Joshua Edward Pearson
I present an intractable puzzle for currently popular view that belief is weak---the view that expressions like 'S believes p' ascribe to S a doxastic attitude towards p that is rationally compatible with low credence that p. The puzzle concerns issues that arise on considering beliefs in conditionals. I show that proponents of weak belief either cannot consistently apply their preferred methodology when accommodating beliefs in conditionals, or they must deny that beliefs in conditionals can be used in reasoning.
I outline a novel counterexample to the plausible principle of belief revision, Anticipation. Anticipation says that if you would not be justified in believing P were you to learn E, and you would not be justified in believing P were you to learn not-E, you cannot now be justified in believing P. If I’m right, then not only is the leading theory of belief revision false, but so are various proposed weakenings. I outline an alternative theory that is able to accommodate the counterexample to Anticipation. Endorsing this theory--or any other theory that rejects Anticipation--has radical consequences; for instance, it means either that it can be rational to refuse free evidence before making a decision, or else it means that justified beliefs play a diminished role in rational decision making.
Inspired by a case from Ginsberg concerning counterfactuals, Stalnaker outlined the now famous “Composers” case which, if what I call the “cautious” judgements concerning it are correct, causes serious trouble in epistemology: various theories of belief revision are false, as is the thesis that rational belief supervenes on rational credence. In contrast, the “bold” judgements regarding Composers cause no such trouble. All proposed models of Composers so far side exclusively with either the cautious or bold judgements. As Stalnaker notes, this is undesirable: both judgements seem reasonable, and which is to be preferred depends only on how epistemically bold or cautious the agent in question is. I outline a new theory of justified belief---independently motivated in my Belief Revision Revised---that can predict this permissive feature of Composers, while still maintaining a substantive connection between rational belief and rational credence.
In ordinary circumstances, on taking a sample from a general pool, one can have "representative knowledge": knowledge that one's sample is representative of the general pool. Combining this anti-skeptical assumption with multi-premise closure principles for justified belief leads to a puzzle: one is in a position to justifiably believe that all possible samples are representative, not just the particular sample one has taken. We can hold onto both representative knowledge and multi-premise closure so long as we are willing to maintain that representative knowledge is asymmetric in various ways: whilst you can believe your sample is representative of the general pool, you cannot believe this of all possible samples; whilst you can believe your next sample is representative, you cannot believe this of your future samples; and whilst you can believe that your sample is representative, you cannot believe this of samples taken by other agents.