Abstract

The most prominent recent model of how concepts can have gradient applicability—that is, apply more fully to some items than to others—is that supplied by the prototype theory. Such a model, however, assumes concepts to be internally individuated and structured, and it might thus be challenged by both concept externalism and conceptual atomism. This paper argues that neither of these challenges presents an obstacle to viewing some concepts as having gradient application, and develops a different model of the conditions for such application. I call the notion of gradient applicability I discuss a feature-based one, and distinguish it from both the prototype theory’s typicality-based notion and a vagueness-based notion. On the model I develop, what determines whether a concept has gradient applicability is just the nature of the property it expresses, not facts about its stereotype. Focusing in particular on Fodor’s Informational Atomism, I argue on this basis that externalism is compatible with gradient applicability for all concepts, and that conceptual atomism is compatible with gradient applicability in the case of natural kind concepts, but not in the case of response-dependent concepts. This does not, however, present an obstacle to viewing some response-dependent concepts as having gradient applicability, for I also argue that in fact atomism cannot be true of response-dependent concepts.