The Role of Similarity in Natural Categorization

The intuitive idea that those things that we put things into categories because we find them similar appears to be non-controversial, if not circular. Cars are clearly more similar to other cars than they are to trees, and trees more similar to other trees than they are to cars. However, a number of theorists have recently questioned the degree to which the notion of similarity is sufficiently clearly defined and constrained to serve as an explanation of our categorization. In this chapter, I discuss the arguments for and against basing categorization on a notion of similarity, and conclude that, construed broadly, similarity may yet be the best explanation of how most of our conceptual categories function. I propose a distinction between concepts viewed as a cultural phenomenon and concepts at the psychological level, and suggest a naive model of conceptual development that starts with concepts as similarity clusters and only in restricted cases replaces these with more causal or theory-based conceptual representations.