1. Introduction

On the one hundredth anniversary of Darwin’s death, Richard Lewontin (Reference Lewontin1983) reviewed a batch of evolution-themed titles for the New York Review of Books. In the course of his discussion, he described what (among other things) had been novel about Darwin’s theory of natural selection, by contrasting “two basic dynamic forms for evolving systems.” One way that a group might evolve is by each member of the group itself changing, for example, “the Harvard class of 1950 is getting grayer and flabbier, because each of its members is doing so.” The other way a group might evolve is by a “change of frequency of different variants, rather than by a set of developmental transformations of every individual” (23). Contrast the Harvard class of 1950 with the Latin boy band Menudo, whose members were replaced as they reached puberty. The band was all Puerto Rican in 1977 and completely non–Puerto Rican by 1996, but not because anybody changed nationality. The Harvard class of 1950 changes through individual transformation, while Menudo changed through (nonrandom) replacement. Darwin’s theory of natural selection was a replacement theory, said Lewontin, whereas previous attempts to explain evolutionary change (e.g., Lamarck) had used individual transformation.

In The Nature of Selection (Reference Sober1984/1993, 147–55), Elliott Sober reformulated Lewontin’s point within the framework of contrastive explanation (‘w explains why p rather than q’). The process by which the Harvard class of 1950 evolves (aging) explains why the class was, in 1983, (a thousand pardons) gray and flabby rather than those very same folks being young and fit. But in the case of Menudo, the process by which the population evolves (selection of new members) does not explain why the 1996 band was completely non–Puerto Rican rather than those very same folks being Puerto Rican. Rather, what the Menudo selection process explains is (among other things) why the 1996 band included no Puerto Ricans rather than the band consisting of different people (who, like the original members, were Puerto Rican). Biological evolution, said Sober, is like Menudo evolution. Looking at a population of organisms today, for certain features of those organisms (i.e., adaptations), natural selection operating over time in the past explains why that population consists of organisms with those features rather than the population consisting of different organisms with different adaptations. Selection does not, however, explain why the organisms in front of us have their adaptations rather than those very same organisms having different adaptations.

Sober’s remarks ignited a debate that is still going after nearly 30 years. My aim in this note is to address one recent and prominent critique of Sober’s position and to argue that this criticism misconstrues and talks past Sober’s position rather than addressing it.

2. Nanay’s New Critique of Sober

Abbreviating Sober’s explicitly contrastive explanandum, Nanay (Reference Nanay2005) asks, for any adaptation and any specific organism that has it, whether selection can explain why that individual organism has that trait. Nanay argues that it can. Since my claim is that Nanay’s conclusion is neither here nor there, I will give only a brief sketch of his argument. Consider a population in which some members have trait A, the rest have trait B, and selection favors A over B. Choose an arbitrary A-bearing organism from the population, and call it X. And suppose that X inherited A from its mother; call her M. Nanay understands Sober to limit selection’s proximate influence to the selected-against individuals and follows suit by attributing to selection only results that are further ramifications of B-bearing individuals dying.Footnote ¹ Selection against B-bearers in the parental generation, the argument goes, leaves more resources for the A-bearers (assuming, reasonably, that total resources are limited). These extra resources in turn help the A-bearers (including M) to survive and reproduce and pass on A. Thus, selection for A in the parental generation explains why X has A.

Nanay takes this conclusion to contradict Sober’s position, but does it really? Sober’s position is framed within an approach to explanation that says proper explanations have the form ‘w explains why p rather than q’. If you inquire about why I sold my antique typewriter to Clyde, I might ask for clarification. Are you asking why I sold rather than gave him the typewriter, why I sold the typewriter rather than the adding machine, or why I sold it to Clyde rather than to Alex? The original explanation request is ambiguous, and different explanations are appropriate depending on what was meant (Dretske Reference Dretske1972). Explananda lacking a contrastive clause (where one is not supplied by pragmatics) are considered ambiguous, and the only appropriate response to ‘Does w explain why p’? is ‘What do you mean by “why p?”’ In terms of A, B, and X, Sober’s explanandum is why X has A rather than X having B. This is not an analysis of ‘why X has A’ but only one thing that a person might mean by that.Footnote ² Since Nanay’s explanandum lacks a contrastive clause, it is unclear how his conclusion relates to Sober’s position. Stegmann (Reference Stegmann2010b) noted this ambiguity, and in response, Nanay (Reference Nanay2010) reformulated his previous argument by appending the clause ‘rather than X having B’ to the end of his conclusion. But the reformulated argument still does not address Sober’s position, as I will now argue.Footnote ³

Nanay’s reformed conclusion says that selection for A in the parental generation explains why X has A rather than X having B. This certainly appears to contradict Sober. But really it does not. That is because Nanay has an idiosyncratic understanding of what this statement means. To see what he means by it, and why it still does not address Sober’s position, we must examine an aspect of Nanay’s argument that I previously glossed over. Nanay does not argue for any explanation claims directly. Rather, he uses the relation of counterfactual dependence (defined below) as a proxy for explanation. His explicit (Reference Nanay2005) conclusion was that ‘X has A’ depends counterfactually on ‘selection favors A’. (From this, Nanay takes it to follow that selection explains why X has A.) Nanay’s reformulation (Reference Nanay2010) simply tacks ‘rather than X having B’ onto the end of his original explanandum, while continuing to treat the whole explanandum as a single proposition. This operation yields the claim: ‘X has A rather than X having B’ depends counterfactually on ‘selection favors A’.

Counterfactual dependence claims can be rephrased as counterfactual conditionals: ‘β depends counterfactually on α’ means that had α not occurred, then β would not have occurred. In shorthand, we write this: .Footnote ⁴ So Nanay’s reformulated (Reference Nanay2010) conclusion is

Notice that this counterfactual has the form

wherein ‘p rather than q’ is a single proposition (or event, depending on your metaphysics). What are the truth conditions of ‘p rather than q’? My best guess is (p & ¬q). (Nanay does not address the issue.) But since p and q are (in this case) mutually exclusive, (p & ¬q) is equivalent to p, making ‘¬(p rather than q)’ equivalent to ¬p. Equation (2) then reduces to

The problem is that equation (3) does not express what Sober (or anybody else) means by explanation claims of the form ‘w explains why p rather than q’. The whole point of adding contrastive clauses to explananda is to discriminate between relevantly different counterfactual outcomes that are lumped together by the uninformative catchall ‘¬p’. Insofar as ‘w explains why p rather than q’ expresses a counterfactual conditional, it is this one:

So to argue, contra Sober, that selection for A explains why X has A rather than X having B, one should argue for

But Nanay has instead argued for equation (1).

In prose, to argue against Sober is to argue that selection explains why ‘X has A’ occurred rather than ‘X has B’ occurring, not to argue (as Nanay appears to have) that selection explains why ‘X has A rather than X having B’ occurred. Sober’s contrastive clause is a specific counterfactual outcome. Nanay’s contrastive clause is a (superfluous) clarification of what actually happened, embedded within a fundamentally noncontrastive explanandum. Nanay’s conclusion may well be correct, but it does not contradict Sober’s position.

One might object to my analysis by saying that Nanay in fact intends counterfactual (5) and that he therefore really is addressing Sober’s genuinely contrastive explanandum. But this interpretation is no more favorable to Nanay’s argument. On this interpretation, Nanay would be claiming that had selection in the parental generation not favored trait A, then X would have trait B. But what difference could a different selection regime make in the parental generation? Regarding M, only that, contrary to what actually happened, M would not survive and reproduce. But this results in X not being born, not in X having B. A desperate critic of Sober could at this point survey the outcome in which M does not reproduce, choose at random from among the B-bearing organisms in the offspring generation, and call the chosen organism X. But this move generates a counterexample to Sober’s claim only by erasing the distinction (between being the same individual and being a different individual) on which Sober’s position is based.Footnote ⁵ In any case, that is not Nanay’s argument.