Abstract
Few studies to date have examined time reallocation in naturalistic, multiresponse human repertoires when one or more responses are restricted. For this experiment, free-operant baseline levels of six responses were measured for four autistic children. The high-probability responses were made unavailable, one at a time, such that subjects had access to five, four, three, and two responses in successive restriction conditions. A return to the six-response free-operant baseline condition completed the experiment. Results were compared to predictions made by four time-reallocation models. These results were described accurately only by the selective substitution model. Further analyses examined alternative explanations for the individual reallocation patterns obtained. An expanded selective substitution definition is proposed that may characterize orderly patterns observed in multiresponse repertoires under restriction conditions more accurately than the other existing models.