doi:10.1111/j.1420-9101.2011.02398.x REVIEW Questioning the cultural evolution of altruism J.-B. ANDRE´*(cid:2) & O. MORIN(cid:3) *CNRSUMR7206Eco-anthropologieetethnobiologie,MNHN,Universite´ParisDiderot,F75005Paris,France (cid:3)DepartmentofCognitiveScience,CentralEuropeanUniversity,Budapest,Hungary (cid:2)LaboratoireEcologieetEvolution,UMR7625,CNRS,EcoleNormaleSupe´rieure,Paris,France Keywords: Abstract conformistimitation; Theevolutionaryfoundationsofhelpingamongnonkininhumanshavebeen cooperation; the object of intense debates in the past decades. One thesis has had a culturalgroupselection; prominent influence in this debate: the suggestion that genuine altruism, culturaltransmission; strictly defined as a form of help that comes at a net fitness cost for the evolutionongraphs; benefactor, might have evolved owing to cultural transmission. The gene– imitation; culturecoevolutionliteratureiswonttoclaimthatculturalevolutionchanges sociallearning. the selective pressures that normally act to limit the emergence of altruistic behaviours.Thispaperaimstorecall,however,thatculturaltransmissionyields altruismonlytotheextentthatitreliesonmaladaptivemechanisms,suchas conformistimitationand(insomecases)payoff-biasedtransmission.Thispoint is sometimes obscured in the literature by a confusion between genuine altruism,maladaptivebydefinition,andmutualisticformsofcooperation,that benefitallpartiesinthelongrun.Theoriesofculturalaltruismdonotliftthe selective pressures weighing on strictly altruistic actions; they merely shift the burdenofmaladaptationfromsocialcognitiontoculturaltransmission. the biological fitness of other(s) at a net fitness cost for Introduction the benefactor (Hamilton, 1964; see also West etal., In models or verbal arguments, many scholars have 2011, section 6.1.1). However, evolutionary theory suggested that cooperation in human societies cannot predictsthataltruismcanevolveonlyifitispreferentially emerge solely because of its benefits to individuals. expressedtowardsgeneticallyrelatedpartners(Hamilton, Mechanisms such as reciprocity or reputation, they 1964; and see also Rousset, 2004; Lehmann & Keller, argue, do not suffice (e.g. Boyd & Richerson, 1985, ch. 2006; West etal., 2007a,b), and the genetic relatedness 7; Boyd & Richerson, 1988; Fehr & Fischbacher, 2003). within most human societies is generally considered They have thus proposed that genuinely altruistic ten- insufficient for this to occur. Therefore, the claim that denciesmustconstitute,inawayoranother,thecement human societies, but not other animal societies, rely ofhumansocieties.Thesetendenciesareoftenreferredto upon genuine altruism must be backed by some specific as ‘strong reciprocity’ (Gintis, 2000; Gintis etal., 2003). evolutionaryarguments.Twoquiteindependentlinesof Humanbeings,sotheyargue,arewillingtosacrificetime research have proposed that the solution to this puzzle andresourcestotheexclusivebenefitofothers,andthis couldbefoundinthefactthathumansocialbehaviours is why they are able to sustain cooperation on a large are culturally ratherthan geneticallytransmitted. Inthe scale(Fehr&Fischbacher, 2003;Gintisetal.,2003). generalattempttogroundsocialsciencesintonaturalistic Notonlydoesthissuggestionentailthathumanbeings foundations,theyplayanimportantrolebyconstituting have altruistic motives in the psychological sense, it arguments in favour of an altruistic view of human entails that they express altruistic behaviours in the cooperation. evolutionary sense of that term: behaviours that increase Inthispaper,ouraimistodiscuss,clarifyandquestion the common hypotheses at the core of all such cultural explanations of altruism. We will first describe the Correspondence:Jean-BaptisteAndre´,46Rued’Ulm,75005Paris,France. mechanisms that may allow cultural transmission to Tel.:+33144322341;fax:+33144272694; e-mail:[email protected] favour the evolution of genuine altruism, even towards ª2011THEAUTHORS.J. EVOL.BIOL. 24(2011)2531–2542 JOURNAL OF EVOLUTIONARYBIOLOGYª2011EUROPEANSOCIETYFOREVOLUTIONARYBIOLOGY 2531 2532 J.-B. ANDRE´ AND O. MORIN nongenetic kin (section ‘The Genetic and Cultural Evo- happenstobegroup-beneficial(i.e.altruistic)ratherthan lution ofAltruism’). individual-beneficial (i.e. selfish). More recently, Leh- Next, we will lay down one key assumption that any mann etal. (2008a) has shown that the same process model of cultural altruism needs: social learning places occursalsounderanotherfamilyofimitationrulescalled strong constraints on our capacity to behave adaptively. ‘one to many’ transmission, in which one, or a few, In other words, altruism is a maladaptive side effect of individual(s)arechosenrandomlyineachgrouptoserve social learning (section ‘When Biological Altruism as cultural model(s). EvolvesCulturally,itisAlwaysaMaladaptiveByproduct Cultural group selection, however, could be said to ofImitation’). come in two distinct versions. Some models predict the As a consequence, biological altruism cannot be con- culturalevolutionofaltruisticbehavioursinastrictsense ceived as arising from a process of gene–culture coevo- (e.g. Boyd & Richerson, 1985, ch. 7; Henrich & Boyd, lution.Ifanything,geneticadaptationshouldpreventthe 2001; Gintis, 2003; Lehmann etal., 2008a). Yet, other spread of cultural altruism. This fact is not straightfor- models, also pertaining to CGS, aim at explaining the wardly denied in the literature we review, but we shall cultural evolution of cooperative behaviours that are not giveexamplesof ambiguous claimsthat seem to suggest altruistic, that is, behaviours that increase both the the contrary (section ‘Ambiguities in the Literature‘). recipient’sandtheactor’sfitness(e.g.Boyd&Richerson, One important source of misunderstanding stems from 1990, 2002, 2009b). The points made in this paper are confusionsbetween mutualisticcooperationthatbenefit meanttoapplytomodelsofthefirstcategory,nottothe all individuals, and genuine biological altruism (section second. Thedistinction between thesetwokinds ofCGS ‘Cultural Group Selection: Disentangling Altruism and (the one based on biological altruism and the one based Cooperation’). onmutuallybeneficialcooperation)iseasilyoverlooked, We will conclude that any explanation for the exis- andsomefrequentconfusionsbetweenthetwowillhave tence of altruistic behaviours expressed towards nonkin to be lifted.This shall bedone atthe endofthis paper. mustinvolvea lackof geneticadaptation.Themaladap- Even though CGS is the only theory explicitly devel- tation may come either from an inability to weigh oped to explain the existence of cultural altruism perfectly the cost and benefits of social actions (malad- towards nongenetic kin, its core mechanism operates in aptations of social cognition) or from a tendency to another important family of models: the study of social imitatemaladaptivebehavioursindiscriminately(malad- evolution in networks, also sometimes called ‘evolution aptationsofculturalcognition).However,culturaltrans- on graphs’ (Skyrms, 2004; Nowak, 2006a,b; Ohtsuki mission in itself does not change, let alone reverse, the etal., 2006; Taylor etal., 2007; Ohtsuki & Nowak, 2008; selectivepressures acting onaltruism. Santos etal., 2008; Tarnita etal., 2009, see also Allison, 1992a for a verbal description of the same process). In these models, individuals interact within networks, and The genetic and cultural evolution of neighbours on the network mutually imitate one altruism another. Innetworkmodels,individualstypicallyfollow a family of rules called payoff-biased imitation. They Twotheoriesof culturalaltruism observe the action and resulting payoff of their neigh- The idea that culture could yield the evolution of bours and preferentially imitate the action played by altruism among nongenetic kin has been primarily put high-payoff individuals (payoff-biased imitation is also forward in the theory of cultural group selection (CGS) sometimes used in CGS – Henrich & Boyd, 2001; Boyd originally developed by Boyd & Richerson (1982, 1985, etal., 2003). Payoff-biased transmission is a cultural 2009a; Richerson & Boyd, 2005; Boyd etal., 2011; see equivalent of biological reproduction, in which payoff also Allison, 1992a,b for similar ideas, Henrich & Boyd, affects cultural influence rather than biological fitness 2001; Fehr & Fischbacher, 2003; Gintis, 2003; Gintis (see Hofbauer & Sigmund, 1998). Network models are etal.,2003;Henrich,2004;Henrich&Henrich,2007,for hence explicitly built to represent indifferently either further applications of CGS; and Lehmann etal., 2007, genetic or cultural evolution. In either case, they show 2008aforapopulationgeneticsformalization).Boydand that altruism can evolve among neighbours because Richerson’s idea is that social behaviours are culturally neighboursdosharecommongenes,orcommoncultural transmitted in a way that tends to homogenize the items. In their cultural version, therefore, these models phenotype of individuals within each group. They con- open the possibility that altruism evolves among genet- sider,inparticular,onetypeof‘imitationrule’thatcould ically unrelated individuals. have this consequence: conformist-biased imitation, whereby individuals imitate whatever behaviour is the Genetic andcultural evolution most frequent in their local group. Conformist imitation generatesastrongphenotypichomogeneitywithinsocial The mechanisms leading to the cultural evolution of groups. Some form of group selection can then act to altruism in all models may be simply captured in the favour the groups in which the shared cultural trait following way. Let us consider a social trait (helping, ª2011THE AUTHORS. J.EVOL.BIOL.24(2011)2531–2542 JOURNAL OF EVOLUTIONARYBIOLOGYª2011EUROPEANSOCIETY FOREVOLUTIONARYBIOLOGY Culturalaltruism? 2533 punishment, or anything else), transmitted genetically. Simply stated, all this merely expresses the fact that, Fromthemostbasicprinciplesofsocialevolutiontheory when two levels of replicators (genes and culture) (detailed in Supporting Information), the effect of selec- coexist,thereisnoapriorireasonwhytheirevolutionary tion on that trait can be measured by a simple and interests should always be aligned. Hence, the direction general equation, Hamilton’s rule (Hamilton, 1963, of selection, measured in Hamilton’s rules, is not the 1964),whichstatesthatthetraitisfavouredbyselection same on both levels. A trait that would be maladaptive when )C +R B >0. In this equation, )C is a partial onthegeneticlevelcanbefavouredbyselectionwhenit g g g g regression coefficient measuring the statistical relation- is transmitted on the cultural level, and vice versa. This ship between the trait value of a focal ‘actor’ individual generalstatementisalmosttrivial,butitsimplicationsare (e.g. altruism or selfishness) and the biological fitness of not, in particular in the case of altruism. Being an thisindividualandiscalledthe‘direct’effectofthetrait autonomous evolving system, culture could, at least in on fitness. B is a regression measuring the statistical principle, yield us to sacrifice ourselves, or at least help g relationshipbetweenthetraitvalueoftheactorandthe perfect strangers, at a net cost to biological inclusive biological fitness of another individual (called the fitness. ‘recipient’ in the case of a helping behaviour) and is called the ‘indirect’ effect of the trait on fitness. Finally, Biological altruism thethirdcoefficient,R ,isthegeneticrelatednessbetween g the actor and the recipient, measured as the regression, Here, we are interested in biologically altruistic traits, on the focal trait, of the heritable component of the defined as traits with a negative effect on the biological recipient’sphenotypeontheheritablecomponentofthe fitness of their carrier ()C <0) and a positive effect on g actor’s phenotype. thebiologicalfitnessofother(s)(B >0).Thesetraitsare g The same principles apply also for a social trait interesting because, under regular genetic evolution, transmitted culturally, that is, encoded by a cultural they should only evolve when they benefit genetically item (see Supporting Information). The trait is favoured related individuals. by selection when )C +RB >0, except that, now, )C As we just explained, however, when a trait is c c c c measures the statistical relationship between the actor’s transmitted culturally, the relevant B and C parameters trait value and the cultural reproductive success of his are not its effects on biological fitness. Hence, a biolog- cultural item (the specific item encoding the trait), B icallyaltruistictraitisnotnecessarilyculturallyaltruistic. c measures the statistical relationship between the actor’s For instance, a costly helping behaviour that would be trait value and the reproductive success of the homolo- easily imitated (e.g. because it is psychologically attrac- gousculturalitemcarriedbyarecipientindividualandR tive)wouldbebiologicallybutnotculturallyaltruistic.In c is the cultural relatedness between the actor and the this paper, however, we are not interested in cultural recipient, measured as the regression of the culturally selfishness/altruism defined in this sense. The aim of transmissiblecomponentoftherecipient’sphenotypeon cultural models, and therefore the aim of the present the culturally transmissible component of the actor’s review,istounderstandhowactualbehavioursthathave phenotype. the paradoxical property to reduce the biological inclu- Aswesee,beingafullygeneralpartitioningofselective sive fitness of their carriers can evolve, that is, to forces, the fundamental structure of Hamilton’s rule is understand how different transmission modes may or unchanged by the transmission mode of a trait (genetic may not change the selective pressure upon biological or cultural). However, the parameters of Hamilton’s rule altruism. Accordingly, in the following, when we speak are affected. First, the relatedness, on a given trait, of ‘altruism’, wealwaysmean‘biological altruism’. between an actor and a recipient, shall depend on the Models found in the literature typically consider the wayinwhichthattraitistransmitted.Aslongascultural evolution of two types of biologically altruistic traits. transmission is not purely vertical, the relatedness it Some consider exclusively the evolution of helping yields probably differs from that yielded by genetic behaviours, whereby a helper provides a good or a transmission (i.e. R „ R). Second, when cultural service to a recipient (Boyd & Richerson, 1985, ch. 7; g c transmission is not purely vertical, the statistical rela- Gintis,2003;Lehmann&Feldman,2008;Lehmannetal., tionshipbetweenthephenotypicexpressionofacultural 2008a). Others consider the joint evolution of helping item (e.g. altruism or selfishness) and its transmission andpunishingbehaviours,wherebyapunisherpaysacost successislikelytodifferfromthestrictbiologicaleffectof to reduce the fitness of other individuals, because they this item. When a given phenotypic trait is transmitted have refused to provide help in a first stage of the game culturally, the parameters that matter to determine the (Henrich&Boyd,2001;Boydetal.,2003;Guzmanetal., effect of natural selection on that trait are not its 2007;Lehmannetal.,2007).Theonlyimportantthingto effects on biological fitness, but its (direct and indirect) understand,here,isthatthecharacterizationofatraitas effects on the cultural fitness of the underlying cultural altruistic (or not) is independent of whether the trait item, and these are likely to be different (i.e. consistsinhelpingorpunishing(Gardner&West,2004; (B ,C ) „ (B,C)). Lehmann etal., 2007). A trait is altruistic if it increases g g c c ª2011THEAUTHORS.J. EVOL.BIOL. 24(2011)2531–2542 JOURNAL OF EVOLUTIONARYBIOLOGYª2011EUROPEANSOCIETYFOREVOLUTIONARYBIOLOGY 2534 J.-B. ANDRE´ AND O. MORIN the direct biological fitness of other(s) at a net cost to tionsmightlastapproximately25years).Thesameholds one’s own. Helping is altruistic when it comes at a net in network models. Ohtsuki etal. (2006) show that personalcostforthehelper.Punishmentcanbealtruistic helpingevolvesiftherelativebenefitofhelp(b/c)islarger also, if it comes at a net personal cost for the punisher, thanthenumber(k)ofsocialpartnersperindividual.This and if it benefits nonpunished individuals by reducing can reasonably be achieved if ‘individuals’ are actually the intensity of local competition, or by increasing the cultural items that reproduce every year (or more) and averageamountof helpexpressed byothers. can interact with as few as approximately 5 other items every cultural generation, not if they are biological organisms reproducing every 25years and interacting Thehypothetical culturalfacilitation of biological with several hundreds of partners in their lives. The altruism migrationratesthathappentofavouraltruismingroup- The effect of selection on any trait depends on its structured populations, the social graphs that happen to transmission mode (see above, section ‘Genetic and favouraltruisminnetworkmodels,areempiricallymore Cultural Evolution’). In itself, this is sufficient to entail reasonableunderculturalthangeneticevolution. thataltruismmay,atleastinprinciple,evolveculturally when its genetic evolution is impossible, and vice versa. Cultural transmission is more homogenizing. The bal- However, models of the cultural evolution of altruism ance between homogenizing and diversifying forces is make a more specific statement (at least in the case of alsotippedbycultureforasecondreason,relatedtothe CGS). Not only do they claim that cultural and genetic specific forms that cultural transmission takes in many evolution differ, they also claim that, under realistic models:conformist-biasedtransmission(Boyd&Richerson, conditions,culturalevolutionisgenerallymorelikelythan 1985,ch.7;Henrich&Boyd,2001;Guzmanetal.,2007; genetic evolution to yield altruism. In the following we Lehmann &Feldman, 2008;Lehmann etal.,2008a)and reviewwhatwetaketobethetwomajormechanisms,at onetomanytransmission(Lehmannetal.,2007,2008a). workseparatelyorinconjunction,thatlendcredibilityto With these modes of transmission, not only is cultural this claim: (i) culture can ‘increase’ relatedness (i.e. reproductionfasterthanbiologicalreproduction,itisalso R >R ) and (ii) culture can ‘reduce’ the direct cost of morehomogenizing(asingleorafewculturalitemsare c g altruism(i.e.C <C ). imitated by every individual). This reinforces even c g further the amount of cultural relatedness relative to Cultural relatedness may be larger than genetic genetic relatedness. relatedness The(geneticorcultural)relatednessinagivengroup,or Theculturalcostsandbenefitsofbiologicalaltruismmay neighbourhood,istheoutcomeofabalancebetweentwo differ from their biological counterparts antagonistic forces: (i) sampling effects that occur each When we think of the effect of culture on biological time reproduction takes place and tend to increase local altruism, we usually exclusively think of the increased homogeneity and (ii) migration and mutation that relatedness. However, when comparing the effect of reintroduce outside polymorphism and tend to decrease selection under genetic vs. cultural transmission, the homogeneity. One of the effect of cultural transmission, other parameters of Hamilton’s rule must also be com- inmodels,istoboosttheimportanceofsamplingeffects pared,namelythecost,C,andbenefit,B,ofaltruism.Let relativetomigration. Thisoccurs intwo distinctways. us insist, however, to avoid any confusion. Here, we considerculturallytransmittedtraitsthathappentohave Cultural reproduction is faster. Itisempiricallyreason- the property to be biologically altruistic (i.e. )C <0 and g abletoassumethatmorereproductioneventstakeplace B >0), and we say that their cultural costs and benefits g for culture than for genes in one given amount of time. (C and B) are likely to differ from their biological costs c c ‘Culturalgenerations’areshorter.Iftheculturalmigration and benefits (C and B ). By no means do we intend to g g rateisnotproportionatelylarger,thenthemigrationrate say that cultural transmission has an effect on the pergenerationismechanicallylowerinculture,andhence biological effectsof traits. cultural relatedness is larger. Interestingly, this occurs Toourknowledge,infact,theonlypaperinwhichthis even if cultural transmission is subject to the same (or second outcome of culture has been explicitly high- similar) selective forces than genetic transmission (i.e. lightedisamodelinwhichculturaltransmissionimpedes under‘payoff-biased’transmission;e.g.Boydetal.,2003; rather than facilitates the spread of altruism. Under a Ohtsukietal.,2006). form of payoff-biased transmission, Lehmann etal. For instance, Boyd etal. (2003) consider a migration (2008a; see also Lehmann etal., 2007) show that the ratebetweengroupswhere<1%ofindividuals‘migrate’ direct cost of altruism is larger under cultural than everygeneration,whichyieldsastrongrelatedness.What genetic transmission (i.e. C >C , compare their equa- c g allows this is, of course, the fact that the generations tions16and22)becausetheeffectoflocalcompetitionis considered in the model are cultural generation. Thus, stronger.Inthiscase,ceterisparibus,altruismislesslikely generations last only a year (whereas biological genera- under culturalthan genetictransmission. ª2011THE AUTHORS. J.EVOL.BIOL.24(2011)2531–2542 JOURNAL OF EVOLUTIONARYBIOLOGYª2011EUROPEANSOCIETY FOREVOLUTIONARYBIOLOGY Culturalaltruism? 2535 In other models, the cultural benefit-to-cost ratio of According to CGS theory, it is an adaptive strategy altruistic traits (B/C) seems more favourable to their becausebehavioursexpressedbythemajoritytendtobe c c success than their biological counterpart (e.g. Boyd & more adaptive than average (Boyd & Richerson, 1985, Richerson, 1985, ch. 7; Henrich & Boyd, 2001; Gintis, ch.7).Yet,CGSscholarsalsorecognizethatfrequencyis 2003). However,thisisnot explicitlyput forwardinthe a highly indirect, limited and easily mistaken proxy. papers, and it would require more formalization to be Many things can cause a majority of individuals to better understood. Essentially, we think this change in expressmaladaptivebehaviours(e.g.foundereffectsand benefit-to-cost ratio arises because (i) altruistic behav- culturaldrift),andthesebehavioursarethenimitatedall iours are assumed to be transmitted just as well (or the same, which is maladaptive. In other words, with almostaswell)asselfishbehaviourswithinlocalgroups, respect to biological fitness, conformist imitation is thatis,theindividualbiologicalcostofaltruism,doesnot mistake-prone, and this mistake-proneness is the very translate into an equivalent cultural cost in local trans- reasonwhyitcanleadtotheacquisitionofmaladaptive mission, whereas (ii) the collective benefit of altruism behaviours suchas altruism. does translate into a cultural advantage that accrues Two theoretical results seem surprisingly at odds with exactly equally to all the individuals in each group (e.g. the above reasoning, claiming to show that conformist via a larger group survival, see e.g. Boyd & Richerson, imitation can be favoured by genetic evolution, even 1985, ch. 7; Gintis, 2003). In other words, and to put it whenitsonlyeffectistotriggertheimitationofaltruism bluntly,inthesemodels,theindividualcostofaltruismis (Henrich & Boyd, 2001; Guzman etal., 2007). Both reduced but its collective benefit is maintained. This modelsconsiderpunishmentratherthanprimaryhelping probably accounts for a significant part of these models’ as the focal altruistic trait. In a model in which outcome. conformist imitation allows the cultural spread of altru- istic punishment, Henrich & Boyd (2001) show that natural selection favours genes that predispose individ- When biological altruism evolves uals to imitate more faithfully the social behaviour of culturally, it is always a maladaptive others, and Guzman etal. (2007) show that natural byproduct of imitation selection favours genes that directly code for conformist Onemightcapturethebasicargumentofthemodelswe imitation itself. In a situation where genetic evolution review as stating that culture, being an autonomous should not lead to altruistic punishment, evolution evolving system, may in principle yield altruism when would favour a genetic predisposition (Henrich & Boyd, genes cannot. They suggest in particular that, under 2001) or a genetic learning rule (Guzman etal., 2007) reasonableassumptions,altruismmightevolveculturally leadingindividualstoacquirealtruisticpunishment.How more easily than genetically. However, the apparent isthat possible? simplicitywithwhichtheyachievethiseffect,bychang- Theexplanationisthatbothresultsarepermittedbyan ing the parameters of Hamilton’s rule, is misleading. It implicitassumption.InGuzmanetal.(2007),individuals might hide the fact that these models rely on a crucial are assumedto be either conformist with respect to both empirical assumption. helping and punishment or not conformist at all. Yet, Genuinealtruismtowardsnonkin(or,moregenerally, whatisadaptiveinthismodelistofaithfullyimitatethe genetically maladaptive altruism) can evolve and/or helping part of others’ behaviour (because when others stabilizeculturallyifindividualsuseimitationrules,such cooperate they generally also punish, and therefore one as conformist, one to many or payoff-biased imitation, shouldcooperateaswell).Butimitatingthepunishment that result in its cultural diffusion. Yet, a maladaptive part of their behaviour is not adaptive (because punish- behaviour, even imitated from someone else, is still a ment is simply always costly). However, being maladap- maladaptive behaviour. When an individual behaves in tivedoesnotkeepthisnormfrombeingadopted,because an altruistic manner towards nonkin, whether because no flexible imitation strategy exists, that would allow she is‘directly’ mistaken,orbecauseshe hasmistakenly agentstocopythebeneficialnormbutnotthecostlyone. imitated another altruist, she is simply expressing a In Henrich & Boyd (2001), punishing is not always genetically maladaptive trait. Thus, cultural evolution detrimental to punishers, because several layers of solves the problem of altruism only to convert it into punishment are authorized: agents who do not punish another problem: the fact that individuals learn from maybepunishedforthis,andsoonupto thenthlevel. others in a way that eventually leads them to acquire Inthiscase,allintermediatelevelsofpunishmentarein maladaptive behaviours. Let us consider three examples thedirectinterestofindividuals’,andonlythenthlevelis before reachinga generalconclusion. genuinelycostlybecauseitfacesnoupper-levelsanction. Why is punishment at this last degree nevertheless adopted by evolved imitation rules? The answer is that Conformist imitation in Henrich & Boyd (2001), just as in Guzman etal. Conformist imitation consists in imitating the behaviour (2007), the same genetic predisposition is assumed to expressed by the majority of individuals in one’s group. affect equally the imitation of all social traits – whereas ª2011THEAUTHORS.J. EVOL.BIOL. 24(2011)2531–2542 JOURNAL OF EVOLUTIONARYBIOLOGYª2011EUROPEANSOCIETYFOREVOLUTIONARYBIOLOGY 2536 J.-B. ANDRE´ AND O. MORIN the more adaptive strategy would consist in imitating adaptivenessoftheirbehaviour.Inasocialnetworkorin onlythosetraitsthatbenefittheindividual.Theimitation a group-structured population, altruists have a larger of the last layer of punishment is hence a costly side payoffthan‘selfish’individuals,butthatisbecausetheir effectofthe (adaptive)imitationof otherlayers. neighbours are more altruistic than average, not because Overall, even though these two models are presented they are altruistic (see Allison, 1992a). This is precisely as exceptional instances in which conformist imitation the difference between a phenotypic correlation emerg- evolves even when its only effect is to trigger the ing from reciprocity and a correlation emerging from imitation of altruism, they have nothing exceptional in co-ancestry. In the latter, the correlation is not the this respect. As in all other models, altruism is a costly reflection of an underlying causal relationship between side effect of a relative maladaptation of imitation the actor’s phenotype and the recipient’s. Helpers do rules. have a larger payoff because they happen to interact NotethatLehmann&Feldman(2008)alsosurprisingly more often with helpers, but helping does not cause an foundthatunbiasedconformistimitation(inwhicheach increase in one’s payoff. Imitating high-payoff indivi- individual imitates a random member of her group), as duals thus leads to the mistaken acquisition of an wellasonetomanytransmission(inwhicheverygroup altruisticbehaviourthatdidnotcausetheirlargepayoffs. member imitates the same peer), can be favoured by Altruism, there again, is a consequence of the learning geneticevolutioninamodelinwhichtheironlyeffectis rule’s proneness to mistakes. to favour the cultural spread of helping (towards nonkin). But a closer look offers an interpretation of Social learning constrains behaviouraladaptation this result. When random imitation occurs (unbiased conformism, or one to many transmission), individuals Being a genetically maladaptive trait, altruism towards candirectly benefit frombeinghelpers, becausethismay nonkin entails a lack of adaptation on the side of genes, influence the phenotype of others in their group and whetherbecauseindividualsmistakenlyactaltruistically, prompt them to help in return through imitation. This orbecausetheymistakenlyimitateotheraltruists.Hence, effectobtainswheneachindividualhasalargeexpected cultural transmission does not change the nature of the influence on others, which is exactly what Lehmann & selective pressures, acting upon genes, with respect Feldman(2008)find(thisoccurswhensocialgroupsare to altruism. What cultural transmission does is only to small; see also Boyd & Richerson, 1988). Under such introduce an intermediate control variable (the rules of circumstances, owing to a type of selective pressurealso social learning) in the evolution of social traits, which at work in the evolution of reciprocity, where the changestheconstraintsapplyingtobehaviours,andlimit expression of helping by one individual increases their potential for adaptation in different ways (see thetendency of others to express it, imitation can be Lehmannetal.,2008b).Thisisinterestingandimportant, favoured because of its effect on helping. However, but doesnot byitself cancelthe paradoxofaltruism. bydefinition, helping is not altruistic in this case, being directly reciprocated. Strong constraintsgenerate strongmaladaptations What is more, in their models, students of cultural Payoff-biasedimitation altruism assume particularly strong constraints over At first, payoff-biased imitation is even more perplexing social learning. They assume that individuals learn from than conformist transmission. Payoff-biased imitation others using stereotyped rules. Individuals copy the most lookslikeaperfectmannertotakebehaviouraldecisions. frequent behaviour of their group, or the behaviour of As long as the observed ‘payoff’ is a faithful measure of theirhighestpayoff-peer,buttheyareunabletoevaluate biological fitness, payoff-biased imitation should be the actual properties of the behaviour (e.g. ‘Does it immune to the mistaken acquisition of genetically involve to spend some resources?’ and ‘Does it entail a maladaptive behaviours. This intuition explains why, in physicalrisk?’).Theabilityofthesesimplerulestosubtly contrast with conformism, scholars who use payoff- control cultural acquisitions is hence limited and opens biased imitation do not (to our knowledge) raise the the possibility of many maladaptive decisions, including question of its evolutionary rationale. However, it turns theimitationofaltruism.Assumingonlysimpleimitation out that this intuition can be erroneous. With payoff- rulesislikeimposingaverystronglimitoverthegenetic biased imitation, one does risk to acquire maladaptive adaptation of behaviour. behaviours, but only a specific category of maladaptive The assumption that evolution has not been able to behaviours: altruistic ones. Altruism is the flaw, so to provide human beings with better mechanisms than speak,of payoff-biasedimitation. stereotypedrulesofimitation,inimportantdomainssuch This flaw manifests itself when interacting individuals associalbehaviour,isquiteasurprisingone.Forinstance, are phenotypically correlated because an individual’s the rule of payoff-biased imitation used in models of payoff then partly depends on her neighbours’ strategy. evolutionongraphsrestsontheassumptionthathumans Consequently, their payoff is a poor indicator of the areabletotrackothers’fitnessgainsincomplexcircum- ª2011THE AUTHORS. J.EVOL.BIOL.24(2011)2531–2542 JOURNAL OF EVOLUTIONARYBIOLOGYª2011EUROPEANSOCIETY FOREVOLUTIONARYBIOLOGY Culturalaltruism? 2537 stances, but unable to understand the simple fact that tion (e.g. Richerson & Boyd, 2005, ch. 5; Boyd & unreciprocated help is costly, while being helped is Richerson, 2006). Yet, they also often seem well aware beneficial(seethesection‘Payoff-BiasedImitation’). of the fact that, as products of evolution, humans are This raises the question: Why do scholars follow unlikely to be so crudely adapted that they are ready to assumptions of this sort? The answer, we think, is that imitate anything. Individuals shall use simple imitation it is part of the implicit way of looking at things in rules, they often agree, only when they have no better cultural evolution. Under the influence of dual-inheri- sourcesofinformation.Forinstance,RichersonandBoyd tance theory (Cavalli-Sforza & Feldman, 1981), scholars state that conformist transmission can operate only if see culture as a supplementary system of inheritance, ‘individuals have difficulty evaluating the costs and parallel to genes. Hence, like genes, cultural items are benefits of alternative cultural variants’ (Richerson & assumed to be transmitted according to simple stereo- Boyd,2005,p.206;seealsoHenrich&Henrich,2007,p. typed rules (e.g. biparental inheritance for nuclear 66). genes). This leaves a lot of autonomy to culture and For this reason, they have put forward the idea of thus yields a lotofmaladaptations. ‘altruistic punishment’ (Boyd & Richerson, 1992; Hen- rich & Boyd, 2001; Fehr & Gachter, 2002), according to which people primarily cooperate because they fear Weaker constrains wouldgenerate weaker punishment (i.e. the helping actions per se are not maladaptations altruistic but directly beneficial to individuals), but the Yet, the ability to modulate one’s social behaviour in norms of punishment are themselves grounded in the functionofothers’couldalsobeseenasacomplexform fact that people are ready to punish even at one’s of adaptive plasticity, rather than a supplementary personal cost (i.e. the second-order retributive actions are system of inheritance. Instead of a general drive to genuinelyaltruistic).Thisisinterestingbecause,punish- imitate others, we could be endowed with a domain- ment being less frequently expressed (in equilibrium, specific ability to construct our social behaviour, in line mostindividualscooperateandareneverpunished),itis with the contingent strategies found in reciprocity biologically‘cheaper’inaveragethanhelping.Therefore, (Trivers,1971).Eventhoughtheycouldneverbeperfect so the argument goes, individuals shall have more and adaptive in all circumstances, such abilities would difficulty detectingthecostofpunishment thanthecost leavemuch lessspace for maladaptive altruism. ofhelpingandarethusmorelikelytousecontent-blind This debate is an old matter in the dual-inheritance imitationintheformer(Henrich&Boyd,2001;Henrich literature and dates back to the foundation of the field & Henrich,2007, pp.66–67). (Cavalli-Sforza & Feldman, 1981; Lumsden & Wilson, Leavingasidethefactthatitisnotclear(atleasttous) 1981).Defendersofsimpleimitationheuristicsarguethat how the mere rarity of punishment makes it more likely we are bound to use stereotyped transmission strategies to be imitated, it is important to understand that the for all purposes, because cultural behaviours and inno- reservationsofthesescholarsontheamountofmistaken vations can take an infinite variety of aspects. No imitations in humans, however interesting and reason- genetically encoded mechanism could have prepared us able they may be, are also, de facto, reservations on the todealwithallofthem(Richerson&Boyd,2005,ch.5; amountofgenuinealtruismpresentinsocieties.IfBoyd, Boyd & Richerson, 2006). We do not aim to enter, or Richerson,Henrichandothersarereadytoconsiderthat evenlesstosettle,thisdebatehere.Wesimplynotethat humansimitatealtruismonlyinrareinstancesinwhich RichersonandBoyd’sgeneralargumentiscertainlywell they have difficulty evaluating its costs (e.g. only for applicabletodecisionsthatwe,asaspecies,haveseldom punishment),theymustalsoagreethatgenuinealtruism (or never) been confronted with (e.g. the making of a israre. kayak or a bow). But it is unclear to us how this same argument applies in a domain, such as social life, to Ambiguities in the literature whichourancestorshavebeenconfrontedforlong.The possible ways of helping others, although they are Gene–culture coevolution infinitely numerous, do share some general properties, and evolution is likely to have endowed us with innate Many statements found in the literature do honestly abilitiesthathelpusdealspecificallywiththesesituations present altruism as a costly side effect of social learning, (as somerecent experiments in preschool children seem which entails that genes that make the cultural acquisi- to suggest, see e.g. Hamlin etal., 2007; Warneken etal., tion of altruism more likely cannot be favoured by 2010; Hamann etal.,2011). selection because of their effect on altruism, but at best in spite of this effect (see e.g. Boyd & Richerson, 1985, p. 227;Gintis, 2003). Weaker maladaptationsare only weaklyaltruistic Otherstatements,however,revolvingaroundtheidea Boyd,RichersonandHenrichdooftenargueinfavourof of gene–culture coevolution, are less clear.For instance, the existence of stereotyped imitation in cultural evolu- inawidelycitedreview,Gintisetal.(2003)definestrong ª2011THEAUTHORS.J. EVOL.BIOL. 24(2011)2531–2542 JOURNAL OF EVOLUTIONARYBIOLOGYª2011EUROPEANSOCIETYFOREVOLUTIONARYBIOLOGY 2538 J.-B. ANDRE´ AND O. MORIN reciprocity as a ‘predisposition to cooperate with others models as in others, a costly side effect of the imitation and to punish those who violate the norms of cooper- of beneficial traits. Therefore, the statement that ‘there ation, at personal cost, even when it is implausible to isnoneedtoassumethatcostlyconformismisaspin-off expect that these costs will be repaid’, which is genet- from individually beneficial conformism’ is simply ically maladaptive by definition, and yet claim that wrong. ‘strong reciprocity is adaptive in the sense of emerging fromagene–culturecoevolutionaryprocess’(theempha- Cultural groupselection: disentangling altruismand sisisours).RichersonandBoydprovideanotherexample cooperation in a recent book, when they state that gene–culture coevolution ‘set up an arms race that drove social Third, as we already acknowledged, CGS comes in two evolution to ever greater extremes of in-group cooper- significantly different versions, and this leads to some ation’suchthateventually‘peoplewerereadytopunish ambiguities. The first version is the target of this paper. [...] even when personal interests were not directly at Here, wecall it the ‘altruistic’ versionbecause itaims to stake’ (Richerson & Boyd, 2005, p. 214). Even more explainhowculturemayyieldtheevolutionofaltruistic surprisingly, Henrich & Boyd (2001) write in their behaviours stricto sensu. In contrast, the second version, abstract that ‘prosocial genes favouring cooperation and called here ‘mutualistic’, aims to explain the cultural punishmentmayinvade’,andGuzmanetal.(2007)write evolution of cooperative behaviours that are not altru- in their conclusion that conformism coevolves geneti- istic (see West etal., 2007b), that is, behaviours that callywithculturalaltruismwith‘noneedtoassumethat increase both the recipient’s and the actor’s fitness costly conformism is a spin-off from individually bene- through incentives such as reciprocity, reputation or ficial conformism’. retribution,performingwhatisknowningametheoryas These are misleading, if not simply false, statements. ‘equilibrium selection’ (e.g. Boyd & Richerson, 1990, They certainly cannot help readers to comprehend the 2002,2009b).Althoughthesetwoversionshaveinitially fact that altruism can be nothing but a costly side effect been explicitly distinguished (e.g. in Boyd & Richerson, of imitation and can even help spreading the false idea 1990), this is no more the case in recent writings that that, thanks to culture, altruism towards nonkin ends neglect to distinguish genuine altruism from mutualistic up being adaptive, even from a gene’s point of view. cooperation(e.g.Richerson&Boyd,2005,ch.6;Boyd& There are at least three sources of ambiguity that play a Richerson, 2009a). role here. We rapidly mention two of them. We then For instance, Richerson & Boyd (2005) begin their devote a specific subsection to the third and most chapter six byexplaining why genetic group selection is important one. limited to small groups of close kin, whereas CGS is First, the idea that the two sides of an interaction possible in larger groups (pp. 201–204). Even though coevolve can mean quite different things. Coevolution theyneverexplicitlydefineit,thisstronglysuggeststhat can be antagonistic when the two sides have conflicting they deal with genuine altruism (they even mention the interests. Or it can be ‘reinforcing’ when the two sides caseofsocialinsects).Butthentheygoontoexplainthat havealignedinterests.Regardingbiologicalaltruismwith human‘tribalsocialinstincts’evolvedgeneticallybecause nonkin,bydefinition,genesandculturehaveconflicting people lacking these instincts were punished and ostra- interests. Hence, altruism can be the outcome of a cized (p. 214), which corresponds to the mutualistic coevolution between genes and culture only in the versionoftheirtheorywherecooperationisindividually former sense, not in the latter. The claim that gene– beneficial.Finally,theyimplicitlymovebacktoaltruism culture coevolution can lead to altruism is thus literally whentheyaskwhygenesthatpreventtheacquisitionof true,butitismisleading.Itgivesthefalseimpressionthat costlycooperativebehavioursdonotinvade(p.214).Ina altruism is even stronger when genes come into play, more recent review, Boyd & Richerson (2009a) seem to whereas it is the exact opposite that is true. The correct be mostly dealing with the mutualistic version of their statement is that cultural evolution can lead to altruism, theory,mentioningtheproblemofequilibriumselection andthatgeneticcounter-evolutionisnotalwaysfullyable (p. 3281), and stating that genetic instincts have to prevent it. This is not what readers are likely to ‘coevolved with culturally transmitted social norms’ understand when they read that ‘strong reciprocity is (p.3287). However, they still choose to define cooper- adaptive in the sense of emerging from a gene–culture ationasa‘costlybehaviourperformedbyoneindividual coevolutionaryprocess’. that increases the payoff of others’ (p. 3283), which is Second, as we already discussed, Henrich & Boyd really altruism and therefore falls under the scope of the (2001) and Guzman etal. (2007) claim to show that other version ofCGS. genes coding for imitation can be favoured even when We fear that such expositions of CGS, which fail to their ‘only’ effect is to yield altruism, but they forget to distinguish mutualism and altruism, could lead to mention that this results from a constraint over the important misunderstandings of the assumptions and adaptation of learning (see section ‘Conformist Imita- explanatorypowerofthetheory.Thealtruisticversionof tion’). The imitation of costly altruism is, in these CGS can explain the existence of genuinely altruistic ª2011THE AUTHORS. J.EVOL.BIOL.24(2011)2531–2542 JOURNAL OF EVOLUTIONARYBIOLOGYª2011EUROPEANSOCIETY FOREVOLUTIONARYBIOLOGY Culturalaltruism? 2539 behaviours (e.g. strong reciprocity), but it relies on the Conclusion existence of a mismatch between genes and culture (the mistaken imitation of costly cultural traits), not on a We do sometimes behave in a strictly altruistic manner reinforcingcoevolutionofthetwo.Onthecontrary,the towards nonkin. For instance, economic experiments mutualisticversionofCGScanonlyexplaintheexistence have shown that, even when perfect anonymity is of cooperative, not altruistic, behaviours, but it can be guaranteed, some experimental subjects still give away the outcome of adaptive learning mechanisms and thus monetary resources to help or punish others, with no emerge from a reinforcing coevolution of genes and clear individual benefit (although anonymity does culture.Confusionsbetweenthetwoversionsarehence strongly reduce subjects’ generosity, see Hoffman etal., dangerous,becauseonecan takethe mildrestrictions of 1996). Our aim here has not been to discuss this oneversiontobethesufficientconditionsfortheother. literature.Wediscussedonespecifickindofevolutionary For instance, and most symptomatically, in their widely explanation of altruism: culturalexplanations. cited review on human altruism (Fehr & Fischbacher, Any explanation for a maladaptive trait, whether it 2003), have a specific section on gene–culture coevolu- involves culture or not, must rely on the fact that tion, where they state that it can ‘provide a solution to naturalselectionhasbeenunabletooptimizeit.Cultural thepuzzleofstrongreciprocity’.Nowhereinthissection altruism is sometimes thought of as an alternative to is the underlying assumption of genetic maladaptation maladaptation, but this is a misunderstanding. Like any mentioned. evolutionary account of a maladaptation, cultural evo- lutionmodelsdorelyontheinabilityofnaturalselection to remove the maladaptation. The only difference Howtodisambiguateculturalgroupselectiontheory between cultural accounts of altruism, and others, has Toavoidanyriskofmisinterpretation,itwouldbeuseful to do with the kind of maladaptation involved (Fig. 1): if, for each version of their theory (or mixture of Do they come from the limitations of our social cogni- versions), proponents of CGS could spell out together (i) tion, or from those of our capacities for cultural learn- its underlying assumptions (e.g. regarding the degree of ing? mismatch between genes and culture) and (ii) its explanatorypower(e.g.regardingtheamountofgenuine Imperfections of socialcognition altruism in human behaviour). In view of most of their recentarticles(inparticularHenrich&Henrich,2007,ch. Genuine altruism could emerge as a consequence of 3;Boyd&Richerson,2009a,b;Boydetal.,2011),itseems imperfections of human social cognition. We are tous(butwemightbewrong)thatBoyd,Richersonand endowed with evolved predispositions to handle Henrich now endorse a primarily mutualistic version of social behaviour. They help us manage our reciprocal their theory, in which helping is mostly individually exchanges and reputation. These predispositions should beneficial, and genetic instincts can coevolve with, and be adapted, in average, to maximize our genetic fitness, reinforce,culturalhelping(notethatthiswasnotthecase but we cannot expect them to be perfect (especially not inearlierwritings,e.g.Boyd&Richerson,1985,p.227). in experimentalsettings). However, if they do currently favour such a view, CGS Virtuallyeveryaspectofourphenotypeistheproduct scholars must also explicitly state that their theory does of compromises: compromises between the costs and not support many claims, found in the literature, about benefits of complexity; compromises between one func- the alleged selflessness of humans. For instance, a tion and another; etc. For instance, a ‘snake-detecting’ moderate version of their theory cannot support the ability cannot both detect all snakes with near-certainty claimthathumancooperationisa‘hugeanomaly’(Fehr and never generate false positives (this has been termed & Fischbacher, 2003), or that it is ‘fundamentally the ‘smoke detector principle’ by Nesse, 2001). Relative incompatible with the biologists’ model of the self- maladaptations are unavoidable in snake detection. regardingreciprocalaltruist’(Gintisetal.,2003;seeWest Similarly, the set of cognitive modules that play a role etal., 2011, for more examples). Initially, the originality inoursocialdecisions(tohelpornothelp,punishornot of CGS lied in its ability to explain cooperation in punish,etc.)mustunavoidablybesubjecttoconstraints. anonymous contexts (in which it cannot have direct Therefore, relative maladaptations are unavoidable in benefits),andithasbeeninvokedinmanyclaimsonthis this domain as well (see West etal., 2011 section 6.6.1). issue(e.g.Fehr&Fischbacher,2003;Gintisetal.,2003). As an illustration, in the vein of the smoke detector Yet, a primarily mutualistic version of CGS loses this principle, one may imagine that it could generally be specific explanatory power. At best, with such a view, adaptive to overestimate the importance of being coop- one can explain that humans sometimes make mistakes erative in social situations, for the sake of one’s reputa- intheirsocialdecisions(e.g.bycooperatingwhenitturns tion. Yet, this would come at the cost of false positives: out to have no benefit), not that they generally and we would sometimes behave in a cooperative manner systematically differ from nonhumans by being ready to when itturns out that there was no benefit to gain (see helpandpunishwithnoinclusivefitnessbenefits. e.g. Hagen &Hammerstein, 2006). ª2011THEAUTHORS.J. EVOL.BIOL. 24(2011)2531–2542 JOURNAL OF EVOLUTIONARYBIOLOGYª2011EUROPEANSOCIETYFOREVOLUTIONARYBIOLOGY 2540 J.-B. ANDRE´ AND O. MORIN Fitness + Social Mistaken reactions Accurate reactions Social cognition to social situations to social situations cognition Cultural Adoption of detrimental Adoption of useful Cultural learning rules behaviours behaviours learning rules COOPERATION Altruism with non-kin mutAulatrliusitsicm c woiothp ekriantion Fig.1Culturedoesnotchangetheselectivepressureongenes,whichalwaystendstomaximizeinclusivefitness.Relativemaladaptationscan neverthelessoccur,becausetheactionofnaturalselectionisconstrained.Inthecaseofcooperativebehaviours(thatbenefitatleastone recipient),relativemaladaptationsmayhaveatleasttwoorigins.Socialcognitionmaybeimperfect,leading(amongothermaladaptive behaviours)toaltruismwithnoinclusivefitnessbenefits.Likewise,culturallearningrulesmaybeimperfect,leadingtotheadoptionof maladaptivebehaviours,someofwhichmayhappentobealtruistic.Influentialversionsofculturalgroupselectionconsiderthatbiological altruismisparticularlyimportantinhumansbecauseofthissecondmechanism. different. It is not, and we think that the respective Imperfectionsof culturalcognition merit of cultural and social explanations of genuine Theoriginalityoftheculturalaccountsofaltruismisthat, altruism are easier to compare when seen as similar in ratherthanrelyingonthenecessaryimperfectionofthe this respect. setofevolvedmechanismsinvolvedspecificallyinsocial decisions, they rely on imperfections of another, more Acknowledgments transversal,setofevolvedmechanisms:themechanisms we use to learn from others. Because we generally We thank Laurent Lehmann for key discussions. We imitate others, so the argument goes, we often end up thank Franc¸ois Rousset, Nicolas Baumard, Nicolas Cla- imitating also their maladaptive courses of actions. idie`re,ClaireElMoudenandFrancescaGiardiniforvery Scholars also explain this relative imperfection by the helpful remarks and ideas. We thank two anonymous existence of a compromise. In the domain of social referees for their sometimes critical but often useful learning,theyargue,wefaceatrade-offbetweentherisk comments ona previous versionofthispaper. to imitate maladaptive behaviours, and the benefit of imitating interesting novelties. Cultural altruism is thus References oneofthecostlyconsequencesofthistrade-off,likeour fearofsnake-shapedwoodsticksisaconsequenceofthe Allison, P.1992a.Theculturalevolutionof beneficientnorms. trade-offsfaced byoursnake-detectingability. Soc.Forces71:279–301. This paper is not the place to discuss in detail the Allison, P. 1992b. Cultural relatedness under oblique and relativemeritsandexplanatorypowerofthesetwolines horizontaltransmission.Ethol.Sociobiol.13:153–169. Boyd,R.&Richerson,P.J.1982.Culturaltransmissionandthe of research. We simply wish to highlight their profound evolutionofcooperativebehavior.Hum.Ecol.10:325–351. similarity. In both cases, constraints limit the adaptive Boyd, R. & Richerson, P.J. 1985. Culture and the Evolutionary perfection of our decisions and introduce unavoidable Process.UniversityofChicagoPress,Chicago. mistakes in our behaviours. In one case, the mistakes Boyd,R.&Richerson,P.J.1988.Theevolutionofreciprocityin are specifically related to social life. In the other, the sizablegroups.J.Theor.Biol.132:337–356. mistakes concern equally all the behaviours we can Boyd, R. & Richerson, P.J. 1990. Group selection among learn from others. Usually, because it has to do with alternativeevolutionarilystablestrategies.J.Theor.Biol.145: culture, the second explanation is seen as profoundly 331–342. ª2011THE AUTHORS. J.EVOL.BIOL.24(2011)2531–2542 JOURNAL OF EVOLUTIONARYBIOLOGYª2011EUROPEANSOCIETY FOREVOLUTIONARYBIOLOGY