I said a while ago that the announcement of the research papers by Bruce Lahn et. al. would be sure to set all the same old intellectually third-rate kooks howling with (entirely unjustified) delight, and events have since proven* this assertion to be correct, but having now read the two papers in question myself in their entirety, I think it's about time I laid out in some detail precisely why any invidious inferences about the differing capabilities of different "races" are not only unjustified, but are actually a mark of the near-total ignorance and utter stupidity of those who make them.
*Note the blanket statement that it is an "incongenial fact that black intelligence is lower" by a certain loudmouth who takes umbrage at being called on his promotion of racist nonsense as "fact" ...
Before getting down to business, I'd like to start off with a brief discussion on some rudiments of population genetics and how it relates to the work done by actual biologists who spend their time sussing out the often extremely convoluted details of how genes are translated into all those proteins which govern how organisms function. A common misconception many people have of how things work is that there are one or a few genes "for" this trait or that one, the presence of absence of which determines whether an organism will or will not manifest a particular trait; a related misconception is that a gene which influences one particular trait will have an effect on that trait only or even primarily, and that therefore any allelic differences we notice must pertain to whatever phenotypical differences we think we already know they're responsible for. If both of these things were true, population genetics would be a much more tractable subject, but unfortunately, neither is even close to accurately describing reality.
The truth about the way genes work is that other than in the case of simple mendelian traits of the kind which geneticists have had most success studying during the last century, it is almost unheard of in genetics research to discover single genes which are able to account for more than a minute fraction of variation in any given trait; furthermore, the very reason why so many traits in any given population happen to abide by a normal distribution is because to the extent said traits aren't determined by environmental differences, they must be under the influence of very many genes, and the smoother the curve, the greater the number of underlying genes must be - this is nothing more than a generalization of the binomial theorem with which we are all familiar. In the case of the human brain, we know that about half of all genes - or some 15,000 of the total - are expressed in the brain during the course of development, and there is no a priori reason to believe that variations in any of them, or even in whatever regulatory regions might govern their behavior, will have no impact on how said organ functions. In light of all the foregoing, anybody who thinks we will find 20 or even 50 genes which rigorously account for, say, 90% of all "genetic" variation in "IQ" or whatever nebulous measure of intellect we decide upon is a fool at best.
Having discussed the "one trait -> one or a few genes" fallacy, let me turn now to its counterpart, namely the belief that a gene's only or main function must be to determine any single particular trait (hence all the idiotic talk of genes "for" IQ, homosexuality, jazz improvisation, etc.). Genes code for proteins, not explicit traits, and it is part of the blind genius of evolution that individual proteins are co-opted to serve multiple roles all the time, so much so that there is even a technical name for the phenomenon, to whit, pleiotropy. Underlying all the ignorant chatter about how the ASPM and microcephalin variants written about by Dr. Lahn must be genes "for" cognition is the assumption that because faults in both genes have been implicated in brain disorders, and because differences exist between humans and chimps in both genes, then "the" purpose of the existence of these new variants has to be to code for "IQ" or some such thing: but the reality is that with an organ as complex as the human brain, there are very many ways for a gene malfunction to lead to devastating consequences, often through causal chains nobody would have guessed beforehand.
To illustrate how things aren't always what they seem, and why it is important to understand the underlying biochemistry before jumping to conclusions, let us consider phenylketonuria: this is a genetic disorder which is characterised by mental retardation, and an uninformed observer might easily jump to the conclusion that this means defects in the gene implicated in it must result in some crucial feature of the brain being wired wrongly, leading to lower IQ scores. And yet, as we now know, the depressed IQ which accompanies phenylketonuria has nothing to do with brain wiring, but is the result of an inability of the sufferers' metabolic systems to produce sufficient levels of phenylalanine hydroxylase: in the presence of a diet which makes up for this deficiency, the IQ scores of the genes carriers turn out to be normal, and what might have been ascribed to an "IQ gene" is in fact just one particularly visible manifestation of an enzyme deficiency which has several other side-effects.
The misconceptions about how genes work extend beyond these two errors, however, and there's a third issue I'd like to discuss which goes by the technical name epistasis. The basic idea behind this term is that if genes at more than one locus govern the expression of a trait, they need not do so in a straightforward, additive fashion like so many dollars which can be netted against each other - even if a gene happens to code for a particular phenotype, it could well be that the trait will not be expressed in the slightest if the allele at some other gene locus isn't the right one. To give an example, suppose there are two genes which govern hair color in mice, with gene A coding for an enzyme which produces the melanin which makes hair black, and gene B coding for another enzyme which modifies the product of gene A so that the resulting hairs are grey (agouti): if a mouse happens to be carrying two broken copies of A, then that mouse is destined to an albino, regardless of how well its copies of B might function, as the enzymes which the products of B alter simply won't be produced. The point here is that an error or variation at some step in a multistep biochemical pathway can suffice to alter the rest of the successive steps in such a way that simplistic totalling of the presence or absence of alleles "for" this or that leads to completely wrong results: genetic background matters, and even if a gene can be shown to affect the expression of a trait in a particular population, there's no reason to believe it will also do so in a different population, even if we are able to adequately control environmental variation (this isn't merely theoretical - see, for instance, this paper, which finds that APoE, although repeatedly implicated as a risk factor for Alzheimer's disease amongst white Americans, is not associated with elevated risk in either African-Americans or Hispanics).
I've discussed three common errors common amongst idiots giving to misusing the abstracts of papers they haven't fully understood to grind their nasty little axes, but I don't want to leave the impression that this is in any way comprehensive. The fact is that I could write a small textbook covering such ground, but I'd much rather turn now to applying some of the above to a rudimentary model which I've selected to weight the argument heavily in favor of those who want to argue that they now have the genetic smoking gun pointing at the "incongenial fact" [sic] of "lower black intelligence"; let's see how their reasoning fares even under the best of circumstances.
A Toy Model of IQ Variation
For the purpose of argument, I'm going to go along with the entirely unfounded assumption that the new ASPM and Microcephalin variants Bruce Lahn's work indicates as undergoing positive selection are in fact "IQ" genes; furthermore, I'm not only going to abide by the implicit assumption that any such genes must be coding for more "IQ" rather than less* (since when have peasant farmers needed to be smarter than hunter-gatherers?), but also that there are only 10 "IQ" genes in total, accounting for all IQ variation which can be ascribed to genetics. I'm going to make the assumption that epistasis is totally unimportant, in order to make the model tractable (this biases things in favor of the IQ cranks because the more complex the genetic interactions which govern "IQ", the less likely it is that they're going to isolate the underlying genes any time soon), and in addition, I'm going to assume that environmental factors have only a marginal effect on IQ, leading to at most 5 points either way in the best and worst circumstances. Finally, I'll assume that each one of 10 loci is responsible for exactly 10% of the total variation (reasonable, as otherwise one would have to imagine the existence of at least one as-yet unmapped "IQ" gene which accounts for an even greater proportion of variation) - say 10 points each - and that there are only 2 alleles at every locus: if a person has two copies of the "good" allele, he gets 20 extra IQ points, if he has just one copy, we give him 10 IQ points, and if he only has two copies of the "bad" gene, he gets no points whatsoever.
Under the model described above, a person who has 1 copy of each good gene will have an IQ between 95 and 105 depending on the quality of his or her environment, a person who has 14 good alleles and 6 bad ones will fall between 135 and 145, while any genetically lucky person who has 2 copies of all 10 good genes will have an IQ between 195 and 205. Given all the above, as well as the existence of two genetically differentiated subpopulations P_A and P_B, what can we logically conclude about the relative intellectual capabilities of the typical members of both groups based on the news that IQ genes 1 and 2 (henceforth IQ_1 and IQ_2) have a frequency of 70% and %80 in P_A, while the same genes occur at a frequency of 40% and 50% in P_B?
If you answered "Nothing", hand yourself a prize: the fact is that we haven't been told anything at all about the frequencies of all the other IQ genes, and for all we know there might be frequency differences in those that suffice to completely swamp any effects which arise from differences in IQ_1 and IQ_2: if IQ_3 = 20%, IQ_4 = 30% and IQ_5 = 40% in P_A while IQ_3 = 65%, IQ_4 = 55% and IQ_5 = 60% in P_B, and if the rest of the IQ_N occur at the same frequency in both groups, then P_B will be the one whose members have the higher average IQ, not P_A, and the fact is that even under the extremely simplistic model we have here, there are an astronomical number of ways in which one can obtain each one of any such outcome even with IQ_1 and IQ_2 distributed as mentioned: if we disregard which particular IQ weighting gets assigned to what locus, this is easy to demonstrate using the theory of integer partitions - taking the distinct orderings into account just inflates the answer by a multiple of n!, where n = 8 if we take IQ_1 and IQ_2 as given. The only circumstance under which we would be safe in assuming that the frequency differences in IQ_1 and IQ_2 implied the lower intellectual capability of P_B would be if we knew that the two groups did not differ at IQ_3, IQ_4, ..., IQ_N, but this assumption is entirely unwarranted in light of the fact that we already know that the two groups have differing allele frequencies at very many loci, a point of which "race realist" cranks are usually eager to remind us; making it even more dubious would be any research report which indicated that the "good" alleles at IQ_1 and IQ_2 were under strong selective pressure in P_A, as an identity of the rest of the "IQ" allele frequencies would leave us with the conundrum of explaining why their benefits should be less evident for population P_B.
Of course, in all I've said so far I've assumed that the the frequency differences were on the order of 30% for each of IQ_1 and IQ_2, but what if the differences turn out to be just 3% each - or, mathematically equivalently, there turn out to be 10 times as many IQ loci, or 100 in total, and each one is responsible for just 1% of IQ variation? This is where our assumptions about the impact of environmental differences come in, as it is obvious that even the extremely restricted influence we attribute to environment would be more than enough to completely swamp the genetic differences between P_A and P_B, so that even though one might think P_A ought to do better on average in IQ tests (an assumption which is itself completely unfounded, as I've explained above), the members of P_B could actually end up with an average IQ as much as 4 points higher than those of P_A! The only way we'd be able to sort out which group had greater intellectual potential would be to place the members of both in identical (and ideally uniform) environments, but this too would only be possible if, say, there weren't any phenotypical differences between the two groups which led us to take it as an "incongenial fact" that the members of P_A were of lower intelligence by nature (with only "politically correct" cowards supposedly stopping we "few", "brave" souls from frankly discussing said "fact") and therefore treating them differently, expecting and demanding less of the members of P_A, and providing them with inferior environments while continually, vocally putting down their capacities in the name of "scientific discussion" [sic] ...
I've said a great deal so far about how population genetics works in theory, but in the next half of this article, I'll be discussing how everything I've written about here applies to research results disclosed here and here. If at the end of that piece any of you are still in doubt as to the laughable pseudo-reasoning behind the puffery coming out of the mouths of certain "race realist" fools, then nothing under heaven and earth will ever convince you.