Rationality and epistemic paradox

This paper provides a new solution to the epistemic paradox of belief-instability, a problem of rational choice which has recently received considerable attention (versions of the problem have been discussed by — among others — Tyler Burge, Earl Conee, and Roy Sorensen). The problem involves an ideally rational agent who has good reason to believe the truth of something of the form:[A_p] p if and only if it is not the case that I accept or believe p.Belief in the latter claim, so the problem runs, must render the agent unable to come to a stable, rationally defensible decision about whether to accept p itself, since each decision can in the event clearly be seen to be unwise. The solution defended in the present paper suggests that in its most serious form the problem beguilingly — but erroneously — assumes that rational agents are always allowed to assume their own rationality when deciding how they should choose.The first draft of this paper was written while I was a visiting fellow at the Department of Philosophy (Research School of the Social Sciences) of the Australian National University. I am grateful to the department and the school, and especially to Frank Jackson and Philip Pettit, for their support during my tenure of the fellowship, and to various commentators at Universities in Australia and New Zealand for their useful criticisms. In this connection I owe special thanks to Barbara Davidson, Philip Pettit, Huw Price, and Richard Sylvan. I have also benefitted from correspondence with Earl Conee and Dick Epstein, and am greatly indebted to a number of anonymous referees forSynthese for helping me to clarify some utterly crucial aspects of my argument.