Computational models of the development of perceptual expertise

Abstract

In a recent article, Palmeri, Wong and Gauthier have argued that computational models may help direct hypotheses about the development of perceptual expertise. They support their claim by an analysis of models from the object-recognition and perceptual-categorization literatures. Surprisingly, however, they do not consider any computational models from traditional research into expertise, essentially the research deriving from Chase and Simon’s chunking theory, which itself was influenced by De Groot’s study of chessplayers. This is unfortunate, as a series of computational models based on perceptual chunking have explained a substantial number of phenomena related to expert behaviour and provide mechanisms that directly address the question of perceptual expertise.