Multiple-Objective Decision Making.

Abstract

Decision makers often approach decisions with a divided mind. Rather than having clear, overall preferences between options, they evaluate them according to many criteria. Worse still, these criteria often conflict in their rankings of options. Such decisions are hard, but it is clear that they can be resolved in better and worse ways. However, decision theoretic accounts of rational decision making have little traction on these realistic cases. This is because a complete, conflict-free preference order over the available actions, representable by a formal structure at least as robust as a weak order is typically understood to be the essential prerequisite for rational action.

There has been very little work on this problem within the philosophical literature; here, I lay the groundwork for an account of good multiple-objective decision-making. The first step is to characterize acceptable methods for constructing overall preference orders from the objective-specific rankings which are accessible to the decision maker. Here, I consider two such methods.

While there has been little work on multiple-objective decision making within philosophy, the problem has received considerable attention in the decision analysis literature. In the first chapter, I argue that decision analytic methods for constructing overall preferences are philosophically well-motivated, and explore how they can be applied to some simple examples of multiple-objective decisions.

In the second chapter, I consider an altogether different approach, which takes at face value the analogy between an individual decision maker trying to reconcile several objectives in her decision and a group of several individuals trying to reach a joint decision. The thought is that multiple-objective decisions can be modeled as social choices---in the sense of Social Choice Theory. The challenge is that such an approach seems to run headlong into the limiting result of Arrow's Theorem. Against earlier work on this approach, I argue that Arrow's Theorem does not apply to individual decisions.