Henrich Evolution EvolAnthrop 2003 2183456

7/24/2019 Henrich Evolution EvolAnthrop 2003 2183456

1/13

ARTICLES

The Evolution of Cultural EvolutionJOSEPH HENRICH AND RICHARD McELREATH

In 1860, aiming to be the rst Euro-peans to travel south to north acrossAustralia, Robert Burke led an ex-tremely well-equipped expedition of

three men (King, Wills and Gray)from their base camp in CoopersCreek in central Australia with vefully loaded camels (specially im-ported) and one horse. Figuring amaximum round trip travel time of three months, they carried twelveweeks of food and supplies. Eight

weeks later they reached tidal swampson the northern coast and began theirreturn. After about ten weeks theirsupplies ran short and they began eat-ing their pack animals. After twelveweeks in the bush, Gray died of illnessand exhaustion, and the group jetti-soned most of their remaining sup-plies. A month later, they arrived backin their base camp, but found thattheir support crew had recently de-parted, leaving only limited supplies.Still weak, the threesome packed theavailable supplies and headed to thenearest outpost of civilization, Mt.Hopeless, 240km south. In less than amonth, their clothing and boots werebeyond repair, their supplies wereagain gone, and they ate mostly camelmeat.

Faced with living off the land, theybegan foraging efforts and tried, un-successfully, to devise means to trapbirds and rats. They were impressedby the bountiful bread and sh avail-able in aboriginal camps, in contrastto their own wretched condition. They

attempted to glean as much as theycould from the aboriginals about nardoo, an aquatic fern bearing sporesthey had observed the aboriginals us-ing to make bread. Despite travelingalong a creek and receiving frequentgifts of sh from the locals, they wereunable to gure out how to catch

them. Two months after departingfrom their base camp, the threesomehad become entirely dependent onnardoo bread and occasional gifts of sh from the locals. Despite consum-ing what seemed to be sufcient calo-ries, all three became increasingly fa-tigued and suffered from painfulbowel movements. Burke and Willssoon died, poisoned and starved fromeating improperly processed nardooseeds. Unbeknown to these intrepidadventurers, nardoo seeds are toxic

and highly indigestible if not properlyprocessed. The local aboriginals, of course, possess specialized methodsfor detoxifying and processing theseseeds. Fatigued and delusional, Kingwandered off into the desert where hewas rescued by an aboriginal group,the Yantruwanta. He recovered andlived with the Yantruwanta for severalmonths until a search party foundhim.

The planning for this expeditioncould not have been more extensive,and these men were not unpreparedBritish schoolboys out on holiday.However, despite their big brains,camels, specialized equipment, train-ing, and seven months of exposure tothe desert environment prior to run-ning out of supplies, they failed to sur- vive in the Australian desert. This bitof history makes a simple point: Hu-mans, unlike other animals, areheavily reliant on social learning toacquire large and important portionsof their behavioral repertoire. Noevolved cognitive modules, evoked

Joseph Henrich received his Ph.D. in1999 from the University of California, Los Angeles, and is currently Assistant Pro-fessor of Anthropology at Emory University.He was recently a fellow in the Society ofScholars at the University of Michigan andat theInstitute for Advanced Study in Berlin.He has conducted ethnographic and exper-imental research among the Machiguengaof Peru and the Mapuche of southern Chile.His theoretical work has involvedconstruct-ing formal models of the evolution of cul-tural learning capacities, cultural evolution,and culture-gene coevolution. E-mail: [email protected] McElreath received his Ph.D. in2001 from the University of California,Los Angeles, and is now Assistant Pro-fessor of Anthropology at the University

of California. He was recently a postdoc-toral fellow at the Center for Adaptive Be-havior and Cognition, Max Planck Insti-tute for Human Development, Berlin. Heconducts ongoing eld work to investi-gate cultural microevolution among sev-eral ethnic groups in southwest Tanzania.E-mail: [email protected]

Key words: social learning; human evolution; cultureand cognition; coevolution; dual inheritance theory

Evolutionary Anthropology 12:123135 (2003)DOI 10.1002/evan.10110Published online in Wiley InterScience(www.interscience.wiley.com).

Humans are unique in their range of environments and in the nature and diversity oftheir behavioral adaptations. While a variety of local genetic adaptations exist withinour species, it seems certain that the same basic genetic endowment produces arcticforaging, tropical horticulture, and desert pastoralism, a constellation that representsa greater range of subsistence behavior than the rest of the Primate Order combined.The behavioral adaptations that explain the immense success of our species arecultural in the sense that they are transmitted among individuals by social learning andhave accumulated over generations. Understanding how and when such culturally

evolved adaptations arise requires understanding of both the evolution of the psycho-logical mechanisms that underlie human social learning and the evolutionary (popu-lation) dynamics of cultural systems.

Evolutionary Anthropology 123


2/13

culture, or generalized cost-bene tcalculators delivered to these men theknowledge of how to detoxify nardoospores or how to make and use rattraps, bird snares, or shing nets fromlocally available materials. Unlike so-cial learning in other animals, humancultural abilities generate adaptivestrategies and bodies of knowledgethat accumulate over generations.Foraging, as it is known ethnographi-cally, would be impossible withouttechnologies such as kayaks, blow-guns, bone tools, boomerangs, andbows. These technological examplesembody skills and know-how that nosingle individual could gure out inhis lifetime. Nonmaterial culture,such as seed processing techniques,tracking abilities, and medicinal plant

knowledge, reveals similar locallyadaptive accumulations. Interest-ingly, this adaptive information is of-ten embodied in socially learnedrules, techniques, and heuristics thatare applied with little or no under-standing of how or why they work.

Thus, understanding a substantialamount of human adaptation requiresunderstanding the cultural learningprocesses that assemble our behav-ioral repertoires over generations.This is not, however, a call to separate

humans from the rest of nature. Aproductive approach should seat hu-mans within the broader context of mammalian and primate evolutionwhile at the same time being able toexplain how and why humans are sodifferent in the diversity and nature of their behavioral adaptations. Our goalin this paper is to review recent devel-opments in understanding both theevolution of the psychological mecha-nisms that make cultural evolutionpossible and the population-level con-sequences of those individually adap-tive mechanisms. Most of the relevantwork occurs within a pair of closelyrelated approaches: gene-culture co-evolution 1 5 and dual-inheritance the-ory. 6,7 These approaches examine theinteractions between genetic and cul-tural inheritance systems. In thesemodels, individual phenotypes arecombinations of both genetic and so-cially transmitted characters, whichin turn affect the transmission rates of different alleles and cultural variants.Early models explored, among other

things, how different modes of cul-tural inheritance affect rates and out-comes of cultural evolution 2 and hownatural selection acting on genes canproduce a semi-autonomous inheri-tance system. 7 Like human behavioralecology, 8 coevolutionary and dual-in-heritance theories are concerned withadaptation. Unlike human behavioralecology, however, these theoriesmodel the proximate mechanismsthat produce adaptations. Like evolu-tionary psychology, these theoriesshare an interest in the design of cog-nition. Unlike most evolutionary psy-chology, however, dual-inheritanceand gene-culture models are rigor-ously formalized, take account of so-cial learning, and explore populationprocesses. For many questions,

strictly outcome-oriented or culture-free models are suf cient and insight-ful. For many others, however, takingaccount of cultural dynamics is essen-tial. As the Burke and Wills story illus-trates, even hunter-gatherer adapta-tion is substantially reliant uponevolved cultural knowledge and tech-nology. To understand adaptation inhuman societies with any time depthseems very dif cult without some at-tempt to account for the evolutionarydynamics that produce such adapta-

tions.Throughout this paper we will usecultural learning and culturaltransmission/acquisition to refer tothe subset of social learning capacitiesthat allow for cumulative cultural evo-lution. We use culture to refer to theinformation acquired by individuals via social learning. Processing nardooand making arrow poison, for exam-ple, are cultural practices because in-dividuals learn them from other mem-bers of their social group. The mentalrepresentations that allow individualsto detoxify the fern spores or bringdown large game with relatively light-weight bows and arrows do not comecoded in their genes, nor are thesecontinually relearned by each individ-ual via trial-and-error experimenta-tion or deduced solely by tness-ori-ented cost-bene t analysis. Instead,such adaptations result from and em-body the cumulative effects of the ef-forts, experiments, errors, insight, andinteractions of many individualsacross generations.

Conceptualizing culture as sociallylearned information stored in people sbrains opens up new sets of evolution-ary questions. We will review the re-search on ve of these: How does so-cial learning in humans increaseadaptability and thereby allow ourspecies to successfully occupy such anenormous range of environments? If cultural learning mechanisms are soadaptive, why are such mechanismsseemingly rare in nature? What cog-nitive processes guide human sociallearning? If cultural variants do notreplicate like genes, can cultureevolve? How does the coevolution of genes and culture in uence humanpsychology and the histories of hu-man societies? These ve questionsbuild a natural progression of puzzles,

from the genetic origins of culturalinheritance to the dynamics of mod-ern cultural and societal evolution.

WHY IS CULTURAL LEARNINGADAPTIVE?

To understand the evolution of so-cial learning, theorists have developedformal models to study how temporallyand spatially changing environmen-tal conditions affect the evolutionarytrade-offs between capacities for in-

dividual learning (for example, trialand error), social learning, and hard-wired behavioral responses. 1,5,7,9 15Most of these models are very abstractand apply to a wide range of animalsocial learning, not just human cul-tural transmission. They show that so-cial learning is favored throughout alarge intermediate range of environ-mental uctuation, especially whenenvironments are highly autocorre-lated. The intuition behind these re-sults is that social learning allows or-ganisms to respond more quickly toenvironmental changes than do hard-wired responses, but only by exploit-ing a body of adaptive knowledge thatis stored in the learned behavioral rep-ertoire of the population. At one ex-treme, when environments uctuateon the order of thousands of genera-tions, social learning serves no pur-pose since raw natural selection act-ing on genes can, on average, do justas well without paying for expensivesocial learning machinery. At theother extreme, when uctuations oc-

124 Evolutionary Anthropology ARTICLES


3/13

cur on the order of single generations,there is little adaptive knowledge forsocial learners to exploit. However,with intermediate rates of change, onthe order of tens or hundreds of gen-erations, social learning mechanismsboth outpace genetic adaptation andhave suf cient time between environ-mental changes to accumulate a bodyof adaptive knowledge in the popula-tion.

When viewed alongside a growingpool of empirical evidence, this theo-retical work suggests that both individual and social learning from an in-tertwined adaptive response toincreasing amounts of environmental variability 16,17 what Potts 18 calls variability selection. First, there isnew evidence that increases in brain

size relative to body size are corre-lated with both social and individuallearning abilities across species. Inprimates, brain size corrected forbody size correlates most stronglywith social learning abilities, but alsowith individual learning ( innova-tion ) and tool use, all three of whichare highly intercorrelated. 19 As far aswe know, no similar studies exist formammals in general, although thereare similar ndings for birds. 20,21 Sec-ond, these data suggest that increases

in brain size in the paleontologicalrecord have been partly driven by in-creases in social learning abilities.Right up to the present, the recordshows that several mammalian lin-eages have undergone increases inbrain size relative to body size. Fi-nally, over the same period, ice-coredata show increasing degrees of cli-matic variation: Over the last fourteenmillion years, which is the limit of thetime depth of the data, increases inclimatic variability are mirrored byincreases in brain ratio. This combi-nation of evidence, alongside the for-mal theory that independently impli-cates environmental variation withincreases in social learning abilities,suggests that human cultural capaci-ties may be a hypertrophied subset of a larger class of learning abilities thathave evolved in many species. 22

Yet humans stand out in the num-ber and diversity of environmentsthey inhabit. What is the role of sociallearning in human adaptability, andhow have these abilities permitted a

tropical primate to spread so rapidlyand successfully into so many habi-tats from the dry savannahs andtropical forests of equatorial Africa tothe Arctic tundra and humid swampsof New Guinea while most othermammals with plausibly well-devel-oped social learning abilities showcomparably restricted ranges? Priorto a clever paper by Rogers, 23 severalresearchers had argued that sociallearning improves human adaptabil-ity by exempting individuals from thecosts of individual learning. 5,7,24 Theargument seems cogent enough: Time

costs and potential mistakes can makeindividual learning quite expensive. If another individual or group of individuals has already paid those costs,learning from that behavior may beconsiderably cheaper. Imagine thetask of selecting among mushroom varieties through individual learning.Because some mushrooms are poi-sonous, the price of choosing thewrong mushroom is quite high. How-ever, an individual who learns fromothers which mushrooms are poison-

ous spares herself those potentialcosts, provided that the behavior of others is adaptive.

However, Rogers showed that thisargument is insuf cient to explain theadaptive success of our cultural spe-cies. Using a very simple model, heproved that sparing individuals thecosts of individual learning will not,on its own, lead to increased overalladaptability in the population themean tness of the population is notincreased. While social learners do very well when they are rare, they dopoorly when they are common. With-out any individual learners, sociallearners cannot track changes in theenvironment, and the rst individuallearner entering a group of sociallearners always has higher tness

than the others. This means that atequilibrium the mean tness of thepopulation as a whole is the same asthat in a population of purely individ-ual learners. Social learning alonedoes not increase adaptability. Box 1explains this mathematical argumentin more detail. Boyd and Richerson 11extended Rogers result to more com-plicated models in which social learn-ers can identify and preferentiallycopy individual learners, the environ-ment varies spatially as well as tempo-

rally, imitation generates errors, andthere are more than two behaviors.None of these changes alter the resultthat the evolution of social learningdoes not lead to a more t population.Cultural capacities, as represented inthese models, do not raise the overalltness of the population, so they areunlikely to explain the adaptive successof our species in the last 200,000 years.

In the same paper, however, Boydand Richerson 11 showed that sociallearning can lead to higher mean t-ness provided either that it allows theaccumulation of behaviors that no in-dividual learner could acquire in itslifetime or improves the ef ciency of individual learning. When either is thecase, social learning may increase themean tness of the population. Therst condition is in fact the questionwe started the paper with, and we willdiscuss it at length in the next section.The second condition is satis ed if learners use individual learning whenit is cheap and reliable, and switch tosocial learning when individual learn-

Without any individuallearners, social learners

cannot track changes inthe environment, andthe rst individuallearner entering a groupof social learners alwayshas higher tness thanthe others. This meansthat at equilibrium themean tness of the

population as a whole isthe same as that in apopulation of purelyindividual learners.Social learning alonedoes not increaseadaptability.

ARTICLES Evolutionary Anthropology 125


4/13

ing is expensive. 7,14 We think both areat work in human cognition. However,

the adaptive gains possible through thesecond mechanism alone seem modestin comparison to those produced by cu-mulative cultural evolution.

WHY ARE CAPACITIES FORCUMULATIVE CULTURALTRANSMISSION RARE?

Several of our colleagues are fond of the Why not baboons? stratagem: If an evolutionary scenario is meant toexplain some unique (or at leastnearly unique) feature of humans,then it must also be able to explainwhy baboons and many other ani-mals do not fall under the same evo-lutionary logic. We have seen manyclever theories crumble before this in-terrogation. The story we outlinedearlier is vulnerable to the same criti-cism. Although human cultural capac-ities can be seen as part of a moregeneral pattern of adaptation forlearning in variable environments,their immense adaptiveness and ap-parent uniqueness poses an evolution-

ary puzzle: Why haven t the sociallearning capacities that generate cu-

mulative cultural adaptations repeat-edly evolved along with other individ-ual and social learning abilities inmany mammalian lineages over thelast fourteen million years? Thus, herewe attempt to explain why human-likecultural capacities should be rare innature, as we believe they are, despitebeing extremely adaptive.

While an increasing amount of eldevidence suggests that other animals,particularly chimpanzees, may main-tain traditions that result from sociallearning, 25 28 there is little reason tobelieve that nonhuman social learningcapacities can generate cumulativeadaptation. 7,29,30 In contrast, accumu-lated cultural skills and knowledge arecharacteristics of all human societies.While the psychological mechanismsthat make cumulative culture possibleare unclear, there are some promisingideas. Tomasello, Kruger, and Rat-ner 29 suggested that true imitation, orobservational learning the directand accurate copying of behaviors,strategies or symbolic knowledge is

necessary for cumulative cultural evo-lution. Other kinds of social learning

may lead to traditions, but not to theaccumulation of adaptive informa-tion. Imagine that individuals are ca-pable of a modest amount of individ-ual learning, so that interaction withthe environment slowly generatesadaptive behavior. If naive individualstend to hand around other individu-als, and some of these individuals pre-fer to hang around certain kinds of food sources, because they have individually learned how to exploit thosefood sources (for example, crackingnuts or termiting), then naive individ-uals would be more likely to devise ameans to exploit that resource. Thiswould be social learning, but since in-dividuals have to reinvent the detailsof the behavior for themselves, albeitaccelerated by proximity to conspecif-ics, the behavior cannot become morecomplex across generations beyond acertain point. Naive individuals do notget a head start, and thus cannotbegin where previous learners left off.If, instead, individuals acquire theirbehavior by directly observing and

Box 1. Mathematical Argument

As we explain in the text, Rogers 23

model demonstrates that sociallearning alone will not increase theaverage tness of a population of cul-tural organisms. A simple graph canmake the argument much clearer.

Figure (a) plots the tnesses of in-dividual learners ( w l , dark line) andsocial learners ( w s , thin line), as wellas the population mean tness ( w ,dotted line), for all frequencies of so-cial learning in the population ( p ).When social learners are rare, they dosigni cantly better than individuallearners, since most potential modelsare practicing the correct behavior. As social learning becomes more

common, however, the populationlags behind the environment moreand more until individual learningpays just as well as social learning, at p . Natural selection will stabilize thepopulation at this equilibrium, atwhich both social and individuallearners receive tness w , which is

the same as the average populationtness when p 0, when there areno social learners.

But if we allow the frequency ofsocial learning to reduce the costs ofindividual learning, a new equilibriumarises at which the population meantness is greater than that of a popu-lation of individual learners. Figure (b)shows the modi ed model in whichthe costs of individual learning de-

crease as the frequency of sociallearning increases. Now the popula-tion mean tness w is greater than the

mean tness in a population of indi-vidual learners, w 0 . In order for sociallearning to increase adaptability, themean tness of the population, itmust also somehow increase the t-ness of individual learning. In the text,we discuss plausible mechanisms forthis effect.



5/13

copying the details of others tech-niques, then individual learning canbuild atop previous innovations. 31 A version of this distinction that allowsfor more continuity with chimpanzeeswould be that chimps possess modesttrue imitative capacities, but the com-plexity of the skills and technologiesthey can represent and the delity of their transmission is less than that of humans.

True imitation is probably not thewhole story, however, at least not inthe long run. In modern humans, asuite of social learning abilities con-tributes to the maintenance and accu-mulation of culture. Simpler forms of observational learning (of physicalskills, for example) likely provided afoundation for more complex kinds of

social learning and inference, such asthose associated with symbolic com-munication and language. Symboliccommunication through proverbs,stories, and myths allows for a greatdeal of cultural transmission withoutobservation in the usual sense. 31,32For example, !Kung hunters knew agreat deal of natural history, includ-ing the fact that porcupines are mo-nogamous. 33 It is hard to imagine thatknowledge of this kind is preservedthrough observational learning alone.

However, Tomasello30

argues thattrue imitation, rooted in a geneticallyevolved capacity for Theory of Mind,generates both linguistic and nonlin-guistic forms of cultural evolution,and that linguistic symbols (includinggrammatical structures) have gradu-ally accumulated, improved, andadapted through a cultural evolution-ary process analogous to that ob-served in the domain of material cul-ture and technology.

Whatever the speci c nature of themechanisms be they true imitationor not it remains puzzling why theyshould be so rare. Boyd and Richer-son 34 constructed a model of the evo-lution of cumulative cultural capaci-ties designed to explore this puzzle. Intheir model, a population lives in a variable environment in which thereis a unique optimal adaptive value of aquantitative trait. Each generation,there is some probability that the environment changes so that a new value of the trait is optimal. Individualphenotypes are a combination of ge-

netic in uences and cultural trans-mission. Other genes affect an individual s reliance on imitation, butcarry an incremental tness cost. Allindividuals engage in some individuallearning, which moves their pheno-types a small amount toward the cur-rent optimum. But individuals with asubstantial reliance on cultural learn-ing can acquire phenotypes muchcloser to the optimum, once such phe-notypes exist in the population. Thesephenotypes are then improved a smallamount by individual learning. Thisprocess repeats every generation.

Unlike the simpler social learningmodels discussed in the previous sec-tion, this work demonstrates that asubstantial reliance on cultural learn-ing is unlikely to spread initially, butgoes to xation and is stable once acritical threshold frequency is sur-mounted (Box 2). Natural selection fa- vors cultural learning only when thecosts of developing and maintainingcultural learning mechanisms aresmaller than the bene ts gained byacquiring simple behaviors that could

be learned on one s own. But, despitebeing dif cult to get started, once areliance on cultural learning is com-mon in the population, it is easy tosustain. Provided that the environ-ment is not too variable, the rate of accumulation of adaptive behaviorthrough cultural learning can easilypay for the cost of the psychologicalcapacities needed to make it possible:Cultural learning mechanisms provide access to the knowledge accumu-lated over generations that simple so-cial learning does not. However,because cultural capacities are not favored when rare, we should not expectthem to be widespread in nature. Apopulation must traverse a tness val-ley before the frequency of true imita-tion is high enough to make it individ-

ually advantageous. Because otherforms of social learning are often builtprincipally out of individual learning,and do not involve inferential recon-structions of behaviors and strategies,they do not face this dilemma butthey also cannot generate cumulativecultural adaptation.

Having offered an explanation of why cumulative cultural abilitiesmight be rare in nature, we are leftwith the question of why it was spe-cically the human ancestral lineagethat crossed the cultural threshold.One possibility is that ancestral hu-mans just happened to drift geneti-cally across the threshold. Randomevents of this kind were likely impor-tant in the evolutionary histories of many species. However, we think it ismore productive to ask if there wassomething particular about the hu-man lineage that made it more likelythan other species to cross this cul-tural threshold. Perhaps our evolvingcultural capacities depended rstupon some other adaptation, which

might have arisen for another reasonentirely. 34 Good answers here areprobably a long way off, but specula-tion based on the existing informationwill help direct future research.

WHAT COGNITIVEMECHANISMS GUIDE CULTURAL

EVOLUTION?Like evolutionary psychology, dual-

inheritance theory combines evolu-tionary theory with empirically

While an increasingamount of eldevidence suggests thatother animals,particularlychimpanzees, maymaintain traditions thatresult from sociallearning, there is littlereason to believe thatnonhuman social

learning capacities cangenerate cumulativeadaptation. In contrast,accumulated culturalskills and knowledgeare characteristics of allhuman societies.



6/13

grounded assumptions about the environments inhabited by ancestral hu-man populations to make predic-tionsa about the details of humanpsychology details that often specifycognitive mechanisms people use toextract adaptive ideas, beliefs, andpractices from their social environ-ments. However, the approach di- verges from mainstream evolutionarypsychology in emphasizing the costlyinformation hypothesis. This hypoth-esis focuses on the evolutionary trade-

offs between acquiring accurate be-havioral information at high cost andobtaining less accurate information atlow cost. When accurate informationis unavailable or too costly, individu-als may exploit the information storedin the behavior and experience of other members of their social group.

By exploring how the costly infor-mation hypothesis generates trade-offs in the evolution of our cognitivecapacities, we can generate produc-tive theories about the details of hu-

man cultural psychology. When infor-mation is costly, natural selection willfavor cognitive mechanisms that al-low individuals to extract adaptive in-formation, strategies, practices, heu-ristics, and beliefs from othermembers of their social group at alower cost than through alternativeindividual mechanisms. Human cog-nition probably contains numerousheuristics and learning biases that fa-cilitate the acquisition of usefulknowledge, practices, beliefs, and be-

Box 2. Cultural Learning

Regions for which cultural learninghas an advantage over individual learn-ing, for the Boyd and Richerson 34

model. The curves show the internalunstable equilibrium of cultural learnersversus individual learners and repre-sent the threshold frequency at whichcultural learning becomes favored byselection. This is shown for two valuesof , the probability that the environ-ment changes each generation andrenders a new behavior adaptive. Ineach case, culture learners will in-crease in the region above the curveand decrease below it.

When individual learning is dif cult(left side of plot), cultural learning

cannot invade the population be-cause too few individuals have theskilled behavior. When cultural learn-ing is rare, the only behaviors thatexist in the population are those thatare solely the result of individuallearning (those that could be guredout by one individual in his lifetime).The cost (for example, in adding met-abolically costly brain tissue) doesnot easily outweigh the bene t be-cause there is little adaptive informa-tion contained in the behavior of oth-ers that the animal cannot gure outon its own. It is important to keep inmind that the developmental or t-ness costs of true imitation mecha-nisms may be quite high, eventhough, from our human perspective,imitation itself strikes us as beingquite easy. It is easy because ourcognition is designed for imitationand social learning (for example, in-fants show both sophisticated imita-

tive and inferential abilities, as well asa strong desire to imitate 70 ).

As we move toward slightly easierproblems, cultural learning still can-not invade, but is stable once com-mon. Once cultural capacities arecommon, population processes willbegin to assemble complex adapta-tions, and individuals who have theability to acquire them will do sub-stantially better than those who can-not. Under these conditions, culturallearning does for genes what theycannot do directly for themselves.

Looking at the far right side of theplot, where individual learning is easy,

essentially everyone acquires highlyadaptive behavior without paying theadditional costs of cultural learningcapacities, so cultural learning israrely adaptive.

In all cases, there are many morecombinations of parameters forwhich cultural learning is stable whencommon, but cannot invade the pop-ulation when rare. In the case where

0.4, implying that the optimal be-havior changes in 40% of genera-tions, there is no dif culty of individ-ual learning for which cultural learninginvades, but a wide range for whichcultural learning is stable.



7/13

havior ( cultural traits or represen-tations ). These mechanisms can beusefully modeled at the algorithmic

level, much as some cognitive scien-tists investigate other kinds of infor-mation processing.

Such cultural learning mecha-nisms, all of which build atop othersocial and cultural learning abilities,can be categorized into content bi-ases and context biases. Box 3 orga-nizes the various forms of culturallearning mechanisms. Content bi-ases, or what Boyd and Richerson 7called direct biases, exploit informa-tive cues of an idea, belief, or behav-ior itself, and thereby in uence thelikelihood of imitation. An equiva-lent perspective prefers to discusscultural learning as adaptive infer-ences triggered by content biases forcues provided in the behavior of oth-ers. 35 Many such biases may haveevolved because they facilitate theacquisition of tness-enhancing cul-tural traits. 2,4,7 Because content bi-ases are likely numerous and gener-ally con ned to particular domainsof culture, for space considerationswe have omitted any substantial dis-

cussion of them here. However, inthinking about content biases, it isimportant to keep in mind a number

of things. First, jury-rigged evolu-tionary products, like human minds,are likely to contain accidental by-products and latent structures thatcreate biases for tness-neutral be-haviors, ideas, beliefs, and val-ues. 36,37 Boyer 38 detailed one kind of by-product content bias in his expla-nation for the universality of reli-gious concepts (like ghosts). Second,even content biases that arose be-cause they led to the adoption of t-ness-enhancing behavior in ancientenvironments may now promote theadoption of quite maladaptive prac-tices. Third, content biases may beeither reliably developing productsof our species-shared genetic heri-tage or they may be culture speci c.People may learn valuable contentcues via cultural learning or, havingacquired one idea or practice via cul-tural transmission, may be morelikely to acquire another because thetwo t together in some cognitivesense.

Context biases, on the other hand,

exploit features of potential models orthe frequencies of alternative behav-iors or strategies, rather than features

of the alternatives themselves, toguide social learning. There is a greatdeal of adaptive information embod-ied in both who holds ideas and howcommon the ideas are. A largeamount of modeling effort has beenexpended in exploring the conditionsunder which different context biasesevolve and how strong natural selec-tion would prefer they be. These mod-els derive from rst principles abouthow individual cognitive biases affectboth individual tness (when theyevolve) as well as the patterns of in-formation in the population (whatthey evolve). Our remaining discus-sion of psychological mechanisms fo-cuses on two categories of context bi-ases in cultural learning: success andprestige bias and conformity bias.

Success and Prestige Bias

If individuals vary in skills (for ex-ample, tool making), strategies (track-ing techniques), or preferences (forexample, for foods) in ways that affect

Box 3. Cultural Learning Mechanisms

Content biases and heuristics arisefrom the interaction of human psy-chology and the characteristics orcues associated with the thing be-ing transmitted (the idea, representa-tion, or behavior). These biases affectthe likelihood of a particular mentalrepresentation being transmitted be-cause of the content of the represen-tation. Content effects can take manyforms. They may re ect the direct ac-tion of natural selection on our pre-pared learning abilities, such as lan-guage, folk biology, and colorcategories. They may also arise asby-products of cognitive evolution:Boyer s 38 approach to cultural phe-

nomena like ghosts and gods is oneexample. Or they may emerge froma kind of more generalized cost-bene t calculation: People prefersteel axes to stone axes because itis much less work to cut down treeswith steel. Such biases may result

for either genetically transmittedcognitive structures (as in Boyer s

argument) or culturally acquiredmental representations. Contextheuristics arise from the learningenvironment or context. Model-based biases result from cues orcharacterist ics of the potentialmodel ( an individual who may be

imitated ) and make the ideas, men-tal representations, or behavior of

their possessor more likely to trans-mit than those held by other individ-uals. Other model-based biasesmay include age, sex, ethnicity, andhealthful appearance. Frequency bi-ases use the commonality or rarityof a behavior as a cue.



8/13

tness, and at least some componentsof those differences can be acquired via cultural learning, then natural se-lection may favor cognitive capacitiesthat cause individuals to learn prefer-entially from more successful individ-uals. The greater the variation in ac-quirable skills among individuals, andthe more dif cult those skills are toacquire via individual learning, thegreater the pressure to preferentiallyfocus one s attention on and imitatethe most skilled individuals. If individuals evaluate potential culturalmodels (individuals they may learnfrom) along dimensions associatedwith competence in underlying skills(such as hunting returns), and focustheir social learning attention onthose who are more successful, they

will be more likely to acquire adaptivestrategies. 31 Interestingly, while theability to rank individuals by foragingsuccess is observed in nonhumans(for better scrounging), 39 there is noevidence that individuals in these spe-cies acquire strategies from successfulforagers. With the rise of cultural ca-pacities in the human lineage, naturalselection needed only to connect theselearning abilities with preexistingranking capacities.

A bias of this kind is a standard

assumption in evolutionary game the-ory, 40 where a preference for copyingthe strategies of successful individualsgenerates an evolutionary dynamicthat is usually mathematically indis-tinguishable from natural selectionacting on genes. However, uncertaintyabout the payoffs and success of otherindividuals complicates success-bi-ased learning. Schlag 41,42 has ex-plored the exact form that such anadaptive bias should take in the pres-ence of noisy feedback about the suc-cess of other individuals, nding thata linear weighting of models by theirobserved payoffs may be more adap-tive than simply imitating the individ-ual with the highest observed payoff.Another solution is for individuals touse aggregate indirect measures of success, such as wealth, health, orfamily size, which integrate overmany instances and smooth out per-ceptual and stochastic errors. Thismay explain the widespread observa-tion that people copy successful individuals, as de ned by local standards.

(See Henrich and Gil-White 31 for asummary of the laboratory and eldevidence.)

However, an additional problemcreated by using indirect indicators of successful strategies is that it is often very unclear which of an individual smany traits have led to success. Arepeople successful because of how theytend their farms, cook their food, ormake sacri ces to the spirits, or allthree? Because of this ambiguity, hu-mans may have evolved the propen-sity to copy successful individualsacross a wide range of cultural traits,only some of which may actually re-late to the individuals success. 7,31,43 If information is costly, it turns out thatthis strategy will be favored by naturalselection even though it may allow

neutral and maladaptive traits tohitchhike along with adaptive culturaltraits. In a world of costly informa-tion, cognitive adaptations do not al-ways produce adaptive behavior fromthe point of view of genes, even inancestral environments. Nevertheless,the theory does allow for predictionsabout the conditions under whichmaladaptive cultural traits willspread.

The evolution of a success bias mayalso be able to explain the formation

of prestige hierarchies. Once success-biased transmission has spreadthrough a population, highly skilledindividuals will be at a premium, andsocial learners will need to competefor access to the most skilled individ-uals. This creates a new selectionpressure on success-biased learners topay deference to those they assess ashighly skilled (those judged mostlikely to possess adaptive informa-tion) in exchange for preferred accessand assistance in learning. Deferencebene ts may take many forms, includ-ing coalitional support, gifts, generalassistance (house-building), and car-ing for offspring. 31

With the spread of deference forhigh skilled individuals, natural selec-tion can take advantage of these ob-servable patterns of deference to fur-ther save on information-gatheringcosts. Naive entrants (say immigrantsor children), who lack detailed infor-mation about the relative skill of po-tential cultural models, may take advantage of the existing pattern of

deference by using the amounts andkinds of deference different modelsreceive as cues of underlying skill. As-sessing differences in deference pro- vides a best guess of the skill rankinguntil more information can be accu-mulated. This also means that skilledindividuals will prefer deference dis-plays that are easily recognized byothers (in public). Thus, along withthe ethological patterns dictated bythe requirements for high delity so-cial learning (proximity and atten-tion), deference displays also includediminutive body positions and socio-linguistic cues. The end point of thisprocess gives us the psychology, soci-ology, and ethology of prestige, which must be distinguished fromthose associated with phylogeneti-

cally older dominance processes.31

From this theory, Henrich and Gil-White 31 derived twelve predictionsabout the interrelationships betweenpreferential imitation or in uence,deference, and other ethological pat-terns, individual characteristics (likeage and sex), and memory. A review of data from psychology, economics,and ethnography turned up a sizableamount of evidence consistent withthese predictions.

Conformist BiasIt is unlikely that success and pres-

tige biases solve all costly informationproblems, however. What do you dowhen any observable differences insuccess and prestige among individu-als do not covary with the observabledifferences in behavior? For example,suppose everyone in your village usesblowguns for hunting except one reg-ular guy who uses a bow and arrowand obtains fairly average hunting re-turns. Do you adopt the bow or theblowgun?

One solution for dealing with suchinformation-poor dilemmas is to copythe behaviors, beliefs, and strategiesof the majority. 7,14 Termed confor-mity bias, this mechanism allows in-dividuals to aggregate informationover the behavior of many individuals.Because these behaviors implicitlycontain the effects of each individual sexperience and learning efforts, con-formist transmission can be the bestroute to adaptation in information-poor environments. To see this, sup-



9/13

pose every individual is given a noisysignal (a piece of information) fromthe environment about what the bestpractice is in the current circum-stances. This information, for any oneindividual, might give them a 60%chance of noticing that blowgunsbring back slightly larger returns thanbows. Thus, using individual learningalone, individuals will adopt the moreef cient hunting practice with proba-bility 0.6. But, if an individual sam-ples the behavior of 10 other individ-uals, and simply adopts the majoritybehavior, his chances of adopting thesuperior blowgun technology increaseto 75%.

Obviously, if everyone uses onlyconformist transmission, no adapta-tion or cultural evolution occurs, but

models of interaction among differentlearning mechanisms indicate thatnatural selection will very often favora mix of social and individual learningwith a substantial reliance on confor-mity. Extending Boyd and Richer-son s7 original model, Henrich andBoyd 14 used simulation to investigatethe interaction and coevolution of ver-tical transmission (parent-offspringtransmission), individual learning,and conformist transmission in spa-tially and temporally varying environ-

ments. These results con

rm that con-formist transmission is likely to evolveunder a very wide range of conditions.In fact, these results show that therange of conditions that favor con-formist transmission are wider thanthose for vertical transmission alone,suggesting that if advanced sociallearning via vertical transmissionevolves at all, we should also expectto observe a substantial conformistbias.

The model of the combination of conformity bias with individual learn-ing and vertical transmission leads tothree predictions: 1. Individuals willprefer conformist transmission over vertical transmission, assuming it ispossible to access a range of culturalmodels at low cost, which is often, butnot always the case; 2. As the accuracyof information acquired through individual learning decreases, reliance onconformist transmission over individ-ual learning will increase; 3. Individu-als should be sensitive to substantialshifts in the relevant environments so

that they decrease their reliance onconformist transmission after recentuctuations or increase it after immi-grating.

Work combining these models withempirical investigations is growing.Kameda and Nakanishi 44 have furtherextended the Henrich and Boyd 14model to predict how human psychol-ogy should respond to changes in thecost of individual learning and de-signed experiments to test their pre-dictions. By analyzing the temporaldynamics of historical cases of the dif-fusion of innovations, Henrich 45 hasfound evidence that is consistent witha strong role for both conformity- andsuccess-biased transmission and in-

consistent with a strong role for individual learning. We imagine futurework will illuminate the complex in-

teractions among conformist andother social learning biases in envi-ronments in which the costs and qual-ities of information vary.

IF CULTURAL VARIANTS DONOT REPLICATE LIKE GENES,

CAN CULTURE EVOLVE?

So far, we have treated the inheri-tance of cultural variants as unprob-lematic. However, because much of the initial work in coevolutionary the-

ory involved tools from population ge-netics and theoretical evolutionary bi-ology, there are good reasons toexamine the strength of the analogybetween genes and memes. Dawkins, in The Extended Pheno-type, 46 described what he saw to bethe necessary characteristics of anyreplicating entity: longevity, fecun-dity, and delity. The structure of thisargument has been used to supportthe analogy between genetic and cul-tural (or memetic ) evolution: Cul-tural ideas can be replicators as well,and hence culture may evolve as dopopulations of alleles. Some cognitiveand evolutionary anthropologists,however, have severely criticized thepower of this analogy, arguing thatcultural ideas are rarely if ever repli-

cated during social learning and thatculture is substantially transformedby human psychology so that ideasare rarely transmitted intact so thereare no or few discrete units in cul-ture. 35,38,47,48 For these reasons, theyargue, cultural variants ( memes orrepresentations ) have little delityand so cannot evolve in a Darwiniansense. Essentially, if cultural inheri-tance involves continuously blending(nondiscrete) traits and mutation-likeprocesses are powerful, memes will

not ful

ll Dawkins requirements for areplicator. Without a replicator, the

argument goes, there can be no cul-tural evolution.

These arguments should be takenseriously. If culture is not an evolvingsystem in the Darwinian sense, thenmany coevolutionary theories (and, of course, substantial portions of this pa-per) require serious rethinking. Build-ing on the preceding points, Sper-ber, 35 Boyer, 48 and Atran 47 haveargued that many existing models of cultural evolution are inappropriate,transmission cannot explain the per-sistence of behavioral variation in hu-mans, and cultural evolution cannotproduce adaptations. If these argu-ments are correct, the story we toldearlier about culture accumulatingpowerful locally adapted skills andtechnologies is somehow mistaken.

We think the arguments we re- viewed earlier are valid in this respect,however. There are good reasons tosuppose that culture is an evolution-ary system, even if the three claims

Obviously, if everyoneuses only conformisttransmission, noadaptation or culturalevolution occurs, butmodels of interactionamong different learningmechanisms indicatethat natural selection will

very often favor a mix ofsocial and individuallearning with asubstantial reliance onconformity.



10/13

above are true. In two recent articles,Henrich and coauthors 49,50 use threemathematical models and severalother lines of argument to show thatthe objections mentioned here do notfollow from their assumptions.Through these analyses, the authorsdemonstrate how Dawkins originalclaims about replicators and Darwin-ian evolution were wrong replica-tors are suf cient for cumulative evo-lution, but not necessary.

In their rst model, Henrich andBoyd 49 address two complaints: thatculturally transmitted ideas are rarely if ever discrete and that inferential biasesin learning (Sperber s strong attrac-tors ) swamp the effects of selectivetransmission and prevent Darwinianadaptation. This model assumes that

individuals possess mental representa-tions ( cultural variants, beliefs, andscripts) that are in uenced by selec-tively learning from some individuals(for example, from successful individu-als). These mental representations arecontinuous (nondiscrete or quantita-tive), so each individual may possess asomewhat different variant of the rep-resentation. There are no copies of variants, only social inuence. Fur-thermore, in learning these representa-tions, individuals use inferential pro-

cesses that strongly bias the

nal formof the representation. Their analysisshows that these complaints are deduc-tively invalid. If cognitive inferential in-uences are suf ciently strong relativeto selective forces (selective learning), acontinuous (quantitative) model re-duces to a discrete-trait replicatormodel commonly used in populationmodels of both culture and genes. Infact, the stronger the effects of inferen-tial bias on learning, the better is thediscrete trait approximation. Moreover,this means that it is the weak effects of selective transmission that determinethe nal equilibrium of the system.

In the second and third models, theauthors construct systems that allowfor large amounts of transmission errorto show that accurate individual-levelreplication of cultural variants is notnecessary for selective forces to gener-ate either cultural inertia or cumulativecultural adaptation. The second modelshows how conformist transmissioncan act to drastically reduce the effectof transmission errors and still generate

either cultural inertia or diffusion of successful variants. The third modelcombines all the potential problemswith models of cultural evolution continuous (nondiscrete) mentalrepresentations, incomplete trans-mission, and substantial inferentialtransformations and shows notonly that adaptive cultural evolutionmay still occur under empiricallyplausible conditions, but that it alsopredicts when such adaptive evolu-tion will not occur.

Many of the insights from these for-mal models have been known forsome time but, unlike Dawkins repli-

cator argument, have not successfully

spread. While Sperber, Boyer and At-ran s criticisms apply to the informaltheorizing of some memeticists, 46,51,52they are wide of the mark for muchformal gene-culture coevolutionarytheory. Continuous trait models goback to the very beginning of the eld.Boyd and Richerson 7 argued in 1985that there is no need to assume par-ticulate units in order to build evo-lutionary models, in fact showing thatblending models best produce herita-ble variation exactly when transmis-

sion is inaccurate. In fact, nineteen of the thirty-eight models presented intheir book are continuous (nondis-crete) trait models that allow for anarbitrary amount of transmission er-ror. Similarly, Cavalli-Sforza andFeldman 2 devoted one of their vechapters entirely to continuous traitmodels. These continuous models al-low for substantial error and otherforms of nonreplication. Similar tocognitivist critics, Boyd and Richer-son also explicitly distinguish publicrepresentations from mental repre-sentations (though using different ter-minology) throughout their book, andrepeatedly specify the inferentialtransformation between observed be-havior and representation formed.They also make explicit reference to

much research in psychology on thenature of social learning and proposethe following pathway for the trans-mission of cultural variants: Modeledevents 3 Attention Processes 3 Re-tention Processes 3 Motor Reproduc-tion 3 Motivation Processes 3 Matching. Chapters 4 and 5 in Boydand Richerson s book discuss howcognitive structures what Sperber 35would later call attractors biascultural change so that some out-comes are more likely than others,and even use some of the same exam-ples as Boyer. 48

The force of arguments like those of Sperber, Atran and Boyer seems to bethat cultural learning requires innate,domain-speci c psychological mecha-nisms (we agree!), and therefore thatmost of the action is in individual psy-chologies and not in the population dy-namics. This conclusion is unfounded:An understanding of cultural evolutionrequires studying both the evolved cog-nitive abilities and inferential mecha-nisms that allow for cultural learning,

as well as the population processes towhich they give rise through social in-teraction. Culture can have heritableproperties and evolve in a Darwiniansense even if it is continuous, error-prone, and individually ephemeral.

HOW DOES COEVOLUTIONINFLUENCE PSYCHOLOGY AND

SOCIETY?

A persistent debate in the social sci-ences is whether the chief causal level

An understanding ofcultural evolutionrequires studying boththe evolved cognitiveabilities and inferentialmechanisms that allowfor cultural learning, aswell as the populationprocesses to which theygive rise through socialinteraction. Culture canhave heritableproperties and evolve ina Darwinian sense evenif it is continuous, error-prone, and individuallyephemeral.



11/13

in social phenomena is the individualor the social. Instead of arguing thatprimary causation exists at eitherlevel, gene-culture population modelstake seriously and treat explicitlyforces at both levels, and sometimesmore. From this perspective, classicfeatures of human cultures and soci-eties, such as culture being shared bymembers of self-ascribed groups, be-come results to derive, rather than a priori assumptions. These dialecticalmodels have helped us to understandhow interactions between cognitionand population processes give rise toethnically marked groups 53,54 and eth-nic psychology, 55 large-scale coopera-tion, prosocial psychologies, and group-bene cial cultural norms. 7,16,56 60Rather than attempting to summarize

this large literature, we focus only onone of the most recent models.

The Coevolution of EthnicallyMarked Groups and EthnicPsychology

In almost all ethnographicallyknown regions and historical periods,humans have organized themselvesinto self-ascribed groups marked byarbitrary symbols. 61 For example, inboth historical and modern East Af-

rica, different pastoralist groups weardifferently colored clothing, whichserves as ethnic markers. In one re-gion of modern Tanzania, Maasaiwear red, Sukuma wear blue, andTaturu wear black. Since no other pri-mate forms such symbolically markedgroups, and existing rates of mixingamong such groups would quicklyerode differences of this kind if theywere transmitted from parent to off-spring in any fashion (culturally or ge-netically), some explanation of theirformation and maintenance is needed.

Prior efforts to explain ethnicityhave proved theoretically unsound.First, the standard approach to ethnicactors as strategic manipulators re-quires that some other processes gen-erate and maintain ethnic groups andtheir markings. If ethnicity weresolely the product of strategic consid-eration or a coalitional psychology, 62ethnicity would rapidly disappear as aphenomenon and there would not beanything to manipulate. 54,63 A moreserious idea is that ethnic markers al-

low actors to select individuals withwhom to cooperate. 64,65 These effortsfail because, unless some process pre- vents out-group members from adopt-ing the same markers, individualswho wear the markers but do not co-operate will destroy the signal value of the symbols. 66,67 So the question re-mains: How do such markers ariseand what are their functions?

In addressing this puzzle, McEl-reath, Boyd, and Richerson 54 con-structed a model of the emergence of ethnic marking in which markersfunction to provide coordination (notcooperation, so there is no free-riderproblem) with other individuals whoshare one s norms. Coordinationmeans that individuals are better off when they practice complementary

behaviors. The familiar example of this occurs in cross-cultural commu-nication, 68 where different expecta-tions in many aspects of interactionroutinely lead to lower payoffs for allparties. The coaching book market forinternational business people atteststo the severity of these problems. It islikely that the same phenomenon oc-curs in many other aspects of cultur-ally inherited behavior. Having thesame norms about child rearing, bar-ter, marriage, inheritance, and con-

ict resolution can be crucial for suc-cessful social relations. Since thenumber of domains of this kind islikely large and many such rules areheld unconsciously, the mutual costsof interactions between individualswith different sets of norms can besubstantial.

The model is sketched as follows.First, imitation of the successful andsocial interaction produces culturallydifferentiated communities. In eachsocial group, whatever norm is ini-tially most common leads to the high-est payoffs, making it more common.Then, provided individuals are biasedto interact with people who share thesame arbitrary symbolic markers asthemselves, symbolically markedgroups that possess different culturalnorms arise endogenously in themodel. Furthermore, even if there isinitially no genetically transmittedpsychological bias to interact withother individuals who share yoursame marker, natural selection, oper-ating in this culturally constructed en-

vironment, will favor genes that reli-ably produce this bias sinceindividuals who prefer to interactwith those with the same marker aremore likely to interact with someonewith the same norms as themselves,and therefore pro t more from socialinteraction.

The model also makes some unex-pected predictions about the nature of ethnic marking. While the model re-quires spatial variation in norms toevolve the association between normsand markers, once markers are asso-ciated with underlying norms, andprovided other processes permit atight linkage between them, spatial variation in norms is no longerneeded to maintain functional ethnicmarkers. Instead, the ethnic groups

in the model merge, forming one largemultiethnic community in which individuals still coordinate their interac-tions based on markers delineatingdistinct ethnic divisions. Since mark-ers in such a situation allow individu-als to assort nearly perfectly with oth-ers who share their norms, membersof smaller norm communities are notat a disadvantage relative to the nor-mative majority. Situations like thisresemble in an abstract way modernmultiethnic cities like Los Angeles or

Detroit,69

in which many ethnicgroups live intermixed but preferen-tially interact among themselves.

The model makes predictions aboutboth evolved psychological propensi-ties and sociological patterns, and ex-plicitly links them. Ethnic markingarises as a side effect of other psycho-logical mechanisms which them-selves have solid individual-level se-lective advantages that happen togenerate behaviorally distinct groups.The strategy of using arbitrary sym-bolic markers to choose interactantsthen evolves because of features of theculturally evolved environment. Cul-tural transmission mechanisms maycreate statistically reliable regularitiesin the selective environments faced bygenes. 4,57 Thus, explaining many im-portant aspects of human psychologyand behavior will require examininghow genes under the in uence of nat-ural selection responded to the regu-larities produced by culture. Thismeans that understanding the behav-ior of a highly cultural species like



12/13

humans will sometimes demand a cul-ture-gene coevolutionary approach.Phenotypic optimality models andmodels that ignore the population dy-namics of social learning certainlyhave their place, and have proven veryuseful. But satisfying answers tomany important questions concerninghuman behavior, from the culturalmicroevolution of foraging adapta-tions to the coevolution of human psy-chology and cultural variation, will re-main elusive unless dual inheritanceor some similar approach is taken se-riously.

ACKNOWLEDGMENTS

This paper coevolved with valuablefeedback from Robert Boyd, Natalie

Henrich, Pete Richerson, Eric Smith,Peter Todd, Annika Wallin, and twoanonymous reviewers.

REFERENCES

1 Cavalli-Sforza LL, Feldman MW. 1973. Cul-tural versus biological inheritance: phenotypictransmission from parent to children (a theory of the effect of parental phenotypes on children sphenotype). Am J Hum Genet 25:618 637.

2 Cavalli-Sforza LL, Feldman MW. 1981. Cul-tural transmission and evolution: a quantitativeapproach. Princeton: Princeton University Press.

3 Feldman MW, Laland KN. 1996. Gene-culturecoevolutionary theory. Trends Ecol Evol 11:453 457.

4 Durham WH. 1991. Coevolution: genes, cul-ture, and human diversity. Stanford: StanfordUniversity Press.

5 Pulliam HR,DunfordC. 1980. Programmed tolearn: an essay on the evolution of culture. NewYork: Columbia University Press.

6 Richerson PJ, Boyd R. 1976. A simple dualinheritance model of the con ict between socialand biological evolution. Zygon 11:254 262.

7 Boyd R, Richerson PJ. 1985. Culture and theevolutionary process. Chicago: University of Chi-cago Press.

8 Winterhalder B, Smith EA. 2000. Analyzingadaptive strategies: human behavioral ecology attwenty- ve. Evol Anthropol 9:51 72.

9 Boyd R, Richerson PJ. 1988. An evolutionarymodel of social learning: the effects of spatial andtemporal variation. In: Zentall T, Galef BG, edi-tors. Social learning: a psychological and biolog-ical approach. Hillsdale, NJ: Lawrence ErlbaumAssociates p 29 48.10 Boyd R, Richerson PJ. 1989. Social learningas an adaptation. Lect Mathematics Life Sci 20:126.11 Boyd R, RichersonPJ. 1995. Whydoes cultureincrease human adaptability? Ethol Sociobiol 16:125 143.12 Stephens DW. 1991. Change, regularity, and value in the evolution of animal learning. BehavEcol 2:77 89.13 Bergman A, Feldman MW. 1995. On the evo-lution of learning: representation of a stochasticenvironment. Theor Popul Biol 48:251 276.

14 Henrich J, Boyd R. 1998. The evolution of conformist transmission and the emergence of between-group differences. Evol Hum Behav 19:215 242.15 Feldman MW, Aoki K, Kumm J. 1996. Indi- vidual and social learning: evolutionary analysisin a uctuating environment. Anthropol Sci 104:209 232.16 Richerson PJ, Boyd R. 1999. The evolutionarydynamics of a crude super organism. Hum Nat10:253 289.17 Richerson PJ, Boyd R, Bettinger RL. 2001.Was agriculture impossible during the Pleisto-cene but mandatory during the Holocene? a cli-mate change hypothesis. Am Antiquity 66:287 411.18 Potts R. 1998. Variability selection in hominidevolution. Evol Anthropol 7:81 96.19 Reader SM, Laland KN. 2002. Social intelli-gence, innovation, and enhanced brain size inprimates. Proc Natl Acad Sci USA 99:4436 4441.20 Lefebvre L, Gaxiola A, Dawson S, Rozsa L,Kabai P. 1998. Feeding innovations and fore-brain size in Australasian birds. Behaviour 135:1077 1097.

21 Lefebvre L, Whittle P, Lascaris E, FinkelsteinA. 1996. Feeding innovations and forebrain sizein birds. Anim Behav 53:549 560.22 Box HO, Gibson KR, editors. 1999. Mamma-lian social learning: comparative and ecologicalperspectives. Cambridge: Cambridge UniversityPress.23 Rogers AR. 1988. Does biology constrain cul-ture? Am Anthropol 90:819 831.24 Lumsden C, Wilson EO. 1981. Genes, mind,and culture. Cambridge: Harvard UniversityPress.25 McGrew WC. 1992. Chimpanzee material cul-ture: implications for human evolution. Cam-bridge: Cambridge University Press.26 Wrangham RW, McGrew WC, de Waal FBM,Heltne P. 1994. Chimpanzee cultures. Cam-bridge: Harvard University Press.27 Whiten A, Goodall J, McGrew WC, Nishida T,Reynolds V, Sugiyama Y, Tutin CEG, WranghamR, Boesch C. 1999. Cultures in chimpanzees. Na-ture 399:682 685.28 Boesch C, Tomasello M. 1998. Chimpanzeeand human culture. Curr Anthropol 39:591 604.29 Tomasello M, Kruger AC, Ratner HH. 1993.Cultural learning. Behav Brain Sci 16:495 552.30 Tomasello M. 2000. The cultural origins of human cognition. Cambridge: Harvard Univer-sity Press.31 Henrich J, Gil-White FJ. 2001. The evolutionof prestige: freely conferred deference as a mech-anism for enhancing the bene ts of culturaltransmission. Evol Hum Behav 22:165 196.

32 Lancy DF. 1996. Playing on the mother-ground: cultural routines for children s develop-ment. New York: The Guilford Press.33 Blurton-Jones N, Konner MJ. 1976. !Kungknowledge of animal behavior. In: Lee R, De- Vore I, editors. Studies of the !Kung San andtheir neighbors. Cambridge: Harvard Univer-sity Press.34 Boyd R, Richerson PJ. 1996. Why culture iscommon, but cultural evolution is rare. In:Runciman WG, Maynard Smith J, Dunbar, RIM,editors. Evolution of social behaviour patterns inprimates and man. Proceedings of The BritishAcademy, Vol. 88. Oxford: Oxford UniversityPress. p 77 93.35 Sperber D. 1996. Explaining culture: a natu-ralistic approach. Oxford: Blackwell.

36 Johnstone RA. 1994. Female preference forsymmetrical males is a by-product of selectionfor mate recognition. Nature 372:172 175.37 Enquist M, Johnstone RA. 1997. Generaliza-tion and the evolution of symmetry preferences.Proc R Soc Lond B 264.38 Boyer P. 1994. Naturalness of religious ideas:a cognitive theory of religion. Berkeley: Univer-sity of California Press.39 Stammbach E. 1988. Group responses to spe-cially skilled individuals in a Macacca fascicularisgroup. Behaviour 107:241 266.40 Gintis H. 2000. Game theory evolving. Prince-ton: Princeton University Press.41 Schlag KH. 1998. Why imitate, and if so,how? J Econ Theory 78:130 156.42 Schlag KH. 1999. Which one should I imitate?J Math Econ 31:493 522.43 Flinn MV, Alexander RD. 1982. Culture the-ory: the developing synthesis from biology. HumEcol 10:383 400.44 Kameda T, Nakanishi D. 2002. Cost-bene tanalysis of social/cultural learning in a non-sta-tionary uncertain environment: an evolutionarysimulation and an experiment with human sub- jects. Evol Hum Behav 23:373 393.45 Henrich J. 2001. Cultural transmission andthe diffusion of innovations: adoption dynamicsindicate that biased cultural transmission is thepredominate force in behavioral change. Am An-thropol 103:992 1013.46 Dawkins R. 1982. The extended phenotype.Oxford: Oxford University Press.47 Atran S. 2002. The religious landscape. Cam-bridge: Cambridge University Press.48 Boyer P. 1999. Cognitive tracks of culturalinheritance: how evolved intuitive ontology gov-erns cultural transmission. Am Anthropol 100:876 889.49 Henrich J, Boyd R. 2002. On modeling cogni-tion and culture: why replicators are not necessaryfor cultural evolution. J Cogn Culture 2:87 112.50 Henrich J, Boyd R, Richerson PJ. n.d. Fivecommon mistakes in cultural evolution. In: Sper-ber D, editor. Epidemiology of ideas: Open CourtPublishing. In press.51 Dennett D. 1995. Darwin s dangerous idea.London: Penguin Press.52 Blackmore S. 1999. The meme machine. Ox-ford: Oxford University Press.53 Boyd R, Richerson PJ. 1987. The evolution of ethnic markers. Cultural Anthropol 2:65 79.54 McElreath R, Boyd R, Richerson PJ. 2003.Shared norms and the evolution of ethnic mark-ers. Curr Anthropol. 44:122 129.55 Gil-White FJ. 2001. Are ethnic groups biolog-ical species to the human brain?: essentialismin our cognition of some social categories. CurrAnthropol 42:515 554.56 Boyd R, Richerson PJ. 1990. Culture and co-operation. In: Mansbridge J, editor. Against self interest. Chicago: University of Chicago Press. p111 132.57 Henrich J, Boyd R. 2001. Why people punishdefectors: weak conformist transmission can sta-bilize costly enforcement of norms in coopera-tive dilemmas. J Theor Biol 208:79 89.58 Boyd R, Richerson PJ. 2002. Group bene cialnorms spread rapidly in a structured population.J Theor Biol 215:287 296.59 BoydR, Richerson PJ. 1992. Punishment allowsthe evolution of cooperation (or anything else) insizable groups. Ethol Sociobiol 13:171 195.60 Soltis J, Boyd R, Richerson PJ. 1995. Can



13/13

group functional behaviors evolve by culturalgroup selection? an empirical test. Curr An-thropol 36:473 494.61 Barth F. 1969. Introduction. In: Barth F, edi-tor. Ethnic groups and boundaries. Boston: LittleBrown.62 Kurzban R, Tooby J, Cosmides L. 2001. Canrace be erased? coalitional computation and so-cial categorization. Proc Natl Acad Sci USA 98:15387 15392.63 Gil-White FJ. 1999. How thick is blood? theplot thickens . . . : if ethnic actors are primordi-

alists, what remains of the circumstantialist/pri-mordialist controversy? Ethnic Racial Stud 22:789 820.64 Nettle D, Dunbar RIM. 1997. Social markersand the evolution of reciprocal exchange. CurrAnthropol 38:93 99.65 Van den Berghe PL. 1981. The ethnic phe-nomenon. Westport, CT: Praeger Publishers.66 Grafen A. 1990. Do animals really recognizekin? Anim Behav 39:42 54.67 Hamilton WD. 1964. The genetical evolutionof social behaviour. J Theor Biol 7:17 52.

68 Gumperz JJ. 1982. Discourse strategies. Cam-bridge: Cambridge University Press.69 Smith N. 2001. Ethnicity, reciprocity, reputa-tion and punishment: an ethnoexperimentalstudy of cooperation among the Chaldeans andHmong of Detroit, Michigan. PhD dissertation.Department of Anthropology, University of Cali-fornia Los Angeles.70 Gergely G, Bekkering H, Kira ly I. 2002. Ratio-nal imitation in preverbal infants. Nature 415:755 756.

2003 Wiley-Liss, Inc.

BOOK REVIEW

Books Received

Pereira, M.E. and Fairbanks,L.A. (Eds.) (2003) Juvenile Pri-mates: Life History, Develop-

ment, and Behavior. xxxiii 428pp. Chicago: University of Chi-cago Press. ISBN 0-22665622-5(paper) $30.00.

Coe, K. (2003) The AncestressHypothesis. xiv 214 pp. NewBrunswick: Rutgers UniversityPress. ISBN 0-8135-3132-2 (pa-per) $29.00.

Kappeler, P.M. and Pereira,M.E. (Eds.) (2003) PrimateLife History and Socioecology.

xxiii 395 pp. Chicago: Univer-sity of Chicago Press. ISBN0-226-42464-2 (paper) $30.00.

Minelli, A. (2003) The Develop-ment of Animal Form: Ontogeny,Morphology, and Evolution.xviii 323 pp. New York: Cam-

bridge University Press. ISBN0-521-80851-0 (cloth) $75.00.

Burkhardt, F., Porter, D.M.,Dean, S.A., Evans, S., Innes, S.,Pearn, A.M., Sclater, A., White,P. and Wilmot, S. (eds.) (2003)The Correspondence of CharlesDarwin. Volume 13: 1865. Sup-plement to the Correspondence1822 1864. New York: Cam-bridge University Press. ISBN0-521-82413-3 (cloth) $90.00.

McKee, J.K. (2003) Sparing Na-ture: The Con ict Between Hu-man Population Growth andEarth s Biodiversity. NewBrunswick: Rutgers UniversityPress. ISBN 0-8135-3141-1(cloth) $28.00.

Drayson, N. (2003) Confessing AMurder. New York: W.W. Nor-ton & Co. ISBN 0-393-32444-3(paper) $13.95.

Schwartz, J.H. and Tattersall, I.(2003) The Human FossilRecord. Volume Two. Cranio-

dental Morphology of Genus Homo (Africa and Asia). NewYork: Wiley-Liss. ISBN 0-471-31928-7 (cloth) $150.00.

Robbins, M.M., Sicotte, P., andStewart, K.J. (2001) MountainGorillas: Three Decades of Re-search at Karisoke. New York:Cambridge University Press.ISBN 0-521-78004-7 (cloth)$85.00.

Lancaster, R.N. (2003) The

Trouble with Nature: Sex in Na-ture and Population Culture.Berkeley: University of Califor-nia Press. ISBN 0-520-23620-3(paper) $21.95.

BOOK REVIEW Evolutionary Anthropology 135

Henrich Evolution EvolAnthrop 2003 2183456

Documents

Transcript of Henrich Evolution EvolAnthrop 2003 2183456