Adaptationists vs. Panglossians

Bryant (
12 Sep 1996 14:05:10 -0600

Paul and others have objected that anybody presenting an adaptationist
hypothesis must be a Panglossian/pan-adaptationist, meaning that they
must believe (as Gould/Lewontin asserted in 1979) that every single trait
we find in nature is of current, optimal utility.

I thought we might get past the name calling and actually discuss our
differences if somebody tried to sort out what we agree with first, and
then approached the nature of our disagreements.

So, here goes. I use the adaptationist program to try to elucidate the
adaptive significance (whether and what) of traits, mostly behavioral.

Here are some points I think most will agree with me about:

*Time lags mean that no trait is necessarily adaptive in the current
environment, even if the trait was originally designed by selection to
deal with ancestral fitness challenges.

*Traits need not be optimal to be functional. Adaptations need only
afford possessors with greater _relative_ fitness than competitors.

*There are multiple adaptive peaks for a given problem, and one should
pay attention to the raw materials ("phylogenetic constraints") a
selection has to work with when studying adaptations.

*Drift can lead to fixation of traits, but not to the formation of
complex traits like eyes, lungs, or brains.

*Selection can act on behavioral phenotype as well as structural phenotype.

*Selection acts on individuals, and indirectly on the genes for
traits--NOT groups of organisms per se.

...and here, I suspect, are the points others will object to. (?)

*Mutation accumulation and metabolic cost will, through deep time,
dispose of comlex traits which are no longer selectively retained (either
through direct selection on the trait in question, or by pleiotropic
selective retention.) Hence, phylogenetic constraint alone is not an
ultimate causation explanation for the presence of complex traits.

*Even if traits are no longer adaptive, adaptationist hypotheses about
their history can be tested by comparative method or simple prediction
testing in the following manner: If postpartum depression is an
investment-withdrawal adaptation for new mothers (ie, a baby
abandonment/killing adaptation), then obvious predicted correlations for
PPD include increased infanticidal ideation and social circumstances
which would have led to low offspring reproductive fitness in ancestral
environs--like unhealthy offspring, or low levels of social support.
These are testable predictions.

So let's get this sorted out, eh?