Genes and Cultures - Stanford University

21 Pages·2003·0.18 MB·English

Current Anthropology Volume44,Number1,February2003
(cid:1)2003byTheWenner-GrenFoundationforAnthropologicalResearch.Allrightsreserved0011-3204/2003/4401-0004$3.50
The recent publication of the first draft of the human
genome (e.g., Venter et al. 2001, Lander et al. 2001) has
✩
CA FORUM ON
broughttopublicattentiontherelationshipbetweentwo
concepts, genotype and phenotype—a relationship that
ANTHROPOLOGY IN PUBLIC
hadpreviouslybeendiscussedlargelybyacademics.The
genotype of an organism is encoded in the DNA thatis
held in chromosomes and other structures inside its
Genes and Cultures cells.Thephenotypeiswhatweareabletoobserveabout
that organism’s biochemistry, physiology, morphology,
and behaviors. We will use the term “phenome” tocir-
cumscribeasetofphenotypeswhosepropertiesandvar-
iabilitywewishtostudy.Ourfocuswillbeonthatpart
What Creates Our Behavioral
ofthehumanphenomethatisdefinedbybehaviorsand
Phenome? especiallyonthebehavioralphenome’sconnectionwith
the human genome.
Our understanding of human behavioral traits has
evolved; explanations of the control of those traits of-
by Paul Ehrlich and
fered50yearsagodifferfromthosemostcommontoday.
In prewar decades genetic determinism—the idea that
Marcus Feldman1
genes are destiny—had enormous influence on public
policyinmanycountries:onAmericanimmigrationand
racial policies, Swedish sterilization programs, and, of
course,Nazilawsonracialpurity(Ehmann2001,Ehrlich
and Feldman 1977, Fisher 2001). Much of this public
Acentralthemeofthefloodofliteratureinrecentyearsin“evo-
policywasbuiltonsupportfrombiological,medical,and
lutionarypsychology”and“behavioralgenetics”isthatmuchor
socialscientists(e.g.,Brigham1923,Goddard1917,Ter-
evenmosthumanbehaviorhasbeenprogrammedintothehu-
mangenomebynaturalselection.Weshowthatthisconclusion man1916),butafterHitler’sgenocidalpoliciesitwasno
iswithoutbasis.Evolutionarypsychologyisaseriesof“just-so” longerpoliticallycorrecttofocusonputativehereditary
storiesrootedinpartintheerroneousnotionthathumanbeings differences. The fading of genetic determinism was an
duringthePleistocenealllivedinthesameenvironmentofevo-
understandable reaction to Nazism and related racial,
lutionaryadaptation.Behavioralgeneticsisbasedonaconfusion
oftheinformationcontainedinatechnicalstatisticcalled“heri- sexual, and religious prejudices which had long been
tability”withthecolloquialmeaningoftheterm,exacerbatedby prevalentintheUnitedStatesandelsewhere.Thus,after
oversimplificationofstatisticalmodelsforthebehavioralsimi- WorldWarII,itbecamethenorminAmericanacademia
larityoftwins.Infact,informationfromtwinstudies,cross-fos-
to consider all of human behavior as originating in the
tering,sexualbehavior,andtheHumanGenomeProjectmakes
itabundantlyclearthatmostinterestingaspectsofthehuman environment—inthewaypeoplewereraisedandtheso-
behavioralphenomeareprogrammedintothebrainbytheenvi- cial contexts in which they lived.
ronment.Thegeneralconfusioncreatedbythegeneticdetermin- Gradually,though,beginninginthe1960s,bookslike
istshashadandwillcontinuetohaveunfortunateeffectson
Robert Ardrey’s Territorial Imperative (1966) and Des-
publicpolicy.
mond Morris’s The Naked Ape (1967) began proposing
explanationsforhumanbehaviorsthatwerebiologically
reductionist and essentially genetic. Their extreme
hereditarianbiasmayhavebeenstimulatedbytherapid
progress at that time in understanding of the role of
DNA, which spurred interest in genetics in both scien-
tists and the public. But perhaps no publication had
broadereffectinreestablishinggeneticcredibilityinthe
behavioral sciences than Arthur Jensen’s (1969) article
“HowMuchCanWeBoostIQ?”Althoughroundlycrit-
icizedbyquantitativegeneticistsandshowntobebased
on the fraudulent data of Sir Cyril Burt (Kamin 1974),
Jensen’s work established a tradition that attempts to
allocate to genetics a considerable portion of the varia-
tioninsuchhumanbehaviorsasforwhomwevote,how
religiousweare,howlikelywearetotakerisks,and,of
course, measured IQ and school performance. This tra-
dition is alive and well today (e.g., Plomin, Owen and
McGuffin 1994, Plomin et al. 1997).
Within the normal range of human phenotypic vari-
1.DepartmentofBiologicalSciences,StanfordUniversity,Stanford,
Calif.94305-5020,U.S.A.([email protected]). ation, including commonly occurring diseases, the role
87
88 F current anthropology Volume44, Number1, February2003
ofgeneticsremainsamatterofcontroversyevenasmore isaboutthedegreetowhichdifferencesintoday’shuman
isrevealedaboutvariationatthelevelofDNA(Pritchard behavioral patterns from person to person, group to
2001, Reich and Lander 2001). Here we would like to group, and society to society are influenced by genetic
reexamine the issue of genetics and human behaviorin differences,thatis,aretraceabletodifferencesinhuman
light of the enormous interest in the Human Genome geneticendowments.Domen“naturally”wanttomate
Project,theexpansionofbehavioralgeneticsasdescribed with as many women as possible while women “natu-
above,andtherecentproliferationofbooksemphasizing rally” want to be more cautious in choosing their cop-
the genetic programming of every behavior from rape ulatorypartners(Bermant1976,Symons1979,Birkhead
(ThornhillandPalmer2000)tothelearningofgrammar 2000;seealsoSmall1993:chap.7)?Istherea“gaygene”
(Pinker 1994). The philosopher Helena Cronin and her (Hameretal.1993,Huetal.1995,Riceetal.1999)?Are
coeditor,OliverCurry,tellusintheintroductiontoYale human beings “innately” aggressive (Ehrlich 2000:
University Press’s “Darwinism Today” series that 210–13)? Are differences in educational achievementor
“Darwinian ideas . . . are setting today’s intellectual income “caused” by differences in genes (Bowles and
agenda” (1999). In the New York Times,NicholasWade Gintis 2001, Jacoby and Glauberman 1995, Lewontin,
(2000) has written that human genes contain the “be- Rose,and Kamin 1984,Taubman1976)?Andarepeople
havioralinstructions”for“instinctstoslaughterorshow ofallgroupsgeneticallyprogrammedtobeselfish(Ham-
mercy, the contexts for love and hatred, the taste for ilton 1964, Richerson and Boyd 1978)? A critical social
obedienceorrebellion—theyarethedeterminantsofhu- issue to keep in mind throughout our discussion is
man nature.” what the response of our society would be if we knew
the answer to these questions. Two related schools of
thought take the view that genetic evolution explains
Genes, Cultures, and Behavior muchofthehumanbehavioralphenome;theyareknown
as evolutionary psychology and behavioral genetics.
Itisincontrovertiblethathumanbeingsareaproductof
evolution,butwithrespecttobehaviorthatevolutionary
process involves chance, natural selection, and, espe- Evolutionary Psychology
cially in the case of human beings, transmission and
alteration of a body of extragenetic information called Evolutionary psychology claims that many human be-
“culture.” Cultural evolution, a process very different haviors became universally fixed as a result of natural
from genetic evolution by natural selection, has played selectionactingduringtheenvironmentofevolutionary
acentralroleinproducingourbehaviors(Cavalli-Sforza adaptation (Tooby and Cosmides 1992), essentially the
and Feldman 1973, 1981; Ehrlich 2000; Feldman and Pleistocene.Ashortcomingofthisargument,asempha-
Cavalli-Sforza 1976; Feldman and Laland 1996). sized by the anthropologist Robert Foley (1995–96),lies
Thisisnottosaythatgenesareuninvolvedinhuman in the nonexistence of such an environment. Our an-
behavior.Everyaspectofaperson’sphenomeisaproduct cestorslivedinawidediversityofhabitats,andtheim-
ofinteractionbetweengenomeandenvironment.Anob- pacts of the many environmental changes (e.g., glacia-
viousexampleofgeneticinvolvementinthebehavioral tions)overthepastmillionyearsdifferedgeographically
phenome is the degree towhich mostpeopleusevision amongtheirvariedsurroundings.Evolutionarypsychol-
to orient themselves—in doing everything from hitting ogists also postulate that natural selection produced
a baseball to selecting new clothes for their children. modules (“complex structures that are functionally or-
Thisisbecausewehaveevolvedgeneticallytobe“sight ganized for processing information” [Tooby and Cos-
animals”—our dominant perceptual system is vision, mides 1992: 33]) in the brain that “tell” us such things
with hearing coming in second. Had we, like dogs, as which individuals are likely to cheat, which mates
evolved more sophisticated chemical detection, we arelikelytogiveusthebestormostoffspring,andhow
might behave very differently in response to the toxic to form the best coalitions (Kurzban, Tooby, and Cos-
chemicals in our environment (Ehrlich 2000). The in- mides2001).Thesebrain“modules,”whichareassumed
formationinourDNArequiredtoproducethebasicmor- to be biological entities fixed in humans by evolution,
phologyandphysiologythatmakesightsoimportantto also have other names often bestowed on them by the
ushasclearlybeenmoldedbynaturalselection.Andthe same writers, such as “computational machines,” “de-
physical increase in human brain size, which certainly cision-making algorithms,” “specialized systems,” “in-
involved a response to natural selection (although the ference engines,” and “reasoning mechanisms” (Du-
precise environmental factors causing this selection re- chaine,Cosmides,andTooby2001).Theresearchclaims
mainsomethingofamystery[Allman1999,Klein1999]), of evolutionary psychology havebeenheavilycriticized
has allowed us to evolve language, a high level of tool by,amongothers,colleaguesinpsychology(e.g.,Bussey
use, the ability to plan for the future, and a wide range and Bandura 1999).
of other behaviors not seen in other animals. Those critics are correct. There is a general tendency
Thus at the very least, genetic evolution both biased for evolutionary psychologists vastly to overestimate
ourabilitytoperceivetheworldandgaveusthecapacity how much of human behavior is primarily traceable to
to develop a vast culture. But the long-running nature- biologicaluniversalsthatarereflectedinourgenes.One
versus-nurturedebateisnotaboutsightversussmell.It reason for this overestimation is the ease with which a
ehrlich and feldman Genes and Cultures F 89
little evolutionary story can be invented to explain al- engraved in our genetic makeup is without basis, espe-
most any observed pattern of behavior. For example, it cially in light of the enormous cultural differences in
seems logical that natural selection would resultinthe sexual preferences.
coding of a fear of snakes and spiders into our DNA,as Foranyculture,Ellis’sevolutionaryargumentswould
the evolutionary psychologist Steven Pinker thinks require that in past populations of women there were
(1997: 386–89). But while Pinker may have genes that DNA-based differences that made some more likely to
makehimfearsnakes,astheevolutionistJaredDiamond choose in those ways and others more likely to seek
pointsout,suchgenesareclearlylackinginNewGuinea mates with other characteristics. And those that chose
natives. As Diamond says, “If there is any single place as Ellis predicts would have to have borne and raised
in the world where we might expect an innate fear of morechildrenthatsurvivedtoreproducethanthosewith
snakes among native peoples, it would be in New other preferences. Might, for example, a woman who
Guinea, where one-third or more of the snake species married a stingy male who kept her barefoot and preg-
are poisonous, and certain non-poisonous constrictor nant out-reproduce the wife of a generous and consid-
snakesaresufficientlybigtobedangerous.”Yetthereis erate mate? That is the way genetic evolution changes
no sign of innate fear of snakes or spiders among the the characteristics of populations over time: by some
indigenouspeople,andchildrenregularly“capturelarge geneticvariants’out-reproducingothers.Whenthathap-
spiders, singe off the legs and hairs, and eat the bodies. pens, we say that natural selection has occurred. But,
The people there laugh at the idea of an inborn phobia unfortunately, there are no data that speak to whether
about snakes, and account for the fear in Europeans as there is (or was) genetic variation in human mate pref-
a resultof theirstupidityinbeingunabletodistinguish erences—variationin,say,abilitytoevaluatespecifically
whichsnakesmightbedangerous”(1993:265).Further- whether a potential mate is “ambitious”—upon which
more, there is reason to believe that fear of snakes in selection could be based. And there are no data for any
other primates is largely learned as well (Mineka, Keir, population showing that women who seek those char-
and Price 1981, Mineka and Cook 1993). acteristics in their sexual partners are more successful
Anotherexampleisthesetofpredictionsadvancedby reproductively—arerepresentedbymorechildreninthe
BruceEllis(1992)aboutthematingbehaviorthatwould subsequent generation—than women who seek hus-
bands with other characteristics.Ellis issimplyconfus-
befoundinapreviouslyunknownculture.Thefirstfive
ing the preferences of women he knows in his society
characteristics that “the average woman in thisculture
with evolutionary fitness.
willseek...inheridealmate,”hepredicts,are(p.283):
1. He will be dependable, emotionally stable and
mature, and kind/consideratetoward her. Behavioral Genetics and Heritability
2. He will be generous. He may communicate a
spirit of caring through a willingness to share time Another reason laypersons tend to overestimate how
and whatever commodities are valued in this culture much of our behavior is genetically determined derives
with the woman in question. from the claims of some scientists that the variation
3. He will be ambitious and perceived by the between individuals in behaviorsisduetotheirgenetic
woman in question as clever or intelligent. differences.Oftenthesesamescientistslooktoadvances
4. He will be genuinely interested in the woman in moleculargeneticsasakindofjustification.Thefol-
in question, and she in him. He may express his in- lowing quote from a recent book by the biochemists
terest through displays of concern for her well-being. Dean Hamer and Peter Copeland (1998) reflects the at-
5. He will have a strong social presence and be titude of many behavioral geneticists: “The emerging
well liked and respected by others. He will possess a science of molecular biology has made startlingdiscov-
strong sense of efficacy, confidence, and self-respect. eriesthatshowbeyondadoubtthatgenesarethesingle
most important factor that distinguishes one person
Evolutionary theory does not support such predictions, from another. We come in large part ready-made from
even if an “average woman” could be defined. First of thefactory.Weacceptthatwelooklikeourparentsand
all,itwouldbenosmalldevelopmentaltrickgenetically other blood relatives; we have a harder time with the
to program detailed, different, and independent repro- idea that we act like them” (p. 11).
ductive strategies into modules in male and female Hamer and Copeland’s view has a long history. For
brains. Those brains, after all are minor variants of the example, in his influential monograph, Arthur Jensen
same incredibly complex structures, and, furthermore, (1969) claimed that the high heritability (a statistical
the degree to which they are organized into modules is value) of IQ made it unlikely that environmentalinter-
far from clear (Ehrlich 2000: 115–19). If the women in ventioncouldsucceedinimprovingtheeducationalper-
the unknown culture actually chose mates meeting El- formanceofdisadvantagedchildren;thefaultlayintheir
lis’s criteria, a quite sufficient alternative evolutionary genes.Thenextthreedecadessawthegrowthofbehav-
explanationwouldbethatwomen(simultaneouslywith ioralgenetics,apartofpsychologybuiltaroundthesta-
men)haveevolvedbigbrains,arenotstupid,andrespond tisticalcomparisonofidenticalandfraternaltwins.Most
tothenormsoftheircultures.Scientifically,thenotion of these studies follow Jensen’s example and produce a
that the detailed attributes of desirable mates must be highheritability,whichiswronglyinterpretedasamea-
90 F current anthropology Volume44, Number1, February2003
sure of how important genes are in determining differ- This broad-sense heritability has no predictive value
ences between individuals in the behavior under study. and indeed cannot be legitimately used in the human
For instance, the law professor Kingsley Browne (1998: behavioral context to predict anything. Ithas,however,
27) tells us that “evidence from behavioral genetics in- been widely misinterpreted as diagnostic of the under-
dicatesthatmanypersonalitytraitsarehighlyheritable; lying causes of variation. Thus, in a recent perspective
thatis,muchoftheirindividualvariationisattributable in the widely read magazine Science,the behavioralge-
to genetic differences among individuals.”Weoftensee neticistsP.McGuffin,B.Riley,andR.Plomin(2001)infer
headlines in major newspapers that summarize the that“DNAvariationsareresponsiblefortheubiquitous
claims of behavioral geneticists with “Gene for Happi- geneticinfluenceinbehavior”fromtheclaimthat“the
nessFound”or“WeVotewithOurGenes.”Butthetext most solid genetic findings about individualdifferences
is not really about genes but about the behavioral ge- in human behavior come from quantitative genetic re-
neticists’ interpretation of their own estimates of heri- search such as twin and adoption studies that consis-
tability computed from twin studies. tentlyconvergeontheconclusionthatgeneticvariation
The behavioral genetics literature is based on studies makesasubstantialcontributiontophenotypicvariation
of identical and fraternal twins combined with a set of for all behavioral domains.” In other words, they claim
statisticalassumptionsaboutgeneticandenvironmental thatthestatisticalmeasureofbroad-senseheritabilityis
contributions that are used to extract estimates of how tellingusaboutthecausesofthebehavioraldifferences,
important genes are in determining behaviors (e.g., in particular how “genetic” they are.
Plomin et al. 1997). We shall examine this heritability The kind of statistical reasoning that underlies this
paradigm in some detail andthenseewhatnewknowl- imputedconnectionbetweentheinformationcodedinto
edge about the human genome can tell us about it. DNAandheritabilityreliesonaparticularmodelofhow
Heritabilitywas originallyintroducedinthe1930sas that information causes behaviors, a model that is not
an indexofamenabilitytoselectivebreedingofagricul- verifiable because we have no idea about how thecom-
tural animals and plants (e.g., Lush 1945; Falconer and plexinteractionsbetweengenes,regulationofgenes,pro-
Mackay 1996: chap. 10). Under carefully controlled en- tein structures, protein concentrations, and environ-
vironmental conditions it measures the fraction of ge- ments would be manifest in a measurable trait or
netic variation that would respond to selection by the behavior. Scientists don’t know what model to use to
breeder on a trait such as fat content in milk or egg compute the degree of“geneticcausation.”Inoneclass
production in chickens. An accurate measure of herita- of such models, for example, Robert Cloninger and col-
bility requires that parents and offspring be raised in leagues(1979)showedthatheritabilitiesweremadevery
identical environments. This original narrowdefinition highbyusingthedoubtfulassumptionthatidenticaland
waspredictive—ittoldtheexperimenterwhathadtobe fraternaltwinshadthesamedegreeofsimilarityintheir
done to move the desired trait in a given direction bya environments.Inanotheranalysis,Devlin,Daniels,and
given amount. The definition of heritability was later Roeder (1997) showed that omission of a contribution
modified,broadenedinfact,toincludegeneticvariation fromthesharedprenatalenvironmentoftwinsalsoleads
thatwasunresponsivetoselectionandtoaccommodate to elevated estimates of heritability. In fact, calculated
thefactthatgenotypesandenvironmentsmightinteract heritabilities give us no information concerning the
in a way that could not be estimated or controlled, es- causes of our actions. The basic reason is that it is im-
pecially in the case of human behaviors2 (Falconer and possibletodistinguishhumanbehavioralphenomesthat
Mackay 1996: 123). are shared because of genetic similarities from those
caused by shared environments. We might act like our
2.Theusualmodelinbehavioralgeneticstakesthephenotypeto parents because they gave us our genes; however, as
bealinearcombinationofgeneticandenvironmentaleffects:Pp RichardLewontinpointedout,“IntheUnitedStates,the
hG (cid:1) eE. In this statistical representation, the square of h is the highestcorrelationsbetweenparentandoffspringforany
broad-senseheritability,anumberbetween0and1.0(or100%),e
socialtraitsareforreligioussectandpoliticalparty.Only
isthecorrespondingfractionofthephenotypeduetothe(nontrans-
mitted)environmentE,andGandEoperateindependently.Itisin the most vulgar hereditarian would suggest that Epis-
the quantityG that the action ofthe DNAis summarized.Fora copalianismandRepublicanismaredirectlycodedforin
single gene, the contributions to an individual’s phenotype that the genes” (Lewontin, Rose, and Kamin 1984: 256).
come from maternal and paternal contributions (alleles) may be
Much has been recently made by behavioral geneti-
independentandsummed,inwhichcasetheactionofthatgeneis
purelyadditive,ortheymayinteractinsomewaythatismeasured cists of heritabilityestimatesforbehavioraltraitsbased
bygeneticdominance.Whentwoormoregenesinteracttoproduce on data compiled in two twin studies: the Minnesota
theircontributionstothephenotype,wecallthegenes“epistatic,” StudyofTwinsRearedApart(MISTRA)andtheSwedish
andthepartoftheircontributiontothevariabilityofthephenotype Adoption-Twin Study of Aging (SATSA). Results from
thatisnotthesumoftheirindividualcontributionsiscalledepis-
these studies are widely cited in textbooks on the ge-
tasis.Theseareallstatisticalnotionsaboutvariancethatarevery
difficulttotranslateintogeneticstructuralorregulatoryphenom- neticsofhumanbehaviors(e.g.,Plominetal.1997),but
ena. The fraction of the variance of P that is due to variation in it is only recently that the statistical assumptions un-
theadditivecontributionstoGiscalledthe“narrow-sense”heri- derlying these analyses and the inconsistencies in the
tability.ThefractionofthevarianceofPduetoGandtopossible
reportingofestimateshavecomeundercarefulscrutiny
interactions between G and E is called the “broad-sense”herita-
bility.Thelatteriswhatismostoftenreferredtoastheheritability (Devlin, Daniels, and Roeder 1997, Feldman and Otto
inbehavioralgenetics. 1997, Goldberger and Kamin 2002). Goldberger and Ka-
ehrlich and feldman Genes and Cultures F 91
min point out that “the only genetical theory involved ius”level.Woulditthenbegoodpolicytogiveremedial
in their analysis are the numbers 1, 1/2, and 1/4 repre- aidtoallwhitesandnonetoanyblacks?Orwoulditbe
senting3 the genotypic correlations for identical twins wisertogiveadditionalhelptothosewhohadlowscores
and the additiveandnon-additivegenotypecorrelations regardlessofskincolor?What,infact,wouldbetherea-
for fraternal twins.” Not only do these authorsfindthe sonforevenbotheringtocalculatethegroupaverageIQ
conclusionsfromMISTRAandSATSAunconvincingbut scores? Wouldwe calculatethemforpopulationsdiffer-
they raise the important question, ignored in the now entiated on the basis of other characteristics, such as
large literature on behavioral correlations among rela- bloodgroups?Infact,theusualphysicaland/orcultural
tives, “What conceivable purpose is served by theflood criteria used to define ethnic groups may have little to
of heritabilityestimates generated by these studies?” dowiththegeneticclassificationofsuchgroups(Wilson
et al. 2001).
It is only because people live in socially stratifiedso-
Policy Implications of Heritability cietiesandhaveafascinationwithskincolor(orheight,
or nose shape—after all, we are sight animals) that dif-
Perhapsmostimportant,degreeofheritabilitycarriesno ferences between certain groups are singled out for in-
message about how easily a characteristic can be vestigationviaheritabilities.IfaveragedifferencesinIQ
changed, and, normally, knowledge of it will have few test scores are correlated withskincolorinoursociety,
if any policy implications. Heritable diseases are rou- shouldwetrytoreducetheincidenceoflowtestscores
tinely treated (e.g., phenylketonuria),asarediseasesbe- by treating skin-color groupsdifferently?Ofcoursenot,
lieved to have little relationship to the victim’sgenetic anymorethanwewouldattempttolowertheincidence
endowment (e.g., endocarditis). Similarly, even if a be- ofskincancer(towhichlighter-skinnedpeoplearemore
havior had a high degree of heritability in one environ- susceptible) by doling out sunscreen on the basis of IQ
ment, a small environmental alteration could totally testscores.Smartsocialpolicywouldbetoaidindividual
change that behavior. The literature on quantitative students with low scores regardless of skin color and
traits in plants, insects, and animals is replete with ex- regardless of what role genes played in determining in-
periments that show the sensitivity of measured heri- dividual IQs.
tability to changes in the environment. Jensen’s proposed heritability of 80% for IQ should
Furthermore,itwouldbefoolishtomakesocialpolicy never have been used to blame the failure in school of
designedtoalterbehavioronthebasisofgroupaverages some groups of students on their genes. Nor should
in characteristics, regardless of the reasons for the dif- Herrnstein and Murray have used their value of 60%
ferencesinthosecharacteristics.Considerathoughtex- (whichtheyfeel“mayerronthelowside”)tounderpin
perimentonthefrequentlypromoted(withnoevidence) their claim that “chances of success in life are increas-
view that there are differences between populations in ingly affected by genes” (1994: 109–10) and their reit-
genes influencing intelligence. Suppose that,counterto eration 25 years later of Jensen’s claim that environ-
everythinggeneticistsknow,thereweresomethingthat mentalinterventioninthelivesofthedisadvantagedin
could be called “genetic IQ” and some way were dis- the United States was doomed to failure (pp. 551–52).
coveredtoassessit—somesortofcognitivelitmuspaper It has been fascinating and disturbing for us as biolo-
on which, when placed on the forehead, a number mi- gists to watch the legacy of Jensen’s 1969 opus unfold
raculously appeared, faultlessly indexing the “genetic in psychology. Thus, one of the leading proponents of
IQ”ofthatindividual.Supposefurtherthataverage“ge- theuseofheritability,RobertPlomin,writingintheThe
netic IQ” litmus-test scores tended to be somewhat Psychologist, claims that “during the 1980s and espe-
higherintheblackpopulation,eventhoughmanywhites ciallythe 1990s psychologybecamemuchmoreaccept-
scoredmuchhigherthanmanyblacks,someatthe“gen- ingofgeneticinfluence,ascanbeseenintheincreasing
numberofbehavioralgeneticarticlesinmainstreampsy-
3.Referringtothemodelofn.1,PphG(cid:1)eE,foranytwoindi- chology journals and in research grants” (Plomin2001).
viduals labeled 1 and 2, we can write P p hG (cid:1) eE and P p He goes on to describe this change in psychology as a
hG (cid:1) eE wherethegeneticandenviro1nmenta1lcontr1ibution2sto “waveofacceptanceofgeneticinfluenceinpsychology.”
2 2
thephenotypesP1andP2ofthetwoindividualsareG1,G2andE1,E2, Thisacceptanceisentirelyduetothewidespreadaccep-
respectively.Now,foridenticalormonozygous(MZ)twinsG and
1 tance of the statistical methodology that leads to the
G are the same because their complete complement of genes is
th2esame;thecorrelationbetweenG andG is1.Ifindividuals1 reportingofbroad-senseheritabilityanditsmisinterpre-
1 2
and2aresibs,thenitcanbeshownthatthecorrelationbetween tation as an index of genetic causality, not to any
the additive contributionstoP andP containedinG andG is neurogeneticadvancesthathavetiedhumanbehavioral
1 2 1 2
1/2 while that between the dominance contributions is 1/4 (see, differences to variation in DNA. In a similar vein, the
e.g., Falconer and Mackay 1996 for more details). Almost all be-
psychologistM.McGue(1997)claims,forexample,that
havioral genetic studies assume that the correlation between en-
vironments E and E when individuals 1 and 2 are MZ twins is “the IQ debate now centers on whether IQ is 50% to
1 2
the same as when they are fraternal or dizygous (DZ) twins. Yet 70% heritable.”
whenthereareenoughdataforthesecorrelationstobecompared, Our point is that the assumptions used to build the
for example,forIQ,theMZvalueislargerthantheDZ value.If
statistical models that produce these estimates do not
this difference in environmental correlations is ignored, a higher
permit us to infer from such heritability estimates the
estimate for heritability is reported (Cloninger, Rice, and Reich
1979,FeldmanandOtto1997). actual extent of “genetic influences” on IQ. Further,
92 F current anthropology Volume44, Number1, February2003
these estimates do not inform potential strategies for ior—the very attributes that make our behavior less
determiningthenatureofsuchgeneticinfluences,ifthey rather than more genetically determined. But to under-
exist. Applications of broad-sense heritabilitytopredic- stand the development of and variation in specific hu-
tive situations are, we repeat, biologically and statisti- man behaviors such as creating charities and cheese-
cally erroneous (Feldman and Lewontin 1975). Evolu- cakes, we must invoke culture, its evolution, and its
tionary psychologists and behavioral geneticists persist potential interaction with biology.
in confounding a technically defined statistic named Itmightbearguedthatsincearelativehandfulofgenes
“heritability” with the colloquial use of thatterm.The can control our basic body plan—one’s height depends
conceptofoverallheritabilityshouldberestrictedinits on millions of the body’s cells’ being stacked pre-
employmenttoplantandanimalbreeding,whereitcan cisely—a handful could also determine our behavioral
bebettermeasuredandtheresultsputtosomepractical phenome. Genes initiate a process of developmentthat
use—such as in applying selection to increase the rate might be analogized with the way a mountain stream
of growth of beef cattle or the weight of swine. entering a floodplain can initiate the development of a
complexdelta.Why,then,couldn’tjustafewgeneshave
evolvedtoprogrammillionsofourbehaviors?Intheory
What Does Determine the Behavioral theymighthave,butinthatcasehumanbehaviorwould
be very stereotyped. Consider the problem of evolving
Phenome?
human behavioral flexibility under such circumstances
of genetic determination. Changing just one behavioral
Geneticists know that a large portion of the behavioral pattern—say, making women more desirous of mating
phenomemustbeprogrammedintothebrainbyfactors with affluent men—would be somewhat analogous to
intheenvironment,includingtheinternalenvironment changing the course of one distributary (branch in the
in which the fetus develops and, most important, the delta) without altering the braidedpatternoftherestof
cultural environment in which human beings spend thedelta.Itwouldbedifficulttodobyjustchangingthe
their entire lives. Behavioral scientists know, for in- flowofthemountainstream(equivalenttochangingthe
stance,thatmanydramaticpersonalitydifferencesmust genes)buteasilyaccomplishedbythrowingbigrocksin
betracedtoenvironmentalinfluences.Perhapsthemost the distributary(changing the environment).
important reason to doubt that genetic variation ac- This partial analogy seems particularly apt in that it
counts for a substantial portion of observed differences is apparently difficult for evolution to accomplish just
in human behavior is simply that we lack an extensive one thing at a time. There are two principalreasonsfor
enoughhereditaryapparatustodothejob—thatwehave this. The first is the complexity of interactions among
a “gene shortage” (Ehrlich 2000). To what extent could alleles and phenotypic traits, especially pleiotropy and
genes control the production of these differences? epistasis. Because there are relatively so few of them,
It is important to remember that behaviors are the most genes must be involved in more than one process
results of charge changes that occur in our network of (pleiotropy).Thenifamutationleadstobetterfunction-
neurons,thespecializedcellsthatmakeupournervous ingofoneprocess,itmaynotbeselectedforbecausethe
system. Behaviors are ultimately under some degree of changemightdegradethefunctioningofanotherprocess.
control in the brain. Neuron networks are the locus of And changes in one gene can modify the influence of
the memories that are also important to our behavior. another in very complex ways (epistasis). Second, be-
That genes can control some general patterns is un- cause they are physically coupled to other genes on the
questioned;theyareobviouslyinvolvedintheconstruc- same chromosome, the fates of genes are not indepen-
tionofourbrains.Theymightthereforealsobuildinthe dent.Selectionthatincreasesthefrequencyofoneallele
potentialforexperiencetoaffectalargepartofthedetails inapopulationwilloften,becauseoflinkage,necessarily
involvedintheneuralcircuitry.Buttheycannotbecon- increase the frequency of another. Selection favoring a
trolling our individual behavioral choices. gene that made one prefer tall mates might also result
Human beings have only three times as many genes in the increase of a nearby gene that produced greater
as have fruitflies (manyofthosegenesappeartobedu- susceptibility to a childhood cancer.
plicatesofthoseintheflies,andthebiochemistryoffly
nerve cells seems quite close to ours) (Zigmond et al.
1999: figs. 9.8, 9.9). But in addition to having sex and The Mysteries of Environmental Control
eating (what flies mostly do) we get married, establish
charities,buildhydrogenbombs,commitgenocide,com- Behavioral scientists are still, unhappily, generally un-
posesonatas,andpublishbooksonevolution.Itisalittle able to determine the key environmental factors that
hardtocreditallthistothedeterminingactionofthose influence the behavioral phenome. For instance, in the
fewadditionalgenes(Ehrlich2000:124–26).Thosegenes case of the Dionne quintuplets, quite subtle environ-
are,however,likelytohavecontributedtotheincreased mental differences—perhaps initiated by different posi-
brainsizeandcomplexitythatsupportthevastcultural tions in the womb or chance interactionsamongyoung
superstructurecreatedbytheinteractionofourneurons quints, their parents, and their observers (Blatz
andtheirenvironments.Theymayalsocontributetothe 1938)—clearly led to substantially different behavioral
wonderful flexibility and plasticity of human behav- and health outcomes in five children with identical ge-
ehrlich and feldman Genes and Cultures F 93
nomes.As theirstoryshows,wereallyknowverylittle havior were completely genetically determined and in-
about what environmental factorscanmodifybehavior. teractions between its genes and its environments did
For example, some virtuallyundetectabledifferencesin not exist. Even single-celled organisms respond to
environments may be greatly amplified as developing changes in their surroundings. Without substantial en-
individualschangetheirownenvironmentsandthoseof vironmental inputs, evolution would not occur and life
their siblings. Equally, subtle and undetected environ- could not exist.
mental factors may put individuals with the same ge- Biological evolution has avoided that problem by al-
netic endowments on similar life courses even if they lowing our behavior to be deeply influenced by the en-
arerearedapart,perhapsexplaininganecdotesaboutthe vironments in which genes operate. In normal human
similarities of some reunited identical twins. environments, genes are heavily involved in creating a
We also know too little about the routes through basic brain with an enormous capacity for learn-
which genes may influence behavior, where again ing—taking in information from the environment and
changes may be behaviorally amplified. Suppose that a incorporatingthatinformationintothebrain’sstructure.
study shows that identical twins, separated at birth, Itislearningthatproceedsafterbirthasaninfant’sbrain
nonetheless show a high correlation of personality uses inputs such as patterns of light from the eyes to
type—both members of twin pairs tend to be either in- wire up the brain so that it can see, patterns of sound
troverted or extraverted. This is interpreted as a high that wire up the brain so that it can speak one or more
heritabilityofintroversionandextraversion.Whatreally languages,andsoon.AsthebrainscientistJohnAllman
is heavily influenced by genetics, however, could be putit,“thebrainisuniqueamongtheorgansofthebody
height, and tall people in that society (as in many in requiring a great deal of feedback from experienceto
societies) may be better treated by their peers and thus developitsfullcapacities”(1999:177).Andthesituation
more likely to become extraverted (Buss 1994: 39–40). is notsodifferentforheight.Therearen’tenoughgenes
Genesinthiscasewillclearlybeinvolvedinpersonality tocontrolachild’sgrowthratefromdaytoday—adding
type but by such an indirect route as to make talk of cells rapidly in favorable (e.g., food-rich) situations and
“genes for introversion or extraversion” essentially slowly or not at all under starvation. And there aren’t
meaningless. enough genes to govern the growth of each column of
And, of course, scientists do know that what appears cells,sometoregulatethoseineachcolumnontheright
to be “genetic” is often simply a function of the envi- side of the spine, some for each in the left. Instead, all
ronment. An example suggested by the philosopher El- growth patterns depend on environmental feedback.
liott Sober (personalcommunication)illustratesthis.In
England before the 18th century, evolutionary psychol-
ogists (had there been any) would have assumed that Does Cortical Mapping Change This View?
maleshadageneticproclivityforknitting.Theknitting
gene wouldhave beenassumedtoresideontheYchro- But hasn’t all the above been shown to be incorrect by
mosome. But by the 19th century, evolutionary psy- recentmappingstudies(Thompsonetal.2001)ofcortical
chologistswouldhaveclaimedthatwomenhadthatge- structuresinthebrainsofmonozygousoridentical(MZ)
netic proclivity, with the knitting gene on the X and dizygousorfraternal(DZ)twins?Thishasbeenthe
chromosome. With historical perspective, we can see interpretationofthosestudiesbythepopularpress(Mo-
that the change was purely culture-driven, not due toa tluk 2001).Thompsonandhiscolleaguescomputeddif-
geneticchange.Asitdidwithknitting,theenvironment, ferences in the quantity of gray matter of MZ and DZ
especiallytheculturalenvironment,seemstodoagood twins and unrelated individuals for various regions of
job of fine-tuning our behavior. A major challenge for thecortex.Notonlydidtheyclaimtohavedemonstrated
sciencetodayistoelucidatehowthatfine-tuningoccurs. that genetic factors significantly influence a number of
structuralregionsofthebrainbuttheyarguedthattheir
gene maps revealed how genes determine individual
brain differences. These are indeed strong claims.
Would Selection Generally Favor Genetic
Thompsonetal.’sdataanalysisofthebrainmapssuf-
Control of Behavior?
fers from many of the defects mentioned above as per-
meating the behavioral genetic literature. Environmen-
Wouldwebebetteroffifwehadmorethanenoughgenes tal contributions are ignored: “Because non-genetic
toplayacontrollingroleineveryoneofourchoicesand familial effects contribute to the resemblance between
actions and those genes could operate independently? relatives, such effects were accommodated, if not en-
Probably not. One could imagine a Hobbesian battle in tirely eliminated, by assuming the same common en-
whichgeneswouldcompetewitheachothertoimprove vironmentalvarianceforMZandDZpairs”(Thompson
theperformanceofthereproducingindividualsthatpos- et al. 2001: 1257). Then, a squared correlation greater
sessedthem—genesforcautionbeingfavoredinoneen- than 0.8 having been claimed for volumes of cortical
vironment one day and genes for impulsiveness in an- structures between MZ twins, the squared correlation
otherenvironmentthenext(“Lookbeforeyouleap,”“He for DZ twins in the same areas varied from 0.6 to 0.89.
whohesitatesislost”).Itisdifficulttoimaginehowany ThereisanobviousdifficultyhereinthatsuchlargeDZ
organism could make the grade evolutionarily if its be- correlations suggest important environmental contri-
94 F current anthropology Volume44, Number1, February2003
butions.4TherelationshipbetweentheMZandDZcor- turalidentitiestothedebatesbetweenmaterialandcul-
relationscertainlydoesnotsuggest“stronggeneticcon- turaldeterminismdescribedbySahlins(e.g.,1976),were
trolofbrainstructure”(p.1254)oreven“tightcoupling proudlynonquantitative.Recentdiscussionsontheide-
of brain structure and genetics” (p. 1256) as claimed. ational or symbolic nature of the subjects of cultural
Strangely, in the analysis by Thompson et al. of the evolution(e.g.,Durham1991),whilecriticalofattempts
relationship between cortical gray matter and Spear- to construct dynamical models of cultural evolution
man’sg,aquantityoftenusedasameasureofcognitive based onindividual-to-individualculturaltransmission,
performance(fromsubtestsoftheWechslerAdultIntel- nevertheless acknowledge the centrality of cultural ev-
ligence Scale), unrelated individuals were omitted and olution to human behavioral analysis. Thus, although
only the 40 twins receivedthe cognitivetests.Ahighly the quantitative paradigms used in behavioral genetics
significant relationship between frontal matter volume donotinformevolutionaryanalysis,thisdoesnotmean
and g is reported. It is not clear how the correlations that we cannot or should not take an evolutionary ap-
between the twins were controlled here, especially in proachtotheunderstandingandmodificationofhuman
lightoftheveryhighDZcorrelationsinthebrainimages behavior.Geneticallyevolvedfeaturessuchasthedom-
and the well-established effects of gender on some con- inanceofourvisualsenseshouldalwaysbekeptinmind,
tributorstog,especiallythegreatervariabilityofmales’ but an evolutionary approach to changing behavior in
scores than females’ (e.g., Jensen 1998:537). In sum, our species must primarily focus on culturalevolution.
Thompson et al. cannot be regarded as having demon- In the last 40,000 years or so, the scale of that cultural
strated a gene-brain relationship, nor do their “genetic evolution has produced a volume of information that
brain maps” contribute to our understanding of how dwarfs whatiscodedintoourgenes.Justconsiderwhat
genes influence cognition. is now stored in human memories, libraries, photo-
graphs, films, video tapes, the Worldwide Web, blue-
prints,andcomputerdatabanks—inadditiontowhatis
inherent in other artifacts and human-made structures.
Conclusions
Althoughtherehavebeenpreliminaryinvestigationsby
Cavalli-Sforza and Feldman (1981) and Boyd and Rich-
WhattherecentevidencefromtheHumanGenomePro-
erson (1985),scientists havebarelybeguntoinvestigate
ject tells us is that the interaction between genes, be-
the basic processes by which that body of information
tween the separate components of genes, and between
changes (or remains constant for long periods)—a task
controllingelementsoftheseseparatecomponentsmust
thatsocialscientistshavebeentakinguppiecemealand
be much more complex than we ever realized. Simple
largely qualitatively for a very long time (e.g., Bischof
additivemodelsofgeneactionoroftherelationshipbe-
1978; Cronk 1999; Durham 1991; Ehrlich 2002; Jacobs
tween genes and environments must be revised. They
and Campbell1961;JohnsonandEarle2000;Kotlerand
have formed the basis for our interpretation of pheno-
Zaltman1971;Murdock1956;PiragesandEhrlich1974;
type-genotyperelationshipsfor84years,eversinceR.A.
Rogers 1995; Stark 1996, 1999). Developing a unified
Fisher’sfamouspaper(1918)thatforthefirsttimerelated quantitativetheoryofculturalchangeisoneofthegreat
Mendelian genes to measurable phenotypes. Newmod- challengesforevolutionaryandsocialscienceinthe21st
els and paradigms are needed to go from the genome to century.
the phenome in any quantitative way. The simplistic Identifying the basic mechanisms by which our cul-
approach of behavioral genetics cannot do the job. We ture evolves will be difficult; the most recent attempts
must dig deeper into the environmental and especially using a “meme” approach (Blackmore 1999, Dawkins
culturalfactorsthatcontributetothephenome.Theas- 1989 [1976]) appear to be a dead end. Learning how to
cendancyofmolecularbiologyhas,unintentionally,mil- influencethatevolutionislikelytobemoredifficultstill
itated against progress in studies of cultural evolution. and fraught with pitfalls. No sensible geneticist envi-
Theories of culture andits evolutioninthe20thcen- sions a eugenic future in which people are selected to
tury, from Boas’s insistence on the particularity of cul- showcertainbehavioraltraits,andmostthinkingpeople
areawareoftheethical(ifnottechnicalandsocial)prob-
4.Referrington.2,ifindividuals1and2areMZtwinsandifthey lems of trying to change our behavior by altering our
are raised in identical environments (so that E p E), thentheir
1 2 genetic endowments. Society has long been mucking
correlationshouldbe1.If,ontheotherhand,theyareDZtwins,
aroundinculturalevolution,despitewarningsofthepo-
all the genetic contributions to the quantity of gray matter are
additive,andE andE arenotcorrelated,theircorrelationshould tential abuses of doing so (e.g., Huxley 1932). Nazi eu-
1 2
be0.5(whichwouldmakethesquaredcorrelation0.25).Dominance genic policies and Soviet, Cambodian, Chinese, and
effects will increase this, but in most cases the increase will not other social engineering experiments stand as monu-
be sufficient to bring the DZ correlation into the ballpark ofthe
ments to the ethical dangers that must be guarded
MZcorrelation.FortheseDZtwins,fromthesimplemodelofnn.
1and2theonlyotherpossibleexplanationforthehighDZtwin againstwhentryingsystematicallytoaltereithergenetic
correlation would be a correlation between environments E and or cultural evolution.
1
E2.ThestudyincludedfivepairseachoffemaleandmaleMZand Nevertheless,wearetodayallinvolvedincarryingout
DZ twins. Despite acknowledging that gender affectsvolumesof
or (with our taxes) supporting experiments designed to
brainstructures,imagesforthesexeswerepooledwithineachzyg-
change behavior. This is attested to by the advertising
osity. It is very difficult to invoke statistical contributions from
genestoexplaintheapparentlycloseMZandDZvaluesreported. business, Head Start programs, and the existence of in-
ehrlich and feldman Genes and Cultures F 95
stitutions such as Sing Sing Prison and Stanford Uni- consensus that phenomena such as exaptation (Gould
versity. The data used by evolutionary psychologiststo 2002), accident as an agent of direction of change itself
infer the biological antecedents of human behavior, ratherthanonlyasourceofvariation(Kimura1983),and
while not telling us anything about genetic evolution, moleculardrive(Dover2000)areallcausalmechanisms
may actually be helpful in improving our grasp of cul- of evolutionarynovelties (Goodwin 2002).
turalevolution.Whatseemscleartoday,however,isthat Assigning adaptive significance to an organ orbehav-
evolutionarypsychologyandbehavioralgeneticsarepro- iour pattern presumesthata problemexiststowhichit
motingavastoveremphasisonthepartplayedbygenetic isasolution(Dubrovsky2002).However,organismsnot
factors(andaseriousunderestimationoftheroleofcul- onlysolveproblemsintheenvironmentbutcreatethem.
tural evolution) in shaping our behavioral phenomes. AsWaddington(1976:18)hasputit,“Asurprisinglylarge
amountoftheenvironmentwhichaffectsnaturalselec-
tionoutcomesonanimalsisthemoreorlessdirectresult
of the animal’s own behavior.” Consideringthesefacts,
Comments
Lewontin (2000) has suggested that a more faithful de-
scription of the organism-environment interaction is
“construction” rather than “adaptation.” In human ev-
bernardo dubrovsky olution, the usual relationship between organism and
McGill University,3445Drummond St.,701, environment has become virtually reversed in adapta-
Montreal, PQ, Canada H3G1X9([email protected]. tion. Cultural invention has replaced genetic change as
mcgill.ca). 5 x 02
the effective source of variation. Consciousness allows
peopletoanalyzeandmakedeliberatealterationsassit-
Ehrlich and Feldman advance serious and valid criti-
uations require, with the result that adaptation of en-
cismsofthemethodsusedbyevolutionarypsychologists
vironmenttoorganismhasbecomethedominantmode
andbehavioralgeneticistsandidentifyfactualerrorsfre-
(Dubrovsky 2002).
quently made by them. While in conceptual agreement
OskarKempthorne(1978)hascriticizedtheexclusive
withthem,Iproposeheretolookatsomeotheraspects
use of observational data in the debate about inherited
of the problem.
and environmental factors contributing to intelligence.
Cosmides and Tooby’s new version of evolutionary
Observational data are used by behavioural geneticists
psychology(1987,1995;ToobyandCosmides2000)com-
(e.g.,Plominetal.1997)notwithstandingtherecognition
bines teleological, adaptationist, and rigid formalist in-
since the beginning of modern science that only exper-
terpretationsofbiologicalevolutionwiththeviewofthe
imentation can test the validity of rival causal hypoth-
mindasasortofcomputerprogramorinformationpro-
eses(Bunge1967).KempthorneandlaterJacquard(1983)
cessor. The adaptationist program considers every evo-
have criticized the use of analysis of variance of one
lutionarynoveltyasafeaturethatfavourssurvivaland/
feature in a population to check for causality, arguing
orreproduction(Gould2002,KirmayerandYoung1999).
that variance only measures dispersion of data around
Adaptationistsregardeachaspectoftheorganism’smor-
thevalueofthemean;itisameasureofdiversity,some-
phology, physiology, and behaviour as a specific adap-
times inappropriately referred to as “variability.” What
tation of the entire organism. For them the problem of
is important in considerations of causality, however, is
evolutionary science is finding out what an adaptation
the magnitude of an effect which can be attributed to
isfor,wheninfactthefirstquestionshouldbewhether
variation or change in one or more independent varia-
it exists (Fodor 2000; Mahner and Bunge 1997:423).
bles. Variance cannot point to any causal factor. The
It is extremely difficult to trace traits back in time,
parallel with statistical correlation is clear. The latter
andanyhypothesisregardingthehistoryofatraitmust
measuresthedegreeofassociationoftwovariables(e.g.,
bebasedonprobability(Northcutt1999).Fewtraitshave
been examined in sufficient detail in enough species in sizeandpopulationofacountry)neitherofwhichcauses
different radiations to allow a meaningful evaluationof the other but variation in one of which can induce
them. The problem of identifying traits is compounded changesintheother.Ithasneverbeendemonstratedthat
by the frequency with which some psychologists and IQs(Jacquard1983)aredeterminedbythegenome,since
psychiatristsarbitrarilyqualifytheconditionofthestate causal relations are valid only for events and not for
or trait for various phenotype components (Paris 1998). attributes.
Questionable concepts such as “brain design in re- Moreover, linear additive models in the absence of a
sponsetoenvironmentaldemands”(KirmayerandYoung theory of interaction are invalid (Lewontin 2000). That
1999) and concepts difficult to verify such as “adaptive both genes and environment produce a given outcome
evolutionoftraits”(Buss1999,Paris1998)lackheuristic isatruism,butweareseriouslymistakenwhenwepre-
value, and there is no evidence whatsoever for an in- sumethatwecanbestexpressthisprinciplebyassigning
structive component in the appearance of evolutionary percentages and stating, for example, that behaviour A
novelties (Dover 2000, Gould 2002, Lewontin 2000). is 40% genetic and 60% environmental. Such reduc-
Adaptationists fail to recognize other factors besides tionist expressions go beyond simple mistakes to enter
natural selection as causally associated with evolution. the domain of the meaningless. Genetics and environ-
Whileitisnotacriterionoftruth,thereisameasureof ment do interact to build a totality, but we need to un-
96 F current anthropology Volume44, Number1, February2003
derstandwhytheresultingwholesareirreducibletosep- It is made possible by innate machinery designed to
arate components. do the learning. The claim that there are several in-
nate modules is a claim that there are several innate
learning machines, each of which learns according
edward hagen to a particular logic.
Institute for Theoretical Biology, Humboldt
Further, the New Guineans’ opinion that “the fear in
University, Berlin, Germany ([email protected].
Europeans [is]a resultoftheirstupidityinbeingunable
hu-berlin.de) 6 ix 02
to distinguish which snakes might be dangerous” sup-
ports rather than refutes Pinker (1997:388):
Because Ehrlich and Feldman fail to provide them, I
sketch evolutionary psychology’s basics here. The world is a dangerous place, but our ancestors
It has long been recognized (e.g., Galen, Paley) that could not have spent their lives cowering in caves;
organisms consist of functional mechanisms—hearts, there was food to gather and mates to win. They had
lungs,livers,bones,intestines,prostates,uteruses,etc.—
to calibrate their fears of typical dangers against ac-
but before 1859 their origin was unknown. Darwinand
tual dangers in the local environment (after all, not
Wallace proposed that these mechanisms—termed ad-
all spiders are poisonous) and against their own abil-
aptations—evolved by natural selection and, thus, nec-
ity to neutralize the danger: their know-how [etc.].
essarily were designed to promote reproduction.
. . . Between the ages of three and five, children be-
Psychologists have demonstrated that cognitive pro-
come fearful of all the standard phobic objects
cesses, like the body’s other mechanisms, have func-
—spiders, the dark, deep water, and so on—and then
tionalstructure.Evolutionarypsychologistsproposethat
master them one by one. Most adult phobias are
this structure evolved by natural selection to serve re-
childhood fears that never went away. That is why
production.Giventhatthebrainmechanismsunderpin-
it is city-dwellers who most fear snakes.
ning vision, hearing, motor control,pain,memory,etc.,
have obvious reproductive utility, this proposition is Finally,Minekaandcolleagues’research,whichEhrlich
compelling. Further, these examples suggest that the and Feldman apparently believe undermines evolution-
brainismadeupofmanyfunctionallyspecializedparts. arypsychologicalhypothesesofspecializedfearlearning,
Anadaptationisrarelydiscoveredordescribedbyiden- actually strongly supports them. Cook and Mineka
tifying the specific genes that directed its construction showed that lab-raised monkeys readily learned to fear
or by documenting heritable variation and differential toysnakesbutnottoyrabbitsorflowers,suggestingthat
reproduction in ancestral populations. Rather, adapta- there is an innate predisposition to learn fears of evo-
tions are recognized by the close functionalfitbetween lutionarily salient dangers, such as snakes. O¨hman and
an adaptive problem and some aspect of phenotypic Mineka(2001)synthesize30yearsofresearchonfearin
structure; it is their evidence of design, not genes, that humans and other primates in an article subtitled “To-
assures us that hearts,lungs,andliversareadaptations. ward an Evolved Module of Fear and Fear Learning.”
Although Ehrlich and Feldman believe that the envi- Against hypothesized sex differences in mating psy-
ronment of evolutionary adaptedness refers to a fixed chology, Ehrlich and Feldman claim that “it would be
time or place, it actually refers to the recurring aspects nosmalldevelopmentaltrickgeneticallytoprogramde-
of the environment that were necessary for the evolu- tailed,different,andindependentreproductivestrategies
tion,development,functioningofaparticularadaptation into modules in male and female brains.” Well, natural
(Tooby and Cosmides 1990a). The environment of evo- selection somehow “programmed” uteruses in females
lutionaryadaptednessofthelung,forexample,includes but not males. Evolutionary psychology argues that
an oxygen atmosphere. Recurring aspects of ancestral men’s and women’s brains, like the restoftheirbodies,
environments that had an impact on reproduction in- are probably identical in most ways but profoundlydif-
clude interactions with the opposite sex, children, par-
ferent in some.
ents, kin, nonkin, plants, animals, predators, and prey
Whereasbehavioralgeneticsfocusesonindividualdif-
and the need to avoid toxins, pathogens, and injuries.
ferences,evolutionarypsychologyfocusesalmostexclu-
Much evolutionary psychological research has been
sively on human universals (age and sex excepted). Ad-
based on the certainty that in the environment of evo-
aptationsaregroundedinthevastmajorityofgenesthat
lutionaryadaptednesswomengotpregnantandmendid
are identical (or nearly identical) in all humans. Impor-
not.
tant individualdifferences arisenotfromminorgenetic
To say “There is reason to believe that fear of snakes
differences but “from exposing the same human nature
in other primates is largely learned”impliesthatlearn-
todifferentenvironmentalinputs”(ToobyandCosmides
ing and psychological adaptations are opposing hypoth-
1990b:23).
eses. But, as Pinker (1997:33) and virtually every other
Accused of spinning “just-so stories,” evolutionary
evolutionary psychologist repeatedly emphasize:
psychologists have, in fact, tested their hypotheses in
Yes, every part of human intelligence involves cul- hundredsofstudieswithmanythousandsofsubjectsin
ture and learning. But learning is not a surrounding scores of different cultures and have published their re-
gas or force field, and it does not happen by magic. sults in the world’s top science journals. Empirical re-