Culture rather than genes provides greater scope for the evolution of large-scale human prosociality

Adrian V. Bella,1, Peter J. Richersonb, and Richard McElreathc

aGraduate

Group in Ecology, bDepartment of Environmental Science and Policy, and cDepartment of Anthropology, University of California, Davis, CA 95616

Edited by Richard E. Nisbett, University of Michigan, Ann Arbor, MI, and approved August 31, 2009 (received for review March 25, 2009)

Whether competition among large groups played an important role in human social evolution is dependent on how variation, whether cultural or genetic, is maintained between groups. Comparisons between genetic and cultural differentiation between neighboring groups show how natural selection on large groups is more plausible on cultural rather than genetic variation.
altruism cultural FST group selection prosociality

uman societies are unusual among vertebrates. While people in small-scale societies exhibit much more cooperation and division of labor than other primates, people in even very large societies also show strong tendencies toward altruism. Warfare, food sharing, and taxation are all examples of prosocial patterns of behavior that are common in human societies but nearly completely absent in other vertebrates. Even when plausible analogues can be found in other vertebrates, the scale of costliness of human altruism is extraordinary (1). Explaining the levels of human altruism observed ethnographically and experimentally has proven to be difficult. Much of this altruism is directed at strangers, and so is difficult to explain as simple reciprocity, or it benefits entire tribes or nations of only distant genealogical kin, and so is difficult to explain as altruism among individuals sharing recent common ancestry. Another scenario many researchers, since at least Darwin (2), are concerned with is competition among residential human groups that are too large to comprise close genealogical kin (26). If groups differ in the frequency of individuals who are willing to sacrifice their own labor, time, or safety in ways that promote the competitive ability of the residential group, then over time groups with higher frequencies of such altruists may tend to replace groups with fewer (79). In this paper, we refer to this scenario as group-level selection, the evolution of behavior that reduces individual fitness but increases the average fitness within large groups of only distantly related individuals. By distantly related, we mean that most individuals within the residential group do not share very recent common ancestors, and so common descent alone does not maintain much genetic variation among residential groups. Nevertheless, given the right population structure and low rates of mixing among groups, individuals within groups may be much more genetically similar to one another than they are to members of other groups, and therefore they may be closely related in one important sense of the term (10). If genetic variation among groups is sufficiently large, evolutionary theory predicts that self-sacrifice on behalf of large residential groups can evolve under the same processes that evolve self-sacrifice on behalf of close kin. This is because all hypotheses about the evolution of altruistic behaviorbehavior that reduces the absolute fitness of the actor but increases the absolute fitness of recipientshinge on processes that change and maintain variation among social groups (1114). Selection for altruism in such large groups, however, remains a controversial topic in part because it is not clear that enough between-group variation existed in human societies to make it an appreciable evolutionary force (15). In very large residential groups, migration can quickly erode between-group genetic
www.pnas.orgcgidoi10.1073pnas.0903232106

variation. Nevertheless, recent work argues that sufficient variation did exist by invoking reproductive leveling (7) [see also (16)]. Reproductive leveling reduces the amount of between group variation needed for selection to favor group-beneficial but individually-costly traits. While it is not known how the estimates of genetic differentiation for small forager groups reported in (7) relate to Pleistocene foraging groups (see SI), it is intriguing to note that reproductive leveling itself already has strong hints of prosociality, begging the question of how it could evolve before altruism (17). This illustrates that for genetic selection to favor altruism in large residential groups, theorists need to invoke particularly strong assumptions. An alternative scenario is that human propensities to cooperate arose through selection on cultural rather than genetic variation (15, 18). Humans developed the capacity for complex culture perhaps beginning 250,000 years ago (19). Since that time, culturally transmitted traits have come, along side of genes, to have a large influence on human behavior. Ever since the advanced human capacity for social learning began, groups of individuals likely began rapid divergence in behavior due to cumulative cultural changes. This behavioral variation between groups can persist, given the right kinds of cultural evolutionary forces (20). Even among our closest living relatives, chimpanzees, plausibly socially-learned traits show some between group variation (21). Selection for culturally-prescribed altruists occurs through the same process as for genes: groups of altruists leave more daughter societies (8, 9). However, one advantage that cultural variation has over genetic is that it does not require violent inter-group competition, nor group extinctions (22, 23). If failed groups were incorporated routinely into successful ones, conformist transmission and other forms of resocialization of failed groups can lead to effective cultural selection on groups even though such a pattern will generate rates of migration that keep genetic FST very low between neighbors. Thus selection on culture can be powerful precisely when genetic selection at the group level is weakest. What is the scope for group-level selection on cultural variation and how does this compare to the equivalent for genes? A number of mechanisms may permit cultural variation to be larger than genetic variation between groups (15, 20). If these mechanisms are important, the scope for group-level selection on culture will be much greater than for genes. Here we compute estimates of cultural variation among human groups and compare these to previous estimates of genetic variation among groups. We restrict ourselves to neighboring groups in the main analysis since only neighbors could compete directly. Despite good reasons to believe our estimates of cultural variation are
Author contributions: A.V.B., P.J.R., and R.M. designed research; A.V.B. performed research; A.V.B. analyzed data; and A.V.B., P.J.R., and R.M. wrote the paper. The authors declare no conict of interest. This article is a PNAS Direct Submission.
1To

whom correspondence should be addressed. E-mail: avbell@ucdavis.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/ 0903232106/DCSupplemental.

PNAS October 20, 2009 vol. 106 no. 42 1767117674

SOCIAL SCIENCES

ANTHROPOLOGY

underestimates, we find much greater scope for multilevel selection on human culture than on human genes. These results call for attempts to produce better estimates relevant to quantitative models of human cultural evolution.
Calculating Cultural and Genetic Variation. The formal condition for

altruism to arise can be expressed using the Price equation (11, 13). Unlike most evolutionary theory, the Price equation is axiomaticit does not depend upon simplifying assumptions, but rather is an exact description of how selection works. Put in terms of regression coefficients and the statistic FST, a measure of genetic differentiation between populations (24), the condition for the frequency of altruism to increase is:

w g, p s 1 FST wig, pig FST

[1]

Here (wg, pg) is the increase in the mean fitness of the group with an increase the frequency of altruists, and (wig, pig) is the fitness decrease of the individual acquiring the altruistic allele. FST estimates the proportion of the total variation in a trait or set of traits (or alleles) that is accounted for by between-group differences. The greater the genetic differentiation (FST) between two groups, the greater the scope for selection at the group level. View the left hand side of the inequality as the benefit-cost ratio for the addition of another altruist in a population at the scale of the group and individual, respectively. The right-hand side of this inequality should be computed for two populations that may compete. There is no reason, in principle, not to use the same F-statistic, FST, for use in describing cultural differentiation between populations, because the derivation of the Price equation makes no assumptions about the nature of the underlying variants. As in the case of genetic FST, cultural FST is the proportion of the total variance in allele (or trait) frequencies found between groups. The higher this number, the greater the cultural differentiation is between groups. By comparing FST for genes and culture, we can assess the relative ability of either inheritance system to respond to group-level selection. An obstacle to computing cultural FST is that social anthropologists have not traditionally sampled individuals explicitly. The one exception known to us is analyzed in the SI and Table S2. Instead, most ethnography consists of statements about normative behavior that is often observed to vary among groups. To compute cultural FST, we require systematic samples of individual beliefs or behavior. We used the World Values Survey (WVS) (25) as a source of data to compute cultural FST for a fairly large sample of national neighbors. The WVS asks a large battery of questions that are likely to be heavily influenced by culture in a large number of countries. The sample size within countries is also large and thus favorable to calculating precise estimates of within and between group variation. We then compare these corresponding genetic FST estimates from (24). Results The top panel of Fig. 1 shows the distribution of FST estimates for culture and genes and the bottom panel shows how these estimates relate to equation (1) and the scope for selection among groups. It is evident that the scope for group-level selection, as described in equation (1), is much greater for culture than genes. Cultural FST (mean 0.0800, median 0.0660) between populations is more than order of magnitude larger than their corresponding genetic FST (mean 0.0053, median 0.0032). In Table S1 in the SI, we list all of the pairwise cultural FST values. The full tables for genetic FST are given in (24). In the case of both culture and genes, the similarity of neighbors is much greater than non-neighbors.
17672 www.pnas.orgcgidoi10.1073pnas.0903232106

Fig. 1. Comparison of genetic and cultural differentiation. Above: Histogram of 150 cultural FST (gray ll) and 59 genetic FST (black border) for neighboring countries calculated from the World Values Survey and in (24), respectively. Bottom: Plot of the cultural against genetic FST for 59 pairs of neighboring countries.

From these estimates we can compute the minimum group benefit over individual cost ratio that would favor altruism, the left-hand side of equation (1). For genes the mean and median benefit across all paired countries is 437 and 311 (range from 312,272), while the respective mean and median for cultural traits is 16 and 14 (range 375). For genes, group beneficial behaviors should be hundreds of times greater than the individual cost to be favored by selection, whereas for culture, grouplevel selection can operate under much less stringent conditions. Discussion Our calculations show much greater scope for cultural rather than genetic group-level selection, although we should acknowledge how this inference may be limited. The low and very low genetic FST values that characterize modern national neighbors might not be typical of ancestral Pleistocene populations. Certainly, much smaller population sizes would have generated more drift. On the other hand, we do not think that the available data from living populations is consistent with neighbors having FST values as high as 0.076, the baseline figure used in (7) (see SI). It is difficult to know how last Glacial population structures might have been like compared to Holocene hunter-gatherers. Human populations densities in most times and places in the Pleistocene were apparently very low. Highly variable climates and a disproportionate emphasis on big game hunting in the last glacial compared to the Holocene would probably have made populations more mobile and more prone to long distance movements in the last glacial. Even in Upper Paleolithic Western Eurasia last glacial populations were apparently only on the order of tens of thousands of people and division into markedly distinct ethnic groups was absent (26). The main Western Eurasian Upper Paleolithic cultures, the Aurignacian and Gravettian, occurred over the whole of Europe and neighboring West Asia without any strongly marked stylistically marked subdivision (27). The culturally innovative Southern African Still Bay and Howiesons Poort Middle Stone Age traditions appear to have been widespread spatially like the Upper Paleolithic but were more restricted in time than the Aurignacian and Gravettian (28). In most parts of the Old World for most of the history of Anatomically Modern Humans most populations made rather
Bell et al.

simple Middle Stone Age or Mode 3 tools (28). Population densities of Mode 3 toolmakers were probably even lower than Upper Paleolithic West Eurasians (29). We are frustrated to have to use estimates of FST derived from late Holocene populations to infer what transpired in Middle and Late Paleolithic contexts. The paleoclimatology and paleoanthropology suggest that Paleolithic populations were structured much differently than contemporary populations, even than the ethnographic and genetic samples we have from small-scale foraging populations. Nevertheless, the same differences in evolutionary processes that allow culture more easily than genes to maintain between-group differences should have operated in the Middle and Upper Paleolithic as well as in contemporary populations. The large differences we find between cultural and genetic FST argue that that the evolutionary processes acting on these two systems of inheritance would have to be very different in the late Pleistocene to affect the qualitative conclusion that the scope for cultural group-level selection was greater than for genetic group-level selection, then as now. The WVS is not the best dataset to use for this purpose. Ideally, we would like to have neighbor cultural FST estimates for small-scale societies as close as possible in structure to those that characterized our Pleistocene ancestors. One set of data from Africa does provide data sufficient to compute cultural FST (see SI and Table S2), but no genetic data are available for these groups (30). Cultural FST estimates reported here are likely the lower bound for ethnic groups as questionnaires typically underestimate behavioral variation across groups (31, 32). Also, most nations have multiple ethnic groups that live within their boundaries and different nations often have the same or similar subcultures as neighboring nations. Some of the variation in the WVS questions may be genetic, as behavior geneticists often report appreciable (genetic) heritabilities for seemingly cultural traits like political preferences (33). As we see above, genetic FST is generally much smaller than cultural FST so a mixture of genetic and cultural effects will lead to an underestimate of cultural FST. The discussions by anthropologists of differences between tribal scale societies that presumably most resemble the late Pleistocene conditions under which our propensities for cooperation arose suggest that cultural differences between tribes were roughly similar to those obtaining between ethnic groups in modern societies [e.g., (34)]. For example, different tribes often have different languages or dialects that may function to limit communication between them (35). As with modern societies, immigration into simple societies was often accompanied by cultural assimilation [e.g., (36)]. Thus, national scale data offer some interesting insights on neighbor cultural and genetic differences. Finding that there is greater scope for inter-group competition to select for prosociality on culture rather than genes does not mean that genes are unimportant to the story. With our early ancestors inheriting both cultural and genetic variants, one inheritance system likely exerted pressure on the other. Support for gene-culture coevolution for well-studied physiological traits can be logically extended to the puzzle of human prosociality. Cattle domestication and the innovation of dairy farming led to selection pressure on genes to produce the enzymes to break down milk sugars beyond weaning (37). Similarly, innate propensities to cooperate might have evolved by gene-culture coevolution rather than by selection among groups for solely genetic variants (38). That is, the evolution of cultural rules mandating cooperation between group members could exert ordinary selection pressures for genotypes that obey cultural rules. Social selection mechanisms such as exclusion from the marriage market, denial of the fruits of cooperative activities, banishment, and execution would have exerted strong selection against genes tending toward antisocial behavior. Social selecBell et al.

tion in favor of genes that predisposed individuals toward prosociality are also easy to imagine. One important function of our calculation is to call attention to the importance and feasibility of quantitatively estimating the important parameters of evolutionary models in human populations. Data from small-scale human societies that better resemble the kinds of social systems important in our evolutionary past would be particularly interesting. (See SI for an analysis of four east-African populations). A conjecture based on ethnographic accounts [e.g., (39)] is that cultural selection among groups is often driven by differences in institutions. One welldescribed case, the Nuer versus Dinka tribes, exemplifies contest-based selection acting on institutional cultural variation among groups. Differences in marriage institutions led to a deeper reckoning of kinship among the Nuer compared to the Dinka, which led to the Nuer raising larger fighting forces and the expansion of the Nuer at the expense of the Dinka. These fights were not genocidal and many defeated Dinka families were incorporated into Nuer tribes, complete with fictive descent to give them a place in the Nuer social order (36). In any case, the Nuer and Dinka peoples were genetically similar, as neighbors would typically have been before long distance mass transport was available. The way the losing side was integrated demographically, and re-socialized culturally, into Nuer society further reduced the possibility of maintaining genetic variation between the groups. Thus the genetic variation between groups in such contexts may be quite small, while cultural variation can remain quite large. The Nuer-Dinka case illustrates this feature of human intergroup competition. The Nuer-Dinka difference was institutional. Human social life is typically regulated by rules of conduct that take the form of self-reinforcing games (40). Few people will behave contrary to the institution and the non-conformists that do exist will not usually be adhering to the institutions of a neighboring group. Thus selection acts on competing equilibria as far as the traits upon which group-level selection is acting. The situation is as if cultural FST was approximately 1 as far as the relevant trait is concerned, as if each population was nearly monomorphic for the same haploid allele. We conclude that the available evidence suggests that direct group-level selection on genes played a smaller role in human evolution than group-level selection on cultural variation. Human genes affecting social behavior, such as our docility compared to chimpanzees (41), probably arose by gene-culture coevolution (42). Materials and Methods
Using data from four phases of the WVS, we computed a pair-wise cultural FST for 154 neighboring pairs of countries sampled therein. We restrict ourselves to neighboring countries in the main analysis since only neighbors could compete directly. We matched these estimates to available estimates of genetic FST from a database of calculated genetic differentiation between nations and ethnic groups (24). For discrete traits, questions in the World Values Survey (WVS) were regarded as loci and responses as alleles. In the WVS, data such as no answer or not asked and the like were not included in the calculation because they were not considered responses. Questions that explored personal idiosyncrasies rather than cultural beliefs were omitted (e.g., selfreported state of health). Further, similar responses were pooled as one response if the choice of responses could not be considered as cardinal. For example, if the possible responses were agree, strongly agree, disagree, and strongly disagree, then the rst two and the last two would be combined as one response. Detailed choice of questions and pooling of responses for all questions in the WVS is available upon request from the corresponding author. For these discrete traits, following (24), cultural FST was computed as follows. For a locus with L alleles and pij is the frequency of allele i in population j,
PNAS October 20, 2009 vol. 106 no. 42 17673

SOCIAL SCIENCES

ANTHROPOLOGY

FST,i
where

varpi p i 1 p i)

is the between group variance in variance in allele frequencies. Across all loci, the FST is

p i

FST

s j1 n j p ij s j1 n j

L i 1 i1 p L i 1 i1 p

p i F ST,i p i

is the average allele frequency across s populations weighted by sample size (n) and
s j1

Cardinal responses in the WVS were treated as quantitative characters, and for each locus (question), an FST was computed from the ratio of the between group (Vg) and total variance (VT), FST Vg/VT. The mean FST across all loci is the reported FST between a pair of populations. A full table of paired bordering countries with their genetic and cultural FST is found in the SI. ACKNOWLEDGMENTS. We thank Samuel Bowles and three anonymous reviewers for commenting on previous versions of this paper. This work was supported in part by a University of California Davis block grant (to A.V.B.) and National Science Foundation award 0340148.

var pi

pij p i 2 s 1

1. Boyd R, Richerson PJ (2006) Culture and the evolution of human social instincts. In Roots of Human Sociality: Culture, Cognition, and Interaction, eds Eneld NJ, Levinson SC (Berg, New York). 2. Darwin C (1874) in The Descent of Man and Selection in Relation to Sex (American Home Library, New York), 2nd Ed. 3. Hamilton WD (1975) Innate social aptitudes of man: An approach from evolutionary genetics. In Biosocial Anthropology, ed Fox R (Wiley, New York). 4. Alexander RD (1987) in The Biology of Moral Systems (Aldine de Gruyter, Hawthorne, NY), p 301. 5. Eibl-Eibesfeld I (1982) Warfare, mans indoctrinability, and group selection. Zeitschrift fur Tierpsychologie 67:177198. 6. Wilson EO (1975) in Sociobiology: The New Synthesis (Harvard Univ Press, Cambridge MA), p 697. 7. Bowles S (2006) Group competition, reproductive leveling, and the evolution of human altruism. Science 314:1569 1572. 8. Boyd R, Gintis H, Bowles S, Richerson PJ (2003) The evolution of altruistic punishment. Proc Natl Acad Sci USA 100:35313535. 9. Boyd R, Richerson PJ (1985) in Culture and the Evolutionary Process (University of Chicago Press, Chicago). 10. Grafen A (1985) A geometric view of relatedness. In Oxford Surveys in Evolutionary Biology, eds Dawkins R, Ridley M (Oxford Univ Press, New York), Vol 2, pp 28 90. 11. Price GR (1970) Selection and covariance. Nature 227:520. 12. Hamilton WD (1963) The evolution of altruistic behavior. Am Nat 97:354. 13. Frank SA (1998) in Foundations of Social Evolution (Princeton Univ Press, Princeton). 14. Wilson DS, Sober E (1994) Reintroducing group selection to the human behavioral sciences. Behav Brain Sci 17:585 654. 15. Henrich J (2004) Cultural group selection, coevolutionary processes and large-scale cooperation. J Econ Behav Organ 53:335. 16. Boehm C (1997) Impact of the human egalitarian syndrome on Darwinian selection mechanics. Am Nat 150:S100 S121. 17. Boyd R (2006) The puzzle of human sociality. Science 314:15551556. 18. Boyd R, Richerson PJ (1982) Cultural transmission and the evolution of cooperative behavior. Hum Ecol 10:325351. 19. McBrearty S, Brooks AS (2000) The revolution that wasnt: A new interpretation of the origin of modern human behavior. J Hum Evol 39:453563. 20. Richerson P, Boyd R (2005) in Not by Genes Alone: How Culture Transformed Human Evolution (University of Chicago Press). 21. Whiten A, et al. (1999) Cultures in chimpanzees. Nature 399:682 685. 22. Boyd R, Richerson PJ (2002) Group benecial norms can spread rapidly in a structured population. J Theor Biol 215:287296. 23. Boyd R, Richerson PJ (2009) Voting with your feet: Payoff biased migration and the evolution of group benecial behavior. J Theor Biol 257:331339.

24. Cavalli-Sforza LL, Menozzi P, Piazza A (1996) in The History and Geography of Human Genes (Princeton Univ Press, Princeton, New Jersey). 25. European Values Study Group and World Values Survey Association (2006) European and World Values Surveys Four-Wave Integrated Data File, 19812004, v.20060423. 26. Bocquet-Appel JP, Demars PY, Noiret L, Dobrowsky D (2005) Estimates of Upper Palaeolithic meta-population size in Europe from archaeological data. J Archeol Sci 32:1656 1668. 27. Klein RG (1999) The Human Career: Human Biological and Cultural Origins (University of Chicago Press, Chicago), 2nd Ed, p 810. 28. Jacobs Z, et al. (2008) Ages for the Middle Stone Age of Southern Africa: Implications for human behavior and dispersal. Science 322:733735. 29. Henrich J (2004) Demography and cultural evolution: How adaptive cultural processes can produce maladaptive losses - The Tasmanian case. Am Antiquity 69:197214. 30. Edgerton RB (1971) in The Individual in Cultural Adaptation; A Study of Four East African Peoples (University of California Press, Berkeley). 31. Cohen D (2007) Methods in cultural psychology. In Handbook of Cultural Psychology, eds Kitayama S, Cohen D (Guilford Press, New York). 32. Heine SJ, Norenzayan A (2006) Toward a psychological science for a cultural species. Perspect Psychol Sci 1:251269. 33. Alford JR, Funk CL, Hibbing JR (2005) Are political orientations genetically transmitted? Am Polit Sci Rev 99:153167. 34. Jorgensen JG (1980) in Western Indians: Comparative Environments, Languages, and Cultures of 172 Western American Indian Tribes (W. H. Freeman, San Francisco). 35. McElreath R, Boyd R, Richerson PJ (2003) Shared norms and the evolution of ethnic markers. Curr Anthropol 44:122129. 36. Kelly RC (1985) in The Nuer Conquest: The Structure and Development of an Expansionist System (University of Michigan Press, Ann Arbor), p 320. 37. Beja-Pereira A, et al. (2003) Gene-culture coevolution between cattle milk protein genes and human lactase genes. Nat Genet 35(4):311313. 38. Richerson PJ, Boyd R, Henrich J (2003) Cultural evolution of human cooperation. In Genetic and Cultural Evolution of Cooperation, ed Hammerstein P (MIT Press, Berlin). 39. Kelly RC (1985) The Nuer Conquest: The Structure and Development of an Expansionist System (University of Michigan Press, Ann Arber). 40. Greif A (2006) in Institutions and the Path to the Modern Economy: Lessons from Medieval Trade (Cambridge Univ Press, Cambridge). 41. Simon HA (1990) A mechanism for social selection and successful altruism. Science 250:16651668. 42. Richerson PJ, Boyd R (1998) The evolution of human ultrasociality. In Indoctrinability, Ideology, and Warfare, eds Eibl-Eibesfeldt I, Salter FK (Berghahn Books, New York).

17674 www.pnas.orgcgidoi10.1073pnas.0903232106

Bell et al.

Supporting Information
Bell et al. 10.1073/pnas.0903232106
SI Text
Genetic Evidence of Differentiation. Our mean genetic distance is

more than an order of magnitude smaller than that reported in Bowles study (1) where non-neighbors were included in the analysis. The source data available to (1) is not sufficient to recompute his FST values for nearest neighbors only. One of Bowles primary sources (2) does report an autocorrelation analysis suggesting that some of the genetic variation in Central Asia and Siberia is clinal. Those data refer to Y chromosome variation. The FST estimates in (1) derived from (2) are corrected for the higher rates of drift for Y (1/4 as many copies in a population as autosomal genes) compared to autosomal loci, but (apparently) not for their higher rates of mutation. Mitochondrial DNA and Y chromosome variants are used to trace the recent ancestry of populations because high rates of mutation lead to many new population specific alleles on time scales of a few thousand years. From the multidimensional scaling analysis in (2), it is apparent that most near neighboring groups are closer genetically that the average populations. Some of the populations in (2) that are most different genetically appear to have no close neighbors in the sample. Groups that tend to have high pairwise FST, such as Lapps in the European Arctic and Chukchi in Eastern Siberia, have probably not been neighbors for many millennia (3). Similarly, when distantly related groups come into contact, migration often lowers genetic distances. For example, in (3, Table 4.11.1), the FST for neighboring Siberian Eskimo and Chukchi is 0.025 although the Chukchi-Alaskan-Eskimo figure is 0.056 and that for more distant North American populations is even higher. The mean FST for neighboring groups in (3, Table 4.11.1) is 0.041 (but on p252 the summary figure for all of N Asia is given as 0.0264!!). The value for Indigenous circumpolar Eurasian populations of 0.076 in (4) appears to be a genetic variance, rather than an inbreeding coefficient which would be a smaller value. Two Wahlund F estimates are given in (5), 0.0067 and 0.0028, based on different subdivisions of the !Kung specified by the authors. The estimate in (6) was not corrected for biases due to sample size. It is not always clear in source material whether corrections for sample sizes have been applied in calculations of FST. Samples from arctic foragers and other small societies that presumably most resemble Pleistocene populations are often very small, as is reflected in standard errors of the estimates for many pairs of populations in (3, see e.g., Table 4.11.1). Summary data reported in (23, p 252) report that the average genetic FST for Arctic populations, mostly derived from small foraging populations similar to the populations used in (1), to be
1. Bowles S (2006) Group competition, reproductive leveling, and the evolution of human altruism. Science 314:1569 1572. 2. Karafet TM, et al. (2002) High levels of Y-chromosome differentiation among native Siberian populations and the genetic signature of a boreal hunter-gatherer way of life. Hum Biol 74:761789. 3. Cavalli-Sforza LL, Menozzi P, Piazza A (1996) in The History and Geography of Human Genes (Princeton Univ Press, Princeton, New Jersey). 4. Rychkov YG, Sheremeteva VA (1980) The genetics of circumpolar populations of Eurasia related to the problem of human adaptation. In The Human Biology of Circumpolar Populations, ed Milan FA (Cambridge Univ Press, Cambridge, New York), p 381.

0.0264 0.0125. The variograms presented in (3, p122124) necessitate that the FST between neighboring groups be far smaller than this average level.
Comparison Between Cultural and Genetic FST. (see Table S1) Cultural FST for four small-scale societies. Twenty-one questions

from (7) on four east-African tribes is used to compute a pairwise cultural FST (Table S1).
Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y

Would you rather own good farming land but no cattle, or good cattle but no farming land? p 169 Is it better to have many friends or many kinsmen (who are not clansmen)? p 174 Under what circumstances can a younger brother tell an older brother that he is wrong? P 178. Responses to values picture 7: Woman watched by man, mention of sex or not, p 185 Responses to values picture 8: Man and woman together inside a house, p 186 How does a psychotic person behave, p 188 What is the most important thing for a young woman to know before marriage? p 60 Do wives obey their husbands and is it right for them to do so? p 143 What makes a man a good friend? p 141 How do people feel about a rich man? p 137 Under what circumstances can a young adult man tell a mzee (old man) that he is wrong, p 135 Do the [population] prefer sons or daughters, and why? p 322 appendix What is the most important thing for parents to teach a toddler? p 322 appendix What is the worst thing that can happen to a man? p 322 Do unmarried females have sexual relations and is it right? p 323 appendix What should a man do if three of his cattle suddenly die? p 323 appendix If a man could have anything he wanted, what would he choose? p 323 appendix To be considered wealthy, what must a [population] own? P 342 appendix Whom can a person trust? p 324 appendix Is there one person you can trust beyond all others? p 325 appendix Why do people kill themselves? p 325 appendix

(see Table S2)

5. Harpending H, Jenkins T (1974) !Kung population structure. In Genetic Distance, eds Crow JF, Denniston C (Plenum Press, New York). 6. Harpending H, Jenkins T (1973) Genetic Distance Among Southern African Populations. In Methods and Theories of Anthropological Genetics, eds Crawford MH, Workman PL (University of New Mexico Press, Albuquerque). 7. Edgerton RB (1971) in The Individual in Cultural Adaptation; A Study of Four East African Peoples (University of California Press, Berkeley).

Bell et al. www.pnas.org/cgi/content/short/0903232106

1 of 2

Table S1. Genetic and cultural FST estimates reported in this study
Genetic FST Albania Albania Albania Albania Algeria Argentina Argentina Argentina Armenia Armenia Australia Austria Austria Austria Austria Austria Austria Austria Azerbaijan Azerbaijan Azerbaijan Bangladesh Belarus Belarus Belarus Belarus Belarus Belgium Belgium Belgium Belgium Brazil Brazil Brazil Brazil Bulgaria Bulgaria Bulgaria Bulgaria Bulgaria Canada Chile China China China China China China China Colombia Colombia Croatia Croatia Croatia Croatia Croatia Czech Republic Czech Republic Czech Republic Denmark Denmark Denmark Denmark Denmark Greece Italy Macedonia, Republic Of Serbia And Montenegro Morocco Brazil Chile Uruguay Georgia Turkey New Zealand Czech Republic Germany Hungary Italy Slovakia Slovenia Switzerland Armenia Georgia Russian Federation India Latvia Lithuania Poland Russian Federation Ukraine France Germany Great Britain Netherlands Colombia Peru Uruguay Venezuela Greece Macedonia, Republic Of Romania Serbia And Montenegro Turkey United States Peru India Japan Pakistan Philippines Republic Of Korea Taiwan Province Of China Viet Nam Peru Venezuela Bosnia And Herzegovina Hungary Italy Serbia And Montenegro Slovenia Germany Poland Slovakia Germany Germany West Great Britain Norway Sweden 0.024 0.02 0.0036 0.0019 0.004 0.0043 0.0012 0.0032 0.0015 0.0015 0.0012 0.00174 0.0315 0.00353 0.0052 0.0064 0.0016 0.0016 0.0036 0.0019 Cultural FST 0.0923 0.0849 0.0571 0.0830 0.0828 0.0509 0.0323 0.0534 0.0484 0.1116 0.0251 0.0842 0.0329 0.0852 0.0415 0.0597 0.0651 0.0518 0.0323 0.0439 0.0666 0.1144 0.0395 0.0539 0.1036 0.0305 0.0212 0.0329 0.0428 0.0391 0.0535 0.0660 0.0469 0.0966 0.0859 0.0725 0.0547 0.0778 0.0437 0.1868 0.0206 0.0455 0.1548 0.1631 0.2545 0.1320 0.1568 0.1165 0.0835 0.0747 0.1531 0.0578 0.0772 0.0355 0.0604 0.0543 0.0692 0.1371 0.0459 0.0708 0.0678 0.0710 0.0590 0.0458

Bell et al. www.pnas.org/cgi/content/short/0903232106

2 of 2

Genetic FST Dominican Republic Estonia Estonia Estonia Estonia Finland Finland Finland France France France France France Georgia Georgia Germany West Germany West Germany West Great Britain Great Britain Greece Greece Greece Hungary Hungary Hungary Hungary Hungary Iceland Iceland Iceland India India Indonesia Iran (Islamic Republic Of) Iran (Islamic Republic Of) Iran (Islamic Republic Of) Iran (Islamic Republic Of) Iraq Iraq Iraq Ireland Ireland Israel Italy Italy Italy Italy Italy Japan Japan Japan Latvia Latvia Latvia Latvia Lithuania Lithuania Luxembourg Luxembourg Luxembourg Macedonia, Republic Of Malta Mexico Morocco Puerto Rico Finland Latvia Russian Federation Sweden Norway Russian Federation Sweden Germany Great Britain Italy Spain Switzerland Russian Federation Turkey Belgium France Great Britain Germany Northern Ireland Italy Macedonia, Republic Of Turkey Romania Serbia And Montenegro Slovakia Slovenia Ukraine Great Britain Ireland Northern Ireland Kyrgyzstan Pakistan Philippines Iraq Pakistan Saudi Arabia Turkey Jordan Saudi Arabia Turkey Great Britain Northern Ireland Jordan Bosnia And Herzegovina France Malta Slovenia Switzerland Republic Of Korea Russian Federation Taiwan Province Of China Belarus Lithuania Russian Federation Sweden Poland Sweden Belgium France Germany Serbia And Montenegro Italy United States Spain 0.0064 0.0153 0.0082 0.0027 0.0024 0.0034 0.0039 0.0023 0.0015 0.0027 0.0022 0.0022 0.0077 0.0136 0.0099 0.00341 0.00168 0.00321 0.00075 0.00116 0.00373 0.00112 0.0034 0.00044 0.0137

Cultural FST 0.0555 0.1059 0.0268 0.0375 0.1432 0.0592 0.1501 0.0658 0.0614 0.0487 0.0670 0.0600 0.0644 0.0634 0.1111 0.0477 0.0698 0.0464 0.0732 0.0579 0.0407 0.0603 0.2498 0.0842 0.0746 0.0585 0.0649 0.0684 0.0653 0.0926 0.1100 0.0579 0.1503 0.1144 0.0997 0.1148 0.1149 0.1830 0.0456 0.0662 0.0999 0.0645 0.0229 0.2083 0.0536 0.0670 0.1240 0.0654 0.0695 0.0954 0.1297 0.0871 0.0395 0.0471 0.0456 0.1573 0.0756 0.1925 0.0132 0.0264 0.0406 0.0509 0.1240 0.0661 0.2211

Bell et al. www.pnas.org/cgi/content/short/0903232106

3 of 2

Genetic FST Netherlands Netherlands Netherlands Netherlands Norway Norway Philippines Philippines Poland Poland Poland Poland Portugal Puerto Rico Puerto Rico Republic Of Moldova Republic Of Moldova Romania Romania Russian Federation Russian Federation Serbia And Montenegro Singapore Slovakia South Africa Spain Sweden Sweden Switzerland Taiwan Province Of China Uganda United States Belgium Germany Germany West Great Britain Great Britain Sweden China Viet Nam Germany Slovakia Sweden Ukraine Spain United States Venezuela Romania Ukraine Serbia And Montenegro Ukraine Turkey Ukraine Bosnia And Herzegovina Indonesia Ukraine Zimbabwe France Germany West Norway Germany Philippines Tanzania, United Republic Of Great Britain 0.0012 0.0016 0.0016 0.0017 0.0025 0.0018 0.0315 0.0047 0.0082 0.0048 0.0039 0.0039 0.0018 0.001

Cultural FST 0.0535 0.0794 0.0406 0.0544 0.0688 0.0439 0.1320 0.1689 0.1607 0.0693 0.2116 0.0644 0.0456 0.0563 0.0591 0.0589 0.0322 0.0850 0.0659 0.1705 0.0139 0.0600 0.1131 0.0568 0.0525 0.0600 0.0565 0.0439 0.0680 0.1372 0.0729 0.0777

Bell et al. www.pnas.org/cgi/content/short/0903232106

4 of 2

Table S2. Estimates for cultural FST derived from (7) on the lower diagonal, with bootstrapped standard errors in the upper diagonal
Hehe Hehe Kamba Pokot Sebei 0.113 0.123 0.087 Kamba 0.003 0.109 0.099 Pokot 0.004 0.004 0.105 Sebei 0.003 0.003 0.004

Bell et al. www.pnas.org/cgi/content/short/0903232106

5 of 2