race and IQ
Jim Chisholm (chisholm@UNIWA.UWA.EDU.AU)
Thu, 27 Oct 1994 08:47:45 +0900
>Jim Chisholm here, with a longish posting on Rushton's views on race and IQ
that I posted on HBES-L several days ago. Most of what follows was published
originally in Current Anthropology (vol 35, Feb. 1994) as a response to
Rushton's comments on my CA artile "Death, Hope, and Sex: Life History
Theory and the Development of Reproductive Strategies" (34(1):1-24).
However, given the ominous implications of Rushton's errors and
misconceptions, I thought it worth repeating on HBES - and now here.
> The concept of race has no value in the analysis of reproductive
strategies. This is because the definition of "race" is ultimately arbitrary
and subjective, and there is no a priori basis for deciding which traits to
use in any classification system. Consequently, one can pick and choose
classificatory traits at will until the groups that result suit one's
particular purpose - which history shows is often exclusionary, or worse.
This is why estimates of the number of different "races" varies so widely,
why there is no such thing as a discrete "race," and why there is usually
more variation in a trait within "races" than between them (see, for
example, Gould's "The Mismeasure of Man"). Ruston himself acknowledged these
problems inhis comment on my CA paper when he said that "no claim is made...
that races are discrete groups" - and noted as example that military rank (a
manifestation of class difference?) is related to brain size. But if "races"
are not very discrete groups, then what are we to make of Rushton's
interpretations of "race" differences in reproductive strategy? Would the
"racial matrix" in brain size, IQ, etc. that Rushton imagines he has found
look the same if he used a different system for racial classification?
> Even if we overlook for the moment Ruston's nebulous concept of
"race," he still hasto explain why he ignores the possibility that
significant group differences in a trait (say, cranial capacity) may be
observed because of simple group differences in the distribution of
environmental or other factors that help give rise to individual differences
in the expression of the trait. When these factors are not known or
controlled for there may be a temptation to assume that unexplained residual
variation between groups constitutes evidence for genetic differences. This
logical mistake gives genetic factors the status of the null hypothesis -
which confuses the UNTESTED genetic "explanation" with the far weaker claim
of "not yer disproved." In his thoughtful and extensive review (on this
network, May 25, 1993) of Rushton's work and of his primary sources, Kim
Hill concluded that that in addition to serious measurement problems, there
have been woefully inadequate attempts to control for environmental factors
that may covary with with race differences in cranial capacity or IQ. Hill
emphasizes in particular that cranial capacity is known to vary with
latitude (climate?) - within racial groups as well as between them - and how
impoverished are the arguments for anything like a direct link between brain
size and IQ.
> Hill also points out that Rushton is inconsistent in his arguments
about how cranial capacity relates to IQ. Rushton believes that "races" with
big brains have higher IQs BECAUSE they have big brains. Even he realzies,
however, that this unconvincing "correlation = causation" argument won't
work with sex differences. The problem is that while males tend to have
larger brains than females, Rushton's data show no sex difference in IQ
scores. How does he resolve this inconsistency? By speculating that despite
their smaller brains, women may be as intelligent as men because their
brains are more densely packed with neurons. Thus, when it comes to brain
size and IQ, as Rushton himself argues, size isn't everything - which raises
the question, if details of neuroanatomy matter for sex differences in IQ,
why not for the putative race differences as well?
> And even if we were to grant that brain size matters, what makes
Rushton think that the "race" differences in brain size (or any other trait)
that he reports are worth explaining? We usually want to explain something
that seems to have "meaning" or "significance." Not only does Rushton assume
that the quality that IQ tests measure today had adaptive significance in
our evolutionary past, he also confuses statistical significance with
biological significance. It is a statistical fact that with very large
sample sizes even very small absolute differences between groups can be
statistically significant. Despite their statistical significance, however,
these same differences can be entirely vapid, without an iota of biological
or evolutionary meaning or significance. Thus, in one of Rushton's samples
(n = 6,325), after adjusting for some covariates, the "race" with the
highest average cranial capacity (1,416 cc) had only 57cc greater capacity
than the "race" with the lowest (1,359cc). Despite the statistical
significance of this very small absolute difference, there is no evidence
for its biological or evolutionary significancwe - especially since the
6,325 people in his sample do not constitute any sort of demographically,
ecologically, or evolutionarily meaningful population. Rushton claims that
the rank ordering of the "races" in brain size is the same as their rank
ordering in IQ and that IQ has biological or evolutionary significance.
Quite aside from the fact that his correlation established no causation, and
evn if IQ scortes measured something accurately and fairly, the same in all
groups, Rushton fails utterly to show how a difference of 57cc could be (or
once was) related to any measure of fitness in the particular environments
in which people with particular IQ scores live or lived.
> Even when genetic factors are implicated in group differences,
their effect may vary both within and between groups because they can only
exert their influence through complec epigenetic pathways that do not
exclude signifiant, potentially adaptive, local environmental input during
development. Rushton seems unaware of the important distinction between
obligate and facultative adaptations, and he ignores all of the
developmental factors known to affect IQ (to say nothing of age at first
intercourse, marital stability, mental health, "rule following," and other
"traits" he imagines are associated with race differences in reproductive
strategies). Evolution is as much about environments as it is genes, and to
show that his is not a deterministic model (as he apparently wants us to
believe) he has to show the gene x environment interactive processes that
produce the traits he is studying. His claim that he is working with a model
that is "50% genetic plus 50% environmental" is without meaning. Such a
model is naive, futile, and in no sense is it a model of development, of
gene x environment interaction. (see Susan Oyama's "The Ontogeny of
Information" for a good critique of such naive moels).
> In his comment on my paper Rushton noted, correctly, that I am not
enthusiastic about r/K theory, but he was wrong to imply that my "avoidance"
(as he put it) had "little justification." Since my paper was not about r/K
theory, I did not dwell on its problems. Such problems nonetheless exist,
and Rushton's claim that his "racial pattern corresponds well the r/K scale"
is nonsensical. Beyond the problems noted above with is "racial pattern,"
his faith in "the r/K scale" is misplaced. Before rushing in even further
whether life history theoreticians fear to tread, he should examine closely
some oftheir work - especially Charnov's "Life History Invariants," Roff's
"The Evolution of Life Histories: Theory and Analysis," and Stearns' "The
Evolution of Life Histories." I have found the latter most useful. It
abundantly justifies questions about existing r/K theory.Among a number of
more technical problems noted by Stearns are these straightforward
shortcomings: r/K theory fails for about 50% of the species for which
adequate data exist, it has failed 75% of artifical-selection experiments
infruit flies, and it confuses the statistical description of population
processes with the explanation of these processes - i.e., "selection
pressures that act on individual organisms" (Stearns, p. 206). More
generally, suggests Stearns, r/K theory fails to specify the mechanisms
whereby habitats 9i.e., r- and K-selecting environments) affect life
histories. The heart of his book, to me, is the suggestion that "one
candidate is the impact of habitats on age- and size-specific mortality
schedules. Thus we seek to understand not habitat > life history but habitat
> mortality regime > life history" (Stearns p 208).
> Finally, in his comment on my CA article Rushton claimed that I
overlooked certain findings from behavior genetics that constradicted my
model. Since in that article I was describing my model, not giving an
exhaustive account of all other conceivable models, this criicism is
irrelevant. But if in making it Rushton meant to suggest that data from
behavior genetic research are somehow inhgerently theratening to
developmental, "environmentalistic" models (in his words), he is simply
wrong, and fall victim again to his inadequate model of development and gene
x environment interaction.