Prior knowledge and exemplar similarity in category learning: Further evidence for their integration

Abstract

Two experiments sought to test the theory that the effects of prior knowledge on category learning are likely to be modified by exposure to novel exemplars. In each experiment, the conditions under which exemplar learning took place were varied and the effects of these contextual factors on the subsequent use of prior knowledge in a categorisation transfer test were examined. In Experiment 1, it was found that prior knowledge had a greater effect when participants were trained to classify a relatively small number of exemplars than when classification learning involved a larger category size. In Experiment 2, it was shown that the contribution of prior knowledge was more significant when participants' expectations about the distribution of exemplars matched their observations of the category structure. These and other findings demonstrate that knowledge-driven processes and exemplar-driven processes are interdependent, and that a comprehensive account of category learning must explain how they are dynamically integrated during the acquisition of new knowledge.