Abstract

It is widely accepted within philosophy of science that indeterministic explanation is possible. In this dissertation I attempt to show that the arguments supporting indeterministic explanation do not warrant the predominance of this view. The arguments play out across two areas of contention. The first concerns meta-theoretical principles that at one time commanded broad acceptance as conditions of adequacy for any proposed model of scientific explanation. These conditions include Principle P, which states that if A explains B, then A cannot also explain -B, and the contrast class condition, which requires that for A to explain B, A must explain why B occurred rather than -B. Under Explanatory Deductivist interpretations, these adequacy conditions block indeterministic explanation, as do Deductive Nomological models of explanation which satisfy these conditions. We find, however, that under non-deductivist interpretations these same adequacy conditions are compatible with indeterministic explanation. This fact shifts the debate to the contention between competing models of explanation, viz. Inductive-Statistical and probabilistic causal models on the one hand accommodating indeterministic explanation, and D-N models that block indeterministic explanation. ;I introduce the debate through a widely cited treatment by Wolfgang Stegmuller. I then confront five distinct arguments against explanatory deductivism and conclude that D-N models remain viable and provide a secure basis for opposing indeterministic explanation. Finally, I engage with arguments defending indeterministic explanation advanced by Carl Hempel, Peter Railton, Paul Humphreys and Wesley Salmon. While I concede that probabilistic causal models provide a coherent and persuasive accommodation of indeterministic explanation, I conclude that the arguments against indeterministic explanation that rely on D-N models maintain significant substance and force. I provide an account of idealization in scientific explanation as a part of my defense of ED, and argue that D-N models properly capture the scientific knowledge represented in examples cited as exemplars of indeterministic explanation