Skip to content

Cognitivism, as an Operation-Analytic Behaviorist Views It

William S. Verplanck

Knoxville, Tennessee, USA

It is a special pleasure to give this paper to the Third International Congress on Behaviorism and the Science of Behavior at this time with Prof. Sato as host. At the last of these meetings, in Palermo, Prof. Sato and his colleagues presented a paper, Behavior analysis of ‘memory,’ that is a pioneering example of the behavioral research that the operation-analytic behaviorist proposes be done on such cognitive–mentalistic–concepts.

This paper will extend Prof. Sato’s exposition of the folk-psychological, or more properly, linguistic roots of cognitivism and will argue that language, dependent as it is on social reinforcement as well as on the objects and events–the interactions of these with the actions of the individual, that is, the behavior–to which verbal behavior is related, cannot be expected to yield a conceptual system adequate to deal with the subject matters that cognitivists purport to describe and explain, that is, the complex behaviors of humans and of members of other species.

We shall, in short, examine the scientific soundness of statements such as the following by a cognitivist: “One of the hallmarks of human evolution may have been the emergence of a radically new psychology that does not merely learn about behavior, but interprets it in a rich and mentalistic way1” (Povinelli, 1995). And we shall offer an operation-analytic, the interbehavior-analytic examination of such a psychology and suggest how an analysis of the origin of its terminology might prove more fruitful than investigation based on that terminology.

Let me start by conceding that the mentalistic language of the cognitivist is a powerful tool for the description and explanation of complex events for the layperson. Here is ‘A note on language’ I distributed to students in a course on Learning I gave somewhere in the mid or later fifties–before the very beginning of the “cognitive revolution.”

“The English language as we use it in ordinary conversation has developed with an obvious function of casual and general communication. Evidently, it does this pretty well, and it has the merit of enabling one to make a certain amount of sense in talking with people devoid of the command of the accurate technical language that is necessary for a scientific grasp of a set of events. That this is so should not lead us into the trap of thinking that conversational terms have validity and appropriateness in talking about our subject matter, even though they may be convenient, easy, and helpful in ‘explaining’ to laymen what is going on. Take the following example, which ‘describes’ and ‘explains’ the behavior of organic molecules.

“‘This marshaling of molecules at the surface of a solution lowers its surface tension, which is a measure of its desire to maintain a minimum surface area.

“‘The ions or molecules at the surface of a detergent solution are rather loosely held; indeed, their hydrophobic portions would like to leave the solution entirely.

“‘The willingness of such solutions to have their surfaces extended also accounts for the formation of suds.

“‘A spectacular demonstration of this power of detergents is to put a duck into a detergent solution. The solution so thoroughly displaces air from the animal’s naturally oily feathers that its buoyancy is sharply reduced and it must struggle to keep from sinking. Our discussion of detergent solutions indicates why this happens. The molecules at the surface of plain water would much rather be in contact with air or water than with a greasy substance. This keeps the water from spreading on a greasy surface, and results in the formation of droplets which present the least possible area of contact between the water and the grease. The detergent solution, however, has those portions of the detergent molecules at its surface which would rather be in contact with the grease than with the air.’

(Quoted from Kushner, L. M. and Hoffman, J. I., Synthetic Detergents, Scientific American, Oct. 1951, pp. 26-30.)

“Anybody can recognize this example of anthropomorphism for what it is–a personification of the molecule designed to give people with no background in chemistry some vague notion of what goes on when the washing gets done. It goes without saying that the writers are not seriously proposing that these statements in any way constitute a scientific description or explanation of the events involved.

“Such use of naive ‘motivational’ and ‘phenomenological’ terms is equally inappropriate if we are to achieve a grasp of the mechanisms involved in the behavior of animals, and of men. The language can and may be used at the most crude descriptive level, as in the example above, and we shall use it this way occasionally. But when one uses it, one must always be able to translate the statement into a more precise, unequivocal, and scientifically meaningful language–even though the translation is a good deal longer, and more cumbersome. If you use terms such as these during your work in this course, be very sure that you can translate them–because you will be called upon to do it. The safe thing, of course, is to give them up and start knowing exactly what you’re talking about before you start talking!”

Kushner and Hoffman presented a radically new chemistry that does not merely learn about the behavior of molecules, but “interprets it in a rich and mentalistic way.”

Just what is this “radically new psychology?” Cognitivists may disagree in detail, but those who observe their behavior in producing papers and books will recognize that, first, they do not question the classical textbook rubrics of motivation, perception, and emotion. They deal specifically with problems in feeling, learning, memory, retention, problem-solving, consciousness, awareness, belief, intention, knowledge, imagery, attention, intelligence, storage, retrieval, information-processing, understanding, thinking, choice, and an all but unlimited variety of ‘problems,’ many of them produced by their theories. Second, they purport to determine how the “mind” or “consciousness” or the point-at-able brain and its parts ‘work.’ And third, they are very scientific. From research in these problem areas, they have generated dozens of concepts, scores of books, milliards of papers, and a goodly number of scientific societies which deal with one or another of the “cognitive sciences” that have proliferated. We can characterize cognitivists’ research by the methodology they pursue.

(No doubt some will complain that the writer of this paper has overgeneralized. He would be the first to point out ‘exceptions’. But that exceptions can be readily identified attests to the verifiability of the generalization. The exception tests (is witness to) the generalization.)

With those few revealing exceptions, cognitivists follow the basic precepts of the behaviorisms; they are indeed “methodological behaviorists.” They carefully specify, insofar as they are able, the independent variables they have identified and think important. These are “physically” specified as objects and events that are related to one or more dependent variables, measures of topographically specified responses of their subjects. These measures of response, the dependent-variables, are carried out in behaviorally appropriate ways. Analysts, however, cannot overlook the unfortunate circumstance that their theories lead cognitivists to neglect variables that behaviorists have proven to be important, and that their experimental methodology minimizes the full variety of responses that a subject may give in an experimental research.2 For cognitivism, a response is ‘significant’ only in that it is a sign that some cerebral or mental event has occurred and produced it, that is, the response is taken as an “operational definition” of the unobserved concept.

Given this methodological behaviorism, however, we can take heart in the fact that operational-analyses of their experiments may enable some findings reported by cognitivists to be incorporated within an operation-analytic behaviorism.

The problem is, however, that in their research, cognitivists invariably accept the hypothetico-deductive methodology of “theory construction,” which has its roots in logical positivism, a philosophical system that purports to be the singular road to ‘truth.’ (They differ from Clark Hull and his coworkers only in the degree to which they have eschewed formal statements.) This method generates both ‘intervening variables’ and ‘hypothetical constructs,’ postulated (guessed at, dreamed up) theoretical entities to which are assigned properties that, rather than being produced by data, produce the ingeniously-designed experiments that then produce the very data that lead to the postulation. These ‘bootstrapped’ data generate statements purporting to explain or describe behavior in terms of agents and processes of mind or of brain (or some parts thereof) which determine the subject’s actions. Environmental objects and events become “information,” which the ‘brain,’ the ‘mind,’ or ‘consciousness’ then “processes,” thus ‘causing’ the behavior measured, which (surprise!) had, in fact, been determined by the apparatus and the instructions to the subject.

The experimental results (and generalizations) so produced are then evaluated on the criteria set by the currently fashionable methodology for “truth-evaluation.” From the special point of view of cognitivists, this methodology is impeccable. But it leaves no space for data from naturalistic as well as much non-cognitive experimental observation. It relies on the application of deductive, rather than inductive logic. It neglects the categorization of behavior in terms other than those imposed by the vocabulary they begin with. The pragmatism that is the earmark of inductive thinking goes by the board.

Let us look at the cognitivists’ concepts:


Some cognitive nominalizations, the verbs from which they are derived, and the verb’s subject and direct objects (the subject of these verbs is usually left open, “it” may be the individual organism, the mind, the brain, or consciousness, depending on theory)

Nominalization          Verb                      Subject      Direct Object 
perception              perceive                   open         unspecified
feeling                 feel                         "               "       
learning                learn that                   "               "       
learning how            learn how                    "               "       
rote-learning           learn by rote                "               "       
memorization            memorize                     "               "       
rehearsal               rehearse                     "               "       
expectancy              expect                       "               "       
intention               intend                       "               "       
problem-solving         solve                        "         problem (env) 
skill-acquisition       acquire                      "         skill (beh)   
attention               attend                       "         unspecified   
knowledge               know                         "               "       
imagination             imagine                      "               "       
belief in               believe in                   "               "       
belief that             believe that                 "               "       
comprehension           comprehend                   "               "       
thinking                think                        "               "       
understanding           understand                   "               "       
categorization          categorize                   "               "       
abstraction             abstract                     "               "       
generalization          generalize                   "               "       
recall                  recall                       "               "       
reminiscence            reminisce                    "               "       
recognition             recognize                    "               "       
reconstruction          reconstruct                  "               "       
retention               retain                       "               "       
storage                 store                        "               "       
memory                  remember                     "               "       
retrieval               retrieve                     "               "       
motivation              motivate                     ?         individual    
emotion                 make (angry/sad)             ?               "       
information             inform                  env.  event          "       
imagery                 image (of)                   "         env. event    


Cognitive nominalization of adjectives describing an individual organism

Nominalization          State of Individual    
 awareness              aware (of, that)       
 consciousness          conscious (of, that)   

First note that the names of these concepts, embedded in the language they use, derive from the rich and mentalistic the warm and fuzzy vernacular, ordinary language. The vocabulary of the language of cognitivism is rooted in the vocabulary of folk-psychology, as Prof. Sato among others has pointed out.

Note that the concept-names in Column 1 are nominalizations of the verbs, the names of actings, in Column 2. Linguistics distinguish between subjects, verbs, and direct objects: Subjects are nouns; verbs are the names of observable, usually transient, actings of an individual; and the direct objects are the consequences of transitive (active) verbs. The nominalizations of Column 1 enable cognitive thinkers to leave open the question whether the individual, the mind, the brain, or parts thereof are the subjects, the ‘agents,’ of the verbs from which the nominalizations are derived. Note, too, that the direct objects of the transitive verbs are equally unspecified. The “concept” floats independently of the specification of either subject or direct object, as the third and fourth columns show. The verbs of descriptive sentences are taken from (abstracted from) sentences and then converted into nouns, the grammatical form taken by thesubjects of further sentences. By this magic of language, actions become entities–agents that act.

One of these nominalizations is “expectancy,” a word with great descriptive and explanatory effectiveness, even now a source of confusion for would-be behavior analysts.

“Expect… expectancy” is an exemplar of a term from ordinary language that becomes a cognitive concept–an explanatory concept. Years ago, Ritchie, a student of Tolman’s, set out a rule for the use of the word “expectancy”–an “operational definition.” Tolman, we note, was a card-carrying logical-positivist and overtly practiced “operational definition” in introducing folk-psychological terms into his behaviorism. This operational definition restricted the use of the word “expectancy” to a rat’s moving down one path of a maze rather than another, a hardly satisfactory restriction on the word’s use. We shall return to fuller consideration of this word in its relationship with behaviors.

This conceptual system is singularly unsuited to naturalistic descriptions of observable behaviors. These nominalizations permit, if not require, the neglect of the specific behaviors of the individual and enable the generation of over-hasty and unfounded abstractions, defensible only by abuse of the notion of operational definition. Thus, one investigator who studied the consequences of teaching children to draw diagrams with paper and pencil described both the behavior and its consequences as “visual reasoning.” (This example may lead us to ask whether, when cognitivists “reason” and “think about” problems, they, like most of us literate humans, have paper and pencil or keyboard at hand and busily take notes and do calculations–except, of course, when they are subjects in an experiment on thinking.)

The ‘big bang’ of the inception of cognitivism has led to a chaotic cosmos of concepts–theoretical entities and processes which offer far less opportunity for being integrated into a system of working generalizations than the subject matter of cosmology itself. The folk-psychological concepts of cognitivism inevitably lead to the perpetuation of anthropocentric thought, sets of statements derived from human verbal behavior. Even more difficult that many of them are based on the specific vocabulary of the folk-psychology of the vernacular, that relates to events observable and describable only by the individual describing them at that moment. Many such description will necessarily use a vocabulary idiosyncratic to the individual making them. Skinner’s 1948 paper provides an understanding of why, in 1913, Watson rejected “consciousness” and “introspection.”

These concepts are nevertheless then extended by uncritical extrapolation to other species, whether or not members of those other species exhibit fewer or more specialized behaviors and specialized structures that participate in behaving. ‘Lloyd Morgan’s Canon’ is now being rejected, or at least being restated, by a new generation of students of ‘Comparative Cognitive Processes3.’

In sum, cognitivism is sliding into a confusion of ideas, concepts, or notions, based on the vernacular, the folk-psychological concepts found in the Indo-European languages of Western Europe (Buck, 1949) which, until recent years, has been the mother-tongue of psychologists, and in only one of which, English, most findings in psychology, cognitivist or behavioral, are now published, whatever the first language of the investigator.

Is there an alternative to the impeccable scientism of the cognitivists, who believe they have freed study of the “mind” from the defects of mentalism? The answer is an unequivocal yes. A methodology that incorporates naturalism with a behavior analytic and operation-analytic inductive approach to the very behaviors that fall within the subject-matters beloved by cognitivists provides the alternative.

This methodology derives from the reciprocal relationships of the activities of the individual and the activities of the environment, that is, from the behaving itself, as it is observed and recorded. Behaving may be thought of as a sort of conversation between individual and environment. In this analysis, the class of environmental events observed to be related to a specific class of the activities of the individual are termed stimuli; the specific class of activities of the individual observed to be related to classes of environmental activities are termed responses.

Stimuli and responses are both classes of transient events identified by recurring instances of them. Any one observation of the relationship of an activity of the individual and of the activity of the environment is of a single instance of each, and of a single instance of behaving, an incident–Kantor’s “psychological event,” Lee’s “things done.” A series of such observations is required to ensure that the apparent reciprocal relationship is not unique, and that stimulus, response, and the relationships have all been identified ostensively. Unique observations give no clue of the class memberships of the observed instances of behaving. Repeated observations enable the investigator to assert such generalizations as ‘discriminative stimuli yield responses, which produce environmental changes that are reinforcing stimuli.’ This statement defines SRS, the three-term contingency. With such repeated observation, the reciprocity of the relationship is evident: We may note, too, that the individual’s response functions as a stimulus to the environment, and that the environment’s response to it functions as a stimulus to the individual. Behavior is a generic term for such interactive, reciprocal relationships.

We must always be alert to this proposition that on any one occasion when one observes an interaction, we are dealing with single instances of both S and R, and that identification of both stimuli and responses is necessarily based on a number of observations. Class membership must be identified and named by demonstrating the referents of the terms stimulus and response; the use of these terms must be based on ostensive definition, and not on verbal formulae alone. When we use words, rather than the events, or records of the events themselves, we must be sure that the words are defined ostensively, whether we are concerned with stimuli and responses, or with cows, jogging, and bar-pressing. Thus, the identity and training of the observer is crucial. One cannot talk or learn about behavior from books, any more than one can become a chemist by memorizing texts and formulas never going to the lab or working in it with somebody. One becomes a behaviorist by doing.

Given this dependence on ostensive definition, on demonstrating, and hence on the experience of the observer, it is not surprising that it has become conventional to deal experimentally with stimuli and responses, instances of which are observably alike, similar to one another, that homogeneity being forced, ensured, by the apparatus. The Skinner box and other such experimental apparatus make it possible to overlook the range of variability among instances of both stimulus and response, of bar press, and pellet. With such apparatus, these variabilities are made experimentally evident in studies of stimulus generalization and (response) induction. The data then appear as counts of response, of the numbers of instances of a response occurring under such and such conditions. The reliability of the apparatus determines the reliability of the count.

Outside the laboratory, however, we find that in specified contexts, the membership of class of both environmental events and actions of the individual is determined by the scientific qualifications and experience of the observer in identifying responses and stimuli from instances of them. Here, the reliablility of data is measured by comparison of counts made by different qualified observers. With such observers, we also find that instances of both S and R may be heterogeneous; they may not be topographically similar (as apparatus ensures) although the members function alike. Specific instances of both stimulus and response may be identified only by functional relationships of each with the other. The stimuli, such as reinforcers or aversors and SD’s, releasers, or elicitors that enable identification of a response may be very unlike one another indeed.

In the usage to be followed in this paper, the term Concept, with a capital C, is applied to stimuli whose class-membership is diverse, and the term Concept-set to responses, instances of which may topographically differ greatly from one another, but which nevertheless can be shown to be related to one or another member of the Concept. In all cases, use of these terms are controlled by ostensive definition, and not by lexical or theoretical definition using other words.4

In dealing with Concepts and their relationships to one another, we have emphasized that words need play no role in the identification of Concepts and Concept-Sets. We communicate with one another, use language-specific words, words that when heard or read and when spoken or written by an observer (whether scientific or not), function as both stimuli and responses. A word or statement of words is thus both a member of a Concept and a member of its related Concept-set. It is thus words with which we work when we deal with Concepts and their Sets. Words function as Concept-names as do the statements that specify rules for determining the class-membership of events, whether those statements distinguish art from science, energy from work, popes from priests, or disjunctive concepts from conjunctive ones. Here, as everywhere, the identity and training of the observer, the individual using words, must necessarily be taken into account in evaluating any reported observation, whether the observation is naturalistic or whether it is generated in an experiment.5

So, down to cases: Let’s start with two different observers, my dog, Diana, and I.

(As I just dictated “dog,” she came to my side; my speaking the word “dog,” I note, produces the same behavior that she shows when I call out “Diana!”…. My behavior repertory includes the word “dog,” and she is an instance of that diverse class of instances, “dog,” just as are all other dogs, all definitions of “dog” I might write or state, and all other writings about and pictures of dogs I have read or seen. My own Concept, “dog” overlaps only in part with the scientific Concept Canis familiarus, whose Concept name/statement limits class-membership of both Concept and Concept-Set.

Diana has a Concept “dog,” too. Diana is illiterate, so her Concept lacks a name. Her Concept is defined by the Concept-Set controlled by instances of a class “dog.” On occasions, when she is in my car and can see out, she will suddenly show loud and persistent barking and simultaneously do a good deal of jumping around, still orienting in one direction. She shows this behavior when and only when there is a dog present somewhere in her visual environment (what she can respond to when there is no object interposed between her eyes and a dog.) She’s very good at this, I seldom see one first. It doesn’t matter what kind of a dog it is, whether it is a mutt or a pedigreed Akita, whether it is standing or moving, and whether is it near or far. When she shows this behavior, if I look in the direction of her orientation, I, too, can see a dog, sometimes near, sometimes very far away. The acuity of her vision, and her accuracy in identifying dogs is remarkable: squirrels and cats get no such response. When Diana is not looking out, but hears the bark of another dog, and there is no dog to be seen, she will show persistent barking but not so loud and with little jumping around. Heard barks, it seems, are like seen dogs, but not exactly.

Diana has another Concept, identified by a Concept-Set. In this case, my name for that Concept is “my car, a gray Audi station wagon.” One subclass of members of her Concept is my shout “car!”: When we are out on our daily walks, leaving the car somewhere in the parking-lot where other cars may be, then at any point, wherever she is and wherever I am, if I shout the word “Car!” she runs to the car, my car, arriving more often at the driver’s side than not (the side of the door she gets in on). She discriminates between my car and other cars that may be in the parking lot, erring only occasionally by going to one similar to it in size and color. There is another set of stimuli in the Concept governing this Concept-Set, I now call “Diana runs to the car door.” That new set is the dog’s seeing (observing) my behavior of changing direction from walking away from the car to walking towards the car. Wherever she is, and wherever I am within earshot or view, she will run to the car door when I shout “Car!” and/or when I change direction of walk. Most interesting, if I shout “Car!” and the car is hidden behind shrubbery she looks for it. Diana, a dog, shows the Concept that I call my car and a Concept-Set I specify as run to the door. She does not have a Concept-name; that is reserved for me as human. She has no schema, nor search-image; she has no rule for the application of the word but she shows both Concept and Concept-Set. Unabashedly, I can state that the events jumping into the car when I open the door and then riding in the car are reinforcers for such “referential” behaviors often observable of dogs, but not of cats. (N.b.of, not in)

In like manner, the behavior of the rat in the Skinner box (more specifically, readily describable in terms of the “three-term contingency,” SRS) clearly shows Concepts and Concept-Sets. Bar, bar-press, magazine click, and pellet are allmy names for the rat’s Concepts (stimuli) and Concept-Sets (responses). Since Herrnstein’s initial research showing that a pigeon can acquire the Concept that English-speakers call “Humans,” its Concept-Set being limited to pecking a key, others have shown even more “complex” and “abstract” Concepts in species other than our own. Even more, Concepts of relationships among Concepts are shown by the work of Sidman and others.In Pavlovian conditioning, two stimuli are presented in sequence to a dog, the first initially producing only orienting, and the second, salivating. Repetition of this operation, pairing, leads the dog to salivate on the presentation of S1. This operation, S1/S2, controls the process “classical conditioning.” The operational behaviorist would state that salivating when the metronome sounds demonstrates the process conditioning, controlled by the operation pairing. Ordinary language, folk psychology (and perhaps all of us speaking informally), says that the dog expects food when the metronome sounds, just as we would say that the rat expects food when it has pressed the bar.

Cognitivism turns these last behaviors, casually labeled “expecting,” into expectancy, an entity, a cognitive event, a thing housed in mind or brain, and then slips it into the proposition that it is the expectancy of food, not the series of pairings, or the metronome, that is the ‘necessary and sufficient’ antecedent to the salivating and the bar-presses respectively. The dog, on hearing the metronome, predicts that food will come, and the rat predicts that pressing the bar will give him a pellet.

The expecting, expectation, expectancy, causes and explains the salivation. Now, one can draw a diagram, and a box in it is labeled “expectancy.” Let’s have a look at the folk language of “expectancy”–first having a look at the connotations of the word “expect” and its derivatives.

Etymology identifies the Latin roots of “expect” as: “look out, look away from.” Roget’s Thesaurus gives us Exhibit Two, the names of the headings, categories, or sections within which the expect words appear. “Expect” deals with statements describing events that have not yet been observed. But, the word expect clearly has connotations. These connotations, unfortunately, slide into the cognitivists’ use of the term after they have defined “expectancy” theoretically.

The “expect” words:
Headings of sections in a Roget’s Thesaurus in which they appear

  expect                 expectation                expectant

  suppose                expectation                pregnant
  think                  imminence                  anticipant
  anticipate             anticipation               eager
  intend                 intention                  hopeful
  hope                   hope
  NOT expect             unastonishment

             expectantly            expectance, expectancy

             anticipatively         expectation
             hopefully              (estate)

So, how do we talk or write about such not-yet observed, ‘expected’ events? Exhibit Three presents an array of such statements.

(three things to be done)

swimming next Thursday
(to) see a movie next Thursday
to the movies next Thursday


vernacular concept root of concept I am going…. I am going to go… I will go… will choose, need, desire I shall go… I plan to go… planning floorplan I anticipate going… anticipating take up before I’m telling you that I’m going… I predict I’ll go… prediction say before I foresee I’ll go… foresight see before I intend to go… intention direct/hear I expect to go… expectancy look out for I’m looking to go…

    B                          		 C                      D

I want to go...                     I may go...             I need to go...
I count on going...                 I might go...           I ought to go...
I'm looking forward to going...     Perhaps I'll go...      I should go...
I hope to go...                     Maybe I'll go...        I must go...
I'm eager to go...                  I'm thinking I'll go... I have to go...
                                    I've a mind to go...    I'm forced to go...
    E                               I guess I'll go...
                                    I think I'll go...
I go/see... every Thursday          I suppose I'll go...
I go/see... every day               I believe I'll go...        F
It's my habit to go...              I imagine I'll go...
I usually go...                     I'm looking to go...    From now on, I am going...
It's my custom to go...             I reckon I'll go...     Beginning now, I am going...
My rule is to go...                 I figure on going...    My New Year's resolution is to go...
I go regularly...                   It's possible that I'll go...                                  

(No need to expand on such statements as “When I go to the movies, I’ll see Emma,” or “When I press the bar, I’ll get food,” and “When I get to the car, I’ll jump in when he opens the door, and I’ll get to ride.”)

Part of each statement identifies it as a description of something that has not-yet happened. And the descriptions of the events vary wildly in their specification of the behaviors that the speaker or writer states will be observable, and hence equally vary in “verifiability.”

As Exhibit Three shows, they can be sorted out, and members of each group can be scaled along identifiable continua. Some of these statements can be identified as mands of the speaker/writer given to the speaker/writer him or herself (‘reflexively’). Some specify events that have previously been defined as aversors, and so on.

Among the many words that turn up in Exhibit Three, especially interesting is the word “plan”–both a noun and a verb. Here, we must note that “plan” refers to a set of instructions not only to the individual who plans, but to others. Oneplans to drive to San Francisco, so he calls AAA, which sends him or her a Trip-tik, a plan–a set of instructions that may be complied with or not. The architect draws plans jointly with his or her client, but it is the contractor who complies with these instructions.

I invite your close and thoughtful examination of these last two exhibits, particularly from the point of view of noting the range of behavioral, not to say philosophical, connotations of the words used. When one says “I intend” to do something or other, the concept ‘intentionality’ of phenomenological philosophies intrudes….

Note, too, that it is possible for either the person addressed by the speaker of such sentences or the speaker himself to be able, on the following Thursday or later, either to confirm or to disconfirm the prediction of behavior, in other words, to evaluate its veridicality. I cannot refrain from pointing out that this question of confirmation, of assessing the veridicality of statement about future behaviors applies to statements about past behaviors. The behaviorist should wonder whether many variables function alike with respect to events that have occurred prior to the present, the “now”, the time of observation, and events that may or may not occur later, in “the future.” The able clinician, we now know, can alter both rememberings and expectings.

The cognitive concepts “memory,” which Prof. Sato has investigated and the now innumerable subclasses of this repository of the past and “expectancy,” a concept relatively uninvestigated by cognitivist, have now been alluded to. A behavioral view of the Concept named “concept” suggests that a descriptive system that does not a priori embody folk-psychological terminology will serve psychology well in ordering our data.

Analysis shows that Concepts such as “expectancy” are inadequate indeed, whether for the control, description, prediction, explanation, or understanding of behaving. Most cognitive concepts, including of course all those we have listed plus a whole army of other concepts derived from them, will prove to be conflatory. Dr. Sato’s “Behavior analysis of memory” shows one such conflation. Remembering, as a broad, ill-defined class of behaviors, refers to no more and no less than a whole family of statements related to past events; these must be as subject to confirmation and disconfirmation as are those related to future events. What people tell us happened to them in the past may be no more or no less verifiable than what they say they will do at some future time. This writer suspects that, after reading or parsing the literature on remembering, cognitivists will find themselves forced to discover or identify as many different “memories” as there are classes of behavior, that is, interactions of stimuli and responses, and of experimental probes of those behaviors.

The Concepts and Concept-sets defined in behavior analysis are a part of behavior. They are shown by dumb animals, as well as by farmers milking cows and telling us what they are doing, by the landscape gardener when he shows his customer shrubs that will not grow too large, and scientists writing of lactation, stimuli, paleolithic tools, and the double helix. Most bear names in ordinary language that are implicitly confusing in that they too often appear in, are members of, more than one Concept, or that all members of too many Concepts and Concept-sets are words.

Concepts are not “mental” entities or neurophysiological events that are found in the head. They are behavioral events that are as identifiable as trees, sodium chloride, tornadoes, work done, lightning, pasta, weaning, and sunrise; they are as publicly observable as any of these, and as measurable, with dimensions of N, the count of instances observed, and of space and time. As the Concepts have developed in the physical sciences, they have evolved, just as investigators find species, languages, and customs evolving and the behaviors of a child developing.

This account of “Concepts,” of Concept-names and sets, is consistent with terms in logic, such as “disjunctive concept.” It underlines the proposition that philosophy, insofar as it neglects the fundamentals of observing and of defining each word by ostentation–pointing at and demonstrating, showing how or what–has become no more than an interesting and entertaining game of words with words.

Behaviorists themselves must be wary at all times of the role of one’s own native, the ‘first,’ language, and of the preexisting terms of the language favored by scientists in their own thinking. As scientists, they should be guided by data and inductive hunch or guess, and not by words. It shows that we must again assert the position that Skinner (19__) took in his paper, Are Theories of Learning Necessary?. Behaving should be categorizable and understandable in terms of other behaviors, in the same language and at the same level of analysis; one need not appeal to such entities as hypothetical constructs or intervening variables.

Cognitivism, rooted in folk-psychology, is the study of behavior through the mediation of language, and not the study of behaving, what all organisms do, and what we humans say and write. It is limited by both its linguistic base and the capacity of its experimental and theoretical methodology to enable a proliferation of concepts that are not often complementary, but too often conflating. That cognitivists may be reaching the same conclusion is given by the following quotation: “Nevertheless, it is important to understand what the motivations for the cognitive revolution, as it developed outside the bounds of university psychology departments, were–exactly which problems it was thought to solve and which problems it was thought to simply leave behind–especially since we seem, once again, to be running into difficulties on the very issues that led us to abandon mentalism in favor of behaviorism, now almost a century ago.” (Green, 1996) That writer is not alone in expressing misgivings on the future of cognitivism, and he may well be on the road to abandoning it. He will not be the first of cognitivists to do so.

The complex behaviors that cognitivists study may be systematically investigated using only ostensively defined concepts. The ordinary methods of the operation- and behavior-analytic system may be expected to perform satisfactorily in the experimental investigation of these behaviors. As the science of behavior progresses, one may expect that some of the scientific concepts to be derived from behavioral investigation may be comparable, parallel or similar to those of cognitivism. This is to be expected if there are systematic regularities among behaviors that transcend both species-memberships and the language in which concepts are described.

Is there a future for cognitivism? The answer must be a qualified “yes.” But cognitivists must first sort out, for example, the several (or more?) concepts all sharing a Concept-name; “consciousness” and “mind” provide good examples. They must recognize that as they find more and more kinds of “memory” (and the like), qualifying “it” by adjectives serves only as face-lift for a scientific concept long past its prime in usefulness. For ‘memory’ and the like, cognitivists need to develop concepts relating specific behaviors at the time of investigatory probe to specific behaviors before the probe, taking note that the probe operations themselves are one of many variables determining the results.

When cognitivists inquire into the nature of the “information” that the ‘brain’ or ‘mind’ ‘processes’, and of the ‘choices’ made, they will discover that they must specify instances of both an environmental event and a second environmental event produced by an instance of an action of the individual and interactions among them of stimuli and responses, by whatever names. The temporal spaces between are ‘filled’ with many events, but none of these are summarizable in neat static “flow charts” of neatly labeled boxes entangled in and linked by carefully placed, neatly labeled arrows. These specify both nothing and anything.

As these changes (now beginning) in the research and theorizing of cognitivists continue, lo and behold, they will become behaviorists, who, as we have, will leave folk-psychological concepts to ordinary talk, relying on data, not words, for scientific conceptualizations.

Perhaps, however, that will not happen until they have produced cognitive and neurocognitive concepts in number equal to the neurons that already occupy the skull that houses the brain and is presumed to house the infinitely extended Cartesian mind.

Operation-analytic behaviorists, we may be sure, will continue to investigate behavior qua behavior inductively, and to employ their findings as they may be needed. We, fully expecting today’s cognitivists to join us tomorrow, will meanwhile consider their conceptual systems as one kind of verbal behavior–which is itself a special, although important class of the many culturally-determined behaviors. The cognitivism of today, it might be said, is no more appropriate as a basis for scientific generalization than spit-balls are for aerodynamics.

An addendum: Concepts derivable from ‘folk-psychology’

Two sentences summarize what I have said: Cognitive psychology, based on the folk-psychology of English, German, and other contemporary languages of Western Europe fails. A science of behavior, ‘psychology,’ must be based on the observation of behaving, with each term used in description and generalizations from such descriptions, defined ostensively, by demonstrating.

Those who base their conceptual systems upon folk-psychology have a great deal of latitude in their choice of Concepts, especially when they use the English language, with its large vocabulary. Folk-psychology, however sanitized by the cognitivists’ scientific methodology, will surely be a function of the native language of the observer, investigator, or theorist and will continue to be so in the absence of ostensive definition of the words they use. These are parts of verbal behavior and are culturally determined by the contingencies of reinforcement that eventuate from their use within their linguistic community. Following Skinner’s argument of 1948, this is especially the case when words relate to “private stimuli,” those descriptions of events observable only by the person speaking or writing–the terms related to the “mental” events of cognitivism. It is no wonder that “consciousness” is so flagrant a conflation.

Cognitive concepts, that is, are products primarily of the orderliness of verbal behaviors and of the contingencies that control them, and only secondarily of the orderliness of behaviors in which words play no part in the interaction of the individual with the environment. They are inevitably anthropocentric–the extrapolation of human-derived concepts to other species–and will lead to errors of both omission and commission in describing the behaviors of species other than Homo sapiens as well as Homo sopiem.

Nevertheless, operation- and behavior-analysts need not avoid all use of the terminology of folk-psychology in ordinary speech. But they must always be wary of, being misled by it. We can continue to say “I’ll make up my mind,” “I’ll take it to heart,” “I’m thinking about what I’ll do today,” and “Imagine what it would be like to be a bat.”

Folk-psychology, the language of ‘common sense,’ ranges from that of the illiterate, through the slang and jargons of specific groups, up through the remarkable characteristics of informal, formal, scholarly, and scientific languages. It is indeed a proper study for behaviorists. Analytic research can disentangle the culturally determined parts of verbal behavior from those that are contingent only on non-verbal variables. This indeed has been done with remarkable success in sciences other than psychology.

A start in this direction has already been provided by data generated by those working in other fields–notably in etymology, lexicography, and related fields.

The behaviorist who pursues these questions must have at hand sources such as, in English, or their equivalent.

(1) The Oxford English Dictionary, which traces the evolution of lexical “meaning,” and usage. It attempts to state in other words the concept or concepts named by the word.

(2) C.D. Buck’s (1949) A Dictionary of Selected Synonyms in the Principal Indo-European Languages: A contribution to the history of ideas. Univ. of Chicago Press (Available in paperback)

(3) One or more editions of Roget’s Thesaurus.

(4) One or more etymological dictionaries.

(5) One or more dictionaries of slang.

The investigator who uses such resources as the OED and Buck’s classic work will compulsively browse, and will not fail to be impressed by the evolution of “abstract” terms from “concrete” ones–from words that are names of point-at-able objects and events.

Consider the relationship among the following concept-names, their related (and often conflated) concepts, their roots, and their usage. Start with the OED, noting each usage it records.

count comptroller control reckon account add up calculate account (for) number figure on tell figure out estimate

How is it that so many words, with so many usages and related to such a variety of concepts and activities, all relate back to counting?

The roots of these words, these Concept-names in one or another language antecedent to English, all have to do with a specific behavior, counting–pointing at and pairing with a number, successive instances of a class of events. Perhapscounting, “enumeration,” is accountable (!!) for many of the behaviors loosely termed “thinking.” I’d guess so. That’s a start.

Cognitive abstractions, “theoretical terms,” are dealt with as the pick of the litter among concepts by many psychologists, to the chagrin of behaviorists. From the first comes “abstraction”–and chagrin comes from “a burr under the saddle.” Looking into the evolution of these words, inspecting their roots, almost invariably leads us back to the name, the Concept-name of specific behaviors. The term “metaphor” comes to mind.

So, too, in slang and in informal speaking, we find the names of the activities of individuals replacing the “abstraction.” People with courage have lots of heart; they show real guts if they continue behaving, even though they began with butterflies in their stomachs, wound up tight, then broke out in a cold sweat and finally dirtied their pants. They don’t refer to mentalisms, such as “anxiety,” “emotion,” or “fear” when they are afraid. (For these words, I refer you to Buck’s classic.) The folk-psychology of slang, be it noted, is clearly behavioral, neither mentalistic nor cognitive. Most noteworthy of all, the James-Lange ‘Theory of emotion’ seems relevant.

Summing up, behaving will be understood if, as, and when we start with the point-at-able, the demonstrable, the ‘I’ll show you what’ and ‘I’ll show you how’ and from these, inductively develop a set of scientific concepts. These will enable the prediction, ‘control’ (qualified probabilistically), and understanding of behaving. This may enable a Baconian taxonomy to emerge. Cognitivism, pursuing a different methodology and based on folk-psychological, ‘ordinary language,’ ‘common sense,’ concepts, is nearing the end of its course. These cognitive concepts and their folk-psychological antecedents are a proper subject matter for investigation by behaviorists interested in Comparative Linguistics.

Finally, is it an accident that an investigator who extensively observed and recorded the behavior of young children playing with one another found that he was able to describe and enumerate their behaviors (other than verbal and facial activities, ‘expressive behaviors’ which he could not observe) wholly in terms of such monosyllabic, ostensively defined concept-names as ‘run,’ ‘hop,’ ‘jump,’ ‘hold,’ ‘throw,’ and so on? Those monosyllabic words alone, complemented with specification of time and location of the kids and the objects they played with (manipulanda), provided a script.

The study of language–or rather, of the evolution of the relationships between ‘words’ and the discriminative and reinforcing stimuli and responses that control them, and the responses that they in turn control–will shed light on many behavioral processes and sets of behavior.


1This is the concluding statement of a paper in which the author, dealing with the “theory of mind” of chimpanzees and four-year-old children, finds that, with “a few notable exceptions,” cage-raised chimpanzees cannot do what the home-raised children can. The paper is a nice example of the mind-sets (a good folk-psychological term of the vernacular) that cognitivists bring to scientific research. They blind themselves to variables that behaviorists have found important.

2 Here, the pot may be calling the kettle black: Experimental behavior analysts are prone to limit both the responses and the measures of them that their subjects can show.

3 For many years, the writer has taught classes that he has no objection to the use of anthropomorphic descriptions or statements of ‘animal behavior,’ provided that it is not considered to explain that behavior, any more than it (folk-psychology) explains human behavior. My dog is jealous of my cat and of other dogs she knows. But ‘jealousy’ does not explain the behavior–and ‘jealous’ is a very inexact description.

4 Koch (19..), in a fine paper on problems of definition, properly emphasized and reemphasizes the crucial importance of ostensive definition of basic scientific concepts, accurately pointing out that this writer, in his Glossary (1957), did not succeed in approximating such definitions to his satisfaction. Well, I do the best I can, with words.

5 The foregoing statements will be recognized as members of the Concept in philosophy whose name is “Aristotelian Nominalism”; they are not instances of the concept named “Platonic Realism.”

Creative Commons License is licensed by Admin under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 United States License.