William S. Verplanck
What I’m going to talk about is my own fifty-seven years of searching among behaviorisms. It will hence be difficult for me to avoid making this a reactional biography, a personal history. I necessarily restrict myself to my possibly idiosyncratic analysis of the various behaviorisms and behaviorists I’ve encountered through that period. I will begin and end with the same topic: the terminology that psychologists who are behaviorists use, which determines the data we gather, how we deal with them, and how we report them to others.
At the University of Virginia
In 1934, I took my first course in psychology at the University of Virginia. Woodworth’s Psychology was the textbook; we had additional readings in Garrett’s Great Experiments in Psychology. I was bothered by a discrepancy in how these two Columbia psychologists used the term “aptitude” and assumed that if I took a second course, I would learn what aptitudes really were: Psychologists, as scientists, surely would know. That was in ’34-’35. (In ’95, people still don’t ‘know,’ except in terms of statistical analyses of how people answer selected questions.)
So, I became a major in U.Va.’s Department of Psychology, with Frank Geldard as Chair and Dick (G. R.) Wendt as the only other member. Wendt, a Woodworth PhD, gave a full-year course on “social psychology.” (The second time he gave it, he did not repeat himself.) Reflecting Woodworth’s empiricism, he covered everything except the kitchen sink. For Dick, all behavior is the subject matter of psychology–a doctrine I’ve never forgotten. Geldard, a product of Clark, a student of John Paul Nafe, and hence a grandson of Titchener, taught E. G. Boring’s version of operationism and carried out experimental research on the cutaneous senses following the strict protocol of Titchenerian introspective structuralism.
By the time I was a senior, I knew all about the “stimulus error” and had been trained to serve effectively as a subject in research on cold sensitivity who gave approvable, introspective accounts of the sensations I experienced: “dull,” “diffuse,” “bright,” “intense” were part of my vocabulary. Frank Geldard was broadly read; he gave a course on Systematic Psychology with Boring’s “Physical Dimensions of Consciousness” as basic reading. He also presented the behaviorism of Watson and Lashley with their exclusion of introspective data (but not the “subject’s report”) from a science of behavior coupled with emphasis on explanation in terms of physiological mechanisms. Watson’s behaviorism, I learned, was a reflexology. I found this psychology more acceptable than structuralism, functionalism, or Gestalt Psychology. Geldard only alluded to Weiss and Kantor, who were characterized as ‘far-out’ behaviorists of minor importance. (He left the impression that they were both figuratively and literally dead.)
Wendt left Virginia before my first year of graduate study; a new faculty member appeared. His first year of teaching graduate students matched my first year of being one. This newcomer was Kenneth W. Spence and all five or six of us graduate students took all of his courses. “Uncle Kenny,” as we called him (behind his back), was indeed avuncular. He led us through Pavlov’s “Conditioned Reflexes” almost line by line, relating it in detail with Clark L. Hull’s theory of behavior and the research of his students–with special emphasis on his own extension of it, a theory of ‘Discrimination Learning.’ This theory confronted Krech’s theory and data; this was developed within Tolman’s behaviorism. Tolman’s vocabulary, derived from ordinary language, added some neologisms. We noted that it avoided the physiological terminology of Hull and Spence and could properly be termed ‘mentalistic.’ Both Hull’s and Tolman’s theories, we students became aware, showed–proved–the importance of theory in psychology. Hull’s theory, we concluded, was truly scientific, so I became a “Hullian,” without full exposure to the logical positivistic doctrines that underlay the work of both Hull and Tolman. That was to come later. In that same year, a minor in philosophy exposed me to Aristotle, as construed by St. Thomas, and then by the neoThomists.
My master’s thesis reported research on the cold sensitivity of the cornea; it was a test of John Paul Nafe’s theory of cold sensitivity. I reviewed the physiological literature, learned a good deal of physiology and, most importantly, learned for the first time that those who have developed a theory and work to advance it by surveying the relevant literature are very likely to misconstrue any finding not consistent with the theory and to seize every ambiguity as supportive of it.
So, I took a Master’s, having been persuaded by Wendt that all behaviors, of both animals and people, were the subject matter of psychology and were closely related, and having been thoroughly grounded by Geldard in structural psychology, both in theory and practice. Spence made me even better informed on the Pavlovian behavioral theory of Clark L. Hull, and convinced me that this was what Psychology should become. Exposure to the mentalistic language of Tolman’s behaviorism and to the glib physiological excesses of Nafe’s vascular theory left an ingrained suspicion of certain kinds of terminology, and of theories based on the physiologizing of some psychologists.
In 1937, “real” jobs were hard to find, so, with offers to remain at Virginia, to go to Harvard, to Yale, or to Brown, where most of Clark’s faculty had moved, I followed Geldard’s advice to work with Clarence Graham, his good friend, as a research assistant on his visual research program. Frank Geldard knew the style and social context of the Brown Department; I’m grateful to him for that recommendation. In the fall of ’38 (preceding a hurricane by a mere 4 weeks), I arrived at Brown to work with Clarence Graham. There, I encountered a new world of behaviorism and was to become a psychophysicist, in the field of vision.
At Brown University
The Department of Psychology at Brown was Walter Hunter’s department. Small in number of faculty and of students, the department was a cohesive group, maintained as such by the daily coffee-time in what had once been the kitchen of one of the two Victorian houses in which the Department lived. Walter S. Hunter, the “Old Doctor,” was a behaviorist of the Watsonian school–a man who viewed psychology as a natural biological science, separated from physiology literally by the skin. His leadership, the stamp he set upon both faculty and students was unequivocal. Psychology he conceived (and practiced) as closely related methodologically to the laboratory sciences of physics and chemistry, and to such biological sciences as neuroanatomy and neurophysiology. He was the experimentalist, not the theorist. He believed in “processes”–as exemplified by a single process of learning, to be accounted for by the function of the brain, conceived as a telephone “associative” network. He admired Ebbinghaus’ research. In this context, he had carried out his controversy with Lashley, who had put forward the concept of “equipotentiality” in brain function. Hunter’s experimental work while I was at Brown was on the behavior of the rat in the double-alternation maze, a problem stemming from his important earlier research on the “delayed reaction”–a “symbolic process.” His behaviorism went beyond the reflexology as it had been presented at Virginia.
Hunter’s carefully controlled experiments demonstrated what I understood later as the limitations of research based on mazes. It seemed somehow to incorporate a proposition that I later viewed as built into Hullian theory: that rats in mazes, elevated or otherwise, had receptors incapable of responding to stimuli presented more than one inch in front of their noses.
Hunter taught us–he pounded into our heads, as did others of the Brown faculty–that one learns, not by studying secondary sources–texts or edited collections–but by analyzing, line by line, research papers that had already been recognized as having made a substantial contribution to the science. He had students read and dissect both the logic and methodology of each report of experimental research. Hunter, I would say, was the exemplary Watsonian behaviorist, one who did not hesitate to investigate behaviors that would now be termed “cognitive.”
The second kind of behaviorism I encountered at Brown was that of Clarence H. Graham, my mentor, “the Young Doctor” (and still my scientific parent). He’d been a student of Hunter, as well as of John Paul Nafe, and early on had written a paper of consequence entitled something like, Psychophysics and Behavior. In this, he subjected the classical psychophysical experimental methods, in which I had been fully indoctrinated by Geldard, to analysis in terms of the science of behavior. Graham’s behaviorism reflected a considerably greater sophistication in the scientific methodology and logic of the thirties than did Hunter’s; it wed structuralism’s methodology (but not theory) to behaviorism. Graham functioned and taught his behaviorism in a broader context: His students read widely; we had to be fully familiar with the problems of methodology in measurement, whether in the physical sciences or in the behavioral sciences. We read Bridgman and Humphrey instead of about Bridgman and Humphrey. We had to learn a good bit about mathematical theories and what they might accomplish, not excluding D’arcy Thompson. We learned the doctrines of logical positivism, or as it was becoming renamed, ‘scientific empiricism’. Hecht’s mathematical theory of vision and its associated research were accepted as examples of the best of science. We kept current with advanced theories and data on the function of the nervous system, subscribing to the electrical theories of nerve conduction and synaptic transmission. Loewi was out. Clarence, who had worked extensively with Keffer Hartline, was, to put it mildly, a reductionist, a devotee of the “one big science” conceived by Carnap. This provided a rationale for studying vision: the retina is embryologically a part of the cerebral cortex, and retinal function should provide clues on cortical function.
As a reductionistic behaviorist, Graham was nevertheless open-minded. His broader interests were reflected in that the largest block of sales of the first edition of Skinner’s Behavior of Organisms went to us graduate students and faculty of Brown. Graham had already grounded us on Skinner’s first two (still most important) papers, on the reflex and on the generic nature of stimulus and response. Indeed, Graham took a shot at developing a mathematical theory using Skinner’s concepts. A good Hullian, I was nevertheless receptive to the broader Skinnerian conceptual system. By now, my interest in behavior was outstripping my interest in vision. My dissertation, using rats on a straight runway, bore earmarks of both Hull and Skinner. Perhaps forecasting my present viewpoint, its title was: The Development of Discrimination in Simple Locomotor Habit.
Clarence Graham, Walter Hunter, Don Lindsley, Joe Hunt, and Eddie Kemp were not my only teachers at Brown. Fellow graduate students kicked problems and concepts around. We worked together, talked together, discussed Guthrie together, and learned from one another. Faculty and students were a single social academic group–the morning coffee house kept us thoroughly mixed. The Seminar Room was also the Departmental Library. I browsed. I skipped from journal to journal, from book to book. Wendt, at Virginia, had branded me; I did not find a single topic without interest.
So, as I left Brown, I was the convinced experimental psychologist, a behaviorist dedicated to research on behavior, well grounded in psychophysical methodology and in the study of the visual systems–a reductionist, a disciple of the Behavior Theory of Clark L. Hull, and persuaded that theory was the driving force, as it were, of productive research. Four further behaviorisms were now in my armamentarium: Hunter’s, Graham’s, Skinner’s, and Guthrie’s. Then Pearl Harbor.
At the Naval Medical Research Lab, U.S. Submarine Base
With a brand new PhD, I began 4 years, 4 months, and 4 days at the U.S. Submarine Base in New London. As a psychologist in uniform, I was confronted with research problems of the real world, problems not successfully dealt with in an experimental laboratory. My colleague there was Neil Bartlett, who had been a fellow graduate student at Brown. After a time, we sorted out the problems that needed to be worked on germane to submarine warfare. I found myself working in the area of vision, specifically on the training of lookouts, training sailors to observe and to search at night when foveal function was lacking. Basic to this was selection–the development of an instrument and a psychophysical method that, when administered by technicians, would demonstrate as quickly as possible whether a given individual could see well enough at night to detect and possibly identify distant targets. And here I came to understand that in visual research, the mode of responding and the responses measured were significant variables in determining whether somebody could see something; it wasn’t just a matter of eyes.
At the same time, I learned a little bit about what people mean by “motivation.” You had to “motivate” look-outs. Looking back I understand how we did it: If lookouts weren’t good, they could get killed. If they were good, they could add a number to their boat’s record, a little Japanese flag to its conning tower. I found, too, a behavioral method for catching malingering naval pilots who were afraid of duty on aircraft carriers, and even more afraid to admit their fear. Being “night-blind” was their way out. This experience set me to thinking about the response in psychophysical research, for it was by manipulating responses that the method caught them out.
Most of all, I learned that if you’re really going to learn about behavior, unless you settle for a kind of myopia, you have to do research on how people behave in the ‘real’ world, that is, the world in which we all live every day, and not solely in the restrictive environment of a laboratory. Neither what is theoretically best nor what works best in the lab is necessarily the best elsewhere.
From this came a field study: the collection of data on the behavior of real sailors searching for, finding, and identifying real targets using real binoculars on real ships. Complicated planning, execution, and data analysis yielded convincing results. Managing of staff, officers, and enlisted men was enlightening. This intensive education showed me that it is possible to carry out successfully a satisfactory experimental design tailored to the restrictions set by the world outside the laboratory. (This stood me in good stead in later years.)
Then, World War II ended and about a year later, I was discharged. Those years put my behaviorism to the test; I found that it worked.
At Indiana University
Visiting the University of Pittsburgh for a job interview, I was advised that if I offered the course in experimental psychology and proposed that students measure the visibility curve, I should have the students learn how to “do science.” That is, I should require them to “state an hypothesis”, gather the data, and then “test the hypothesis” statistically. This fatuity was my first lesson in the cookie-cutter doctrine of “scientific research.” It made no sense to me and ignited a skepticism in such doctrines.
I was fortunate; I wound up at Indiana University in what were to become that Department’s greatest years. Through those years at Indiana, I remained a total, unreconstructed Hullian. My faith (and it was a quasi-religious faith) was only slightly jarred by the Hull-Yamaguchi paper on the measurement of E-bar-dot, which required, as I recall by actual count, 28 consecutive assumptions, not all of them plausible.
At the same time, I was a very close associate, both administratively and socially, of Fred Skinner, as well as of his students. (Disciplehood had begun.) Some of these, I would attempt in one way or another to convert to my Hullian faith and to the eternal verities of logical positivism and the doctrines on research developed by philosophers. I gave a course, God forgive me, on “Scientific Method and Theory Construction.”
Necessarily, I followed closely what all the Skinnerians were up to and was an interested onlooker as the Skinner box and EAB research style evolved. I talked regularly with Bill Estes, my closest friend through those years, telling him about W. J. Crozier’s statistical theory of vision. (Crozier, brilliant student of Jacques Loeb, whose students included Hecht and Hoagland, and who was Skinner’s patron/host at Harvard, was a constant, remote, and important background figure through all these years.) I got moral support from Clete Burke, the ultimate Hullian; from Sid Bijou, then another Hullian; and from Bill Jenkins, who had been at Brown, but finished up at Yale and represented Thorndikian research with a Hullian touch.
My own research with George Collier and John Cotton was twofold: One was a complete and utterly ultimate factor analysis of all the data gathered at New London on sailors seeing ships. This demonstrated that factor analysis can be a lot of fun and that nobody should ever get hypnotized by the p = .20 view of acceptance and rejection of a “hypothesis.” (One fifth of our armies of correlations were significant at that level.) George, John, and I collected data on psychophysical methodology that showed very clearly that the “judgments” of an individual, the responses of the individual to each successive stimulus presentation in “threshold” measurement, were not solely dependent on the physical properties of the stimulus. Each successive response was also dependent on the preceding responses. An assumption basic to psychophysical measurement was invalid.
As a good Hullian, I followed in the footsteps of Clark L. Hull, doing casual research on hypnosis and suggestibility, and discovered how easy (and misleading) it can be to bring a subject under one’s verbal control by this methodology. Confirmation of a suggestion by the subject’s response was critical–it was a reinforcer. More, subjects would do things when in a hypnotic state that in ‘real life’ it would never occur to them to try. When they found they could do it without “hypnosis,” they were astonished.
But perhaps the most significant contribution to my later thinking and research came from a graduate student. This was Parker Lichtenstein, who was accustomed to dropping by the Department office at night when I was there in my administrative office next to Fred’s. Standing in the door, he had me as a captive audience and systematically tried to explain what that fellow down the hall named J. R. Kantor was talking about. I listened–I had no easy alternative, except to return to dull paperwork. I argued, exhibiting the great skepticism that any good logical-positivistic Hullian, dedicated to the revealed scientific methodology of the laboratory, would show to the writings of this fellow who wrote all those unreadable books. I, for one (not to say Fred and all other faculty members), didn’t understand his writings and wasn’t prepared to try. We couldn’t understand–we already knew about scientific methodology. (Eventually, I was able to report on a confrontational seminar lead jointly by Skinner and Kantor.)
As an aside, one of the secretaries I had during those years was a daughter of one A. C. Kinsey. From Kinsey and a few of his interviewees, I learned more about what you could and could not find out about people by asking them questions, if you knew how to do it. More, his basic research on the taxonomy of insects taught me something about classes, sets of events.
Thus, at Indiana I developed a first-hand acquaintance with the methodology of the Skinnerian behaviorism that was later to become The Experimental Analysis of Behavior, a fuller grasp of his approach to behavior, and achieved a small background in the interbehavioral teachings of Robert Kantor. And Clark Hull’s behaviorism no longer seemed to have all the answers. Most important for my own future work, I had also reached the conclusion that the more I knew about vision, the less I knew about seeing. I was “into” behavior and out of the psychophysiological bent of visual research.
Through these years, several meetings of behaviorists at other Universities were held. At these I made acquaintance with Fred Keller and his group at Columbia. As disciples, they were more devout than Fred’s own group at Indiana.
The year Fred was away from Indiana giving the William James lectures at Harvard, I served as Assistant Head to Robert Kantor. I came to know this man well personally and observed how he went about his work. In due course, it came about that Fred accepted an appointment at Harvard. After a delay, I also accepted an appointment to start at Harvard in 1950.
At Dartmouth, Summer of 1950
At those I. U. meetings of behaviorists from other universities, some of us had a dream about behaviorism, learning, and theory. It came true in the summer of 1950. As an outcome of the interests of us dreamers, A. T. Poffenberger was able to arrange SSRC funding for a conference to be held throughout the summer of 1950 at Dartmouth University. Seven well-qualified people met to examine and to evaluate five of the then-current major “theories of learning.” From these efforts, the book, Modern Learning Theory, eventuated. Through that summer, I was able to exchange words and “thinking” with a number of extraordinarily talented contemporaries. As a consequence, I became more conversant with a number of theories, all but one a behaviorism, and with the behaviorism of Bill Estes, Ken MacCorquodals, Paul Meehl, and Connie Mueller, and with the anti-behaviorism of Sig Koch–as methodological critic.
Mornings, we talked, reviewing with one another the previous day’s work and library research. Afternoons, we read, making notes. We assigned to one another responsibilities for full analysis of the theories we thought were important. We collaborated in writing the preface, which can be taken as the absolute, end-of-the-world, ultimate summary of the standards upon which all evaluations of theory in psychology should be based. This preface probably remains the most complete, yet succinct, summary of the “received wisdom” of theory construction–it is easy to relate “intervening variables” and “hypothetical constructs” to its statements.
Hull’s theory fell victim to Sig Koch, the critic whose high intellectual functions have never been sullied by thought of constructive criticism, but have enabled him to remain unparalleled as a critic. (Still a Hullian then, I thought him a nit-picker.) Guthrie, Tolman, and Lewin (the only non-behavioral theorist) came under similar, but less compulsively detailed scrutiny. I learned a lot about them and found myself further attracted by E. R. Guthrie’s down-to-earth, nuts and bolts methodology and exposition.
My own subject was B. F. Skinner. I necessarily reviewed carefully and critically all he had published by that date and had had the benefit of working closely with him and his students. After analyzing the body of his work, I reached the conclusion that Skinner had not written a theory, but rather presented a systematic viewpoint that took a fresh uncharted approach to the development of an account of behaving. Even then, I identified Skinner’s as being related to another non-theory–so much a non-theory that we didn’t even think of considering it at this meeting. That was the interbehaviorism of Robert Kantor.
At Darmouth then, I was exposed in detail to a five-membered set of behaviorisms and to critical analyses of them. My confidence in both the specifics of Hull’s reductionistic system and in the gospel of theory construction was shaken, but remained more or less intact. None of them, except perhaps Skinner, fared very well at our hands. It took another year or so for me to write up my findings and to state clearly to myself and for others the difference between “theories” and systematic viewpoints.
Although of approximately the same size as the Brown Department, Harvard’s proved very different. The faculty were more diverse in interests and values, and its ethos was set, not by the morning coffee for faculty and students alike, but by Boring’s lunch-hour in the Seminar Room for faculty, distinguished associates, and visitors. It was informal, but Boring sat at the head of the table. Participation in this was effectively mandatory for faculty members, and the daily discussions ranged wide, with measured and rancor-free arguments on any and all issues. How could we help Tolman in his battle with MacCarthyism at Berekley? Could systematic reinforcement enable Percherons to win horse-races? What about euphoropsia? How’s the teaching-machine coming? Do the body-builds of Japanese ensure that they will dominate the Chinese?
At Harvard I soon ran into another behaviorism, one that, ironically, seemed at first to be no behaviorism at all. That was, of course, S. S. Stevens’ Methodological Behaviorism in its application to the measurement of “sensation.” That methodology was exemplary; its results showed me that a subject’s assignment of a number to an event that is presented to the eye or ear followed a behavioral method that was applicable to the measure of any behavioral event on which research is being carried out, and with which both subject and researcher are interacting. Stevens’ work, in short, provided behaviorists with tools essential for the measurement of events that, in the strictly “materialistic” and “reductionistic” context, cannot be measured.
A second kind of behaviorism that I encountered at Harvard for the first time had its roots at Indiana University. It was the don’t-look, don’t-observe mode of behaviorism in which, in the interests of “experimental control” of the environment, the behavior of the individual is reduced to a set of marks on paper made by a relay or equivalent; records whose marks are treated as “responses”–the real thing. (This behaviorism I encountered later at the University of Wisconsin, where a response was also defined as anything the subject does that leaves a mark on paper. In this case, however, the mechanical and electronic linkage between the subject and recorder could vary.) In this behaviorism, everything was done automatically: the “program” was set, the reinforcers and recorders functioned, and out came the records to be analyzed and ordered. I well remember the day when some wise-foolish laboratory assistant walked into the soundproofed laboratory where the genetically programmed pigeons were working and heard a strange sound that came at regular intervals. She went to the source of the sound and peeked through the tiny window into the box from which the sound was coming. This bird was making his or her marks on the cumulative record by retreating to the rear of the box and then hurling itself at the front of the box, thus activating the relay. This behaviorism I deemed as a peculiarly parochial behaviorism: It investigates the behavior of the individual by programming the behavior of the environment in ways most of which have minimal relationship to how real-world environments behave.
At some point along here, in communicating the futility of theories developed to account for the phenomena of “reinforcement” (e.g., “Drive Reduction Theory”), Fred remarked, “If anything, Edwin (Guthrie) is right,” or words to that effect.
Another behaviorist I encountered there was someone whom I had already met, but had only passing familiarity with. In the course of preparing the final draft of my paper on Skinner for Modern Learning Theory, I finally got acquainted with, more familiar with, or came to know (but still to a limited degree) the position of J. R. Kantor. The grounding in his thinking that I had picked up at Indiana was illuminated by working on the analysis of Skinner’s conceptual system. Kantor’s concept of the interactive relationship between stimulus and response began to make sense, as did the concept of the interbehavioral field and (perhaps most importantly) his views on the relationships between psychology, biology, and physics. When I later encountered Wittgenstein’s views as expressed in Philosophical Investigations, I found myself on familiar ground.
Harvard had little equipment for undergraduate experimental classes, nor were there funds for its purchase. The shops were busy–working on apparatus for Békésy, Skinner, and Stevens. I had to make do using what little equipment there was. I managed to put together a course on experimental psychology that embodied a graded series of experiments in which the students participated as both subjects and experimenters. The course followed straightforward progressively ordered “classical” experimental designs and procedures of data collection, with parallel progressively developing statistical and other methods of data analysis. Beginning with simple psychophysical measurement, the experiments became more complex, touching the basics as equipment permitted.
From putting together and giving this course, I learned that when one manipulates verbal materials as stimuli, they behave as stimuli, and that when they are manipulated as responses, they behave as responses. That is, the operations carried out by an experimenter are critical in the investigation of the interaction of stimuli with responses. These students were not asked to “test hypotheses,” but to describe or state what they wanted to find out about; they were not told what they either should or would find. Their research taught me that operant conditioning was exactly that, and that subjects can become conditioned even though they do not notice the behaviors they are so regularly emitting. They taught me that responses were identified by the way they function, and that their function depends on how they are used and not by an experimenter’s arbitrary choices. An environmental event that one chooses as a “stimulus” and an activity of the individual that one measures as a “response” may not function as such. They are “putative” stimuli and “putative” responses.
I encountered yet another behaviorism during those years, but only at second hand, thanks to a student. A senior philosophy major, star of that department, took my course in experimental psychology. After a few meetings, he told me that what I was teaching in classes was already familiar; I was talking in the language of Gilbert Ryle. So, I read The Concept of Mind.
Sure enough, Ryle was another behaviorist, a man who analyzed the vocabularies of ordinary language and of philosophical concepts in terms of the verbal behaviors of those who used those terms and dealt with those issues. (In 1995, I finally learned that a name has been applied to these endeavors: it is Logical Behaviorism.) Reading Ryle led me into reading Wittgenstein’s later works. These are far more behaviorial than not: they surely provided me with a fuller context for the evaluation of Kantor’s writings.
About then, in a problem on verbal conditioning following the Greenspoon method, a graduate student’s subjects’ high rates of saying “plural nouns” occurred as the subjects ran through a series of concepts whose members were plurals. That is to say, concept-sets appeared.
Through the first two and a half years at Harvard, rats taught me that theories of drive based on logical and physiological considerations, as in Hull’s and Tolman’s, are deflated when one examines the eating and drinking they are derived from. Other unexpected results turned up: simple methods and simple apparatus, with “simple” animals whom one could watch, showed, for example, that if a rat is presented with a movable stimulus (an S-delta) during periods of nonreinforcement of barpressing, s/he will first accidentally remove it by vigorous attack, but in very short order will settle for an efficient and calm removal. Such results suggested strongly that the methodology that the behaviorists practiced was remarkably likely to lead them to overlook or to ignore a variety of behaviors.
Through this period, Frank Beach, yet another behaviorist (Lashley-trained) was visiting, and I spent a good deal of time with him. He, too, had found that the “rigorious”, “carefully experimentaly controlled” research on the copulatory behavior of rats was not only misleading–it had been misguided.
The program of research on the sequential dependencies of response found in threshold measurement continued. The ultimate apparatus, designed to deliver insofar as possible physically constant short-duration visual stimuli neared completion. Final analysis of data collected at Indiana on several hundred students “guessing” showed that successive “guesses” showed the same response-to-response dependencies that had turned up in measures of the visual threshold.
In early ’52, the Harvard faculty was invited to a series of seminars at Tufts to hear the presentation of a recently arrived British biologist. He talked about a new brand of behavioral research–ethology. Ethological research was done in real life, in the field, with the subject’s free to do things with minimal or no experimental manipulanda. Fresh from completing the analysis of Skinner’s work, with a growing understanding of Robert Kantor, and concerned with the theoretical niceties of “stimulus,” “response,” “habit-strength,” and even “reflex reserve,” I found the research he reported remarkably parallel, both conceptually and in findings, to behaviorisms with which I was already familiar, i.e., Hull, Kantor, and Skinner. I had to go see. And so, I spent an eight-month sabbatical in Europe with the ethologists.
That seminar at Tufts opened the door to a new behaviorism, one that built on Walter Hunter’s course on animal behavior, which pretty well covered rat behavior, and on the browsing I had done in the more broadly based journals on the subject of the first quarter century.
Of the eight months abroad, most were spent in Tinbergen’s laboratory at Oxford with a number of several-day stays at his field research stations. Thorpe’s and Hinde’s work at Maddingly were visited. A couple of days in London, visiting J. B. S. Haldane, raised questions on the antecedents to “innate” behaviors. A several week tour of the continent took me to van Iersel at Leiden, Baerends at Groningen, Lorenz at Buldern, Otto Kohler at Freiburg, von Frisch’s lab in Munich, and Heinroth at his zoo in Zurich. (My visit to von Holst in Wilhelmshaven, although arranged through Lorenz, was a failure. Just hours before my arrival at his lab, he had suffered an acute attack of Anglophonophobia, an undiplomatic disorder.)
These were behaviorists. All gathered solid data, skillfully conducting simply designed research, whether on a sand dune, in a pond, or in aquaria that simulated their subjects’ “natural” ecological environment. They spoke a common language, tied ostensively to the behaviors they observed. They identified stimuli and taught how to recognize and identify responses that could be seen but could leave no marks on paper. Inclined to theorize in one way or another, their variant theories did not bridle their research, which was carried out in the broader taxonomic context of evolution. Their language, the terminology they used, however, differed substantially from that of the behaviorists I knew; it needed translation.
I observed gulls, ducks, kittiwakes, newts, geese, bats, Lorenz’s “Nazi-chickens,” chaffinches, and gorillas, and learned how to look and to see. I also observed the observers and how they interacted with their subject matter–and with one another.
I learned that there are different classes of responses interacting with different classes of stimuli–these differences being distinguishable not so much topographically as by measures of interaction between them. More, the relationships shown were dependent on context, and such concepts as “drive” and “instinct” did little but misconstrue.
In sum, I learned both that research results on a few species selected as “model”–whether rats, pigeons, or standard college sophomores–can lead only to a miniscience, however persuasive and elegant some of its findings might be. Behavioral research in the laboratory may produce, I came to think, something very like scrimshaw.
This visit, too, taught me much about the interrelationships among ideology, politics, and scientific problems and theory–more than one would wish to know. There is a politics of science that may determine both the problems investigated and the evaluation and interpretation of data. Scientific investigations yield results that have political consequences, and too often it is these planned political consequences that have determined the research that is done–the choice of experimental conditions found in previously published results that would most surely lead to the desired conclusion. (Does this ring a Bell curve?)
Back at Harvard-2
Back from Europe, my first concern was to bring the behaviorism of the ethologists into compatability with the behaviorisms I had previously encountered and to render those behaviorisms accessible to the ethologists. For me, that was a problem of vocabulary–so I embarked on writing a brief glossary of terms from each of these languages, defining them in a way that would be intelligible to both groups, a requirement that forced me to draw on all I had learned. The result was what I termed empirical–“operational”–definitions, following rules given in the Preface of that Glossary.
This endeavor began a long process of assimilation–digestion, if you will–and was the first step in the search for the development of a behaviorism–a behaviorial science–that continues to this day.
Of these, the straightforward methodologically behavioral research and the hard-headed approach of Smitty Stevens taught me most about research proper. At Boring’s daily lunches, I had learned to value most highly the thinking of Smitty Stevens (or rather, some of it). What he was writing then and published later, and what he did later has guided my own work in many ways. Explicitly, he forged ahead and found quanitative empirical laws without feeling constrained to develop a physiological theory to account for those findings. His constant admonition to investigators to “stay out of the grass,” to look at, and to look for the trees and the forests, was an apt metaphor. This I have taken as gospel.
Logical behaviorism, with its stress on the language we use in talking and writing, provided the tools for reporting–for communicating, for thinking–about behavior. The broad conceptual system put forward by Kantor has made most sense in “putting it all together”.
A summary: the first seventeen years of search
By the fall of 1955, then, I had encountered and evaluated many behaviorisms and had reached several conclusions. First, the standard “philosophies of science” reflected a kind of Sunday-morning quarterbacking that has often led active researchers astray. The emphasis on polished theory and on the need to develop new theories and to follow the strict guidelines of inferential statistical analysis of data to account for (“explain”) observations, rather than to rely on less formalized ‘guesswork’ and to use “rough and ready” common-sense methods of summarizing and ordering results, have limited our capacity to make fresh observations, ones that open new ground. Observations, ordered not only in their gathering but also in the language in which they are described, suggest new directions as well as new methods of research without assistance from philosophers. Operationism, that is, reliance on the procedures, the hardware and software of measurement, is critical for the clarification of the terminology of a behavioral language. Numbers persuade. Numbers show equivalence and lack thereof. Second, complex apparatus, like complex theory, equally militates against the development of an empirical science of behavior that has direct relationship to behaving as it occurs outside the laboratory. Research on behavior that can be carried out only with “high-tech, state-of-the-art” apparatus is likely to become research limited to the investigation of the behavioral capabilities of the apparatus rather than of the subject, just as research based on elaborate theories similarly degenerates to the investigation of the behavior of the theory rather than of the subject matter.
The third conclusion came from a quite different source. The student who introduced me to Ryle was not the only student I learned from. Students collectively taught me much on how to do behavioral research efficiently. From them, I learned, first, that students are capable of collecting data on one another that are highly reliable, and second, that the basic orderliness of behavior is accessible under what might be called variable conditions. Basic behavior is robust in its orderliness. Repeatable findings can be developed under a variety of circumstances certainly lacking the usual “experimental control of the environment,” provided that the procedures carried out in getting those findings are carefully specified and followed. The ecological niche of the student is the classroom, and in the classroom, one can carry out sound research. But there is a never-to-be-overlooked caveat: the student investigatiors must not be told what to expect. There must be no “hypothesis” to test. They must want to find out for themselves, “What will happen if . . . .”
In short, I developed what may be a different kind of behavioral research, one that grew out of all that I had encountered up to that time. The results of such research, done with students, can replicate the results of psychological research already published in journals. It can extend them. Or it can find them lacking due to over-design based on theory and on the convictions of the experimenter and of the assistants who actually conduct his or her experiments, who are “testing an hypothesis.”
One’s best investigative colleagues, then, are students. This served me in good stead over the next forty years, a period in which I found myself following this course. During this period, many of those students have themselves become behavioral psychologists.
So, I came to distrust elaborations, whether of apparatus or theory, in sum, to keep the execution of research as simple as possible and to use equally simple (but exact) methods for data analysis. Simple behavioral methods can (and should) be applied to the behavioral problems too often seemingly inaccessible except through heavy-handed theory, complex experimental design, and even more complex apparatus and statistical analysis. These define the grass that Smitty Stevens has counselled psychologists to stay out of.
A four-year “sabbatical”
Through the next four years, fall ’55-’59 inclusive, I necessarily put to good use what I had learned to that date. This was a challenging period, a sort of four-year sabbatical away from the facilities and ambiance of a graduate faculty. I did not have the usual opportunities to interact with fellow faculty members or to engage in graduate teaching. Nor was I subjected to the pressures and temptations of “publish or perish.”
Through a one-year appointment at Stanford, although faculty offices were adjacent, there was little interaction among their occupants. A short-termer, I had to make do with pick-up apparatus and to work with students, undergraduate and graduate. The second year found me circling the globe–to wind up with a second several-months stay at Oxford with Tinbergen and his group. The last two years at Hunter provided me with no resources other than large classes of very bright, able, and hard-working students. (Is it sexist to note that these were all women?) Other faculty members were nowhere to be seen. There, I learned even more about the capacities of students as a research resource and how to enable them to generate sound data. At Hunter, I discovered that the simple arrangement of terms that are empirically defined may produce what looks remarkably like a set of laws, a descriptive “theory” of behavior.
During these years, two summers were spent in Seattle at the Univ. of Washington, where Sid Bijou showed me his behavioral research with children–a first encounter with an ‘applied’ behaviorism. His own children showed me that intellectual labor can be a game to be enjoyed, with confirmation the sole reinforcer. One summer at Wisconsin demonstrated that the “dust-bowl empiricism” of that time and place was indeed dusty–dry and unproductive–and once again that graduate students are eager for innovative research and the opportunity to do it. My time there, however, was largely spent in getting the organization that became the Psychonomic Society started. This also took up most of my time when (in 1959) I went to the Univ. of Maryland as Professor. Four years later, at the Univ. of Tennessee, both the Psychonomic Society and new responsibilities as Department Head left limited time.
Searching and sorting at the Universities of Maryland and Tennessee (Knoxville) and since retirement in 1981
Through those years, I kept searching and, although I found no new behaviorisms, my suspicion that theories–whether Hullian or neurophysiological–serve primarily as blinders was amply confirmed, if only by the two distinguished behaviorists, both wed to theories, who proved unable to recognize the importance of any research that didn’t fit their favored theory.
I found no behaviorisms other than those I had already encountered. No doubt, it would be possible to write a history of how each of those has developed in these past years. Such histories would need to be developed on the basis, not of research data and theory, but of the identity and subsequent research careers and contributions of those working within each tradition. (I must note that one behaviorism that emerged in the early ’50s, behavior-based “mathematical theories of learning,” survived less than a score of years and that another, “foraging theory,” one segment of an “economic theory of behavior,” is still extant. I did not search among either.)
Such histories might note the consequences for these behaviorisms–the persisting effects of logical positivistic dogma on research methodology and theory and of the “cognitive revolution” that was unleashed in the late fifties.
Through these past forty years, I have devoted my own scientific activities, not to searching for further behaviorisms, but to searching among the behaviorisms I’d already encountered and among the behaviors I’ve observed. To these, I’ve added the writings of critical thinkers-about-behavior who seem to complement the achievements of those to whom I have already acknowledged a heavy debt. Most important among these philosophers and psychologists have been Steinar Kvale, Kurt Riegel, and necessarily William James, whose later papers on “radical empiricism” have reached broad readership only recently. Too, I have revisited the research of both Tolman and Guthrie, and even the basic observations of the first man to define psychology as the science of behavior–William MacDougall. (MacDougall’s deplorably literary theory of “instincts” demonstrates unequivocally all that is counter-productive in theorizing and in research based solely on theory. The closely interrelated concomitant changes in the stimuli we respond to, how we respond to them, and how we communicate to others and to ourselves the behavioral state into which identifiable and manipulatable events, some simple and some complex, have thrust us, must be incorporated within any behavioral science.)
The research methods that began as necessity during the first four years–the absence of the apparatus and the style of graduate universities–proved indeed to be the benevolent and fruitful mother of invention. Growing experience with and respect for the work of students as both researchers and subjects enabled me to collaborate with them in carrying out experiments simple in design but unequivocal in result, enabling the exploration of further areas of behavior and to evaluate, pragmatically and progressively, a systematic ordering as it developed. Such research demonstrated that following the simplest methods of data analysis is critical for reaching conclusions upon which one can build. More complex analyses, on the other hand, almost invariably produce results consistent with only the preconceptions of the experimenter and have little further application.
Pragmatism, I found, is the test of the validity of the findings of research. Results must be accepted as valid and reliable, not on the basis of statistical tests, but on their utility: given a set of results, can a further experiment be carried out successfully that could not be done if those results were not sound?
Evaluating and reevaluating the methods taught, searching among behavioral findings, devaluating some and accepting others on the basis of their congruence with atheoretical empirical principles of research has produced an ordering of behaviors. Sound observations, one finds, may lead to bad theories; much less often do bad theories yield observations of consequence beyond the theory.
In short, over these past years, I’ve been assembling a behaviorism that I believe will live. It has been put together with bits and pieces of some of all those I’ve encountered, with large chunks of three others: “interbehaviorism,” “the experimental analysis of behavior,” and “logical behaviorism.”
This review of past research findings, behavioral or not, has produced what seems to be a taxonomic system for the ordering of results. That system has developed concurrently (and interactively) with the extension of the vocabulary of an empirical, descriptive language of behavior. The taxonomic system is summarized in Operation Analysis; the vocabulary is defined and ordered in an associated Glossary/Thesaurus of Behavior.
A science of behavior must develop its own language and methodology. It must be skeptical of the logic and methodology borrowed from those physical sciences which have developed within the confines of the laboratories of Universities. Its methods of gathering, analyzing, and organizing data are simple, and their ordering, at this point in the development of the science, must be largely classificatory, that is, taxonomic.
Explanation for such a behaviorism is description in terms that enable each new finding to be placed in its context, that is, assimilated with related findings. It is not the behaviorist’s business to explain what we do, or how and “why” we do it, in terms of either events hypothesized in a “mind” or of events observed in the nervous system. The laws of behaviorism will prove self-explanatory and its findings will serve to guide the neuroanatomical and neurophysiological research of the future.
At the University of Maryland, I began an analysis and categorization of the interactions of the individual and the environment as they have been investigated in the laboratory, attempting to identify those experimental procedures that control processes. These are the paradigmatic operations. Such analysis required a close, more demanding analysis of the operations of measurement. These enabled a parallel analysis and categorization of the behavioral interactions observed as they occur “in real life,” that is, outside the laboratory and without experimental manipulation.
The evaluation of the results of published experimental research based on these principles requires examination of how the experimenters dealt with each of the classes of variables that have been identified as pertinent to experimental findings. These reflect decisions of the experimenter(s) on each of these categories of variables–hence, they are termed categoric operations, operations that must be replicated by any other researcher who wishes to confirm reported results by repeating the research.
Applying such operation analysis to several hundred randomly selected published papers had disheartening results. My students and I found, first, that most of the research could not be repeated because these categoric operations were inadequately reported, and too often not at all. Second, the operations carried out were so closely tailored to fit the experimenter’s theory that they were of limited generalizability, that is, of no consequence independent of the theory to which they were designed. And there are many, many theories. It is not at all difficult to design and carry out an experiment, and by the use of inferential statistical tests, find out what you wished to.
I expect to publish a monograph summarizing Operation Analysis and the methodology developed for this analysis.
Parallel to Operation Analysis has been the preparation of a greatly extended glossary of behavioral terms and of a thesaurus in which they are ordered. This began almost thirty years ago and has been articulated with the work on Operation Analysis. It is a continuing effort. Following the empirical principles of definition of the 1957 Glossary, the definitions, when read in order in the thesaurus, effectively form a treatise on an empirical science of behavior.
Near completion, the Glossary/Thesaurus is now undergoing final editing by Brady Phelps, who is working with me, and will work in the future with Peter Harzem, Edward K. Morris, and Roger Ray for its publication.
Together, Operation Analysis and the Glossary/Thesaurus set forth the behaviorism I found after this fifty-seven year search.
I exit through the same door through which I entered: the vocabulary of psychology, that is, of behavior.