A Day of Great Illumination: B. F. Skinner's Discovery of ... · A DAY OF GREAT ILLUMINATION: B. F. SKINNER’SDISCOVERY OF SHAPING GAIL B. PETERSON UNIVERSITY OF MINNESOTA—TWIN

317

JOURNAL OF THE EXPERIMENTAL ANALYSIS OF BEHAVIOR 2004, 82, 317–328 NUMBER 3 (NOVEMBER)

A DAY OF GREAT ILLUMINATION: B. F. SKINNER’S DISCOVERY OF SHAPING

GAIL B. PETERSON

UNIVERSITY OF MINNESOTA—TWIN CITIES

Despite the seminal studies of response differentiation by the method of successive approximationdetailed in chapter 8 of The Behavior of Organisms (1938), B. F. Skinner never actually shaped anoperant response by hand until a memorable incident of startling serendipity on the top floor of aflour mill in Minneapolis in 1943. That occasion appears to have been a genuine eureka experiencefor Skinner, causing him to appreciate as never before the significance of reinforcement mediatedby biological connections with the animate social environment, as opposed to purely mechanicalconnections with the inanimate physical environment. This insight stimulated him to coin a newterm (shaping), and also led directly to a shift in his perspective on verbal behavior from an emphasison antecedents and molecular topographical details to an emphasis on consequences and moremolar, functional properties in which the social dyad inherent to the shaping process became thedefinitive property of verbal behavior. Moreover, the insight seems to have emboldened Skinner toexplore the greater implications of his behaviorism for human behavior writ large, an enterprise thatcharacterized the bulk of his post-World War II scholarship.

Key words: Skinner, shaping, mediated reinforcement, verbal behavior

Most, if not all, JEAB readers are familiarwith the story of B. F. Skinner’s venture intothe world of weapon-systems design and de-velopment during World War II (cf. Skinner,1960). To pursue that wartime research pro-ject, Skinner took a 1-year leave of absencefrom his duties on the faculty of the Univer-sity of Minnesota and moved his workplaceoff campus to a secret laboratory provided byGeneral Mills, Inc. Figure 1 is a photographfrom that time that shows the large GoldMedal Flour railroad yard and flour-millingcomplex in downtown Minneapolis immedi-ately adjacent to St. Anthony Falls on the Mis-sissippi River. Skinner’s lab was located in thehalf-floor space at the top of the buildingholding the large water tanks in the uppercenter portion of the photo, on the same lev-el on which the gigantic sign saying ‘‘Even-tually’’ was mounted. The photograph ofSkinner (see Figure 2)shows him standing onthe roof just outside his top floor lab, with

An earlier version of this essay appeared under the title‘‘The Discovery of Shaping, or B. F. Skinner’s Big Sur-prise’’ in The Clicker Journal: The Magazine for Animal Train-ers, Issue # 43 ( July/August 2000), pp. 6–13. That versionis also available on the website of the Cambridge Centerfor Behavioral Studies, www.behavior.org. I am especiallygrateful to my colleague Edward K. Morris of the Uni-versity of Kansas and to the Harvard University Archivesfor providing me with copies of the Schrader/Skinnercorrespondence presented in this article.

Reprint requests should be addressed to Gail B. Peter-son, 75 East River Road, Minneapolis, Minnesota 55455(e-mail: [email protected]).

some of the script of Eventually (in reverse)visible above and behind him. (Eventually re-ferred to Gold Medal Flour’s popular mar-keting slogan of that era: ‘‘Eventually you willuse Gold Medal Flour, so why not now?’’ Inresponse, rival Pillsbury Mills, located directlyacross the river, adopted as their marketingslogan: ‘‘Because Pillsbury’s Best!’’).

Much less well known than the Project Pi-geon story, however, is a smaller but ultimate-ly more important story within that story:Skinner’s discovery of what he later dubbedshaping, and the impact that discovery had onhis outlook on the determinants of complexhuman behavior. The purpose of this articleis to tell that inside story.

Rather remarkably often in his various writ-ings over the years (e.g., 1958, 1972, 1979,1983), Skinner recounted the following inci-dent:

In 1943 Keller Breland, Norman Guttman,and I were working on a war-time projectsponsored by General Mills, Inc. Our labora-tory was the top floor of a flour mill in Min-neapolis, where we spent a good deal of timewaiting for decisions to be made in Washing-ton. All day long, around the mill, wheeledgreat flocks of pigeons. They were easilysnared on the window sills and proved to bean irresistible supply of experimental subjects.. . . This was serious research, but we had ourlighter moments. One day we decided to teacha pigeon to bowl. The pigeon was to send awooden ball down a miniature alley toward aset of toy pins by swiping the ball with a sharp

318 GAIL B. PETERSON

Fig. 1. The General Mills, Inc., railroad yard and Gold Medal Flour milling complex in Minneapolis circa 1943.B. F. Skinner’s wartime laboratory was located in the half-floor space immediately beneath the large water tanks onthe top story of the building marked by the ‘‘Eventually’’ sign in the top center part of the picture. (Photo courtesyof the Minnesota Historical Society.)

sideward movement of the beak. To conditionthe response, we put the ball on the floor ofan experimental box and prepared to operatethe food-magazine as soon as the first swipeoccurred. But nothing happened. Though wehad all the time in the world, we grew tired ofwaiting. We decided to reinforce any responsewhich had the slightest resemblance to aswipe—perhaps, at first, merely the behaviorof looking at the ball—and then to select re-sponses which more closely approximated thefinal form. The result amazed us. In a few min-utes, the ball was caroming off the walls of thebox as if the pigeon had been a championsquash player. The spectacle so impressed Kel-ler Breland that he gave up a promising careerin psychology and went into the commercialproduction of behavior. (1958, p. 94)

This anecdote is only mildly interesting inits own right. It is somewhat amusing, for ex-

ample, to imagine the great and famous B. F.Skinner and two of his most prominent stu-dents snaring pigeons from the window sillsof a flour mill and taking time to teach oneof them to bowl. However, there is also some-thing tantalizingly intriguing in Skinner’s ex-position here that makes this story potentiallymore significant: ‘‘The result,’’ he wrote,‘‘amazed us.’’ Why? Why should they havebeen amazed by what had happened? And whywas it that the ‘‘spectacle so impressed KellerBreland that he gave up a promising careerin psychology and went into the commercialproduction of behavior’’?

The questions above are prompted by atleast three considerations. First, note that theoccasion Skinner describes here occurred in1943, 5 years after the publication of The Be-havior of Organisms (1938). That book had

319THE DISCOVERY OF SHAPING

Fig. 2. B. F. Skinner standing on the roof of the Gen-eral Mills Utility Building, just outside his wartime labo-ratory. The script of the ‘‘Eventually’’ sign (see Figure 1)is visible behind him. (Photo courtesy of Robert Bailey.)

contained a whole chapter (chapter 8) on‘‘The differentiation of a response’’ in whichSkinner had described the results of his ex-periments using the method of successive ap-proximation to develop differentiated oper-ant lever-pressing behavior in rats. In thatwork he had found that, when he graduallyincreased response force or duration require-ments, the rat pressed more forcefully orheld the lever down longer, respectively.Wasn’t that shaping? Of course it was. Butwhy, then, was he so amazed by what he sawin the bowling pigeon in 1943? One mightunderstand his being impressed in 1933, 10years earlier, when he was still in the processof discovering the basic principles of operantconditioning. But 1943?

Second, even before the publication of TheBehavior of Organisms, Skinner had been in-volved in an historic academic exchange withKonorski and Miller over whether operantconditioning was distinctively different fromrespondent conditioning (Konorski & Miller,1937; Miller & Konorski, 1928, 1969; Skinner1935, 1937). Part of that exchange includedSkinner’s vivid description of what soundslike the process of hand shaping a rat’s lever-press response:

But elaborate and peculiar forms of responsemay be generated from undifferentiated op-erant behavior through successive approxi-mation to a final form. This is sometimes trueof the example of pressing the lever. A rat maybe found (very infrequently) not to press thelever spontaneously during a prolonged peri-od of observation. The response in its finalform may be obtained by basing the reinforce-ment upon the following steps in succession:approach to the site of the lever, lifting thenose into the air toward the lever, lifting thefore-part of the body into the air, touching thelever with the feet, and pressing the leverdownward. When one step has been condi-tioned, the reinforcement is withdrawn andmade contingent upon the next. (1937, p.277)

This sounds like the description of an ac-tual, empirical event. But if Skinner had nev-er actually shaped a behavior by hand until1943, then this description in the 1937 articlemust have been entirely speculative. As weshall see, that, indeed, appears to be the case.

The third thing that makes his expressionof surprise at this 1943 event intriguing is thefact that the power of positive reinforcementfor creating unlikely behavioral performanc-es by an animal had already been widely pub-licized in a 1937 story in Life magazine aboutSkinner’s lab rat Pliny. Skinner had trainedPliny to perform a behavior chain consistingof:

. . . pulling a string to obtain a marble from arack, picking the marble up with the fore-paws, lifting it to the top of the tube, anddropping it inside. Every step in the processhad to be worked out through a series of ap-proximations, since the component responseswere not in the original repertoire of the rat.(1938, p. 340)

Skinner characterized this demonstrationwith Pliny as a tour de force of operant con-


ditioning. Surely he had used shaping intraining Pliny—hadn’t he? In fact, hadn’t hesaid as much? True, he hadn’t actually usedthe word shaping, but surely that is what hemeant when he described the behavior ashaving been ‘‘worked out through a series ofapproximations’’? If so, why, then, was he soamazed years later when he trained a pigeonto perform a very similar but considerablysimpler behavior?

Hints at answers to some of these questionsare contained in one of Skinner’s other ren-derings of the story:

Possibly our most impressive experiment con-cerned the shaping of behavior. I had usedsuccessive approximation in my experimentson the force and duration of lever-pressing,and we had seen how important it was inteaching a pigeon to peck hard. Pliny’s com-plex behavior had been put together step bystep by making slight changes in the appara-tus. But one exciting day on the top floor ofthat flour mill we programmed contingenciesby hand.

We put a pigeon in a large cardboard car-ton, on one wall of which was a food dispenseroperated by a hand switch. We put a woodenball the size of a Ping-Pong ball on the floorand undertook to teach the pigeon to knockit about the box. We began by reinforcingmerely looking at the ball, then moving thehead toward it, then making contact with it,and eventually knocking it to one side with aswiping motion. The pigeon was soon battingthe ball about the box like a squash player. Wehad shaped a very complex topography of be-havior through successive approximation in amatter of minutes, and we ‘‘gazed at one an-other in wild surmise.’’

I remember that day as one of great illu-mination. We had discovered how much easierit was to shape behavior by hand than bychanging a mechanical device. (1979, p. 268)

Again, very emphatic and, indeed, dramat-ic language: ‘‘Possibly our most impressive ex-periment’’ from among many innovativetechniques explored during Project Pigeon;‘‘we ‘gazed at one another in wild surmise,’’’a literary allusion to John Keats’ famous 1816sonnet On First Looking into Chapman’s Homerexalting the exhilaration that accompanies pi-oneering exploration and great discovery;and ‘‘I remember that day as one of greatillumination,’’ a significant statement indeedcoming from a man whose career was char-acterized by many penetrating insights.

A description of the same event in anotherpiece provides further clarification:

I well remember the day when Norman Gutt-man, Keller Breland, and I discovered howwrong all this [making small changes in thephysical environment in order to implementa program of successive approximation] wasby dispensing with the mechanical contingen-cies and reinforcing successive approxima-tions to a complex response by hand. By op-erating a food dispenser with a hand switchwe taught a pigeon to strike a wooden ballwith a swiping motion of its beak and to knockthe ball about its cage, rather in the mannerof a squash player. Shaping complex behaviorthrough a programmed sequence of contin-gencies is now so commonplace that it is hardto understand why we should have beenamazed at the speed with which this was done.(1972, p. 3)

This particular passage makes two pointsabout Skinner’s illuminating surprise: it re-volved around (a) the efficacy of implement-ing a program of successive approximation bysimply watching the animal and operating thereinforcement-delivery device by hand, ratherthan making small mechanical adjustments ofthe physical environment, as he had alwaysdone before, and (b) the rapidity with whichdramatic changes in response topographycan be brought about when one does this. Inaddition, it validates the bewilderment con-temporary readers might have at the fact thathe was so surprised that day in 1943; in hind-sight, Skinner himself admitted that ‘‘it ishard to understand why we should have beenamazed.’’

Evidently, then, it was the efficacy and ef-ficiency of free-form shaping by hand thathad amazed Skinner and company. But thatimplies that, despite all the lever- pressing ex-periments he had performed with rats be-tween 1930 and 1943, he had never actuallyshaped a behavior by hand before that fatefulday in the flour mill. Could that be true? Itappears that it well may be. Close reading ofSkinner’s biography reveals that, yes, indeed,it is most likely the case that, in his 1937 ‘‘re-ply’’ to Konorski and Miller, the descriptionof how one could go about shaping leverpressing by a rat had been purely hypotheti-cal. In commenting on the matter some 40years later, Skinner confessed:

I do not remember actually shaping lever-


pressing in such explicit stages, but I was sureit could be done, and I had certainly changedthe ‘‘value of a single property’’ through suc-cessive approximation in producing veryforceful responses. (1979, p. 185)

The final set of historical texts consistentwith the main thesis of this article has beenuncovered only recently. Shortly after I hadsubmitted the original draft of this manu-script to JEAB, Professor Edward Morris ofthe University of Kansas sent me copies of twointeresting letters he had come across inad-vertently this past spring while perusing thehistorical material on Skinner held by theHarvard University Archives. Aware of my ear-lier research on this story, he immediatelyrecognized these documents as being directlypertinent. They are reproduced here in Fig-ures 3 and 4.

A brief correspondence was initiated in thespring of 1965 by a Miss Peggy Schrader, thenan undergraduate at Illinois State University.Her letter (Figure 3) reads:

4/5/65Dear Dr. Skinner,

I am a beginning student of psychology.Right now I am being overwhelmed by thethought of all I have to learn before I willknow a little about the subject. There is asmall question, though, that has gotten me tothinking and wondering so much that I can-not wait to learn the answer.

My professor used your experiments withteaching pigeons to bowl as an example in arecent lecture. I later realized that this was theearliest use of successive approximations assuch that I had heard of so I ran up to hisoffice and asked him if this was the first ex-perimental use of the technique. He wasn’tcertain but referred me to several articles suchas the one written by you in the American Psy-chologist in 1958. My curiosity has not yetbeen satisfied. Could you please tell me—ortell my professor so that he can tell me—whether successive approximations were firstused as an experimental technique in 1943.My professor is:

Dr. Walter H. FriedhoffProfessor of PsychologyIllinois State UniversityNormal, Illinois, 61761

Thank you very much.Sincerely,

Peggy Schrader402 Broadway

Normal, Illinois, 61761

Skinner responded to Peggy within the week(see Figure 4):

Dear Miss Schrader:The date 1943 is, I think, correct for the

first deliberate use of successive approxima-tion. I assume that the article you mention isthe enclosed. The description of our experi-ence with a pigeon there is, I think, accurate.Sincerely yours,B. F. Skinner

Clearly, Peggy Schrader was an unusuallyastute student, as well as a charmingly conciseand meticulous letter writer. She obviouslyrecognized something that appears to havebeen largely overlooked and underappreci-ated by academic psychologists, includingdedicated Skinner aficionados. Strictly speak-ing, however, the 1943 occasion was not ‘‘theearliest use of successive approximations assuch,’’ given the work with Pliny and the stud-ies of response differentiation noted above.However, it very likely truly was the earliestuse of free-form hand shaping, as such, by anacademic psychologist. In this connection,Skinner’s qualifying the occasion as ‘‘the firstdeliberate use of successive approximation’’is, I think, significant given conventional def-initions of the adjective deliberate, that is, care-fully weighed or considered, studied, inten-tional, careful or slow in deciding, steady inmovement or action, slow and even, unhur-ried.

Thus it appears that neither the lever press-ing by rats in Behavior of Organisms nor Pliny’sbehaviors had been shaped by hand. All thatpre-1943 behavior, including Pliny’s complexrepertoire, had come about purely as a resultof the way the physical environment hadbeen set up, together with contingencies con-trolled by mechanical and electrical program-ming apparatus. Skinner had taken a com-pletely hands-off approach in all that earlywork. The first real hands-on shaping was thatday in the flour mill.

Frankly, this isn’t all that hard to believe.As we all know, in his early writings Skinneroften characterized operant behavior as be-havior that is spontaneously emitted, as opposedto being elicited by antecedent stimuli in theenvironment. On this spontaneous emissionview, a given operant was conceived as havingsome natural base rate of occurrence (the op-erant level ) prior to that rate being modified


Fig. 3. Illinois State University student Peggy Schrader’s April 1965 letter to B. F. Skinner inquiring about thefirst use of the successive approximations technique. (Courtesy of Edward K. Morris and the Harvard UniversityArchives.)


Fig. 4. The unsigned file copy Skinner retained of his reply to Peggy Schrader. (Courtesy of Edward K. Morrisand the Harvard University Archives.)

by response consequences. The operant levelmay be very low, on this view, but it’s neverzero. If one waits long enough, to paraphrasethe old Gold Medal Flour slogan, eventually,a response will be emitted. In this connec-tion, recall Skinner’s description of the situ-ation immediately preceding the decision to

do hand shaping that day in the flour mill:‘‘Though we had all the time in the world,we grew tired of waiting.’’ The classic lab ratlever-pressing operant conforms quite nicelyto this spontaneous-emission formulation. Inactual lab practice, as opposed to instruction-al lab exercises, it is virtually never necessary


to shape lever pressing in a rat. A magazine-trained but otherwise naive lab rat will beginto press a lever quite nicely on its own if itsimply is put in the box with the control ap-paratus programmed on a fixed-ratio 1 sched-ule. And that’s the way Skinner used to do it.Thus it is well within the realm of possibilitythat seeing that pigeon knock that littlewooden bowling ball around that day in 1943was, in fact, the first time that Skinner or any-one else in the history of academic psychol-ogy had seen behavior shaped by hand and hadrecognized the significance of the observation. Andthe significance of the observation went be-yond the speed and efficiency that Skinnerexplicitly noted above: also brought into sud-den and sharp relief was the potential of thesocial dyad as an unusually effective mediumof operant conditioning.

But surely behavior had been shaped byhand before. Yes, one would certainly thinkso. Indeed, even Skinner seems to havethought so. In reflecting critically on his ap-parent discovery of hand shaping, he com-mented that ‘‘Thorndike had shaped the be-havior of a cat by hand’’ (1979, p. 268).However, unless Thorndike related some-thing to that effect to Skinner personallywhen he visited Skinner’s lab during the 1937APA convention on the University of Minne-sota campus, there is little reason for Skinnerto have thought that Thorndike had everdone any actual shaping. To be sure, Thorn-dike may have administered reinforcing con-sequences by hand (he did not have the ben-efit of electrical control apparatus), andperhaps that’s what Skinner was referring to,but delivering reinforcers by hand does not,in and of itself, qualify as shaping. Further-more, close reading of Thorndike’s classicwork (1911) finds no description of his gen-uinely shaping the behavior of cats to anypurpose. Also, because his dependent vari-able was response latency, Thorndike’s catswere required to make the full response (e.g.,operating a latch) from the beginning oftraining, trial after trial. Response topogra-phies may well have changed as the latenciesdecreased, but not because Thorndike waswatching and deliberately reinforcing select-ed variations. The closest thing to hand shap-ing Thorndike appears to have done was themethod he used to tame his monkeys:

In getting them so that they would let them-selves be handled it was of almost no serviceto take them and feed them while holdingthem or otherwise make that state pleasant forthem. By far the best way is to wait patientlytill they do come near, then feed them; waitpatiently till they do take hold of your arm,then feed them. If you do take them and holdthem partly by force, you must feed them onlywhen they are comparatively still. In short, intaming them one comes unconsciously toadopt the method of rewarding certain oftheir impulses rather than certain conditionswhich might be associated in their minds withideas, had they such. (p. 234)

Mazes and straight runways would be evenless likely to call for hand shaping thanThorndike’s puzzle boxes, and therefore it isunderstandable why one finds no descriptionof a shaping process in the early writings ofTolman or Hull. It is worth noting, however,that Hull (1943) was much impressed bySkinner’s experiments on the differentiationof lever pressing force, replicating those stud-ies and relating the findings to his concept ofbehavior oscillation. And in his 1952 book,Hull devoted an entire chapter to the theo-retical problem of response differentiation,again relying considerably on Skinner’s workon response force and also on the Pliny dem-onstration, concluding that chapter with thestatement:

. . . the principles of behavior oscillation andcorrelated reinforcement as stated in the pre-ceding paragraph have yielded an understand-ing of how needed novel acts never previously per-formed may come into existence so that theirreinforcement may occur in the conventionalmanner, a problem that has greatly disturbedsome theorists. (p. 214)

Thus Hull obviously appreciated the theo-retical significance of response shaping forthe genesis of new behavior (cf. Logan, 1959,pp. 339–348), but nowhere in Hull’s writingsor in the writings of the other classic earlylearning theorists does one encounter a de-scription of hand shaping and the signifi-cance of the dyadic process it entails.

Although the concept of response shapingmay have gone unappreciated by academicpsychologists until the middle of the 20thcentury, people had no doubt shaped one an-other’s behavior as well as that of their‘‘beasts’’ for eons prior to that time. Burch


and Pickel (1990), for example, have notedthe early work of the German dog trainerColonel Konrad Most (1910/1954) and itsanticipation of many modern operant con-ditioning concepts, including shaping. InAbrichtung des Hundes: Individuell und ohneStrafen (1910), which translates as ‘‘TheTraining of Dogs: Individually and withoutpunishments’’ (the subtitle prophetically em-blematic of Skinner’s future approach), Mostdescribed both response prompting andshaping, as this excerpt from the 1954 trans-lation clearly shows:

. . . In providing it [a physical responseprompt] the trainer will find his best chanceto give rein to the most exuberant expressionof the warmth of his feelings. He ought, assoon as the required act is performed and alsowhen even the slightest progress is apparent,not only to utter such words as ‘‘good boy’’repeatedly in caressing tones, and fondle thedog, but also, if the exercise in hand permitsit, to execute a dance of joy with the animal.. . . All the physical movements which excitea dog’s instinct to play are capable of arousingthe utmost delight in him. (pp. 33–34)

Evidence of an intuitive understanding andapplication of shaping can be found in thewritings of early popular authors as well. Forexample, in his novel White Fang, Jack Lon-don (1906) referred to ‘‘the plasticity of hisclay, of his capacity for being moulded by thepressure of the environment’’ (p. 143) as hedescribed the behavior of the story’s maincharacter, the wolf dog White Fang. Londonalso provided a clear, detailed, and extendeddescription in that book (pp. 164–170) ofwhat we today would call systematic desensi-tization.

Thus the essential elements of hand shap-ing had been described both in popular fic-tion and in technical manuals prior to 1943,and the method had no doubt been em-ployed pragmatically by at least some of ourancestors for millennia. But it is one thing fora phenomenon to occur in nature, be ob-served, and even be exploited deliberately,and quite another for it to be recognized assomething of fundamental scientific signifi-cance. For example, people certainly knewthat response consequences affect behaviorlong before Thorndike formulated the Lawof Effect; surely no one has ever been sur-prised to hear that rewards strengthen behav-

ior. What was unappreciated, however, washow this simple fact could be employed toaccount for an enormous volume of behaviorin a wide range of species. That’s what Thorn-dike was the first to recognize. Similarly, it wasthe scientific significance of direct shaping,both theoretical and practical, that Skinnerappears to have been the first to recognize.

As discussed in more detail below, Skin-ner’s discovery that day in the flour mill ap-pears to have been very much a personal in-tellectual watershed for him. He obviouslysaw it as something special because he adopt-ed a new term for it: shaping. He had usedthe term successive approximation before (forthe first time in the ‘‘Reply to Konorski &Miller’’ paper, 1937) as well as the phrase dif-ferentiation of a response, but he never onceused the word shaping until after that ‘‘dayof great illumination.’’ The word does not ap-pear anywhere in The Behavior of Organisms1,nor can it be found in any other papers bySkinner prior to 1951. Although he neversaid so, the term may have been suggested tohim by his reading of J. B. Watson’s Behavior-ism (1930, e.g., pp. 98, 102). In any event, thefirst time it appears in the literature of mod-ern psychology denoting the specific processof behavior-change we all associate with it to-day was in Skinner’s little 1951 Scientific Amer-ican article titled ‘‘How to Teach Animals’’:

The reinforcement gives you a means ofcontrol over the behavior of the animal. Itrests on the simple principle that wheneversomething reinforces a particular activity of anorganism, it increases the chances that the or-ganism will repeat that behavior. This makes itpossible to shape an animal’s behavior almost as asculptor shapes a lump of clay [italics added].There is, of course, nothing new in this prin-ciple. What is new is a better understandingof the conditions under which reinforcementworks best. (pp. 26–27)

It is irresistible to conclude that Skinnerdecided a special term was in order for whathe had observed up there in the flour mill,

1 Skeptical readers will probably doubt this assertionand immediately consult their copies of Behavior of Or-ganisms, where they will find the word ‘‘shaping’’ on p.xiv in the Preface to the Seventh Printing. But this is theonly place in the book where the word appears, and thispreface was written in 1966, 28 years after the originalpublication and 23 years after the episode in the flourmill.


one that would suggest a distinction betweenthe process of behavioral elaboration direct-ed by constraints in the physical environmentwith mechanical connections to sources of re-inforcement from behavioral elaboration di-rected by the social environment, a parent, ateacher, a trainer, a therapist, who controlledreinforcer delivery. His analogy with the be-havior of a sculptor, a creative artist, conveysa hint of this, and he employed that meta-phor again later and at greater length (cf.,Skinner, 1953). The inanimate, physical en-vironment can correctly be said to shape be-havior, of course, and indeed it does. But themost complex repertoires are those that areshaped by the social environment, by otherliving, breathing, behaving creatures. It wasthis special dynamic that struck Skinner likea bolt from the blue that day in 1943, and itturned his attention to the analysis of com-plex human behavior to a significantly great-er extent than before.

For example, there was a definite shift inSkinner’s approach to verbal behavior afterthe ‘‘day of great illumination.’’ He had beeninterested in the psychology of language andliterature since very early on, had created andtaught a special course on it, and had pub-lished some novel analytical and empiricalstudies of it (Skinner, 1936, 1939, 1942). Buthis initial approach to the analysis of verbalbehavior now seems uncharacteristically sta-tistical and molecular for Skinner. It focusedalmost exclusively on the role of antecedentevents and on the frequencies of certain to-pographies of response (e.g., the frequencyof a specific consonant in a line of poetry).Concern with base rates and what we wouldtoday probably call behavioral momentum pre-dominate in those Shakespeare and Swin-burne studies, for example, but there’s abso-lutely nothing in them about the role ofsocial reinforcement. Much the same can besaid for the verbal summator work. And, asBjork (1993) has cogently pointed out, Skin-ner’s early course on the psychology of liter-ature was actually more psychoanalytic thanbehavior-analytic. Although the concept ofmultiple causation continued to figure prom-inently in his overall analysis of verbal behav-ior, the role of consequences and contingen-cies of reinforcement in verbal behaviorbecame much more important after 1943.

Skinner began concentrated effort on his

manuscript on verbal behavior under the aus-pices of a Guggenheim Fellowship in 1944,which then became the William James Lec-tures at Harvard in 1947, and from thereevolved into Science and Human Behavior(1953) and ultimately his 1957 treatise VerbalBehavior. Reinforcement by other people be-came definitive of verbal behavior. ‘‘Verbalbehavior always involves social reinforcementand derives its characteristic properties fromthis fact’’ (Skinner, 1953, p. 299). In VerbalBehavior, Skinner defined verbal behavior ge-nerically as ‘‘behavior shaped and maintainedby mediated consequences’’ (p. 2). (Pleasenote the conspicuous presence of the S-wordin the preceding quotation.) By mediated con-sequences, of course, he meant consequencescontrolled by another person, as was the case,for the first time, in the bowling pigeon epi-sode in the flour mill. The social mediationof the reinforcement process became the pri-mary defining factor whereas all other aspectsbecame secondary:

In defining verbal behavior as behavior rein-forced through the mediation of other per-sons we do not, and cannot, specify any oneform, mode, or medium. Any movement ca-pable of affecting another organism may beverbal. (1957, p. 14)

The definitiveness of the dyadic interactionwas given further emphasis when he wrote of‘‘the behavior of a speaker’’ and ‘‘the behav-ior of a listener,’’ the two most fundamentalaspects of ‘‘the total verbal episode’’:

In explaining the behavior of the speaker weassume a listener who will reinforce his behav-ior in certain ways. In accounting for the be-havior of the listener we assume a speakerwhose behavior bears a certain relation to en-vironmental conditions. The interchanges be-tween them must explain all the conditionsthus assumed. The account of the whole epi-sode is then complete. (1957, p. 34)

Thus the behavior of the speaker is shapedby reinforcers administered by the listener,just as the behavior of the bowling pigeon wasshaped by Skinner or Breland or Guttman orwhichever one of them it was who operatedthe feeder. At the same time, the behavior ofthe speaker changes the immediate environ-ment of the listener so as to set the occasionfor the listener’s behavior of presenting thereinforcer, just as that pigeon’s successive ap-


proximations to pecking the ball had set theoccasion for the shaper to operate the feeder.All the controlling variables of the behavioralepisode were encapsulated in this dyadic in-teraction or ‘‘interchange.’’ The scientific ac-count was complete, its potential scope mon-umental. No wonder they ‘‘gazed at oneanother in wild surmise.’’

But, it seems reasonable to ask, did theflour mill episode really prompt a major shiftin Skinner’s perspective on verbal behavior,or is this merely a fanciful suggestion derivedfrom a wily assemblage of selected historicaltidbits? In his own words:

The techniques of shaping behavior and ofbringing it under stimulus control which canbe traced, as I have suggested elsewhere, to amemorable episode on the top floor of thatflour mill in Minneapolis needed only a de-tailed reformulation of verbal behavior to bedirectly applicable to education. (1960, p. 37)

Skinner was an erudite etymologist and ameticulous writer. He chose his words andcrafted his sentences carefully. Please notethat, in the quotation above, he did not saythe techniques of shaping behavior andbringing it under stimulus control ‘‘neededonly to be incorporated into my existing for-mulation of verbal behavior’’; he said they‘‘needed only a detailed reformulation [italicsadded] of verbal behavior.’’ And in that re-formulation of verbal behavior, the dyadic in-terchange inherent to the shaping processbecame the central and defining feature.

Skinner’s insightful appreciation of theubiquity and power of socially mediated con-sequences of behavior did more than prompthim to reformulate and finally publish his ac-count of human verbal behavior. It also em-boldened him to extend the basic conceptsof operant conditioning from the behavior ofrats and pigeons in the laboratory to the be-havior of people writ large. He had beenquite reticent on this front prior to the dis-covery of shaping. Aside from a few, extreme-ly cautious statements in the closing pages ofThe Behavior of Organisms (1938, pp. 441–442), Skinner unapologetically confined hisanalysis to the lever pressing of his rats: ‘‘Thebook presents nothing more than an experi-mental analysis of a representative sample ofbehavior. Let him extrapolate who will’’ (p.442). But following the 1943 episode in the

flour mill came large extrapolations in theform of Walden Two (1948), Science and Hu-man Behavior (1953), and Verbal Behavior(1957). In the preface of the second printing(1976) of Walden Two, in commenting on hisoutlook in 1945 following the end of the war,Skinner wrote: ‘‘In Behavior of Organisms, pub-lished 7 years earlier, I had refused to applymy results outside the laboratory’’ (p. v). Hispolicy on this changed dramatically afterWorld War II, with most of the rest of his ca-reer being devoted to extensions outside thelaboratory to the behavior of people as indi-viduals, groups, societies, and cultures. He be-came the champion of the cause for extend-ing the principles of behavior analysis to thebetterment of the human condition on per-sonal, local, national, and global scales. Incommenting on Walden Two as well as in clos-ing his account of Project Pigeon, Skinnerwrote:

I still believe the same kind of wide-rangingspeculation about human affairs, supported bystudies of compensating rigor, will make a sub-stantial contribution toward that world of thefuture in which, among other things, therewill be no need for guided missiles. (1960, p.37)

Skinner’s ingenuity as an inventor, creativegadgeteer, and classic bench scientist is justi-fiably legendary. The clever pieces of auto-matic apparatus he devised for recordingbehavior, programming contingencies, deliv-ering reinforcers, and graphically depictingthe course of behavior change truly revolu-tionized and advanced the field. This earlyemphasis on automation, however, also bi-ased him toward thinking primarily in termsof mechanical connections between an or-ganism’s behavior and its environment, effec-tively preventing him from considering thepotential of biological connections with thesocial environment—until that day of greatillumination in the flour mill. In the morethan 60 years that have transpired since then,shaping has come to be taken largely forgranted, probably as much because of its in-tuitive appeal and sheer plausibility as for itsdemonstrated validity. It is hard to imagine atime in the formal field of psychology whenthe shaping process was essentially unknown,but that was clearly once the case. At the sametime, we continue to be rightly impressed by


demonstrations of its effectiveness, for ex-ample, in teaching speech to individuals wholack it (e.g., Isaacs, Thomas, & Goldiamond,1960; Lovaas, Berberich, Perdoff, & Schaef-fer, 1960), or, more recently, in helping in-dividuals with brain damage recover lost limbfunction even after many years of nonuse(e.g., Taub et al., 1994). And in the futuristicfield of artificial intelligence, a generic capac-ity for behavior to be shaped by its conse-quences is recognized as key to eventual suc-cess in the design of versatile robots (e.g.,Saksida, Raymond, & Toutetzky, 1997; Savage,1998, 2001). Thus B. F. Skinner’s studies ofresponse differentiation and his subsequentinsight about shaping, each a genuinely orig-inal set of observations, deserve to be delin-eated among the most important findings, interms of both theoretical and practical signif-icance, in the history of behavioral science.

REFERENCES

Bjork, D. W. (1993). B. F. Skinner: A life. New York: BasicBooks.

Burch, M. R., & Pickel, D. (1990). A toast to Most: Kon-rad Most, a 1910 pioneer in animal training. Journalof Applied Behavior Analysis, 23, 263–264.

Hull, C. L. (1943). Principles of behavior. New York: Apple-ton-Century-Crofts.

Hull, C. L. (1952). A behavior system. New Haven: YaleUniversity Press.

Isaacs, W., Thomas, J., & Goldiamond, I. (1960). Appli-cation of operant conditioning to reinstate verbal be-havior in psychotics. Journal of Speech and Hearing Dis-orders, 25, 8–12.

Konorski, J., & Miller, S. (1937). On two types of condi-tioned reflex. Journal of General Psychology, 16, 264–272.

Logan, F. A. (1959). The Hull-Spence approach. In S.Koch (Ed.), Psychology: A study of a science. Study I. Con-ceptual and systematic. Volume 2. General systematic for-mulations, learning, and special processes (pp. 293–358).New York: McGraw Hill.

London, J. (1906). White Fang. New York: Macmillan.Lovaas, O. I., Berberich, J. P., Perdoff, B. F., & Schaeffer,

B. (1960). Acquisition of imitative speech by schizo-phrenic children. Science, 151, 705–706.

Miller, S., & Konorski, J. (1928). Sur une forme particu-liere des reflexes conditionnels. Comples Rendus des Se-ances de La Societe Polonaise de Biologie, 49, 1155–1157.

Miller, S., & Konorski, J. (1969). On a particular form ofconditioned reflex (B. F. Skinner, Trans.). Journal ofthe Experimental Analysis of Behavior, 12, 187–189.

Most, K. (1954). Training dogs: A manual. London: Pop-

ular Dogs Publishing Company Ltd. (Original workpublished 1910 as Abrichtung des Hundes: Individuellund ohne Strafen.)

Saksida, L. M., Raymond, S. M., & Toutetzky, D. S. (1997).Shaping robot behavior using principles from instru-mental conditioning. Robotics and Autonomous Systems,22, 231–249.

Savage, T. (1998). Shaping: The link between rats androbots. Connection Science, 10, 321–340.

Savage, T. (2001). Shaping: A multiple contingenciesanalysis and its relevance to behaviour-based robotics.Connection Science, 13, 199–234.

Skinner, B. F. (1935). Two types of conditioned reflexand a pseudo-type. The Journal of General Psychology, 12,66–77.

Skinner, B. F. (1936). The verbal summator and a meth-od for the study of latent speech. Journal of Psychology,2, 71–107.

Skinner, B. F. (1937). Two types of conditioned reflex: Areply to Konorski and Miller. The Journal of GeneralPsychology, 16, 272–279.

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

Skinner, B. F. (1939). The alliteration in Shakespeare’ssonnets: A study of literary behavior. Psychological Re-cord, 3, 186–192.

Skinner, B. F. (1942). A quantitative estimate of certaintypes of sound-patterning in poetry. The American Jour-nal of Psychology, 30, 64–79.

Skinner, B. F. (1948/1976). Walden two. New York: Mac-millan.

Skinner, B. F. (1951). How to teach animals. ScientificAmerican, 185, 26–29.

Skinner, B. F. (1953). Science and human behavior. NewYork: Macmillan.

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

Skinner, B. F. (1958). Reinforcement today. American Psy-chologist, 13, 94–99.

Skinner, B. F. (1960). Pigeons in a pelican. American Psy-chologist, 15, 28–37.

Skinner, B. F. (1972). Some relations between behaviormodification and basic research. In S.W. Bijou & E.Ribes-Inesta (Eds.), Behavior modification: Issues and ex-tensions (pp. 1–6). New York: Academic Press.

Skinner, B. F. (1979). The shaping of a behaviorist: Part twoof an autobiography. New York: New York UniversityPress.

Skinner, B. F. (1983). A matter of consequences: Part three ofan autobiography. New York: New York University Press.

Taub, E., Crago, J. E., Burgio, L. D., Groomes, T. E.,Cook, E. W. III, DeLuca, S. C., et al. (1994). An op-erant approach to rehabilitation medicine: Overcom-ing learned nonuse by shaping. Journal of the Experi-mental Analysis of Behavior, 61, 281–293.

This smart University of Minnesota rat works slot ma-chine for a living. (1937, May 31). Life, 80–81.

Thorndike, E. L. (1911). Animal intelligence. New York:Macmillan.

Watson, J. B. (1930). Behaviorism. University of ChicagoPress.

Documents

A Day of Great Illumination: B. F. Skinner's Discovery of ... · A DAY OF GREAT ILLUMINATION: B. F. SKINNER’SDISCOVERY OF SHAPING GAIL B. PETERSON UNIVERSITY OF MINNESOTA—TWIN