Kinship and altruism: A cross-cultural experimental study Madsen, Elainie; Tunney, Richard J.; Fieldman, George; Plotkin, Henry C.; Dunbar, Robin I. M.; Richardson, J-M; McFarland, David Published in: British Journal of Psychology DOI: 10.1348/000712606X129213 2007 Link to publication Citation for published version (APA): Madsen, E., Tunney, R. J., Fieldman, G., Plotkin, H. C., Dunbar, R. I. M., Richardson, J-M., & McFarland, D. (2007). Kinship and altruism: A cross-cultural experimental study. British Journal of Psychology, 98(2), 339-359. https://doi.org/10.1348/000712606X129213 Total number of authors: 7 General rights Unless other specific re-use rights are stated the following general rights apply: Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal Read more about Creative commons licenses: https://creativecommons.org/licenses/ Take down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. LUND UNIVERSITY PO Box 117 221 00 Lund +46 46-222 00 00 Copyright © The British Psychological Society Reproduction in any form (including the internet) is prohibited without prior permission from the Society 339 The British Psychological BritishJournalofPsychology(2007),98,339–359 Society q2007TheBritishPsychologicalSociety www.bpsjournals.co.uk Kinship and altruism: A cross-cultural experimental study Elainie A. Madsen1,2*, Richard J. Tunney1,3, George Fieldman4, Henry C. Plotkin1, Robin I. M. Dunbar5, Jean-Marie Richardson1 and David McFarland6 1University College London, UK 2University of St Andrews, Scotland, UK 3University of Nottingham, UK 4Buckinghamshire Chilterns University College, UK 5University of Liverpool, UK 6University of Oxford, UK Humansarecharacterizedbyanunusuallevelofprosociality.Despitethis,considerable indirect evidence suggests that biological kinship plays an important role in altruistic behaviour.Allpreviousreportsoftheinfluenceofkinselectiononhumanaltruismhave, however, used correlational (rather than experimental) designs, or imposed only a hypothetical or negligible time cost on participants. Since these research designs fail eithertocontrolforconfoundingvariablesortomeetthecriteriarequiredasatestof Hamilton’sruleforkinselection(thatthealtruistpaysatruecost),theyfailtoestablish unequivocally whether kin selection plays a role. We show that individuals from two different cultures behave in accordance with Hamilton’s rule by acting more altruistically (imposingahigherphysicalcostuponthemselves)towardsmoreclosely related individuals. Three possible sources of confound were ruled out: generational effects, sexual attraction and reciprocity. Performance on the task however did not exhibit a perfect linear relationship with relatedness, which might reflect either the intrusionofothervariables(e.g.culturaldifferencesinthewaykinshipiscosted)orthat ourbehaviouralmeasureisinsufficientlysensitivetofine-tuneddifferencesintheway individuals view their social world. These findings provide the first unequivocal experimental evidence that kinship plays a role in moderating altruistic behaviour. Kinship thus represents a baseline against which individuals pitch other criteria (includingreciprocity,prosociality,obligationandamoralsense)whendecidinghowto behavetowardsothers. *Correspondence should be addressed to E. A. Madsen, School of Psychology, University of St Andrews, South Street, StAndrews,FifeKY169JP,Scotland,UK(e-mail:[email protected]). DOI:10.1348/000712606X129213 Copyright © The British Psychological Society Reproduction in any form (including the internet) is prohibited without prior permission from the Society 340 ElainieA.Madsenetal. Recent research, using small-scale experiments based on standard game formats from economics,hasshownthatpeopleoftenapplycriteriaoffairnesstosocialinteractions, accedetosocialpressureandengagein‘altruisticpunishment’(punishmentthatincurs abenefittothecommunityasawhole,butatacosttothepunishingindividual:Fehr& Fischbacher, 2004; Fehr, Fischbacher, & G¨achter, 2002; Fehr & G¨achter, 2000, 2002; Gintis, Bowles, Boyd, & Fehr, 2003; Orstrom, Gardner, & Walker, 1994). Humans are thuscharacterizedbyanunusuallevelofprosociality:individualsare,ingeneral,willing toadheretosocialnormsofbehaviourdelineatedbytheircommunity,eventhoughthey mayincur personal costs bydoing so. Thesekindsofprosocialpredispositionsarehoweversetintoacomplexbackground ofmoreexplicitbiologicalinterestsandsocialobligations(e.g.thoseestablishedbythe customary modes of the society). Indeed, even in these micro-economic tests, individuals were not immune to the costs of altruism: when the costs to the punisher increased, the frequencies of altruistic punishment declined (Fehr & G¨achter, 2002). More importantly, however, even though kinship provides an important strand of the socialnexuswithinwhichanyprosocialbehaviourisnecessarilyembedded,ithasbeen explicitly excluded from all of these studies (as is conventional in micro-economic experiments). Altruisticbehaviouroriginallyproveddifficulttoexplaininevolutionaryterms:ina simple Darwinian world, any gene promoting altruism should be rapidly driven to extinction. The solution was provided by Hamilton (1964; popularized by Dawkins, 1976), who showed that evolution occurs as the result of copies of genes being propagated,andthatthereplicationofgenesintofuturegenerationscanoccurthrough either personal reproduction or the reproduction of individuals carrying the same gene(s).FormallyknownasHamilton’sruleofkinselection,Hamilton’scorefindingwas that a genefor altruism can evolve in apopulation, providing: rB.C where B is the benefit to the recipient, C the cost to the altruist (both measured as the number of offspring gained or lost) and r the coefficient of relationship (i.e. the probability that they share the same gene by descent). In essence, altruism evolves when the altruist’s actions allow it to contribute more copies of its genes to future generations by enabling a relative to reproduce more successfully than by reproducing itself. In effect, when all else is equal, more closely related individuals should be favoured over less closely related individuals. While the currency in Hamilton’s (1964) technical definition of altruism is fitness (for which even reproductive success is only a proxy), this cannot easily be measured directly in most real-world studies; most studies of altruism in animals have therefore used as an operational definition, the expenditure of energy or resources that may increase the welfare of one individual at the expense of another (Barash, 1977; Dawkins, 1989). Hamilton’srulehasbeenafoundationstoneofmodernevolutionarybiologyandthe study of animal behaviour for nearly half a century (e.g. Sober & Wilson, 1998), and there iswidespread observational andexperimentalevidenceofits role inmoderating altruisticbehaviourinnon-humananimals(e.g.Cheney&Seyfarth,1990).Thereisalso abundantevidencethathumansbehavemorealtruisticallytowardscloserthantowards more distant relatives. In societies reliant on subsistence agriculture, cooperation between farmers is twice as likely between relatives as non-relatives (Berte´, 1988; Copyright © The British Psychological Society Reproduction in any form (including the internet) is prohibited without prior permission from the Society Cross-culturalstudyofaltruism 341 Hames,1987).Inindustrialsocieties,beneficiariesofawilltendtoreceiveslicesofthe deceased’s estate that are broadly proportional to their genetic relatedness, r, to the deceased (Smith, Kish, & Crawford, 1987). Similarly, kinship patterns correlate with: willingnesstomurderpoliticalrivalsandformstablealliances(Dunbar,Clark,&Hurst, 1995; Johnson & Johnson, 1991), taking sides in disputes (Chagnon & Bugos, 1979; Hughes, 1988), emotional and material support within social networks (Barber, 1994; Berte´, 1988; Betzig & Turke, 1986; Dunbar & Spoors, 1995; Hames, 1987; Hill & Hurtardo, 1996), cooperation under catastrophic circumstances (Grayson, 1993; McCullough & Barton, 1991) and combat situations (Shavit, Fischer, & Koresh, 1994), the coherence and fission of cohabitational units (Chagnon, 1981; Hurd, 1983; Koertvelyessy, 1995), membership of cooperative labour units (Morgan, 1979), exchange of information among US inshore fishermen (Palmer, 1992), safeguarding againstinfanticide(Hill&Hurtado,1996)andtheinvestmenttendenciesofbothparents (Chagnon,1992;Firth,1957;Hill&Kaplan,1988)andmoredistantkin(Euler&Weitzel, 1996; Gaulin, McBurney, & Brakeman-Wartel, 1997; Hawkes, O’Connell, & Blurton- Jones,1989;Hill&Hurtado,1996;Pashos,2001).Finally,acrossmostsocieties,women arefarmorelikelytoprovidechildcarefortheirrelativesthanforunrelatedneighbours (Bereczkei, 1998; Burton, 1990; Ivey, 2000; Judge, 1995; Pennington & Harpending, 1993; Silk, 1990;Stack, 1975). Whilethesefindingsareinteresting,allthesestudiessufferonecommondrawback: theevidenceisentirelyobservational.Innocase isitpossibletoexcludecategorically all confounding variables (interaction time, reciprocity, etc.), so as to make an unequivocal test of the claim that human altruistic behaviour is modulated by the geneticrelatedness,whenotherfactorsareheldconstant.Anexceptiontothiswasan experiment reported by Caporeal, Dawes, Orbell, and van de Kagt (1989), who found that individuals were more likelyto donate their fee for participating to a local charity whendoingtheexperimentingroupsthanalone.Althoughinterpretedasevidencefor a natural tendency to behave altruistically, thus posing a challenge to Hamilton’s rule, the experiment suffered from design flaws (principally, failure to impose a cost) sufficienttocompromisethesignificanceofitsfindings.Asimilarcriticismappliestothe experimentalstudiesofprosocialitybyFehrandcolleagues(Fehr&Fischbacher,2004; Fehr et al., 2002; Fehr & G¨achter, 2002; Gintis et al., 2003): none meet the crucial criterionnecessaryforaformaltestofHamilton’srule(thatthealtruistpaysapersonal cost). Although it might be suggested that participants incur a cost by forgoing their fees,thisprovidesonlyforaveryweakformofaltruismbecausetheexperimentiscost- neutraltothealtruist(wholeavesinthesamestateashe/shearrived,apartfromatrivial lossoftime).Moreimportantly,becauseeachoftheseexperimentsfocusesonaltruism among unrelated individuals, none provide an experimental test of Hamilton’s rule in humans. In contrast, self-reported frequencies of giving and receiving assistance (financial, emotional,etc.)fromfemale residents intheUSsuggested thatintimes ofneedcloser kin were a weightier source of assistance than more distant kin and non-kin (Essock- Vitale & McGuire, 1985). This survey, however, relied on self-reports of helping behaviourthatmayhavebeenaffectedbysocialdesirabilityandleavesmanypotentially confounding variables uncontrolled. A number of studies have attempted to test the significance of kinship and Hamilton’s rule using classic social psychology vignette or questionnaire designs (Burnstein, Crandall, & Kitayama, 1994; Kruger, 2003; Park & Schaller, 2005; Webster, 2004).Inaseriesofhypotheticalscenariosinwhichparticipantsreportedthelikelihood Copyright © The British Psychological Society Reproduction in any form (including the internet) is prohibited without prior permission from the Society 342 ElainieA.Madsenetal. of helping an individual as a function of genetic relatedness (with costs ranging from trivialcourtesiestolife-and-deathheroicdeeds),collegestudentsweremoreinclinedto help relatives than non-relatives, and particularly so in life-and-death situations (Burnstein et al., 1994; Korchmaros & Kenny, 2001; Kruger, 2003). While providing primafacie evidenceinsupportofHamilton’srule,these experimental paradigms are opentoquestionongroundsofbothinternalandecologicalvalidity.Onestudy(Park& Schaller,2005),forexample,failedtoimposeanykindofcost(actualorhypothetical); others failed to eliminate non-genetic benefits (e.g. social desirability) as confounding variables. As anthropologists and social psychologists have repeatedly discovered, humans commonly say one thing, but do something entirely different (e.g. Bliezner & Adams,1992;Freeman,1999;Wolf,2001).Hence,noneofthesestudiesprovidessecure evidence on which to base a claim about the role of kin selection in actual human altruistic behaviour. Inthispaper,wereportaseriesofthreeexperimentsdesignedtotestthehypothesis that,whenallelseisequal,anindividual’swillingnesstobearagenuinecost(physical pain) to the benefit of another is a function of the recipient’s biological relatedness. Clearly, it would be improper to influence the reproductive success of either our participants or their relations to any significant degree. However, we believe that compelling evidence for altruism of this form can only be demonstrated under controlled conditions in which the costs to the participant and benefits to their recipientsareinbothcasesrealandmeasurable(asisconventionalinstudiesofaltruism in animals). While these may not be sufficient to affect reproductive success, the fact thatthecostsandbenefitsarenotmerelyhypotheticalcanbeusedtotesttheprediction that altruistic behaviour is mediated by Hamilton’s rule. We do not seek to contrast preferences for kin with other explanations for altruism, such as prosociality or reciprocity, whether direct (e.g. Trivers, 1971) or indirect (e.g. Alexander, 1987). Our aim is to assess the strength of the kinship effect so as to provide a firm background against whichthe role ofother modulating influencescan be assessed. As such,this is thefirstattempttoassessexperimentallytheimpactofkinshiponaltruisticbehaviour, rather than on declared inclinations. In addition, we consider two further issues. First, kinship has significant culture- specific overtones (Keesing, 1975), which makes it important to assess the extent to whichkin-biasedaltruismmightbeaculture-specificphenomenon.Forinstance,there is wide cross-cultural variation in kinship classification systems: social relatedness, as defined linguistically and culturally, often differs from biological relatedness (Jones, 2000; Sahlins, 1977), thereby failing to provide a reliable genealogical guide. In some terminologies, unrelated individuals may be classified as kin, while in others collateral relativesareassignedtothesamekincategoryasmuchclosergenetickin(e.g.afather’s brother’s sons are classified as brothers: Jones, 2000); furthermore, some kinship classificationsystems(e.g.thatoftheSouthAfricanZulu;Gluckman,1950)distinguish between kin of similar genealogical distance (a mother’s and a father’s sister), while others(e.g.English)donot.Suchinconsistenciesbetweenbiologicalandsocialkinship, and associated norms of behaviour risk confounding tests of genetic kinship and nepotism. We therefore conducted the experiments in two very different cultural environments (a UK student and a South African Zulu population) and asked whether the same patternsoccurred despite these cultural differences. Second,sincereciprocityisanimportantbasisforaltruisticbehaviour,weattempted tocontrolforitslikelyimpactintwoseparateways:inoneexperiment,weincludedan unrelatedbestfriendofthesamesexasaprospectiveexchangepartner(r 0)aswell ¼ Copyright © The British Psychological Society Reproduction in any form (including the internet) is prohibited without prior permission from the Society Cross-culturalstudyofaltruism 343 as a children’s charity (r 0, but no potential for future exchange). In another ¼ experiment, participants specified, for each recipient of their altruism, whether they had received helpfrom them inthe pastor expected to do so inthe future. Overview ofexperiments All the experiments followed the same basic repeated measures design. Participants were asked to impose a cost upon themselves in the form of pain from a physical exercise in return for a proportionate material reward given to an individual or organization,whoseidentitywasdisclosedtotheparticipantinadvanceofcarryingout the exercise. The exercise was a standard isometric ski-training exercise (sitting as thoughonachairwiththebackagainstawall,withthecalvesandthighsatrightangles toeachother;Cabanac,1986).Participantswereaskedtoholdthispositionforaslong aspossible.Thepositionbecomesincreasinglypainfulwithtime,becausetheweightof thetrunkisbornbythethighmusclesactingatrightanglestogravity.Thepainincurred does not bear a linear relationship to the duration for whichthe position is held: after approximately 100 seconds, it increases massively. The duration for which the participantheldthepositionwasrecordedfromthemomentthepositionwasadopted until the participant stood up or sat down on the floor. In the UK, this duration was measuredviaaninfraredlight-beam(mountedontwolaboratorystands45centimetres from the ground) that passed beneath the participants’ buttocks/thighs when the correctpositionwasadopted.Acomputerconnectedtotheinfraredcellsrecordedthe times between onset and offset of the beam. A lack of electricity at the South African sites prevented the use of electronic apparatus. Instead, the apparatus consisted of a nylontripwiremounted45centimetresfromthegroundonametalframe;durationwas recorded on a stopwatch. Participants were advised that they could cease holding the posture atanytime. The length of time a participant maintained the position on a given trial was transformedintoamaterialbenefitforarecipient,whovariedinbiologicalrelatedness to the participant (the coefficient of relationship, r, the independent variable). Each participant repeated the experiment with the recipient varied systematically between four principal categories: the participant (r 1), a sibling or parent (r :5), a ¼ ¼ grandparent, aunt, uncle, niece or nephew (r :25), a cousin (r :125). Before the ¼ ¼ experiment, each participant drew up a list of individuals for the three categories of geneticrelatedness.Care was taken to ensure that onlybiological kin wereincluded. For each experiment, the relatedness category for the recipient of the payout was randomizedacrosstrials(usingLatinsquares).Beforeeachtrial,participantsweretold thattheproceedswouldbesentdirectlytoanominatedindividual,chosenatrandomby theexperimenterfromamongthoselistedbytheparticipant.Recipientswerelimitedto individuals who were of reproductive age (16–45 years) and not cohabiting with the participant. At the end of each trial, the duration for which the position was held was determined, and the fee appropriate to this calculated. Where the participant was the beneficiary,thiswaspaiddirectlytothem;otherwise,achequefortheappropriatefee was sent to the nominated beneficiary along with a note explaining the origins of the donation. In South Africa, weused cash equivalents. Participantswerenotinformedofthetimeelapseduntilatrialhadbeencompleted,or toldthepurposeoftheexperimentuntilaftertheyhadcompletedthefullseriesoftrials. The design of the project was formally approved by the University College London Copyright © The British Psychological Society Reproduction in any form (including the internet) is prohibited without prior permission from the Society 344 ElainieA.Madsenetal. and University College Hospital Committee on the Ethics of Human Research. Participants were treated in accordance with the ‘Ethical Principles of Psychologists and Code of Conduct’ (APA, 1992) and the ‘Code of Conduct, Ethical Principles and Guidelines’(BPS,2000). If participants follow Hamilton’s rule, investment (time for whichthe position was held) should increase with the recipient’s relatedness to the participant. In effect, we tested whether investment flows differentiallydown channels of relatedness. EXPERIMENT 1 In this experiment, we tested the central prediction that willingness to invest altruistically declines as relatedness declines. We controlled for two possible confounding variables:participant gender and trial order. Method Participants Participants were recruited from the student population at the University of Oxford. Participants completed a health questionnaire, and were excluded if they were older than 40, married or declared a medical condition. Sample sizes were 11 males and 13 females. Procedure Participantsattendedonsuccessivedaysandcarriedouttheexperimentforoneofthe nominated categories of relatedness on each day. Participants received £0.40 per 20 secondsforwhichtheposturewasheld,inadditiontoaflatratefeeof£1.00forattending eachtrial.Attheendofeachtrial,participantsratedhowmuchtherecipientwouldvalue thereceiptofmoney(intermsoftheirpersonalwealthrelativetothatoftheparticipant, onascaleof1–5,withascoreof3 sameasparticipant:welfarevalue). ¼ Results The length of time (in seconds) that male and female participants maintained the posture for isshown in Figure1as a functionof the relatednessof the recipients. Thedatawerenormalizedbylog transformationsandenteredintoa2 4split-plot e £ ANOVAwithgender(malevs.female)asabetween-subjectsfactorandrecipient(degree of relatedness) as a within-subjects factor. There was no effect of gender (F 1; 21 0:14, MSE 0:82, p :71, h2 ,:01), indicating that males and females ð Þ¼ ¼ ¼ p maintained theposture for similar periods. There was a main effect of recipient (F 3; 63 8:34, MSE 0:06, p,:01, ð Þ¼ ¼ h2 :28,allpvaluesrelatingtotestsofthedirectionalityHamilton’srule’spredictions p ¼ about kinship effects are one-tailed, all others are two-tailed). The interaction failed to reach significance (F 3; 63 2:11, MSE 0:06, p :06, h2 :09). In line with self- ð Þ¼ ¼ ¼ p ¼ reports that university students are more inclined to help kin with whom they feel emotionallyclose(Korchmaros&Kenny,2001)werankedcharityaslesscloselyrelated than friend and found a reliable linear trend for recipient (F 1; 21 18:55, ð Þ¼ MSE 0:07, p,:01, h2 :47), showing that the cost that participants imposed on ¼ p ¼ themselveswas proportional to the degreeofrelatedness of therecipients. Copyright © The British Psychological Society Reproduction in any form (including the internet) is prohibited without prior permission from the Society Cross-culturalstudyofaltruism 345 Figure 1. Experiment 1: Mean (^ one SE) duration for which participants held a painful isometric skiingexerciseasafunctionoftherelatednesscategoryoftherecipientofthemoneythatparticipants earnedfromdoingso. To examine the linear trend further, we conducted planned contrasts to compare differencesindurationforeachconsecutivelevelofrecipient.Inlinewithconvention, we used the pooled error term to control for Type 1 errors. Participants imposed a higher cost when the recipient was themselves (r 1) than r :50 (F 1; 63 9:68, ¼ ¼ ð Þ¼ MSE 0:06, p,:01) and when the recipient was a r :50 than r :25 ¼ ¼ ¼ (F 1; 63 3:38, MSE 0:06, p,:05). There was no difference in duration between ð Þ¼ ¼ r :25 andr :125 (F ,1:0). ¼ ¼ Although the interaction between recipient and gender did not reach significance we examined each comparison separately for male and female participants. Males imposed a higher cost on themselves than r :50 (F 1; 63 11:52, ¼ ð Þ¼ MSE 0:06, p,:01); there was also a reliable difference between r :50 and ¼ ¼ r :25(F 1; 63 15:68,MSE 0:06,p,:01),andsurprisinglymalesinvestedmore ¼ ð Þ¼ ¼ in r :125 than r :25 (F 1; 63 6:48, MSE 0:06, p,:05). While female ¼ ¼ ð Þ¼ ¼ participants imposed a higher cost on themselves than r :50 (F 1; 63 6:48, ¼ ð Þ¼ MSE 0:06, p,:05), they showed no difference between either r :50 and r :25 ¼ ¼ ¼ (F,1.0) or r :25 vs. r :125 (F 1; 63 2:00, MSE 0:06,p..05). ¼ ¼ ð Þ¼ ¼ Discussion Thisexperimentdemonstratedthat,inlinewithourprediction,therewasasignificant andreliablelineartrendbetweentheeffortinvestedinthetask(indexedastheduration forwhichthepositionwasheld)andtherelatednessofthebeneficiary(i.e.participants differentiallyfavouredmorecloselyrelatedindividuals).Althoughthereweresignificant differences between successive categories of recipient, that between the two least relatedcategories(r :25andr :125)was,however,notitselfsignificant.Thismight ¼ ¼ be because our behavioural measure is not sufficiently sensitive to pick up the finer Copyright © The British Psychological Society Reproduction in any form (including the internet) is prohibited without prior permission from the Society 346 ElainieA.Madsenetal. differences in relatedness, or because other social variables (such as reciprocity) becomemoreimportantthanbiologicalrelatednessperse.Therewassomesuggestion of a gender difference, with females acting more equitably across the relatedness categoriesthanmales,butthemodestsamplesizemakesitdifficulttobecertainthatthe absence of a significant difference between the two genders is not simply a consequenceofgreatervarianceinthewomen’sdata.Thismightbeduetooneofthree possiblereasons:(1)womenexhibitedgreatervariancethanmen,whichmightreduce the slope of the correlation for women; (2) women may discriminate on more dimensions (e.g. friendship and time spent together, as well as kinship) than men do, andhencetheirresultsmaybelessclear-cutwhenanalysedagainstasingleindependent variable; and (3) the gender difference may be real: women may discriminate less in relation to relatedness than men do. In Experiment 2, we sought to control for these effectsto explore gender differences moreexplicitly. Wemaynotefourotherpossiblesourcesofconfoundthatmightexplaintheobserved results.First,ifparticipantsfavouredthemselvesbutdidnotdiscriminatebetweenother individuals,thismightcreateasignificantalbeitspuriouspositiverelationshipbetween relatednessandduration.Second,evolutionarytheorydoesnotconsiderrelatednessas the only criterion for investment under Hamilton’s rule. Since the central issue is maximizingthenumberofcopiesofagenepropagatedintofuturegenerations,theage ofthebeneficiaryisacriticalconsideration:olderindividualsinevitablyproducefewer future offspring than younger ones and therefore have a lower reproductive value (Fisher, 1930), making them of potentially less interest for altruistic investment (for reviews, see Barrett, Dunbar, & Lycett, 2002; Hughes, 1988). The presence of both parents and siblings in relatedness category r :5, and grandparents as well as ¼ aunts/uncles in relatedness category r :25, could have the effect of progressively ¼ reducingtheaverageperformanceifparticipantsactedsolelyintermsoftherecipients’ reproductivevalue.Thethirdpossibleconfoundisthatreciprocity,ratherthankinship, might explain the results, as reciprocity is an important factor regulating human relationships.Reciprocitymightalsoexplainthegenderdifferenceifonesex(men)acts in response to kinship, but the other (women) places greater weight on reciprocity. Fourth,recipient gender might playa significant role if disproportionately more males acted as recipientsintrialswith lesscloselyrelatedrecipients:for instance menmight behave more generously towards women than towards other men, either because of socialnormsorbecausetheywereshowingoff(aformofsexualdisplay:Hawkes,1991). EXPERIMENT 2 Inordertoexaminefurthertheapparentgenderdifferenceandthepossiblesourcesof confoundsidentifiedabove,weranasecondexperimentwithanincreasedsamplesize for both genders while controlling for reproductive value, reciprocity and recipient gender to remove some ofthe potentialconfounds inExperiment 1. Method Participants Participants were 20 male and 20 female students from the University of London. All werehealthy,singlevolunteersagedunder30years(mean 20.8,SD 2:7years).Of ¼ ¼ Copyright © The British Psychological Society Reproduction in any form (including the internet) is prohibited without prior permission from the Society Cross-culturalstudyofaltruism 347 the participants 66% described themselves as British, 15% as North American, 5% as European and 12% as Asian. Procedure TheprocedurewasasinExperiment1,exceptthattwoadditionaltrialswithunrelated recipients were included: a same-sex best friend and a national UK children’s charity. In addition, recipients were differentiated by kinship category (grandmother vs. aunt, parent vs. sibling) aswellas by relatedness (r :25or r :5). ¼ ¼ Participants were rewarded with £0.70 per 20 seconds for which the posture was held, in addition to aflatrate fee of £1.00 for attendingeachtrial. Results The mean length of time in seconds that participants maintained the posture for each levelofrecipientareshowninFigure2.Followingnormalizationbylog transformation, e these data were entered into a 2 6 split-plot ANOVA with gender as the between- £ subjectsfactorandrecipientasthewithin-subjectsfactor.Therewasnomaineffectof gender (F 1; 38 ,1:0). There was a main effect of recipient (F 5; 190 10:52, ð Þ ð Þ¼ MSE 0:05, p,:01, h2 :22). The interaction between the two failed to reach ¼ p ¼ significance(F 5; 190 1:76,MSE 0:05,p :12,h2 :04).AsinExperiment1and ð Þ¼ ¼ ¼ p ¼ consistent with our hypothesis, the linear trend of recipient was reliable (F 1; 38 31:27,MSE 0:06,p,:01,h2 :45). ð Þ¼ ¼ p ¼ Planned contrasts revealed no reliable difference between r 1 and r :50 ¼ ¼ (F 1; 190 1:54, MSE 0:05, p..05). There was a reliable difference between ð Þ¼ ¼ Figure 2. Experiment 2: Mean (^ one SE) duration for which participants held a painful isometric skiingexerciseasafunctionoftherelatednesscategoryoftherecipientofthemoneythatparticipants earnedfromdoingso.