367

JOURNAL OF THE EXPERIMENTAL ANALYSIS OF BEHAVIOR 2002, 77, 367–371 NUMBER 3 (MAY)

PHILOSOPHY OF BEHAVIORISM

G. E. ZURIFF

WHEATON COLLEGE

I was privileged to have been associated inone form or another with the Harvard Pi-geon Lab from September 1964, when I ar-rived as a graduate student, until my friendand colleague Gene Heyman moved out ofWilliam James Hall on June 30, 1998. Aftercompleting my graduate degree in 1968, Icontinued to live in Cambridge and wastherefore able to continue my affiliation withthe Pigeon Lab, frequenting the lab meetingsto learn about the latest research and occa-sionally to present some ideas of my own.During the academic year 1981–1982, I was afull-time visiting scholar at the lab, regularlyattending lab meetings as well as weekly con-versations with Fred Skinner. These lattermeetings, or ‘‘pow-wows’’ as Fred calledthem, were attended by a small group con-sisting of Will and Maggie Vaughan, Pere Ju-lia, another visiting scholar at the time, andme, as well as occasional invited guest speak-ers (Skinner, 1983, p. 394).

When I arrived at Harvard in 1964, the Pi-geon Lab was about to move from the base-ment of Memorial Hall into William JamesHall, and it had completed its transition fromSkinner’s leadership to Herrnstein’s. Skinnerhad accepted a government Career Awardand was retired from teaching and depart-mental responsibilities, although he re-mained Professor of Psychology. Not only wasthis a change in style, but it also meant a basicshift in research direction. For Skinner, thefundamental dependent variable in operantresearch was the absolute response rate of thefree-operant response. He conceptualized re-sponse rate as closely associated with re-sponse probability, or response strength, thesuccessor to the earlier reflex reserve. As avisual representation of the moment-to-mo-ment changes in response rate, the cumula-tive record was the ideal recording instru-ment. Under this paradigm, the research

Address correspondence to the author at 120 FosterSt. Cambridge, Massachusetts 02138 (e-mail: zuriff@mit.edu).

program consisted of the systematic explora-tion of how response rate changes as a func-tion of contingencies of reinforcement, typi-cally in the form of schedules of reinforcement.

However, it gradually became apparent thatthis conception of response rate is internallyproblematic. On the one hand, response ratewas supposed to measure response strengthas a function of reinforcement contingencies.On the other hand, however, research was re-vealing that response rate was itself suscepti-ble to control by contingencies of reinforce-ment. If response rate, or more technically,the interresponse time, could be manipulat-ed by changes in the contingencies, it couldnot at the same time reflect responsestrength. The seeds of this contradiction werealready apparent in Skinner’s own work(Ferster & Skinner, 1957, pp. 7, 10; Skinner,1938, p. 284), which was devoted to the studyof response rate as a measure of responsestrength but also often explained responserate as due to response features prevailing atthe time of reinforcement.

Herrnstein’s early work with two-key con-current schedules of reinforcement offered away out of this dilemma. In contrast to theprevious research program, Herrnstein’s de-pendent variable was the relative rate of re-sponse on one of the keys, and this variablewas a molar variable in that it was calculatedover a session rather than moment to mo-ment. Skinner had long opposed the use ofrelative frequency measures, or what hecalled the behavior ratio (Skinner, 1950), asit was used by Hull and Tolman in their maze-running experiments. Furthermore, relativeresponse rate was a steady-state variable, anddid not lend itself to the analysis of changerevealed by the cumulative records Skinnerfavored (Skinner, 1976). Moreover, Skinner(1983, p. 362) doubted that a molar princi-ple, rather than a simple process, could havefigured in natural selection.

The transition from absolute rates to rela-tive rates of response and reinforcement and

368 G. E. ZURIFF

from moment-to-moment changes to moremolar measures had profound implications.First, and foremost, it initiated a coherent re-search program that characterized the pigeonlab for its remaining years. Herrnstein’smatching law, a direct consequence of thistransition, became the organizing principlefor decades of research, much of it emanat-ing from the Pigeon Lab and its graduates. Asecond consequence was the effect on Skin-ner’s relationship to the lab. Although therewas a host of reasons for his estrangementfrom the lab, it seems likely that his opposi-tion to the change in research direction con-tributed to his absenting himself from labmeetings. As a result, graduate students inthe 1960s onward had very little, if any, con-tact with Skinner and any direct influence hemight have exerted on their thinking, exper-imental or conceptual.

Skinner’s absence from the Pigeon Lab wasa source of disappointment for me. I had ar-rived at Harvard with the single-minded in-tention of learning more about Skinner’s phi-losophy of behaviorism. As an undergraduateat Columbia, I had been inspired by Kellerand Schoenfeld who offered behaviorism asnot only a scientific theory and method butalso as a way to conceptualize all of psychol-ogy. They led a department of true believersin this philosophy, including Bill Cumming,a radical behaviorist if there ever was one,Ralph Hefferline, who devoted an entire se-mester seminar to Skinner’s (1957) Verbal Be-havior, and Herb Terrace, newly arrived fromthe Harvard Pigeon Lab. Psychology at Co-lumbia in those days was synonymous withSkinnerian behaviorism, and introductorypsychology consisted of reading Keller andSchoenfeld’s (1950) Principles of Psychology ac-companied by a rat lab in operant condition-ing, much to the consternation of many a Co-lumbia undergraduate. But to me, aphilosophy major interested in philosophy ofpsychology, this was exciting and challenging.After a few of these courses, and some phi-losophy of science with Ernest Nagel and phi-losophy of mind with Arthur Danto, I was offto Harvard to study with Skinner himself.

To my chagrin, not only was Skinner notaccessible, but Dick Herrnstein, my mentor,did not share my interest in philosophy ofbehaviorism. To be sure, if pushed into a phil-osophical discussion, he could hold his own,

displaying the intellectual brilliance hebrought to everything he approached. But hisheart was with the science of behavior, not itsconceptual foundations. Because of my inter-ests in philosophy, he often compared me toa spider, spinning a web of ideas out of myown mind, in contrast to empiricists like him-self, who like squirrels, construct a science bycollecting one fact after another from the ex-ternal natural world.

Another source of irritation in my quest tolearn philosophy of behaviorism was locatedright down the corridor in Memorial Hallwhere S. S. Stevens ran the psychophysics lab.Using his magnitude estimation methods, heclaimed to be measuring ‘‘sensations,’’ or‘‘psychological magnitudes’’ in contrast tophysical magnitudes. Nevertheless, Stevens(1935) claimed that these apparently mentalterms all had perfectly good operational def-initions. What irked me was that his defini-tions, derived ultimately from Bridgman’s no-tion of operational definition, wereunintelligible to me. I didn’t understandthen, and still do not, how a sensation can beequivalent to a set of experimental operationsand behavioral responses. The issue contin-ued to trouble me, and it was no accident thatmy first publication in philosophy of behav-iorism was ‘‘A Behavioral Interpretation ofPsychophysical Scaling’’ (Zuriff, 1972) inwhich I offered a behaviorist understandingof psychophysical research.

Whereas the Pigeon Lab generated two co-herent research programs, one under Skin-ner and later one under Herrnstein, no com-parable coherent philosophy of behaviorismemerged, other than Skinner’s own work. Un-doubtedly, there were many reasons for this,but I will mention only three. First, as Skinnerincreasingly absented himself from the lab,the opportunities for him to enlist membersof the lab into his vision of behaviorism de-creased. His philosophical discussions wereconducted primarily with his followersaround the world rather than with membersof the Pigeon Lab down the hall. Second,Herrnstein was committed to empirical be-havioral research rather than the explorationof the conceptual foundations of behavior-ism, and as a consequence, no institutionalcontext existed for the development of a co-herent philosophy. Third, graduate studentsand postdoctoral fellows came to Harvard for

369PHILOSOPHY OF BEHAVIORISM

the empirical behavioral research, not the be-haviorism. Those seeking conceptual trainingchose to go elsewhere, for example, to theUniversity of Nevada to work with WillardDay. Furthermore, as Skinner’s interest inphilosophy of behaviorism shifted to social is-sues, the Harvard students had even less in-terest because of the tenuous connection be-tween his social ideology and the science ofoperant conditioning he founded. To besure, many an animated philosophical discus-sion took place in the Pigeon Lab, but noconsensus emanated from them.

Obviously the most important contributionto philosophy of behaviorism to emerge fromthe Pigeon Lab is Skinner’s own corpus ofwriting. Much has been said about his radicalbehaviorism, and I will not attempt to addmore here other than to outline his philoso-phy as a baseline to which later developmentscan be compared. First are his brilliant be-haviorist interpretations of the mental world.This includes his interpretations of mentalconcepts such as meaning, purpose, and ex-pectation in terms of contingencies and his-tories of reinforcement, his introduction ofprivate events into behaviorism to account forfirst-person reports of internal episodes, andhis behavioral epistemology. Second is hisconception of psychological theory and the-orizing. This includes, on the one hand, hisstrictures against theoretical inference, hissubstitution of functional relation for causa-tion, his notion of a theory as a parsimoniousset of laws, and his view of explanation as pre-diction and control. On the other hand areincluded his views of effective scientific prac-tice as staying close to the data and observa-tion, his opposition to the hypothetico-de-ductive method, his championing anautonomous science of behavior, and his in-terpretation of science as the behavior of sci-entists. Third is his social philosophy includ-ing his ideas on the application of behavioraltechnology, his advocacy of behavioral con-trol, his opposition to the use of punishment,and his rejection of traditional mental con-cepts such as ‘‘freedom’’ and ‘‘personal re-sponsibility,’’ which he believed interferedwith human progress.

Although the Pigeon Lab did not prove tobe a place for the continued incubation, dis-cussion, and development of these ideas as itwas for Skinner’s science of operant condi-

tioning, many members of the lab made im-portant individual contributions to the dis-course on the philosophy of behavior overthe years. For example, of the more than 200articles published in the journal Behaviorism,a journal devoted to philosophical behavior-ism, between its founding by Willard Day in1972 and its demise in 1989, more than 10%were authored by members and former mem-bers of the Pigeon Lab. Nine of them alsoserved on the editorial board.

In addition, several of them published im-portant textbooks on learning and behavior.Although these books were devoted primarilyto the science of behavior, they typically in-cluded an introductory chapter describingthe history and nature of behaviorism. Oftenbrief passages touched on philosophical is-sues from a Skinnerian perspective. Two ex-amples are textbooks by Reynolds (1975, pp.2, 3, 131) and Catania (1998, pp. 254, 374).In addition to the several textbooks on learn-ing and behavior, five graduates of the PigeonLab also published books primarily devotedto the development of philosophy of behav-iorism. To be sure, many other alumni alsopublished articles with philosophical content,but I shall restrict myself to these majorworks, which alone are sufficient to illustrateboth the important role played by alumni ofthe Pigeon Lab in articulating the philosophyof behaviorism and the lack of any single uni-fying conception.

Robert Boakes (1984) authored From Dar-win to Behaviourism, a very readable history ofthe study of animal psychology from Darwinto about 1930. The later sections of this bookdeal with the beginnings of behaviorism, es-pecially the work of John Watson, but theyalso touch on the thought of Tolman, Lash-ley, and Hull. Skinner, however, is not dis-cussed. During the period covered by Boakes,researchers in animal psychology had tostruggle with many philosophical issues cen-tral to behaviorism, including the definitionof behavior, the inference of mental opera-tions and contents from behavior, the rela-tions between behavior and physiology, sci-entific method, the nature of consciousness,and the relations between mind and body.

William Baum’s (1994) Understanding Be-haviorism comes very close to representing adirect descendant and developer of Skinner’sthought while modernizing and clarifying

370 G. E. ZURIFF

Skinner’s radical behaviorism (Zuriff, 1995).Baum reconceptualizes several of Skinner’sbasic empirical concepts, including reinforce-ment, and embeds them into modern evolu-tionary biology. For example, he works out indetail the parallels among natural selection,reinforcement, and cultural selection, and hesubstitutes the contemporary theoreticalframework of biological fitness for Skinner’searlier concept of survival. Similarly, Baumpresents the best exposition available of Skin-ner’s concept of rule-governed behavior, fill-ing in the gaps in Skinner’s thought and cor-recting the weaknesses. It is in the area ofsocial philosophy that Baum makes his mostoriginal contribution to Skinnerian behavior-ism. In his discussion of freedom and happi-ness, Baum brings to bear the latest studiesin adaptation-to-the-norm and equity theoryto deal with some of the anomalies in behav-iorist treatments. Finally, in a somewhat idio-syncratic philosophical analysis, Baum iden-tifies radical behaviorism with a version ofpragmatism.

Whereas Baum attempted to extend Skin-nerian behaviorism, John Staddon’s (2001)The New Behaviorism sets out to refute nearlyevery aspect of radical behaviorism and to re-place it with his ‘‘new behaviorism’’ (Zuriff,2001). Staddon calls into question Skinner’sbasic concept of reinforcement contingency,challenging the adequacy of contiguity to ex-plain the effects of reinforcement. He rejectsprediction and control as sufficient goals fora scientific theory, and he argues for a deepertheoretical understanding. Contrary to Skin-ner’s proscriptions against theorizing, Stad-don advocates theoretical inferences of inter-nal states and mechanisms to supply causalexplanations rather than the teleological ex-planations provided by contingencies of re-inforcement (and see discussion of Rachlinbelow).

Likewise, in his discussion of social philos-ophy, Staddon (2001) disagrees with most ofSkinner’s recommendations. He rejects Skin-ner’s use of survival as a selection criterionfor social practices, preference for positive re-inforcement over punishment, advocacy forsocial control, and his call for the overthrowof traditional values such as freedom and per-sonal responsibility. Underlying Staddon’s op-position are his suspicion of revolutionarychange and his skepticism over grand extrap-

olations from animal laboratory research tothe complex world of everyday human action.

In his densely written Behavior and Mindand an important series of papers, HowardRachlin (1985, 1992, 1994) presents a highlyoriginal and intriguing version of behavior-ism. Using an Aristotelian analysis, he arguesfor unusual behaviorist understandings of ex-planation and the mental. He claims that toexplain behavior in terms of contingencies ofreinforcement, and by extension, in terms ofprinciples such as the matching law or maxi-mization, is to give a final cause for the be-havior rather than an antecedent efficientcause. It is comparable to explaining why theviolinist played the notes she did by sayingshe was playing Debussy’s String Quartet in Gminor. This explanation places the behaviorin the larger context of what was happening,past as well as future, and the Quartet is thusthe final cause rather than an antecedentevent that produces the violinist’s playing.Similarly, to explain behavior in terms ofmaximization, or a utility function, or a con-tingency is to place it in the context of eventsand correlations extended over a period oftime and space.

This teleological behaviorism also offers aninteresting interpretation of mentalistic lan-guage. When we explain a person’s behaviorin terms of beliefs, feelings, thoughts, or sen-sations, we seem to be suggesting an anteced-ent efficient cause. According to Skinner, thiscause may be a covert event or an antecedenthistory of reinforcement. For Rachlin, in con-trast, the mentalistic explanation is interpret-ed as teleological in that it suggests a finalcause rather than an efficient cause. It em-beds the person’s behavior in the larger con-text of this person’s overt behavior and its in-teractions with the environment overextended time and space. Thus, Rachlin dis-penses with Skinner’s world of internal re-sponses and stimuli (as well as the informa-tion processing apparatus of the cognitivists)and substitutes final cause explanation for ef-ficient cause explanation. Interestingly, Rach-lin’s shift away from single responses and re-inforcements to extended patterns ofbehavior–environmental interactions in hisaccounts of explanation and the use of men-tal terms parallels (and was perhaps suggest-ed by?) the theoretical shift in the Pigeon Labfrom moment-to-moment analysis to more

371PHILOSOPHY OF BEHAVIORISM

molar variables measured over long periodsof time.

In contrast to Rachlin, Baum, and Staddon,Zuriff’s (1985) Behaviorism does not advocateone particular version of behaviorism. In-stead, it attempts to present a conceptual re-construction of all varieties of behaviorism,revealing their logical geography. Its thesis isthat behaviorism is an attempt to develop anobjective and empirical psychology and thatthere are many ways to achieve this. In work-ing out the meaning of objective and empirical,the behaviorist encounters a number of phil-osophical choice points, for example, the se-lection of a data language and how to con-ceptualize first-person reports. As each choicepoint is negotiated, others materialize as alogical consequence. Thus, a version of be-haviorism, including Skinner’s radical behav-iorism, can be conceptualized as a branchingtree diagram, with each choice point repre-sented by a node. Zuriff attempts to evaluatethe strengths and weaknesses of each versionin light of contemporary philosophy and sci-ence and to correct flaws where possible.

Each of these books has played an impor-tant role in contemporary philosophy of be-haviorism. Although these products of the Pi-geon Lab do not represent a coherentbehaviorism, they do share two features incommon (with the exception of Boakes’ his-tory). First, each in its own way has tried tomodernize behaviorism in light of recent de-velopments in philosophy and science, espe-cially evolutionary biology. Second, each hasattempted to defend behaviorism against theavalanche of criticism leveled at it by philos-ophers and psychologists, especially cognitiv-ists. Although these defenses are, I believe,successful in the sense that they are for themost part intellectually valid, they have notbeen successful in the sense of terminatingthe endless recycling of the same criticisms

or inducing psychologists to pay serious atten-tion to behaviorism. Nevertheless, in assess-ing the legacy of the Pigeon Lab, these con-tributions to the philosophy of behaviorismshould be remembered as significant achieve-ments.

REFERENCESBaum, W. M. (1994). Understanding behaviorism: Science,

behavior, and culture. New York: HarperCollins.Boakes, R. A. (1984). From Darwin to behaviourism: Psy-

chology and the minds of animals. Cambridge, England:Cambridge University Press.

Catania, A. C. (1998). Learning (4th ed.). Upper SaddleRiver, NJ: Prentice Hall.

Ferster, C. B., & Skinner, B. F. (1957). Schedules of rein-forcement. New York: Appleton-Century-Crofts.

Keller, F. S., & Schoenfeld, W. N. (1950). Principles ofpsychology. New York: Appleton-Century-Crofts.

Rachlin, H. (1985). Pain and behavior. Behavioral andBrain Sciences, 8, 43–83.

Rachlin, H. (1992). Teleological behaviorism. AmericanPsychologist, 47, 1371–1382.

Rachlin, H. (1994). Behavior and mind: The roots of modernpsychology. New York: Oxford University Press.

Reynolds, G. S. (1975). A primer of operant conditioning(Rev. ed.). Glenview, IL: Scott, Foresman.

Skinner, B. F. (1938). The behavior of organisms. New York:Appleton-Century-Crofts.

Skinner, B. F. (1950). Are theories of learning necessary?Psychological Review, 57, 193–216.

Skinner, B. F. (1957). Verbal behavior. New York: Apple-ton-Century-Crofts.

Skinner, B. F. (1976). Farewell, my lovely! Journal of theExperimental Analysis of Behavior, 25, 218.

Skinner, B. F. (1983). A matter of consequences. New York:Knopf.

Staddon, J. E. R. (2001). The new behaviorism: Mind, mech-anism, and society. Philadelphia: Psychology Press.

Stevens, S. S. (1935). The operational definition of psy-chological concepts. Psychological Review, 42, 517–527.

Zuriff, G. E. (1972). A behavioral interpretation of psy-chophysical scaling. Behaviorism, 1, 118–133.

Zuriff, G. E. (1985). Behaviorism: A conceptual reconstruc-tion. New York: Columbia University Press.

Zuriff, G. E. (1995). A review of Baum’s UnderstandingBehaviorism: Science, Behavior, and Culture. The BehaviorAnalyst, 18, 395–400.

Zuriff, G. E. (2001). Conservative behaviorism: A reviewof Staddon’s The New Behaviorism: Mind, Mechanism,and Society. The Behavior Analyst, 24, 263–269.