Abstract

The aim of my dissertation is to develop a model of mechanistic explanation that makes sense of the causal nature of evolutionary explanations. I argue that evolutionary explanations are mechanistic: they are intended to describe the causal production of distributions of properties across populations of objects. ;I define mechanisms as sets of sequences of causal interactions. Mechanisms are described in explanatory contexts by a sequence of distributions of properties and of causal interactions in populations of physical objects. The first distribution is generatively connected to the last distribution by way of the causal processes described by intermediate distributions. ;A mechanistic explanation is modeled as a derivation of an explanandum distribution from a specification of an initial distribution and a mechanism. A putative mechanistic explanation is correct when the mechanism it specifies was operative, false when the mechanism it specifies was not operative, and fails to be an explanation at all if it does not specify a mechanism. ;I apply this explanatory model to selection explanations in evolutionary biology. This application requires a new definition of natural selection. On the suggested definition the state of any evolving population is mechanistically explained by a statistical description of the causal mechanism driving evolution in the population. ;The model of explanation yields a defensible account of causal, mechanistic explanation in statistical sciences. Moreover, the definition of natural selection motivates solutions to three long-standing problems in evolutionary biology: it obviates the circularity objection; it allows a resolution of the controversy over units and levels of selection; and, it allows for selection in causally non-homogeneous environments