Summary: Harris 1979 What happened to little Albert.pdfkim roy
Whatever Happened to Little Albert?
BEN HARRIS Vassar College
ABSTRACT: John B. Watson and Rosalie Rayner's 1920 conditioning of the infant Albert B. is a well- known piece of social science folklore. Using pub- lished sources, this article reviews the study's actual procedures and its relationship to Watson's career and work. The article also presents a history of psycholo- gists' accounts of the Albert study, focusing on the study's distortion by Watson himself, general textbook authors, behavior therapists, and most recently, a prominent learning theorist. The author proposes pos- sible causes for these distortions and analyzes the Albert study as an example of myth making in the history of psychology.
Almost 60 years after it was first reported, Watson and Rayner's (1920) attempted conditioning of the infant Albert B. is one of the most widely cited experiments in textbook psychology. Under- graduate textbooks of general, developmental, and abnormal psychology use Albert's conditioning to illustrate the applicability of classical conditioning to the development and modification of human emotional behavior. More specialized books focus- ing on psychopathology and behavior therapy (e.g., Eysenck, 1960) cite Albert's conditioning as an ex- perimental model of psychopathology (i.e., a rat phobia) and often use Albert to introduce a dis- cussion of systematic desensitization as a treat- ment of phobic anxiety.
Unfortunately, most accounts of Watson and Rayner's research with Albert feature as much fabrication and distortion as they do fact. From information about Albert himself to 'the basic ex- perimental methods and results, no detail of the original study has escaped misrepresentation in the telling and retelling of this bit of social science folklore.
There has recently been a revival of interest in Watson's conditioning research and theorizing (e.g., MacKenzie, 1972; Seligman, 1971; Weimer & Palermo, 1973; Samelson, Note 1), and in the mythology of little Albert (Cornwell & Hobbs, 1976; Larson, 1978; Prytula, Oster, & Davis, 1977). However, there has yet to be a complete examination of the methodology and results of the
Vol. 34, No. 2, 151-160
Albert study and of the process by which the study's details have been altered over the years. In the spirit of other investigations of classic studies in psychology (e.g., Ellenberger, 1972; Parsons, 1974) it is time to examine Albert's con- ditioning in light of current theories of learning. It is also time to examine how the Albert study has been portrayed over the years, in the hope of discovering how changes in psychological theory have affected what generations of psychologists have told each other about Albert.
As described by Watson and Rayner (1920), an experimental study was undertaken to answer three questions: (1) Can an infant be conditioned to fear an animal that appears simultaneously with a loud, fear-arousing sound? ( 2 ) Would such fear transfer to other animals or to inanimate objects? (3) How long would such fears persist? In at- tempting to answer these questions, Watson and Rayner selected an infant named Albert B., whom they described as "healthy," and "stolid and un- emotional" (p. 1). At approximately 9 months of age, Albert was tested and was judged to show no fear when successively observing a number of live animals (e.g., a rat, a rabbit, a dog, and a monkey), and various inanimate objects (e.g., cotton, human masks, a burning newspaper). He was, however, judged to show fear whenever a long steel bar was unexpectedly struck with a claw hammer just behind his back.
Two months after testing Albert's apparently unconditioned reactions to various stimuli, Watson and Rayner attempted to condition him to fear a
Preparation of this article was aided by the textbook and literature searches of Nancy Kinsey, the helpful com- ments of Mike Wessels, and the bibliographic assistance of Cedric Larson. The author also thanks Bill Wood- ward and Ernest Hilgard for their comments on earlier versions of this work.
Requests for reprints should be sent to Ben Harris, Box 368, Vassar College, Poughkeepsie, New York 12601.
AMERICAN PSYCHOLOGIST • FEBRUARY 1979 • 151 Copyright 1979 by the American Psychological Association, Inc.
white rat. This was done by presenting a white rat to Albert, followed by a loud clanging sound (of the hammer and steel bar) whenever Albert touched the animal. After seven pairings of the rat and noise (in two sessions, one week apart), Albert reacted with crying and avoidance when the rat was presented without the loud noise.
In order to test the generalization of Albert's fear response, 5 days later he was presented with the rat, a set of familiar wooden blocks, a rabbit, a short-haired dog, a sealskin coat, a package of white cotton, the heads of Watson and two assist- ants (inverted so that Albert could touch their hair), and a bearded Santa Glaus mask. Albert seemed to show a strong fear response to the rat, the rabbit, the dog, and the sealskin coat; a "nega- tive" response to the mask and Watson's hair; and a mild response to the cotton. Also, Albert played freely with the wooden blocks and the hair of Watson's assistants.
After an additional 5 days, Watson recondi- tioned Albert to the rat (one trial, rat paired with noise) and also attempted to condition Albert di- rectly to fear the previously presented rabbit (one trial) and dog (one trial). When the effects of this procedure were tested in a different, larger room, it was found that Albert showed only a slight reaction to the rat, the dog, and the rabbit. Consequently, Watson attempted "to freshen the reaction to the rat" (p. 9) by presenting it with the loud noise. Soon after this, the dog began to bark loudly at Albert, scaring him and the experi- menters and further confounding the experiment.
To answer their third question concerning the permanence of conditioned responses over time, Watson and Rayner conducted a final series of tests on Albert after 31 days of neither condition- ing nor extinction trials. In these tests, Albert showed fear when touching the Santa Claus mask, the sealskin coat, the rat, the rabbit, and the dog. At the same time, however, he initiated contact with the coat and the rabbit, showing "strife be- tween withdrawal and the tendency to manipulate" (Watson & Rayner, 1920, p. 10). Following these final tests, Albert's mother removed him from the hospital where the experiment had been conducted. (According to their own account, Watson and Rayner knew a month in advance the day that Albert would no longer be available to them.)
The Context of Watson and Rayner's Study What was the relationship of the Albert experi- ment to the rest of Watson's work? On a per-
sonal level, this work was the final published project of Watson's academic career, although he supervised a subsequent, related study of the de- conditioning of young children's fears '(M. C. Jones, 1924a, 1924b). From a theoretical per- spective, the Albert study provided an empirical test of a theory of behavior and emotional de- velopment that Watson had constructed over a number of years.
Although Watson had publicly declared himself a "behaviorist" in early 1913, he apparently did not become interested in the conditioning of motor and autonomic responses until late 1914, when he read a French edition of Bekhterev's Objective Psychology (see Hilgard & Marquis, 1940). By 1915, Watson's experience with conditioning re- search was limited to this reading and his collabora- tion with his student Karl Lashley in a few simple studies. Nevertheless, Watson's APA Presidential Address of that year made conditioned responses a key aspect of his outline of behaviorism and seems to have been one of the first American ref- erences to Bekhterev's work (Hilgard & Marquis, 1940, p. 24; Koch, 1964, p. 9; Watson, 1916b). Less than a year after his APA address, two ar- ticles by Watson (1916a, 1916c) were published in which he hypothesized that both normal de- fense mechanisms and psychiatric disorders (e.g., phobias, tics, hysterical symptoms) could be under- stood on the basis of conditioning theory.
Six months later, the American Journal of Psy- chology featured a more extensive article by Wat- son and J. J. B. Morgan (1917) that formulated a theory of emotion, intended to serve both ex- perimentalists and clinicians. Its authors hypothe- sized that the fundamental (unlearned) human emotions were fear, rage, and love; these emotions were said to be first evoked by simple physical manipulations of infants, such as exposing them to loud sounds (fear) or restricting their movements (rage). Concurrently, they hypothesized that "the method of conditioned reflexes" could explain how these basic three emotions become transformed and transferred to many objects, eventually result- ing in the wide range of adult emotions ithat is evoked by everyday combinations of events, per- sons, and objects. In support of these theoretical ideas, Watson and Morgan began to test whether infants' fears could be experimentally conditioned, using laboratory analogues of thunder and light- ning. In the description of this work and the related theory, a strong appeal was made for its practical importance, stating that it could lead to a standard experimental procedure for "bringing
152 • FEBRUARY 1979 • AMERICAN PSYCHOLOGIST
the human emotions under experimental control" (p. 174).
By the early months of 1919, Watson appears not yet to have found a reliable method for experi- mentally eliciting and extinguishing new emotional reactions in humans. However, by this time he had developed a program of research with infants to verify the existence of his hypothesized three fundamental emotions. Some early results of this work were described in May 1919, as part of a lengthy treatise on both infant and adult emotions. Anticipating his work with Albert,1 Watson (1919b) for the first time applied his earlier prin- ciples of emotional conditioning to children's fears of animals. Based on a case of a child frightened by a dog that he had observed, Watson hypothe- sized that although infants do not naturally fear animals, if "one animal succeeds in arousing fear, any moving furry animal thereafter may arouse it" (p. 182). Consistent with this hypothesis, the results of Watson and Rayner's experiments with Albert were reported 9 months later.
Although Watson's departure from Johns Hop- kins prematurely ended his own research in 1920, he continued to write about his earlier findings, including his work with Albert. In 1921, he and Rayner (then Rosalie Rayner Watson) summa- rized the results of their interrupted infant re- search program, concluding with a summary of their experience with Albert. Although this was a less complete account than their 1920 article, it was the version that was always referenced in Watson's later writings. These writings included dozens of articles in the popular press (e.g., Wat- son, 1928b, 1928c), the books Behaviorism (1924) and Psychological Care of Infant and Child (1928a), and a series of articles in Pedagogical Seminary (Watson, 192Sa, 192Sb, 1925c). Many of these articles retold the Albert story, often with photographs and with added comments elabo- rating on the lessons of this study.
Introductory-Level Textbook Versions of Albert
A selective survey of textbooks2 used to introduce students to general, developmental, and abnormal psychology revealed that few books fail to refer to Watson and Rayner's (1920) study in some manner. Some of these accounts are completely accurate (e.g., Kennedy, 197S; Page 1975; White- hurst & Vasta, 1977). However, most textbook versions of Albert's conditioning suffer from in- accuracies of various degrees. Relatively minor
details that are misrepresented include Albert's age (Calhoun, 1977; Johnson & Medinnus, 1974), his name (Galanter, 1966), the spelling of Rosalie Rayner's name (e.g., Biehler, 1976; Helms & Turner, 1976; McCandless & Trotter, 1977; Pa- palia & Olds, 1975), and whether Albert was ini- tially conditioned to fear a rat or a rabbit (CRM Books, 1971; Staats, 1968).
Of more significance are texts' misrepresenta- tions of the range of Albert's postconditioning fears and of the postexperimental fate of Albert. The list of spurious stimuli to which Albert's fear response is claimed to have generalized is rather extensive. It includes a fur pelt (CRM Books, 1971), a man's beard (Helms & Turner, 1976), a cat, a pup, a fur muff (Telford & Sawrey, 1968), a white furry glove (Whittaker, 1965), Albert's aunt, who supposedly wore fur (Bernhardt, 1953), either the fur coat or the fur neckpiece of Albert's mother (Hilgard, Atkinson, & Atkinson, 1975; Kisker, 1977; Weiner, 1977), and even a teddy bear (Boring, Langfeld, & Weld, 1948). In a number of texts, a happy ending has been added to the story by the assertion that Watson removed (or "reconditioned") Albert's fear, with this pro- cess sometimes described in detail (Engle & Snell- grove, 1969; Gardiner, 1970; Whittaker, 1965).
What are the causes of these frequent errors by the authors of undergraduate textbooks? Pry- tula et al. (1977) cataloged similar mistakes but offered little explanation of their source. Corn- well and Hobbs (1976) suggested that such dis- tortions, if not simply due to overreliance on sec- ondary sources, can be generally seen as authors'
1 In tracing the development of Watson's ideas about conditioning, it would be helpful to know whether the experiments with Albert had already begun when Watson wrote his 1919 Psychological Review article. Unfortu- nately, there is no hard evidence of exactly when the Albert study was completed. Watson and Rayner's origi- nal report was published in the February 1920 Journal of Experimental Psychology, suggesting that the research was completed in 1919. Also, M. C. Jones (1975, Note 2) remembers that Watson lectured about Albert as early as the spring of 1919 and showed a film of his work with infants at the Johns Hopkins University (Watson, 1919a). Individual frames of this film published later ("Behavior- ist Babies," 1928; "Can Science Determine Your Baby's Career Before It Can Talk'?," 1922; Watson, 1927, 1928a) suggest that at some date this film contained footage of Albert's conditioning. Since the work with Albert lasted for approximately 4 months, there seems to be a strong possibility that Watson's 1919 prediction was not en- tirely based on theoretical speculation.
2 After this survey of texts was completed, similar re- views by Cornwell and Hobbs (1976) and by Prytula et al. (1977) were discovered. Interested readers should con- sult these articles for lists of additional textbook errors.
AMERICAN PSYCHOLOGIST • FEBRUARY 1979 • 153
attempts to paint the Albert study (and Watson) in a more favorable light and to make it believable to undergraduates. Certainly, many of the com- mon errors are consistent with a brushed-up image of Watson and his work. For example, not one text mentions that Watson knew when Albert would leave his control—a detail that might make Watson and Rayner's failure to recondition Albert seem callous to some modern readers.
However, there are other reasons for such errors besides textbooks' tendencies to tell ethically pleas- ing stories that are consistent with students' com- mon sense. One major source of confusion about the Albert story is \Watson himself, who altered and deleted important aspects of the study in his many descriptions of it. For example, in the Sci- entific Monthly description of the study (Watson & Watson, 1921), there is no mention of the con- ditioning of Albert to .the dog, the rabbit, and the rat that occurred at 11 months 20 days; thus Al- bert's subsequent responses to these stimuli can be mistaken for a strong generalization effect (for which there is little evidence). A complementary and equally confusing omission occurs in Psycho- logical Care oj Infant and Child (Watson, 1928a). There, Watson begins his description of ithe Albert study with Albert's being conditioned to a rabbit (apparently the session occurring at 11 months 20 days). As a result, the reader is led to believe that Albert's fear of a rat (a month later) was the product of generalization rather than the initial conditioning trials. Besides these omissions, Wat- son and Rayner (1920) also made frequent edi- torial comments, such as the assention that fears such as Albert's were "likely to persist indefinitely, unless an accidental method for removing them is hit upon" (p. 12). Given such comments, it is understandable that one recent text overestimates the duration of the Albert experiment by 300% (Goldenberg, 1977), and another states that Al- bert's "phobia became resistant to extinction" (Kleinmuntz, 1974, p. 130).
A second reason for textbook authors' errors, it seems, is the desire of many of us to make experi- mental evidence consistent with textbook theories of how organisms should act. According to popu- lar versions of learning theory (as described by Herrnstein, 1977), organisms' conditioning should generalize along simple stimulus dimensions; many textbooks list spurious fear-arousing stimuli (for Albert) that correspond to such dimensions. To illustrate the process of stimulus generalization, Albert is often said to have feared every white,
furry object—although he actually showed fear mostly of nonwhite objects (the rabbit, the dog, the sealskin coat, Watson's hair), and did not even fear everything with hair (the observers). But to fit a more simplified view of learning, either new stimuli appear in some texts (e.g., a white rabbit, a white glove) or it is simply asserted that Albert's conditioning generalized to all white and furry (or hairy) stimuli (see Biehler, 1976; Craig, 1976; Helms & Turner, 1976). Though it might seem as if Albert's fear did generalize to the cate- gory of all animate objects with fur (e.g., the rabbit) or short hair (e.g., Watson's head), this is impossible to show conclusively. The only ex- perimental stimuli not fitting this category were the blocks and the observers' hair. Apparently the blocks were a familiar toy (thus not a proper stimulus), and Albert's familiarity with the ob- servers is not known (although we may guess that one might have been his mother).
Behavior Therapists' Views of Albert
Unfortunately, misrepresentations of Watson and Rayner's (1920) work are not confined to intro- ductory-level texts. For proponents of behavioral therapies, Albert's conditioning has been a fre- quently cited reference, although its details have often become altered or misinterpreted. Joseph Wolpe, for example, is well known for his condi- tioned-anxiety model of phobias and his treatment of various neurotic disorders by what was origi- nally termed "reciprocal inhibition" (Wolpe, 19S8). According to Wolpe and Rachman (1960):
Phobias are regarded as conditioned anxiety (fear) reac- tions. Any "neutral" stimulus, simple or complex, that happens to make an impact on an individual at about the time that a fear reaction is evoked acquires the ability to evoke fear subsequently, (p. 145)
In support of .this model Wolpe and Rachman cited the Albert study to "indicate that it is quite possible for one experience to induce a phobia" (p. 146). Also, Eysenck (1960) asserted that "Albert developed a phobia for white rats and in- deed for all furry animals" (p. 5). Similar inter- pretations of Watson and Rayner's (1920) experi- ment are found in subsequent writings by Wolpe and other behavior therapists (e.g., Rachman, 1964; Sandier & Davidson, 1971; Ullman & Kras- ner, 1965; Wolpe, 1973).
Critical reading of Watson and Rayner's (1920) report reveals little evidence either that Albert developed a rat phobia or even that animals con-
154 • FEBRUARY 1979 • AMERICAN PSYCHOLOGIST
sistently evoked his fear (or anxiety) during Wat- son and Rayner's (1920) experiment. For ex- ample, 10 days after the completion of the initial (seven-trial) conditioning to a white rat, Albert received an additional trial of conditioning to the same rat. Immediately following this, his reaction to the rat was described as: "Fell over to the left side, got up on all fours and started to crawl away. On this occasion there was no crying, but strange to say, as he started away he began to gurgle and coo, even while leaning far over to the left side to avoid the rat" (p. 7 ) .
On the same day as this, Albert received a trial of conditioning to the rabbit he had seen previ- ously (using the clanging steel bar). When shown the rabbit itwice again, he whimpered but did not cry. Immediately after this, his reactions were tested in a different (larger) room. When shown the rabbit, Albert's response was described as: "Fear reaction slight. Turned to left and kept face away from the animal but the reaction was never pronounced" (p. 9).
Finally, 31 days later and after having received an additional conditioning trial to the rat at the end of the preceding session, Albert's reactions to the (same) rat were:
He allowed the rat to crawl towards him without with- drawing. He sat very still and fixated intently. Rat then touched his hand. Albert withdrew it immediately, then leaned back as far as possible but did not cry. When the rat was placed on his arm he withdrew his body and began to fret, nodding his head. The rat was then al- lowed to crawl against his chest. He first began to fret and then covered has eyes with both hands, (p. 11)
Not only does Albert's response seem lacking in the strength that we associate with phobia (pos- sibly due to Watson's alternation of acquisition and extinction trials) but on a qualitative basis it seems unlike the classically conditioned anxiety on which some behavior therapists base their theoretical models of phobias.
Of course, it might be argued by proponents of a two-factor theory of phobias that Albert's re- actions to the rat and the rabbit were successful escape responses from the anxiety-arousing stimuli, thus explaining Albert's relative calm (no rapid breathing, crying, etc.). However, Albert did not consistently avoid the animals to which he was conditioned. On his final day of testing, for ex- ample, Albert initially did not avoid the rabbit to which he had been conditioned; he then attempted to avoid it, but then "after about a minute he reached out tentatively and slowly and touched
the rabbit's ear with his right hand, finally manip- ulating it" (Watson & Rayner, 1920, p. II). 3
A more serious problem' with clinicians' citing of the Albert study is the failure of Watson's con- temporaries to replicate his work. Although H. E. Jones (1930) subsequently demonstrated persistent galvanic skin response (GSR) conditioning with an infant (using a mild electric current as an un- conditioned stimulus, and a light and various sounds as conditioned stimuli), attempts to repli- cate the Albert study using variations of Watson's own method were unsuccessful. Valentine (1930), for example, used extensive naturalistic observa- tion and failed to find conditioned fear of infants to loud noises; he criticized both Watson's meth- odology and his simplistic theory of emotional de- velopment. Bregman (1934) was also unsuccess- ful in her attempts to condition even 1 of IS in- fants to fear wooden and cloth objects, using a disagreeable noise as an unconditioned stimulus (see Thorndike, 1935). Finally, whatever our ret- rospective view of Albert's conditioned reactions, a conditioned-avoidance model of phobias (with fear as a necessary component) is not consistent with more recent experimental and clinical litera- ture (see Costello, 1970; Hineline, 1977; Marks, 1969,1977).
Albert and Preparedness Theory
One of the reasons that Albert is so well known is that he is rediscovered every S or 10 years by a new group of psychologists. In the early 1960s, Wolpe and Eysenck were the curators and analysts of the Albert myth. Ten years later, Wolpe and Eysenck were supplanted by M. E. P. Seligman,
3 Another model that has been applied to the Albert study is that of operant or instrumental conditioning. For example, Larson (1978) and Reese and Lipsitt (1970) cited a paper by R. M. Church (Note 3) on this point (see also Kazdin, 1978). Such an interpretation is ap- parently based on Watson's notes indicating that at least for the first two trials, the loud noise was contingent on Albert's active response (i.e., touching the rat). Also, the one trial of conditioning to the rabbit occurred when Albert had begun "to reach out and manipulate its fur with forefingers" (Watson & Rayner, 1920, p. 8). The attractiveness of an (aversive) instrumental model of Al- bert's conditioning is that it would not necessarily predict any emotional reaction by Albert and would help ex- plain his reluctance to touch the experimental animals. Strong support for this model is lacking, however, with Watson and Rayner describing at least four conditioning trials on which the loud sound was not contingent on Albert's instrumental response, and a number of trials the character of which is uncertain.
AMERICAN PSYCHOLOGIST • FEBRUARY 1979 • 155
who has seized control of the Albert story and uses it (in slightly revised form) to attack the views of its former proponents. At the same time, Seligman both challenges traditional theories of learning and proposes his own reformulation, known as "preparedness theory."
Briefly stated, preparedness theory (Seligman, 1970, 1971; see also Schwartz, 1974) posits that traditionally held laws of learning cannot be uni- formly applied to all stimuli interacting with all organisms. In a classical conditioning paradigm, organisms may be physiologically or cognitively "prepared" to form certain conditioned stimulus- unconditioned stimulus associations and "contra- prepared" to develop others. In the former case (e.g., rats learning taste aversion to food causing illness) the association is easily formed, but in the latter case (e.g., rats learning taste aversion to food paired with footshock) it is difficult if not im- possible to form. Similarly, Seligman (1970) sum- marized evidence from instrumental-learning para- digms to suggest that for a particular organism, certain behaviors differ in their potential to be successfully conditioned (see Shettleworth, 1973).
Relevant to Albert, Seligman (1971) hypothe- sized that the strength of human phobic reactions (i.e., their resistance to extinction) is due to the high degree of preparedness of certain stimuli (e.g., snakes). This conditioning to phobic objects oc- curs very quickly, whereas conditioning to other stimuli (assumed to be of low preparedness or contraprepared) results in fear reactions that are less intense, take longer to establish, and extinguish more quickly. As Marks (1977) noted, there is some evidence that objects differ in their ability to produce conditioned GSR in humans over time (e.g., Ohman, Erixon, & Lb'fberg, 1975). It also makes sense that evolution 'may have made it easier for humans to learn some responses than others (see Herrnstein, 1977). However, much of Seligman's (1971) discussion of human phobias is based on an erroneous interpretation of Watson and Rayner's (1920) work.
As described in his article "Phobias and pre- paredness," Seligman's version of Albert's condi- tioning is generally consistent with the exaggerated claims for the study made by Watson (e.g., Wat- son, 1924). According to preparedness theory, the existence of strong animal phobias in the hu- man clinical literature is evidence that "furry things" (Seligman, 1971, p. 315) are strongly prepared phobic stimuli for humans. If furry things are highly prepared and Watson and Rayner
(1920) used furry things in their study, then Al- bert must have quickly developed a strong fear of animals and other furry things. Consistent with this logic is Seligman's (1971) assertion that "Al- bert became afraid of rats, rabbits, and other furry objects" (p. 308, italics added) and that Watson "probably did not" become an aversive stimulus to Albert. In fact, Albert was "com- pletely negative" to Watson's hair (Watson & Rayner, 1920, p. 7), and of course, Albert's fear was only tested to a single rat, a single rabbit, and to no previously neutral, nonfurry objects.
In addition to presenting this inaccurate picture of how Albert's fear initially generalized, Selig- man's account also misrepresents the ease with which Albert was conditioned, the durability of his reactions, and the details of an attempt to repli- cate the Albert study. According to Seligman (1971), Albert's "conditioning occurred in two trials" and this "prepared fear conditioning [did] not extinguish readily" (p. 315). In fact, "seven joint stimulations were given [to Albert] to bring about the complete reaction" (Watson & Rayner, 1920, p. 5), and there is little if any evidence either that the .reactions of Albert were resistant to a formal extinction procedure (or to the pas- sage of time) or that he was tested with valid contraprepared stimuli. Further, in describing a similar study that actually used a contraprepared stimulus (a wooden duck), Seligman erred in his statement that the experimenter "did not get fear conditioning to a wooden duck, even after many pairings with a startling noise" (1971, p. 315). In fact, the experimenter himself admitted that this failure was due to the inadequancy of his unconditioned stimulus, not to the inappropriate- ness of a wooden duck as a phobic stimulus:
We did not succeed in establishing a conditional fear re- sponse to the duck for the simple reason that the noise failed to evoke fear. Once only in something over fifty trials did the child exhibit what might be called a wor- ried look. (English, 1929, p. 222)
One can understand how the Albert study could be selectively misperceived by Seligman, since the errors that he committed result in a historical account that provides more support for the pre- dictions of his preparedness theory than does (sub- sequent) clinical observation (DeSilva, Rachman, & Seligman, 1977; Rachman & Seligman, 1976). It seems ironic that in making his case for the new theory of preparedness, Seligman first had to strengthen the old Watsonian interpretation of the Albert study: that it was a successful laboratory
156 • FEBRUARY 1979 • AMERICAN PSYCHOLOGIST
demonstration of fear conditioning, its generaliza- tion, and resistance to extinction.
What can be deduced from reviewing the many versions of Watson and Rayner's study of Albert? One somewhat obvious conclusion is that we should be extremely wary of secondhand (and more re- mote) accounts of psychological research. As Cornwell and Hobbs (1976) suggested, 'this may be most relevant to often-cited studies in psychology, since we may be more likely to overestimate our knowledge of such bulwarks of textbook knowledge.
What about the process by which secondary sources themselves come to err in their description of classic studies? A simple explanation might assume that more recent authors, like any recipi- ents of secondhand information (e.g., gossip), are more likely to present an account of much-cited research that has "drifted well away from the original" (Cornwell & Hobbs, 1976, p. 9). For the Albert study at least, this relatively passive model of communication is an oversimplified view. Not only was Watson quick to actively revise his own description of his research (e.g., Watson, 1928a; W(atson & Watson, 1921) but it took little time for textbook authors to alter the details of Albert's conditioning. For example, within a year after Watson's original article, one text (Smith & Guthrie, 1921) had already invented spurious stim- uli to which Albert's initial fear generalized; such errors were also contained in early texts by H. L. Hollingworth (1928) and J. W. Bridges (1930).
There has undoubtedly been some distortion due to the simple retelling of the Watson and Rayner study, but a more dynamic influence on textbook accounts seems to have been the authors' opinions of behaviorism as a valid theoretical view- point. For example, the agreement of Harvey Carr's (1925) text with Watson's overgeneraliza- tions about Albert was consistent with Carr's (1915) relatively favorable review of Watson's early work. Similarly, as behaviorism's influence grew, even relative skeptics seem to have been willing to devote more attention to the Albert study. For example, the fourth edition of Robert S. Woodworth's (1940) text, Psychology, men- tioned that Albert's "conditioned fear was 'trans- ferred' from the rat to similar objects" (p. 379), though the previous edition of the text (Wood- worth, 1934) did not mention this generalization and was more critical of Watson's theory of emo-
tional development. Woodworth's 1934 text also had Albert initially conditioned to a rabbit, while the 1940 one correctly described the conditioned stimulus of a rat. This greater accuracy in Wood- worth's later account is an indication of at least one author's ability to resist any general drift toward increasing misinformation.
Any attempted explanation of textbook errors concerning Albert raises the question of the role of classic studies and the nature of historical changes in psychology. As discussed by Samel- son (1974) and Baumgardner (1977), modern citations of classic studies can often be seen as attempts by current theorists to build a false sense of continuity into the history of psychology. In social psychology, for example, claiming Auguste Comte as a founder of the field (see Allport, 1968) gives the impression that our contemporary mo- tives (especially the wish for a well-developed behavioral science) have directed the field's prog- ress for almost a century (Samelson, 1974). To cite another classic "origin," the Army's psycho- logical testing program during World War I is taken by some clinical psychologists as an early example of how the profession of psychology has always grown in relation to its increased useful- ness. However, it has recently been shown that World War I intelligence testing was of little prac- tical use at the time (Samelson, 1977).
In reviewing these classic studies or origin myths in psychology, it should be emphasized that this myth^making process is not anyone's attempt to defraud the public. Instead, it arises "as largely a byproduct of pedagogy: as a means to elucidate •the concepts of a scientific specialty, to establish its tradition, and to attract students" (Samelson, 1974, p. 223). 4 This seems a fair explanation of the Albert case—one that casts the frequent "re- discoverers" of Watson and Rayner (Garrett, 1941; Harlow, 1949; Salter, 1949; Seligman, 1971; Watson, 1928a; Wolpe, 1958) as participants in the process of building historical support for new theoretical perspectives (e.g., preparedness theory).
As Samelson (1974) noted, the major difficulty with such reevaluations of classic studies is that they obscure the actual factors that determine the course of scientific research. In the case of the Albert study, debate still surrounds the question, How did behaviorism become a dominant force in American psychology (MacKenzie, 1972, 1977;
* Samelson noted that he was paraphrasing T. S. Kuhn (1968) here.
AMERICAN PSYCHOLOGIST • FEBRUARY 1979 • 157
Weimer & Palermo, 1973)? The answer is beyond the scope of this study, since it involves much more than an evaluation of the Albert study.5
However, it is now possible to assent that by itself the Albert study was not very convincing proof of the correctness of Watson's general view of personality and emotions. In addition to the study's reliance on only one subject, the experi- mental stimuli were insufficient to test for general- ization effects, the observers' accounts were too subjective, and the technology did not exist to permit reliable assessment of emotional responses (see Sherman, 1 9 2 7 ) ; there was insufficient fol- low-up and there was a confounding of instru- mental and classical conditioning paradigms (see Footnote 3). These methodological flaws were also apparent to critical reviewers of the day (e.g., English, 1929; Valentine, 1930) and surely to Watson and Rayner themselves. However, they are worth emphasizing here because of continuing attempts to integrate the study into the early conditioning literature (e.g., Seligman, 1971). It may be useful for modern learning theorists to see how the Albert study prompted subsequent research (i.e., Bregman, 1934), but it seems time, finally, to place the Watson and Rayner data in the category of "interesting but uninterpretable results."
5 Other relevant factors include the relation of Watson's theory to psychoanalysis (Bakan, 1966), the state in 1920 of- other theories of emotion, the impact of Watson's popu- lar writing (Buckley, Note 4; Harris & Morawski, Note 5), and Watson's national reputation.
1. Samelson, F. Reactions to Watson's behaviorism: The early years. Paper presented at the meeting of the Cheiron Society, Wellesley, Massachusetts, June 1978.
2. Jones, M. C. Personal communication, April 1, 1978. 3. Church, R. M. The role of fear in punishment. In
J. R. Braun (Chair), The effects of punishment on be- havior. Symposium presented at the meeting of the American Psychological Association, New York, Sep- tember 1966.
4. Buckley, K. W. Behaviorism and the professionaliza- tion of American psychology: A study of John Broadus Watson (1878-1958). Doctoral dissertation in prepara- tion, University of Massachusetts—Amherst, 1979.
5. Harris, B., & Morawski, J. John B. Watson's pre- dictions for 1979. Unpublished manuscript, Vassar Col- lege, 1978.
Allport, G. The historical background of modern social psychology. In G. Lindzey & E. Aronson (Eds.), The
handbook of social psychology (2nd ed., Vol. 1). Read- ing, Mass.: Addison-Wesley, 1968.
Bakan, D. Behaviorism and American urbanization. Journal of the History of the Behavioral Sciences, 1966, 2, 5-28.
Baumgardner, S. R. Critical studies in the history of social psychology. Personality and Social Psychology Bulletin, 1977, 3, 681-687.
Behaviorist babies. Review of Reviews, November 1928, pp. 548-549.
Bernhardt, K. S. Practical psychology (2nd ed.). New York: McGraw-Hill, 1953.
Biehler, R. F. Child development: An introduction. Bos- ton: Houghton Mifflin, 1976.
Boring, E. G., Langfeld, H. S., & Weld, H. P. (Eds.). Foundations of psychology. New York: Wiley, 1948.
Bregman, E. O. An attempt to modify the emotional attitudes of infants by the conditioned response tech- nique. Journal of Genetic Psychology, 1934, 45, 169*- 198.
Bridges, J. W. Psychology: Normal and abnormal. New York: Apple ton, 1930.
Calhoun, J. F. Abnormal psychology (2nd ed.). New York: CRM Books, 1977.
Can science determine your baby's career before it can talk? New York American Sunday Magazine, January 8, 1922.
Carr, H. A. Review of Behavior, an introduction to comparative psychology by J. B. Watson. Psychological Bulletin, 1915,12, 308-312.
Carr, H. A. Psychology: A study of mental activity. New York: Longmans, Green, 1925.
Cornwell, D., & Hobbs, S. The strange saga of little Albert. New Society, March 18, 1976, pp. 602-604.
Costello, C. G. Dissimilarities between conditioned avoid- ance responses and phobias. Psychological Review, 1970, 77, 250-254.
Craig, G. J. Human development. Englewood Cliffs, N. J.: Prentice-Hall, 1976.
CRM Books. Developmental psychology today. Del Mar, Calif.: Author, 1971.
DeSilva, P., Rachman, S., & Seligman, M. E. P. Prepared phobias and obsessions: Therapeutic outcome. Be- haviour Research and Therapy, 1977, 15, 65-77.
Ellenberger, H. F. The story of "Anna O": A critical review with new data. Journal of the History of the Behavioral Sciences, 1972, 8, 267-279.
Engle, T. L., & Snellgrove, L. Psychology: Its principles and applications (5th ed.). New York: Harcourt, Brace & World, 1969.
English, H. B. Three cases of the "conditioned fear response." Journal of Abnormal and Social Psychology, 1929, 34, 221-225.
Eysenck, H. J. Learning theory and behaviour therapy. In H. J. Eysenck (Ed.), Behaviour therapy and the neuroses: Readings in modern methods of treatment derived from learning theory. Oxford, England: Per- gamon Press, 1960.
Galanter, E. Textbook of elementary psychology. San Francisco: Holden-Day, 1966.
Gardiner, W. L. Psychology: A story of a search. Bel- mont, Calif.: Brooks/Cole, 1970.
Garrett, H. E. Great experiments in psychology. New York: Appleton-Century-Crofts, 1941.
Goldenberg, H. Abnormal psychology. Monterey, Calif.: Brooks/Cole, 1977.
Harlow, H. F. The formation of learning sets. Psycho- logical Review, 1949, 56, 51-65.
Helms, D. B., & Turner, J. S. Exploring child behavior. Philadelphia, Pa.: Saunders, 1976.
158 • FEBRUARY 1979 • AMERICAN PSYCHOLOGIST
Herrnstein, R. J. The evolution of behaviorism. Ameri- can Psychologist, 1977, 32, 593-603.
Hilgard, E. R., Atkinson, R. C., & Atkinson, R. L. Psy- chology (6th ed.). New York: Harcourt Brace Jovano- vich, 1975.
Hilgard, E. R., & Marquis, D. G. Conditioning and learning. New York: Appleton-Century, 1940.
Hineline, P. N. Negative reinforcement and avoidance. In W. K. Honig & J. E. R. Staddon (Eds.), Handbook of operant behavior (Vol. 2). Englewood Cliffs, N.J.: Prentice-Hall, 1977.
Hollingworth, H. L. Mental growth and decline. New York: Appleton, 1928.
Johnson, R. C., & Medinnus, G. R. Child psychology: Behavior and development (3rd ed.). New York: Wiley, 1074.
Jones, H. E. The retention of conditioned emotional reactions in infancy. Journal of Genetic Psychology, 1930, 37, 485-497.
Jones, M. C. The elimination of children's fears. Journal of Experimental Psychology, 1924, 7, 383-390. (a)
Jones, M. C. A laboratory study of fear: The case of Peter. Pedagogical Seminary, 1924, 31, 308-315. (t>)
Jones, M. C. A 1924 pioneer looks at behavior therapy. Journal of Behavior Therapy and Experimental Psy- chiatry, 1975, 6, 181-187.
Kazdin, A. E. History of behavior modification: Experi- mental foundations of contemporary research. Balti- more, Md.: University Park Press, 19i78.
Kennedy, W. A. Child psychology (2nd ed.). Englewood Cliffs, N.J.: Prentice-Hall, 1975.
Kisker, G. W. The disorganized personality (3rd ed.). New York: McGraw-Hill, 1977.
Kleinmuntz, B. Essentials of abnormal psychology. New York: Harper & Row, 1974.
Koch, S. Psychology and emerging conceptions of knowledge as unitary. In T. Wann (Ed.), Behaviorism and phenomenology: Contrasting bases for modern psy- chology. Chicago: University of Chicago Press, 1964.
Kuhn, T. S. Science: I. The history of science. In D. L. Sills (Ed.), International encyclopedia of the social sciences (Vol. 14). New York: Macmillan, 1968.
Larson, C. Some further notes on the "rat rabbit" problem and John B. Watson. Teaching of Psychology, 1978, S, 35.
MacKenzie, B. D. Behaviourism and positivism. Journal of the History of the Behavioral Sciences, 1972, 8, 222- 2*1.
MacKenzie, B. D. Behaviourism and the limits of sci- entific method. Atlantic Highlands, N.J.: Humanities Press, 1977.
Marks, I. Fears and phobias. New York: Academic Press, 1969.
Marks, I. Phobias and obsessions: Clinical phenomena in search of a laboratory model. In J. D. Maser & M. E. P. Seligman (Eds.), Psycho pathology: Experimental models. San Francisco: Freeman, 19'77.
McCandless, B. R., & Trotter, R. J. Children: Behavior and development (3rd ed.). New York: Holt, Rinehart & Winston, 1977.
Oilman, A., Erixon, G., & Lofberg, I. Phobias and pre- paredness: Phobic versus neutral pictures as condi- tioned stimuli for human autonomic responses. Journal of Abnormal Psychology, 1975, 84, 41-^5.
Page, J. D. Psychopathology: The science of understand- ing deviance (2nd ed.). Chicago: Aldine, 19i75.
Papalia, D. E., & Olds, S. W. A child's -world: Infancy through adolescence. New York: McGraw-Hill, 1975.
Parsons, H. M. What happened at Hawthorne? Science, 1974, 183, 922-932.
Prytula, R. E., Oster, G. D., & Davis, S. F. The "rat rabbit" problem: What did John B. Watson really do? Teaching of Psychology, 1977, 4, 44-46.
Rachman, S. Learning theory and child psychology: Therapeutic possibilities. In H, J. Eysenck (Ed.), Ex- periments in behaviour therapy. Oxford, England: Pergamon Press, 1964.
Rachman, S., & Seligman, M. E. P. Unprepared phobias: "Be prepared." Behaviour Research and Therapy, 1976, 14, 333-338.
Reese, H. W., & Lipsitt, L. P. Experimental child psy- chology. New York: Academic Press, 1970.
Salter, A. Conditioned reflex therapy. New York: Crea- tive Age Press, 1949.
Samelson, F. History, origin myth and ideology: "Dis- covery" of social psychology. Journal for the Theory of Social Behaviour, 1974, 4, 2J7-231.
Samelson, F. World War I intelligence testing and the development of psychology. Journal of the History of the Behavioral Sciences, 1977, 13, 274-282.
Sandier, J., & Davidson, R. S. Psychopathology: An analysis of response consequences. In H. D. Kimmel (Ed.), Experimental Psychopathology: Recent research and theory. New York: Academic Press, 1971.
Schwartz, B. On going back to nature. Journal of the Experimental Analysis of Behavior, 1974, 21, 183-198.
Seligman, M. E. P. On the generality of the laws of learning. Psychological Review, 19>70, 77, 406-418.
Seligman, M. E. P. Phobias and preparedness. Behavior Therapy, 1971, 2, 307-320.
Sherman, M. The differentiation of emotional responses in infants: I. Judgments of emotional responses from motion picture views and from actual observation. Journal of Comparative Psychology, 1927, 7, 265-284.
Shettleworth, S. J. Food reinforcement and the organiza- tion of behaviour in golden hamsters. In R. A. Hinde & J. Stevenson-Hinde (Eds.), Constraints on learning. London: Academic Press, 1973.
Smith, S., & Guthrie, E. R. General psychology in terms of behavior. New York: Appleton, 1921.
Staats, A. W. Learning, language -and cognition. New York: Holt, Rinehart & Winston, 1968.
Telford, C. W., & Sawrey, J. M. Psychology. Belmont, Calif.: Brooks/Cole, 1968.
Thorndike, E. L. The psychology of wants, interests, and attitudes. New York: Appleton-Century, 1935.
Ullman, L. P., & Krasner, L. Introduction. In L. P. Ullman & L. Krasner (Eds.), Case studies in behavior modification. New York: Holt, Rinehart & Winston, 1965.
Valentine, C. W. The innate bases of fear. Journal of Genetic Psychology, 1930, 37, 394-420.
Watson, J. B. Behavior and the concept of mental dis- ease. Journal of Philosophy, 1946, 13, 589-597. (a)
Watson, J. B. The place of the conditioned reflex in psy- chology. Psychological Review, 1916, 23, 89-116. (b)
Watson, J. B. The psychology of wish fulfillment. Sci- entific Monthly, 1916, 3, 479-487. (c)
Watson, J. B. Experimental investigation of babies (Film). Chicago: Stoelting, 1919. (a) (Psychological Abstracts, 1937,11, No. 6061.)
Watson, J. B. A schematic outline of the emotions. Psychological Review, 1919, 26, 165-196. (b)
Watson, J. B. Behaviorism. New York: Norton, 19'24. Watson, J. B. Experimental studies on the growth of
the emotions. Pedagogical Seminary, 1925, 32, 328- 348. (a)
Watson, J. B. Recent experiments on how we lose and change our emotions. Pedagogical Seminary, 19'25, 32, 349-371. (b)
AMERICAN PSYCHOLOGIST • FEBRUARY 1979 • 159
Watson, J. B. What the nursery has to say about in- stincts. Pedagogkal Seminary, 1925, 32, 293-327. (c)
Watson, J. B. Can psychology help me rear my child? McCall's, September 1927, pp. 44; 72.
Watson, J. B. Psychological care of infant and child. New York: Norton, 1928. (a)
Watson, J. B. The heart or the intellect. Harper's Mag- azine, February 1928, pp. 345-352. (b)
Watson, J. B. What about your child? Cosmopolitan, October 1928, pp. 76-77; 108; 110; 112. (c)
Watson, J. B., & Morgan, J. J. B. Emotional reactions and psychological experimentation. American Journal of Psychology, 1917, 28, 163-174.
Watson, J. B., & Rayner, R. Conditioned emotional re- actions. Journal of Experimental Psychology, 1920, 3, 1-14.
Watson, J. B., & Watson, R. R. Studies in infant psy- chology. Scientific Monthly, 1921, 13, 493-515.
Weimer, W. B., & Palermo, D. S. Paradigms and normal
science in psychology. Science Studies, 1973, 3, 211- 244.
Weiner, B. (Ed.). Discovering psychology, Chicago: Science Research Associates, 1977.
Whitehurst, G. J., & Vasta, R. F. Child behavior. Bos- ton: Houghton Mifflin, 1977.
Whittaker, J. O. Introduction to psychology. Philadel- phia, Pa.: Saunders, 1965.
Wolpe, J. Psychotherapy by reciprocal inhibition. Stan- ford, Calif.: Stanford University Press, 1958.
Wolpe. J. The practice of behavior therapy (2nd ed.). New York: Pergamon Press, 1973.
Wolpe, J., & Rachman, S. Psychoanalytic "evidence": A critique based on Freud's case of Little Hans. Journal of Nervous and Mental Disease, 1960, 130, 135-148.
Woodworth, R. S. Psychology (3rd ed.). New York: Holt, 1934.
Woodworth, R. S. Psychology (4th ed.). New York: Holt, 1940.
160 • FEBRUARY 1979 • AMERICAN PSYCHOLOGIST