The cultural wealth of nations

Humans are the virtuosos of culturaldiversity. We fish, hunt, shepherd,forage and cultivate. We practise

polygyny, polyandry and monogamy, paybride-prices and dowries, and have patri-lineal and matrilineal wealth inheritance.We construct or inhabit all manner of shel-ters, speak about 7,000 different languagesand eat everything from seeds to whales.And this is not counting many unique, andsometimes bizarre, belief systems andbehavioural practices1.

Contemporary cultural diversity is prob-ably a fraction of the diversity that has everexisted, and there may be less cultural diver-sity now than there was 10,000 years ago, justbefore the advent of agriculture. Because of their superior reproductive rate, agri-culturalists replaced many indigenoushunter-gatherer cultures as they spreadacross Europe and other parts of the globe2,3.

If the picture of human cultures is one ofvariability, the human genetic landscape isone of homogeneity. All of humanity variesless genetically than does a typical wild pop-ulation of chimpanzees4,5. This may reflectour youthfulness as a species. Anatomicallymodern Homo sapiens emerged only about75,000–100,000 years ago, and may have suffered a demographic ‘bottleneck’ in therecent past, meaning that in evolutionaryterms we are all descended from a not-so-distant common ancestor.Also,of course,wecan interbreed throughout our entire world-wide range. Add the facts that we regularlytrade, migrate across each other’s territoriesand wage war against each other,and a puzzleemerges: where does our extreme culturaldiversity come from,and what maintains it?

The answers can perhaps be found inthinking about human cultures as if they arecollections of distinct biological species. Justas species carry genetic adaptations to theirenvironments, we believe that cultural adap-tations have evolved in response to social life,and that such adaptations work to maintaincultural identity and coherence. Like speciesthat do not interbreed, human cultures aresurprisingly resistant to influences fromother cultures and often act as barriers togene flow. Important elements of cultureshow dominantly vertical modes of inheri-tance, much like the vertical transmission ofgenetic information. Collections of differentspecies are observed to distribute themselvesgeographically in accordance with clear law-like patterns, and so, too,do human cultures.At a psychological level, humans displayforms of social behaviour conducive to livingin small groups, such as rewarding coopera-tion, punishing those who deviate fromnorms,and being wary of outsiders.

Cultures and speciesBut what is culture? This is one of thoseconcepts that resists easy definition, eventhough most of us know which culture webelong to and who its other members are.We will use a simple definition that linkscultures to distinct language groups (Box 1,overleaf), but it almost certainly underesti-mates their true diversity.

To see how human cultures partition theworld like so many species, one need onlylook to linguistic diversity.Language–culturegroups are not evenly distributed. Some 700to 1,000 different languages,about 15% of thetotal on Earth, are spoken in the 312,000

square miles of the island of New Guinea6.In the northwest coastal areas a different language may be spoken every few miles.Similar densities of languages can be foundon many Pacific island archipelagoes. Bycomparison, about 500 different languageswere spoken in all of North America at thetime of European contact, and only about 90are spoken in China despite its vast popula-tion and area about 12 times the size of NewGuinea. What determines variation in thedensity (numbers per unit area) of cultures?

One powerful factor is the environment.By 1953, the anthropologist J. B. Birdsell7

had documented the greater density ofAustralian Aboriginal tribal groups in wetterareas, and others have since reported highercultural densities in coastal and equatorialregions8. In animals, a trend called Rapa-port’s rule9 holds that the density of animalspecies is highest in equatorial regions anddeclines steadily towards the poles. Polarregions support less biodiversity.

Some years ago, we tested Rapaport’s rulein humans using information on NativeAmerican tribal groups at the time of Euro-pean contact10. For each line of latitude, werecorded the number of languages spoken,and plotted its trend (Fig. 1, overleaf). Thenumber of languages (human cultures)found per unit area is high in the lower latitudes and declines as one moves north-wards. At extreme northern latitudes only ahandful of different cultures inhabit theentire east–west expanse of northern Canada.An almost identical trend emerges in thediversity of different mammal species with latitude11,12. In both cases the declines indiversity are matched by increasing sizes of

news and views feature

The cultural wealth of nationsMark Pagel and Ruth Mace

Why, when the human race shows comparatively little genetic variation, arecultural differences so widespread and enduring? Thinking about cultures interms of biological species provides some provocative answers.

NATURE | VOL 428 | 18 MARCH 2004 | www.nature.com/nature 275

PIC

TU

RE

CR

ED

ITS

PAG

E 2

78

18.3 Pagel NV feature MH 12/3/04 5:49 pm Page 275

© 2004 Nature Publishing Group

276 NATURE | VOL 428 | 18 MARCH 2004 | www.nature.com/nature


geographical ranges. Northerly dwelling animals are found over a large area,and so arenortherly dwelling human cultural groups.Asingle species — humans — carved up NorthAmerica as might 500 different species ofmammal. Studies of African language–culture groups also find that biological andcultural diversity co-vary13.

Ecologists explain the trend in speciesdiversity by noting that northern regionsexperience wider annual variation in cli-mate. This favours species that are ecologicalgeneralists, able to cope with a range of con-ditions. The reverse is true for equatorialregions, where the relative annual stabilityselects for specialization in a particular habi-tat or niche. This explanation may be correctfor animal species11 but cannot work forhumans. Apart from being just a singlespecies, we are unequalled generalists,having colonized virtually the entire Earththrough our ability to build fires, makeclothes and shelters, and eat a variety offoods.

We believe that a more fundamental ten-dency accounts for the cultural patterns withlatitude. The low cultural diversity seen innortherly latitudes is understandable, evenexpected. In this relatively unproductiveenvironment, individuals must range overlarge areas to eke out a living.This movementof people tends to homogenize culture andlanguage. The puzzle is to explain the highercultural densities in more productive areas.Why do humans not just form one large andhomogeneous cultural group in ecologicallyricher areas such as New Guinea?

It may be that human subpopulationscontinually split off from larger groups, andform around defensible resources — be theytracts of forest or fishing grounds — becauseindividuals seek to control those resources.Similarities between human cultural andmammalian species diversity may thenemerge because the same ecological factorsmake it more or less likely that small groupscan be viable. Over time, these self-sufficient

groups emerge as daughter cultures and then as fully fledged different societies withtheir own customs, specializations andbehavioural rules. The speed of cultural evolution makes this plausible. Even a new language can emerge after as few as 500 years of isolation. But we also suggestthat it happens because human cultures have evolved mechanisms for maintainingtheir separation from other cultures.

Cultures and gene flowFor cultural variation to arise in the way wesuggest, cultures must act to exclude eachother. Anthropologists and linguists canconstruct family trees of major languagegroups based on the similarities betweenthe languages. These trees — called phylo-genies — chart the genealogical relation-ships between language groups. Since thelate 1980s it has been known that phylo-genetic trees of major language groups correspond closely to phylogenetic treesconstructed from human genetic markers14.This result is consistent with our view, yet it may not be surprising: when, owing to

isolation for whatever reasons, peoplediverge from one another on backgroundgenetic markers, their languages and cultures also tend to drift apart.

But studies of European genetic diversityshow that language differences may reducegenetic exchange between populations.Robert Sokal15,16 and colleagues measuredthe frequency of 63 genetic markers in samples taken from 3,119 different locationsacross Europe. They then applied themethod of ‘wombling’ (after W. H.Womble17) to measure the rate of change ingene frequencies between these differentlocations. Standard genetic theory tells usthat if people diffuse over an area such asEurope,then no boundaries of abrupt geneticdifference are expected. However, barriers tothe movement of people can produce zonesof abrupt change.

Sokal et al. identified 33 boundaries sepa-rating areas of especially sharp changes ingene frequencies across Europe, and drewthem on a map (Fig. 2). Twenty-two of theboundaries correspond to physical bound-aries such as the Alps or the English Channel.They are also linguistic boundaries. For afurther 11 boundaries, no physical or otherbarrier could be detected, and yet nine ofthese correspond to linguistic boundaries. Inall, then, language is associated with regionsof abrupt genetic change in 31 of the 33 cases.Not all studies support the separation ofgenes and language18, but it does seem thathumans are more likely to mate with peoplethey can talk to! The consequence is that cultural differences become self-reinforcing,prompting further cultural divergence.

Cultural transmissionCultures also separate themselves fromother cultures by the ways in which theytransmit cultural information. Convention-ally, we can describe vertical and horizontalmodes of transmission of both genetic andcultural traits. The dominant mode of

0

50

100

150

200

No.

of s

pec

ies

0

5

10

15

20

25

30

35

No.

of l

angu

ages

0

1

2

3

4

5

Sp

ecie

s/ar

ea

0

0.1

0.2

0.3

0.4

0.5

0.6

Lang

uage

s/ar

ea

0 10 20 30 40 50 60 70 80 90

Latitude (°N)

Species/areaLanguages/area

LanguagesSpecies

A definition of culture hasproved elusive, and yet thisdoes not make the conceptunusable. Biologists havesimilar difficulties definingspecies. The biological-speciesconcept proposes that groupsof animals form distinct specieswhen they cannot interbreedsuccessfully. Many apparentlydistinct species — such ashorses and zebras or wolvesand dogs — can interbreed.But what makes the concept

useful nonetheless is that thesespecies normally do notinterbreed if other options areavailable.

In this article we adopt asimilarly pragmatic approach tocultures. Collins EnglishDictionary, in line with othersources (see refs 27, 28),defines a culture as “the sumtotal of a set of shared beliefs,values, practices”. Preciselywhat is shared and what is notwill never be easy to document,

and, like species, not everyonewill operate solely within theirculture.

A rough guide to thenumber of different culturalgroups can be obtained bytaking them to be synonymouswith distinct language groups,for which classification we rely on standard handbooks ofthe world’s languages29. Thisdefinition almost certainlyunderestimates the true numberof distinct cultures. M.P. & R.M.

Box 1Culture and language

Figure 1 Language, mammals and Rapaport’srule. Top, Numbers of languages and ofmammal species at each degree of northlatitude in North America. The trends reflectthe shape of the continent, being narrow inthe south regions and growing wider athigher latitudes. Both trends peak at about40° N, where North America is about 3,000miles wide. (Data from refs 10 and 11.)Bottom, Densities of languages and mammalspecies, calculated as the number of either ata given latitude divided by the area of thecontinent for a 1° latitudinal slice at eachlatitude. Having controlled for area, bothtrends show the decline in densitycharacteristic of Rapaport’s rule9. Similarlatitudinal trends for language and speciesdiversity have been reported for Africa13.



transmission for both kinds of trait is verti-cal: parents transmit genes to offspring, andcultural traits are transmitted down genera-tions by parents, elders and teachers. Nearlyall of us speak our parents’ language, andpolitical and religious beliefs are surpris-ingly stable over generations. Humans areespecially likely to copy common traits, atendency that works to reduce variationwithin a culture but increase it between cultures. Genetic information is also some-times transmitted horizontally, such aswhen a virus or other vector carries geneticmaterial from one species to another.Normally, however, species barriers to interbreeding prevent large-scale horizontaltransfer of genetic information, and it is anoddity when it is observed. We share morethan 98% of our genes with chimpanzeesbecause our genome is largely a closed shopto other genomes.

Barriers to horizontal transmission ofculture are, in principle, far weaker. Hori-zontal transmission may sometimes beimposed, as when one culture conquersanother. More prosaically, however, culturaltraits can diffuse among geographically closeneighbours, as when one culture borrows oradopts words, ideas, customs and technolo-gies from another.The English language is anexample, with about half of its vocabularybeing of Germanic origin (vertical transmis-sion) and half of Romance origin, reflecting,among other forces, the Norman conquest ofEngland in the eleventh century. Of course,cultures do not always survive contact withone another, because a politically, techno-logically or economically dominant culturefrequently displaces a weaker one.

It is revealing for our argument to con-sider how cultures treat events of horizontaltransmission.Whereas vertical transmissionof cultural traits goes largely unnoticed,horizontal transmission is far more likely tobe regarded with suspicion or even indigna-tion. France, for example, devotes a ministryto slowing or banning what is portrayed as anoverwhelming march of English words, cus-toms and phrases into the French languageand culture. While the other nations of theworld work away on their computers, theFrench sit at their ordinateur. The British aresimilarly alert to what they perceive to be‘Americanisms’. In both cases the home pop-ulation is seen as an unwilling victim. How-ever, a recent study of the English languagefound that it had admitted at least 90,000new meanings over the past century, withonly 5% derived from borrowings19. If thereis an enemy, it is within. Cultures, it seems,like to shoot messengers.

These anecdotal accounts can be placedupon a firmer footing. If horizontal trans-mission is strong, cultures should tend to share their traits with those of their nearest geographical neighbours. If ver-tical transmission predominates, cultures

should tend to have the traits of the culturesfrom which they descend; that is, oftheir ancestral culture (Box 2, overleaf).Although geography and ancestry are oftencorrelated, a statistic known as the Manteltest20 can distinguish the two hypotheses.In astudy of 47 cultural traits in 277 African soci-eties, most traits examined, especially thoseaffecting means of subsistence, family struc-ture and kinship (traits closely associatedwith reproductive success), were conservedover generations21. The only traits to showclustering that was dominantly geographicalpertained to sexual division of labour.Another investigation, using a worldwidesample of cultures, found that subsistencepractices, mating systems and sexual divi-sion of labour flowed down generations22.These results are striking: even under theinfluence of close geographical neighbours,cultures can remain stable and coherentunits. Borrowing and imitation certainlyoccur. But cultural evolution is not a free-for-all in which all traits become equallyavailable for adoption each generation.

Extreme socialityHuman cultures are not as impermeable tooutside influences as genomes, but they doerect barriers to the movement of peopleand ideas. What is so valuable as to call forthis much protection? In a word, the answeris the group itself. It seems likely thathuman evolution in general, and humancultural evolution in particular, is dis-tinguished by its sophisticated group behaviour. Historically, groups have hadcooperative jobs to do, such as huntinglarge prey or warring with other humangroups for access to territory or mates.These jobs are too large for any one person,placing a premium on group coherence,communication and cooperation.

The trouble is that this also makes cheat-ing more profitable. The evolutionary theorist William Hamilton23 argued that,to protect themselves, cooperative groupsevolve strategies to make admission into theirranks difficult. These can take the form ofbeing wary of outsiders, long periods of probation and costly (to the initiate)


NATURE | VOL 428 | 18 MARCH 2004 | www.nature.com/nature 277

Figure 2 Genes and language in Europe. This map shows lines that divide regions of abrupt differencesin the frequencies of 63 genetic markers in Europe’s populations. The 22 blue lines identify regionsthat are separated by both physical barriers, such as the Alps or the English Channel, and linguisticdifferences. The 11 red and orange lines divide regions that have no detectable physical barriers, yetnine of the 11 (those in red) correspond to linguistic differences. In this case, then, language isassociated with abrupt genetic change in 31 of 33 cases. Some lines fall in the sea (such as betweenIreland and Iceland), indicating that the theoretical point of rapid genetic difference falls somewherebetween the two locales and not on either one of them. (Map, redrawn from ref. 16, shows Europe of1990, and provides a detailed key to the linguistic and other boundaries corresponding to the lines.)



278 NATURE | VOL 428 | 18 MARCH 2004 | www.nature.com/nature


initiation ceremonies.This works to a degree,but cheats can still arise from within thegroup. Theorizing about group behaviourhas therefore also emphasized the impor-tance of genetic relatedness between groupmembers.When it is high (as can be achievedif immigration is kept at a low level),altruisticand cooperative behaviours flourish becausethey tend to benefit one’s own relatives. Butgenetic relatedness has its limitations. Whenit gets too high within a group, inbreedingbegins to take its toll, and the benefits ofsexual reproduction, namely the mixing andrecombining of genes,are diminished.

Humans may have worked out a way todiscard the need for relatedness as a means ofensuring cooperation: uniquely among theanimals, humans may carry a set of behav-ioural adaptations specific to promotingcooperation and reciprocity,even when relat-edness is low between group members24,25.According to the doctrine known as strongreciprocity,humans are predisposed to coop-erate with others, to make fair distribution ofgains,and to punish those who fail to cooper-ate, even at a cost to themselves and with noexpectation that these costs will be repaid24,25.

This is a remarkable assertion, but shouldwe believe it? Theoretical studies show thatnorms of cooperation and punishment ofcheats can arise and be maintained in groupsby a process of cultural ‘group selection’, inwhich more cohesive and cooperative groupsoutcompete groups riven by selfish cheats.Over time, these kinds of cooperative groupscome to dominate. Laboratory experimentswith volunteers and cross-cultural studiesseem to support the strong-reciprocity view.

There are implicit norms of cooperation ingroups, and individuals seem willing to punish cheats altruistically — that is, thepunisher pays the cost of punishing — eventhough all group members reap the benefits.

This extreme sociality can make coopera-tion a stable strategy resistant to cheating evenwhen group members are not related. But itdoes depend upon one key demographic fea-ture: migration between groups must be keptlow. If it is not,groups become homogenized,cheats can prosper and the driving force ofgroup selection — differences betweengroups — fails. Wariness of strangers, for allits potentially ugly manifestations, may bedeep in our psychological make-up.

An unscientific postscriptWe end by taking our cue from the existen-tial philosopher Søren Kierkegaard26. If theview put forward here is correct, thenhuman cultural diversity arises from twomain forces. One is the drive to secede fromlarger groups whenever possible, the betterto control some defensible resource; this iswhat gives rise to the geographical patternsof diversity. The second set of forces issocial and behavioural. They maintaincooperation within groups and create cul-tural identity and coherence, causing barri-ers to gene flow and meaning that verticalcultural transmission dominates.

Putting these forces together, we get apicture of humans as a highly social andgroup-focused species. None of this is to saythat selfish behaviour has been erased orthat all cultures survive intact. The all-too-common ‘tragedies of the commons’, in

which individual over-exploitation of com-mon resources results in their collapse,remind us of the price of selfishness. But thispicture of the nature of cultures suggeststhat they are surprisingly robust against out-side influences (although not invincible)and that, at least for large cultures, worriesabout cultural swamping are overstated.Nevertheless, our ancient cultural practicesmay also be telling us that, in a world inwhich mass movements of people frompoorer to richer areas will become ever morecommon, we must be especially vigilantabout our own tendencies to protect the status quo ante. ■

Mark Pagel is in the School of Animal andMicrobial Sciences, University of Reading,Whiteknights, Reading RG6 6AJ, UK.e-mail: [email protected] Mace is in the Department of Anthropology,University College London, Gower Street,London WC1E 6BT, UK.e-mail: [email protected]. Harris, M. Cows, Pigs, Wars and Witches: The Riddles of Culture

(Vintage, New York, reissue edn, 1989).

2. Renfrew, C. Archaeology and Language: The Puzzle of Indo-

European Origins (Penguin, London, 1987).

3. Diamond, J. & Bellwood, P. Science 300, 597–603 (2003).

4. Kaessmann, H., Wiebe, V. & Pääbo, S. Science 286,

1159–1162 (1999).

5. Kaessmann, H., Heissig, F., von Haeseler, A. & Pääbo, S.

Nature Genet. 22, 78 (1999).

6. Wurm, S. in Atlas of the World’s Languages (eds Moseley, C. &

Asher, R. E.) 93–158 (Routledge, London, 1994).

7. Birdsell, J. B. Am. Nat. 87, 171–207 (1953).

8. Nichols, J. Evol. Anthropol. 3, 206–215 (1995).

9. Rapaport, E. H. Science 201, 679–686 (1982).

10.Mace, R. & Pagel, M. Proc. R. Soc. Lond. B 261, 117–121 (1995).

11.Pagel, M. D., May, R. M. & Collie, A. Am. Nat. 137, 791–815 (1991).

12.Pagel, M. in The Evolutionary Emergence of Language

(eds Knight, C., Studdert-Kennedy, M. & Hurford, J.)

391–416 (Cambridge Univ. Press, 2000).

13.Moore, J. L. et al. Proc. R. Soc. Lond. B 269, 1645–1653 (2002).

14.Cavalli-Sforza, L. L., Piazza, A., Menozzi, P. & Mountain, J.

Proc. Natl Acad. Sci. USA 85, 6002–6006 (1988).

15.Sokal, R. R. et al. Am. Nat. 135, 157–175 (1990).

16.Barbujani, G. & Sokal, R. R. Proc. Natl Acad. Sci. USA 87,

1816–1819 (1990).

17.Womble, W. H. Science 114, 315–322 (1951).

18.Rosser, Z. H. et al. Am. J. Hum. Genet. 67, 1526–1543 (2000).

19.Dent, S. The Language Report (Oxford Univ. Press, 2003).

20.Mantel, N. Cancer Res. 27, 209–220 (1967).

21.Guglielmino, C. R., Viganotti, C., Hewlett, B. & Cavalli-Sforza,

L. L. Proc. Natl Acad. Sci. USA 92, 7585–7589 (1995).

22.Holden, C. & Mace, R. Am. J. Phys. Anthropol. 10, 27–45 (1999).

23.Hamilton, W. H. in ASA Studies 4: Biosocial Anthropology

(ed. Fox, R.) 133–153 (Malaby, London, 1975).

24.Fehr, E. & Fischbacher, U. Nature 425, 785–791 (2003).

25.Gintis, H., Bowles, S., Boyd, R. & Fehr, E. Evol. Hum. Behav. 24,

153–172 (2003).

26.Kierkegaard, S. Concluding Unscientific PostScript to

Philosophical Fragments, 1846 (transl. Hong, H. V. &

Hong, E. H.; Princeton Univ. Press, 1992).

27.Durham, W. H. Coevolution: Genes, Culture and Human

Diversity (Stanford Univ. Press, 1991).

28.Cavalli-Sforza, L. L. & Feldman, M. W. Cultural Transmission and

Evolution: A Quantitative Approach (Princeton Univ. Press, 1981).

29.Ruhlen, M. A Guide to the World’s Languages Vol. 1 Classification

(Arnold, London, 1991).

30.Mace, R. & Pagel, M. Curr. Anthropol. 35, 549–564 (1994).

31.Pagel, M. in Time–Depth in Historical Linguistics

(eds Renfrew, C., MacMahon, A. & Trask, L.) 189–207

(McDonald Inst. Archaeology, Cambridge, 2000).

Picture credits. Top row (left to right): F. Naumann/Panos;

K. Mueller/Still Pictures; J. Miles/Panos; J.-L. Dugast/Still Pictures;

E. Rassias/Gamma/Katz; J. Horner/Panos. Bottom row: Gamma/Katz;

D. Tatlow/Panos; K. Robinson/Panos; J. Cancalosi/Still Pictures;

C. & J. Lenars/Corbis; A. Wolfe/Still Pictures.

The ancestor–descendantrelationships between a groupof cultures can be representedby a phylogenetic tree.Phylogenies are branchingdiagrams, like a genealogy or afamily tree, that describerelatedness between a group ofspecies or cultures. Linguisticdata can be used to constructphylogenies of cultures forcomparative cross-culturalstudies30,31 in much the sameway that genetic data can beused to construct phylogeniesof animals. The diagram showsa hypothetical geographicalarea with four cultures labelleda–d. If geography determinesthe traits that any two cultureshave in common, then, forexample, cultures a and cshould share the trait (the two

colours can be taken torepresent two different valuesof some trait — for example,farming versus hunting andgathering as the culture’ssubsistence practice). Ifancestry determines the valueof the trait, then cultures that are near to one another on the phylogenetic tree should have the same value

of the trait, independently ofgeography. Studies show that ancestry (phylogeny) isoften more influential thangeography for core culturaltraits21,22. This is the outcomeshown here: despite being next to culture c, culture ashares its trait with its nearest phylogenetic relative. M.P. & R.M.

Box 2 Transmission of cultural traits

a

a b c d

cb

d



Documents

The cultural wealth of nations