Proceedings of the Annual Meeting of the Cognitive Science Society

A key function of categories is to help predictions about unob-served features of objects. At the same time, humans often findthemselves in situations where the categories of the objectsthey perceive are uncertain. How do people make predictionsabout unobserved features in such situations? We propose arational model that solves this problem. Our model comple-ments existing models in that it is applicable in settings wherethe conditional independence assumption does not hold (fea-tures are correlated within categories) and where the featuresare continuous as opposed to discrete. The qualitative predic-tions of our model are borne out in two experiments.