The Brain as a Statistical Inference Engine—and You Can Too

Abstract

There are several possible templates for award talks.1 The most common is an intellectual history—how I came to make all these wonderful discoveries. However, I am never completely happy with my work, as it seems a pale shadow of what I think it should have been. Thus I am picking a different model—things we all know but do not say out loud because we have no evidence to support them; and besides, making such bold claims sounds pretentious. Thus I do not expect to say anything too novel here. I hope all my readers already know that the brain exploits statistics, and most of them suspect that we in statistical computational linguistics have something to say about how this works out in the case of language. My goal is therefore not to say anything you do not believe, but to cause you to believe it more passionately.