Play training project : the effects of training method and contingency of reinforcement on toy manipulation skills in profoundly retarded institutionalized children

Abstract

Play has been recognized as a therapeutic modality with the mentally retarded. The objectives of this study were to investigate the effects of four training methods and two reinforcement conditions on play. The eight profoundly retarded subjects who participated individually in 26 daily training sessions were randomly assigned to unique combinations of training method and reinforcement condition. The trainer interacted with the subject for 10 seconds with each of 4 toys, 3 times per session. For subjects in the verbal training method, the trainer told the subject what actions to take with the toys. Appropriate play was modeled by the trainer in the demonstration method. Manual guidance was provided for the subjects in the physical training method. The presence of the toys without intervention constituted the fourth training method. Reinforcement was given either contingent upon 10 seconds of specified play actions or on a noncontingent variable interval schedule. Observational frequency data were collected during each session, classifying play behaviors into four levels of play on the Toy Manipulation Scale, adapted from Barnes (1971). Session scores were weighted for further analysis. The frequency data were analysed by chi-square. The weighted total scores were assessed by Mann-Whitney U and Kolmogorov-Smirnov tests. Individual subject results were analysed by chi-square and interpreted as a series of single subject designs. Subjects in the physical and toys training method had significant results; results of the subjects in the verbal and demonstration methods were invalid. When data for both subjects in each training method were combined for analysis, the verbal method was most powerful in eliciting responses at the highest levels of play. Within each training method, the data for the two reinforcement conditions were compared by the Mann-Whitney U test. In all training methods the subject who was assigned the contingent reinforcement condition performed at higher response levels than did the subject in the noncontingent reinforcement condition. Suggestions for future research included training in small groups over a longer time interval, to maximize treatment effects. To apply this to a residential setting, treatments must be matched to the subject, and may be combined for optimal results.