Is Natural Selection a Population-Level Causal Process? - Mechanism and Causality in Biology and Economics - page 172

If the preceding analysis is correct, then we have a ready answer to the epiphe-
nomenon problem. Suppose someone raises the epiphenomenon problem in the
following way:
One may wonder: to explain why the relative frequency in population size
between group N and I changes in accordance with a yearly increasing ratio, is it
really necessary to invoke the tyrant's law? It seems that annual genocides,
committed by the imperial army, suffice to explain why the relative frequency
changes in a yearly increasing way. If one has each and every causal story about
how, each year, each officer of the imperial army kills his victims and how
survivors escape death, then such many causal stories, in addition to a little
statistics, suffice to show why the relative frequency changes in a yearly increasing
way. So the tyrant's law is causally redundant: it has no effect at all. The change in
the relative frequency is not caused by the law. Rather, it is a summation result due
to many causes that one can only find in each gory and detailed causal story.
The epiphenomenon problem, in this case, fails to cast doubt on the causal efficacy
of the tyrant's law. Few, if any, would deny that the tyrant's law is part of the cause
of each annual change in the relative population size between group N and I .
On the other hand, though the many detailed causal stories, in addition to a little
statistics, suffice to sum up the annual relative frequency and calculate the way the
relative frequency changes, they do not suffice to explain causally why it changes in
the way that it does. A complete explanation of why the relative frequency changes
in a yearly increasing way must invoke the tyrant's law. The distinction between
“suffice to sum up” and “suffice to explain causally” permits an answer to the
overdetermination problem: the annual change in the relative frequency is
not overdetermined, for the many detailed causal stories, along with statistics,
do not suffice to explain causally why the relative frequency changes in a yearly
increasing way.
As mentioned above, the view that natural selection is a population-level cause
of the change in the relative frequency among a population from one generation to
the next faces a dilemma: it gets into trouble either with the epiphenomenon
problem or with the overdetermination problem. However, with respect to causal
efficacy, if the tyrant's law can serve as an analogy for natural selection, then a way
out of the dilemma is accessible to natural selection. So my argument, if successful,
would be good news for the population-level cause view of natural selection. More
specifically, it would be an argument in support of Millstein's view that natural
selection is a population-level causal process in the sense that it acts as a cause of
evolution and that it impinges on comparative and thus population-level
properties. 4 The fact that natural selection operates at the population level provides
no support for it being a shadow process. Despite appearances, natural selection is
part of the cause of the change in the relative frequency.
4 I leave open an interesting question as to whether natural selection, operating at the population
level, is as causally robust as the many events occurring at the individual level. Stuart Glennan
( 2009 ) distinguishes causal relevance and causal productivity and argues that at the population
level of natural selection, there can be causal relevance without causal production. In response,
Millstein ( 2013 ) argues that there is causal production at the population level of the natural
selection process.