Peter Godfrey-Smith

Both biologists and philosophers often make use of simple verbal formulations of necessary and sufficient conditions for evolution by natural selection (ENS). Such summaries go back to Darwin's Origin of Species (especially the "Recapitulation"), but recent ones are more compact.1 Perhaps the most commonly cited formulation is due to Lewontin.2 These summaries tend to have three or four conditions, where the core requirement is a combination of variation, heredity, and fitness differences. The summaries are employed in several ways. First, they are often used in pedagogical contexts, and in showing the coherence of evolutionary theory in response to attacks from outside biology. Second, they are important in discussions of extensions of evolutionary principles to new domains, such as cultural change. The summaries also have intrinsic scientific and philosophical interest as attempts to capture some core principles of evolutionary theory in a highly concise way. Despite their prominence, both the proper formulation and status of these summaries are unclear. Standard formulations are subject to counterexamples, and their relations to formal models of evolutionary change are not straightforward. Here I look closely at these verbal summaries, and at how they relate to formal models. Are the summaries merely rough approximations that have no theoretical role of their own? Perhaps they could operate as theoretical statements in Darwin's time, but have now been superseded by more exact treatments.

"Additive variance in fitness" is an important concept in the formal apparatus of population genetics. Wimsatt and Lloyd have argued that this concept can also be used to decide the "unit of selection" in an evolutionary process. The paper argues that the proposed criteria of Wimsatt and Lloyd are ambiguous, and several interpretations of their views are presented. It is argued that none of these interpretations provide acceptable criteria for deciding units of selection. The reason is that additive variance in fitness can be both a cause of evolution, but also a byproduct of selection at another level.

The commentaries by Dennett, Sterelny, and Queller on Darwinian Populations and Natural Selection (DPNS) are so constructive that they make it possible to extend and improve the book’s framework in several ways. My replies will focus on points of disagreement, and I will pick a small number of themes and develop them in detail. The three replies below are mostly self-contained, except that all my comments about genes, discussed by all three critics, are in the reply to Queller. Agential views of evolution, discussed by Queller and Dennett, are addressed in my reply to Dennett

A wide range of ecological and evolutionary models predict variety in phenotype or behavior when a population is at equilibrium. This heterogeneity can be realized in different ways. For example, it can be realized through a complex population of individuals exhibiting different simple behaviors, or through a simple population of individuals exhibiting complex, varying behaviors. In some theoretical frameworks these different realizations are treated as equivalent, but natural selection distinguishes between these two alternatives in subtle ways. By investigating an increasingly complex series of models, from a simple fluctuating selection model up to a finite population hawk/dove game, we explore the selective pressures which discriminate between pure strategists, mixed at the population level, and individual mixed strategists. Our analysis reveals some important limitations to the ESS framework often employed to investigate the evolution of complex behavior.

One version of pluralism was defended in a well-known paper by Sterelny and Kitcher. In this sense, pluralism is the view that any given selective process can be described at a variety of different levels in the biological hierarchy. On Sterelny and Kitcher’s view, one can explain giraffe necks in terms of competition among longer-necked and shorter-necked giraffes, and one can also explain them in terms of competition among the genes that lead to these differences in neck size. Although these descriptions might have different heuristic virtues, neither description has a special explanatory status that the other lacks.