Choosing what to decide

This chapter addresses what is sometimes referred to as the ‘framing’, or ‘problem specification’, or ‘editing’ phase of rational decision making. The point of departure is simple. Before you make a decision, you have to choose what to decide, i.e. determine the relevant alternatives, states, and outcomes of your decision problem. Consider, for example, Savage’s famous omelette example.1 An agent intending to cook an omelette has just broken five good eggs into a bowl, and now has to decide whether to break a sixth egg, that might be rotten, into the omelette. Before deciding what to do, the agent has to determine exactly what sets of alternatives, states, and outcomes to take into account, and how to represent these entities in a formal representation.

This chapter is entirely devoted to developing a normative theory of the predeliberative phase of rational decision making. At first glace it might be thought that the pre-deliberative phase is of little relevance for the controversy over Bayesianism and non-Bayesianism, which is supposed to be the main topic of this book. However, a number of concepts developed in the present chapter will play a crucial role in the axiomatisation of the expected utility principle undertaken in Chapter 7, and in the formulation of some of the impossibility theorems in Chapter 8. The present discussion can thus be seen as a preparation for future chapters.