Abstract

A classic feature-set model of similarity, the contrast model of Tversky (1977), is applied to a range of phenomena in category-based induction. These phenomena include basic similarity effects, typicality and asymmetry effects, diversity effects, effects of projectibility of properties, and contextual influences on similarity. These analyses help to extend the contrast model to a new area of research as well as to place constraints on the model itself. Although categorization research has largely focused on people's ability to infer taxonomic category labels, categories facilitate a number of cognitive abilities and functions. One of our most important abilities is inductive inference (Anderson, 1991; Billman & Heit, 1988; Heit, 1992; Osherson, Smith, Wilkie, Lopez, & Shafir, 1990), and category-level information enables a rich set of inferences. For example, you might not be able to infer much about Peter until you are told that Peter is a goldfish, in which case you could...

Citations