental chemistry). Given the focus of early experimental psychology on the study of sensation, it is not surprising that there was some psychological debate about sensations as basic elements of the mind (see Boring, 1933, 1942). According to Boring (and cited by Stevens), scientific interest in attributes of sensation goes back to Wilhelm Wundt (1893, Chapter 10), who
characterized sensation in terms of attributes (quality and intensity), without, however, giving a
systematic account of this analysis.11 Klpe (1893) gave a more systematic treatment of attributes, distinguishing between quality, intensity, and duration (for all senses), and adding extension for the senses of vision and touch. In addition, there were attempts to introduce mental el9. Stevens rejected Borings idea that substance dualism could be disproved on scientific grounds. In his private notebook of early 1933, he mentions a discussion he and Boring had about this (HUG (FP)-2.45, Box 1).
10. More recent predecessors dealing specifically with auditory experience were Helmholtz (1863) and Stumpf
(1883,1890).
11. While it has been suggested that Borings historical writings had an agenda (e.g., ODonnell, 1985), I am here relying on his accounts, as my interests do not lie with giving a historically accurate account so much as understanding
how Stevenswho learned his history from Boringsituated his own approach. (In his obituary for Boring, Stevens
professes that Borings History of Experimental Psychology is one of two books that he read twice.)

140

ULJANA FEEST

ements other than sensations.12 The Titchenerian tradition of American psychology later ruled
out images and feelings as basic elements of consciousness, thereby equating consciousness
with sensation (for a more detailed account, see Boring, 1933). The structuralist focus on sensations as basic elements of consciousness came under attack with the rise of Gestalt psychology. Boring attributed to Titchener a theoretical response to this state of affairs, which, however,
Titchener never fully worked out. According to this view, consciousness is to be described in
terms of four basic dimensions (quality, intensity, extensity, and protensity; see Titchener, 1929),
with each sense modality having different attributes that can be classified under these dimensions.13 It was this idea that Boring wanted to argue for in his 1933 book.
Unfortunately, Borings book was a flop and only sold 105 copies in the first 17 years (cf.
Boring, 1961, p. 53). According to Stevens, this lack of success was, in part, due to the fact
that Boring had not been able to provide experimental evidence for his claim that conscious experience has certain attributes, or dimensions. At that point in time, Stevenss work on tonal attributes, which would have demonstrated this point, had not yet been completed: [S]ome of
the research that [Boring] was directing [i.e., Stevenss own research] was soon to clarify the
relation between tonal sensation and its four attributes: pitch, loudness, volume, and density
(Stevens, 1968, p. 597). The relevance of Stevenss research on tonal attributes to Borings
work on dimensions of consciousness can also be gathered from the fact that Boring made this
work a central point of reference in his 1935 article Attributes of Sensation (Boring, 1935).
The relevance of Stevenss operationism to his experimental work on tonal attributes, in turn,
can be inferred from the fact that Stevens devoted a whole section of his paper The
Operational Definition of Psychological Concepts (Stevens, 1935c) to the discussion of tonal
attributes. Thus, my thesis is that the emergence of Stevenss operationism was closely related
to his psychophysical work on tonal attributes, which in turn raised methodological issues
about how to conceptualize and measure such attributes.
Tolmans Notions of Instincts and Instrumental Reasoning
Important cues as to the origins and enabling conditions for Tolmans operationism can be
found in the nature of the intervening variables that he proposed (i.e., demands and cognitions). In this section, I argue that (a) attempts to get a scientific grip on demand concepts go
back to the mid-1920s and can be traced back even farther to his student days, and (b) his answer
as to how to investigate (i.e., operationalize) the referents of such concepts is closely related to
his notion of cognition, which also had an interesting development within his thought.14
Within Tolmans work, we find reference to both of these variables (albeit under different
names) as early as 1925 (e.g., Tolman, 1925/1958h). Tolman believed that any given action is
always going to be determined by both of these types of intervening variables, where the former provides the motivating force for a behavior and the latter enables the organism to represent knowledge about the world. Tolman needed to conceptualize these notions such that they
could be experimentally teased apart (e.g., by holding one constant and observing the effect on
the other, which obviously required some prior assumptions as to what are typical causes and
effects of such internal states). Equally important, however, he needed to come up with some
behavioristically respectable account of why it was permissible to appeal to such mentalistic
12. For example, Klpe tried to find elements of (imageless) thought, and Titchener argued that there were three
classes of mental elements: sensations, images, and feelings (Titchener, 1910).
13. The distinction between dimensions and attributes is not completely clear. Boring admits that [t]he dimensions
of consciousness are the immediate successors to the old attributes of sensation (Boring, 1933, p. 23).
14. The parallel between Tolmans epistemological views and his conceptualization of cognition has also been noted
by Laurence Smith (1986). The present account is inspired by (but not identical with) Smiths analysis.

OPERATIONISM IN PSYCHOLOGY

141

notions at all. This is particularly apparent in his writings about demands. This question, for
him, had two aspects: (1) what criteria could be used when describing a behavior as purposive
(this was related to his notion of molar behavior) and (2) on what grounds might it be permissible to explain a behavior by appeal to an internal motivating state (see Tolman,
1925/1958a; 1926). In his development of how he dealt with these questions, it is important to
recognize the impact of McDougalls notion of instinct. In fact, I believe that Tolmans thinking about demands has to be viewed as an attempt to put McDougalls notion of instinct on
an objective, behavioral footing. Tolman had first come across McDougalls work in a seminar
he took with one of his Harvard professors, Ralph Barton Perry (18761957), while he was a
graduate student there (Tolman, 1952). Perrys own attempts to develop a scientific notion of
value are also significant here (see Perry, 1926). In his articles leading up to this work (Perry,
1918, 1921a, 1921b, to which Tolman, 1920, 1922, appealed), Perry explicitly endorsed both
a doctrine of behaviorism and the notion that behavior has to be scientifically explained by appeal to beliefs and desires. Lastly, one of Tolmans other Harvard mentors, Edwin B. Holt
(18731946), has to be mentioned, whoamong other thingsfocused on the issue of how it
is possible to describe purposive behavior in objective (nonteleological) terms (1915).15
Tolman thought of demands as providing the drives that motivate biological organisms to
engage in instrumental reasoning. His notion of cognition, thus, was one of instrumental reasoning. According to Tolmans theory of purposive behavior, environmental features are cognitively represented in terms of how they can be used or manipulated in order to attain certain
goals. Or, to use Tolmans own terminology, environmental features are represented as meansobjects, which figure in cognitive postulations (or hypotheses) as to what would happen if
the object were to be manipulated in a certain way (Tolman, 1932a). Thus, for Tolman, cognitive representations are always formed relative to certain goals (e.g., the goal to get to the food)
and they involve expectations as to the outcomes of hypothetical actions (e.g., if I were to take
the left lane, this would get me to the food). I believe that we can make out a variety of factors that may have enabled Tolman to move toward this view.
One such factor was Robert Yerkess (18761956) research on primate intelligence. In 1916,
Yerkes published a monograph, The Mental Life of Monkeys and Apes, based on research carried
out in the spring of 1915, in which he reported several experiments that he conducted with orangutans, which (while quite ambiguous) he took to demonstrate their capacity for instrumental reasoning. It is known that Tolman took a class on comparative psychology with Yerkes at Harvard
around 1915.16 Another important intellectual stimulation was German Gestalt psychology.
During his early career, Tolman undertook two visits to Germany (one after his first year of graduate school, in 1911, and one in 1923), where he stayed with Kurt Koffka, though each of these
stays was only brief.17 In his later work, Tolman emphasized the importance of Kurt Lewins notion of objects having a certain valence (see Tolman, 1952, but see also Tolman, 1932a, 1932b),
but also his notion of molar behavior as being more than the sum of physiological processes, as
well as his usage of the expression sign-gestalts, indicates his closeness to certain Gestalt psychological ideas, though he took this latter notion to be broader than theirs, in that it emphasized
the relationship between perception and action. For example, a chair might be mentally represented as that chair, if placed against the wall, can be stood upon like a stepladder to reach this
15. See Feest (in press) for a more detailed account of the impact of Perry and Holt on Tolman, and a discussion of
how Tolmans treatment of drives was located in the instinct debate of the 1920s.
16. It may also be speculated that Wolfgang Khlers research on primate intelligence, conducted on Tenerife around
the same time, may have had an indirect impact on Tolman, as Khler and Yerkes were corresponding (Ash, 1995;
Kohler, 1917).
17. I owe this information to Nancy Innis.

142

ULJANA FEEST

picture (Tolman, 1933/1958e, p. 80). Two other important figures to be mentioned here are Egon
Brunswik (19031955) and David Krech (19091977). Tolman met Brunswik while on sabbatical in Vienna in 1933/1934 (after the publication of his Purposive Behavior, but before the publication of his papers on operationism), and they published a paper together (Tolman &
Brunswik, 1935). According to Tolmans own account, Brunswik gave me new insight into the
essentially achievement character of behavior (Tolman, 1952). Apart from significant differences between the two, it seems to me that Brunswiks impact consisted of underscoring an element that was already present in Tolmans thinkingi.e., the notion that all behavior is based
on hypotheses about the nature of the environment, which may or may not be correct. The major
new insight that Tolman picked up here (which had, of course, been implied by his previous
work) was the notion that all knowledge is fallible. Tolman recognized that this idea was similar to his own notion that behavior is guided by sign-gestalts. Since sign-gestalts were essentially hypotheses about the outcomes of hypothetical manipulations, there was always a chance
that they might not lead to the desired outcomes (i.e., that they were mistaken). In their joint
publication, Tolman and Brunswik spelled out these basic ideas in more detail, essentially stating that both with respect to perception as well as action, the organism faces a degree of uncertainty, which the organism copes with by trying to come up with hypotheses that maximize the
probability of success. The notion that sign-gestalts are hypotheses had already begun to appear
in the context of Tolmans experimental work shortly before his encounter with Brunswik. In
particular, it was tied to work that he conducted with his student Isodore Krechevsky
(Krechevsky, 1932a, 1932b; Tolman & Krechevsky, 1933).18 This (and more) work was later
summarized in Tolmans article on cognitive maps in rats (Tolman, 1948).
With this overview, I hope to have given some plausibility to my thesis that Tolmans operationism was an attempt to legitimate his talk about intervening variables, and that his views
on how to conceptualize these variables went back to his student days. Furthermore, I think that
his views about cognitive representations in rats (conceived as hypotheses as to what would
happen if the object in question were manipulated) were similar to his own practice of operationalizing conceptsi.e., as referring to hypothetically posited objects, which, if experimentally manipulated in a certain way, are expected to behave a certain way (thereby confirming
hypotheses about the existence of such objects).
The Impact of Positivism
The above outlines of the origins of Stevenss and Tolmans operatonisms are not intended
to be exhaustive. For example, I have neglected to mention a variety of factors that may have
contributed to my main actors intellectual development.19 Furthermore, by pointing to the contexts in which certain methodologies originated I have not thereby provided an analysis of why
they were adopted. For example, while claiming that the basic tenets of these two varieties of
operationism were similar, I have only hinted at possible explanations for why these positions
ended up looking so similar (namely, by reference to the methodological distrust of traditional
introspectivism and mentalism, which both of these scientists had adopted from behaviorism),
despite the very different origins and research interests of their proponents.
While my historical outlines were supposed to show (among other things) that these operationisms were already being contemplated before the scientists in question encountered repre18. Today, Krechevsky is better known by the name David Krech, which he adopted in 1943 (the year he got married) in order to spare his son the anti-Semitism he had encountered throughout his professional life (Krech, 1974).
19. For example, Stevenss religious background as a Mormon, or his involvement with the American Acoustical
Society, or Tolmans friendships with the pragmatist philosophers, C. I. Lewis and S. Pepper.

OPERATIONISM IN PSYCHOLOGY

143

sentatives of logical positivism, this is not to deny that both Stevens and Tolman did indeed encounter representatives of logical positivism, and that this had an impact on how they formulated
their operationisms. I believe that those references were largely rhetorical, aimed at backing up
their views by appealing to cutting-edge philosophy of science. Let us begin with Stevens.
Herbert Feigl, a member of the Vienna Circle, was at Harvard in 1930, but this was before
Stevenss time. While it has been asserted that Feigl brought Bridgmans operationism to the attention of Harvard psychologists (see Green, 1992; Moyer, 1991), this is contradicted by Borings
own retrospective account, according to which both he and Stevens were unaware of Bridgmans
operationism when they were working on the Consciousness book (Boring, 1961). At any rate, if
Feigl (via Boring) left any traces in Stevenss thinking, this is not reflected in Stevenss early notebooks of 1932 and 1933. Hardcastle (1995) has made the case that Stevenss knowledge of
German was limited and that he was therefore unlikely to have been exposed to logical positivism
before Carnaps 1934 contribution to the journal Philosophy of Science (Carnap, 1934). However,
by the mid- to late 1930s, Stevens did know positivist philosophers of science, got involved with
the Unity of Science movement in the late 1930s, and explicitly related his operationism to this
movement in his 1936 paper.20 But this strikes me as an ex post adoption of terminology, rather
than reflecting a genuine impact on the content of his own operationism.
With respect to Tolman, I would like to make a similar point. As I showed in the case
study, the essential components of Tolmans operationism were already implicit in his writings from at least the mid-1920s onward. The remaining question is what prompted him to
articulate this position as an explicit doctrine in the mid- and late 1930s. An obvious answer,
which Tolman confirmed in his 1937 paper, is that in the 1930s he came across S. S.
Stevenss and Bridgmans formulations of operationism. Another obvious suggestion is that
while in Vienna, Tolman got acquainted with members of the Vienna circle, and possibly
read Carnaps Psychology in Physical Language (Carnap, 1932). There is good reason to
believe that Tolman attended meetings of the Vienna Circle, based on his friendship with
Brunswik, as well as the fact that Schlick had been a visitor at the Philosophy Department
at Berkeley a couple of years earlier. Also, Tolman was approached to present a paper at the
1936 Unity of Science congress, but was unable to attend (the paper he had planned to present was his 1937 paper on operationism). However, I would like to make the case that he was
only adopting a label for a position that was already fairly consolidated and that differed
from the kind of position that is usually associated with positivism.
EPISTEMOLOGICAL PROBLEMS WITH OPERATIONISM
The aim of my historical case studies was to argue that operationism in psychology was
not identical with, nor was it significantly influenced by, an empiricist epistemology or a verificationist theory of meaning. According to this analysis, both scientists in question wanted to
legitimate their research on mentalistic topics by providing empirical criteria of application
for their concepts.21 These empirical criteria of application were definitions insofar as they
specified features that these authors took to be indicative of their purported subject matter
thereby enabling them to do research on that subject matterbut they were not intended to exhaustively provide the meanings of the concepts in question, or to be unrevisable. However,
20. Stevens presented a paper, entitled On the Problem of Scales for the Measurement of Psychological Attitudes,
at the Fifth International Congress for the Unity of Science, which was held at Harvard University in September
1939 (see Stadler, 1997, p. 431).
21. I do not claim that this motivation can be generalized to other operationists. As was remarked by an anonymous
referee, my interpretation might prove to be problematic when it comes to Skinners operationism.

144

ULJANA FEEST

while my analysis may imply that psychological operationism cannot have inherited the specific problems of empiricism and verificationism, it does not imply that it does not have problems of its own. The problem I want to discuss here is that of circularity.
The issue is that it would be circular to justify the validity of a concept by reference
to the operations that led to its introduction.22 This was one of the problems that behaviorists had made out in traditional mentalism. As argued above, operationism was, in part,
an attempt to deal with this problem. So, if operationism cannot avoid circularity (as has
been suggested by critics), it would have failed to meet this goal (this problem is also
pointed out by Green, 1992, p. 300). One possible way of avoiding circularity would be to
provide evidence for a given thing (e.g., Stevenss tonal density and volume or Tolmans
demand) by showing that it can be detected by different means (i.e., different experimental operations). If I am correct in my contention that both Stevens and Tolman were realists about the referents of their concepts, then it is plausible to assume that they would
have believed this to be permissible. Indeed, by the mid-1940s, this was explicitly allowed
by proponents of operationism (Boring, 1945). But none of the early operationists were
able to provide criteria that would settle how to determine that two operations do indeed
get at the same object (i.e., operationalize the same concept). I believe that part of the
appeal of operationism was its cautionary aspect, in that it advised scientists not simply to
assume that one and the same concept applies under very different circumstances (i.e.,
when different operations are used).23
A related problem is thatgiven the assumption that the cognitive system is very complex, and given all the assumptions that go into an experimental setupone may be skeptical
of whether a particular experiment really individuates any one cognitive process in pure, uncontaminated form (or at all, for that matter). In other words, did Tolman really have grounds
for assuming that his defining experiments on demands, even if such a cognitive entity really exists, actually reflected demands, rather than being the output of a complex interaction
of any number of cognitive processes that are required when the rat performs in a maze? As
described by Grace (2001), both of these problems were addressed by subsequent methodological developments within psychology. On the one hand, the convergent/discriminant operations (or multitrait-multimethod) approach suggested correlating the results of different
tests of the same purported phenomenon, and of similar tests about different purported phenomena (Campbell & Fiske, 1959). On the other hand, the converging operations approach
suggested designing experiments that would specifically test alternative hypotheses to account for the data (Garner et al., 1956).
I do not intend to go into the details of these methods or whether they successfully deal
with the problems just outlined, let alone whether psychologists today can generally be said
to apply them correctly. Rather, the point of this section was to isolate what I take to be a
problem that critics of operationism might legitimately be concerned about. However, as
mentioned earlier, in response to Graces article, several authors suggested that there is
something deeply wrong with operationism, which cannot be fixed by superficial method-

22. By the notion of validity, psychologists mean that a concept or test refers to what it is thought to refer to. This
is different from the notion of validity in logic.
23. For example, Operationism is the manner in which the present generation utters the familiar cry of science, be
careful (Pratt, 1939, p. 81). Even Bridgman later claimed that all he had intended to say in his 1927 book was that
[operationism] is a technique of analysis which endeavors to attain the greatest possible awareness of everything
involved in a situation by bringing out in the light of day all our activity or operations when confronted with the situation (Bridgman, 1938, p. 130).

OPERATIONISM IN PSYCHOLOGY

145

ological innovations. In the following section, I will discuss the questions of what this debate might be about.
WHAT IS THE DEBATE ABOUTWHAT SHOULD IT BE ABOUT?
Let us start with the charge that operationismcontrary to its own assertionsis not
metaphysically neutral (Bickhard, 2001; Green, 1992, p. 314). I believe that this charge is directed at the positivist rejection of metaphysics, and the idea of a neutral observation language.
Once again, it might be helpful to keep in mind the difference between the positivist and the
operationist project. Positivists attempted to formulate an epistemology that was metaphysically neutral (e.g., Carnap, 1931). Operationists (on my construal) attempted to formulate a
methodology for an empirical psychology, which was sensitive to problems with introspectivism and mentalism, and which forced psychologists to explicate their prior assumptions
about how to interpret observations. Kendler (1981) also emphasizes this point when he argues
that when there is a disagreement between empirical psychologists about a given phenomenon,
it may turn out that the disagreement is not about the empirical results, but about how to operationalize the concept.
Let us turn to a second, related critique of operationismthat it implies a particular notion of what it means to be scientific, which in turn has an impact on how psychological phenomena are conceptualized and what kinds of psychological phenomena can be scientifically
investigated at all. According to this critique, the charge is not merely that operationism is not
metaphysically neutral, but that it promotes a particular metaphysical picture. One version of
this charge is that operationism is intricately tied to quantitative and experimental methods
i.e., an understanding of psychology as a natural science. For example, Leahey (1983) argues
that a naturalistically conceived psychology (which operationismaccording to himstands
for) cannot account for the phenomenon of intentionality. What I take him to be saying here is
that operationism is incapable of employing qualitative or hermeneutic methods, which would
do more justice to the psychological subject matter. The first of these two assertions does not
strike me as self-evidently true, but it certainly warrants further discussion. More importantly,
though, I would like to argue that operationism is not committed to any particular notion of the
subject matter of psychology. While it is true that Stevens and Tolman (as we have seen) were
indeed proponents of a quantitative and experimental method in psychology, I believe that this
coincidence is historically contingent. To put it quite bluntly, all empirical psychologists have
to operationalize their concepts. And all empirical psychologists have to argue for their results
by laying open both their conceptual presuppositions and their empirical data. Viewed this way,
the issue seems to be whether psychology can be an empirical science at all.
One last critique of operationism is that psychologists frequently confuse definitions and
factsi.e., they think that operational definitions can substitute for theoretical work about the
referent of the concept. In this vein, Christopher Green quite appropriately points out that although operational definitions might have a role to play in piloting nascent thought about a
given phenomenon, they cannot ultimately replace the fruits of hard, rigorous thought (Green,
1992, p. 315). While agreeing with this assessment, I believe that he is expressing a critique of
certain widespread research practices, not of operationism per se.24

24. Gerd Gigerenzer has suggested to me that the proliferation of operationally defined empirical concepts at the expense of theoretical work might be due to funding policies that reward empirical work more than theoretical work
and provide an incentive for psychologists to inflate the significance of a given empirical effect. Following up this
suggestion might itself be an interesting historical and sociological project.

146

ULJANA FEEST

Summing up this section, I believe the debate mistakenly focuses on the tenability of operationism, where in fact it should be about the question of preconditions for psychology as an
empirical science and about the question of whether psychologists exhibit the requisite
thoughtfulness in their research. Insofar as there is agreement that psychology can be an empirical science, the debate should then be about what are adequate concepts and how (not
whether) to operationalize them, and how (not whether) to validate them. This task is far from
trivial, and it may turn out that the gaps between different (e.g., qualitative vs. quantitative) approaches within psychology cannot be bridged. However, I believe that the gap, ultimately, is
not one between proponents and opponents of operationism.
CONCLUSION
In this article, I have provided a conceptual analysis of the notion of operationism in
psychology. This analysis was based on a historical investigation of the operationisms of
two early proponents, S. S. Stevens and E. C. Tolman. Two construals of operationism were
introduced, a positivist and a methodological reading. It was argued that while critics
of operationism frequently assume a positivist reading of this position, an analysis of the
ways in which operationism was applied in Stevenss and Tolmans research suggests a
methodological reading, according to which operational definitions have a pragmatic role
to play in psychological research, but are not taken to be necessarily true or unrevisable.
This analysis was further backed up by a historical study of the origins of both Stevenss
and Tolmans operationisms, which revealed that neither one of them had any significant
contact with proponents of positivism in philosophy, until the substance of their position
was already in place. I then presented an analysis of some epistemological problems with
operationism, and the ways in which methodological writings in the 1950s tried to address
these problems. Lastly, I claimed that the current debate about operationism is really about
a couple of deeper points of contention, and I suggested that these should be addressed independently of the issue of operationism.
ACKNOWLEDGMENTS
I would like to thank Bob Olby who first encouraged me to pursue this project in a
graduate class I took with him in 1998. I also thank the members of my dissertation committee, especially my advisor, Peter Machamer, and my friend Stephanie Koerner. Thanks
to audiences at the 2003 ESHHS-meeting and at the Max Planck Institute for the History
of Science as well as to two anonymous referees for helpful remarks. I am grateful to Nancy
Innis for sharing information about Tolman with me. Last but not least, many thanks to Ray
Fancher and to the members of the Wiley production department.
REFERENCES
Amundsen, R. (1983). E. C. Tolman and the intervening variable: A study in the epistemological history of psychology. Philosophy of Science, 50, 268282.
Ash, M. (1995). Gestalt psychology and German culture, 18901967. Cambridge, UK: Cambridge University Press.
Bergmann, G., & Spence, K. (1941). Operationism and theory in psychology. Psychological Review, 48, 114.
Bickhard, M. (2001). The tragedy of operationalism. Theory and Psychology, 2(1), 3544.
Boring, E. (1921). The stimulus error. American Journal of Psychology, 32, 449471.
Boring, E. (1923). Intelligence as the test tests it. New Republic, 6, 35.
Boring, E. (1929). A history of experimental psychology. New York and London: D. Appleton-Century Co.
Boring, E. (1933). The physical dimensions of consciousness. New York and London: The Century Co.

OPERATIONISM IN PSYCHOLOGY

147

Boring, E. (1935). The relation of the attributes of sensation to the dimensions of the stimulus. Philosophy of Science,
2, 236245.
Boring, E. (1942). Sensation and perception in the history of experimental psychology. New York: Appleton-CenturyCrofts.
Boring, E. (1945). The use of operational definition in science. Psychological Review, 52(5), 243245.
Boring, E. (1952). Edwin Garrigues Boring. In E. Boring, H. Langfeld, H. Werner, & R. Yerkes (Eds.), A history of
psychology in autobiography (Vol. IV, pp. 2752). Worcester, MA: Clark University Press.
Boring, E. (1961). Psychologist at large. In E. Boring, Psychologist at large. An autobiography and selected essays
(pp. 383). New York: Basic Books.
Boring, E., Bridgman, P., Feigl, H., Israel, H., Pratt, C., & Skinner, B. (1945). Symposium on operationism.
Psychological Review, 52, 241294.
Bridgman, P. (1927). The logic of modern physics. New York: Macmillan.
Bridgman, P. (1938). Operational analysis. Philosophy of Science, 5, 114131.
Campbell, D. T., & Fiske, D. W. (1959). Convergent and discriminant validation by the multitrait-multimethod matrix. Psychological Bulletin, 56, 81105.
Carnap, R. (1932). Psychology in physical language. Erkenntnis, 3, 107142. (Reprinted in A. Ayer (Ed.), Logical
positivism (pp. 165198). Glencoe, IL: Free Press.)
Carnap, R. (1934). On the character of philosophical problems. Philosophy of Science, 1, 519.
Carnap, R. (1936/1937). Testability and meaning. Philosophy of Science, 3(4), 419471; 4(1), 140.
Carnap, R. (1956). The methodological character of theoretical concepts. In H. Feigl & M. Scriven (Eds.), Minnesota
studies in the philosophy of science and the concepts of psychology and psychoanalysis, Vol. I., pp. 3876.
Minnesota: University of Minnesota Press.
Carnap, R. (1931). berwindung der Metaphysik durch logische Analyse der Sprache. In Erkenntnis 2, 219241.
Danziger, K. (1990). Constructing the subject: Historical origins of psychological research. Cambridge, UK:
Cambridge University Press.
Dingler, H. (1988). Method instead of epistemology and philosophy of science. Science in Context, 2, 369408.
(Original work published 1936)
Fechner, G. (1860). Elemente der Psychophysik. Leipzig: Breitkopf & Hrtel.
Feest, U. (2002). Die Konstruktion von Bewutsein: Eine psychologiehistorische Fallstudie. In C. Zittel (Ed.), Wissen
und soziale Konstruktion (pp. 129153). Berlin: Akademie Verlag.
Feest, U. (in press). Giving up instincts in psychologyor not? In B. Gmez-Ziga & A. Mlberger (Eds.), Recent
contributions to the history of the human sciences. Munich: Profil-Verlag.
Frank, P. (Ed.). (1956). The validation of scientific theories. Boston: The Beacon Press.
Friedman, M. (1999). Reconsidering logical positivism. Cambridge, UK: Cambridge University Press.
Garner, W. R., Hake, H. W., & Erikson, C. W. (1956). Operationism and the concept of perception. Psychological
Review, 63, 149159.
Grace, R. (2001). On the failure of operationism. Theory and Psychology, 11(1), 533.
Green, C. (1992). Of immortal mythological beasts: Operationism in psychology. Theory & Psychology, 2, 291320.
Green, C. (2001). Operationism again: What did Bridgman say? What did Bridgman need? Theory and Psychology,
11(1), 4551.
Hanson, R. (1958). Patterns of discovery: An inquiry into the conceptual foundations of science. Cambridge, UK:
Cambridge University Press.
Hardcastle, G. (1995). S. S. Stevens and the origins of operationism. Philosophy of Science, 62, 404424.
Helmholtz, H. (1863). Lehre von den Tonempfindungen. Braunschweig: F. Vieweg und Sohn.
Hempel, C. (1950). Problems and changes in the empiricist criterion of meaning. Revue International de Philosophie,
11, 4163.
Hempel, C. (1952). Fundamentals of concept formation in empirical science. International Encyclopedia of Unified
Science, Vol. II, No. 7.
Hempel, C. (1954). A logical appraisal of operationism. Scientific Monthly, 79, 215220.
Hempel, C. (1965). Fundamentals of taxonomy. In Aspects of scientific explanation and other essays in the philosophy of science (pp. 137154). New York: Free Press
Holt, E. (1915). The Freudian wish and its place in ethics. New York: Henry Holt.
Israel, H. (1945). Two difficulties in operational thinking. Psychological Review, 50, 273291.
Israel, H., & Goldstein, B. (1944). Operationism in psychology. Psychological Review, 51, 177188.
Kendler, H. (1981). The reality of operationism: A rejoinder. The Journal of Mind and Behavior, 2, 331341.
Kendler, H. (1983). Operationism: A recipe for reducing confusion and ambiguity. The Journal of Mind and
Behavior, 4, 9197.
Koch, S. (1992). Psychologys Bridgman and Bridgmans Bridgman. An essay in reconstruction. Theory and
Psychology, 2(3), 261290.
Khler, W. (1917). Intelligenzprfungen an Anthropoiden. Abhandlungen der Kniglich-Preussischen Akademie der
Wissenschaften, phys.-math., Kl. Nr. 1.
Krech, D. (formerly I. Krechevsky). (1974). David Krech. In G. Lindzey (Ed.), A history of psychology in autobiography (Vol. VI., pp. 221250). Englewood Cliffs, NJ: Prentice Hall.

148

ULJANA FEEST

Krechevsky, I. (1932a). Hypothesis versus chance in the pre-solution period in sensory discrimination learning.
University of California Publications in Psychology, 6, 2744.
Krechevsky, I. (1932b). The genesis of hypothesis in rats. University of California Publications in Psychology, 6,
4564.
Klpe, O. (1893). Grundriss der Psychologie. Nrnberg: Engelman. (Outlines of Psychology, Based on the Results
of Experimental Investigation. E. Titchener, trans. New York: Macmillan).
Leahey, T. (1980). The myth of operationism. The Journal of Mind and Behavior, 1, 131140.
Leahey, T. (1981). Operationism still isnt real: A temporary reply to Kendler. The Journal of Mind and Behavior, 2,
343348.
Leahey, T. (1983). Operationism and ideology: Reply to Kendler. The Journal of Mind and Behavior, 4, 8190.
MacCorquodale, K., & Meehl, P. (1948). On a distinction between hypothetical constructs and intervening variables.
Psychological Review, 55, 95107.
McDougall, W. (1914). An introduction to social psychology (8th ed.). London: Menuhen.
Michell, J. (1999). Measurement in psychology. A critical history of a methodological concept. Cambridge, UK:
Cambridge University Press.
Moyer, A. (1991). Bridgmans operational perspective on physics. Part I: Origins and developments. Studies in
History and Philosophy of Science, 22, 373397.
ODonnell, J. (1985). The origins of behaviorism in American psychology, 18701920. New York and London: New
York University Press.
Pennington, L., & Finan, J. (1940). Operational usage in psychology. Psychological Review, 47, 254266.
Perry, R. (1918). Docility and purposiveness. Psychological Review, 25, 120.
Perry, R. (1921a). A behavioristic view of purpose. Journal of Philosophy, 18, 85105.
Perry, R. (1921b). The independent universality of purpose and belief. Journal of Philosophy, 18, 169180.
Perry, R. (1926). General theory of value. Its meaning and basic principles construed in terms of interest. New York:
Longman, Green & Co.
Pratt, C. (1939). The logic of modern psychology. New York: MacMillan.
Quine, W. O. (1951). Two dogmas of empiricism. Philosophical Review, 60, 2043.
Rheinberger, H.-J. (1997). Toward a history of epistemic things: Synthesizing proteins in the test tube. Stanford, CA:
Stanford University Press.
Rogers, T. (1989). Operationism in psychology: A discussion of contextual antecedents and a historical interpretation of its longevity. Journal of the History of the Behavioral Sciences, 25, 139150.
Rogers, T. (2001). On the importance of considering investigative practices in attempting to understand operationism.
Theory & Psychology, 11(1), 5966.
Salmon, W. C. (1985). Empiricism: The key question. In N. Rescher (Ed.), The heritage of logical positivism (pp.
121). Lanham, MD: University Press of America.
Smith, L. (1986). Behaviorism and logical positivism. A reassessment of their alliance. Stanford, CA: Stanford
University Press.
Smith, R. (1997). The Norton history of the human sciences. New York and London: W. W. Norton & Company.
Stadler, M. (1997). Studien zum Wiener Kreis. Ursprung, Entwicklung und Wirkung des Logischen Empirismus im
Kontext. Frankfurt: Suhrkamp.
Stevens, S. S. (1934a). The volume and intensity of tones. American Journal of Psychology, 46, 397408.
Stevens, S. S. (1934b). Tonal density. Journal of Experimental Psychology, 17, 585592.
Stevens, S. S. (1934c). The attributes of tones. Proceedings of the National Academy of Sciences, 20, 457459.
Stevens, S. S. (1935a). The relation of pitch to intensity. Journal of the Acoustical Society of America, 6, 150154.
Stevens, S. S. (1935b). The operational basis of psychology. American Journal of Psychology, 47, 323330.
Stevens, S. S. (1935c). The operational definition of psychological concepts. Psychological Review, 42, 517552.
Stevens, S. S. (1936). Psychology, the propaedeutic science. Philosophy of Science, 3, 90103.
Stevens, S. S. (1939a). Psychology and the science of science. Psychological Bulletin, 36, 221263.
Stevens, S. S. (1968). Edwin Garrigues Boring: 18861968. American Journal of Psychology, 81, 589606.
Stevens, S. S., & Davis, H. (1936). Physiological acoustics: Pitch and loudness. Journal of the Acoustical Society of
America, 8, 113.
Stevens, S. S., & Davis, H. (1938). Hearing: Its psychology and physiology. New York: Wiley.
Stumpf, C. (1883). Tonpsychologie, Vol. I. Leipzig: Hirtzel.
Stumpf, C. (1890). Tonpsychologie, Vol. II. Leipzig: Hirtzel.
Suppe, F. (1977). The structure of scientific theories (2nd ed.). Urbana, IL: University of Illinois Press.
Titchener, E. (1910). A text-book of psychology. New York: Macmillan.
Titchener, E. (1929). Systematic psychology: Prolegomena. New York: Macmillan.
Tolman, E. (1920). Instinct and purpose. Psychological Review, 27, 217233.
Tolman, E. (1922). Can instincts be given up in psychology? Journal of Abnormal Psychology and Social
Psychology, 17, 139152.
Tolman, E. (1926). The fundamental drives. Journal of Abnormal Psychology, 20, 349358.
Tolman, E. (1932a). Purposive behavior in animals and men. New York: Century.
Tolman, E. (1932b). Lewins concept of vectors. Journal of General Psychology, 7, 3-15.
Tolman, E. (1937). An operational analysis of demand. Erkenntnis, 6, 383392.

OPERATIONISM IN PSYCHOLOGY

149

Tolman, E. (1948). Cognitive maps in rats and men. Psychological Review, 55(4), 189208.
Tolman, E. (1952). Edward Chace Tolman. In E. Boring, H. Werner, H. Langfeld, & R. Yerkes (Eds.), A history of
psychology in autobiography (Vol. 4, pp. 323339). Worcester, MA: Clark University Press.
Tolman, E. (1958a). Behaviorism and purpose. The Journal of Philosophy, Reprinted in Behavior and psychological
man (pp. 3237). Berkeley and Los Angeles, CA: University of California Press. (Original work published
1925)
Tolman, E. (1958b). A behaviorists definition of consciousness. Psychological Review. Reprinted in Behavior and
psychological man (pp. 6368). Berkeley and Los Angeles, CA: University of California Press. (Original work
published 1927)
Tolman, E. (1958c). A behavioristic account of the emotions. Psychological Review. Reprinted in Behavior and psychological man (pp. 2331). Berkeley and Los Angeles, CA: University of California Press. (Original work published 1923)
Tolman, E. (1958d). The determiners of behavior at a choice point. Reprinted in Behavior and psychological man
(pp. 144178). Berkeley and Los Angeles, CA: University of California Press. (Original work published 1938)
Tolman, E. (1958e). Gestalt and sign gestalt. Psychological Review. Reprinted in Behavior and psychological man
(pp. 7793). Berkeley and Los Angeles, CA: University of California Press. (Original work published 1933)
Tolman, E. (1958f). Operational behaviorism and current trends in psychology. Reprinted in Behavior and psychological
man (pp. 115129). Berkeley and Los Angeles, CA: University of California Press. (Original work published 1936)
Tolman, E. (1958g). Psychology and immediate experience. Philosophy of Science. Reprinted in Behavior and psychological man (pp. 94114). Berkeley and Los Angeles, CA: University of California Press. (Original work
published 1935)
Tolman, E. (1958h). Purpose and cognition: The determiners of animal learning. Psychological Review. Reprinted in
Behavior and psychological man (pp. 3847). Berkeley and Los Angeles, CA: University of California Press.
(Original work published 1925)
Tolman, E., & Brunswik, E. (1935). The organism and the causal texture of the environment. Psychological Review,
42, 4377.
Tolman, E., & Krechevsky, I. (1933). Means-end-readiness and hypothesisA contribution to comparative psychology. Psychological Review, 40, 6070.
Waters, R., & Pennington, A. (1938). Operationism in psychology. Psychological Review, 45, 414423.
Weber, C. (1940). Valid and invalid conceptions of operationism in psychology. Psychological Review, 49, 5468.
Wundt, W. (1893). Grundzge der Physiologischen Psychologie (Vol. 1; 4th ed.). Leipzig: Verlag von Wilhelm
Engelmann.
Yerkes, R. (1916). The mental life of monkeys and apes. Behavior Monographs, 3(1).

Vous aimerez peut-être aussi