Geographical Centrality as an Explanation for Racial Differences in Intelligence

Edward M. Miller

Department of Economics and Finance

University of New Orleans

New Orleans, La. 70148

emmef@uno.edu (E Mail)

December 4, 1995

Contents

Stylized Facts 1

Access to Intelligence Raising Mutations Determines Population Average Intelligence 6

Diffusion of Genes 6

Foraging Populations with Population Density Varying 8

The Role of Agriculture 12

Boat Migrations and Trading 22

Negative Selection for Intelligence 24

Implications for Other Genes 25

Testable Implications 26

Implications for Variability in Intelligence 28

Conclusions 30

References 31

Summary

Intelligence is affected by many different genes. It has also plausibly been subject to unidirectional selection. Calculations show that favorable mutations would move at a rate that was slow relative to the time since modern human symbolic culture emerged. This makes it very likely that geographical differences in the frequencies of various intelligence related genes exist. With unidirectional selection in a polygenetic system, it is meaningful to talk about some areas being more advanced than others (since there is a direction in which all are moving). Centrally located populations will normally be more advanced. Genes will move faster in thinly populated areas. The thinly populated areas can serve as genetic freeways that carry genes rapidly across continents.

New technologies, including agriculture, the horse and the ship, accelerated the spread of mutations. The horse caused the Eurasian steppes to become a genetic highway that transported favorable mutations across Eurasia. This probably caused these areas to reach high levels of intelligence ahead of other areas. Areas without horses or ships, such as sub-Saharan Africa lagged. Peripheral areas such as Australia and the Americas also lagged due to isolation from the large populations of Eurasia. Keywords: Intelligence, race, population genetics, unidirectional selection.

In his survey of the intelligence of the worldUs peoples Lynn (1991a) found that the highest levels were found in people that evolved in Eurasia (Mongoloids and Caucasoids), with low values found for those that evolved in Africa (Negroids).

The few explanations that have been offered for the evolution of racial differences in intelligence have involved differing strengths of selection for intelligence in various regions. Climate has been the most common source for differential selection for intelligence. These theories have argued that the intellectual demands of life in cold climates was greater than in warm climates. Lynn (1991b) has placed emphasis on the intellectual abilities needed to survive cold, to build fires, and to hunt in groups. Rushton (1995) has presented a theory involving r versus K selection. Miller (1991) has pointed to the need to store food to survive the winter and how this may have selected for intelligence. He has also (Miller, in press) argued that one of the advantages of intelligence was that it helped in detecting deception in mates and potential mates, and that this ability was more important in cold climates than in warm ones. The implicit assumption in these models is that the same alleles were present in virtually all populations. Thus, intellectual differences between populations must reflect differences in the strength of selection for intelligence.

The alternative to be presented here is that some populations were reached more quickly by more of the mutations that produce high intelligence. These became the more intelligent populations. Other populations, those that were less accessible to intelligence increasing mutations, lagged in intelligence. Thus, the populations reached by the largest numbers of such mutations would have the highest average intelligence. Most populations experienced selection for intelligence, although its strength may have differed.

Stylized Facts

There are several stylized facts (well established generalizations) that will be used in the argument.

1. Much of the current human variation in intelligence is genetic (Bouchard, 1993; Bouchard, Lykken, McGue, Sega, & Tellegen, 1990; Jensen, 1981, Plomin, & Loehlin, 1989; Rowe, 1994).

2. Behavior genetics research suggests the absence of a single gene for intelligence. Instead, intelligence is affected by a large number of different genes (Plomin, Owen, & McGuffin,1994; Plomin, McClearn, Smith, Vignetti, Chorney, Venditti, Kasarda, Thompson, Detterman, Daniels, Owen, & McGuffin, 1994; Plomin R., McClearn, G., Smith, D., Skuder, P., Vignetti, S., Chorney, M., Chorney, K., Kasarda, S., Thompson, L., Detterman, D., Petrill, S., Daniels J., Owen, M., & McGuffin P., 1995; Skuder, P., Plomin R., McClearn, G., Smith, D., Vignetti, S., Chorney, M., Chorney, K., Kasarda, S., Thompson, L., Detterman, D., Petrill, S., Daniels J., Owen, M., & McGuffin P., 1995).

Wills (1991) suggests 50 genes, each contributing about 3 IQ points is of a plausible order of magnitude. Jinks & Fulker (1970, p. 343) conclude that at least 22 loci seem to be controlling IQ. Later (p. 344), using data for inbreeding depression, they conclude that about 100 genes seem to be showing dominance for high IQ.

3. There has been unidirectional selection for intelligence in much, if not all, of the world. The fact of inbreeding depression suggests that many of the alleles that contribute to low intelligence are recessive, with the alleles contributing to high intelligence being dominant. This suggests that the genes for high intelligence have been the subject of a continual process of directional selection (Jinks & Fulker, 1970, p. 343). Because directional selection acts very slowly in eliminating recessive genes, but quickly to increase the frequency of dominant alleles, a high average level of dominance suggests long continued directional selection for a trait. It is not known exactly what selected for intelligence in the course of human intelligence, but plausible candidates include the intellectual requirements of survival, and the social needs to get along with fellow humans.

4. It will be presumed that each individual has an equal probability of experiencing an intelligence raising mutation (regardless of the population they live in). This is standard genetic theory, since no population differences in vulnerability to mutations are known. Weakening this assumption would not change the nature of the argument.

5. Favorable genes spread slowly under prehistoric conditions in which humans were divided into tribes that only occasionally exchange genes with each other. There is evidence that humans have built in mechanisms that make them ethnocentric and suspicious of outsiders (Reynolds, Falger, & Vine, 1987). While this does not completely prevent contact and interbreeding between human populations, it does greatly reduce it.

Hiorns & Harrison (1977) compute gene frequencies for 300 generations for 10 populations in a row with one starting out with a gene frequency of 100% and the other zero. Their graphs show that for a migration rate of 5% each generation (the percentage of the populations that marries into the adjacent tribe) and a selection factor of .01, even after 300 generations, the populations are still easily distinguishable. Their analysis led the authors to conclude (pp. 440-441), RThis analysis clearly stresses the limitations of migration and reinforcing selection as homogenizing influences on between-population variety in short-term evolution. It seems unlikely, for instance, that should some populations have become fixed for a Tgeneral improvementU gene since the end of the Palaeolithic, or even since the advent of the species Homo sapiens as we know it, gene flow and selection would have distributed the genes very far in space or that it would have achieved an appreciable frequency in many of the populations it had reached, through these agencies alone.S

Rouhani (1989) uses FisherUs (1937) wave of advance model for the spread of advantageous genes to make this point. Fisher showed, by using a diffusion model, that after a gene was established in a deme there would be a wave of advance for which V= (1/2)s(s)1/2, where V is the velocity of the wave, and s is the selective advantage of the allele. The measure of the parent offspring distance is provided by s which in turns equals me2, where e2 is the area of the deme, and m is the migration rate between demes. The parent offspring distance depends on the average distance between centers of adjacent demes, and the percentage of the population that migrates between demes. Increasing the area of the demes increases the distance between them. This increases the average parent offspring distance, and hence the rate of gene flow. This effect will be important later in the argument when the effects of introducing the horse and the ship are considered.

Rouhani uses plausible parameters (selection coefficient of .01, 5% migration between demes, demes 500 people and 5000 square miles) based on characteristics of current hunter-gatherers, and concludes that advantageous genes would advance at .8 miles per generation. For a favorable mutation to go from South Africa to the China coast would require 400,000 years. Weiss & Maruyama (1976) and Livingstone (1992) use similar assumptions to arrive at somewhat faster rates of advance, but still quite a slow one.

It should be noticed that this is a long period of time relative to the time that many authorities believe has passed since humans left Africa. For instance, if humans reached China 100,000 years ago, a favorable mutation that occurred in China would not have reached Europe or Africa yet, nor would mutations occurring in the latter areas have reached China. Thus, if the original hunter-gatherer social pattern had remained in place, there could be many favorable mutations that are found in only certain parts of the world, simply because there has not been time for them to spread to other parts of the world. This makes regional variation in frequencies of intelligence relevant genes virtually certain.

There is one possibility that should be noted. If one population had a sufficient advantage over other populations, possibly because of their intelligence, or the weapons or organization it made possible for them, this population could expand at the expense of the other populations. Their expansion just distributes the genes for intelligence faster than they would diffuse in the standard stepping stone models. The replacement of the Neandertals by anatomically modern humans could be such an episode. Such replacement can be seen in the fossil record (although even here disproving evolution in place is difficult).

If the differences between populations left no evidence in the fossil record (and the differences between more and less intelligent individuals in modern populations are typically not the sort that would be apparent in skeletal remains), several such rapid replacements could have occurred that left no fossil record, and these may have played a role in disseminating genes for intelligence. Later in this paper, dispersals due to the coming of agriculture, the horse, and ocean going ships will be discussed. The period since the emergence of modern symbolic intelligence is short relative to the time required for mutations to spread around the world (see below).

6. Judging from when they adopted their current symbolic culture, humans have had their current level of intelligence for a relatively short period of time. Of course, there is no direct measure of early human intelligence. However, Stringer & Gamble (1993) document the absence of modern symbolic culture before the upper Paleolithic. Noble & Davidson (1991) argue that there are no signs of symbols in the archaeological record before 32,000 years ago. White (1982) summarizes the differences between the middle and upper Paleolithic, most of which can be interpreted as evidence for greater intelligence in the upper Paleolithic. Binford (1982, p. 178) states his impression that "the ability to anticipate events and conditions not yet experienced was not one of the strengths of our ancestors prior to the appearance of clear evidence for symboling." Intelligence might almost be defined as "the ability to anticipate events and conditions not yet experienced".

The material culture of prehistoric man was at a very low level before the emergence of anatomically modern man, and gradually increased. The rate of progress was very slow. Although it is just barely possible that humans had a high level of intelligence long before they developed evidence of a sophisticated material culture, and merely did not display their intelligence, the simplest explanations for the long period of a primitive, non-symbolic culture is that humans had not yet developed sufficient intelligence to do more (again see discussion in Miller, 1995).

The interesting thing is that the period in which humans have had a symbolic culture appears to be short in comparison with the time required for genes to diffuse around the world. If high intelligence is recent, the mutations that produced the high intelligence would not have time to reach all populations.

7. The above discussion shows why genes, even subject to favorable selection, are likely to diffuse slowly, with at any given time there being many favorable mutations that have reached some populations but not others.

To make the nature of the argument to come clearer, imagine that each favorable mutation raises the IQ of the individual carrying the mutation by one IQ point. The IQ of an individual will then be determined by the number of favorable alleles he has inherited. The average intelligence of a population will then be determined by the number of favorable alleles that have reached it, weighted by the percentage of the population that has inherited each allele. If, as appears to be the case, genes can go to fixation in one population before they have even reached other populations (see the simulations in Livingstone, 1992), the intelligence of a population can be conceptualized as determined by the number of favorable mutations that have reached it. Let us explore the implications of this simple idea.

Access to Intelligence Raising Mutations Determines Population Average Intelligence

Consider what the above stylized facts imply for the distribution of intelligence among populations. With selection for intelligence, a major determinant of the average intelligence level in a population will be the number of genes favorable for intelligence that have reached the population. As will be shown below, a major determinant of the number of favorable alleles that reach a population is that populationUs location, with the more peripheral populations receiving fewer favorable alleles.

Diffusion of Genes

Thinking about the diffusion of advantageous genes on a straight line, it would seem that the middle would be likely to have been reached by more favorable genes than either end.

As the diagram shows, the largest number of advantageous mutations would be expected to have reached the center. The easiest way to see this is to consider a period of time just sufficient for a gene originating at either end (i.e. either A or C) to have reached the center. Now consider a point not at the center, say at B. The genes originating at the left side B would still have reached it, but there would be an area on the extreme right, near C, from whence mutations would not yet have been received. Since this argument can be made for all points not at the center, it follows that the population with the highest expected number of favorable mutations will be the one located at the center. The highest value for a polygenetic trait such as intelligence subject to unidirectional selection is expected to be at that point.

The above point is simple, but most discussions of the evolution of human traits seem to have presumed that the lines in the above diagram would be horizontal, and that human intelligence was rising in a uniform manner. The slowness of gene flow makes this unlikely. If we think of the world as three lines joined at the Middle East (Africa, Asia, Europe), the Middle East would be expected to have received the largest number of advantageous mutations, and the peripheral regions to have received the fewest. The prediction is that during foraging times, the Middle Eastern populations would have had the largest number of alleles conducive to high intelligence. If agriculture had not come, one might have expected this pattern to have persisted into modern times.

For a flat plane the highest level of intelligence would be expected at the center. The argument can be seen on Figure 2. It shows how a favorable mutation originating at C spreads outwards in concentric circles. The diagram can also be interpreted as showing the areas from which which favorable mutations will have been received. At point C, all mutations occurring before T1 will have been received from the area encompassed by the circle labeled T1. All mutations occurring since time T2 inside the circle labeled T2 will have reached point C. The further back one looks, the larger the area there is to draw on for mutations. Now imagine a continent shaped like the ellipse. Consider points A and C. All mutations arising within the areas encircled will have reached both points. The circles are the same size, with the result that the maximum distances from which mutations can reach them are the same.

However, for the hypothetical population at C, the land area within its circle (the circle labeled with B and D) is much less than for the circle centered at A. It follows the peripheral population at C will have benefitted (on average) from fewer mutations than the population at point A. Thus, it would be expected that the population at point A would have a higher intelligence than the population at point C.

Again, this is a simple point but the prediction is that the more centrally located populations will have been reached by the most mutations. Hence, they will be the most intelligent.

It should be noticed that the above prediction that intelligence should be higher in the more centrally located regions is a logical deduction from several generally accepted facts: that intelligence is affected by numerous genes, that intelligence has been selected for during relatively recent prehistoric times, and that favorable mutations diffuse slowly. Anyone wishing to dispute the conclusion, that the centrally located populations should have a higher frequency of the genes for intelligence, has to dispute one of the generally accepted facts, or dispute the logic. Either is hard. The predicted forager pattern might roughly fit the data if Australia and America are thought of as peripheral regions with small populations (i.e. few advantageous mutations) and slow diffusion of advantageous genes originating elsewhere. Both Australia and America have lower intelligences than Eurasia (Lynn, 1991a).

Before discussing the implications of agriculture, let us discuss further how fast favorable genes might migrate in a foraging population. This speed should not be taken to be constant.

Foraging Populations with Population Density Varying

If intermarrying tribes are roughly constant in population size (as they are believed to be), the distance to the boundaries of a tribe will be much less in low population density areas. In such low density regions the population ranges over larger areas, and bands will be separated by larger distances. Those of marriageable age have to look much further to find mates. Thus genes would actually flow fastest through such areas. In essence, they would travel many more miles before they hit a tribal boundary.

A few wide ranging tribes could pick up the genes at one end, have them increase in frequency within their populations, and then transmit them to the other end of their territory. Once this had been done, many more miles would have been covered than would have happened in a more densely populated region, where many tribal boundaries would have had to be crossed. One implication is that thinly populated areas like the Sahara may have been less of a barrier to the flow of favorable genes than thought (they could remain a barrier to neutral genes) (Miller, 1994a).

This argument of course makes the differences observed between the north and south of the Sahara desert more puzzling Cavalli-Sforza, Menozzi, & Piazza (1994). Probably most of the genes for which it appears a barrier are selectively neutral, or subject to only weak selection. With only diffusion, a small desert population can have little impact on the adjacent populations.

Thinly populated areas actually may act as genetic freeways. Consider the situation shown in Figure 3:

Each rectangle is a tribe. The wide rectangles (only one of which is completely shown) represent low population density tribes. A mutation appearing in any square is equally likely to be transported to any adjacent square, where it then proceeds towards fixation. After a period of time, the mutation is carried to an adjoining tribe, represented by a rectangle. Thus, the arrows leaving square A show how a favorable mutation might be transported. After the first period, the mutation is present in adjacent populations only. If the whole diagram was like the lower two rows, there would be waves of mutations moving across the diagram in the manner Fisher described, but it would take a long time for a mutation arising at one end to reach the other end.

Now consider the situation with low density populations to the north. Mutations arising in the lower squares will quickly reach the upper rectangles. They will then move rapidly east and west. These low density regions then serve as freeways transporting advantageous mutations to other regions. As illustrated, the mutation reaches B and D in the first period. In the second period, the mutation is transported to each of the rectangles adjacent to those reached in the first period. It can be seen that the distant rectangle E is reached at the same time as the much closer rectangle C. The mutation was transported to the low density area D, where in the course of dispersing within the tribe it was quickly carried to the far boundaries. From there, it readily diffused to the population E. Thus, the low population density area served as a genetic freeway transmitting the favorable mutation from A to E, much quicker than if it had to traverse the intervening series of small squares. The tribal boundaries are major obstacles to gene flow. The number of boundaries to be crossed determines the speed with which favorable mutations move. The number of boundaries to be crossed in going A, B, C is the same as the number crossed going A, D, E. Thus, tribes C and E will receive the mutations at about the same time, even though E is much further away.

The transmission might be even more rapid if the populations were very mobile themselves, as might happen for pastoralists (see discussion below), or northern groups following mobile herds of game (i.e. reindeer). The steppes of Eurasia might have been such a freeway, picking up genes from populations around the boundaries of Eurasia and carrying them to the other ends. They may have delivered them to Europe and North Asia at either end.

Empirical evidence on the size of tribes is thin, but it does appear that the area occupied by a tribe increases as the environment deteriorates. There is a tradition in anthropology that tribes have about 500 members. If tribes have an equilibrium size of about 500 (see Kelly, 1994, for an evaluation of this tradition), in low density areas the tribes must occupy larger areas in order to reach this population size (which is argued to be roughly the number required to provide adequate mating opportunities).

Empirically, Birdsell's (1953) examination of the relationship between the area occupied by an Australian tribe and the rainfall showed that tribal areas were larger in the drier climates. After excluding the tribes which depended on resources that were not likely to be affected by local rainfall (island tribes, coastal tribes, and tribes on large rivers fed from outside the tribal territory) and Rby the elimination of tribes in which cultural factors modify the size of the population from the assumed constant of 500 persons,S the area occupied by a tribe (Y) and the rainfall (X) was: Y=7,112.8 X-1.58451

Since area goes up as the square of a linear dimension, the linear dimensions of the territory of the tribe (L) will vary as approximately X-3/4. Thus, the distance between tribal boundaries increases as the rainfall declines. In very dry areas, such as the interior of Australia or the prehistoric Sahara, distances between tribal boundaries would be appreciably greater.

In a low density area, like the Sahara, the genes should have to cross fewer tribal boundaries to cover any given distance. As Birdsell (1951: 282) put it in discussing Australia in an earlier paper, RConsidered in terms of the rate and ease of gene flow, the great, forbidding, arid desert spaces of the central portion of the continent represent freeways, rather than obstacles, to gene exchange between distant populations.S

He provides evidence, from the spread of Carpentarian characteristics in Australia, that genetic diffusion is indeed as would be predicted from a model where tribal boundaries are an obstacle to gene flow, and the boundaries are further apart in areas of low rainfall. This leads to the hypothesis that in Eurasia, as in Australia, the central part of the continent with its low population densities may have constituted a freeway that permitted favorable mutations to move faster than they would have if population densities were higher. For instance, mutations originating in the densely populated area of China could have moved through thinly populated Central Asia to Central Europe faster than if they had to diffuse along the coast of China, reach India, traverse northern India, and then moved through the Middle East, and the Balkans.

The Role of Agriculture

It was discussed above how Rouhani (1989) estimated that the rate of progress of advantageous genes would be only .8 miles per generation for a hunter-gatherer population. This slow speed is predominantly due to the low intermarriage rates across tribal boundaries, which in turn arises from tribes being endogenous. Genes can move faster if whole populations move, since even on foot, people can move faster than .8 miles per generation. It is not known how often favorable genes were carried forward by the movement of tribes, although it is easy to imagine that this occurred.

Movement of genes could have been faster with the coming of agriculture. It has been argued that agriculture was spread by demic diffusion. A settled way of life increases the population growth rate, and farming populations would be expected to expand into adjacent areas that were inhabited only by foraging populations. Cavalli-Sforza, Menozzi, & Piazza (1994, pp. 108-111) present evidence that the expansion of agriculture into Europe was at about the rate of one kilometer per year. The kilometer per year advance of genes by demic diffusion would far exceed the .8 miles (1.3 kilometers) per generation estimated (see above) for a foraging population. With a generation of 20 years, this is about 25 times as fast as genes would diffuse through a foraging population.

Their evidence suggests genes of the agricultural population were carried along with the expansion of agriculture. There was a continual mixing of the expanding agricultural population with the hunter-gather populations of the regions they were moving into. Thus, any advantageous mutations that had occurred in the hunter-gather populations would probably be picked up and carried by the expanding farming populations into new areas. This wave of advance carried both the original genes of the first population to adopt farming, and of the intervening foraging populations. The Middle East is likely to have been a central area during the hunter-gathering period when intelligence was high.

It should be noticed that the accelerated pace at which genes move during a demic diffusion of agriculture is a one time effect. After the spread of agriculture, population density would increase. It would then be expected that tribes would shrink in size and the distances to be traveled to find a mate would decrease. In addition, agriculture involves a sedentary life which would reduce the natural movements of foraging people which might bring them into contact with other groups. The situation might come to resemble that in New Guinea with a very large number of tribes each speaking their own languages, and doing relatively little intermarrying (Cavalli-Sforza et al. 1994). This would cause the post-agricultural revolution rate of spread of new mutations to decrease to a level below that of the foraging period.

However, not all areas would have adapted agriculture. Those areas that were too dry, or otherwise unsuitable would have retained a foraging life style. These thinly populated desert and steppe areas would remain areas of low population density able to serve as genetic freeways moving genes from one area to another. In particular, the thinly populated Central Asian steppes may have served as a genetic freeway connecting the densely populated agricultural areas of China, India, the Middle East, and Europe. With mutation rates being the same in the different populations, most of the favorable mutations would occur in the agricultural areas where most of the worldUs population lived. However, the thinly populated Central Asian area would have received these genes and transmitted them to other parts of Asia.

If within the farming population there was selection for intelligence, the genes once introduced into a population would move towards fixation. This demic farming diffusion model predicts higher intelligence levels in the farming communities after the agricultural expansion, because more of the genes for intelligence would have reached them.

It would also be expected that the greatest number of genes for intelligence would have reached the areas that were settled at the end of the agricultural expansion. The reason is that the wave of advance would have picked up the largest number of favorable mutations. To illustrate, consider a series of tribes arrayed along a straight line A, B, C, D, E, F. An agricultural expansion begins among A. Its expansion brings it into C. A favorable gene that had emerged in C passes into the expanding population, and begins to increase in frequency. Meanwhile, the favorable gene is carried with the wave of advance. Additional favorable genes may be picked up from D and E. Thus, when the wave reaches F, it is likely to have picked up any favorable genes that were in the foraging populations that the wave traversed. Thus, it is the populations that are last reached by the wave of advance that are likely to have picked up the largest number of favorable genes.

This argument would hold even if the first population to adopt agriculture was no more intelligent than other foraging populations. It could even work if the farmers were less intelligent. However, the first farmers may have been more intelligent. It is possible that intelligence was needed to conceive of the idea of planting crops and cultivating them. This is not to say that the first to conceive of planting seeds and farming actually did so.

Hunting and collecting takes less work than farming. It is likely that the fact that seeds grew into plants, and that farming was possible was discovered several times. It was then promptly forgotten since it was easier to gather what nature had already planted. However, eventually population may have risen to the point where adequate food was not gathered by hunting and gathering, and planting increased the food supply. At that point, someone conceived of farming, implemented the idea, and encouraged his descendants to do so. The first farmer was very likely more intelligent than most. He was also likely more farsighted since he could visualize the harvest vividly enough to inspire him to do the work of planting for a return that will come only many months from now. Once farming was adopted by a few pioneers, their descendants were more numerous, and their genes spread. The initial advantage was partially a better food supply, but it may have also been the adoption of a sedentary way of live that permitted women to have the next child before the first was old enough to walk long distances on its own (Lee, 1972 as cited by Ammerman & Cavalli-Sforza, 1984, p. 64). The problem of carrying two children at once in a migratory lifestyle is believed to have limited foragers to having only one child young enough that could not walk at his parents' pace. Because adopting farming required intelligence, and because farmers probably out reproduced others, farming's appearance would have been accompanied by increased intelligence.

However, even if the initiation of farming did not require any special intelligence, the first farming population may have been unusually intelligent. The first farmers are believed to have been Middle Eastern. As discussed earlier, the Middle East is a central region receiving favorable mutations from Asia, Europe, and Africa. Thus, at any given time, populations in this area are likely to have received more favorable mutations than populations in other regions. Evidence is that farming was carried into Europe from the Middle East by movement of farming populations. Such movements would have carried the genes for intelligence that had reached the Middle East from Africa and Asia into Europe (Europe presumably already had the genes that had emerged there). This demic expansion from the Middle East into Europe can explain why modern Middle Easterners do not appear to be more intelligent than Europeans, even though the earlier theoretical argument suggested that a greater number of favorable genes should have reached them.

Cavalli-Sforza et al. (1994) report that the gene frequency differences between European populations are relatively small and that European gene frequencies resemble those of the Middle East. In dendograms (descent trees) the branch leading to Europeans is often relatively short. A very interesting study discussed by Cavalli-Sforza et al., (1994) analyzed only a few populations (including Chinese, Europeans, two populations of African pygmies, and Melanesians), but collected data on a very large number of alleles. A tree constructed from this data showed a very short branch leading to the Europeans (p. 91). Several explanations were considered, but the most plausible was mixture. Calculations showed that the European gene frequencies could be explained well by a mixture of Chinese with a smaller percentage of pygmies. Obviously, this is not the actual racial history of the Europeans (who are both taller and lighter skinned than either group, for instance). The pygmies are fairly close to other Africans in the frequency of their measured genes (the set of measured genes frequencies includes no genes that affect height) according to their data.

The above surprising result is most easily explained by the current European gene frequencies reflecting a relatively recent (in prehistoric terms) migration of a Middle Eastern population that was in turn a mixture of Asians and Africans (or at least one which regularly received genes from both). After expansion, the gene frequencies were frozen, and drift did not change them much from those that had existed in the Middle East.

Renfrew (1991) and Barbujani, Pilastro, Domenico, & Renfrew (1994) argue that not only do European gene frequencies suggest demic diffusion from the Near East, but evidence of such demic diffusion can also be found in the areas occupied by the speakers of Altaic languages, and by the Asian speakers of Indo-European and Elamo-Dravidian languages, and possibly the Afro-Asian languages. The basic argument is that agriculture emerged in the Near East among several groups in the Fertile Crescent. One wave of expansion swept into Europe carrying the Indo-European languages with it. The existence of this wave is documented by the distribution of gene frequencies in Europe, and by archaeological evidence which shows a steady advance of farming at about 1 kilometer per year. It is hypothesized that another wave, possibly starting in the Zagros foothills of Iran, led to an expansion of the populations that become Elamo-Dravidian speakers. This wave reached as far as southern India. Later, Indo-European speakers expanded into Iran and North India, leaving the Dravidian speakers isolated in South India, with a couple of relict populations isolated along the expansion path. Of course, these theories are highly controversial, with most linguists rejecting the idea that the origins of the language groups go back as far as the origins of agriculture.

The Altaic speakers are argued to have spread north from the Fertile Crescent area, expanding all the way to Korea and Japan. Along the way they would have had the chance to pick up and spread genes for intelligence in the large area from the Middle East to Japan. Finally, the Afro-Asian speakers spread from the Levant into Egypt and then on to the rest of North Africa. Even if these arguments are rejected, the existence of these language groups is generally agreed to be due to the languages having a common origin. The spread of the language from the area of common origin must have been accomplished by the movement of people, even if only small groups of conquerors (see discussion below). Of course, if the movements were later than the origin of agriculture, there would be less time for favorable genes to be selected for.

In the case of all of these expansions, the argument being made is not that the early farmers were necessarily any more intelligent than the foragers whose territory they expanded into. Instead, the argument is simply that, due to the mixing of genes from a larger area, a larger selection of advantageous genes would have reached the populations affected by the agricultural expansion. If the genes were merely neutral, the resulting mix would be a weighted average of the gene frequencies of the constituent populations. However, with directional selection for higher intelligence, having a wider selection of intelligence related genes for selection to work on would have resulted in the eventual evolution of higher intelligence. Thus, even if the evidence of a common language family reflects only a conquest, a few new genes would have been introduced.

Once agriculture had spread into an area, the rate of gene flow would be expected to decline again. Agriculture would support a denser population, and one that was less mobile. In a denser population the distance that must be traveled for a mate is less, and the average distance between the partners in a marriage is less. Languages and dialects would differentiate, and these differences would prevent marriages between different groups. Tribal groups would come to occupy smaller areas, and would be expected to be endogamous. Thus, the rate of diffusion of genes would be limited by the boundaries of the tribal groups, and the short distances from one boundary to the other would again limit the gene diffusion. The situation might come to resemble that in New Guinea where there are large number of tribes occupying a relatively small area, with large linguistic and genetic differences between the tribes. In such an environment new intelligence raising genes would spread very slowly. (That is, they would spread slowly unless something again happened to cause large scale migrations of new peoples).

It should be noticed that if the numerical size of demes remains constant (say at the traditional 500), changing population density uniformly does not change the rate of advance of a trait undergoing unidirectional selection, such as intelligence. The total number of mutations at any given distance is increased as population density increases, but the number of deme boundaries to be crossed is also increased as each deme comes to occupy a smaller area. A way to see this is to think of the demes as being in a hexagonal grid, with the demes arranged in concentric circles around the deme one is interested in. At any given time, the mutations (if any) from a certain number of demes away are reaching the target deme. Changing the size of the demes does not change the number of boundaries that must be crossed for mutations arising say 20 demes away to reach the target deme. If, say after 100,000 years, mutations 20 demes away are just reaching the target deme, it makes no difference how large the demes are in a model in which members of a deme are equally likely to mate with any other member of the deme. The distance to the deme that is 20 demes away just happens to be less when the population density is lower.

Of course, as pointed out, where the problem is expressed as time to cross a specified distance, lowering population density lowers the number of boundaries to be crossed, thus speeding up the time required for a gene to cross the boundaries.

Increasing the heterogeneity in the population densities will decrease the number of boundaries to be crossed to connect two distant demes, since the gene flow will be through the low density areas between the centers for high population. The coming of agriculture probably did increase the heterogeneity of population density. The areas that adapted agriculture were the areas of higher rainfall, which were probably already areas of relatively high population density. Agriculture just increased their population densities further. The areas of low rainfall, which were already areas of low population density, would have remained foraging areas of low population density. Thus, the heterogeneity increased.

Horse Based Migrations

There is evidence for several migrations after the early spread of agriculture. These are the migrations that are usually interpreted as giving rise to various major language groups (The Renfrew hypothesis discussed earlier that agricultural expansion gave rise to the Indo-European language groups is a minority view). For instance, there is a linguistic similarity between the various languages of the Indo-European group which extend from India to Western Europe. This is usually explained by these languages having a common origin, implying that the speakers of the proto-Indo-European language once lived in an area small enough to have a common language, estimated by Mallory (1989, p. 146) at 250,000 to 1,000,000 square kilometers. Obviously, for the Indo-European language to now cover the very large area they do cover, there must have been an expansion of the language, which was almost certainly caused by a movement of at least some people, even if just a few conquers. Mallory estimated that the proto-Indo-Europeans were in their homeland 4500-2500 years B. C. They expanded from this homeland. Why they expanded is not definitely known, but one plausible explanation is the domestication of the horse, and the advantage this gave them in warfare.

Domestication of the horse occurred around 5,000 BC or earlier. RThis innovation cut traveling times by a factor of five or more, nullifying whatever territorial boundaries had previously existed. . . . . Riding provided the ability to strike out over great distances, instigated cattle looting or horse-stealing raids, the accumulation of wealth, trading capacities, and the development of violence and warfare. Material remains of the first half of the 5th millennium B. C. show that in an enormous territory east of the Don River and between the Middle Volga, the Caucasus Mountains, and the Ural Mountains there spread a uniform culture.S (Gimbutas, 1991, p. 354). Very likely this uniform culture arising from the mobility horses permit mixed the genes thoroughly, and much more quickly than normal diffusion could have mixed them. Gimbutas and others have argued that the advantage of the horse would have led to the expansion of the first peoples to have mastered it. This would have rapidly spread intelligence promoting genes. The effects of the initial expansion were followed by a period of faster gene flow resulting from the horse based culture.

Not only is the horse a major asset in warfare, but a pastoral economy seems to lead to an emphasis on fighting. This is basically because the development of an economy based on livestock changes the cost benefit-ratio for raiding, making it a much more economical source of food. Livestock is easily driven away. In contrast grains and tubers must be carried away (and perhaps even harvested). Foraging people seldom have much worth raiding for (other than women). Faced with the threat of raids, those owning livestock are forced to develop fighting skills to defend their livestock. Even today, herding people seem more oriented towards fighting. Since the horse domesticators were probably a pastoral people, they would be expected to have developed a livestock raiding culture. With the military advantage of horseback riding adding to culture oriented toward fighting, they very plausibly could have expanded into surrounding peoples, as Gimbutas (1991), Mallory (1989), Anthony (1986) and others have argued. The case for such an expansion is based on both archaeological evidence, and the widespread prevalence of the Indo-European languages. Such an expansion presumably carried genes.

Of course, for the spread of Indo-European languages, it is not necessary for the original inhabitants of an area to be displaced. Conquest by a relatively small group can lead to the adoption of the conquerorUs language. The classic example is the adoption of Turkish in what is now Turkey, which is known to be the result of a conquest by a relatively small number of Turks.

However, the introduction of a small number of advantageous genes would not require many people, especially if the leaders of the conquering army were more likely to be carrying the desirable genes. It is very likely that achieving and retaining leadership of a conquering army was facilitated by intelligence. It is also very likely that the conquerorUs leadership had an above average chance of leaving their genes, through either marriage or rape. Once the genes had been introduced into a population, if there was selection for such genes, they would gradually increase in frequency. Genetic evidence for an expansion from the steppes exists.

Cavalli-Sforza et al. (1994, p. 293 and fig. 5.11.3) found that the third principal component for European gene frequencies showed an area of extreme values north of the Black Sea, with what appear to be roughly concentric circles around this area. They point out that this is consistent with an expansion of the Kurgan culture from the steppes of Europe such as Gimbutas argued for. They also note that Scythians were in the same area later, and also invaded Europe. It could also have been the original homeland, or an intermediate long-term homeland for some of the other barbarian populations that later invaded Europe.

The Indo-EuropeansU original homeland is a subject of disagreement. For the sake of discussion, imagine it was in the steppes north of the black Sea (as above) or north of the Caspian Sea as hypothesized by Gimbutas, with expansion from here going into Europe and further east into Asia. Any mutations for intelligence between the homeland and Western Europe would have been swept up by the migrating populations and spread into Europe. In the long period of time since 4500-2500 years BC there would have been time for these genes to benefit from selection, and to increase in frequency.

Similar effects could occur with other homelands, although the magnitude of the effect might vary. For instance, if nomadism was introduced into the steppes from the farming populations on its western edge (in the Ukraine or Rumania), the steppe populations might initially have had gene frequencies similar to those populations and a later movement into parts of Western Europe might have brought fewer new genes. However, they still might have picked up genes from further east in the course of subsequent movements, and then brought these into Western Europe. Of course, with a more western origin of the steppe nomads, the latter movement into Iran and India would have brought into these areas genes originating in western Europe. Thus, regardless of where the population that spoke proto-Indo European is believed to have originally lived, the movements of the parts of this population that spread the Indo-European languages would have spread intelligence promoting mutations, the more favorable of which would have been selected for.

Notice, for the above effect to occur, it need not be argued that the new arrivals were more intelligent that the conquered. They may have been less intelligent, with the conquest's immediate effect being to lower the average intelligence. However, if the new arrivals had genes for intelligence that had not yet reached the original population, the net effect after a long period of time, could have been to have raised the intelligence of the combined populations above that of any of the original populations. For instance, consider the unrealistic case where genes have become completely fixed in a population. The settled farmers have two favorable mutations fixed, and the invading populations one. The invaders are lower in intelligence. The immediate effect of mixing the populations lowers the intelligence below that of the original residents. However, with selection for intelligence after many generation the newly merged population may come to have all three alleles in high proportion, and to have an intelligence higher than that of either of the predecessor populations.

Thus, to argue that the Indo-European expansions contributed to raising intelligence one does not necessarily have to argue that the Indo-Europeans were themselves superior in intelligence. They may have merely played a role in spreading desirable mutations widely. Of course, the original Indo-European expansion is just one of many expansions by steppe horse riding populations.

Many expansions of steppe peoples are known to have occurred in history such as the barbarians that invaded the Roman Empire (Cavalli-Sforza et al. 1994, Fig. 5.2.6), and later the Magyars and Mongols. It is very likely that the steppes of Eurasia were traversed in both direction several times by horse mounted conquerors. China was repeatedly conquered by horseman from the steppes of Asia (Cavalli-Sforza et al. 1994, p. 201-202; Darlington, 1966). In particular the Hun that sacked Rome, under attila, those that attacked India under Mihirakula, and the Hsuing-nu who threatened China were apparently the same group, spreading their genes over this vast Eurasian area (Kust, 1983, p. 36). Any favorable mutations arising between China at one end, and Europe at the other were probably diffused throughout the Eurasian region. Similar arguments could be made for these other expansions, although after the later expansions there would be less time for natural selection to increase the frequency of any desirable genes introduced. Again, if the conquerorsU leaders were more likely to carry genes for intelligence, and these leaders fathered many offspring, newly introduced genes might have had a head start that was more important than might have been guessed from the numbers in the conquering army.

More recent horse borne long distance movements are known, including that of the Turks (see Mallory, 1989, p. 147 map), or that of the Arabs out of Arabia into North Africa, Spain, etc. Each of these could have spread favorable genes.

It is also very likely that horse riding societies permitted choice of spouses over relatively large areas. The horse provided mobility, and the lack of the attachment to a fixed place that farming people had would permit groups to gradually change their locations. Pastoralism often leads to seasonal movements which may cover from 20 to 1000 miles or more (Cavalli-Sforza et al. 1994, p. 200). These may bring the nomads into contact with different sedentary populations at different times, or into contact with other nomadic populations. Hence, genes can be presumed to have moved very rapidly in the areas of the world populated by horse mounted people. Thus, the great steppes of Eurasia appear to have been a freeway that transported desirable mutations from one end of Eurasia to the other end. This happened long enough ago so that there were often several hundred generations for selection to increase the frequency of the genes that made for intelligence. The result was the evolution of high intelligence within the peoples within the reach of the horse riding Eurasian populations. These appear to have extended from Japan, Korea, and China at one end to Western Europe and North Africa at the other end.

Could lack of access to the horse have slowed down the spread of genes? Many areas of the world lacked access to horses till recently. The Americas, Australia, and New Guinea lacked horses because they were isolated by water from Eurasia, where the horse was domesticated. Sub-Saharan Africa is contiguous to areas that used horses, but because of the tsetse fly did not have domesticated horses. In these areas populations would have moved on foot. Tribal size would have been smaller. Favorable mutations would have moved slowly. Over time they would come to have a lower level of intelligence than the lands whose gene flows were facilitated by horses.

Boat Migrations and Trading

Another method that could carry favorable mutations over long distances is boats. Once long distance boat transport had emerged, mutations could cover long distances without having to diffuse slowly through populations.

The earliest example of such long distance boat based migrations is that of the Phoenicians who settled such places as Carthage. This could have moved mutations at an even more rapid rate than horse based migrations.

It might be noted that the Phoenicians were located at the Middle Eastern crossroads where they may have received genes from Asia, Europe, and Africa. Their early colonies could have transported these genes to distant places from whence they spread. This may have contributed to the spread of Middle Eastern genes throughout Europe.

As with other innovations, it is plausible that the people that first perfected long distance ship transport may have been above average in intelligence. If this was so, the ability of the ship to carry them long distances would have dispersed their genes widely. From the initial colonies, the genes would have spread to adjacent peoples, and then spread rapidly in frequency.

Of course, after the Phoenician era, the Greeks established a far flung set of colonies, extending to the Black Sea Coast and around the Mediterranean. After that the Roman Empire emerged. It was centered on the Mediterranean and experienced large scale migrations of peoples. This could have easily transported favorable mutations from one end of the Empire to the other. With a large number of generations since this era, natural selection could have served to raise the frequency of the desirable genes throughout the Mediterranean region, and areas in contact with it.

It should be noticed that the ship also led to extensive long distance trade. If ancient sailors and merchants were like modern sailors, they left genes behind them. From the ports, genes could have easily spread inland.

Certainly, the traders and ship captains were well above average in intelligence. It is very likely that soon after a intelligence increasing gene reached a population, some of those receiving the gene went into the intelligence requiring profession of trading (where those with the high IQ have a comparative advantage). Since the traders tended to be travelers, the gene may have been at an over 50% frequency among those going on trading trips even when it had a much lower frequency in the population as a whole. This could speed up the diffusion of the gene among people who did long distance trading.

It is also very likely that offspring of traders became traders themselves, and that the traders in a community intermarried extensively. This could make an advantageous gene move faster than with random marriage. Suppose for instance an advantageous gene emerged in Central Asia and was carried to the Black SeaUs east coast. One could easily imagine a Phoenician trader bringing home a concubine or slave carrying the gene. This match could easily give rise to a son who then signed on to participate in a trading trip to England's Cornwall tin mining district, where he mated with a prostitute. Thus, in two generations a gene could make it half way across Eurasia. This sequence of events is much more likely if traders were drawn from sons of traders, than if they were randomly selected from the whole Phoenician population. In the latter case, it might take many generations for the gene to slowly increase in frequency before it reached someone who was making a Cornwall trip. Such long distance gene transport would be of little importance for neutral genes, since the percentage of genes in Cornwall that could be traced to the Black Sea Coast would be small. However, if the gene did raise intelligence and was hence selected for, the gene could come to have a high frequency in Cornwall, and diffuse from there to the rest of England. The parts of Eurasia (and North Africa) connected by long distance trading routes would tend to have their intelligence raised as intelligence promoting genes were spread over long distances.

Negative Selection for Intelligence

The above discussion has been on the assumption that high intelligence was selected for. Such positive selection for intelligence is plausible in many societies. Intelligence would help in attracting mates, and in achieving positions of leadership that led to mating opportunities. It probably also assisted in earning a living and hence promoting the feeding and survival of one's offspring, as well as in arranging advantageous matches for them. However, in more recent times intelligence may not have contributed to reproductive success. In most modern industrial societies, the high socioeconomic status and educated individuals have fewer children than those with low status and poor educations (Herrnstein & Murray, 1994; Itzkoff, 1994). This seems to occur especially with females. Females postpone marriage and child bearing to obtain an education. They find a conflict between a high status occupation and child rearing. (Women also find a conflict between low status occupations and child rearing, but find it easier to sacrifice a low status, uninteresting occupation for child rearing). Also, low intelligence seems to lead to more failures at contraception and additional births. However, these conditions appear to have risen only recently with changing status for women, and the emergence of modern contraception.

There may have been earlier selection against intelligence. Most likely, the population of cities in Medieval Europe and early modern times failed to reproduce themselves, primarily because the high population density facilitated the spread of disease. The population was maintained by continual immigration from the surrounding countryside.

In at least some circumstances, it is likely that those that immigrated to the cities (and remained there) were of higher intelligence than those that remained in the cities. The cities probably had a higher proportion of occupations for which high intelligence was an asset, including craftsmen, traders, and government officials. Intelligence was probably not as much of an asset in peasant agriculture. It is possible (but unproven) that high intelligence by encouraging movement to a city (and being able to earn enough to stay there) was actually selected against in some times and places. If the selection for intelligence disappeared, the areas of the world where genes were being well mixed would no longer have a tendency to have their intelligence raised.

Implications for Other Genes

The argument has been developed for genes that raise intelligence because that is a socially desirable trait that many believe to have been subject to unidirectional selection for most of human history. However, the principle is perfectly general that conditions producing the long distance importation of new alleles lead to the increase in the receiving area of any trait subject to unidirectional selection.

In a region where malaria was endemic and malaria resistance was being selected for, one would expect more mutations resistant to malaria to have reached the areas that were exchanging genes over long distances than the areas that were relatively reproductively isolated.

There may be other traits that have been subject to selection in much of the world. In northern climates, a common strategy for getting through the winter was storage of food. It was also desirable to devote effort to the building of homes that would protect from the cold. In modern economies, this ability leads to saving and investment in productive resources. In prehistoric tropical societies there was little opportunity for planning ahead. Thus, the populations that moved out of the human cradle in Africa have probably been under selection for the ability to defer gratification. This trait would be expected to be most common in the areas that have received genes from a large part of the world.

Testable Implications

The above account appears to be congruent with what is known about the worldwide distribution of intelligence. High intelligence is reported for the populations of Europe and Northeast Asia (China, Japan, Korea) which are at each end of the Eurasian steppes (Lynn, 1991a). Areas that are isolated from Eurasia by water, and of smaller populations (Australia and the Americas) have lower scores even when the populations are living in relatively cold areas that might be thought to have selected for intelligence.

In the above model, Australia and the Americas are peripheral areas. Their populations were probably too small (in relation to the Eurasian populations) to generate many mutations, and they were probably too far, or too isolated to receive many mutations from the Eurasian land mass. Elsewhere (Miller 1995), I argued that these continents were sufficiently isolated that mutations occurring on the Eurasian land mass since their initial settlement probably had not reached them. Even if I am wrong, and there has been some gene flow since initial settlement, it is likely that their positions far from the Eurasian population centers caused them to be very peripheral, limiting their access to intelligence increasing mutations. Thus, they would have been expected to have lagged behind Eurasia in the development of intelligence, as the data shows them to do (see Miller, 1995 for detailed documentation).

Lower intelligence is found in Africa, and among those of African descent. This can be plausibly argued to be due to weaker selection for intelligence in a tropical climate (Miller, 1991, in press) along with the isolation caused by lack of horses and poorer access to water borne trade (and traders) in earlier eras. The poorer access to water borne trade would be due to poor harbors on the coast, a lack of inland seas, and a lack of navigable rivers flowing down to the water.

What other predictions emerge? Right now, while the evidence is quite strong that there are genes that contribute to intelligence, exactly what these genes are and where they are located is unknown. However, evidence has recently been presented that certain genetic markers are statistically more common in those of high intelligence than in those of low intelligence (Plomin, et al, 1994; Plomin et al, 1995; Skuder, et al, 1995), and one has been found that appears to affect spatial ability without affecting intelligence (Berman, & Noble, 1995). Given the rate of progress in molecular genetics, it is likely that several alleles that have a positive or negative effect on intelligence will soon be located. If the above theory is right, not only will these genes prove to differ in frequency between populations in different parts of the world, but some of the ones identified in European or northeast Asian populations (the populations most commonly studied, simply because they are the populations that are most convenient to the leading laboratories) will be found to be essentially absent (a low frequency may be the result of recent mixing with Europeans) in the original aboriginal populations in such areas as Australia and the Americas.

The above theory raises the possibility that certain alleles with a favorable effect on intelligence may have become fixed in European or Northeast Asian populations if they originated in these regions, (and possibly even if they originated elsewhere but reached these populations early enough for natural selection to fix them). Studies that are limited to just one group (such as Caucasians or Japanese) may not detect a correlation of these genes with intelligence.

A good example is provided by the high-affinity aldehyde dehydrogenase gene, which comes in two versions in Orientals (Tu & Israel, 1995). One version provides protection against alcoholism because they cannot easily digest the aldehyde that is produced after alcohol consumption. The aldehyde makes them mildly sick. This simple genetic difference can explain most of the difference in drinking within the Oriental population in North America. However, the allele that is common in Orientals is virtually unknown in Caucasians. Studies limited to Caucasians would not have discovered this genetic effect.

The above argument would suggest that mixed populations (such as American blacks, or those of mixed Australian aboriginal and Caucasian descent) might very profitably be investigated. A finding that possession of a particular genetic marker was correlated with intelligence would suggest that that marker either directly affected intelligence, or was close to a gene that affected intelligence. Of course, in populations that are a mixture of two populations that differ in intelligence, any gene that differs in frequency may be merely serving as a marker for the extent of admixture (not to mention for the extent of acculturation).

It would be necessary to control for this. For instance, if there were other genes that were believed to be unrelated to intelligence (possibly from studies in other ethnic groups), but which did differ in frequency between the two parent groups, these could be used to estimate the degree of admixture. Many genetic markers, including blood group, human leukocyte antigen genes, and restriction length polymorphisms, are known to differ between populations (Cavalli-Sforza et al. 1994). Thus, it should be possible to estimate the extent of admixture independently of the genes believed to be linked with intelligence. Independent measures of acculturation would have to be sought as a control. This differs from the procedure of the major quantitative tract loci study of intelligence so far (Plomin, et al. 1994), which limited itself only to Caucasians.

It was argued that some isolated areas such as Australia may have received few new mutations after settlement. However, if they experienced continued selection for intelligence, some of the alleles that the population arrived with may have become fixed, or nearly fixed in their populations. In this case, the standard deviation of intelligence should be smaller in such populations than in the populations that have been continually receiving new genes from other populations. This is a testable prediction.

Africans are generally found to have somewhat lower standard deviations for intelligence than Caucasians (Jensen, 1980). This might be explained if a slower migration of alleles into Africa and within Africa had resulted in African populations having fewer polymorphic intelligence relevant genes. Many intelligence relevant alleles would have reached them so long ago that they had become fixed, and many other alleles would not have reached them yet, even if they accounted for appreciable variation in other populations. In the areas that have had continual access to new mutations there will be more alleles that have not become fixed, causing a greater standard deviation of intelligence.

Implications for Variability in Intelligence

Incidentally, this ongoing process of new mutations coming into a population followed by selection for them may be the way to resolve the paradox of why there is so much genetic variation for intelligence (g), if g is a trait that is beneficial. Some have pointed out that variables that are subject to strong selection normally show little variability. Usually such variables reach their equilibrium values quickly, and are now observed in the process of reaching equilibrium.

For instance Patterson (1995, p. 210) gives great weight to Vale's (1980, p. 435) rhetorical question, "If IQ is a fitness character, why should the additive variance be anywhere near .71?". Vale goes on to argue, "The answer of course is that it should not, if indeed IQ is closely related to fitness. If it is not so related, then presumedly it has not been selected for throughout human evolution. If it has not been selected for, then it evidently has not played a very great role in that evolution."

In general, a trait can be contributing to fitness and be being selected for without the trait having reached its genetic limit, although powerful selection makes it more likely that the limit will be rapidly approached, making it harder to observe the organism in the process of being selected. For a trait subject to the type of selection in which one animal having the trait increases the benefit of a even higher level of the trait in another individuals (the so called arms race or red queen effect, see Ridley, 1994), the period of adaptation is increased. If intelligence is subject to unidirectional selection in which people with a higher intelligence benefit reproductively from being able to outwit those of lower intelligence, it is likely that at any given time there will be some of higher intelligence than others, thus solving the problem. Still, in general Vale and Patterson have a point for virtually all traits except intelligence.

Intelligence, since it is needed to discover its own existence, occupies a special position among all traits. As intelligence gradually increases, it is to be expected that a few individuals with sufficient intelligence to do psychometrics, and discover the concept of g will emerge. When the distribution of intelligence has risen to the point where some individuals investigate intelligence, others individuals will be of much lower intelligence. At this time, only a small fraction of the population is likely to have sufficient intelligence to do psychometrics and to understand the concept of g. Thus, the finding of a wide range in intelligence, a variable that contributes to fitness, is perhaps not as surprising as it might appear at first.

Thus, Patterson's (1995, p. 196) argument "The problem which Herrnstein, Jensen, and all hereditarian psychologists face them, from the discipline on which they have so heavily drawn, is that IQ scores are too hereditary if they are to sustain the claim that these tests have any significance beyond the test center and classroom." This would be a much more powerful argument if applied to any trait other than intelligence.

The same argument can be extended to populations. Because of the wide geographical area Homo sapiens occupies, its long generations, and the obstacles to gene flow across tribes, there are likely to be differences in the intelligence of different populations at any time. When some populations have reached the point of having the technology to explore the world, they are likely to discover that other populations have not yet developed to this point, and they can be expected to conclude that there are differences between the world's various populations in intelligence.

Conclusions

Intelligence is a genetically influenced variable that is affected by many different genes. It has also plausibly been subject to unidirectional selection. Calculations indicate that for a small hunter-gatherer population that genes would move at a rate that was slow relative to the time since modern human symbolic culture emerged. This makes it very likely that geographical differences in the frequencies of various intelligence related genes will exist. With unidirectional selection in a polygenetic system, it is meaningful to talk about some areas being more advanced than others (since there is a direction in which all are moving). Centrally located populations will normally be more advanced. Genes will move faster in thinly populated areas. The thinly populated areas can serve as genetic freeways that carry genes rapidly across continents.

Given the very slow progress of genes with a stable population structure, occasional waves of advance caused by new technologies or the movement of populations can greatly accelerate the movement of mutations. The spread of agriculture was one such event. The coming of the horse and the ship were other similar events. The horse caused the steppes of Central Asia to become a genetic highway that transported favorable mutations from China, Europe, India, and the Middle East to other areas. This caused these areas to reach high levels of intelligence before other areas. Areas without the horse, such as sub-Saharan Africa, would have lagged.

Peripheral areas such as Australia and the Americas probably also lagged due to isolation from the large populations of Eurasia.

References

Alexander, Richard D. (1990). How did humans evolve. Ann Arbor, Museum of Zoology, Special Publication No. 1.

Ammerman, A. J. & Cavalli-Sforza, L. L. (1984). The neolithic transition and the genetics of populations in Europe. Princeton: Princeton University Press.

Anthony, D. W. (1986). The "Kurgan culture," Indo-European origins, and the domestication of the horse." Current Anthropology, 27, 291-304.

Barbujani, G., Pilastro, A., Domenico, S. D., & Renfrew, (1994). Genetic variation in North Africa and Eurasia: neolithic demic diffusion vs. paleolithic colonization. American Journal of Physical Anthropology, 95: 137-154.

Berman, S. M. & Noble, E. P. (1995). Reduced visuospatial performance in children with the D2 dopamine receptor A1 allele. Behavior Genetics, 25, 45-58.

Binford, L. R. (1982) Comment. Current Anthropology, 23, 177-181.

Biraben, J.-N. (1980). An essay concerning mankindUs evolution. Population 4: 1-13.

Birdsell, Joseph B. (1951). Some implications of the genetic concept of race in terms of spatial analysis. In Cold springs harbor symposia on quantitative biology, Vol. XV, Origin and evolution of man, Cold Spring Harbor, New York: Long Island Biological Association.

Birdsell, Joseph B. (1953). Some environmental and cultural factors influencing the structuring of Australian aboriginal populations. American Naturalist 87:171-207.

Bouchard, T. J., Jr., (1993). The genetic architecture of human intelligence. In Vernon, P. A. (Ed.). The biological basis of intelligence, Norwood: Ablex, 33-94.

Bouchard, T. J., Jr., Lykken, D. T., McGue, M., Sega, N. L., & Tellegen, A. (1990). Sources of human psychological differences: The Minnesota study of twins reared apart. Science, 250, 223-250.

Buss, David M. (1994). The evolution of desire. New York: Basic Books.

Cavalli-Sforza, L. L., Menozzi, P., & Piazza, A. (1994). The history and geography of human genes. Princeton: Princeton University Press.

Darlington, C. D. (1969). The evolution of man and society. New York: Simon & Schuster. Gimbutas, M. (1991). The civilization of the goddess. San Francisco: Harper. Herrnstein, Richard J. & Murray, Charles (1994). The bell curve: intelligence and class structure in american life. New York: The Free Press.

Hiorns, R. W. & Harrison, G. A. (1977). The combined effects of selection and migration in human evolution. Man, 12, 438-445.

Jensen, A. R. (1981). Straight talk about mental tests, New York: The Free Press.

Jinks, J. L. & Fulker, D. W. (1970). Comparison of the biometrical genetical, MAVA, and classical approaches to the analysis of human behavior. Psychological Bulletin, 73, 311-349.

Knight, C. , Power, C., & Watts, I. (1995). The human symbolic revolution: A Darwinian account. Cambridge Archaeological Journal, 5, 75-114.

Kust, M J. (1983). Man and Horse In History. Alexandria, Va.: Plutarch Press.

Lee, R. B. (1972). Population growth and the beginnings of sedentary life among the !Kung bushmen. In Population growth: anthropological implications. Ed. B. Spooner, Cambridge: MIT Press. Livingstone, F. B. (1992). Gene flow in the pleistocene. Human Biology, 64, 67-80.

Lynn, R., (1991a). Race differences in intelligence: a global perspective, Mankind Quarterly, 31, 254-296.

Lynn, R., (1991b). The evolution of racial differences in intelligence, Mankind Quarterly, 32, 99-121.

Mallory, J. P. (1989). In search of the indo-europeans. London: Thames & Hudson.

McShane, D. & Berry, J. W. (1988). Native north americans: indian and inuit abilities. In Irvine, S. H. & Berry, J. W. (Eds.) Human abilities in cultural context (pp. 385-426) Cambridge: Cambridge University Press

Miller, E., (1991). Climate and intelligence. Mankind Quarterly, 32 127-132.

Miller, E., (1993). Could r selection account for the African personality and life cycle? Personality and Individual Differences 15, 665-676.

Miller, E. M, (1994). Did the Neandertals separate the Africans from the Eurasians? Manuscript

Miller, E. M, (1994). Paternal Provisioning versus Mate Seeking in Human Populations, Personality and Individual Differences, 17, No. 2, 227-255.

Miller, E. M, (1995). The evolution of Australian and Amerindian intelligence. Manuscript Miller, E. M, (in press). Environmental Variability Selects for Large Families only in Special Circumstances: Another Objection to Differential K Theory, Personality and Individual Differences.

Noble, W. & Davidson, I. (1991). The evolutionary emergence of modern human behavior: Language and its archaeology. Man, 26, 223-253.

Patterson, O. (1995). For whom the bell curves. In Fraser, S. (Ed.) The bell curve wars. New York: Basic Books.

Plomin, R., & Loehlin, J. C. (1989). Direct and indirect IQ heritability estimates: a puzzle. Behavior Genetics, 19, (3), 331-342.

Plomin, R., McClearn, G., Smith, D., Vignetti, S., Chorney, M., Venditti C., Kasarda, S., Thompson, L., Detterman, D., Daniels J., Owen, M., & McGuffin P. (1994). DNA markers associated with high versus low IQ: The IQ quantitative trait loci (QTL) project. Behavior Genetics, 24, 107-118.

Plomin, R., McClearn, G., Smith, D., Skuder, P., Vignetti, S., Chorney, M., Chorney, K., Kasarda, S., Thompson, L., Detterman, D., Petrill, S., Daniels J., Owen, M., & McGuffin P. (1995). Allelic associations between 100 DNA markers and high versus low IQ. Intelligence, 21, 31-48.

Renfrew, C. (1991). Before babel: speculations on the origins of linguistic diversity. Cambridge Archaeological Journal 1: 3-23.

Reynolds, V., Falger, V., & Vine, I. (1987). The sociobiology of ethnocentrism. London: Croom Helm.

Ridley, M. (1994). The Red Queen: Sex and the Evolution of Human Nature. New York: Macmillan. Rouhani, S. (1989). Molecular genetics and the pattern of human evolution: plausible and implausible models. In The human revolution, Mellars, P. & Stringer, C., Eds., 47-61.

Rowe, D. C. (1994). The limits of family influence: genes, experience, and behavior. (New York: Guilford: 1994).

Skuder, P., Plomin R., McClearn, G., Smith, D., Vignetti, S., Chorney, M., Chorney, K., Kasarda, S., Thompson, L., Detterman, D., Petrill, S., Daniels J., Owen, M., & McGuffin P. (in press). A polymorphism in mitochrondrial DNA associated with IQ? Intelligence, 21, 1-12.

Slurnik, P. (1993). Ecological dominance and the final sprint in hominid evolution. Human Evolution, 8, 265-273.

Stringer, C. & Gamble, C. (1993). In search of the Neanderthals: Solving the puzzle of human origins. London: Thames and Hudson.

Takahata, N. (January 1993). Allelic geneology and human evolution, Molecular Biology and Evolution, 10, 2-22.

Tu, G. & Israel, Y. (1995). Alcohol consumption by orientals in north America is predicted largely by a single gene. Behavior Genetics, 25(1), January, 59-65.

Vale, J. R. (1980). Genes, environment, and behavior: An interactionist approach. Cambridge, Harper and Rower.

Vraniak, Damian A. (1993). Native Americans. In Encyclopedia of human intelligence, Sternberg, Robert J. New York: Macmillan Publishing Co.

Weiss, K. M., & Maruyama, T. (1976). Archeology, population genetics and studies of human racial ancestry. American Journal of Physical Anthropology, 44, 31-50.

White, R. (1982). Rethinking the middle/upper Paleolithic transition. Current Anthropology, 23, 169-180.

Wills, Christopher (1991). Exons, introns, and talking genes. New York: Basic Books.