Research Group
in Analytic Philosophy

Abstract

Many philosophers are prepared to admit that some thoughts—demonstrative thoughts and perceptual beliefs—depend directly on perception to individuate their content. If there are such thoughts, they must play a key role in explaining how perceptions justify beliefs, how the reference of singular thoughts is fixed by perception, and how subjects make judgments about properties and relations using perceptual discrimination. But the nature of perception-dependent thoughts remains largely mysterious. Thoughts are supposed to be structured representations modeled on language, composed of concepts and inheriting nothing from sensation. Perceptions are often described as unstructured, analog and nonconceptual. Given these differences, what exactly does it mean to say that the thoughts depend on perception to individuate their content? I examine and reject two existing proposals, then defend a new proposal.

The first proposal is relationism, adopted by many accounts of demonstrative thoughts: perception individuates content for the thoughts relationally, without epistemic representations of that content playing any semantic role in the thought. I reject this proposal on the grounds that it cannot account for demonstrations of properties and relations, in which perceptual modes of presentation do determine the semantic content of demonstrative thoughts. The second proposal is conceptualism: the content of perceptual experience is captured by conceptual states and enters thoughts directly under conceptual (yet perceptual) modes of presentation. This proposal is rejected because the alleged conceptual states do not qualify as concepts.

The third proposal is that nonconceptual modes of presentation are parts of demonstrative thoughts, that is, they determine the semantic content of those thoughts. I’ll ask under what conditions nonconceptual modes of presentation could allow subjects to consciously determine extensions in thoughts, and whether they would violate two constraints which are designed to ensure that thought constituents are recombinable: generality and compositionality. I conclude that those conditions have chances of being met, and that nonconceptual perceptual modes of presentation are likely to determine the semantic content of demonstrative thoughts.