Acquisition of basic object categories by severely handicapped children.

The present study was designed to investigate category acquisition (a) as a function of initial exposure to only good exemplars as opposed to exemplars varying in goodness-of-example and (b) as a function of exposure to single as opposed to multiple exemplars. This research was undertaken within the framework of the best-example theory of categorization. Severely handicapped prelinguistic children and teenagers were selected as subjects because of the extreme difficulty they experience in generalizing. Experimental control was obtained through a repeated-measures Latin square design. The data indicated that training based on 1 or more good exemplars resulted in significantly more accurate generalization than training with a range of exemplars. In fact, training with a range of exemplars did not result in generalization above chance levels. Training based on multiple good exemplars tended to lead to more accurate generalization than training based on a single good exemplar. The results are explained in the context of the best-example theory and are briefly related to teaching practices.