Dynamics of dimension weight distribution and flexibility in categorization
Koen Lamberts, Steven Chong · 2000
Abstract As the other chapters in this volume will have explained, rational analysis attempts to understand cognition in terms of its adaptation to the environment. In this chapter, we are particularly concerned with rational analysis of categorization. The purpose of categorization is to predict category membership of objects, which in turn allows for the prediction of new object features. If one categorizes an object as a tiger, for instance, on the basis of its colour and shape, one can infer that the object is probably dangerous, carnivorous, has sharp teeth, and so forth. Ideally, categorization should take the structure of the environment into account, and exploit the regularities that can be found. Anderson (1990, 1991) has provided a detailed analysis of the structure of categorization, and has shown how a Bayesian framework can form the basis of a detailed and powerful model of category learning and categorization.