Abstract

Classical empiricists are notorious for claiming that cognition is perceptually based. This dissertation reassesses that claim against the background of contemporary cognitive science. More specifically, it develops a broadly empiricist theory of concepts and argues that this theory outperforms other theories. In the first chapter, I lay down seven desiderata. An adequate theory of concepts should lend itself to an explanation of expressive scope, intentionality, cognitive content, acquisition, categorization, compositionality, and publicity. I argue that none of the leading theories that have been developed by philosophers and psychologists can satisfy all of these desiderata . ;In the second chapter, I propose an alternative theory, which identifies concepts with what I call 'proxytypes'. A proxytype is a collection of stored, high-level perceptual representations, which represent features typically exhibited by members of the category it designates. I argue that proxytypes lend themselves to an economical account of concept acquisition by bootscrapping on independently motivated theories of perception, and I provide an account of categorization by exploiting the kinship between proxytypes and prototypes. The third chapter raises three objections to proxytype theory, involving three of the remaining desiderata: expressive scope, publicity, and compositionality. In responding to these objections, I develop a viable successor to the reviled analytic/synthetic distinction, and I outline a three-stage model of concept combination. The final chapter completes my defense of proxytype theory, by taking on the last two desiderata. It provides a naturalistic account of intentionality and cognitive content inspired by Locke's distinction between real and nominal essences. The real content of a proxytype is the property of belonging to some unique kind, and the nominal content is a set of properties typically exhibited by members of that kind. I contrast my account of nominal content with three theories of narrow content and conclude that it has all of their merits and none of their flaws.