Abstract: | After being pretested to determine base levels of imitation, 32 9-14 yr old retarded children were reinforced for imitating a model in 9 training sessions. Ss in a single model condition were reinforced by the same model across all sessions, whereas Ss in a multiple model condition were reinforced by 3 different models (3 sessions per model). A posttest to assess levels of imitation was then conducted by a model with whom the Ss had not had contact and who demonstrated a new set of behaviors. Results during training sessions show that (a) Ss learned to imitate, and this learning was not inhibited by multiple models; and (b) Ss generalized and imitated nonreinforced behaviors, and this response generalization was facilitated by multiple models. Most importantly, pre-posttest comparisons indicated that generalized use of the new response class (imitation) with new models was 8 times greater for Ss trained with multiple as opposed to single models. Implications for the maintenance and generalized effectiveness of social intervention programs are discussed. (PsycINFO Database Record (c) 2010 APA, all rights reserved) |