253

JOURNAL OF APPLIED BEHAVIOR ANALYSIS 2003, 36, 253–257 NUMBER 2 (SUMMER 2003)

USING VIDEO MODELING AND REINFORCEMENT TOTEACH PERSPECTIVE-TAKING SKILLS TO

CHILDREN WITH AUTISM

LINDA A. LEBLANC AND ANDREA M. COATES

WESTERN MICHIGAN UNIVERSITY

SABRINA DANESHVAR, MARJORIE H. CHARLOP-CHRISTY, AND CAROLINE MORRIS

CLAREMONT MCKENNA COLLEGE

AND

BLAKE M. LANCASTER

WESTERN MICHIGAN UNIVERSITY

We evaluated video modeling and reinforcement for teaching perspective-taking skills to3 children with autism using a multiple baseline design. Video modeling and reinforce-ment were effective; however, only 2 children were able to pass an untrained task, indi-cating limited generalization. The findings suggest that video modeling may be an effec-tive technology for teaching perspective taking if researchers can continue to developstrategies for enhancing the generalization of these new skills.

DESCRIPTORS: autism, perspective taking, video modeling, theory of mind

Video modeling involves showing a vid-eotape of a person providing an exact versionof a behavior for a child to imitate. Videomodeling has been combined with otherstrategies (e.g., reinforcement) to teach con-versational and self-care skills to childrenwith autism, and may result in faster acqui-sition and better generalization than in vivomodeling (e.g., Charlop-Christy, Le, & Free-man, 2001). Charlop-Christy et al. suggestthat video modeling may be effective be-cause television is an engaging medium (i.e.,results in longer sustained attention) thatdoes not require social interaction duringlearning. Video also permits accentuation ofcertain stimulus features and minimizationof distracting or irrelevant features to help

We thank Patrick Ghezzi and James Carr for theirhelpful comments and Ivy Chong and James Carr forassistance with preparing graphs.

Address correspondence to Linda A. LeBlanc, West-ern Michigan University, Department of Psychology,1903 W. Michigan Ave., Kalamazoo, Michigan49008-5439.

combat the problem of stimulus overselec-tivity (e.g., Spradlin & Brady, 1999). Theseadvantages may also be valuable for teachingcomplex social behavior such as perspectivetaking.

Perspective taking has been defined as un-derstanding that another person’s beliefsabout events may be different from realityand that those beliefs will guide future be-havior (Sigman & Capps, 1997). This phe-nomenon is illustrated by the classic Sally-Anne false-belief task (Baron-Cohen, Leslie,& Frith, 1985) in which a child predicts thesearch behavior of a puppet that has seen anobject hidden in one location but has notseen that it was later moved. To demonstrateperspective taking, the child answers that thepuppet will search in the original location,behaving based on its own directly observedinformation even though it is false. Perspec-tive taking appears to serve an integral func-tion in social behaviors such as sharing, turntaking, and empathy, which involve taking

254 LINDA A. LEBLANC et al.

another’s perspective. From a more behavior-analytic viewpoint, perspective taking in-volves (a) observing the behaviors of anotherindividual in a given situation and (b) pre-dicting the individual’s subsequent behavior(e.g., ‘‘He’s going to look where the ball washidden before he left’’) or responding in ac-cordance with the private thoughts or emo-tions another individual might typically ex-perience in that situation (e.g., making con-soling remarks, such as ‘‘better luck nexttime,’’ after the observed individual lost atennis match).

Children diagnosed with autism oftenhave difficulty with false-belief tasks and alsodo not develop related social behaviors suchas sharing, turn taking, empathy, and verbalinitiations (Shabani et al., 2002; Spradlin &Brady, 1999). The current study evaluatedthe effectiveness of video modeling and re-inforcement for teaching perspective takingto children with autism and examined gen-eralization among novel tasks.

METHOD

Participants

Three boys, aged 7 to 13 years, who hadbeen diagnosed with autism, participated.Timmy was 7 years 3 months of age with averbal mental age of 4 years 10 months. Heproduced simple complete sentences whenprompted, but rarely initiated speech. Bobbywas 7 years old with a verbal mental age of6 years 6 months. He held simple conver-sations about his preferred topic, trains. Da-vid was 13 years of age with a verbal mentalage of 15 years. He read and wrote in com-plete sentences and constructed brief para-graphs about a variety of topics. All childrenwere primarily served in special educationand were included in general educationclassrooms at least part of every day. Timmyreceived additional services at a specializedafterschool program.

Setting

Timmy’s sessions were conducted at hisafterschool program with testing in onesmall room and training in another. ForBobby and David, the study was conductedin their respective special education class-rooms. Testing occurred at a table in onecorner of the classroom, and video modelingsessions were held in another corner. Chil-dren were seen two to three times weekly.Sessions were videotaped for Timmy andBobby but not David per parental request.

Task Descriptions

A staff member administered three com-mon measures of perspective taking while heor she sat directly across from the participantat a small table. The experimenter promptedand praised attentiveness (e.g., eye contactand looking at stimuli) and in-seat behaviorthroughout each task. No feedback was everprovided during any testing session. Com-plete scripts of all tasks and all instructionalvideos are available from the first authorupon request.

Sally-Anne. The Sally-Anne task includedcommon animal puppets (e.g., Barney,Cookie Monster, Tigger, Franklin) presentedin dyads. Initially, both puppets were pres-ent, and the participant saw one of themplace an object under a bowl. One puppetleft, and the child saw the other puppetplace the object under a box. The observingchild was asked where the departed puppetwould look for his object. The correct an-swer was ‘‘under the bowl’’ or pointing atthe bowl. This task was administered onceas a pretest before any baseline or trainingoccurred and once as a posttest after testingand training occurred for the other tasks.

M&Mst. This task was based on the‘‘Smarties’’ task using a substitute candy(M&Ms; Perner, Frith, Leslie, & Leekam,1989). The participant was shown an over-sized box of candy and was asked what he

255VIDEO MODELING TO TEACH PERSPECTIVE TAKING

thought was inside. Then, the box wasopened to show that there was really a pencilinside rather than candy. The participant wasasked to predict what another person, whowas not present, would think the box con-tained. The correct answer was ‘‘M&Ms.’’

Hide and seek. Two experimenters and thechild were present and saw that a puppet leftfootprints when it walked (Chandler, Fritz,& Hala, 1989). Those footprints were erasedand the second experimenter left the roombefore the puppet walked, again leavingfootprints, to place treasure in a chestmarked ‘‘1.’’ The puppet then moved thetreasure to a nearby box marked ‘‘2’’ but leftno footprints. The second experimenter re-turned and the child was asked to predictwhere this person would look for the trea-sure. The correct answer was ‘‘1’’ or pointingto Chest 1.

Stimulus variations. Each variation in-volved a stimulus change in the task, andseveral required a response not observed inthe video. The variations were tested but nottrained and were used to evaluate generaliza-tion and to determine whether the child wasimitating a rote response from the video orusing the strategy in the video. At least twovariations for each task were used with eachchild.

Three variations were developed for thehide-and-seek task. Variant 1 (V1) had foot-prints leading to a red cup with the coinsmoved to a white cup. Variant 2 (V2) hadthe footprints leading to Chest 1 with thecoins moved to the experimenter’s pocket.Variant 3 (V3) had the footprints leading toChest 2 and the coins moved to Chest 1.V1 and V3 required a response not observedin the training video.

Five variations were developed for theM&M task. Variant 1 (V1) had pennies inthe candy box. Variant 2 (V2) had a crayonin the box. Variant 3 (V3) had a pencil in aCheetost canister. Variant 4 (V4) had a pen-cil in a cereal box, and Variant 5 (V5) had

a toy in a cereal box. V3 through V5 re-quired the child to give a response differentthan the trained response.

Data Collection and Interobserver Agreement

The participant’s response was scored aspass or fail. A second trained observer scoreda 100% of sessions for Timmy, 34% of ses-sions for Bobby, and 100% of sessions forDavid. Agreements were divided by agree-ments plus disagreements and multiplied by100%. Mean interobserver agreement was100% for all participants.

Research Design

We used a multiple baseline design acrosstwo tasks (M&M and hide and seek) foreach participant. In addition, the Sally-Annetask, which was never trained, served as pre-and posttests. Baseline with the two trainingtasks began immediately following the pre-test. The order of task training was counter-balanced across subjects. Each session inbaseline, intervention, and follow-up lastedapproximately 4 to 10 min.

Pretest, posttest, and baseline. The experi-menter provided no feedback on perfor-mance. After completion of session, thechild was praised for effort and given a self-nominated preferred item (e.g., sticker, smallcandy) for participating regardless of his per-formance.

Video modeling and reinforcement. Testingsessions continued as in baseline. Prior totesting, the child viewed a video of an adultcorrectly completing the task. The video fo-cused on relevant visual cues (e.g., zoom-inon the footprints), and the model explainedthe strategy (e.g., ‘‘he looks in 1 because thefootprints lead to 1—it’s a clue’’). The ex-perimenter paused the video to have thechild respond to perspective-taking ques-tions immediately after the correct responsewas modeled. Correct answers resulted inpraise and preferred edible items or stickers.Incorrect responses resulted in a replay of the

256 LINDA A. LEBLANC et al.

Figure 1. Performance during testing sessions for Timmy, Bobby, and David. Original tasks are depictedas closed circles, and additional task variations are depicted as open circles, squares, triangles, and closed squares.

257VIDEO MODELING TO TEACH PERSPECTIVE TAKING

video and a prompt to pay attention untilcorrect imitation occurred. The training ses-sion was complete at three consecutive cor-rect responses to the video. The range ofviewings per modeling session was three toeight. The training phase continued until achild responded correctly in the testing ses-sion on three consecutive trials. Then, all ex-emplars for that task were tested. It shouldbe noted that a child could consistently passin the training session and then fail in test-ing. Follow-up was conducted 1 month afterthe final training session. No video was usedprior to maintenance sessions. If a childfailed a question at follow-up, a booster vid-eo modeling session was conducted and thechild was retested.

RESULTS AND DISCUSSION

The results are depicted in Figure 1. Vid-eo modeling with reinforcement was an ef-fective teaching procedure for these perspec-tive-taking tasks. All children consistentlyfailed the primary tasks and variants in base-line. All children eventually mastered thetasks and passed variants even when novelvocal or motor responses were required. Fol-low-up evaluations were successful for Bob-by and David (one booster session required).All participants failed the Sally-Anne pretest;however, 2 of the 3 participants (Timmy andBobby) passed the task after intervention forthe other two tasks, suggesting generaliza-tion (data not shown in the figure). Inter-estingly, Knoll and Charman (2000) dem-onstrated the same pattern of respondingand skill transfer with typically developing3-year-olds, indicating that this stimuluscontrol problem is not unique to childrenwith autism.

One limitation of the current investiga-tion is that we taught perspective taking butdid not directly address a behavioral expla-nation for the target response. Additional re-search should focus on behavioral explana-tions such as stimulus control and the roleof equivalence relations (Spradlin & Brady,1999). Finally, future studies might evaluatestrategies for increasing perspective-takingskills in natural social situations.

REFERENCESBaron-Cohen, S., Leslie, A. M., & Frith, U. (1985).

Does the autistic child have theory of mind? Cog-nition, 21, 37–46.

Chandler, M., Fritz, A. S., & Hala, S. (1989). Small-scale deceit: Deception as a marker of two-, three-,and four-year olds’ early theories of mind. ChildDevelopment, 60, 263–277.

Charlop-Christy, M. H., Le, L., & Freeman, K.(2001). A comparison of video modeling with in-vivo modeling for teaching children with autism.Journal of Autism and Developmental Disorders, 30,537–552.

Knoll, M., & Charman, T. (2000). Teaching falsebelief and visual perspective taking skills in yourchildren: Can a theory of mind be trained? ChildStudy Journal, 30, 273–304.

Perner, J., Frith, U., Leslie, A. M., & Leekam, S.(1989). Exploration of the autistic child’s theoryof mind: Knowledge, belief, and communication.Child Development, 60, 689–700.

Shabani, D. B., Katz, R. C., Wilder, D. A., Beau-champ, K., Taylor, C. R., & Fischer, K. J. (2002).Increasing social initiations in children with au-tism: Effects of a tactile prompt. Journal of AppliedBehavior Analysis, 35, 79–83.

Sigman, M., & Capps, L. (1997). Children with au-tism: A developmental perspective. Cambridge, MA:Harvard University Press.

Spradlin, J. E., & Brady, N. C. (1999). Early child-hood autism and stimulus control. In P. M. Ghez-zi, W. L. Williams, & J. E. Carr (Eds.), Autism:Behavior analytic perspectives (pp. 49–65). Reno,NV: Context Press.

Received June 25, 2002Final acceptance February 18, 2003Action Editor, Robert Stromer