Is Natural Selection a Population-Level Causal Process? - Mechanism and Causality in Biology and Economics - page 159

individuals? Bouchard and Rosenberg ( 2004 ) argue that natural selection is causal
and at the level of individuals, as opposed to Matthen and Ariew ( 2002 ) and Walsh
et al. ( 2002 ), who argue that natural selection is at the population level and purely
statistical.
In addition to these two polar extreme positions, Millstein ( 2006 ) tries to steer a
middle course by arguing that natural selection is a population-level causal process.
On this view, (1) natural selection can be a cause of evolution, namely, it is able to
make a change in the frequency of traits among a population from one generation to
the next; (2) natural selection is by nature comparative: whenever it acts as a cause
of evolution, it impinges on comparative and thus population-level properties (e.g.,
variation in fitness, frequency of traits) rather than on individual-level properties
(e.g., fitness, traits). According to Millstein ( 2002 , 2005 , 2006 ), who endorses a
modified version of Beatty's ( 1984 ) account, “natural selection should be
characterized as a discriminate sampling process whereby physical differences
between organisms are causally relevant to differences in reproductive success.
Drift, by contrast, is an indiscriminate sampling process whereby physical
differences between organisms are causally irrelevant to differences in reproductive
success” ( 2006 , p. 640). Such a sampling process, discriminate or not, operates at
the population level.
I will make three points in this chapter: first, Millstein's account of natural
selection is incomplete, in the sense that nowhere in her account can one find a
place for cases of natural selection- of . Second, we should prefer Brandon's account
of natural selection and drift over Millstein's, on the grounds that her account fails
to meet a plausible requirement that Brandon's account succeeds in meeting:
namely, whenever natural selection and drift operate together, a change in the
strength of natural selection implies an inverse change in the strength of drift, and
vice versa. Third, the prospects of the view that natural selection is a population-
level causal process depend on a satisfactory solution to both the epiphenomenon
and the overdetermination problems. With the help of an analogy, I will show how
the two problems can be dealt with.
1 Why Is Millstein's Account of Natural Selection Incomplete?
Sober ( 1984 ) draws a contrast between selection- of and selection- for . To see why,
consider the following two propositions:
(1) There is selection- of trait T in population p if and only if T is fitter than not-T
in p.
(2) There is selection- for trait T in population p if and only if T is fitter than not-T
in p.