ABSTRACT
Consider the following description: B faces a choice point where she can either X and get payoff R or not X and get payoff S. Though this description might seem mundane, reflection on it raises some interesting issues. Given that virtually all actions or courses of action take place over time, B can be capable of (intentionally) X-ing only if B is a temporally extended agent that can have and retain the goal of X-ing from the initiation of the action or course of action to its completion. If, however, X-ing takes more than just a brief moment, and there are, practically speaking, multiple potential points of reconsideration, it might be wondered whether what is described as B’s X-ing is better described as a series of momentary agents {B1, B2…, Bn}, performing a series of momentary actions {X1, X2,…, Xn}. Relatedly, it might be wondered whether what is described as a payoff to B is better described as a potentially unequal distribution of payoffs to some group of momentary agents that may fail to include some or even all of the momentary agents performing {X1, X2,…, Xn}. However things are described, the following interesting questions arise: Are there rational constraints regarding the discounting of future utility? Are there rational constraints regarding the adoption of goals that can only be carried out in the future? Are there rational constraints regarding reconsidering or abandoning previously adopted goals? This chapter will consider these questions and closely related questions that are tied to the following facts: although agents, like actions, are normally at least somewhat temporally extended, they need not fully identify with their past and future selves; moreover, although agents are capable of non-deliberatively following through with goals they have adopted, they are also capable of reconsidering and changing courses.
