Abstract

A set of constraints forces trade-offs which prevent us from achieving the best possible definitions of the ‘level’ and ‘unit’ of natural selection. This set consists in decisions concerning conflicting pre-analytic intuitions in problematic cases, the relative roles of various conceptual resources in the definitions, which facts need to be accounted for using the definitions, how the relation between selection and evolution orients the definitions, and the relation between the level and unit concepts. Systematic reconstruction and evaluation of leading analyses along these dimensions favors a new functional analysis over Williams’ consequentialist analysis, Sober’s causal analysis, and Dawkins’ teleological analysis