The simplest way of compromising the realism implicit in our framework is to adopt a non-realist view of causation. On such a view, statements of the form 'X causally affects Y' do not report objective facts about the world at all. Anyone holding such a view would deny that it is an objective matter whether a given character causally affects fitness. Similarly, if one held that causation is nothing but correlation, one would deny the reality of the distinction between character-fitness covariance that does and does not reflect direct selection. Given such metaphysical views, the levels-of-selection question can at most be a question about the heuristic utility of one description over another.
Non-realist views of causation, and a refusal to accept the causation/correlation distinction, have a long precedent among empiricist philosophers and scientists, but they enjoy little popularity today.16 Few if any writers on the levels of selection explicitly endorse non-realism about causation. Nonetheless, it is sometimes suggested that to talk about the 'true' levels of selection involves a mistaken reification of causal relations, or ignores the fact that causal chains can be chopped up in multiple ways. For example, Kitcher (2004) argues that it makes no sense to ask about 'the real locus of causation in selection processes' (p. 89, emphasis in original). He writes: 'one can tell all the facts about how genotype and phenotype frequencies change across the generations—including the causal explanations of the changes—without any commitment to a definite level at which selection acts' (ibid. p. 89). Far from being a natural default position, realism about the levels of selection is the result of inflationary metaphysics, according to Kitcher.
This argument seems to me unconvincing. To ask about the 'real locus of causation' is to ask about the hierarchical level or levels at which character differences cause fitness differences. So presuming we are realists about questions of the form 'does X causally affect Y?', Kitcher's assertion seems wrong. If the explanation for why a given particle character has changed in frequency is that it causally affects the fitness of particles, for example, then selection has acted at the particle level; if not, then not. Given this conception of the levels-of-selection question, it is clearly not possible to give causal explanations of evolutionary changes 'without any commitment to a definite level at which selection acts'.
16 For example, Hume, Mach, Pearson, and Russell all defended such views. See Price (2006) for a recent, sophisticated version of non-realism about causation.
For different causal explanations embody different such commitments, and thus conflict.
However, this may be slightly too quick. From the context of the above quotation, Kitcher is alluding to the version of pluralism associated with 'genic selectionism'. As usually formulated, this view does not say that the level of selection is a wholly indeterminate matter, but rather than any selection process, at whatever level, can also be described as a case of genic selection (Dawkins 1982; Buss 1987; Sterelny and Kitcher 1988). So some levels-of-selection disputes do have objectively correct answers; it is only those disputes in which one of competing levels is the genic level that receive a pluralistic resolution. This view is examined in Chapter 5; properly understood, it turns out to be compatible with the realism implicit in our framework.
One might think that a causality-based argument for pluralism can be extracted from the concept of a cross-level by-product. Since selection at one level can have effects that filter up or down the hierarchy, perhaps the notion of the real level(s) of selection, or the real 'locus of causation' loses its grip? But this argument is incorrect, for it conflates direct and indirect selection. The filtering of effects from one level to another does not imply that there is no fact about whether any given character-fitness covariance is the result of a direct causal link. Cross-level by-products make it hard to discover the causal basis of evolutionary change, but do not undermine the idea that there is one true causal story to be told.
Finally, another possible causality-based route to pluralism must be eliminated. Recall the supervenience argument, which says that any higher-level character-fitness covariance must be spurious, because collective properties supervene on particle ones, so the causal arrow is always at the lower level. One could respond to this argument in a pluralistic way, that is, by saying that there is no objective fact about what level the causal arrow is at, or that causal arrows at both levels can peacefully coexist. In the philosophy of mind, these responses to the supervenience argument have sometimes been touted.17 If they work, they presumably must apply in biology too. But the resulting pluralism would not be pluralism about the levels of selection. For as stressed previously, the lower-level causal story need not be a selective story. So even if the choice
17 The idea that causal arrows at both levels can peacefully coexist is sometimes called 'causal overdetermination'. Loewer (2002) defends this idea as a way of responding to the supervenience argument developed by Kim (1998), in the context of the mind/body problem.
between lower- and higher-level causal stories is conventional, this does not make the level of selection a matter of convention.
Was this article helpful?