Conceptual flux: The case against mental misrepresentation.

Naturalistic theories of the content of mental representations almost universally hold that mental content is a function of the use of mental representations. However, use theories of meaning have a problem explaining how misrepresentation could be possible. If all uses count in fixing meaning, then none of them can be misuses, and there can be no misrepresentation (as well as no conceptual error or false belief). Typical use theories seek to limit the uses which count towards meaning, and they propose criteria which are supposed to make some uses not count in the determination of the meaning of a mental representation, so that some of those non-meaning-fixing uses could be misrepresentations. I argue that none of these attempts can succeed. They are either inconsistent, or question-begging and arbitrary. This includes all two-factor conceptual role theories, causal theories, informational theories, and adaptational role theories. Most also cannot allow for misrepresentation, and those which can do so by invoking some non-naturalistic source of meaning. So, I conclude that no naturalistic theory of content is consistent with the possibility of misrepresentation. Moreover, I propose a way to make sense of content without misrepresentation, by adopting a pragmatic view of representation, inference, and learning. On such a view, identification of an 'error' is interpreted as the adoption of a new concept and discontinuation of the use of the old concept. Concepts are judged with an eye not to truth, but rather, to their utility. We adopt some concepts and not others in virtue of their success in maneuvering through the world, and the social pressure to speak and think using the concepts which others use. Communication between people holding different concepts is still possible provided that the concepts are similar enough for the differences to be irrelevant in the context of the communication. This pragmatic view of meaning and concept change has significant implications for epistemology, metaphysics, truth, philosophy of science, psychology, and artificial intelligence. I pay special attention to the issue of meaning holism, explaining the various levels of holism that the no-misrepresentation can entail