The reliability of inference to the best explanation.

I analyze inference to the best explanation (IBE), and defend it against three problems. First, it seems we have no reason to think that the explanatory virtues exemplified by a theory count as evidence that the theory is true. Second, for all we know, the explanations we consider do not include the true explanation, and so our inference to the best of these explanations is doomed to failure. This is the problem of the bad lot recently defended by Bas van Fraassen. These two difficulties demand an argument that beliefs generated by IBE are justified. I answer this demand by developing a version of comparative reliabilism according to which a belief is justified if it is produced by a reliable process and there is no reliable, competing process. Thus, if IBE is reliable, and there is no reliable, competing process, beliefs generated by it are justified. The pessimistic induction is the induction from the falsity of previously-accepted scientific theories to the probable falsity of currently-accepted scientific theories. The pessimistic induction threatens to undermine the justification of beliefs generated by IBE, for if it is correct, IBE is an unreliable form of inference. I meet this challenge by arguing that IBE is reliable conditional on a background of true beliefs, including beliefs concerning which explanatory virtues are evidential. The past unreliability of IBE when these conditions were not met does not impugn the conditional reliability of IBE. I then follow Richard Boyd in arguing that the best explanation of the instrumental success of science, given the dependence of IBE on background beliefs, is that IBE is conditionally reliable. Because this is an abductive argument, concerns about circularity arise. I argue that these are defused by reliabilism.PhDPhilosophyPhilosophy, Religion and TheologyUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/128511/2/3029420.pd