What should you do if you are uncertain about which axiology is true? One popular view is Expected Value Maximisation (EVM), the claim that you should maximise the expected moral value of your actions. In this presentation, I examine how we can understand the notion of credences that EVM presupposes. I suggest that one way to understand it is via a specific interpretation of representation theorems from decision theory. I defend the significance of representation theorems against recent objections, argue that they may indeed provide the best way to understand credences, and show that this would have important implications for the normative structure and significance of EVM.