Christopher Peacocke

The principal thesis of A Study of Concepts is that a concept is individuated by its possession condition. Concepts are here understood to be sliced as finely as epistemic possibility. So now and 6 o’clock are different concepts, even if, in context, they pick out the same time; likewise for the observational concept circular and the complex concept locus of coplanar points equidistant from a given point. In the simplest cases, a possession condition is stated by giving a true, individuating statement of the form.

Only computational explanations of a content‐involving sort can answer certain ‘how’‐questions; can support content‐involving counterfactuals; and have the generality characteristic of psychological explanations. Purely formal characteriza‐tions of computations have none of these properties, and do not determine content. These points apply not only to psychological explanation, but to Turing machines themselves. Computational explanations which involve content are not opposed to naturalism. They are also required if we are to explain the content‐involving properties of mental states.

This paper has five theses, which are intended to address the claims in Jerry Fodor's paper. (1) The question arises of the relation between the philosophical theory of concepts and epistemology. Neither is explanatorily prior to the other. Rather, each relies implicitly on distinctions drawn from the other. To explain what makes something knowledge, we need distinctions drawn from the theory of concepts. To explain the attitudes mentioned in a theory of concepts, we need to use the notion of knowledge. (2) Concepts can have a normative dimension without their nature being interpretation‐dependent, and without their nature being mind‐dependent. The normative dimension traces back instead to the role of truth as a constitutive aim of judgement. (3) Conceptual combination is to be explained at the level of reference and semantic value. When we respect this fact, there is no difficulty in explaining compositionality for epistemically‐constrained concepts. (4) Epistemically‐constrained theories of concepts are committed to the applicability of the notion of the a priori. They are not committed to the applicability of the notion of analyticity, where this is construed as truth purely in virtue of meaning or the identity of concepts. (5) Possession of a concept has causal‐explanatory powers, whether given by a possession condition in the form presented in A Study of Concepts, or by possession of an implicit conception. A possession‐condition theory of concepts need not be a form of dispositionalism.

It is just over fifty years since the publication of Quine’s ‘Two Dogmas of Empiricism’. That paper expresses a broad vision of the system of relations between meaning, experience, and the rational formation of belief. The deepest challenges the paper poses come not from the detailed argument of its first four sections – formidable though that is – but from the visionary material in its last two sections.1 It is this visionary material that is likely to force the reader to revise, to deepen, or to rethink her position on fundamental issues about the relations between meaning, experience, rationality, and, above all, the a priori. Does what is right in Quine’s argument exclude any rationalist view of these relations? How should a rationalist view be formulated? Those are the questions I will be addressing. I start with the critical part of this task, a consideration of the strengths and weaknesses of Quine’s vision. Drawing on the constraints emerging from that critical discussion, I will then turn to the positive task of articulating and defending a rival conception. The rival conception can be described as a Generalized Rationalism