Godfrey-Smith, Peter - Conditions for Evolution by Natural Selection

03/14/2008

Journal of Philosophy, Vol 54 No 10 October 2007

A paper involving mathematically technical formulations of the components of evolution by natural selection, this paper has a modes conclusion. The paper aims to clarify the status of the various conditions given for evolution by natural selection (ENS). Some condition sets should be considered summaries, or descriptions of all the various ways change in a population can be considered ENS. Other condition sets give sufficient conditions for ENS, outlining a recipe ENS. These condition sets are idealizations, since they might not capture all of the various cases of ENS. Author's main point is to distinguish between these two and warn against not distinguishing them.

Generally, there are three main components to ENS as given in condition sets:
1. Phenotypic variation
2. Differential fitness
3. Heritability/heredity

Author brings up a number of cases that certain phrasings of ENS don't get right:
Case 1: Culling: no reproduction therefore no heredity, yet there is differential fitness because some organisms die more rapidly than others
Reply: this isn't a case of ENS, at least not yet (pg494)

Case 2: differential reproduction with no differential fitness: organisms reproduce the same amounts, but faster than others
Reply: the standard condition sets of ENS are idealized with discrete generations that don't overlap and reproduce at the same time (pg496)

Author discusses the problems with heredity and heritability mixed with fitness. It appears the two concepts should be kept separate, but differential fitness that is heritable should show up in future generations. Differential fitness itself doesn't suggest heritability-- a certain disposition needs to be present for a population to 'respond' to selection. (pg500) This is supposed to be heritability, and it is has a set of difficulties.
Case 3: 'Biased Inheritance': a heritable trait that improves fitness sometimes is heritable, sometimes not. Overall fitness remains unchanged after a new generation.
Case 4: A heritable trait that also improves fitness doesn't come from parent to child in the right way (pg502)
Replies: tie together fitness with heritability in the recipes of evolution, unfortunately connecting two concepts that we wanted to keep separate.

Author discusses arcane equations that are meant to separate heritability from fitness. (pg506-509) [If anyone thinks I'm going to commit the hours need to understand this, forget it.]

The final problems author brings up is heritability by accident and genetic drift. (pg510) These can be fixed by saying that genuine cases of ENS has a causal link between the phenotypic trait's fitness value and the inheritance of that trait. The problem with 'causal link' is that it is vague and not easily expressed in an equation. Author examines an alternate equation and finds it inadequate.