| This dissertation explores issues relating to enriching our theory of rational belief change to accommodate a broader set of doxastic attitudes than mere first-order beliefs. In Chapter 2, I introduce the orthodox AGM theory of belief revision by contrasting it with an alternative (broadly Bayesian) approach to qualitative belief revision. This alternative approach is defined by way of a diachronic version of the Lockean thesis, which requires coherence between the agents beliefs and credences. A number of surprising and novel results concerning the relationship between Lockean and AGM revision. The remaining chapters explore some peculiarities for the diachronic norms of epistemic rationality that result from 'future-directed higher-order beliefs' --- i.e. agents' beliefs about their future beliefs. In Chapter 3, I consider their effects on the norms of introspective inference and argue that they introduce the possibility that the agent's procedure for belief change is relevant for determining which rules of inference the agent may employ with the guarantee of truth-preservation. This suggests the familiar account of validity, according to which only the meanings of the logical parts of the language can determine the validity of an inference rule, does not fully capture guaranteed truth-preservation. This chapter additionally includes an application to the surprise examination paradox and offers a novel solution. In Chapter 4, I show that the peculiarities generated by future-directed higher-order beliefs are not limited to introspective inference and establish a result involving agents with inconsistent beliefs. By constructing a revised version of the surprise examination paradox, in which the agent begins with straightforwardly inconsistent beliefs, I demonstrate that there are situations in which the agent has no choice but to give up some true belief if she is to resolve the inconsistency in her prior beliefs. This is surprising since one might think that when an agent has logically inconsistent beliefs, there should be some way to "surgically" remove the false beliefs she has without also cutting away some true beliefs. But, in the case discussed, all of the logically possible options for resolving the agent's inconsistency include the removal of some true beliefs. Additionally, Chapter 4 includes several alternative demonstrations of the main result of this chapter, which taken together provide a compelling argument that future-directed higher-order beliefs are responsible for this behaviour rather than other contingent features of the case. |
