Title: Indecisive Decision Theory
Abstract: I argue that the correct decision theory should be indecisive -- for some decision problems it should decline to say either that one choice is best, or that several choices are tied. The main argument concerns decision problems involving a predictor that look a lot like Stag Hunt games. In these problems there are two ratifiable choices, and there is no way for a decision theory to systematically say which of them one should take without that theory violating some basic principles of dynamic coherence. The right lesson is that both choices are permissible, and the correct theory is indecisive about which of them should be taken.
Following the lecture, there will be an hour for seminar discussion.