Introduction

The preceding chapter focused exclusively on distinctions and notation conventions that should prove useful for characterizing the logic of dynamic choice. In this chapter I shall explore a set of conditions that some have thought might plausibly be imposed on any pattern of dynamic choice that is to qualify as rational.

By way of anticipating the substance of the next chapter, the conditions in question can be read as providing one way in which one can factor Hammond's consequentialist principle for dynamic choice, although one must proceed with caution in this regard, since (as indicated at the close of the preceding chapter) Hammond's own construction is considerably more elaborate, and (as I shall indicate more carefully as I proceed) he views the notion of “dynamic consistency” somewhat differently than I do. Roughly speaking, however, the conditions to be explored do capture what Hammond regards as a consequentialist approach, and collectively they can be seen to suffice for the derivation of CF and CIND, the choice-set versions, respectively, of CFO and IND. As such, they manage to provide a simplified and thus, one would hope, perspicuous model of the sort of consequentialist argument that he and others have thought could be constructed in favor of these two basic presuppositions of the modern theory of rationality.

In the presentation to follow, however, I shall develop these conditions in isolation from Hammond's own arguments.