On the similarity of features

Tversky's featural contrast model for similarity judgments is discussed and its featural definition of entities and Boolean characterization of judgments are criticized. Features are themselves identified on the basis of appraisals of similarity and so cannot be defined in an a priori or context-free manner. Tversky's model may assist the discovery of orderly psychological patterns, but misses the central cognitive issues of concern. Empirical findings relevant to this critique are reviewed. The criticisms of Tversky's model lead to a more general critique of the representational-computational model of mind and its associated methods.