Search Results

This chapter develops and defends a new account of essentially indexical, de se, or self-locating belief with the aim of using it in later chapters to clarify the analogy between self-locating information and phenomenal information. The account is a modified version of David Lewis's analysis of self-locating information in terms of centered possible worlds, which are abstract objects consisting of a possible world plus a designated person and time in the world. It is argued that the modified analysis is better able than Lewis's to account for the way knowledge and belief change over time, and for one person's knowledge and belief about the knowledge and belief of others. The analysis is applied to two famous examples from the literature on essentially self-locating belief — David Lewis's case of the two omniscient gods, and Adam Elga's Sleeping Beauty puzzle. The chapter concludes with an appendix that sketches some of the formal details.

The main constructive task of this chapter is to provide a way of modeling self-locating or essentially indexical information in a way that allows for the communication of this kind of information, and for a representation of common ground that includes information about where and when in the world the participants in a conversation locate themselves and each other. The model uses the resources introduced in David Lewis’s theory of self-locating belief, which models belief with centered worlds, but modifies his account in a way that allows for the explanation of the interaction of the self-locating attitudes of different agents, and of the change of propositional attitudes over time.

It is argued, following David Lewis, that we should model a cognitive state by a set of centered possible worlds, since this is required to represent the believer’s self-locating or indexical knowledge and belief. But it is also argued, contra Lewis, that we should take the contents of belief to be propositions, represented by sets of uncentered possible worlds, since this is required to give a perspicuous account of agreement and disagreement of different agents, and of change of belief over time. Reconciling these two thoughts requires a defense of Propositionality: roughly, the thesis that any ignorance of where one is in the world is also ignorance about what the world in itself is like. This thesis is defended against some criticisms, and motivated by an externalist picture of knowledge and intentionality.

This chapter compares the Certainty-Loss Framework (CLF) developed over previous chapters to various other Bayesian schemes for updating self-locating degrees of belief. In particular, it considers frameworks proposed by Joseph Halpern and Christopher Meacham, Sarah Moss, and Robert Stalnaker. CLF is found in some cases to get particular examples right the other frameworks get wrong, and in other cases to apply more clearly or without the need for supplementary apparati. These comparisons shed additional light on the motivations for CLF and on its strengths as a modeling framework. The Stalnaker discussion also completes an argument (begun in earlier chapters) that Bayesians cannot handle context-sensitivity simply by retaining Conditionalization and reconstruing the objects of degrees of belief.

This chapter explains how to apply the Certainty-Loss Framework (CLF) to stories involving self-locating belief and context-sensitivity. This formal modeling framework and its distinctive updating rules (Generalized Conditionalization and the Proper Expansion Principle) were defined and defended in previous chapters. Here the framework is used to model rational requirements in a number of stories, most importantly Adam Elga’s highly controversial Sleeping Beauty Problem. After analyzing solutions to that problem offered by Elga and by David Lewis, the chapter shows two different ways of constructing CLF models that refute Lewis’s solution (without invoking indifference principles or even Lewis’s Principal Principle). Objections to the CLF models are considered and rebutted.

This chapter examines the difficulties self-locating degrees of belief cause for traditional Bayesian modeling frameworks. It argues that these problems are caused not by Bayesianism’s view of the contents to which degrees of belief are assigned, but instead by the way Conditionalization (the traditional Bayesian updating norm) interacts with context-sensitive claims. Solving these problems requires us first to understand the interactions between formal models of the same story based on different modeling languages. A Proper Expansion Principle (PEP) is proposed that governs such interactions. Combining PEP with Generalized Conditionalization (an updating rule defended in previous chapters) yields the Certainty-Loss Framework (CLF), a formal framework that correctly models rational requirements in stories involving context-sensitivity. The version of PEP presented in this chapter improves on earlier versions proposed by the author, which were susceptible to a counterexample invented by Sarah Moss.

It is often assumed that experiential reference, in particular the references we make using so-called essential indexicals (I, here, and now), is irreducible to other forms or reference. In focusing on Donnellan’s insights concerning the referential use of definite descriptions and empirical evidence coming from cognitive sciences (in particular Pylyshin’s work on situated vision), Eros Corazza discusses and defends this view. In so doing, he shows how experiential reference rests on a form of egocentric immersion underpinning agent-centered behaviours. It is further argued that our capacity to express de se thoughts (i.e. thoughts about ourselves) supervenes on the ability we have to master viewpoint-dependent thoughts. This constitutes the cognitive grounds upon which philosophical insights concerning the notions of essential indexicals, self-locating beliefs, and self-centered behaviours should be understood.

This chapter presents a variant of the “Sleeping Beauty” case that shows that those who are “halfers” with respect to the original Sleeping Beauty problem are committed to holding that rationality can be perspectival in a rather extreme and surprising way. Specifically, halfers are committed to the possibility of “robustly perspectival” contexts, contexts where two maximally rational disputants in perfect communication are rationally required to disagree despite knowing that their epistemic positions are equally strong. Unlike halfers, “thirders” are committed to a different result in the new case that does not involve any counterintuitive perspectivalism. While this case could be seen as providing a new and powerful argument against the halfer position, it is argued that there are good reasons for thinking that the case does involve a robustly perspectival context, a conclusion that calls into question a key principle often taken for granted in the disagreement literature.