Behavioral contrast and generalization across settings in the treatment of autistic children

Abstract

Previous research on the generalization of treatment gains across settings has typically focused on the question of whether generalization occurs or does not occur. However, the experimental literature suggests another possibility: that behavioral contrast may occur in extra-therapy settings if the reinforcement procedures in the therapy setting are highly discriminable from those in other settings. Therefore, this investigation was designed to systematically assess whether: (1) a highly discriminable treatment procedure in one setting would produce a behavioral contrast effect in other unmanipulated settings; and (2) such contrast-like trends could be eliminated if initially different reinforcement procedures in two settings were subsequently made similar. The results for the eight autistic children who participated in this investigation showed that: (1) When very different reinforcement procedures (primary rewards or punishment in therapy settings, and no rewards or punishment in extra-therapy settings) were in effect, the children showed contrast-like behavior changes in the untreated extra-therapy settings; and (2) such trends in responding could be eliminated, resulting in generalization of treatment gains, if the reinforcement procedures were subsequently made relatively similar in the two settings, even if the procedures consisted essentially of noncontingent reinforcement.