Speaker: Isaac Wilhelm 

Abstract: How should agents change their credences after learning centered propositions like ``It is morning'' or ``I am late''? How should agents update, that is, on centered information? In this paper, I use tools developed in the accuracy literature to answer these questions. According to the theory which I propose, after obtaining evidence that includes a centered proposition like ``It is morning,'' agents should adopt credences which---among all credences compatible with the evidence---are closest to their original credences. This theory generalizes to cases where the agent's evidence includes information about chances. And this theory establishes interesting connections between accuracy and rational updating on centered propositions.