JOURNAL OF APPLIED BEHAVIOR ANALYSIS

MODIFICATION OF SOCIAL WITHDRAWALTHROUGH SYMBOLIC MODELING'

ROBERT D. O'CONNOR

STANFORD UNIVERSITY2

The present experiment was designed to test the efficacy of symbolic modeling as a treatmentto enhance social behavior in preschool isolates. Nursery school children who displayed markedsocial withdrawal were assigned to one of two conditions. One group observed a film depictingincreasingly more active social interactions between children with positive consequences ensu-ing in each scene, while a narrative soundtrack emphasized the appropriate behavior of themodels. A control group observed a film that contained no social interaction. Control childrendisplayed no change in withdrawal behavior, whereas those who had the benefit of symbolicmodeling increased their level of social interaction to that of non-isolate nursery schoolchildren.

Recent years have witnessed increasing ap-plications of principles of learning to psycho-pathology. Ample evidence has accumulatedto indicate that behavioral approaches holdconsiderable promise for the treatment of di-verse psychological conditions (Bandura, 1969;Eysenck, 1964; Krasner and Ullmann, 1967;Wolpe and Lazarus, 1967). Many of these ap-plications, however, have been concerned withthe treatment of highly circumscribed dis-orders. Only recently have researchers begunto investigate the modifications of interper-sonal modes of behavior.

Social interaction, an obviously importantfactor in personality development, has becomethe focus of much attention among social-learning theorists, developmentalists, andtherapists. There are several reasons for high-lighting the role of interpersonal behavior inpersonality development. First, a child who

"This research was supported by Public Health Re-search Grant M-5162 from the National Institute ofMental Health to Albert Bandura, and by the LouisHaas Research Fund, Stanford University. The authoris grateful to Professor Bandura, whose enthusiasticassistance and many suggestions were invaluable duringall phases of this project, and to Professor Eleanor E.Maccoby, who provided support and helpful commentsduring the initial stages of the experiment. ProfessorsRobert R. Sears and Edith Dowley generously assistedin the design and implementation of observationalprocedures, and Marian O'Connor collaborated onnumerous resources which insured the success of theprogram.

'Reprints may be obtained from the author, Dept.of Psychology, University of Illinois, Urbana, Illinois61801.

is grossly deficient in social skills will be seri-ously handicapped in acquiring many of thecomplex behavioral repertoires necessary foreffective social functioning. Second, childrenwho are unable to relate skillfully to othersare likely to experience rejection, harrassment,and generally hostile treatment from peers.Such negative experiences would be expectedto reinforce interpersonal avoidance re-sponses which, in turn, further impede thedevelopment of competencies that are sociallymediated. Current theories concerning thedeterminants of personality patterns (Ban-dura, 1969; Bandura and Walters, 1963; Mi-schel, 1968; Peterson, 1968) emphasize socialvariables and underscore the general impor-tance of social interaction.

Several attempts have been made to enhancethe social behavior of isolate children (Allen,Hart, Buell, Harris, and Wolf, 1964; Hart,Reynolds, Baer, Brawley, and Harris, 1968;Hartup, 1964; Patterson and Anderson, 1964)through differential reinforcement. Thesestudies have shown that if peer interaction isreinforced, either socially or otherwise, andisolate play is either punished or ignored,children eventually display a higher level ofsocial behavior. The utilization of a treat-ment program based solely on reinforcementprocedures may encounter difficulties in thedevelopment of social responsiveness in ex-treme isolates. However, while a series ofpreliminary observations which served as apilot for the present study found 20%, or moreof nursery school children exhibiting rela-

15

1969, 2, 15-22 NUMBER I (SPRING 1969)

ROBERT D. O'CONNOR

tively low levels of social responsiveness, manyof whom could be helped by arranging favor-able response consequences; a smaller per-centage of children either perform no socialinteraction response or provide only rare op-portunities for the application of reinforce-ment. When such gross deficits exist, the re-inforcing agent must either introduce a ratherlaborious set of "shaping" procedures, whichrequires waiting for the emission of a rein-forceable social response, or resort to moreactive means for establishing the desired be-havior.

Lovaas (1966) showed that relatively com-plex repertoires can be established to replacegross deficits through a combination of model-ing and reinforcement procedures. Of greaterinterest and relevance to the approach usedin the present study is evidence that childrencan acquire new patterns of behavior on thebasis of observation alone (Bandura and Hus-ton, 1961; Bandura and McDonald, 1963;Bandura and Mischel, 1965; Bandura, Ross,and Ross, 1963; Hicks, 1965). Since repertoirescan thus be learned on a non-response basis,with no reinforcement to the observer, amodeling program may be particularly effec-tive in the case of gross behavior deficits.

It was noted earlier that in most cases,severe withdrawal reflects both deficits in so-cial skills and avoidance of interpersonal sit-uations. An optimal treatment, therefore,should transmit new social competencies andalso extinguish social fears. Modeling proce-dures are also ideally suited for this purpose.A series of studies by Bandura (1968a) dem-onstrated that various patterns of avoidancebehavior can be successfully eliminatedthrough modeling, and that such proceduresare readily applicable to therapeutic situa-tions.By devising a carefully constructed film

sequence, the therapist can stage rather com-plicated situations and events in a dramaticmanner that controls the viewer's attention torelevant cues. While the exclusion of extrane-ous events provides much of this attentioncontrol, the enthusiastic and emotionally ex-pressive behavior of models can further en-hance attention and vicarious learning in theviewer (Bandura, 1962; Berger, 1962; Bergerand Johansson, 1968). Once the observer's at-tention has thus been directed toward thefilmed events, the therapist may introduce

repeated exposure to clearly defined stimuli.Positive response consequences to the

model, such as social praise or material rein-forcements for modeled behaviors, which havebeen shown to increase the performance ofsimilar behavior in observers (Bandura, 1965;Bandura, 1968b), can also be incorporatedinto filmed events. Symbolic modeling proc-esses employing such principles have effec-tively extinguished severe avoidance behaviorin children (Bandura and Menlove, 1968).The present experiment sought to extend

the use of symbolic modeling to the modifica-tion of social withdrawal. This approach ap-pears particularly well suited for achievingboth of the desired outcomes indicated in thepathology described above; i.e., the transmis-sion of social skills and the extinction of socialfears. These two modeling processes, alongwith the facilitation of interpersonal behav-iors which may exist in the observers' reper-toire, provided the rationale for the manip-ulation of social interaction behavior, whichin fact comprised the focus of the experi-mental change assessment.

Children who displayed extreme socialwithdrawal were shown a sound-film depictingpeers engaged in progressively more activesocial interaction. The viewers were childrenwho not only had very low base-rates of socialinteraction, according to the dual assessmentprocedure (below), but whose frequent re-treats into corners, closets, and lockers gaveobservers and teachers a similar impression ofactive, purposive withdrawal in many in-stances. The filmed behavior of peers pre-sented to these "isolates" was actively fol-lowed by reinforcing outcomes such as peerapproval, either verbal or expressional (smil-ing, nodding, etc.); peer acceptance of themodel into a game or a conversation, i.e., in-vitations to join or offering play materials orreaching out to take the model's hand, etc.;in most scenes the model behavior resulted insome tangible reinforcement such as a blockor other toy, a book to read together withthe peers, a dish to wash or dry in a coopera-tive homemaking activity; and so on. A secondgroup of equally withdrawn children vieweda film that contained no human characters.After exposure to their respective films, thechildren's social behavior was observed in thenursery school situation. It was predicted thatchildren having the benefits of symbolic mod-

16

SYMBOLIC MODELING AND SOCIAL WITHDRAWAL

eling would display a significant increase insocial interaction with their peers.

METHOD

Selection of IsolatesHead teachers in each of nine nursery

school classes were asked to choose from theirenrolment lists the five most socially with-drawn children in their class. Each teacherwas to rank order these five children "whointeract the least with their peers". Of the 365children enrolled, the 45 nominated in thispreliminary selection and 26 children ran-domly chosen from the remaining non-isolateswere then observed in the nursery school set-ting at randomly selected times throughoutthe day.Each child was observed for a total of 32

intervals, each interval consisting of 15 sec,over a period of eight days (with the time ofday counterbalanced on consecutive days).During each 15-sec interval, the children's be-havior was scored in terms of five separateresponse categories; one every 3 sec. These in-cluded physical proximity, verbal interaction,"looking at", "interacting with", and the sizeof the group involved in any interaction se-quence. Although the first three response cat-egories, considered social orienting responses,were scored, the measure of social behaviorwas based entirely on the frequency withwhich the children interacted with their peers(i.e., category four). The three categories of"orienting" responses and the "size of group"category were included in the observations forintra-sample reliability checks and for pos-sible assessment purposes in case the primary"interacts" category had not reflected suchnotable change in the children's behavior.Obviously, a score in the "interaction" cate-gory necessitated "proximity and looking at"responses, and increases in the less critical"verbalizes" and "number in group" categor-ies had to accompany most interactions per-formed by these former isolates. The majoremphasis of the experimental manipulation,therefore, was on the interaction scores, andthe changes assessed were in this categoryalone, although changes in the other behav-ioral categories were at least as significant asthose reported in the response class of interesthere. A social interaction was defined as anybehavior directed toward another child which

involved a reciprocal quality. Neither parallelplay nor solitary verbalizations qualified. Thetwo-way nature of a scorable interaction neces-sitated not only the output of the subjectchild, but some indication of recognition andattention from the second child in the inter-action. Thus, if a subject spoke to or other-wise directed his behavior to another child,but the second child did not respond in anyway, at least appearing to be aware of theintended interaction, no score was given foran interaction.

Six trained observers performed the ratings.During a randomly chosen 50% of thesessions, these observers were paired andobserved a given child independently butsimultaneously. Inter-scorer reliability on eachof these sessions was r = > 0.92. (# of agree-ments/the 32 possible agreements on a givenchild's interactions), which allowed formatched observer correlations across subjectsof r = 0.90 + (product-moment) in all pairs ofobservers.

Children who scored fewer than five of 32possible interactions and who had been ratedby teachers as isolates were selected as subjectsand randomly assigned to either the modelingor the control film conditions. Thus, to qualifyfor the experiment, children had to meet thedual criteria of having exhibited extremewithdrawal over a long period of time asjudged by their teachers, and to have dis-played isolate behavior as measured by ob-jective behavioral observations. Of the 20 "iso-lates" who met these criteria, 13 were includedin the experiment; four of the remaining sevenwere omitted because they were frequentlyabsent from school; and three of the childrenvigorously refused to leave the nursery room.The 26 non-isolate children were primarily

included to furnish an additional baseline forevaluating any changes produced by the treat-ment program. These children displayed amean of 9.1 social interactions, while themeans for children assigned to the modelingand control conditions were 1.75 and 1.50 re-spectively, with the scores in both groupsranging from 0 to 5. The modeling groupcontained four girls and two boys and the con-trol group, four girls and three boys.

Treatment ConditionsChildren in both conditions were brought

individually to the experimental room where

17

ROBERT D. O'CONNOR

they were told they could watch a televisionprogram. Each child was seated before a largeTV console while the experimenter pluggedin the set and ostensibly tuned in the picture.The films were shown on a glass lenscreen bymeans of a rear projector arrangement. Asthe apparatus was plugged in, the hidden pro-jector and tape recorder were activated simul-taneously. An extension speaker directed thesound through the TV set.The experimenter left the child alone in

the room to view the film on the pretext thathe had some work to complete and would re-turn before the film had ended. The experi-menter then observed the session through aone-way mirror from an adjoining room. Allchildren appeared to be highly attentive tothe "TV show" throughout the film. The at-tention apparently commanded by the televi-sion presentation was quite impressive andobviously advantageous to the experimentalprocedure.

Children in the modeling condition saw asound-color film lasting approximately 23min. The film portrayed a sequence of 11scenes in which children interacted in a nur-sery school setting. In each of these episodes,a child is shown first observing the interactionof others and then joining in the social activi-ties, with reinforcing consequences ensuing.The other children, for example, offer himplay material, talk to him, smile and generallyrespond in a positive manner to his advancesinto the activity. The scenes were graduatedon a dimension of threat in terms of the vigorof the social activity and the size of the group.The initial scenes involve very calm activitiessuch as sharing a book or toy while two chil-dren are seated at a table. In the terminalscenes, as many as six children are showngleefully tossing play equipment around theroom.

Multiple modeling has been shown to bemore efficacious than single modeling (Ban-dura and Menlove, 1968). Also, a second pilotstudy conducted as a preliminary to the pres-ent experiment suggested a powerful effect onsocial behavior of subjects as a result of mul-tiple, live peer-modeling; the child displayingthe social approach behavior was varied fromscene to scene in terms of age and sex, in-cluding a total of six different models, fourgirls and two boys, with their ages rangingfrom 4 to 7 yr.

To accent further the modeling cues andthe positive consequences associated with thesocial behavior of the approaching child, anarrative sound track was prepared in whicha woman's voice, judged by the experimenterto be very soothing, described the actions ofthe model and the other children in the on-going sequence. The script consisted entirelyof descriptions of ongoing social responsesand outcomes and was a further attempt tofocus the viewers' attention to relevant cues.The control film depicted 20 min of the

acrobatic performances of Marineland dol-phins and was accompanied by a musicalsoundtrack. Since the film contained no hu-man characters, it provided a control for anyeffects which might have been derived solelyfrom the presentation of a film in the ex-perimental procedure and contact with experi-menters. The control group further provideda basis on which to measure any change in thesocial behavior of isolate subjects which mighthave occurred as a result of nursery schoolparticipation during the course of the project.3

Post-Treatment AssessmentImmediately after being shown their respec-

tive films, the children were returned to theirregular classrooms. They were given 2 min toadapt to the classroom situation, after whichthey were observed for 32 consecutive inter-vals, each lasting 15 sec, according to the sameobservation procedure employed in the pre-treatment assessment. The social interactionscore was again the number of 15-sec intervalsin which the child displayed a direct socialinterchange (defined according to the "recip-rocal quality" criteria mentioned earlier) withone or more children. In order to control forany bias in ratings, observers were kept un-aware of condition assignments, and each ob-served a random combination of treated andcontrol subjects. Aside from the usual "blindassessment" control, this randomizing tech-nique was thought to reduce further any pos-sible observer bias.

RESULTSFigure 1 represents the mean number of

social interaction responses performed by chil-

3A copy of the 400 ft super-8-mm film and sound-track can be obtained from the author for approxi-mately $66.00 (cost) plus $6.00 for packing and mailing.

18

SYMBOLIC MODELING AND SOCIAL WITHDRAWAL

12 r

z0

0

Ir-

w

z

-~SYMBOLIC MODELING--- CONTROL CONDITION

.NONISOLATE BASELINE

K0

81-

6F

4

2 _

I

PRE-TEST POST-TEST

Fig. 1. Mean number of social interactions displayedby subjects in the modeling and control conditions, be-fore and after the experimental sessions. The dottedline represents the level of interactions manifested by26 non-isolate children who were observed at thepretest phase of the study.

dren in the modeling and control conditionsduring the pretest and immediately after theexperimental session.

An analysis of within-group changes showedthat the seven control children remained es-sentially unchanged, whereas the six childrenwho had viewed the modeling film markedlyincreased their level of social interaction(t = 2.29; p < 0.03). The post-test interactionscores of treated subjects were in fact similarto those of non-isolates who had been ob-served during the pretest period.A between-groups analysis revealed the

change in social interaction to be significantlygreater for subjects in the modeling treatmentthan for controls (t = 2.53; p = 0.015). Abetween-groups comparison of the levels ofsocial interaction achieved at post-test indi-cated significant differences as evaluated byeither the t-test (t = 2.70; p = 0.01) or theMann Whitney U-test (U = 3.0; p < 0.004).Having measured the powerful effects of the

modeling film in terms of the group compari-sons above, data for the individual subjects inthe modeling condition were found to indicateconsistent positive changes across subjects. Allsix children in the modeling condition exhib-ited increased social interaction behavior in

the post-film assessment. Children in the con-trol condition performed essentially the samenumber of interactions before and after view-ing the control film, with only slight increasesor decreases in the behavior.

DISCUSSIONThe present results established symbolic

modeling as a highly efficacious procedure formodifying social withdrawal. Children whodid not receive the modeling treatment re-mained socially withdrawn, whereas those whoobserved a systematic filmed presentation ofpeer interactions associated with reinforcingconsequences displayed a significant increasein social responsiveness.Follow-up observations could not be made

because the nursery school term was com-pleted. However, a second set of teachers' rat-ings was obtained at the end of the schoolyear. The teachers, who were kept uninformedas to which conditions children had been as-signed, again rated the five most withdrawnchildren from their enrollment lists as theyhad done in the preliminary selection. Onlyone of the six subjects who had been in themodeling condition was still rated as an iso-late. It is interesting to note that this child,who also improved the least as measured bybehavior observation, viewed the modelingfilm with the sound track 20-sec behind thepicture due to a mechanical failure. Four ofthe seven control subjects were again judgedto be extreme isolates. Although these findingshave some suggestive value, they should beaccepted with reservations because of theglobal nature of the ratings and the fact thatchanges in classroom enrollment may affecttheir comparability.Immediate treatment effects achieved by

symbolic modeling may produce lastingchanges in social interaction without the needof additional procedures, provided that thechildren's initial social behavior is favorablyreceived by peers. However, the applicationof systematic reinforcement of appropriate so-cial responses would ensure the maintenanceof the induced behavioral changes. Bandurahas suggested "the combined use of modelingand reinforcement procedures" as the mostefficacious mode of therapy in "eliminatingsevere behavioral inhibitions" (Bandura, Gru-sec, and Menlove, 1967). In order to substan-

19

ROBERT D. O'CONNOR

lJ BASE-RATEPOST-TEST

Cl)z25-0

20O-j

0

IL0

:OD

z 5

1 2 3 4 5 6MODELING FILM

Fig. 2. Actual number of social interactions displayedposure to the modeling or control film.

tiate this hypothesis a study needs to beconducted which extends the design of thepresent experiment to include durational(follow-up) assessment periods, as well as acomparison of modeling, shaping, and com-bined treatment procedures. In the presentstudy, social inhibitions were reduced andappropriate social responses were facilitatedthrough symbolic modeling with built-in re-inforcement to the model. New responses tofamiliar social stimuli which were formerlyassumed to have elicited avoidance behaviorwere acquired vicariously within one treat-ment session.The subsequent performance of these newly

acquired behaviors is seen as primarily result-ing from the facilitation and extinction effectswhich derive from the observation of modelsperforming the target behaviors with no aver-sive consequences ensuing. This theoreticalexplanation of the modeling effects achieved

1 2 3 4 5 6 7CONTROL FILM

by each child in either condition, before and after ex-

is based on the same set of experimental datawhich guided the construction of the treat-ment film (an important point in the compar-ison of differential treatment outcomes). Adistinction is drawn between response acqui-sition and response performance according toexperimental data derived from a recent ex-periment (Bandura, 1965b) which suggestedthat novel behaviors may be acquired by ob-servers, even though the models are punished,but that these responses may not be performedreadily without the addition of strong incen-tives beyond those of the situational stimuliin the modeling presentation. The conclusionfrom this study is theoretically relevant to thepresent discussion, in that it suggests that amodel's reinforcement contingencies may benegative enough to inhibit performance ofnewly acquired responses, but have little or noeffect on their acquisition. Other data, suchas the Berger (1962, 1968) studies mentioned

20

SYMBOLIC MODELING AND SOCIAL WITHDRAWAL

earlier and a recent demonstration of arousalreduction in snake phobics through live mod-eling (Blanchard, 1968), provide support forthe additional modeling effects of disinhibi-tion and the facilitation of responses exist-ing in the observer's repertoire, which com-plete the three-fold theory of modeling effectsincorporated into the design of the presentfilm and the theoretical explanation of itseffects.An alternative explanation for the increased

social behavior of children who observed themodeling film might be based strictly on theprinciples of reinforcement theory. Since thedata here indicate only changes in the rate ofsocial interaction responses, it might be arguedthat the simplest behavioral description wouldidentify the filmed presentation as a discrim-inative stimulus for appropriate matching re-sponses. (Baer and Sherman, 1964; Baer, Peter-son, and Sherman, 1967; Peterson, 1968). Thisapproach would be based upon what is oftenconsidered the most parsimonious behavioralanalysis of observable events, i.e., the perform-ance of measurable behaviors and a descrip-tion of the stimulus situation in which thematching behaviors, as a "functional responseclass", occurred. The rationale for such a de-scription might be based upon the observer'sprior history of reinforcement for matchingmodels' behavior, or similar conceptions of"imitation", per se, with emphasis on thechange in reinforcement value of models' be-havior, the controlling power of modeledbehavior as a discriminative stimulus, etc.(Staats, 1963, 1968). Aside from recent sug-gestions concerning the presumptive nature ofthese "heavyweight" reinforcement-theory ex-planations (Glucksberg, 1968); the parsimonyderiving from reinforcement approaches, interms of therapeutic efficiency, has not yetbeen demonstrated in instances where gross be-havior deficits are identified as the change-worthy phenomena. The expense of "shapingtime" must be compared to a treatment whichmay provide for the acquisition of possiblynovel skills according to principles of associa-tive learning (contiguous presentation of mod-eling stimuli), as well as facilitating theperformance of modeled responses and otherappropriate behaviors in the observer's reper-toire (discriminative stimulus function of non-aversive modeled behavioral outcomes), whilereducing negative arousal responses to feared

stimuli, all in one treatment session. The ex-planatory value of reinforcement principlesmay thus be relevant to performance variables,while an approach that is intended to effectinput (learning) variables as well may providetreatment procedures with markedly greaterapplicability to behavior deficit conditions.The allowance for possible learning deficits intreatment strategies designed to modify be-havior deficits, i.e., attention to input as wellas output deficits, may be much more than atheoretical distinction. The powerful effectsof the modeling presentation reported hereunderscore what appears to be a very practical,therapeutically useful reason to allow for thenotion of mediational processes as well as re-inforcement principles when these factors maybe relevant to the therapeutic strategy.This brief discussion of two possible expla-

nations of the modeling effects achieved in thepresent study may serve to direct the reader'sattention to further analyses of the modelingprocess in general. Thoroughly detailed pre-sentations may be found in more appropriatepublications (Bandura, 1968c; Bandura, 1969;Mischel, 1968; O'Connor, 1969; Staats, 1968;Ullmann, 1968).

It should be noted in passing that the pres-ent experiment achieved significant changesin social behavior among children with rela-tively severe deficits without developing atherapeutic relationship. Until recently, afairly intimate client-therapist relationshipand the attainment of insight have been con-sidered necessary conditions for personalitychange. In contrast, the results and discussionabove indicate that the social behavior ofchildren can be effectively enhanced by effortsto arrange social stimulus conditions whichmay ensure the acquisition of requisite compe-tencies, the reduction of inhibiting fears, andthe facilitation of appropriate responses. Itshould also be apparent that attention to learn-ing variables provides for treatment proce-dures which can optimistically be applied toany program of behavior change. Teachers andother social agents might greatly increase theefficacy of their work with individual children,as well as in group procedures, by employingsome of the principles of social learning andsymbolic modeling presented here. The use ofcarefully designed therapeutic films in class-room and experimental situations alike mayprovide significantly more efficient modifica-

21

22 ROBERT D. O'CONNOR

tion of various behavior deficits and otherdeviant behaviors.

REFERENCES

Allen, K. E., Hart, B. M., Buell, J. S., Harris, F. R., andWolf, M. M. Effects of social reinforcement on iso-late behavior of a nursery school child. Child Devel-opment, 1964, 35, 511-518.

Bandura, A. Influence of models' reinforcement con-tingencies on the acquisition of imitative responses.Journal of Personality and Social Psychology, 1965,1, 589-595. (b)

Bandura, A. Modeling approaches to the modificationof phobic disorders. Ciba Foundation symposium:The role of learning in psychotherapy. London:Churchill, 1968. (a)

Bandura, A. Principles of behavior modification. NewYork: Holt, Rinehart & Winston, 1969, (in press).

Bandura, A. A social learning interpretation of psy-chological dysfunctions. In P. London and D. Rose-han (Eds.) Foundations of abnormal psychology.New York; Holt, Rinehart & Winston, 1968. Pp.293-344.

Bandura, A. Social learning theory of identificatoryprocesses. In D. A. Goslin (Ed.), Handbook of social-ization theory and research. Chicago: Rand McNally,1968. Pp. 213-262 (b)

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

Bandura, A. Vicarious processes: A case of no-triallearning. In L. Berkowitz (Ed.), Advances in experi-mental social psychology. Vol. II. New York: Aca-demic Press, 1965. Pp. 1-55.

Bandura, A., Grusec, J., and Menlove, F. L. Vicariousextinction of avoidance behavior. Journal of Per-sonality and Social Psychology, 1967, 5, 16-23.

Bandura, A. and Huston, Aletha C. Identification asa process of incidental learning. Journal of Abnor-mal and Social Psychology, 1961, 63, 311-318.

Bandura, A. and McDonald, F. J. Influence of socialreinforcement and the behavior of models in shap-ing children's moral judgments. Journal of Abnor-mal and Social Psychology, 1963, 67, 274-281.

Bandura, A. and Menlove, F. L. Factors determiningvicarious extinction of avoidance behavior throughsymbolic modeling. Journal of Personality and SocialPsychology, 1968, 8, 99-108.

Bandura, A. and Mischel, W. Modification of self-imposed delay of reward through exposure to liveand symbolic models. Journal of Personality andSocial Psychology, 1965, 2, 698-705.

Bandura, A., Ross, D., and Ross, S. A. Transmission ofaggression through imitation of aggressive models.Journal of Abnormal and Social Psychology, 1961,64, 575-582.

Bandura, A. and Walters, R. Social learning and per-sonality development. New York: Holt, Rinehart &Winston, 1963.

Baer, D. M. and Sherman, J. A. Reinforcement controlof generalized imitation in young children. Journalof Experimental Child Psychology, 1964, 1, 37-49.

Baer, D. M., Peterson, R. F., and Sherman, J. A. Thedevelopment of imitation by reinforcing behavioralsimilarity to a model. Journal of the ExperimentalAnalysis of Behavior, 1967, 10, 405-416.

Berger, S. M. Conditioning through vicarious instiga-tion. Psychological Review, 1962, 69, 450-466.

Berger, S. M. and Johansson, S. L. Effect of model'sexpressed emotions on an observer's resistance toextinction. Journal of Personality and Social Psy-chology, 1968, 10, 53-58.

Blanchard, E. B. Relative contributions of modeling,informational influences, and physical contact inthe extinction of phobic behavior. Unpublisheddoctoral dissertation. Stanford University, 1968.

Eysenck, H. J. Experiments in behavior therapy. NewYork: Macmillan, 1964.

Glucksberg, S. A self-made straw man. ContemporaryPsychology, 1968, 13, 624-625.

Hart, B. M., Reynolds, N. H., Baer, D. M., Brawley, E.R., and Harris, F. R. Effect of contingent and non-contingent social reinforcement on the cooperativeplay of a preschool child. Journal of Applied Be-havior Analysis, 1968, 1, 73-76.

Hartup, W. W. Peers as agents of social reinforce-ment. Young Children, 20, 1965.

Hicks, D. J. Imitation and retention of film-mediatedaggressive peer and adult models. Journal of Per-sonality and Social Psychology, 1965, 2, 97-100.

Krasner, L. and Ullmann, L. P. Research in behaviortherapy. New York: Macmillan, 1964.

Lovaas, I., Berberich, J. P., Perlof, B. F., and Schaeffer,B. Acquisition of imitative speech by schizophrenicchildren. Science, 1966, 151, 705-707.

Mischel, W. Personality and assessment. New York:Wiley & Sons, 1968.

O'Connor, R. D. Modeling treatment of non-behaviordisorders. Paper presented at 41st annual meetingof the Midwestern Psychological Association, Chi-cago, March 1969.

Patterson, G. R. and Anderson, D. Peers as social re-inforcers. Child Development, 1964, 35, 951-960.

Peterson, D. R. The clinical study of social behavior.New York: Appleton, Century, Crofts, 1968.

Peterson, R. F. Some experiments on the organizationof a class of imitative behaviors. Journal of AppliedBehavior Analysis, 1968, 3, 225-235.

Staats, A. W. Learning, language, and cognition. NewYork, Holt, Rinehart & Winston, 1968.

Staats, A. W. and Staats, C. K. Complex human be-havior. New York, Holt, Rinehart & Winston, 1963.

Ullmann, L. P. Making use of modeling in the thera-peutic interview. Paper read at A.A.B.T. meetings,San Francisco, 1968.

Wolpe, J. and Lazarus, A. A. Behavior therapy tech-niques. London: Pergamon Press, 1966.

Received 18 September 1968.