Introduction

What arguments can be offered for taking WO and IND (or alternatively CF and CIND) as normative for choice? The question of the validity of normative principles of choice poses a methodologically troublesome problem. Since a norm prescribes rather than describes human behavior, it cannot be treated as a hypothesis or a theory about human behavior that is subject to the usual sort of empirical test. On the more traditional way of thinking about the logic of the justification of norms, there is no alternative but to appeal to even more “fundamental” principles from which the one in question can be derived. And, of course, to avoid an infinite regress, this appeal to higher authority must at some point come to an end with principles that simply recommend themselves to intuition.

For some, the end comes much sooner, with a suggestion that either WO or IND is intuitively acceptable. Most theorists, however, have tried to say something by way of defending the principle in question – by relating it to other, more familiar conditions, by responding to counterarguments, or simply by telling some sort of story. Arrow, for example, defends the ordering condition by appeal to the notion of what is necessary if choice is to be possible, and Samuelson (as already noted) suggests that reflection on the logical nature of disjunctive prospects reveals that the sort of noncomplementarity that is presupposed by IND is very plausible.