The Problem of Old Evidence for Bayesianism and Its Relationship to Interpretations of Probability

The problem of old evidence allegedly poses the most serious challenge to the Bayesian confirmation theory. All proposed solutions to this problem can be divided into two types: classical (treating the challenge as serious and trying to meet it) and non-classical (with denying that there is a real problem and trying to dissolve it in one way or another). Classical solutions have been proposed by Garber, Jeffery, and Niiniluoto, and have been criticized by many, among them, Eells and Earman. One of the non-classical solutions is to choose an objective (rather than Bayesian’s subjective) interpretation of probability; this view has been proposed by Rosenkrantz. In this paper, we thoroughly examine the classical solutions and objections that have been raised against them, trying to show that the classical approach is deficient. In the end, we try to make a case for Rosenkrantz’s proposal as the only solution which, we believe, gets to the root of the problem correctly