摘要
Bayesianism is a theory of probabilistic reasoning that attempts to capture the logic of confirming and disconfirming hypotheses. I first argue that Bayesianism reveals striking parallels between structures universally held as paradigms of rational belief systems and structures typically considered clear examples of irrational belief systems. I next explain that the crucial difference between these two types of belief systems is found not inside the systems but outside them, in the dynamics, i.e., the attitudes, by which such systems are revised and maintained. The principal attitude that distinguishes these belief systems is ＂open-mindedness.＂ I conclude that rationality and irrationality are primarily properties of attitudes, and derivatively of persons （who exhibit such attitudes） and of beliefs （that are maintained by such attitudes）. It turns out then that, on the one hand, the Bayesian approach reveals important truths about the nature of rationality and irrationality, but, on the other hand, it is inadequate as a theory of rationality, since it leaves some aspects of rationality and irrationality unaccounted for. The Bayesian analysis on the basis of which these conclusions are reached arises from a careful examination of the Duhem problem, which is the problem of determining the disconfirmation impact on the plausibility of hypotheses collectively responsible for a false observational consequence.
