Abstract

Positing representations and operations on them as a way of explaining behavior was one of the major innovations of the cognitive revolution. Neuroscience and biology more generally also employ representations in explaining how organisms function and coordinate their behavior with the world around them. In discussions of the nature of representation, theorists commonly differentiate between the vehicles of representation and their content—what they denote. Many contentious debates in cognitive science, such as those pitting neural network models against symbol processing accounts, have focused on the types of vehicles proposed for mental representation and whether they have the appropriate structure to succeed in bearing their contents. Philosophers, in contrast, have focused their debates on content and the particular way in which vehicles might bear content—that is, the process of representing rather than the format of representations. I will offer a novel answer to the question of how it is that a representation has content by focusing on the architecture of representation. My proposal is that representations occur in a particular type of mechanism—one in which a control system regulates a plant—and that we can gain traction on cognitive systems of representation by considering how this works in physical systems more generally.