Re: Rushton

Wed, 19 Oct 1994 12:27:00 PDT

Wilk refers to Jared Diamond:

" Jared Diamond's article is especially strong, arguing that
the genetic data, fed into numerical taxonomy programs, lumps together
ALL non-Africans with some Africans, and distinguishes several taxonomic
groups within Africa that have higher taxonomic status than any of the
differences between non-Africans. In other words, if you are looking for
races, there are three or four African groups, then one other that
includes some Africans and everyone else."

Previously on this list I have discussed some of the problems with the
concept of "race" (emic) being treated as if it is a valid etic concept.
Unfortunately, the kind of analysis done by Jared Diamond (doing a numeric
taxonomy analysis of a large number of genetic measures) only establishes the
old dictum: "garbage in, garbage out" and doesn't really show very much.

While the EMIC notion of race would seem to suggest that ANY measure ought to
distinguish race A from race B, any ETIC notion of race would reject that
assumption from the beginning. From an etic viewpoint one could ask: are
allele frequencies for human breeding populations everywhere the same, or is
there variation in allele frequencies across breeding populations? The
answer is clearly: there is variation. One could ask: if breeding population
A differs in its allele frequency for trait X from breeding population B,
does it follow that population A differs from population B for ALL traits?
The answer in this case is in the negative--population A might agree with
population B for some traits, but not for others.

This line of argument could be expanded to include historical
questions. Do all current breeding populations genetically descended
from a common ancestral breeding population have identical allele
frequencies? Answer: no. Drift and selection will lead to differention
among breeding poulations (whereas migration between breeding
populations tends to reduce differentiation). Contrariwise, are all such
current breeding populations so distinct that there common ancestry has been
obliterated, as it were? The answer here is essentially no: Current allele
frequencies do reflect past genetic connections, though the degree of current
similarity may vary from one trait to another (i.e., a so called neutral
trait will lead to differentiation only via drift and current frequency
values will reflect past frequency values; a trait under selection will
change in accordance with selective pressures so that the current frequencies
will relate poorly to original values).

Once the role of selective factors is recognized, the argument can be
expanded to include selection. Will breeding populations subjected to the
same selection be similar with respect to allele frequencies affected by
those selective pressures? Answer: yes.

We can include geography. Will breeding populations in the same geographic
area tend to be similar? Yes and no. If the selective pressure is
essentially the same over the geographic region, then for traits affected by
that selective pressure there will be similarity. For traits where the
selective pressure has variation on a scale smaller than the geographic
region then for such traits there will be heterogeneity.

Now we can, I suggest, begin to talk about race etically as roughly
identifying a set of breeding populations that show similarity for some set
of traits, yet differ from other such breeding populations for the same
traits, and where the size of the sets is relatively large. (The latter
condition prevents a race from collapsing into a few breeding populations and
is not precise. It seems that race-etic (as say used by ecologists for
animal populations) refers to a set of breeding populations that fit the
similarity criterion and also represent at least 1/10(?) [1/10 is a
guesstimate on my part--I don't know if anyone has actually looked at animal
races in terms of what proportion they represent of the whole species] (1/10
is used solely for the sake of illustration).

Now with this notion of etic-race the question becomes: Are there (large)
sets of breeding populations of homo sapiens that show similarity with
respect to a trait or suite of traits and differ from other breeding
populations for the same trait. Note that for such a set of breeding
populations, the set of breeding populations may be highly variable with
regard to allele frequencies for other traits, and some of the breeding
populations may show high degree of similarity to other breeding populations
not in this set for yet other traits.

Such etic races are not "fixed" in that interbreeding will lead to their
demise (in that interbreeding will lead to all affected breeding populations
to have similar allele frequencies); hence the presence of etic races
requires two conditions: (1) common selective pressures for the breeding
populations in question and (2) isolation from other breeding populations.
Futher, the set of breeding populations are a race ONLY WITH REGARD TO THE
TRAIT(S) IN QUESTION, and the grouping together of that set of breeding
populations as an etic race may be totally meaningless for other traits.

Here we have the nub, I think, of what is so problematic with using emic
concepts of race: the assumption is usually made that the identification of
person X with race A somehow informs us about not only the trait(s) used for
the classification, but other traits as well. Let us see how this plays out
with two traits: IQ and Schizophrenia.

Mike Lieber wrote a long piece about the problems of connecting alleles with
the trait as it is observed. What he says is correct with regard to teh
complexity, but misleading with regard to linking genetics with traits. Were
his argument applied rigorously, we would conclude (much like the cigaretter
manufacturers did with their denial that there is a causal linkage between
smoking and lung cancer) that we can say nothing about most traits and
possible genetic underpinnings. Now studies of schizophrenia have not (I
believe--but I may be wrong) not carried with them racism arguments.
Consequently, studies of schizophrenia look for people who have the trait
first, and then their characteristics (e.g., relatedness) second. I.e.,
looking at purported races (even if etically defined as above) is not very
useful as the genetics involved in schizophrenia are quite likely to have
little to do with the traits used for the etic race definitions (assuming
such definitions are made). Even if it should be the case that frequency of
schizophrenia differs from one etic race to another, this is not very
informative about either the underlying genetics or about how to medically
deal with person X who has schizophrenia.

COntrast this with IQ studies which first consider not even etic definitions
of race, but emic definitions and then ask about distributions of the trait
with respect to those emic definitions. Is the reason for so doing to better
understand the genetics involved in brain functioning and how there may be
variation in brain functioning (or what ever might be the supposed research
question) or is it better seen as an attempt to give an invalid definition
(an emic definition taken as if it is equally valid as an etic definition
when demonstrably such is not the case) the appearance of validity?

D. Read