Imitation-baer-et-al-1967 roca

JOURNAL OF THE EXPERIMENTAL ANALYSIS OF BEHAVIOR

THE DEVELOPMENT OF IMITATION BY REINFORCINGBEHAVIORAL SIMILARITY TO A MODEL1

DONALD M. BAER,2,3 ROBERT F. PETERSON, AND JAMES A. SHERMAN3

UNIVERSITY OF KANSAS, UNIVERSITY OF ILLINOIS,AND UNIVERSITY OF KANSAS

This research demonstrated some of the conditions under which retarded children can betaught to imitate the actions of adults. Before the experiment, the subjects were withoutspontaneous imitative behavior, either vocal or motor. Each subject was taught, with foodas reinforcement, a series of responses identical to responses demonstrated by an experimenter;i.e., each response was reinforced only if it was identical to a prior demonstration by anexperimenter. Initially, intensive shaping was required to establish matching responses bythe subjects. In the course of acquiring a variety of such responses, the subjects' probabilityof immediate imitation of each new demonstration, before direct training, greatly increased.Later in the study, certain new imitations, even though perfect, were never reinforced;yet as long as some imitative responses were reinforced, all remained at high strength. Thisimitativeness was then used to establish initial verbal repertoires in two subjects.

CONTENTS

MethodSubjectsFirst training proceduresFurther training procedures

Probes for imitationNon-reinforcement of all imitationImitative chainsVerbal imitationsGeneralization to other experimenters

ResultsReliability of scoring imitative responsesFirst training proceduresDRO proceduresImitative chainsVerbal behaviorGeneralization to other experimenters

DiscussionThe development of a class of behaviors

which may fairly be called "imitation" is aninteresting task, partly because of its relevanceto the process of socialization in general andlanguage development in particular, andpartly because of its potential value as a train-ing technique for children who require specialmethods of instruction. Imitation is not a spe-cific set of behaviors that can be exhaustivelylisted. Any behavior may be considered imita-tive if it temporally follows behavior demon-strated by someone else, called a model, and if

its topography is functionally controlled bythe topography of the model's behavior. Spe-cifically, this control is such that an observerwill note a close similarity between the topog-raphy of the model's behavior and that of theimitator. Furthermore, this similarity to themodel's behavior will be characteristic of theimitator in responding to a wide variety ofthe model's behaviors. Such control could re-sult, for example, if topographical similarityto a model's behavior were a reinforcing stim-ulus dimension for the imitator.There are, of course, other conditions which

can produce similar behaviors from two orga-nisms on the same occasion, or on similar occa-sions at different times. One possibility is that

1A portion of this research was presented at thebiennial meeting of the Society for Research in ChildDevelopment, Minneapolis, Minnesota, March, 1965.This research was supported by PHS grant MH-02208,National Institute of Mental Health, entitled AnExperimental Analysis of Social Motivation. Mr. FrankJunkin, Superintendent, Dr. Ralph Hayden, MedicalDirector, and other members of the staff of the Fir-crest School, Seattle, Washington, made space andsubjects available. We wish to thank Mrs. Joan Beaversfor her help as a "new" experimenter in the testsof generalization and for assistance in the preparationof this manuscript.

2Reprints may be obtained from Donald M. Baer,Department of Human Development, University ofKansas, Lawrence, Kansas 66044.3Research Associates, Bureau of Child Research,

University of Kansas.

405

1967, 10, 405-416 NUMBER 5 (SEPTEMBER)

DONALD M. BAER, et al.

both organisms independently have beentaught the same responses to the same cues;thus, all children recite the multiplication ta-bles in very similar ways. This similarity doesnot deserve the label imitation, and hardlyever receives it; one child's recitation is notusually a cue to another's, and the similarityof their behavior is not usually a reinforcerfor the children. Nevertheless, the children ofthis example have similar behaviors.The fact that the world teaches many chil-

dren similar lessons can lead to an arrange-ment of their behaviors which comes closer toa useful meaning of imitation. Two childrenmay both have learned similar responses; onechild, however, may respond at appropriatetimes whereas the other does not. In that case,the undiscriminating child may learn to usethis response when the discriminating onedoes. The term imitation still need not be ap-plied, since the similarity between the twochildren's responses is not functional for eitherof them; in particular, the second child is notaffected by the fact that his behavior is similarto that of the first. This arrangement ap-proaches one which Miller and Dollard (1941)call "matched-dependent" behavior. One or-ganism responds to the behavior of anothermerely as a discriminative stimulus with re-spect to the timing of his own behavior; manytimes, these behaviors will happen to be alike,because both organisms will typically use themost efficient response, given enough experi-ence.

It should be possible, however, to arrangethe behavior of two organisms so that one ofthem will, in a variety of ways, produce precisetopographical similarity to the other, butnothing else. A study by Baer and Sherman(1964) seemingly showed the result of suchprior learning in several young children. Inthat study, reinforcements were arranged forchildren's imitations of three activities of ananimated, talking puppet, which served bothas a model and a source of social reinforce-ment for imitating. As a result of this rein-forcement, a fourth response of the puppetwas spontaneously imitated by the children,although that imitation had never before beenreinforced. When reinforcement of the otherthree imitations was discontinued, the fourth,never-reinforced imitation also decreased instrength; when reinforcement of the originalimitations was resumed, imitation of the

fourth response again rose in rate, although itstill was never reinforced. In short, these chil-dren apparently generalized along a stimulusdimension of similarity between their behav-iors and the behaviors of a model: when simi-larity to the model in three different ways wasreinforced, they thereupon displayed a fourthway of achieving similarity to the model.Thus, similarity between their behavior andthe model's was a functional stimulus in theirbehavior.Metz (1965) demonstrated the development

of some imitative behavior in two autistic chil-dren who initially showed little or no imita-tive response. In this study, responses similarin topography to demonstrations by the ex-perimenter were reinforced with "Good" andfood. Metz found that, after intensive training,several imitative responses could be main-tained in strength even when not reinforcedwith food, and that the subjects had a higherprobability of imitating new responses aftertraining than before. However, in one of theconditions used to evaluate the subjects' imi-tative repertoire before and after imitativetraining, "Good" was still said contingentupon correct new imitations. Thus, for onesubject who initially showed a non-zero rateof imitation, it could be argued that the in-creased imitation in the test after training wasdue to an experimentally developed reinforc-ing property of "Good", rather than to theimitation training as such. Further, in theMetz study, due to a lack of extinction orother manipulation of the behavior, it is dif-ficult to specify that the higher probability ofimitating new responses, and the maintenanceof unreinforced imitative responses, were infact due to the reinforcement of the initialimitative responses during training.

Lovaas, Berberich, Perloff, and Schaeffer(1966) used shaping and fading procedures toestablish imitative speech in two autistic chil-dren. They reported that as training progressedand more vocal behavior came under the con-trol of a model's prior vocalization, it becameprogressively easier to obtain new imitativevocalizations. When reinforcement was shiftedfrom an imitative-contingent schedule to abasically non-contingent schedule, imitativebehavior deteriorated. In an additional ma-nipulation, the model presented Norwegianwords interspersed with English words for thechildren to imitate. Initially, the children did

406

THE DEVELOPMENT OF IMITATION

not reproduce the Norwegian words perfectly.However, the authors judged that the subjectsgradually improved their imitations of theNorwegian words even though these imitationswere not reinforced.The studies by Baer and Sherman (1964),

Metz (1965), Lovaas et al (1966), and other re-ports (Bandura, 1962) suggest that for childrenwith truly imitative repertoires, induction hasoccurred, such that (1) relatively novel behav-iors can be developed before direct shaping,merely by providing an appropriate demon-stration by a model, and (2) some imitativeresponses can be maintained, although unrein-forced, as long as other imitative responses arereinforced.The purpose of the present study was to ex-

tend the generality of the above findings andto demonstrate a method of producing a trulyimitative repertoire in children initially lack-ing one.

METHOD

SubjectsThree children, 9 to 12 years of age, were

selected from several groups of severely andprofoundly retarded children in a state school.They were chosen not because they were re-tarded, but because they seemed to be the onlychildren available of a practical age who ap-parently showed no imitation whatsoever.(The success of the method to be describedsuggests that it may have considerable practi-cal value for the training of such children.)The subjects were without language, but madeoccasional grunting vocalizations, and re-sponded to a few simple verbal commands("Come here", "Sit down", etc.). They wereambulatory (but typically had developed walk-ing behavior relatively late in their develop-ment, in the sixth or seventh year), could dressthemselves, were reasonably well toilet trained,and could feed themselves. Fair eye-hand coor-dination was evident, and simple manipula-tory skills were present.The subjects were chosen from groups of

children initially observed in their wards froma distance over a period of several days. Noinstances of possible imitation were noted inthe subjects finally selected. (That is, on nooccasion did any subject display behavior sim-ilar to that of another person, except in in-stances where a common stimulus appeared to

be controlling the behaviors of both persons,e.g., both going to the dining area when foodwas displayed on the table.) Subsequently, anexperimenter approached and engaged thesubjects in extended play. In the course of thisplay, he would repeatedly ask them to imitatesome simple response that he demonstrated,such as clapping his hands, or waving. Thechildren failed to imitate any of these re-sponses, although they clearly were capable ofat least some of them. Finally, during thetraining itself, every sample of behavior wasinitially presented to the child as a demon-stration accompanied by the command, "Dothis"; at first, none of these samples was imi-tated, despite extensive repetition.

First Training ProceduresEach subject was seen at mealtimes, once or

twice a day, three to five times a week. Thesubject's food was used as a reinforcer. It wasdelivered a spoonful at a time by the experi-menter, who always said "Good" just beforeputting the spoon into the subject's mouth.The subject and experimenter faced eachother across the corner of a small table, onwhich were placed the food tray and the ex-perimenter's records. Elsewhere in the roomwas another small table on which were placedsome materials used later in the study, a deskwith a telephone on it, a coat rack holding oneor more coats, a wastebasket, and a few otherchairs.The basic procedure was to teach each sub-

ject a series of discriminated operants. Eachdiscriminated operant consisted of three ele-ments: a discriminative stimulus (SD) pre-sented by the experimenter, a correct responseby the subject, and reinforcement after a cor-rect response. The SD was the experimenter'scommand, "Do this", followed by his demon-stration of some behavior. The response re-quired was one similar to the experimenter's.Thus, the operant learned was always topo-graphically imitative of the experimenter'sdemonstration. The reinforcement was food,preceded by the word "Good".

Since none of the subjects was imitative,none of the initial SD's was followed by anybehavior which resembled that demonstratedby the experimenter. This was true even forthose behaviors which the subjects were clearlycapable of performing. Subject 1, for example,would sit down when told to, but did not imi-

407


tate the experimenter when he said "Do this",sat down, and then offered her the chair.Hence, the initial imitative training for allsubjects was accomplished with a combinationof shaping (Skinner, 1953) and fading (Ter-race, 1963a, 1963b) or "putting through" pro-cedures (Konorski and Miller, 1937).The first response of the program for Sub-

ject 1 was to raise an arm after the experi-menter had raised his. The subject waspresented with a series of arm-raising demon-strations by the experimenter, each accompa-nied by "Do this", to which she made no re-sponse. The experimenter then repeated thedemonstration, reached out, took the subject'shand and raised it for her, and then immedi-ately reinforced her response. After severaltrials of this sort, the experimenter begangradually to fade out his assistance by raisingthe subject's arm only part way and shapingthe completion of the response. Gradually, theexperimenter's assistance was faded until thesubject made an unassisted arm-raising re-sponse whenever the experimenter raised hisarm. The initial responses for all subjects weretaught in this manner whenever necessary.

Occasionally during the very early trainingperiods a subject would resist being guidedthrough a response. For example, with a re-sponse involving arm raising, Subject 3 at firstpulled his arm downward whenever the ex-perimenter attempted to raise it. In this case,the experimenter merely waited and triedagain until the arm could be at least partiallyraised without gteat resistance; then the re-sponse was reinforced. After subjects had re-ceived a few reinforcements following theexperimenter's assistance in performing a re-sponse, they no longer resisted. As the numberof responses in the subjects' repertoire in-creased, the experimenter discontinued theguiding procedure and relied only on shapingprocedures when a response did not match thedemonstration.A number of responses, each topographi-

cally -similar to a demonstration by the experi-menter, was taught to each subject. Trainingof most responses was continued until its dem-onstration was reliably matched by the subject.The purpose of these initial training proce-dures was to program reinforcement, in asmany and diverse ways as practical, whenevera subject's behavior was topographicallyf simi-lar to that demonstrated by the experimenter.

Further Training ProceduresProbes for imitation. As the initial training

procedures progressed, and the subjects beganto come under the control of the experiment-er's demonstrations, certain responses weredemonstrated which, if imitated perfectly ontheir first presentation, were deliberately notreinforced on the first or any future occasion.These responses served as probes for the devel-oping imitative nature of the subject's reper-toire. A list of the responses demonstrated, in-cluding the reinforced ones for the initialtraining procedure and the unreinforcedprobe demonstrations, is given in Table 1 forSubject 1. These responses are listed in theorder of first demonstration. Subject 1 had 95reinforced and 35 unreinforced responses. Sim-ilar responses were used with Subjects 2 and 3.Subject 2 had 125 reinforced and five unrein-forced probes; Subject 3 had eight reinforcedresponses and one unreinforced probe.During the probes, the experimenter con-

tinued to present SD's for imitation. If the re-sponse demonstrated belonged to the group ofreinforced responses and the subject imitatedwithin 10 sec, reinforcement ("Good" andfood) was delivered and the next response wasdemonstrated. If the subject did not imitatewithin 10 sec, no reinforcement was deliveredand the experimenter demonstrated the nextresponse. If it belonged to the unreinforcedgroup of responses (probes), and if the subjectimitated it, there were no programmed conse-quences and the experimenter demonstratedthe next response no sooner than 10 sec afterthe subject's imitation. If it was not imitated,the experimenter performed the next demon-stration 10 sec later. The purpose of the 10-secdelay was to minimize the possibility that thesubjects' unreinforced imitations were beingmaintained by the possible reinforcing effectsof the presentation of an SD for a to-be-rein-forced imitative response. Demonstrations forreinforced and unreinforced responses werepresented to subjects in any unsystematicorder.Non-reinforcement of all imitation. After

the probe phase, and after stable performancesof reinforced and unreinforced imitative re-sponses were established, non-reinforcement ofall imitative behavior was programmed. Thepurpose of this procedure was to show the de-pendence of the imitative repertoire on the

408

Table 1The Sequence of Responses Demonstrated to Subject 1

(Asterisks indicate unreinforced responses.)1. Raise left arm. 66. Throw box2. Tap table with left hand *67. Walk to telephone3. Tap chest with left hand *68. Extend both arms (sitting)4. Tap head with left hand 69. Walk and tap head with left hand5. Tap left knee with left hand 70. Walk and tap head with right hand6. Tap right knee with left hand *71. Walk and clap hands7. Tap nose *72. Open mouth

*8. Tap arm of chair 73. Jump9. Tap leg of table 74. Pat radiator10. Tap leg with left hand *75. Nod no11. Extend left arm 76. Pick up phone12. Make circular motion with arm 77. Pull drawer13. Stand up 78. Pet coat14. Both hands on ears 79. Tear kleenex15. Flex arm 80. Nest four boxes16. Nod yes 81. Point gun and say "Bang"17. Tap chair seat *82. Put towel over face18. Extend both arms *83. Put hands over eyes19. Put feet on chair *84. Tap floor20. Walk around *85. Scribble21. Make vocal response *86. Move toy car on table22. Extend right arm sideways 87. Place circle in form board23. Tap shoulder 88. Place circle, square, and triangle in form board24. Tap head with right hand *89. Crawl under table25. Tap right knee with right hand *90. Walk and clap sides26. Tap leg with right hand *91. Lie on floor27. Tap left knee with right hand *92. Kick box28. Raise right arm overhead *93. Put foot over table rung29. Tap chest with right hand *94. Fly airplane30. Tap table with right hand *95. Rock doll31. Move chair *96. Burp doll32. Sit in chair *97. Tap chair with bat33. Throw paper in basket *98. Open and close book34. Pull up socks 99. Work egg beater35. Tap desk 100. Put arm through hoop36. Climb on chair 101. Build three block tower37. Open door *102. Stab self wth rubber knife38. Move ash tray 103. Put blocks in ring39. Put paper in chair 104. Walk and hold book on head40. Sit in two chairs (chained) 105. Ride kiddie car41. Tap chair with right hand 106. Sweep with broom42. Move paper from basket to desk 107. Place beads around neck43. Move box from shelf to desk 108. Ride hobby horse44. Put on hat *109. Put on glove45. Move hat from table to desk 110. Use whisk broom on table46. Move box from shelf to desk 111. Work rolling pin47. Nest three boxes *112. Push large car48. Put hat in chair 113. Put beads on doorknob49. Tap wall *114. Put hat on hobby horse50. Move waste basket 115. Sweep block with broom51. Move paper from desk to table 116. Place box inside ring of beads52. Stand in corner 117. Put glove in pocket of lab coat53. Pull window shade 118. Push button on tape recorder54. Place box in chair *119. Bang spoon on desk55. Walk around desk 120. Lift cup56. Smile 121. Use whisk broom on a wall57. Protrude tongue *122. Put a cube in a cup58. Put head on desk 123. Rattle a spoon in a cup59. Ring bell *124. Throw paper on the floor60. Nest two boxes *125. Hug a pillow61. Crawl on floor 126. Tap pegs into pegboard with hammer*62. Walk with arms above head *127. Wave a piece of paper63. Sit on floor *128. Shake a rattle64. Put arm behind back (standing) *129. Hit two spoons together65. Walk with right arm held up 130. Shake a tambourine

409


food reinforcement which was apparently re-sponsible for its development.

Non-reinforcement of imitation was insti-tuted in the form of reinforcement for any be-havior other than imitation. Differential rein-forcement of other behavior is abbreviatedDRO (Reynolds, 1961). The experimenter con-tinued saying "Good" and feeding the subject,but not contingent on imitations. Instead, theexperimenter delivered reinforcement at least20 sec after the subject's last imitation hadtaken place. Thus, for the group of previouslyreinforced responses, the only change betweenreinforcement and non-reinforcement periodswas a shift in the contingency. For the groupof unreinforced or probe responses there wasno change; food reinforcement still did notfollow either the occurrence or non-occurrenceof an imitative response. This procedure in-volved simultaneously the extinction of imi-tation and also the reinforcement of whateverother responses may have been taking placeat the moment of reinforcement.

For Subject 1, the DRO period was 30 sec.For Subject 2, DRO periods were 30, 60, and0 sec' (DRO Q-sec meant reinforcement was de-livrered immediately after the SD, before animitative response could occur.) This sequenceof DRO intervals was used because, as dis-played in the- Results section, Subject 2 main-tained stable imitation under the initial DROprocedures, unlike the other subjects. For Sub-ject 3, the DRO period was 20 sec. After theDRQ procedure for each subject, contingentreinforcement of imitation was resumed andthe procedures described below were insti-tuted.

Imitative chains. After reinforcement forimitative behavior was resumed with Subjects1 and 2, the procedure of chaining togetherold and new imitations was begun. At firstonly two-response chains were demonstrated;then three-response chains, after two-responsechains were successfully achieved; and so on.During chaining, the experimenter demon-strated the responses the subject was to imi-tate as an unbroken series. In all cases, thedemonstrated chain contained both responsespreviously learned by the subject and rela-tively new ones. Walking from one locale toanother in the process of performing these be-'haviors was not considered part of the imita-tive chain and was not judged for imitativeaccuracy.

Verbal imitations. Late in the training pro-gram for Subjects 1 and 3, when virtually anyAnew motor performance by the experimenterwas almost certain to be imitated, vocal per-formances were begun with simple sounds.The experimenter, as usual, said "Do this",but instead of making some motor responsemade a vocal one, for example, "Ah". Sub-jects I and 3 repeatedly failed to imitate suchdemonstrations. Different procedures werethen employed to obtain vocal imitations. ForSubject 1, the vocal response to be imitatedwas set into a chain of non-vocal responses.For example, the experimenter would say,."Do this", rise from his chair and walk to thecenter of the room, turn towards the subject,say "Ah", and return to his seat. To such ademonstration Subject 1 responded by leav-ing her seat, walking toward the center of theroom, turning toward the experimenter, andthen beginning a series of facial and vocal re-sponses out of which eventually emerged an"Ah" sufficiently similar to the experimenter'sto merit reinforcement. This coupling ofmotor and vocal performances was maintainedfor several more demonstrations, during whichthe motor performance was made successivelyshorter and more economical of motion; fin-ally, the experimenter was able to remainseated, say "Do this", say "Ah", and immedi-ately evoke an imitation from the subject.Proceeding in this manner, simple sounds wereshaped and then combined into longer ormore complex sounds and finally into usablewords.

Subject 3, like Subject 1, initially failed toimitate vocalizations. In his case, the experi-menter proceeded to demonstrate a set ofmotor performances which moved successivelycloser to vocalizations. At first the experi-menter obtained imitative blowing out of alighted match, then blowing without thematch, then more vigorous blowing whichincluded an initial plosive "p", then added avoiced component to the blowing which wasshaped into a "Pah" sound. Proceeding in thismanner, a number of vocalizations were pro-duced, all as reliable imitations.

Generalization to other experimenters.When the imitative repertoire of Subject 1had developed to a high level, new experi-menters were presented to her, of the oppositeor the same sex as the original male experi-menter. These novel experimenters gave the

410


same demonstrations as the original experi-menter in the immediately preceding session.The purpose of this procedure was to investi-gate whether the subject's imitative repertoirewas limited to demonstrations by the originalmale experimenter. During this procedure,the new experimenters delivered reinforce-ment in the same manner as the original ex-perimenter; i.e., previously reinforced imita-tions were reinforced and probes were not.

RESULTS

Reliability of Scoring Imitative ResponsesChecks on the reliability of the experiment-

er's scoring of any response as imitative weremade at scattered points throughout the studyfor Subjects 1 and 2. The percentage of agree-ment between the experimenter's scoring andthe independent records of a second observerexceeded 98%.

First Training ProceduresThe initial training procedure contained

occasions when the extent of the developingimitative repertoire of each subject could beseen. These were occasions when behavior wasdemonstrated by the experimenter to thesubject for the first time. Any attempt by thesubject to imitate such new behavior beforedirect training or shaping could be attributedto the history of reinforcement for matchingother behavior of the experimenter. Thus, itwas possible to examine the sequence of initialpresentations to each subject to discover anyincreasing probability that new behaviorwould be imitated on its first presentation.The sequence of 130 responses in Subject 1's

program was sufficient to increase her proba-bility of imitating new responses from zero atthe beginning of the program to 100% at theend. This was demonstrated by grouping the130 responses into 13 successive blocks of 10each. As shown in Fig. 1, the proportion imi-tated on the first presentation within eachblock rose, not too steadily, but nonethelessclearly, to 100% by the 13th block.The proportion of new responses success-

fully imitated by Subject 2 upon their firstpresentation rose from 0% to 80%, througha sequence of 130 new responses, as shown inFig. 2.

Subject 2 displayed both more variable andless thorough imitation of new responses on

BLOCKS OF10 NEW RESPONSES

Fig. 1. The development of imitation in Subject 1.

their first presentation than did Subject 1, al-though the general form of the data is similar.

Subject 3 was taught only eight discrimi-nated operants of imitative topography, whichhe acquired much more rapidly than dideither Subject 1 or 2. He imitated the ninthspontaneously on its first presentation, al-though he had not imitated it before training.The progressive development of imitation

was apparent in other aspects of the data aswell. The number of training sessions requiredto establish new imitations was displayed byplotting this number of sessions for each suc-cessive block of 10 new responses. The cri-terion for establishment of a new imitativeresponse was that, for one trial, a subject dis-played the response demonstrated by the ex-perimenter with no shaping or fading pro-cedures require(l for that trial. This is shownin Fig. 3 for Subject 1 and in Fig. 4 for Sub-ject 2, as solid lines. Both graphs show a syste-matically decreasing number of sessions re-quire(l to establish successive new imitations.The dotted portions of each graph representdeviations from the usual type of training

0 _ SUBJECT #2w

0e0

0 ~ ~ ~~1 NE-EPNE

Fizg

z ~0t

ILI

BLOCKS OF10 NEW RESPONSES

Fig. 2. The development of imitation in Subject 2.

'"ISUBJECT #1

'UU) zo 2w0IA 0 '

U Z

z ae wO0 0

S.- Uz I.-

U ce

'U 90.

411


procedure and thus are plotted differently.For Subject 1 the dotted portion represents aperiod in which verbal responses were intro-duced (not plotted as part of Fig. 3, but dis-cussed later in this report). For Subject 2 thedotted portion represents a sequence of ses-sions in which few new imitative responseswere introduced. Rather, two previously es-tablished imitative responses of similar to-

pography, which the subject no longer clearlydisplayed, were worked on intensively.

DRO ProceduresFor all subjects, both reinforced and unrein-

forced imitative behavior was maintained overcontinuing experimental sessions as long asfood reinforcement was contingent upon at

2 3 4 S a to a

BLOCKS OF

10 NEW RESPONSESrate of development of imitation in

a

g 0 U2

cs F- zA '

0

0i *- W.LUi Z 4

Fig. 4. TISubject 2.

SUBJECT #2

2 3 4a3 a

BLOCKS OF

10 NEW RESPONSES

he rate of development of imitation in

:7

SUBJECT # I100 nVW

0@

.6*

SO

50~~~~~~

REINFORCED MIT.

UNREINFORCED IMIT.

00

0V0

0

1oo 105

SESSIONSFig. 5. The maintenance and extinction of reinforced and unreinforced imitation in Subject 1. (The breaks

in the data before and after session 160 represent periods of experimentation aimed at other problems.)

412

4

n-

n-

The

a

hi U)

at

0

z

0°

4c

Z) a

U) U0 3

Fig. 3.Subject 1.

(I)

z

0

--C')

z ow

ZLd0

.z

w

w

0


least some imitative behavior. When reinforce-ment was no longer contingent upon imitativebehavior during the DRO periods, both thepreviously reinforced imitations and the never-

reinforced probe imitations decreased mark-edly in strength.

Figure 5 is a plot of the percentages of eachtype of imitative response by Subject 1. Itshows that her probability of imitating the 35probes varied between 80 and 100%, as longas the other 95 imitations, within which theprobes were interspersed, were reinforced. Theapplication of the DRO 30-sec procedure ex-

tinguished virtually all imitative behaviorwithin about 20 hr. The previously reinforcedimitations and the probe imitations extin-guished alike in rate and degree. All imitativebehavior recovered when, with a small amountof shaping, reinforcement was again made con-tingent upon imitative behavior.

Figure 6 is a similar plot of the imitative be-havior of Subject 3. It shows the maintenanceof the one probe imitation and eight rein-forced imitations during reinforcement of imi-

tation, a marked decrease in both types ofimitative behavior during the DRO 20-secperiod, and a recovery when contingent rein-forcement of imitations was resumed.

Figure 7 is a plot of the imitative behaviorof Subject 2. Her results were similar to thoseobtained for Subjects 1 and 3, in terms of themaintenance of 125 reinforced and five probeimitations, under conditions of reinforcementof imitations. However, her data depart fromthe others' during the DRO period. Initially,this subject showed no reliable signs of ex-tinction after four sessions of DRO with a 30-sec delay. Next, DRO 60-sec was instituted forfour sessions, still without any reliable effect.At that point, a procedure of DRO 0-sec was

begun, meaning that the experimenter demon-strated some behavior, and instantly, beforethe subject could respond, said "Good" anddelivered the food to her mouth. Thus, rein-forcement served to forestall the durable imi-tative responses this subject was displaying.Figure 7 demonstrates the immediacy of effectof this procedure. After four sessions of DRO

100 - SUBJECT #3

REINFORCED IMIT.*0O000 UNREINFORCED IMIT.

r I I I I I I I I

' 15 20

0

0

0~~~~~~~~~~~~~~~~~~~~~~~~~

00

0' 00

I ItI If I

25 30

-- DRO

SESSIONSFig. 6. The maintenance and extinction of reinfoyced and unreinforced imitation in Subject 3.

cn)z0

C/)z0

w

z

w

CL

w

a-

a

LLJ

0

35FX--|* E E * * * E * * * * . . - *E*-m-* sF,,

I I . I I IFA

413

F,


100

SO

0

SUBJECT #2

- NREINFORCED IMIT.*----0 UNREINFORCED IMIT.

105 110 115

30"

DRO

SESSIONSFig. 7. The maintenance and extinction of reinforced and unreinforced imitation in Subject 2.

0-sec, it was possible to resume the proceduresof DRO 30-sec and produce only a brief andpartial recovery of the rate of imitation, whichthen declined to zero. A return to contingentreinforcement, with a small amount of shap-ing, quickly reinstated the high rate of imita-tion previously displayed.

In all cases, then, it is clear that the imita-tive repertoire depended on reinforcement ofat least some of its members. It is noteworthythat those responses which had developed andbeen maintained previously without direct re-

inforcement could not survive extinction ap-

plied to the entire class of behaviors.

Imitative ChainsSubjects 1 and 2 were exposed to the pro-

cedure of chaining together old and new imi-tative responses. At the end of 10 hr of theprocedure for Subject 1, lengthy chains con-

taining already established and new imitativeresponses became practical. It was possible toobtain perfect imitation on 90% of the chains,

some of which involved as many as five re-

sponses. Subject 2 received only 2 hr of train-ing on chains. At the end of this time, shewould imitate 50% of the three-responsechains demonstrated to her, and 80% of thetwo-response chains.

Verbal BehaviorSubjects 1 and 3 were used in the procedures

for the development of verbal imitation.Verbal imitations were established for Sub-ject 1 by chaining together motor and vocalbehaviors and then fading out the motor com-

ponents. Twenty hours of training resulted in

10 words which were reliably imitated such as,

"Hi", "Okay", the subject's name, and thenames of some objects. Subject 3's training invocal imitations was accomplished by evokinga set of motor imitations which moved suc-

cessively closer to vocalizations. Approximately10 hr of training produced the reliable imita-tive vocalizations of seven vowel and conso-

nant sounds.

COz0-

CY

zw0-

IJ

zw

w

ri:I

000

S00S

0000S

0000000000S00

125

414


Generalization to Other ExperimentersWhen Subject 1 was presented with new

experimenters, of both the opposite and samesex as the original male experimenter, sheshowed approximately the same degree ofimitation displayed to the original experi-menter. That is, she imitated all of the threeprobe demonstrations given by one new maleexperimenter and imitated 12 of 15 reinforceddemonstrations by a second new male experi-mented on the first demonstration and theremaining three by the third demonstration.On another occasion, the second new male ex-

perimenter re-presented the 15 demonstra-tions; all were imitated on their first demon-stration. The subject also imitated all of a

series of demonstrations by a female experi-menter.

DISCUSSIONThe procedures of this study were sufficient

to produce highly developed imitation in theexperimental subjects. However, a noteworthypoint is the relative difficulty experienced inobtaining initial matching responses from a

subject even when the response required (e.g.,arm raising) clearly was in the subject's cur-

rent repertoire. This suggests that the subjectswere not so much learning specific responsesas learning the instruction, "Do as the experi-menter does." Initially, then, the proceduresof this study seem to have involved bringinga number of the subjects' responses under theinstructional control of the experimenter'sdemonstration.4 To establish this type of in-structional control by demonstration requiresthat the subjects either have or develop re-sponses of observing their own behavior as

well as the experimenter's behavior.As an increasing number of the subjects' be-

haviors came under the instructional controlof demonstration, additional behavior, not pre-viously observed in the subjects' repertoires,became increasingly probable, merely as a re-

sult of presenting an appropriate demonstra-tion by a model. In the terminology suggestedby Miller and Dollard (1941), a sufficientlyextensive arrangement of one child's behaviorinto matched-dependent response with a

'The authors are indebted to Israel Goldiamondfor his suggestions in clarifying this point.

model's behavior was sufficient to induce atendency to achieve similarity in more waysthan were originally taught.The development of imitative repertoires,

including the unreinforced imitation of probedemonstrations, could be accounted for by theeffects of conditioned reinforcement. Condi-tioned reinforcement may have operated inthe present study in the following way: thebasic procedure was that of teaching the sub-ject a series of responses, each of which wastopographically similar to a demonstrationjust given by a model. Initially, each responsehad to be established separately. When estab-lished, such responses were imitative onlytopographically and would better be calledmatched-dependent behavior; the fact that asubject's response was similar to the experi-menter's behavior at that point had no func-tional significance for any of the subject'sother responses. Nevertheless, topographicalsimilarity between child and experimenter wasthere to be attended to by the child, and thissimilarity was potentially discriminative withrespect to the only reinforcement delivered inthe experimental situation. One of the mosteffective ways of giving a stimulus a reinforc-ing function is to make it discriminative withrespect to reinforcement. In these applications,the stimulus class of behavioral similarity was,in numerous examples, made discriminativewith respect to positive reinforcement. Hence,similarity could be expected to take on a posi-tive reinforcing function as well as a discrimi-native function. As a positive reinforcer, itshould strengthen any new behavior that pro-duced or achieved it. Behaviors that achievesimilarity between one's self and a model are,of course, imitative behaviors; furthermore,they are imitative by function and not by coin-cidence.This analysis is simple only at first inspec-

tion. In particular, it should be noted that"similarity" is not a simple stimulus dimen-sion, like the frequency of sound or the in-tensity of light. Similarity must mean a corre-spondence of some sort between the stimulusoutput of the child's behavior and the stimu-lus output of the model's. A correspondencebetween two stimuli is not too esoteric a stim-ulus to consider as functional in controllingbehavior. However, for an imitative repertoireto develop, a class of correspondences must be-come- functional as stimuli, The child must

4}5

416 DONALD M. BAER, et al.

learn to discriminate a correspondence be-tween the appearance of his hand and themodel's hand, his arm and the model's arm,his leg and the model's leg, his voice and themodel's voice, etc. It would seem reasonablethat each of these kinds of difference must re-quire some prior experience on the child's partto appreciate. A scantiness of such experiencemay well be characteristic of retarded chil-dren, and makes them intriguing subjects forsuch studies. The ability to generalize simi-larities among a considerable variety of stim-uli, which the children of these studies evi-denced, suggests that the training they weresubjected to was adequate to the problem. Animmediate next problem, it would seem, isthe detailed analysis *of those procedures tofind out which of them accomplished whatpart of this generalization. That analysismight yield a fair understanding of imitativebehavior.

REFERENCESBaer, D. M. and Sherman, J. A. Reinforcement con-

trol of generalized imitation in young children.J. exp. Child Psychol., 1964, 1, 37-49.

Bandura, A. Social learning through imitation. InM. R. Jones (Ed.) Nebraska symposium on motiva-tion. Lincoln: University of Nebraska Press, 1962,Pp. 211-269.

Konorski, J. and Miller, S. On two types of con-ditioned reflex. J. gen. Psychol., 1937, 16, 264-272.

Lovaas, 0. I., -Berberich, J. P., Perloff, B. F., andSchaeffer, B. Acquisition of imitative speech byschizophrenic children. Science, 1966, 151, 705-707.

Metz, J. R. Conditioning generalized imitation inautistic children. J. exp. Child Psychol., 1965, 2,389-399.

Miller, N. E. and Dollard, J. Social learning and imi-tation. New Haven: Yale University Press, 1941.

Reynolds, G. S. Behavioral contrast. J. exp. Anal.Behav., 1961, 4, 57-71.

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

Terrace, H. S. Discrimination learning with andwithout "errors". J. exp. Anal. Behav., 1963, 6,1-27. (a)

Terrace, H. S. Errorless transfer of a discriminationacross two continua. J. exp. Anal. Behav., 1963,6, 223-232. (b)

Received 28 October 1965

Documents

Imitation-baer-et-al-1967 roca