| This dissertation critically examines the popular model of non-demonstrative inference known as "Inference to the Best Explanation" (IBE), and more broadly, the role that explanatory considerations play in scientific practice. Despite its seeming pervasiveness in science, IBE remains poorly understood. Given the frequency with which IBE appears to show up, not only in science, but also in daily life and in philosophical debates, an investigation into the nature of IBE will also have implications for traditional epistemology and philosophical methodology.;My conclusions about the status of IBE, however, are mixed. While I show how some traditional objections that have plagued IBE can be solved (Chs. 1-2), I conclude that IBE, at least in its canonical formulation, is untenable as an account of scientific reasoning. The central obstacle that IBE faces is that certain "explanatory virtues" cannot be justified by Bayesianism, the dominant approach in philosophy of science to explicating theory confirmation (Chs. 3-4). As I demonstrate, only some explanatory virtues can be given a Bayesian explication. As a potential reply, I formulate a novel argument that appeals to the epistemic value of understanding, whose conclusion is that IBE is superior to Bayesianism; however, as I show, this argument fails (Ch. 5). Finally, I attempt salvage some aspects of IBE that are difficult to justify from the Bayesian framework by arguing that certain explanatory virtues are important in determining which hypotheses are worthy of pursuit (Ch. 6).;To defend this view, I appeal to the recent high-profile dispute over the scientific legitimacy of string theory. Ultimately then, we are left with a two-tiered view of the explanatory virtues, one according to which some virtues relevant to what theory ought to be believed, whereas others are relevant to what theory ought to be pursued. |
