Abstract

Two experiments were conducted to examine the effects of retention interval and the familiarity of stimuli on classification and old-new recognition of ill-defined category instances using the learning-transfer phase paradigm. The generalized context model(GCM) and the prototype abstraction model were formalized to quantitatively contrast in predicting classification performance. In experiment 1, random dot patterns were used as category instances reflecting the characteristics of ill-defined category well. In experiment 2, Reed's(1972) schematic faces were used as category instances that were more meaningful than the random dot patterns. The retention interval was manipulated in immediacy, a week, and a month. The results of experiment 1 using dot patterns showed that prototype information was not used in the classification and old-new recognition of category instances. However, there was a tendency that the fitness of GCM was decreased and that of prototype model was increased as the retention interval increased. In contrast, the results of experiment 2 using schematic faces showed that prototype information contributed significantly to the classification of category instances. Especially, the combined models showed that prototype information made more contribution than exemplar information to the classification and old-new recognition of category instances as the retention interval increased. The implications of this study were discussed.