Representations Are Rate-Distortion Sweet Spots

Abstract

Theorists working on information and representation are often not centrally concerned with “Shannon” information, as it is often put, but with some other, sometimes called “semantic,” kind of information. This perception is wrong. Shannon’s theory of information is the only one we need. I sketch a (Shannon) informational account of representation, for a certain important family of cases. This account, which represents a significant departure from the Dretskean philosophical mainstream, will show how a number of popular proposals about the purportedly noninformational ingredients in representation actually belong in the same coherent, purely information-theoretic picture.