THE BEHAVIORAL AND BRAIN SCIENCES (1984) 7, 547-581

Printed in the United States of America

The operational analysis of

psychological terms
B. F. Skinner
Department of Psychology and Social Relations, Harvard University,
Cambridge, Mass. 02138

Abstract: The major contributions of operationism have been negative, largely because operationists failed to distinguish logical
theories of reference from empirical accounts of language. Behaviorism never finished an adequate formulation of verbal reports and
therefore could not convincingly embrace subjective terms. But verbal responses to private stimuli can arise as social products
through the contingencies of reinforcement arranged by verbal communities.
In analyzing traditional psychological terms, we need to know their stimulus conditions ("finding the referent"), and why each
response is controlled by that condition. Consistent reinforcement of verbal responses in the presence of stimuli presupposes stimuli
acting upon both the speaker and the reinforcing community, but subjective terms, which apparently are responses to private
stimuli, lack this characteristic. Private stimuli are physical, but we cannot account for these verbal responses by pointing to
controlling stimuli, and we have not shown how verbal communities can establish and maintain the necessary consistency of
reinforcement contingencies.
Verbal responses to private stimuli may be maintained through appropriate reinforcement based on public accompaniments, or
through reinforcements accorded responses made to public stimuli, with private cases then occurring by generalization. These
contingencies help us understand why private terms have never formed a stable and uniform vocabulary: It is impossible to establish
rigorous vocabularies of private stimuli for public use, because differential reinforcement cannot be made contingent upon the
property of privacy. The language of private events is anchored in the public practices of the verbal community, which make
individuals aware only by differentially reinforcing their verbal responses with respect to their own bodies. The treatment of verbal
behavior in terms of such functional relations between verbal responses and stimuli provides a radical behaviorist alternative to the
operationism of methodological behaviorists.
Keywords: awareness; behavior, verbal; behaviorism, methodological; behaviorism, radical; operationism; philosophy of psychology;
private events; reference; semantics; subjectivity-objectivity; verbal community

Operationism may be defined as the practice of talking the corresponding set of operations" cannot be taken
about (1) one's observations, (2) the manipulative and literally, and no similarly explicit but satisfactory state-
calculational procedures involved in making them, (3) the ment of the relation is available. Instead, a few round-
logical and mathematical steps which intervene between about expressions recur with rather tiresome regularity
earlier and later statements, and (4) nothing else. So far, whenever this relation is mentioned: We are told that a
the major contribution has come from the fourth provi- concept is to be defined"in terms of certain operations,
sion and, like it, is negative. We have learned how to that propositions are to be "based upon" operations, that
avoid troublesome references by showing that they are a term denotes something only when there are "concrete
artifacts which may be variously traced to history, philos- criteria for its applicability," that operationism consists
ophy, linguistics, and so on. No very important positive in "referring any concept for its definition to . . . con-
advances have been made in connection with the first crete operations," and so on. We may accept expressions
three provisions because operationism has no good defi- of this sort as outlining a program, but they do not provide
nition of a definition, operational or otherwise. It has not a general scheme of definition, much less an explicit
developed a satisfactory formulation of the verbal behav- statement of the relation between concept and operation.
ior of the scientist. The weakness of current theories of language may be
Operationists, like most contemporary writers in the traced to the fact that an objective conception of human
field of linguistic and semantic analysis, are on the fence behavior is still incomplete. The doctrine that words are
between logical "correspondence" theories of reference used to express or convey meanings merely substitutes
and empirical formulations of language in use. They have "meaning" for "idea" (in the hope that meanings can then
not improved upon the mixture of logical and popular somehow be got outside the skin) and is incompatible
terms usually encountered in casual or even supposedly with modern psychological conceptions of the organism.
technical discussions of scientific method or the theory of Attempts to derive a symbolic function from the principle
knowledge (e.g. Bertrand Russell's An Inquiry into of conditioning (or association) have been characterized
Meaning and Truth, 1940). Definition is a key term but is by a very superficial analysis. It is simply not true that an
not rigorously defined. Bridgman's (1928; see also 1945) organism reacts to a sign "as it would to the object which
original contention that the "concept is synonymous with the sign supplants" (Stevens 1939). Only in a very limited

© 7984 Cambridge University Press O14O-525X/84IO4O547-35I$O6.OO 547

Skinner: Psychological terms
area (mainly that of autonomic responses) is it possible to language from society, but the reinforcing action of the
regard a sign as a simple substitute stimulus in the verbal community continues to play an important role in
Pavlovian sense. Modern logic, as a formalization of maintaining the specific relations between responses and
"real" languages, retains and extends this dualistic theory stimuli which are essential to the proper functioning of
of meaning and can scarcely be appealed to by the verbal behavior. How language is acquired is, therefore,
psychologist who recognizes his own responsibility in only part of a much broader problem.
giving an account of verbal behavior. We may generalize the conditions responsible for the
The operational attitude, in spite of its shortcomings, is standard "semantic" relation between a verbal response
a good thing in any science, but especially in psychology and a particular stimulus without going into reinforce-
because of the presence there of a vast vocabulary of ment theory in detail. There are three important terms: a
ancient and nonscientific origin. It is not surprising that stimulus, a response, and a reinforcement supplied by
the broad empirical movement in the philosophy of the verbal community. (All of these need more careful
science, which Stevens has shown to be the background definition than are implied by current usage, but the
of operationism, should have had a vigorous and early following argument may be made without digressing for
representation in the field of psychology - namely, be- that purpose.) The significant interrelations between
haviorism. In spite of the differences which Stevens these terms may be expressed by saying that the commu-
claimed to find, behaviorism has been (at least to most nity reinforces the response only when it is emitted in the
behaviorists) nothing more than a thoroughgoing opera- presence of the stimulus. The reinforcement of the re-
tional analysis of traditional mentalistic concepts. We sponse "red," for example, is contingent upon the pres-
may disagree with some of the answers (such as Watson's ence of a red object. (The contingency need not be
disposition of images), but the questions asked by behav- invariable.) A red object then becomes a discriminative
iorism were strictly operational in spirit. I also cannot stimulus, an "occasion" for the successful emission of the
agree with Stevens that American behaviorism was response "red."
"primitive." The early papers on the problem of con- This scheme presupposes that the stimulus act upon
sciousness by Watson, Weiss, Tolman, Hunter, Lashley, both the speaker and the reinforcing community; other-
and many others, were not only highly sophisticated wise the proper contingency cannot be maintained by the
examples of operational inquiry, they showed a willing- community. But this provision is lacking in the case of
ness to deal with a wider range of phenomena than do many "subjective" terms, which appear to be responses
current streamlined treatments, particularly those of- to private stimuli. The problem of subjective terms does
fered by logicians (e.g. Carnap 1934) interested in a not coincide exactly with that of private stimuli, but there
unified scientific vocabulary. But behaviorism, too, is a close connection. We must know the characteristics of
stopped short of a decisive positive contribution - and for verbal responses to private stimuli in order to approach
the same reason: It never finished an acceptable formula- the operational analysis of the subjective term.
tion of the "verbal report." The conception of behavior The response "My tooth aches" is partly under the
which it developed could not convincingly embrace the control of a state of affairs to which the speaker alone is
"use of subjective terms." able to react, since no one else can establish the required
A considerable advantage is gained from dealing with connection with the tooth in question. There is nothing
terms, concepts, constructs, and so on, quite frankly in mysterious or metaphysical about this; the simple fact is
the form in which they are observed - namely, as verbal that each speaker possesses a small but important private
responses. There is then no danger of including in the world of stimuli. So far as we know, responses to that
concept the aspect or part of nature which it singles out. world are like responses to external events. Nevertheless
One may often avoid that mistake by substituting term for the privacy gives rise to two problems. The first difficulty
concept or construct. Meanings, contents, and references is that we cannot, as in the case of public stimuli, account
are to be found among the determiners, not among the for the verbal response by pointing to a controlling
properties, of response. The question, What is length? stimulus. Our practice is to infer the private event, but
would appear to be satisfactorily answered by listing the this is opposed to the direction of inquiry in a science of
circumstances under which the response "length" is behavior in which we are to predict a response through,
emitted (or, better, by giving some general description of among other things, an independent knowledge of the
such circumstances). If two quite separate sets of circum- stimulus. It is often supposed that a solution is to be found
stances are revealed, then there are two responses having in improved physiological techniques. Whenever it be-
the form "length," since a verbal response class is not comes possible to say what conditions within the orga-
defined by phonetic form alone but by its functional nism control the response "I am depressed," for example,
relations. This is true even though the two sets are found and to produce these conditions at will, a degree of
to be intimately connected. The two responses are not control and prediction characteristic of responses to ex-
controlled by the same stimuli, no matter how clearly it is ternal stimuli will be made possible. Meanwhile, we must
shown that the different stimuli arise from the same be content with reasonable evidence for the belief that
"thing." responses to public and private stimuli are equally lawful
What we want to know in the case of many traditional and alike in kind.
psychological terms is, first, the specific stimulating con- But the problem of privacy cannot be wholly solved by
ditions under which they are emitted (this corresponds to instrumental invasion. No matter how clearly these inter-
"finding the referents") and, second (and this is a much nal events may be exposed in the laboratory, the fact
more important systematic question), why each response remains that in the normal verbal episode they are quite
is controlled by its corresponding condition. The latter is private. We have not solved the second problem of how
not entirely a genetic question. The individual acquires the community achieves the necessary contingency of

548 THE BEHAVIORAL AND BRAIN SCIENCES (1984) 7:4

 Skinner:Psychological terms
reinforcement. How is the response "toothache" appro- again not an answer, for we are interested in how re-
priately reinforced if the reinforcing agent has no contact sponses to private stimuli are normally, and noninstru-
with the tooth? There is, of course, no question of mentally, set up.) There are two important possibilities.
whether responses to private stimuli are possible. They The surviving covert response may be regarded as an
occur commonly enough and must be accounted for. But accompaniment of the overt one (perhaps part of it), in
why do they occur, what is their relation to controlling which case the response to the private stimulus is im-
stimuli, and what, if any, are their distinguishing charac- parted on the basis of the public stimulus supplied by the
teristics? overt responses, as in (1). On the other hand, the covert
There are at least four ways in which a verbal communi- response may be similar to, though probably less intense
ty with no access to a private stimulus may generate than, the overt one and hence supply the same stimulus,
verbal behavior in response to it: albeit in a weakened form. We have, then, a third
1. It is not strictly true that the stimuli which control possibility: A response may be emitted in the presence of
the response must be available to the community. Any a private stimulus, which has no public accompaniments,
reasonably regular accompaniment will suffice. Consid- provided it is occasionally reinforced in the presence of
er, for example, a blind man who learns the names of a the same stimulus occurring with public manifestations.
trayful of objects from a teacher who identifies the objects Terms falling within this class are apparently descrip-
by sight. The reinforcements are supplied or withheld tive only of behavior, rather than of other internal states
according to the contingency between the blind man's or events, since the possibility that the same stimulus
responses and the teacher's visual stimuli, but the re- may be both public and private (or, better, may have or
sponses are controlled wholly by tactual stimuli. A satis- lack public accompaniments) seems to arise from the
factory verbal system results from the fact that the visual unique fact that behavior may be both covert and overt.
and tactual stimuli remain closely connected. 4. The principle of transfer or stimulus generalization
Similarly, in the case of private stimuli, one may teach a supplies a fourth explanation of how a response to private
child to say "That hurts" in agreement with the usage of stimuli may be maintained by public reinforcement. A
the community by making the reinforcement contingent response which is acquired and maintained in connection
upon public accompaniments of painful stimuli (a smart with public stimuli may be emitted, through generaliza-
blow, tissue damage, and so on). The connection between tion, in response to private events. The transfer is based
public and private stimuli need not be invariable; a not on identical stimuli, as in (3), but on coinciding
response may be conditioned with intermittent reinforce- properties. Thus, we describe internal states as "agi-
ment and even in spite of an occasional conflicting con- tated," "depressed," "ebullient," and so on, in a long list.
tingency. The possibility of such behavior is limited by Responses in this class are all metaphors (including spe-
the degree of association of public and private stimuli cial figures like metonymy). The term metaphor is not
which will supply a net reinforcement sufficient to estab- used pejoratively but merely to indicate that the differen-
lish and maintain a response. tial reinforcement cannot be accorded actual responses to
2. A commoner basis for the verbal reinforcement of a the private case. As the etymology suggests, the response
response to a private stimulus is provided by collateral is "carried over" from the public instance.
responses to the same stimulus. Although a dentist may In summary, a verbal response to a private stimulus
occasionally be able to identify the stimulus for a tooth- may be maintained in strength through appropriate rein-
ache from certain public accompaniments as in (1), the forcement based upon public accompaniments or conse-
response "toothache" is generally transmitted on the quences, as in (1) and (2), or through appropriate rein-
basis of responses which are elicited by the same stimulus forcement accorded the response when it is made to
but which do not need to be set up by an environmental public stimuli, the private case occurring by generaliza-
contingency. The community infers the private stimulus, tion when the stimuli are only partly similar. If these are
not from accompanying public stimuli, but from collat- the only possibilities (and the list is here offered as
eral, generally unconditioned, and at least nonverbal exhaustive), then we may understand why terms refer-
responses (hand to jaw, facial expressions, groans, and so ring to private events have never formed a stable and
on). The inference is not always correct, and the accuracy acceptable vocabulary of reasonably uniform usage. This
of the reference is again limited by the degree of associ- historical fact is puzzling to adherents of the "correspon-
ation. dence school" of meaning. Why is it not possible to assign
3. Some very important responses to private stimuli are names to the diverse elements of private experience and
descriptive of the speaker's own behavior. When this is then to proceed with consistent and effective discourse?
overt, the community bases its instructional reinforce- The answer lies in the process by which "terms are
ment upon the conspicuous manifestations, but the assigned to private events," a process we have just ana-
speaker presumably acquires the response in connection lyzed in a rough way in terms of the reinforcement of
with a wealth of additional proprioceptive stimuli. The verbal responses.
latter may assume practically complete control, as in None of the conditions which we have examined per-
describing one's own behavior in the dark. This is very mits the sharpening of reference which is achieved, in the
close to the example of the blind man; the speaker and the case of public stimuli, by a precise contingency of rein-
community react to different, though closely associated, forcement. In (1) and (2) the association of public and
stimuli. private events may be faulty; the stimuli embraced by (3)
Suppose, now, that a given response recedes to the are of limited scope; and the metaphorical nature of those
level of covert or merely incipient behavior. How shall we in (4) implies a lack of precision. It is, therefore, impossi-
explain the vocabulary which deals with this private ble to establish a rigorous scientific vocabulary for public
world? (The instrumental detection of covert behavior is use, nor can the speaker clearly "know himself" in the

THE BEHAVIORAL AND BRAIN SCIENCES (1984) 7:4 549

Skinner:Psychological terms
sense in which knowing is identified with behaving dis- fact is of extraordinary importance in evaluating tradi-
criminatively. In the absence of the "crisis" provided by tional psychological terms.
differential reinforcement (much of which is necessarily The response "red" is imparted and maintained (either
verbal), private stimuli cannot be analyzed. (This has casually or professionally) by reinforcement which is
little or nothing to do with the availability or capacity of contingent upon a certain property of stimuli. Both
receptors.) speaker and community (or psychologist) have access to
The contingencies we have reviewed also fail to pro- the stimulus, and the contingency can be made quite
vide an adequate check againstfictionaldistortion of the precise. There is nothing about the resulting response
relation of reference (e.g. as in rationalizing). Statements which should puzzle anyone. The greater part of psycho-
about private events may be under control of the depriva- physics rests upon this solid footing. The older psycholog-
tions associated with reinforcing consequences rather ical view, however, was that the speaker was reporting,
than antecedent stimuli. The community is skeptical of not a property of the stimulus, but a certain kind of
statements of this sort, and any attempt to talk about one's private event, the sensation of red. This was regarded as a
private world (as in psychological system making) is later stage in a series beginning with the red stimulus.
fraught with self-deception. The experimenter was supposed to manipulate the pri-
Much of the ambiguity of psychological terms arises vate event by manipulating the stimulus. This seems like
from the possibility of alternative or multiple modes of a gratuitous distinction, but in the case of some subjects a
reinforcement. Consider, for example, the response "I similar later stage could apparently be generated in other
am hungry." The community may reinforce this on the ways (by arousing an "image"), and hence the autonomy
basis of the history of ingestion, as in (1), or on the basis of of a private event capable of evoking the response "red"
collateral behavior associated with hunger, as in (2), or as in the absence of a controllable red stimulus seemed to be
a description of behavior with respect to food, or of proved. An adequate proof, of course, requires the elim-
stimuli previously correlated with food, as in (3). In ination of other possibilities (e.g. that the response is
addition the speaker has (in some instances) the powerful generated by the procedures which are intended to
stimulation of hunger pangs, which is private since the generate the image).
community has no suitable connection with the speaker's Verbal behavior which is "descriptive of images" must
stomach. "I am hungry" may therefore be variously be accounted for in any adequate science of behavior. The
translated as "I have not eaten for a long time" (1), or difficulties are the same for both behaviorist and subjec-
"That food makes my mouth water" (2), or "I am raven- tivist. If the private events are free, a scientific descrip-
ous" (3) (compare the expression "I was hungrier than I tion is impossible in either case. If laws can be dis-
thought" which describes the ingestion of an unexpected- covered, then a lawful description of the verbal behavior
ly large amount of food), or "I have hunger pangs. " While can be achieved, with or without references to images. So
all of these may be regarded as synonymous with "I am much for "finding the referents"; the remaining problem
hungry,' they are not synonymous with each other. It is of how such responses are maintained in relation to their
easy for conflicting psychological systematists to cite referents is also soluble. The description of an image
supporting instances or to train speakers to emit the appears to be an example of a response to a private
response "I am hungry" in conformity with a system. stimulus of class (1) above. That is to say, relevant terms
Using a stomach balloon, one might condition the verbal are established when the private event accompanies a
response exclusively to stimulation from stomach con- controllable external stimulus, but responses occur at
tractions. This would be an example of either (1) or (2) other times, perhaps in relation to the same private
above. Or speakers might be trained to make nice obser- event. The deficiencies of such a vocabulary have been
vations of the strength of their ingestive behavior, which pointed out.
might recede to the covert level as in (3). The response "I We can account for the response "red" (at least as well
am hungry" would then describe a tendency to eat, with as for the "experience" of red) by appeal to past conditions
little or no reference to stomach contractions. Everyday of reinforcement. But what about expanded expressions
usage reflects a mixed reinforcement. A similar analysis like "I see red " or "I am conscious of red"? Here "red"
could be made of all terms descriptive of motivation, may be a response to either a public or a private stimulus
emotion, and action in general, including (of special without prejudice to the rest of the expression, but "see"
interest here) the acts of seeing, hearing, and other kinds and "conscious" seem to refer to events which are by
of sensing. nature or by definition private. This violates the principle
When public manifestations survive, the extent to that reinforcement cannot be made contingent upon the
which the private stimulus takes over is never certain. In privacy of a stimulus. A reference cannot be narrowed
the case of a toothache, the private event is no doubt down to a specifically private event by any known method
dominant, but this is due to its relative intensity, not to of differential reinforcement.
any condition of differential reinforcement. In a descrip- The original behavioristic hypothesis was, of course,
tion of one's own behavior, the private component may be that terms of this sort were descriptions of one's own
much less important. A very strict external contingency (generally covert) behavior. The hypothesis explains the
may emphasize the public component, especially if the establishment and maintenance of the terms by supplying
association with private events is faulty. In a rigorous natural public counterparts in similar overt behavior. The
scientific vocabulary private effects are practically elimi- terms are in general of class (3). One consequence of the
nated. The converse does not hold. There is apparently hypothesis is that each term may be given a behavioral
no way of basing a response entirely upon the private part definition. We must, however, modify the argument
of a complex of stimuli. Differential reinforcement cannot slightly. To say "I see red" is to react, not to red (this is a
be made contingent upon the property of privacy. This trivial meaning of "see"), but to one's reaction to red.

550 THE BEHAVIORAL AND BRAIN SCIENCES (1984) 7:4

 Skinner: Psychological terms
"See" is a term acquired with respect to one's own Some afterthoughts on methodological and
behavior in the case of overt responses available to the radical behaviorism
community, but according to the present analysis it may
be evoked at other times by any private accompaniment In the summer of 1930, two years after the publication of
of overt seeing. Here is a point at which a nonbehavioral Bridgman's The Logic of Modern Physics, I wrote a paper
private seeing may be slipped in. Although the com- called "The Concept of the Reflex in the Description of
monest private accompaniment would appear to be the Behavior" (Skinner 1931), later offered as thefirsthalf of a
stimulation which survives in a similar covert act, as in (3), doctoral thesis. Although the general method, particu-
it might be some sort of state or condition which gains larly the historical approach, was derived from Mach's
control of the response as in (1) or (2). The Science of Mechanics (1893), my debt to Bridgman
The superiority of the behavioral hypothesis is not was acknowledged in the second paragraph. This was, I
merely methodological. That aspect of seeing which can think, the first psychological publication to contain a
be defined behaviorally is basic to the term as established reference to The Logic of Modern Physics (1928), and it
by the verbal community and hence most effective in was thefirstexplicitly operational analysis of a psychologi-
public discourse. A comparison of cases (1) and (3) will cal concept.
also show that terms which recede to the private level as Shortly after the paper was finished, I found myself
overt behavior becomes covert have an optimal accuracy contemplating a doctoral examination before a committee
of reference, as responses to private stimuli go. of whose sympathies I was none too sure. Not wishing to
The additional hypothesis follows quite naturally that wait until an unconditional surrender might be neces-
being conscious, as a form of reacting to one's own sary, I put out a peace feeler. Unmindful or ignorant of
behavior, is a social product. Verbal behavior can be the ethics of the academy, I suggested to a member of the
distinguished, and conveniently defined, by the fact that Harvard department that if I could be excused from
the contingencies of reinforcement are provided by other anything but the most perfunctory examination, the time
organisms rather than by a mechanical action upon the which I would otherwise spend in preparation would be
environment. The hypothesis is equivalent to saying that devoted to an operational analysis of half a dozen key
it is only because the behavior of the individual is impor- terms from subjective psychology. The suggestion was
tant to society that society in turn makes it important to received with such breathless amazement that my peace
the individual. One becomes aware of what one is doing feeler went no further.
only after society has reinforced verbal responses with The point I want to make is that at that time - 1930 - I
respect to one's behavior as the source of discriminative could regard an operational analysis of subjective terms as
stimuli. The behavior to be described (the behavior of a mere exercise in scientific method. It was just a bit of
which one is to be aware) may later recede to the covert hackwork, badly needed by traditional psychology, which
level, and (to add a crowning difficulty) so may the verbal I was willing to engage in as a public service or in return
response. It is an ironic twist, considering the history of for the remission of sins. It never occurred to me that the
the behavioristic revolution, that as we develop a more analysis could take any but a single course or have any
effective vocabulary for the analysis of behavior we also relation to my own prejudices. The result seemed as
enlarge the possibilities of awareness, so defined. The predetermined as that of a mathematical calculation.
psychology of the other one is, after all, a direct approach I am of this opinion still. I believe that the data of a
to "knowing thyself." science of psychology can be defined or denoted unequiv-
The main purpose of this discussion has been to define ocally, and that some one set of concepts can be shown to
a definition by considering an example. To be consistent, be the most expedient according to the usual standards in
psychologists must deal with their own verbal practices scientific practice. Nevertheless, these things have not
by developing an empirical science of verbal behavior. been done in thefieldwhich was dominated by subjective
They cannot, unfortunately, join logicians in defining a psychology, and the question is, Why not?
definition, for example, as a "rule for the use of a term" Psychology, alone among the biological and social sci-
(Feigl 1945); they must turn instead to the contingencies ences, passed through a revolution comparable in many
of reinforcement which account for the functional relation respects with that which was taking place at the same time
between a term, as a verbal response, and a given in physics. This was, of course, behaviorism. The first
stimulus. This is the "operational basis" for their use of step, like that in physics, was a reexamination of the
terms; and it is not logic but science. observational bases of certain important concepts. But by
Philosophers will call this circular. They will argue that the time Bridgman's book was published, most of the
we must adopt the rules of logic in order to make and early behaviorists, as well as those of us just coming along
interpret the experiments required in an empirical sci- who claimed some systematic continuity, had begun to
ence of verbal behavior. But talking about talking is no see that psychology actually did not require the redefini-
more circular than thinking about thinking or knowing tion of subjective concepts. The reinterpretation of an
about knowing. Whether or not we are lifting ourselves established set of explanatory fictions was not the way to
by our own bootstraps, the simple fact is that we can make secure the tools then needed for a scientific description of
progress in a scientific analysis of verbal behavior. behavior. Historical prestige was beside the point. There
E antually we shall be able to include, and perhaps to was no more reason to make a permanent place for terms
understand, our own verbal behavior as scientists. If it like "consciousness," "will," or "feeling" than for "phlo-
turns out that ourfinalview of verbal behavior invalidates giston" or "vis anima." On the contrary, redefined con-
our scientific structure from the point of view of logic and cepts proved to be awkward and inappropriate, and
truth value, then so much the worse for logic, which will Watsonianism was, in fact, practically wrecked in the
also have been embraced by our analysis. attempt to make them work.

THE BEHAVIORAL AND BRAIN SCIENCES (1984) 7:4 551

SkinnenPsychological terms
Thus it came about that while the behaviorists might philosophy of "truth by agreement." The public, in fact,
have applied Bridgman's principle to representative turns out to be simply that which can be agreed upon
terms from a mentalistic psychology (and were most because it is common to two or more agreers. This is not
competent to do so), they had lost all interest in the an essential part of operationism; on the contrary, opera-
matter. They might as well have spent their time in tionism permits us to dispense with this most unsatisfy-
showing what an 18th-century chemist was talking about ing solution of the problem of truth. Disagreements can
when he said that the Metallic Substances consisted of a often be cleared up by asking for definitions, and opera-
vitrifiable earth united with phlogiston. There was no tional definitions are especially helpful, but opera-
doubt that such a statement could be analyzed opera- tionism is not primarily concerned with communication
tionally or translated into modern terms, or that subjec- or disputation. It is one of the most hopeful of principles
tive terms could be operationally defined, but such mat- precisely because it is not. The solitary inhabitant of a
ters were of historical interest only. What was wanted was desert isle could arrive at operational definitions (pro-
a fresh set of concepts derived from a direct analysis of the vided he had previously been equipped with an ade-
newly emphasized data, and this was enough to absorb all quate verbal repertoire). The ultimate criterion for the
the available energies of the behaviorists. Besides, the goodness of a concept is not whether two people are
motivation of the enfant terrible had worn itself out. brought into agreement but whether the scientist who
I think the Harvard department would have been uses the concept can operate successfully upon his mate-
happier if my offer had been taken up. What happened rial - all by himself if need be. What matters to Robin-
instead was the operationism of Boring and Stevens. This son Crusoe is not whether he is agreeing with himself
has been described as an attempt to climb onto the but whether he is getting anywhere with his control over
behavioristic bandwagon unobserved. I cannot agree. It nature.
is an attempt to acknowledge some of the more powerful One can see why the subjective psychologist makes so
claims of behaviorism (which could no longer be denied) much of agreement. It was once a favorite sport to quiz
but at the same time to preserve the old explanatory him about intersubjective correspondences. "How do
fictions. It is agreed that the data of psychology must be you know that O's sensation of green is the same as E's?"
behavioral rather than mental if psychology is to be a And so on. But agreement alone means very little. Vari-
member of the Unified Sciences, but the position taken is ous epochs in the history of philosophy and psychology
merely that of "methodological" behaviorism. According have seen wholehearted agreement on the definition of
to this doctrine the world is divided into public and psychological terms. This makes for contentment but not
private events; and psychology, in order to meet the for progress. The agreement is likely to be shattered
requirements of a science, must confine itself to the when someone discovers that a set of terms will not really
former. This was never good behaviorism, but it was an work, perhaps in some hitherto neglected field, but this
easy position to expound and defend and was often does not make agreement the key to workability. On the
resorted to by the behaviorists themselves. It is least contrary, it is the other way round.
objectionable to the subjectivist because it permits him to 3. The distinction between public and private is by no
retain "experience" for purposes of "nonphysicalistic" means the same as that between physical and mental.
self-knowledge. That is why methodological behaviorism (which adopts
The position is not genuinely operational because it thefirst)is very different from radical behaviorism (which
shows an unwillingness to abandon fictions. It is like lops off the latter term in the second). The result is that
saying that although the physicist must admittedly con- whereas the radical behaviorist may in some cases consid-
fine himself to Einsteinian time, it is still true that er private events (inferentially, perhaps, but nonetheless
Newtonian absolute time flows "equably without relation meaningfully), the methodological operationist has ma-
to anything external." It is a sort of E pur si muove in neuvered himself into a position where he cannot. "Sci-
reverse. What is lacking is the bold and exciting behav- ence does not consider private data," says Boring (1945). I
ioristic hypothesis that what one observes and talks about contend, however, that my toothache is just as physical as
is always the "real" or "physical" world (or at least the my typewriter, though not public, and I see no reason
"one" world) and that "experience" is a derived construct why an objective and operational science cannot consider
to be understood only through an analysis of verbal (not, the processes through which a vocabulary descriptive of a
of course, merely vocal) processes. toothache is acquired and maintained. The irony of it is
It may be worthwhile to consider four of the principle that, whereas Boring must confine himself to an account
difficulties which arise from the public-private of my external behavior, I am still interested in what
distinction. might be called Boring-from-within.
1. The relation between the two sets of terms which are 4. The public-private distinction apparently leads to a
required has proved to be confusing. The pair most logical, as distinct from a psychological, analysis of the
frequently discussed is "discrimination" (public) and verbal behavior of the scientist, although I see no reason
"sensation" (private). Is one the same as the other, or why it should. Perhaps it is because the subjectivist is still
reducible to the other, and so on? A satisfactory resolu- not interested in terms but in what the terms used to
tion would seem to be that the terms belong to conceptual stand for. The only problem a science of behavior must
systems which are not necessarily related in a point-to- solve in connection with subjectivism is in the verbal
point correspondence. There is no question of equating field. How can we account for the behavior of talking
them or their referents, or reducing one to the other, but about mental events? The solution must be psychological,
only a question of translation - and a single term in one rather than logical, and I have tried to suggest one
set may require a paragraph in the other. approach in my present paper.
2. The public-private distinction emphasizes the arid The confusion which seems to have arisen from opera-

552 THE BEHAVIORAL AND BRAIN SCIENCES (1984) 7:4

 Commentary/ Skinner: Psychological terms

tionism - a principle which is supposed to eliminate "This is a chair" or "That is a Ming vase" - it does not happen
confusion - is discouraging. But upon second thought it often, and there is no reason to take it as paradigmatic of
appears that the possibility of a genuine operationism in linguistic behavior, or as central or basic in it.
psychology has not yet been fully explored. With a little Let us set that fact aside also, and attend to the tiny fragment
effort I can recapture my enthusiasm of some years ago/ of linguistic behavior that does fit this pattern. Still there is
trouble for Skinner's theory of meaning. I am confronted by
(This is, of course, a private event.) something red; it is a stimulus, to which I respond by saying
"(That is) red." In calling these items a "stimulus" and a
NOTE "response" respectively, Skinner is implying that the former
This article is slightly revised from the original, which ap- causes the latter: Like most stimulus-response meaning theo-
peared in Psychological Review 52: 270-277; 291-294, 1945. rists, he is apparently attracted by the idea that the meanings of
our utterances are determined by the very same items that
cause them. In his own words, the "referents" of what we say
"control" our saying it, and he ties control to prediction, speak-
ing of a "science of behavior in which we are to predict response
through, among other things, an independent knowledge of the
Open Peer Commentary stimulus."
The phrase "among other things" is needed in that sentence.
Without it, Skinner would be implying that linguistic behavior
Commentaries submitted by the qualified professional readership of is vastly more predictable than it really is, in the manner of the
this journal will be considered for publication in a later issue as stimulus-response meaning theorist who once wrote: "If you
Continuing Commentary on this article, lntegrative overviews and want a person to utter the word chair, one of the best ways is to
syntheses are especially encouraged.
let him see an unusual chair" (Miller 1951, p. 166). That is
plainly false, of course, and no one would write it who was not in
thrall to a bad theory. In a large range of situations we can
predict something about the world from a fact about what is said
Stimulus-response meaning theory - for example, someone's saying "This is a chair" is evidence
that he is probably in the presence of a chair - but predictions
Jonathan Bennett running the other way are nearly always quite hopeless (this
Department of Philosophy, Syracuse University, Syracuse, N.Y. 13210 point is made by Ziff 1970, p. 73; see also Ziff 1960, sees. 46 and
Skinner's account of how subjective psychological terminology 54). But Skinner says "among other things." We are to suppose
gets its meaning relies on his views about meaning in general. that the causally sufficient conditions for a person's uttering
Though not extensively laid out in "Terms," their general "(That is) red" consist in (i) a red stimulus in conjunction with (ii)
outline emerges clearly enough to show how radically mistaken a set of circumstances C which always mediates between a
they are. So there must be a lot wrong also with Skinner's stimulus and an utterance whose meaning is somehow given by
account of the meanings of psychological terms, but I shall not the stimulus. If the theory is not that there is a single value of C
follow out those consequences; my topic is the underlying such that someone who undergoes a red stimulus in C circum-
stimulus-response approach to meaning in general. stances says something like "That is red, " someone who sees a
To evaluate Skinner's views about meaning we must first chair in C circumstances says "That is a chair," and so on, then
cleanse them of their most unrealistic assumption, namely that there is no theory. The aim is to say something systematic about
the basic linguistic performance is the uttering of a single word. how the meanings of utterances relate to their causes, and that
requires a general rule enabling us to read off the meaning of an
When Skinner speaks of "the circumstances under which the
utterance from the facts about the causal chain that produced it.
response 'length' is emitted" he is not discussable. Apart from We shan't get that merely by learning that in each case the
certain highly specialized circumstances, such as helping with a causal chain includes, together with a lot of other stuff, some-
crossword puzzle or displaying reading skills, there are no thing constitutive of the meaning of the utterance. We need a
circumstances under which that one word is uttered in isolation. systematic way offilteringout the "other stuff' in order to isolate
And when he implicitly contrasts "I see red" with "red," calling the element that gives the meaning; and so, as I said, we need a
the former an "expanded expression," he puts the cart before single value of C that tells us in each case which part of the causal
the horse. Although we grasp sentences only through under- chain gives the meaning and which part belongs to the all-
standing their constituent words, the notion of meaning attaches purpose "other stuff." (For a fuller defense of this, see sec. 6 of
primarily to whole sentences and only derivatively to smaller Bennett 1975.)
units such as words. Our primary concept of meaning is that of
something's meaning that P, and the notion of word meaning That is the project of Skinner's kind of stimulus-response
must be understood through the idea of the effect on a sen- meaning theory. (There is another kind - no better but different
tence's meaning of replacing this word in it by that. Try to - according to which meaning is determined not by the stimuli
imagine a tribe that has a word for trees, a word for sand, a word to which an utterance is a response but rather by the responses
forfire,and so on, but that does not use these words in sentences to the utterance considered as stimulus. For more on this, and
to say anything about trees, sand, orfire.The supposition makes on relations between the two, see sees. 7 - 9 of Bennett 1975.)
no sense: If the noises in question are not used to say anything, As a project, it has no hope of success: There is no reason to think
to express whole "that-P" messages, there is nothing to make it there is anything remotely resembling a general truth of the
form "Whenever anyone encounters an F item in C circum-
the case that the noises are words at all.
stances he utters something meaning that the item is F." Let C
However, when Skinner and other stimulus-response mean- be somewhat vague and tattered around the edges; let it also be
ing theorists focus on the single word, perhaps they are really less than perfectly unitary, consisting perhaps of about 17
thinking not of the word "red," say, but rather of the one-word disjuncts; lower your sights by looking only for a rule that applies
sentence "Red!," meaning something like "That thing (in front about 20% of the time; help yourself to two or three further
of me) is red." Let us suppose this, and forget that it still makes indulgences as well. Still the project will have no chance of
no sense of "the response 'length.'" success. It assumes a world-to-meaning relationship that simply
The activity of labeling whatever public or private item one is doesn't exist.
presented with is a rare event. Even if we allow for it to be done
in normal sentences with several words each - for example, This is not to deny that when a person says something

THE BEHAVIORAL AND BRAIN SCIENCES (1984) 7:4 553