The Final Taboo: Race Differences in Ability
Jon Entine's book, Taboo, is a direct, powerful, and, effective challenge to an orthodoxy that has developed over the last 50 years or so for very understandable, though not valid, reasons. This orthodoxy says, "Yes, racial variation does exist in our species, but it is quantitatively and functionally inconsequential; furthermore, there is every reason to believe that it couldn't be otherwise." Entine provides good evidence and solid arguments that should convince the reader that the orthodoxy is wrong.
My goal here is to put the issues Entine raises in a broader quantitative context from more of an evolutionary perspective. The evidence that race is a meaningful biological concept, not a mere social construction, and that racial differences in abilities and behaviors have been important over the course of human evolution, has never been stronger. If anything, Entine himself is somewhat wary of the final taboo
Why is there "every reason to believe" that racial variation couldn't be anything other than quantitatively and functionally inconsequential? Why should social scientists feel comfortable penning statements such as "Bipedalism (or intellectual ability) is such a critical aspect of the human adaptation that one would not expect to see great differences in bipedal (or cognitive) performances at either the individual-to-individual level, or between populations"? Well, ideological desires aside, they would argue that the fuel for natural selection is heritable and a functionally significant variation (which is obviously true), and that therefore adaptation will use up some portion of that variation, decreasing the amount left for us to observe.
Further, the more important the feature (bipedalism, a large brain, language ability), the more the genetic variation underlying it would be used up, and the less variation would be seen in it today. If this argument holds at the individual level, and individual variation is much greater than group variation (which it is), then gene-based group variation with respect to those features critical to our adaptation simply cannot be worth worrying about when trying to explain such variation as we do observe in, say, the quality of bipedalism (running performance) or the level of mental ability (IQ scores). All of this sounds reasonable, and the basic theory underlying it has indeed for decades been a standard part of the thinking of evolutionary biologists, and a fixture in evolutionary biology textbooks.
The Race is to the Swiftest and the Swiftest are the Kalenjin
On the other hand, let's analyze some real data that should make the case for racial differences in sports performance obvious to all. The venue is the World Cross-Country Championships in Turin, Italy in 1997. The competition included 275 runners from 60 countries. They took part in a test of human bipedal ability about as similar to what our ancestors faced in the Environment of Evolutionary Adaptation (EEA) as one could imagine.
The results? For the 12th consecutive year, the team title went to Kenya. Kenyans took two of the three medals, and five of the first 10 places, finishing 1st, 3rd, 4th, 6th, 7th, 17th, 19th, 24th, 28th, and 47th (no country being allowed more than 10 entrants). The winning time was 35 minutes, 11 seconds. The 10 Kenyans averaged 36:04, which if obtained by a single runner would have given him 12th place overall. And this doesn't even begin to tell the whole story on the magnitude of group differences, as most of the Kenyan runners come from one tribe, the Kalenjin, who number only 3 million and make up about 10% of Kenya's population.
Over those 12 years of Kenyan domination, if the Kalenjin runners had been a team of their own, they would have won the team title 8 times, and they have taken 18 of the 36 medals. In other words, the Kalenjin have been "over-represented" among the medalists by a factor of about 1700. (Here's the math: 3 million Kalenjin versus 5 billion of the rest of us humans. The Kalenjin took 18 of 36 possible medals; and the entire rest of the world together got the other 18. Their "over-representation" is then as 5 billion is to 3 million, or a factor of about 1700.)
But what about the received dogma that bipedalism is such a critical aspect of the human adaptation that one would not expect to see great differences either from individual to individual, or between populations? Well, it seems to me that a factor of 1700 can legitimately be seen as a "great difference."
We can crunch the numbers even more finely to compute the difference in running ability between the average Kalenjin and the average for the rest of the world. The most useful metric here is the standard deviation, where, for a normal (i.e., bell curve type) distribution, the area that falls between one standard deviation (SD) on each side of the mean includes 68.26% of the sample; two SDs, 95.6%; three, 99.3%; and so on.
So how many standard deviations is the Kalenjin mean beyond the mean of the rest of the world? How do you get to a ratio of 1700? Assume that the cross-country medalists are ultimately selected from, in effect, 20 million males of the appropriate age all around the world. That means that the winner is about 5.3 SD above the world mean; that is, at +5.3 SD on the bell curve we are looking at the best individual among 20 million competitors. But only about 12,000 of those 20 million (one in 1700) are Kalenjin, and the best among 12,000 is only 3.8 SDs above the Kalenjin mean. The Kalenjin mean then would be about 1.5 SD (5.3 SDs - 3.8 SDs = 1.5 SDs) beyond that of the rest of the world. That is an enormous difference for any measure of human performance. Put that figure in the more familiar IQ context and the Kalenjin have an RQ (Running Quotient) of about 122 compared to a world mean of 100.
And just how much is 1 SD of RQ worth in cross-country finishing time? If the best runners are 5.3 SD beyond the world mean (a probability of 1 in 20 million on the bell curve), then 1 SD slower equals 4.3 SD above the world mean, which is about 150-fold more frequent. As the finishing times in the 1997 race were: 1st = 35:11; 150th = 38:44, 1 SD was worth about 3.5 minutes. So a properly conditioned male with average genetic potential and of an appropriate age (that is, one at the world mean), would require 35:11 plus 3.5 minutes per SD x 5.3 SDs, or about 55 minutes to complete the courseóa full 20 minutes behind the winner!
Just How Big Are Race Differences?
What about other, less contentious realms? For example, what is the degree of racial variation with respect to morphology in humans as compared to other taxa? I answered that question five years ago in Skeptic (3:3:87) as follows: "racial morphological distances within our species are, on the average, about equal to the distances among species within other genera of mammals, as, for example, between pygmy and common chimpanzees. I am not aware of any other mammalian species where the constituent races are as strongly marked as they are in ours."
To which I should have added "except those few races heavily modified by recent human selection; in particular, dogs." I remain as unaware as I was then of any other mammalian species that exhibits such extensive variation, and I have come to think that there really aren't any. Since I wrote that sentence in 1995 I have addressed the matter of "racial morphological distances" in a more quantitative manner, and I present some of those data and conclusions here.
Previous investigators have not bothered to quantify these issues, which allows ideology to prevail. To avoid this problem I present some comparisons below that apply the same metric to various cranial and facial measurements compiled by world recognized authorities in their respective fields. The metrics designed here represent a novel methodological approach for studying this problem. The 29 human samples were provided courtesy of physical anthropologist W.W. Howells, using 28 measurements on some 2500 individuals; the large chimpanzee sample (17 measurements on 347 individuals) and gorilla sample (25 measurements on 590 individuals) were both provided courtesy of primate taxonomist Colin Groves of the Australian National University in Canberra. Taxonomists classify chimps into two species, the common chimpanzee (Pan troglodytes), and the so-called pygmy chimp or bonobo (Pan paniscus). The common chimp species, Pan troglodytes, is itself divided into three subspecies or races, called P. t. verus, P. t. troglodytes, and P. t. schweinfurthii. Gorillas are classified as belonging to either the western or one or more of four eastern populations. The formal taxonomy for gorillas is still in dispute, with some recent classifications seeing three speciesóone in the west (Gorilla gorilla) and two in the east (G. graueri and the so-called "mountain gorilla," G. berengi), while others classify the eastern forms as races of Gorilla gorilla.
The metric I have used here is the percentage difference per size-corrected measurement (expressed in SD units), and the numbers given are the percent increases in distance going from within-group to between-group comparisons of individuals. Thus, for example, the increase in average distance in going from the paired comparisons of two males to similar comparisons for male and female pairings in some human populations is about 1-3%; for example, for Zulu, 1.15%; Australia, 1.87%; and Santa Cruz Island, California, 2.16%.
For pygmy chimps, the corresponding increase for male-female versus male-male comparisons is 4.7%; for common chimps it is 10.4%. These figures are consistent with the general sense that the degree of sexual dimorphism increases as we go from humans to pygmy chimps to common chimps.
Comparing distances among the three common chimp subspecies or races (for males) gives us about 6% between verus and either schweinfurthii or troglodytes, and 1.6% (which is getting very close to noise level) between the latter two. From bonobo to verus is about 20%; to troglodytes, 14.6%; and to schweinfurthii, 8.8%. These three bonobo morphological distances are correlated with the corresponding geographical distances, and one has to wonder to what extent the smaller bonobo (P. paniscus)óchimpanzee (P. t. schweinfurthii) statistic is due to parallel evolution and how much to gene flow between the two sometime after the basic chimp/bonobo split took place. Neither I nor anyone else as yet presumes to have the answer to that one.
For gorillas the basic split is between the western form (G. gorilla) and four eastern populations (three races of G. graueri and G. berengi, the "mountain gorilla"). The relevant statistics for these comparisons are: 17.3%, 19.8%, 22.9%, and 24.7%, respectively. Between the first graueri race and the other three eastern forms, the percentages are 7.9%, 12.8%, 12.3%. Between the second graueri race and the others, the differences are 4% and 11.1%. Between the third graueri race and berengi, one gets a value of 8.4%.
Those are the ape statistics I have calculated to date. Now, what about us humans? Let's start by comparing three African samples, using the data provided by W. W. Howells. The three samples include: the Dogon of Mali, the Teita (Kenyan Bantu speakers), and the South African Bushman. The morphological distances are as follows: Dogon to Teita = 9.9%; Dogon to Bushman = 13.4%; Teita to Bushman = 14.9%. Thus the Dogon or Teita to Bushman "racial" distance is very similar in magnitude to the 14.6% separating two chimpanzee species (using the Pan troglodytes troglodytes to bonobo number). Similar figures (16.3% and 15.5%, respectively) are obtained when comparing a sample from Hokkaido (Japanese, not Ainu) and two Amerind groups (the Arikara from South Dakota and a sample from Santa Cruz Island). Other comparisons that help put these distances in perspective are: South Australia to Tolai (a New Britain group) = 10%; Dogon to Norse = 19.4%; and South Australia to Norse = 26%.
The largest distance among chimpanzees or gorillas (the 24.7% between G. gorilla and G. berengi) is slightly less than that separating Howells' Norse and the South Australian samples (26%), but even this set of statistics doesn't begin to exhaust the range of human variation. The largest differences of all in Howells' sample are found when comparing Africans with either Asians or Asian-derived (Amerind) populations. Thus, the distance between Teita and Tierra del Fuegans is 32.4%, while Zulu to Tierra del Fuegans or Santa Cruz yields a distance of about 36%. The largest difference for any of the human data sets is 46%, which comes from comparing Teita with Buriat (who live in the Lake Baikal area of Siberia and speak a Mongolian language).
I will also note here that putting these results on the bell curve would mean that each increase of 15% in distance is worth about 1 SD, and that the absolute within-group variation does not vary significantly among the three genera (humans, chimps, and gorillas) involved. I think this is the first time any such comparisons have ever been carried out (and it is certainly the first time any such data have been published), and the results seem well worth the effort.
Why? Who would have predicted that the racial morphological distances in our species could be much greater than any seen among chimpanzees or gorillas, or, on the average, some 10-fold greater than those between the sexes?
I think it fair to say "no one." But they do exist, and they also seem to speak directly to the question of race differences in athletic performanceóin the sense that we again see those one to two, or more, SD spans coming up. If the prevailing orthodoxy were correct and race were just a recent social construction based on a few unimportant characteristics such as skin color, one couldn't find such differences in so critical a behavior as bipedalism or the anatomical structures on which it is based. Remember that the 1.5 SD advantage the Kalenjin have over the average human population suggests that there must be some population that is 1.5 SD below the mean, and therefore separated from the Kalenjin by 3 SDs (as are Buriat from Teita in the cranial data analysis above).
In the PBS website debate between a leading forensic anthropologist, George Gill, and physical anthropologist Loring Brace on Kennewick Man and the reality of race, Gill describes how it is possible to determine race from skeletal remains, including those critical to bipedalism. "Numerous individual methods involving midfacial measurements, femur traits [which I note here are related to running ability], and so on are over 80% accurate alone, and in combination produce very high levels of accuracy. ÖI have been able to prove to myself over the years, in actual legal cases, that I am more accurate at assessing race from skeletal remains than from looking at living people standing beside me. ÖThe idea that race is ëonly skin deep' is simply not true, as any experienced forensic anthropologist will affirm." Were the orthodoxy true, Gill's statement could not be.
At this point the skeptic might bring up the quite correct point that the genetic distances among human races are minimal, and argue that we might therefore be fooling ourselves with this concentration on phenotype rather than genotype. As President Clinton said in the first State of the Union address of the new millennium, referring to the 99.9% genetic identity of all living humans: "Modern science has confirmed what the old religious teachers told us [which, of course, they didn't], that what unites us is our common humanity." Well, the 99.9% figure is true enough. So is our common humanity. We do, after all, all belong to one species and so share a lengthy common evolutionary history. But the thrust of the message is not correct. And to see that it isn't, we need only to think of our closest friends. Dog breeds can obviously be much more different from one another, both morphologically and behaviorally, than human races; no one doubts that those differences are genetically based; and yet they remain genetically indistinguishable. That is, we cannot yet look at a dog's DNA and determine whether it is a Pit Bull or a Pekinese, but any child of five can, and so should any potential burglar.
The point is that there need not be any strong correlation between genetic distance and morphological or behavioral distance. Therefore, one cannot legitimately argue, as many do for our species, that the minimal levels of interpopulational genetic variation observed are telling us something about the levels of such functional phenotypic variation present. Again, just as was the case for Darwin's thinking, looking at "artificial" (human-guided) selection gives us a much better sense of what natural selection can accomplish.
Back To The Orthodoxy
The evidence I have presented shows that the orthodoxy with which this piece began is clearly wrong, and, in a way, that really isn't news. We have long known of any number of characteristics that surely have been under very intense selection for very long times, without this selection having reduced variation in them. We've just seen the situation with respect to bipedalism, and I'll mention one otheróbrain size. The tripling in size of the human brain over the last 2 million years is one of the most rapid evolutionary changes that can be documented in the fossil record, and given the costs involved (delayed maturation, birthing problems, energy costs to run the brain, the travails involved in training it), we can be sure that the selective processes involved must have been intense. Yet brain size within our species remains one of our most, not least, variable features. It is easy to find pairs of perfectly normal humans whose brain sizes differ by enough grey matter to run a whole chimpanzee (400cc), and there are populations that differ in their mean brain sizes by close to 300cc (something more than 2 SDs).
There are innumerable other well-known examples from agriculture and animal husbandry. I'll just mention one study that has been going on at the University of Illinois Champaign-Urbana for more than a century. Scientists there have been selecting for oil content (both high and low) in maize (corn) kernels, with the results seen in Figure 1. Note that in the line selected for higher oil content it has increased almost 5-fold (20 SDs), and at a pretty steady rate, indicating that the underlying genetic variability has not been running out. Thus today's average is something close to three times the highest values one might have seen in the 1880s when the experiment started, one might note in closing, with a grand total of 12 ears.
I should note here that it has taken natural selection some 20,000 centuries to increase human brain size the 20 SDs that intense human-guided selection has managed for oil content in one. That kind of comparison gives us a much better sense of what is possible and strongly reinforces the point about not using up heritable variation under intense selectionóespecially when one considers that over 20,000 centuries there is going to be a fair amount of new variation added by new mutations.
Now, by definition, short-term human examples are going to be much harder to come by. We don't deliberately breed humans for particular characteristics, and with long generation times its difficult to find meaningful data even in natural experiments. Nonetheless, there is one of these natural experiments that provides very obvious results in a rather short period of time. In his recent book, The Lapita Peoples, my Berkeley colleague, Patrick V. Kirch, has summarized the linguistic, archaeological, morphological, and genetic evidence showing that today's Polynesians derive from a population, or populations, living off the eastern end of New Guinea about four thousand years ago. In the interim, they have spread out to cover a vast triangle of islands from New Zealand in the south, to Hawaii in the north, and Easter Island in the east. Doing so meant crossing thousands of miles of ocean that could be chillingly cold at night, and in large outriggers where upper body strength would have been at a premium, thus apparently selecting primarily for larger body size (Bergmann's rule), and, by extension, proportionately even larger upper bodies (there is positive allometry between overall body size and upper body size in apes and ourselves). Thus we now find big-bodied people in the tropics. At first glance this seems odd because it is a violation of Bergmann's rule, but they are there, and Jon Entine notes that there have been more than 50 Polynesians in the National Football League, making them by far its most "overrepresented minority."
So the actual data, be they from corn or people, don't seem to be consistent with the idea that strong selection on important features reduces the amount of heritable variation in them. But why doesn't it? Well, things might work that way if only one or a small number of genes were involved in producing a complex character like the brain or the various structures involved in bipedalism. But we know that isn't true. And we can dispose of the argument that selection was involved at only one or a small number of genetic loci with a simple thought experiment. Imagine that this were the case. Then the idea would be correct, and variation would get rapidly used up. And then? Well, yes, that's the problem, isn't it? No more variation, no more natural selection, no more adaptive changeórapid extinction. Or, more formally, this sort of process would rapidly lose out to one in which the variation used to produce the adaptation was in fact spread out among as many genes as possible, such that new mutations could easily reconstitute such variation as was lost to selection. And the more important the adaptation, the more likely it would evolve that way; that is, by involving as many loci as possible in the selective process. That's why we see all that functional variation out there in precisely those features that have been so important in our evolution.
Out of Africa
Now that we have dealt with the question of why variation does not get used up, we can go back to the question of "why do Africans dominate some sports, such as sprinting and marathoning?" The answer is: (1) genes, genes, and more genes; and (2) nobody really knows why Africans should be so dominant. How do we know that statement (1) is correct? Because the results over the past few decades are consistent with a genetic model, and inconsistent with the "it's all society and culture" model. If it were the latter, then as opportunities for individuals of African descent began opening up, first on the track, and then in various professional sports, and finally, and especially, in Africa itself, we would have gotten an increase in African participation and competitiveness, up to where their numbers would be strongly correlated with their numbers in the arena from which the participants were being drawn. The NBA would be about 10% black, we'd see a Kalenjin only every few years at the cross-country championships. If, on the other hand, the athletic superiority was a product of genes, then we'd would have gotten exactly what we have gottenóparticular genetic populations in very different social and cultural settings producing very similar results.
At least some factors that could very likely be contributing to African dominance are clear. First, we have to remember that our lineage arose in Africa some 4-5 million years ago and, since then, has spent most of its time exclusively in Africa. Bipedalism began in Africa, stone tool making began in Africa, we first spoke in Africa, humanness began in Africa, much of the increase in brain size occurred in Africa, and so on. Our basic adaptations are African. Given that, it would seem that we would have had to make adaptive compromises, such as to cold weather, when populating other areas of the world, thus taking the edge off our "African-ness." Thus body fat levels seem to be at a minimum among African populations; they do not increase with age in them, and Africans in training can apparently achieve lower body fat levels more readily than is the case for Europeans and Asians. That would appear to be plenty to begin with in explaining such African dominance as exists in the running world.
The Final Taboo
The final taboo, to which Jon Entine devotes Chapter 18 (entitled "Sports and IQ"), results from the fear people have that allowing a genetic explanation in one area leads to the possibility of genetic contributions in all areas, and one of thoseóthe most touchy of allóis going to be brain power. The idea that being smart and being a good athlete are mutually exclusive has a strong appeal, and its presence is, often unthinkingly, taken for granted. The stereotype then is not so much "blacks as physical brutes and whites as thinking beings," but the more general one going back perhaps to Platonic types of "thinking beings" being designed by nature to rule over "physical brutes."
I can see no necessary, or even likely, negative correlation between the physical and the mental. On the contrary, the data show that there is an obvious, strong, positive correlation among class, physical condition, and participation in regular exercise in this country. On the "runner" end for example, our fitness/running booms were not driven by mindless "physical brutes." At the "sprinter" end, professional football teams have, in recent years, been known to use the results of IQ tests as one indicator of potential in rookies. Research going back to the study of intellectually gifted California schoolchildren begun by Lewis Terman in the 1920s show, if anything, a low positive correlation. In any case, it is long past time to get this "brains or brawn" business out of our systems. Doing so would let us look at the "why Africans in sports?" matter far more intelligently, if for no other reason than that it would bring the power of many more previously timid individual brains to bear on it.
Jon Entine has broken the taboo in the realm of athletic achievement. In a calm, measured, and reasoned discussion spread over some 375 pages, he makes public what just about all the sports authorities he quotes in his book have acknowledged, mostly in private, for decades: Africans are better than the rest of us at some of those things that most make us human, and they are better because their separate African histories have given them, in effect, better genes for some of our recently developed tests of some basic human adaptations. The rest of us, or, more fairly, our ancestors, have had to compromise some of those African specializations in adapting to more temperate climates and more varied environments. Contemporary Africans, through their ancestors, are then advantaged in not having had to do so, and their bodies, along with their resulting performances, show it. Let's go on to learn from this example that genetic variation is the norm, and that competition provides us with a procedure for discovering how variation occurs. We can't even begin to know what could be until we encourage and allow as much of that variation as possible to express itself. Taboo illustrates for us how much we have all gained by allowing African genetic variation to express itself via athletic competition. That lesson should not be ignored in the name of racial equality. Individuals are not equal; nor are races. They cannot be. That they are not can sometimes be a problem; far more often, it is an opportunity. But there will be neither an opportunity made available, nor results to take advantage of, if we cannot accept that we can't make it come out "even."
So be it.
Copyright © 19992011 Jon Entine all rights reserved