Abstract

Representation is a pivotal concept in cognitive science, yet there is a serious obstacle to a naturalistic account of representations’ semantic content and intentionality. A representation having a determinate semantic content distinguishes correct from incorrect representation. But such correctness is a normative matter. Explaining how such norms can be part of a naturalistic cognitive science is what I call the normativity problem. Teleosemantics attempts to naturalize such norms by showing that evolution by natural selection establishes neural mechanisms’ functions, and such functions provide the normativity requisite for a determinate semantic content. I argue that such attempts fail, because when specifying functions, and thus semantic contents, that are determinate enough to enable misrepresentation, they must tacitly appeal to human normative practices, especially the practice of giving intentional states as reasons for actions. I present a different tactic: using evolution by natural selection to avoid rather than solve the normativity problem. Representations’ semantic contents and their intentional targets are irreducibly normative. Semantics and intentionality are constituted within human normative practices. However, evolution by natural selection can be used to naturalistically explain the transition from a world without human beings and human normative practices — and thus without any distinction between thoughts that may be called “correct” or “incorrect” — to a world in which such human practices and distinctions are commonplace