Stanford Encyclopedia of Philosophy

EvolutionaryGameTheory
FirstpublishedMonJan14,2002substantiverevisionSunJul 19,2009
Evolutionary game theory originated as an application of the mathematical theory of games to
biological contexts, arising from the realization that frequency dependent fitness introduces a strategic
aspect to evolution. Recently, however, evolutionary game theory has become of increased interest to
economists,sociologists, and anthropologists--and social scientists in general--as well as philosophers.
The interest among social scientists in a theory with explicit biological roots derives from threefacts.
First,theevolutiontreatedbyevolutionarygametheoryneednotbebiologicalevolution. Evolution
may,inthiscontext,oftenbeunderstoodasculturalevolution,wherethisrefersto changesinbeliefs
andnormsovertime.Second,therationalityassumptionsunderlyingevolutionary gametheoryare,in
manycases,moreappropriateforthemodellingofsocialsystemsthanthose assumptionsunderlying
thetraditionaltheoryofgames.Third,evolutionarygametheory,asan explicitlydynamictheory,
providesanimportantelementmissingfromthetraditionaltheory.Inthe prefacetoEvolutionandthe
TheoryofGames,MaynardSmithnotesthat[p]aradoxically,ithas turnedoutthatgametheoryis
morereadilyappliedtobiologythantothefieldofeconomicbehaviour forwhichitwasoriginally
designed.Itisperhapsdoublyparadoxical,then,thatthesubsequent developmentofevolutionary
gametheoryhasproducedatheorywhichholdsgreatpromiseforsocial scientists,andisasreadily
appliedtothefieldofeconomicbehaviourasthatforwhichitwas originallydesigned.

1. HistoricalDevelopment
2. TwoApproachestoEvolutionaryGameTheory
2.1Definitionsofevolutionarystability
2.2Specifyingdynamicsforthepopulation
3. WhyEvolutionaryGameTheory?
3.1Theequilibriumselectionproblem
3.2Theproblemofhyperrationalagents
3.3Thelackofadynamicaltheoryinthetraditionaltheoryofgames
4. ApplicationsofEvolutionaryGameTheory
4.1Asenseoffairness
4.2Theemergenceoflanguage.
5. PhilosophicalProblemsofEvolutionaryGameTheory
5.1Themeaningoffitnessinculturalevolutionaryinterpretations
5.2Theexplanatoryirrelevanceofevolutionarygametheory
5.3Thevalueladennessofevolutionarygametheoreticexplanations
Bibliography
AcademicTools
OtherInternetResources
RelatedEntries

1. HistoricalDevelopment

EvolutionarygametheorywasfirstdevelopedbyR.A.Fisher[seeTheGeneticTheoryofNatural
Selection(1930)]inhisattempttoexplaintheapproximateequalityofthesexratioinmammals.The
puzzleFisherfacedwasthis:whyisitthatthesexratioisapproximatelyequalinmanyspecieswhere
themajorityofmalesnevermate?Inthesespecies,thenonmatingmaleswouldseemtobeexcess
baggagecarriedaroundbytherestofthepopulation,havingnorealuse.Fisherrealizedthatifwe
measureindividualfitnessintermsoftheexpectednumberofgrandchildren,thenindividualfitness
dependsonthedistributionofmalesandfemalesinthepopulation.Whenthereisagreaternumberof
femalesinthepopulation,maleshaveahigherindividualfitnesswhentherearemoremalesinthe
population,femaleshaveahigherindividualfitness.Fisherpointedoutthat,insuchasituation,the
evolutionarydynamicsleadtothesexratiobecomingfixedatequalnumbersofmalesandfemales.
Thefactthatindividualfitnessdependsupontherelativefrequencyofmalesandfemalesinthe
populationintroducesastrategicelementintoevolutions.
Fisher'sargumentcanbeunderstoodgametheoretically,buthedidnotstateitinthoseterms.In1961,
R.C.Lewontinmadethefirstexplicitapplicationofgametheorytoevolutionarybiologyin
EvolutionandtheTheoryofGames(nottobeconfusedwiththeMaynardSmithworkofthesame
name).In1972,MaynardSmithdefinedtheconceptofanevolutionarilystablestrategy(hereafter
ESS)inthearticleGameTheoryandtheEvolutionofFighting.However,itwasthepublicationof
TheLogicofAnimalConflict,byMaynardSmithandPricein1973thatintroducedtheconceptof
anESSintowidespreadcirculation.In1982,MaynardSmith'sseminaltextEvolutionandtheTheory
ofGamesappeared,followedshortlythereafterbyRobertAxelrod'sfamousworkTheEvolutionof
Cooperationin1984.Sincethen,therehasbeenaveritableexplosionofinterestbyeconomistsand
socialscientistsinevolutionarygametheory(seethebibliographybelow).

2.TwoApproachestoEvolutionaryGameTheory
Therearetwoapproachestoevolutionarygametheory.Thefirstapproachderivesfromtheworkof
MaynardSmithandPriceandemploystheconceptofanevolutionarilystablestrategyastheprincipal
toolofanalysis.Thesecondapproachconstructsanexplicitmodeloftheprocessbywhichthe
frequencyofstrategieschangeinthepopulationandstudiespropertiesoftheevolutionarydynamics
withinthatmodel.
Thefirstapproachcanthusbethoughtofasprovidingastaticconceptualanalysisofevolutionary
stability.Staticbecause,althoughdefinitionsofevolutionarystabilityaregiven,thedefinitions
advanceddonottypicallyrefertotheunderlyingprocessbywhichbehaviours(orstrategies)change
inthepopulation.Thesecondapproach,incontrast,doesnotattempttodefineanotionof
evolutionarystability:onceamodelofthepopulationdynamicshasbeenspecified,allofthestandard
stabilityconceptsusedintheanalysisofdynamicalsystemscanbebroughttobear.

2.1Definitionsofevolutionarystability
Asanexampleofthefirstapproach,considertheproblemoftheHawkDovegame,analyzedby
MaynardSmithandPriceinTheLogicofAnimalConflict.Inthisgame,twoindividualscompete
foraresourceofafixedvalueV.(Inbiologicalcontexts,thevalueVoftheresourcecorrespondstoan
increaseintheDarwinianfitnessoftheindividualwhoobtainstheresourceinaculturalcontext,the
valueVoftheresourcewouldneedtobegivenanalternateinterpretationmoreappropriatetothe
specificmodelathand.)Eachindividualfollowsexactlyoneoftwostrategiesdescribedbelow:
Hawk Initiateaggressivebehaviour,notstoppinguntilinjuredoruntilone'sopponent
backsdown.
Dove Retreatimmediatelyifone'sopponentinitiatesaggressivebehaviour.
Ifweassumethat(1)whenevertwoindividualsbothinitiateaggressivebehaviour,conflicteventually

resultsandthetwoindividualsareequallylikelytobeinjured,(2)thecostoftheconflictreduces
individualfitnessbysomeconstantvalueC,(3)whenaHawkmeetsaDove,theDoveimmediately
retreatsandtheHawkobtainstheresource,and(4)whentwoDovesmeettheresourceisshared
equallybetweenthem,thefitnesspayoffsfortheHawkDovegamecanbesummarizedaccordingto
thefollowingmatrix:
Hawk Dove
Hawk (VC) V
Dove
0
V/2
Figure1:TheHawkDoveGame
(Thepayoffslistedinthematrixareforthatofaplayerusingthestrategyintheappropriaterow,
playingagainstsomeoneusingthestrategyintheappropriatecolumn.Forexample,ifyouplaythe
strategyHawkagainstanopponentwhoplaysthestrategyDove,yourpayoffisVifyouplaythe
strategyDoveagainstanopponentwhoplaysthestrategyHawk,yourpayoffis0.)
Inorderforastrategytobeevolutionarilystable,itmusthavethepropertythatifalmostevery
memberofthepopulationfollowsit,nomutant(thatis,anindividualwhoadoptsanovelstrategy)
cansuccessfullyinvade.Thisideacanbegivenaprecisecharacterizationasfollows:LetF(s1,s2)
denotethechangeinfitnessforanindividualfollowingstrategys1againstanopponentfollowing
strategys2,andletF(s)denotethetotalfitnessofanindividualfollowingstrategysfurthermore,
supposethateachindividualinthepopulationhasaninitialfitnessofF0.Ifisanevolutionarily
stablestrategyandamutantattemptingtoinvadethepopulation,then
F()=F0+(1p)F(,)+pF(,)
F()=F0+(1p)F(,)+pF(,)
wherepistheproportionofthepopulationfollowingthemutantstrategy.
Sinceisevolutionarilystable,thefitnessofanindividualfollowingmustbegreaterthanthe
fitnessofanindividualfollowing(otherwisethemutantfollowingwouldbeabletoinvade),and
soF()>F().Now,aspisverycloseto0,thisrequiresthateitherthat
F(,)>F(,)
orthat
F(,)=F(,)andF(,)>F(,)
(ThisisthedefinitionofanESSthatMaynardSmithandPricegive.)Inotherwords,whatthismeans
isthatastrategyisanESSifoneoftwoconditionsholds:(1)doesbetterplayingagainstthan
anymutantdoesplayingagainst,or(2)somemutantdoesjustaswellplayingagainstas,but
doesbetterplayingagainstthemutantthanthemutantdoes.
Giventhischaracterizationofanevolutionarilystablestrategy,onecanreadilyconfirmthat,forthe
HawkDovegame,thestrategyDoveisnotevolutionarilystablebecauseapurepopulationofDoves
canbeinvadedbyaHawkmutant.IfthevalueVoftheresourceisgreaterthanthecostCofinjury
(sothatitisworthriskinginjuryinordertoobtaintheresource),thenthestrategyHawkis
evolutionarilystable.Inthecasewherethevalueoftheresourceislessthanthecostofinjury,thereis
noevolutionarilystablestrategyifindividualsarerestrictedtofollowingpurestrategies,although
thereisanevolutionarilystablestrategyifplayersmayusemixedstrategies.[1]

IntheyearsfollowingtheoriginalworkofMaynardSmithandPrice,alternateanalyticsolution
conceptshavebeenproposed.Ofthese,twoimportantonesaretheideaofanevolutionarilystableset
(seeThomas1984,1985a,b),andtheideaofalimitESS(seeSelten1983,1988).Theformer
providesasetwisegeneralizationoftheconceptofanevolutionarilystablestrategy,andthelatter
extendstheconceptofanevolutionarilystablestrategytothecontextoftwoplayerextensiveform
games.

2.2Specifyingdynamicsforthepopulation
Asanexampleofthesecondapproach,considerthewellknownPrisoner'sDilemma.Inthisgame,
individualschooseoneoftwostrategies,typicallycalledCooperateandDefect.Hereisthe
generalformofthepayoffmatrixfortheprisoner'sdilemma:
Cooperate Defect
Cooperate (R,R)
(S,T)
Defect
(T,S)
(P,P)
Figure2:PayoffMatrixforthePrisoner'sDilemma.
Payoffslistedas(row,column).
whereT>R>P>SandT>R>P>S.(Thisformdoesnotrequirethatthepayoffsforeachplayer
besymmetric,onlythattheproperorderingofthepayoffsobtains.)Inwhatfollows,itwillbe
assumedthatthepayoffsforthePrisoner'sDilemmaarethesameforeveryoneinthepopulation.
HowwillapopulationofindividualsthatrepeatedlyplaysthePrisoner'sDilemmaevolve?Wecannot
answerthatquestionwithoutintroducingafewassumptionsconcerningthenatureofthepopulation.
First,letusassumethatthepopulationisquitelarge.Inthiscase,wecanrepresentthestateofthe
populationbysimplykeepingtrackofwhatproportionfollowthestrategiesCooperateandDefect.
Letpcandpddenotetheseproportions.Furthermore,letusdenotetheaveragefitnessofcooperators
anddefectorsbyWCandWD,respectively,andletWdenotetheaveragefitnessoftheentire
population.ThevaluesofWC,WD,andWcanbeexpressedintermsofthepopulationproportionsand
payoffvaluesasfollows:
WC=F0+pcF(C,C)+pdF(C,D)
WD=F0+pcF(D,C)+pdF(D,D)
W=pcWC+pdWD
Second,letusassumethattheproportionofthepopulationfollowingthestrategiesCooperateand
Defectinthenextgenerationisrelatedtotheproportionofthepopulationfollowingthestrategies
CooperateandDefectinthecurrentgenerationaccordingtotherule:

Wecanrewritetheseexpressionsinthefollowingform:

Ifweassumethatthechangeinthestrategyfrequencyfromonegenerationtothenextaresmall,these
differenceequationsmaybeapproximatedbythedifferentialequations:

TheseequationswereofferedbyTaylorandJonker(1978)andZeeman(1979)toprovidecontinuous
dynamicsforevolutionarygametheoryandareknownasthereplicatordynamics.
ThereplicatordynamicsmaybeusedtomodelapopulationofindividualsplayingthePrisoner's
Dilemma.ForthePrisoner'sDilemma,theexpectedfitnessofCooperatingandDefectingare:
WC =F0+pcF(C,C)+pdF(C,D)
=F0+pcR+pdS
and
WD =F0+pcF(D,C)+pdF(D,D)

=F0+pcT+pdP.

SinceT>RandP>S,itfollowsthatWD>WCandhenceWD>W>WC.Thismeansthat

and

SincethestrategyfrequenciesforDefectandCooperateinthenextgenerationaregivenby

and

respectively,weseethatovertimetheproportionofthepopulationchoosingthestrategyCooperate
eventuallybecomesextinct.Figure3illustratesonewayofrepresentingthereplicatordynamical
modeloftheprisoner'sdilemma,knownasastatespacediagram.

Figure3:TheReplicatorDynamicalModelofthePrisoner'sDilemma
Weinterpretthisdiagramasfollows:theleftmostpointrepresentsthestateofthepopulationwhere
everyonedefects,therightmostpointrepresentsthestatewhereeveryonecooperates,andintermediate
pointsrepresentstateswheresomeproportionofthepopulationdefectsandtheremaindercooperates.
(OnemapsstatesofthepopulationontopointsinthediagrambymappingthestatewhenN%ofthe
populationdefectsontothepointofthelineN%ofthewaytotheleftmostpoint.)Arrowsontheline

representtheevolutionarytrajectoryfollowedbythepopulationovertime.Theopencircleatthe
rightmostpointindicatesthatthestatewhereeverybodycooperatesisanunstableequilibrium,inthe
sensethatifasmallportionofthepopulationdeviatesfromthestrategyCooperate,thenthe
evolutionarydynamicswilldrivethepopulationawayfromthatequilibrium.Thesolidcircleatthe
leftmostpointindicatesthatthestatewhereeverybodyDefectsisastableequilibrium,inthesense
thatifasmallportionofthepopulationdeviatesfromthestrategyDefect,thentheevolutionary
dynamicswilldrivethepopulationbacktotheoriginalequilibriumstate.
Atthispoint,onemayseelittledifferencebetweenthetwoapproachestoevolutionarygametheory.
Onecanconfirmthat,forthePrisoner'sDilemma,thestatewhereeverybodydefectsistheonlyESS.
Sincethisstateistheonlystableequilibriumunderthereplicatordynamics,thetwonotionsfit
togetherquiteneatly:theonlystableequilibriumunderthereplicatordynamicsoccurswheneveryone
inthepopulationfollowstheonlyESS.Ingeneral,though,therelationshipbetweenESSsandstable
statesofthereplicatordynamicsismorecomplexthanthisexamplesuggests.TaylorandJonker
(1978),aswellasZeeman(1979),establishconditionsunderwhichonemayinfertheexistenceofa
stablestateunderthereplicatordynamicsgivenanevolutionarilystablestrategy.Roughly,ifonlytwo
purestrategiesexist,thengivena(possiblymixed)evolutionarilystablestrategy,thecorresponding
stateofthepopulationisastablestateunderthereplicatordynamics.(Iftheevolutionarilystable
strategyisamixedstrategyS,thecorrespondingstateofthepopulationisthestateinwhichthe
proportionofthepopulationfollowingthefirststrategyequalstheprobabilityassignedtothefirst
strategybyS,andtheremainderfollowthesecondstrategy.)However,thiscanfailtobetrueifmore
thantwopurestrategiesexist.
TheconnectionbetweenESSsandstablestatesunderanevolutionarydynamicalmodelisweakened
furtherifwedonotmodelthedynamicsbythereplicatordynamics.Forexample,supposeweusea
localinteractionmodelinwhicheachindividualplaystheprisoner'sdilemmawithhisorher
neighbors.NowakandMay(1992,1993),usingaspatialmodelinwhichlocalinteractionsoccur
betweenindividualsoccupyingneighboringnodesonasquarelattice,showthatstablepopulation
statesfortheprisoner'sdilemmadependuponthespecificformofthepayoffmatrix.[2]
WhenthepayoffmatrixforthepopulationhasthevaluesT=2.8,R=1.1,P=0.1,andS=0,the
evolutionarydynamicsofthelocalinteractionmodelagreewiththoseofthereplicatordynamics,and
leadtoastatewhereeachindividualfollowsthestrategyDefectwhichis,asnotedbefore,theonly
evolutionarilystablestrategyintheprisoner'sdilemma.Thefigurebelowillustrateshowrapidlyone
suchpopulationconvergestoastatewhereeveryonedefects.

Generation1

Generation2

Generation3

Generation4

Generation5

Generation6

Figure4:Prisoner'sDilemma:AllDefect
[Viewamovieofthismodel]
However,whenthepayoffmatrixhasvaluesofT=1.2,R=1.1,P=0.1,andS=0,theevolutionary
dynamicscarrythepopulationtoastablecycleoscillatingbetweentwostates.Inthiscycle
cooperatorsanddefectorscoexist,withsomeregionscontainingblinkersoscillatingbetween
defectorsandcooperators(asseeningeneration19and20).

Generation1

Generation2

Generation19

Generation20

Figure5:Prisoner'sDilemma:Cooperate
[Viewamovieofthismodel]
Noticethatwiththeseparticularsettingsofpayoffvalues,theevolutionarydynamicsofthelocal
interactionmodeldiffersignificantlyfromthoseofthereplicatordynamics.Underthesepayoffs,the
stablestateshavenocorrespondinganalogueineitherthereplicatordynamicsnorintheanalysisof
evolutionarilystablestrategies.
AphenomenonofgreaterinterestoccurswhenwechoosepayoffvaluesofT=1.61,R=1.01,P=
0.01,andS=0.Here,thedynamicsoflocalinteractionleadtoaworldconstantlyinflux:underthese
valuesregionsoccupiedpredominantlybyCooperatorsmaybesuccessfullyinvadedbyDefectors,
andregionsoccupiedpredominantlybyDefectorsmaybesuccessfullyinvadedbyCooperators.In
thismodel,thereisnostablestrategyinthetraditionaldynamicalsense.[3]

Generation1

Generation3

Generation5

Generation7

Generation9

Generation11

Generation13

Generation15

Figure6:Prisoner'sDilemma:Chaotic
[viewamovieofthismodel]
Thesemodelsdemonstratethat,althoughnumerouscasesexistinwhichbothapproachesto
evolutionarygametheoryarriveatthesameconclusionregardingwhichstrategiesonewouldexpect
tofindpresentinapopulation,thereareenoughdifferencesintheoutcomesofthetwomodesof
analysistojustifythedevelopmentofeachprogram.

3. WhyEvolutionaryGameTheory?
Althoughevolutionarygametheoryhasprovidednumerousinsightstoparticularevolutionary
questions,agrowingnumberofsocialscientistshavebecomeinterestedinevolutionarygametheory
inhopesthatitwillprovidetoolsforaddressinganumberofdeficienciesinthetraditionaltheoryof
games,threeofwhicharediscussedbelow.

3.1Theequilibriumselectionproblem
TheconceptofaNashequilibrium(seetheentryongametheory)hasbeenthemostusedsolution
conceptingametheorysinceitsintroductionbyJohnNashin1950.Aselectionofstrategiesbya
groupofagentsissaidtobeinaNashequilibriumifeachagent'sstrategyisabestresponsetothe
strategieschosenbytheotherplayers.Bybestresponse,wemeanthatnoindividualcanimproveher
payoffbyswitchingstrategiesunlessatleastoneotherindividualswitchesstrategiesaswell.This
neednotmeanthatthepayoffstoeachindividualareoptimalinaNashequilibrium:indeed,oneof
thedisturbingfactsoftheprisoner'sdilemmaisthattheonlyNashequilbriumofthegamewhenboth
agentsdefectissuboptimal.[4]
YetadifficultyariseswiththeuseofNashequilibriumasasolutionconceptforgames:ifwerestrict
playerstousingpurestrategies,noteverygamehasaNashequilbrium.ThegameMatching
Penniesillustratesthisproblem.
Heads Tails
Heads (0,1) (1,0)
Tails (1,0) (0,1)
Figure7:PayoffmatrixforthegameofMatchingPennies
(Rowwinsifthetwocoinsdonotmatch,whereasColumnwinsifthetwocoinsmatch).
WhileitistruethateverynoncooperativegameinwhichplayersmayusemixedstrategieshasaNash
equilibrium,somehavequestionedthesignificanceofthisforrealagents.Ifitseemsappropriateto
requirerationalagentstoadoptonlypurestrategies(perhapsbecausethecostofimplementinga
mixedstrategyrunstoohigh),thenthegametheoristmustadmitthatcertaingameslacksolutions.
AmoresignificantproblemwithinvokingtheNashequilibriumastheappropriatesolutionconcept
arisesbecausegamesexistwhichhavemultipleNashequilibria(seethesectiononSolutionConcepts
andEquilibria,intheentryongametheory).WhenthereareseveraldifferentNashequilibria,howis
arationalagenttodecidewhichoftheseveralequilibriaistherightonetosettleupon?[5]Attempts
toresolvethisproblemhaveproducedanumberofpossiblerefinementstotheconceptofaNash
equilibrium,eachrefinementhavingsomeintuitivepurchase.Unfortunately,somanyrefinementsof
thenotionofaNashequilibriumhavebeendevelopedthat,inmanygameswhichhavemultipleNash
equilibria,eachequilibriumcouldbejustifiedbysomerefinementpresentintheliterature.The
problemhasthusshiftedfromchoosingamongmultipleNashequilibriatochoosingamongthe

variousrefinements.Some(seeSamuelson(1997),EvolutionaryGamesandEquilibriumSelection)
hopethatfurtherdevelopmentofevolutionarygametheorycanbeofserviceinaddressingthisissue.

3.2Theproblemofhyperrationalagents
Thetraditionaltheoryofgamesimposesaveryhighrationalityrequirementuponagents.This
requirementoriginatesinthedevelopmentofthetheoryofutilitywhichprovidesgametheory's
underpinnings(seeLuce(1950)foranintroduction).Forexample,inordertobeabletoassigna
cardinalutilityfunctiontoindividualagents,onetypicallyassumesthateachagenthasawelldefined,
consistentsetofpreferencesoverthesetoflotteriesovertheoutcomeswhichmayresultfrom
individualchoice.Sincethenumberofdifferentlotteriesoveroutcomesisuncountablyinfinite,this
requireseachagenttohaveawelldefined,consistentsetofuncountablyinfinitelymanypreferences.
Numerousresultsfromexperimentaleconomicshaveshownthatthesestrongrationalityassumptions
donotdescribethebehaviorofrealhumansubjects.Humansarerarely(ifever)thehyperrational
agentsdescribedbytraditionalgametheory.Forexample,itisnotuncommonforpeople,in
experimentalsituations,toindicatethattheypreferAtoB,BtoC,andCtoA.Thesefailuresofthe
transitivityofpreferencewouldnotoccurifpeoplehadawelldefinedconsistentsetofpreferences.
Furthermore,experimentswithaclassofgamesknownasabeautypageantshow,quite
dramatically,thefailureofcommonknowledgeassumptionstypicallyinvokedtosolvegames.[6]
Sinceevolutionarygametheorysuccessfullyexplainsthepredominanceofcertainbehaviorsof
insectsandanimals,wherestrongrationalityassumptionsclearlyfail,thissuggeststhatrationalityis
notascentraltogametheoreticanalysesaspreviouslythought.Thehope,then,isthatevolutionary
gametheorymaymeetwithgreatersuccessindescribingandpredictingthechoicesofhuman
subjects,sinceitisbetterequippedtohandletheappropriateweakerrationalityassumptions.

3.3Thelackofadynamicaltheoryinthetraditionaltheoryofgames
AttheendofthefirstchapterofTheoryofGamesandEconomicBehavior,vonNeumannand
Morgensternwrite:
Werepeatmostemphaticallythatourtheoryisthoroughlystatic.Adynamictheory
wouldunquestionablybemorecompleteandthereforepreferable.Butthereisample
evidencefromotherbranchesofsciencethatitisfutiletotrytobuildoneaslongasthe
staticsideisnotthoroughlyunderstood.(VonNeumannandMorgenstern,1953,p.44)
Thetheoryofevolutionisadynamicaltheory,andthesecondapproachtoevolutionarygametheory
sketchedaboveexplicitlymodelsthedynamicspresentininteractionsamongindividualsinthe
population.Sincethetraditionaltheoryofgameslacksanexplicittreatmentofthedynamicsof
rationaldeliberation,evolutionarygametheorycanbeseen,inpart,asfillinganimportantlacunaof
traditionalgametheory.
Onemayseektocapturesomeofthedynamicsofthedecisionmakingprocessintraditionalgame
theorybymodelingthegameinitsextensiveform,ratherthanitsnormalform.However,formost
gamesofreasonablecomplexity(andhenceinterest),theextensiveformofthegamequicklybecomes
unmanageable.Moreover,evenintheextensiveformofagame,traditionalgametheoryrepresentsan
individual'sstrategyasaspecificationofwhatchoicethatindividualwouldmakeateachinformation
setinthegame.Aselectionofstrategy,then,correspondstoaselection,priortogameplay,ofwhat
thatindividualwilldoatanypossiblestageofthegame.Thisrepresentationofstrategyselection
clearlypresupposeshyperrationalplayersandfailstorepresenttheprocessbywhichoneplayer
observeshisopponent'sbehavior,learnsfromtheseobservations,andmakesthebestmovein
responsetowhathehaslearned(asonemightexpect,forthereisnoneedtomodellearningin
hyperrationalindividuals).Theinabilitytomodelthedynamicalelementofgameplayintraditional
gametheory,andtheextenttowhichevolutionarygametheorynaturallyincorporatesdynamical

considerations,revealsanimportantvirtueofevolutionarygametheory.

4. ApplicationsofEvolutionaryGameTheory
Evolutionarygametheoryhasbeenusedtoexplainanumberofaspectsofhumanbehavior.Asmall
samplingoftopicswhichhavebeenanalysedfromtheevolutionaryperspectiveinclude:altruism
(FletcherandZwick,2007Gintisetal.,2003SanchezandCuesta,2005Trivers,1971),behaviorin
publicgoodsgame(ClemensandRiechmann,2006Hauert,2006Hauertetal.,2002,2006
HubermanandGlance,1995),empathy(PageandNowak,2002Fishman,2006),humanculture
(EnquistandGhirlanda,2007Enquistetal.,2008),moralbehaviour(Alexander,2007Boehm,
1982HarmsandSkyrms,2008Skyrms1996,2004),privateproperty(Gintis,2007),signaling
systemsandotherprotolinguisticbehaviour(Barrett,2007HauskenandHirshleirfer,2008Hurd,
1995Jager,2008Nowaketal.,1999Pawlowitsch,2007,2008Skyrms,forthcomingZollman,
2005),sociallearning(KamedaandNakanishi,2003Nakahashi,2007Rogers,1988Wakanoand
Aoki,2006Wakanoetal.,2004),andsocialnorms(Axelrod,1986Bicchieri,2006Binmoreand
Samuelson,1994Chalubetal.,2006Kendaletal.,2006Ostrum,2000).
Thefollowingsubsectionsprovideabriefillustrationoftheuseofevolutionarygametheoretic
modelstoexplaintwoareasofhumanbehavior.Thefirstconcernsthetendencyofpeopletoshare
equallyinperfectlysymmetricsituations.Thesecondshowshowpopulationsofprelinguistic
individualsmaycoordinateontheuseofasimplesignalingsystemeventhoughtheylacktheability
tocommunicate.Thesetwomodelshavebeenpointedtoaspreliminaryexplanationsofoursenseof
fairnessandlanguage,respectively.Theywereselectedforinclusionhereprimarilybecauseofthe
relativesimplicityofthemodelandapparentsuccessatexplainingthephenomenoninquestion.

4.1Asenseoffairness
Onenaturalgametouseforinvestigatingtheevolutionoffairnessisdividethecake(thisisthe
simplestversionoftheNashbargaininggame).Inchapter1ofEvolutionoftheSocialContract,
Skyrmspresentstheproblemasfollows:
Herewestartwithaverysimpleproblemwearetodivideachocolatecakebetweenus.
Neitherofushasanyspecialclaimasagainsttheother.Outpositionsareentirely
symmetric.Thecakeisawindfallforus,anditisuptoustodivideit.Butifwecannot
agreehowtoshareit,thecakewillspoilandwewillgetnothing.(Skyrms,1996,pp.34)
Moreformally,supposethattwoindividualsarepresentedwitharesourceofsizeCbyathirdparty.
Astrategyforaplayer,inthisgame,consistsofanamountofcakethathewouldlike.Thesetof
possiblestrategiesforaplayeristhusanyamountbetween0andC.Ifthesumofstrategiesforeach
playerislessthanorequaltoC,eachplayerreceivestheamountheaskedfor.However,ifthesumof
strategiesexceedsC,noplayerreceivesanything.Figure8illustratesthefeasiblesetforthisgame.

Figure8:ThefeasiblesetforthegameofDividetheCake.Inthisfigure,thecakeisof
sizeC=10butallstrategiesbetween0and10inclusivearepermittedforeitherplayer
(includingfractionaldemands).
WehaveaclearintuitionthattheobviousstrategyforeachplayertoselectisC/2thephilosophical
problemliesinexplainingwhyagentswouldchoosethatstrategyratherthansomeotherone.Evenin
theperfectlysymmetricsituation,answeringthisquestionismoredifficultthanitfirstappears.Tosee
this,firstnoticethatthereareaninfinitenumberofNashequilibriaforthisgame.Ifplayer1asksfor
pofthecake,where0pC,andplayer2asksforCp,thenthisstrategyprofileisaNash
equilibriumforanyvalueofp[0,C].(Eachplayer'sstrategyisabestresponsegivenwhattheother
haschosen,inthesensethatneitherplayercanincreaseherpayoffbychangingherstrategy.)
However,theequalsplitisonlyoneofinfinitelymanyNashequilibria.
Onemightproposethatbothplayersshouldchoosethatstrategywhichmaximizestheirexpected
payoffontheassumptiontheyareuncertainastowhethertheywillbeassignedtheroleofPlayer1or
Player2.Thisproposal,Skyrmsnotes,isessentiallythatofHarsanyi(1953).Theproblemwiththisis
thatifplayersonlycareabouttheirexpectedpayoff,andtheythinkthatitisequallylikelythatthey
willbeassignedtheroleofPlayer1orPlayer2,thenthis,too,failstoselectuniquelytheequalsplit.
Considerthestrategyprofile p,Cp whichassignsPlayer1pslicesandPlayer2Cpslices.If
aplayerthinksitisequallylikelythathewillbeassignedtheroleofPlayer1orPlayer2,thenhis
expectedutilityisp+(Cp)=C/2,forallvaluesp[0,C].
Nowconsiderthefollowingevolutionarymodel:supposewehaveapopulationofindividualswho
pairupandrepeatedlyplaythegameofdividethecake,modifyingtheirstrategiesovertimeinaway
whichisdescribedbythereplicatordynamics.Forconvenience,letusassumethatthecakeisdivided
into10equallysizedslicesandthateachplayer'sstrategyconformstooneofthefollowing11
possibletypes:Demand0slices,Demand1slice,,Demand10slices.Forthereplicatordynamics,
thestateofthepopulationisrepresentedbyavector p0,p1,,p10 whereeachpidenotesthe
frequencyofthestrategyDemandislicesinthepopulation.

Thereplicatordynamicsallowsustomodelhowthedistributionofstrategiesinthepopulation
changesovertime,beginningfromaparticularinitialcondition.Figure9belowshowstwo
evolutionaryoutcomesunderthecontinuousreplicatordynamics.Noticethatalthoughfairdivision
canevolve,asinFigure9(a),itisnottheonlyevolutionaryoutcome,asFigure9(b)illustrates.

(a)Theevolutionoffairdivision.

(b)Theevolutionofanunequaldivisionrule.
Figure9:Twoevolutionaryoutcomesunderthecontinuousreplicatordynamicsforthe
gameofdividethecake.Oftheelevenstrategiespresent,onlythreearecolourcodedso
astobeidentifiableintheplot(seethelegend).Theinitialconditionsforthesolution
shownin(a)wasthepoint 0.0544685,0.236312,0.0560727,0.0469244,0.0562243,
0.0703294,0.151136,0.162231,0.0098273,0.111366,0.0451093 ,andtheinitial
conditionsforthesolutionshownin(b)wasthepoint 0.410376,0.107375,0.0253916,
0.116684,0.0813494,0.00573677,0.0277155,0.0112791,0.0163166,0.191699,
0.00607705 .
Recallthatthetaskathandwastoexplainwhywethinktheobviousstrategychoiceinaperfectly

symmetricresourceallocationproblemisforbothplayerstoaskforhalfoftheresource.Whatthe
aboveshowsisthat,inapopulationofboundedlyrationalagentswhomodifytheirbehavioursina
mannerdescribedbythereplicatordynamics,fairdivisionisone,althoughnottheonly,evolutionary
outcome.Thetendencyoffairdivisiontoemerge,assumingthatanyinitialconditionisequallylikely,
canbemeasuredbydeterminingthesizeofthebasinofattractionofthestatewhereeveryoneinthe
populationusesthestrategyDemand5slices.Skyrms(1996)measuresthesizeofthebasinof
attractionoffairdivisionusingMonteCarlomethods,findingthatfairdivisionevolvesroughly62%
ofthetime.
However,itisimportanttorealisethatthereplicatordynamicsassumesanypairwiseinteraction
betweenindividualsisequallylikely.Inreality,quiteofteninteractionsbetweenindividualsare
correlatedtosomeextent.Correlatedinteractioncanoccurasaresultofspatiallocation(asshown
aboveforthecaseofthespatialprisoner'sdilemma),thestructuringeffectofsocialrelations,or
ingroup/outgroupmembershipeffects,tolistafewcauses.
Whencorrelationisintroduced,thefrequencywithwhichfairdivisionemergeschangesdrastically.
Figure10illustrateshowthebasinofattractionofAllDemand5changesasthecorrelationcoefficient
increasesfrom0to0.2.[7]Oncetheamountofcorrelationpresentintheinteractionsreaches=0.2,
fairdivisionisvirtuallyanevolutionarycertainty.Notethatthisdoesnotdependonthereonlybeing
threestrategiespresent:allowingforsomecorrelationbetweeninteractionsincreasestheprobability
offairdivisionevolvingeveniftheinitialconditionscontainindividualsusinganyoftheeleven
possiblestrategies.

(a)=0

(c)=0.2

(b)=0.1

Figure10:Threediagramsshowinghow,astheamountofcorrelationamong
interactionsincreases,fairdivisionismorelikelytoevolve.
What,then,canweconcludefromthismodelregardingtheevolutionoffairdivision?Italldepends,
ofcourse,onhowaccuratelythereplicatordynamicsmodelstheprimaryevolutionaryforces(cultural
orbiological)actingonhumanpopulations.Althoughthereplicatordynamicsareasimple
mathematicalmodel,itdoessufficeformodellingbothatypeofbiologicalevolution(seeTaylorand
Jonker,1978)andatypeofculturalevolution(seeBrgersandSarin,1996Weibull,1995).As
Skyrms(1996)notes:
Inafinitepopulation,inafinitetime,wherethereissomerandomelementinevolution,
somereasonableamountofdivisibilityofthegoodandsomecorrelation,wecansaythat
itislikelythatsomethingclosetoshareandsharealikeshouldevolveindividingthe
cakesituations.Thisis,perhaps,abeginningofanexplanationoftheoriginofour
conceptofjustice.
Thisclaim,ofcourse,hasnotgonewithoutcomment.Foraselectionofsomediscussionsee,in
particular,D'Arms(1996,2000)D'Armsetal.,1998Danielson(1998)Bicchieri(1999)Kitcher
(1999)Gintis(2000)Harms(2000)Krebs(2000)AlexanderandSkyrms(1999)andAlexander
(2000,2007).

4.2Theemergenceoflanguage.
InhisseminalworkConvention,DavidLewisdevelopedtheideaofsenderreceivergames.Such
gameshavebeenusedtoexplainhowlanguage,andsemanticcontent,canemergeinacommunity
whichoriginallydidnotpossessanylanguagewhatsoever.[8]Hisoriginaldefinitionisasfollows
(withportionsofextraneouscommentarydeletedforconcisionandpointsenumeratedforclarityand
laterreference):
AtwosidedsignalingproblemisasituationSinvolvinganagentcalledthe
communicatorandoneormoreotheragentscalledtheaudience,suchthatitistruethat,
anditiscommonknowledgeforthecommunicatorandtheaudiencethat:
1. Exactlyoneofseveralalternativestatesofaffairss1,,smholds.The
communicator,butnottheaudience,isinagoodpositiontotellwhichoneitis.
2. Eachmemberoftheaudiencecandoanyoneofseveralalternativeactionsr1,,
rmcalledresponses.Everyoneinvolvedwantstheaudience'sresponsestodepend
inacertainwayuponthestateofaffairsthatholds.Thereisacertainonetoone
functionFfrom{si}onto{rj}suchthateveryoneprefersthateachmemberofthe
audiencedoF(si)onconditionthatsiholds,foreachsi.
3. Thecommunicatorcandoanyoneofseveralalternativeactions1,,n(nm)
calledsignals.Theaudienceisinagoodpositiontotellwhichonehedoes.Noone
involvedhasanypreferenceregardingtheseactionswhichisstrongenoughto
outweighhispreferenceforthedependenceFofaudience'sresponsesuponstates
ofaffairs.[]
4. Acommunicator'scontingencyplanisanypossiblewayinwhichthe
communicator'ssignalmaydependuponthestateofaffairsthatheobservesto
hold.ItisafunctionFcfrom{si}into{k}.[]
5. Similarly,anaudience'scontingencyplanisanypossiblewayinwhichthe
responseofamemberoftheaudiencemaydependuponthesignalheobservesthe
communicatortogive.ItisaonetoonefunctionFafrompartof{k}into{rj}.

[]
WheneverFcandFacombine[]togivethepreferreddependenceoftheaudience's
responseuponthestateofaffairs,wecall Fc,Fa asignalingsystem.(Lewis,1969,
pp.130132)
SincethepublicationofConvention,itismorecommontorefertothecommunicatorasthesender
andthemembersoftheaudienceasreceivers.Thebasicideabehindsenderreceivergamesisthe
following:Natureselectswhichstateoftheworldobtains.ThepersonintheroleofSenderobserves
thisstateoftheworld(correctlyidentifyingit),andsendsasignaltothepersonintheroleof
Receiver.TheReceiver,uponreceiptofthissignal,performsaresponse.IfwhattheReceiverdoesis
thecorrectresponse,giventhestateoftheworld,thenbothplayersreceiveapayoffof1ifthe
Receiverperformedanincorrectresponse,thenbothplayersreceiveapayoffof0.Noticethat,inthis
simplifiedmodel,nochanceoferrorexistsatanystage.TheSenderalwaysobservesthetruestateof
theworldandalwayssendsthesignalheintendedtosend.Likewise,theReceiveralwaysreceivesthe
signalsentbytheSender(i.e.,thechannelisnotnoisy),andtheReceiveralwaysperformsthe
responseheintendedto.
WhereasLewisallowedtheaudiencetoconsistofmorethanoneperson,itismorecommonto
considersenderreceivergamesplayedbetweentwopeople,sothatthereisonlyasinglereceiver(or,
inLewisianterms,asinglememberoftheaudience).[9]Forsimplicity,inthefollowingwewill
consideratwoplayer,senderreceivergamewithtwostatesoftheworld{S1,S2},twosignals{1,
2},andtworesponses{r1,r2}.(Weshallseelaterwhylargersenderreceivergamesareincreasingly
difficulttoanalyse.)
Noticethat,inpoint(2)ofhisdefinitionofsenderreceivergames,Lewisrequirestwothings:that
therebeauniquebestresponsetothestateoftheworld(thisiswhatrequiringFtobeonetoone
amountsto)andthateveryoneintheaudienceagreesthatthisisthecase.Sinceweareconsideringthe
casewherethereisonlyasingleresponder,thesecondrequirementisotiose.Forthecaseoftwo
statesoftheworldandtworesponses,thereareonlytwowaysofassigningresponsestostatesofthe
worldwhichsatisfyLewis'srequirement.Theseareasfollows(whereXYdenotesinstateofthe
worldX,thebestresponseistodoY):
1. S1r1,S2r2.
2. S1r2,S2r1.
Itmakesnorealdifferenceforthemodelwhichoneofthesewechoose,sopicktheintuitiveone:in
stateoftheworldSi,thebestresponseisri(i.e.,function1).
Astrategyforthesender(whatLewiscalledacommunicator'scontingencyplan)consistsofa
functionspecifyingwhatsignalhesendsgiventhestateoftheworld.Itis,asLewisnotes,afunction
fromthesetofstatesoftheworldintothesetofsignals.Thismeansthatitispossiblethatasender
maysendthesamesignalintwodifferentstatesoftheworld.Suchastrategymakesnosense,froma
rationalpointofview,becausethereceiverwouldnotgetenoughinformationtobeabletoidentify
thecorrectresponseforthestateoftheworld.However,wedonotexcludethesestrategiesfrom
considerationbecausetheyarelogicallypossiblestrategies.
Howmanysenderstrategiesarethere?Becauseweallowforthepossibilityofthesamesignaltobe
sentformultiplestatesoftheworld,therearetwochoicesforwhichsignaltosendgivenstateS1and
twochoicesforwhichsignaltosendgivenstateS2.Thismeanstherearefourpossiblesender
strategies.Thesestrategiesareasfollows(where'XY'meansthatwhenthestateoftheworldisX
thesenderwillsendsignalY):

Sender1:S11,S21.
Sender2:S11,S22.
Sender3:S12,S21.
Sender4:S12,S22.
Whatisastrategyforareceiver?Here,itprovesusefultodeviatefromLewis'soriginaldefinitionof
theaudience'scontingencyplan.Instead,letustakeareceiver'sstrategytobeafunctionfromthe
setofsignalsintothesetofresponses.Asinthecaseofthesender,weallowthereceivertoperform
thesameresponseformorethanonesignal.Bysymmetry,thismeansthereare4possiblereceiver
strategies.Thesereceiverstrategiesare:
Receiver1:1r1,2r1.
Receiver2:1r1,2r2.
Receiver3:1r2,2r1.
Receiver4:1r2,2r2.
IftherolesofSenderandReceiverarepermanentlyassignedtoindividualsasLewisenvisaged
thenthereareonlytwopossiblesignalingsystems: Sender2,Receiver2 and Sender3,Receiver
3 .Allotherpossiblecombinationsofstrategiesresultintheplayersfailingtocoordinate.The
coordinationfailureoccursbecausetheSenderandReceiveronlypairtheappropriateactionwiththe
stateoftheworldinoneinstance,aswith Sender1,Receiver1 ,ornotatall,aswith Sender2,
Receiver3 .
WhatiftherolesofSenderandReceiverarenotpermanentlyassignedtoindividuals?Thatis,whatif
natureflipsacoinandassignsoneplayertotheroleofSenderandtheotherplayertotheroleof
Receiver,andthenhasthemplaythegame?Inthiscase,aplayer'sstrategyneedstospecifywhathe
willdowhenassignedtheroleofSender,aswellaswhathewilldowhenassignedtheroleof
Receiver.SincetherearefourpossiblestrategiestouseasSenderandfourpossiblestrategiestouseas
Receiver,thismeansthatthereareatotalof16possiblestrategiesforthesenderreceivergamewhen
rolesarenotpermanentlyassignedtoindividuals.Here,aplayer'sstrategyconsistsofanorderedpair
(SenderX,ReceiverY),whereX,Y{1,2,3,4}.
ItmakesadifferencewhetheroneconsiderstherolesofSenderandReceivertobepermanently
assignedornot.Iftherolesareassignedatrandom,therearefoursignalingsystemsamongsttwo
players[10]:
1. Player1:(Sender2,Receiver2),Player2:(Sender2,Receiver2)
2. Player1:(Sender3,Receiver3),Player2:(Sender3,Receiver3)
3. Player1:(Sender2,Receiver3),Player2:(Sender3,Receiver2)
4. Player1:(Sender3,Receiver2),Player2:(Sender2,Receiver3)
Signalingsystems3and4arecurious.System3isacasewhere,forexample,IspeakinFrenchbut
listeninGerman,andyouspeakGermanbutlisteninFrench.(System4swapsFrenchandGerman
forbothyouandme.)Noticethatinsystems3and4theplayersareabletocorrectlycoordinatethe
responsewiththestateoftheworldregardlessofwhogetsassignedtheroleofSenderorReceiver.
Theproblem,ofcourse,withsignalingsystems3and4isthatneitherPlayer1norPlayer2woulddo
wellwhenpittedagainstacloneofhimself.Theyarecaseswherethesignalingsystemwouldnot
workinapopulationofplayerswhoarepairwiserandomlyassignedtoplaythesenderreceivergame.
Infact,itisstraightforwardtoshowthatthestrategies(Sender2,Receiver2)and(Sender3,Receiver
3)aretheonlyevolutionarilystablestrategies(seeSkyrms1996,8990).
Asafirstapproachtothedynamicsofsenderreceivergames,letusrestrictattentiontothefour

strategies(Sender1,Receiver1),(Sender2,Receiver2),(Sender3,Receiver3),and(Sender4,
Receiver4).Figure11illustratesthestatespaceunderthecontinuousreplicatordynamicsforthe
senderreceivergameconsistingoftwostatesoftheworld,twosignals,andtworesponses,where
playersarerestrictedtousingoneofthepreviousfourstrategies.Onecanseethatevolutionleadsthe
populationinalmostallcases[11]toconvergetooneofthetwosignalingsystems.[12]

Figure11:Theevolutionofsignalingsystems.

Figure12illustratestheoutcomeofonerunofthereplicatordynamics(forasinglepopulationmodel)
whereallsixteenpossiblestrategiesarerepresented.Weseethateventuallythepopulation,forthis
particularsetofinitialconditions,convergestooneofthepureLewisiansignallingsystemsidentified
above.

Figure12:Theevolutionofasignallingsystemunderthereplicatordynamics.
Whenthenumberofstatesoftheworld,thenumberofsignals,andthenumberofactionsincrease
from2,thesituationrapidlybecomesmuchmorecomplex.IfthereareNstatesoftheworld,N
signals,andNactions,thetotalnumberofpossiblestrategiesequalsN2N.ForN=2,thismeansthere
are16possiblestrategies,aswehaveseen.ForN=3,thereare729possiblestrategies,andasignalling
problemwhereN=4has65,536possiblestrategies.Giventhis,onemightthinkthatitwouldprove
difficultforevolutiontosettleuponanoptimalsignallingsystem.
Suchanintuitioniscorrect.HofbauerandHutteger(2008)showthat,quiteoften,thereplicator
dynamicswillconvergetoasuboptimaloutcomeinsignallinggames.Inthesesuboptimaloutcomes,a
poolingorpartialpoolingequilibriumwillemerge.ApoolingequilibriumoccurswhentheSender
usesthesamesignalregardlessofthestateoftheworld.Apartialpoolingequilibriumoccurswhen
theSenderiscapableofdifferentiatingbetweensomestatesoftheworldbutnotothers.Asan
exampleofapartialpoolingequilibrium,considerthefollowingstrategiesforthecasewhereN=3:
SupposethattheSendersendssignal1instateoftheworld1,andsignal2instatesoftheworld2and
3.Furthermore,supposethattheReceiverperformsaction1uponreceiptofsignal1,andaction2
uponreceiptofsignals2and3.Ifallstatesoftheworldareequiprobable,thisisapartialpooling
equilibrium.GiventhattheSenderdoesnotdifferentiatestatesoftheworld2and3,theReceiver
cannotimprovehispayoffsbyrespondingdifferentlytosignal2.Giventheparticularresponse
behaviouroftheReceiver,theSendercannotimproveherpayoffsbyattemptingtodifferentiatestates
oftheworld2and3.

5.PhilosophicalProblemsofEvolutionaryGameTheory
Thegrowinginterestamongsocialscientistsandphilosophersinevolutionarygametheoryhasraised
severalphilosophicalquestions,primarilystemmingfromitsapplicationtohumansubjects.

5.1Themeaningoffitnessinculturalevolutionaryinterpretations
Asnotedpreviously,evolutionarygametheoreticmodelsmayoftenbegivenbothabiologicalanda
culturalevolutionaryinterpretation.Inthebiologicalinterpretation,thenumericquantitieswhichplay
aroleanalogoustoutilityintraditionalgametheorycorrespondtothefitness(typicallyDarwinian
fitness)ofindividuals.[13]Howdoesoneinterpretfitnessintheculturalevolutionaryinterpretation?
Inmanycases,fitnessinculturalevolutionaryinterpretationsofevolutionarygametheoreticmodels
directlymeasuressomeobjectivequantityofwhichitcanbesafelyassumedthat(1)individuals
alwayswantmoreratherthanlessand(2)interpersonalcomparisonsaremeaningful.Dependingon
theparticularproblemmodeled,money,slicesofcake,oramountoflandwouldbeappropriate
culturalevolutionaryinterpretationsoffitness.Requiringthatfitnessinculturalevolutionarygame
theoreticmodelsconformtothisinterpretativeconstraintseverelylimitsthekindsofproblemsthat
onecanaddress.Amoreusefulculturalevolutionaryframeworkwouldprovideamoregeneraltheory
whichdidnotrequirethatindividualfitnessbealinear(orstrictlyincreasing)functionoftheamount
ofsomerealquantity,likeamountoffood.
Intraditionalgametheory,astrategy'sfitnesswasmeasuredbytheexpectedutilityithadforthe
individualinquestion.Yetevolutionarygametheoryseekstodescribeindividualsoflimited
rationality(commonlyknownasboundedlyrationalindividuals),andtheutilitytheoryemployedin
traditionalgametheoryassumeshighlyrationalindividuals.Consequently,theutilitytheoryusedin
traditionalgametheorycannotsimplybecarriedovertoevolutionarygametheory.Onemustdevelop
analternatetheoryofutility/fitness,onecompatiblewiththeboundedrationalityofindividuals,thatis
sufficienttodefineautilitymeasureadequatefortheapplicationofevolutionarygametheoryto
culturalevolution.

5.2Theexplanatoryirrelevanceofevolutionarygametheory
Anotherquestionfacingevolutionarygametheoreticexplanationsofsocialphenomenaconcernsthe
kindofexplanationitseekstogive.Dependingonthetypeofexplanationitseekstoprovide,are
evolutionarygametheoreticexplanationsofsocialphenomenairrelevantormerevehiclesforthe
promulgationofpreexistingvaluesandbiases?Tounderstandthisquestion,recognizethatonemust
askwhetherevolutionarygametheoreticexplanationstargettheetiologyofthephenomenonin
question,thepersistenceofthephenomenon,orvariousaspectsofthenormativityattachedtothe
phenomenon.Thelattertwoquestionsseemdeeplyconnected,forpopulationmemberstypically
enforcesocialbehaviorsandruleshavingnormativeforcebysanctionsplacedonthosefailingto
complywiththerelevantnormandthepresenceofsanctions,ifsuitablystrong,explainsthe
persistenceofthenorm.Thequestionregardingaphenomenon'setiology,ontheotherhand,canbe
consideredindependentofthelatterquestions.
Ifonewishestoexplainhowsomecurrentlyexistingsocialphenomenoncametobe,itisunclearwhy
approachingitfromthepointofviewofevolutionarygametheorywouldbeparticularily
illuminating.Theetiologyofanyphenomenonisauniquehistoricaleventand,assuch,canonlybe
discoveredempirically,relyingontheworkofsociologists,anthropologists,archaeologists,andthe
like.Althoughanevolutionarygametheoreticmodelmayexcludecertainhistoricalsequencesas
possiblehistories(sinceonemaybeabletoshowthattheculturalevolutionarydynamicsprecludeone
sequencefromgeneratingthephenomenoninquestion),itseemsunlikelythatanevolutionarygame
theoreticmodelwouldindicateauniquehistoricalsequencesufficestobringaboutthephenomenon.
Anempiricalinquirywouldthenstillneedtobeconductedtoruleouttheextraneoushistorical
sequencesadmittedbythemodel,whichraisesthequestionofwhat,ifanything,wasgainedbythe
constructionofanevolutionarygametheoreticmodelintheintermediatestage.Moreover,evenifan
evolutionarygametheoreticmodelindicatedthatasinglehistoricalsequencewascapableof
producingagivensocialphenomenon,thereremainstheimportantquestionofwhyweoughttotake

thisresultseriously.Onemaypointoutthatsincenearlyanyresultcanbeproducedbyamodelby
suitableadjustingofthedynamicsandinitialconditions,allthattheevolutionarygametheoristhas
doneisprovideonesuchmodel.Additionalworkneedstobedonetoshowthattheunderlying
assumptionsofthemodel(boththeculturalevolutionarydynamicsandtheinitialconditions)are
empiricallysupported.Again,onemaywonderwhathasbeengainedbytheevolutionarymodel
woulditnothavebeenjustaseasytodeterminetheculturaldynamicsandinitialconditions
beforehand,constructingthemodelafterwards?Ifso,itwouldseemthatthecontributionsmadeby
evolutionarygametheoryinthiscontextsimplyareaproperpartoftheparentsocialscience
sociology,anthropology,economics,andsoon.Ifso,thenthereisnothingparticularabout
evolutionarygametheoryemployedintheexplanation,andthismeansthat,contrarytoappearances,
evolutionarygametheoryisreallyirrelevanttothegivenexplanation.
Ifevolutionarygametheoreticmodelsdonotexplaintheetiologyofasocialphenomenon,
presumablytheyexplainthepersistenceofthephenomenonorthenormativityattachedtoit.Yetwe
rarelyneedanevolutionarygametheoreticmodeltoidentifyaparticularsocialphenomenonasstable
orpersistentasthatcanbedonebyobservationofpresentconditionsandexaminationofthehistorical
recordshencethechargeofirrelevancyisraisedagain.Moreover,mostoftheevolutionarygame
theoreticmodelsdevelopedtodatehaveprovidedthecrudestapproximationsoftherealcultural
dynamicsdrivingthesocialphenomenoninquestion.Onemaywellwonderwhy,inthesecases,we
shouldtakeseriouslythestabilityanalysisgivenbythemodelansweringthisquestionwouldrequire
oneengageinanempiricalstudyaspreviouslydiscussed,ultimatelyleadingtothechargeof
irrelevanceagain.

5.3Thevalueladennessofevolutionarygametheoreticexplanations
Ifoneseekstouseanevolutionarygametheoreticmodeltoexplainthenormativityattachedtoa
socialrule,onemustexplainhowsuchanapproachavoidscommittingthesocallednaturalistic
fallacyofinferringanoughtstatementfromaconjunctionofisstatements.[14]Assumingthatthe
explanationdoesnotcommitsuchafallacy,oneargumentchargesthatitmustthenbethecasethat
theevolutionarygametheoreticexplanationmerelyrepackagescertainkeyvalueclaimstacitly
assumedintheconstructionofthemodel.Afterall,sinceanyargumentwhoseconclusionisa
normativestatementmusthaveatleastonenormativestatementinthepremises,anyevolutionary
gametheoreticargumentpurportingtoshowhowcertainnormsacquirenormativeforcemustcontain
atleastimplicitlyanormativestatementinthepremises.Consequently,thisapplicationof
evolutionarygametheorydoesnotprovideaneutralanalysisofthenorminquestion,butmerelyacts
asavehicleforadvancingparticularvalues,namelythosesmuggledinthepremises.
Thiscriticismseemslessseriousthanthechargeofirrelevancy.Culturalevolutionarygametheoretic
explanationsofnormsneednotsmuggleinnormativeclaimsinordertodrawnormative
conclusions.Thetheoryalreadycontains,initscore,apropersubtheoryhavingnormativecontent
namelyatheoryofrationalchoiceinwhichboundedlyrationalagentsactinordertomaximize,as
bestastheycan,theirownselfinterest.Onemaychallengethesuitabilityofthisasafoundationfor
thenormativecontentofcertainclaims,butthisisadifferentcriticismfromtheabovecharge.
Althoughculturalevolutionarygametheoreticmodelsdoactasvehiclesforpromulgatingcertain
values,theywearthoseminimalvaluecommitmentsontheirsleeve.Evolutionaryexplanationsof
socialnormshavethevirtueofmakingtheirvaluecommitmentsexplicitandalsoofshowinghow
othernormativecommitments(suchasfairdivisionincertainbargainingsituations,orcooperationin
theprisoner'sdilemma)maybederivedfromtheprincipledactionofboundedlyrational,self
interestedagents.

Bibliography
Ackley,DavidandMichaelLittman(1994).InteractionsBetweenLearningandEvolution,in

ChristopherG.Langton,ed.,ArtificialLifeIII.Reading,MA:AddisonWesley,pp.487509.
Adachi,N.andMatsuo,K.(1991).EcologicalDynamicsUnderDifferentSelectionRulesin
DistributedandIteratedPrisoner'sDilemmaGames,ParallelProblemSolvingFromNature
(LectureNotesinComputerScience:Volume496),Berlin:SpringerVerlag,pp.388394.
Alexander,J.McKenzie(2000).EvolutionaryExplanationsofDistributiveJustice,
PhilosophyofScience,67:490516.
(2007).TheStructuralEvolutionofMorality,Cambridge:CambridgeUniversityPress.
Alexander,JasonandBrianSkyrms(1999).BargainingwithNeighbors:IsJustice
Contagious?JournalofPhilosophy,96(11):588598.
Axelrod,R.(1984).TheEvolutionofCooperation.NewYork:BasicBooks.
(1986).Anevolutionaryapproachtonorms,AmericanPoliticalScienceReview,80(4):
10951111.
Axelrod,RobertM.andDion,Douglas(1988).TheFurtherEvolutionofCooperation,
Science,242(4884):13851390.
Axelrod,RobertM.andHamilton,WilliamD.(1981).TheEvolutionofCooperation,Science,
211(4489):13901396.
Banerjee,AbhijitV.andWeibull,JoergenW.(1993).EvolutionarySelectionwith
DiscriminatingPlayers,WorkingPaper#375,ResearchInstituteofIndustrialEconomics,
UniversityofStockholm.
Barrett,JeffreyA.(2007).DynamicPartitioningandtheConventionalityofKinds,
PhilosophyofScience,74(4):527546.
Bergin,J.andLipman,B.(1996).EvolutionwithStateDependentMutations,Econometrica,
64:943956.
Bicchieri,Cristina(2006).TheGrammarofSociety,Cambridge:CambridgeUniversityPress.
Binmore,KenandSamuelson,Larry(1991).EvolutionaryStabilityinRepeatedGamesPlayed
ByFiniteAutomata,JournalofEconomicTheory,57:278305.
(1994).AnEconomist'sPerspectiveontheEvolutionofNorms,JournalofInstitutional
andTheoreticalEconomics,150(1):4563.
Bjrnerstedt,J.andWeibull,J.(1993).NashEquilibriumandEvolutionbyImitation,in
Arrow,K.andColombatto,E.(eds.),RationalityinEconomics,NewYork:Macmillan.
Blume,L.(1993).TheStatisticalMechanicsofStrategicInteraction,GamesandEconomic
Behaviour,5:387424.
Blume,LawrenceE.(1997).PopulationGames,inW.BrianArthur,StevenN.Durlauf,and
DavidA.Lane(eds.),TheEconomyasanEvolvingComplexSystemII(SFIStudiesinthe
SciencesofComplexity:Volume27),Reading,MA:AddisonWesley,,pp.425460.
Boehm,C.(1982).Theevolutionarydevelopmentofmoralityasaneffectofdominance
behaviorandconflictinterference,JournalofSocialandBiologicalStructures,5:413421.
Bgers,TilmanandSarin,R.(1996).NaiveReinforcementandReplicatorDynamics,
WorkingPaper,CentreforEconomicLearningandSocialEvolution,UniversityCollege
London.
(1997).LearningThroughReinforcementandReplicatorDynamics,Journalof
EconomicTheory,77(1):114.
Boyd,RobertandLorberbaum,JeffreyP.(1987).NoPureStrategyisEvolutionarilyStablein
theRepeatedPrisoner'sDilemmaGame,Nature,32(7)(May7):5859.
Boylan,RichardT.(1991).LawsofLargeNumbersforDynamicalSystemswithRandomly
MatchedIndividuals,JournalofEconomicTheory,57:473504.
Busch,MarcL.andReinhardt,EricR.(1993).NiceStrategiesinaWorldofRelativeGains:
TheProblemofCooperationunderAnarchy,JournalofConflictResolution,37(3):427445.
Cabrales,A.andPonti,G.(1996).Implementation,EliminationofWeaklyDominated
StrategiesandEvolutionaryDynamics,WorkingPaper,CentreforEconomicLearningand
SocialEvolution,UniversityCollegeLondon.
Canning,David(1988).RationalityandGameTheoryWhenPlayersareTuringMachines,
ST/ICERDDiscussionPaper88/183,London:LondonSchoolofEconomics.
Canning,David(1990c).Rationality,ComputabilityandtheLimitsofGameTheory,

EconomicTheoryDiscussionPaperNumber152,DepartmentofAppliedEconomics,
UniversityofCambridge,July.
Canning,David(1992).Rationality,ComputabilityandNashEquilibrium,Econometrica,
60(4):877888.
Chalub,F.A.C.C.,Santos,F.C.andJ.M.Pacheco(2006).Theevolutionofnorms,Journalof
TheoreticalBiology,241:233240.
Cho,I.K.andKreps,DavidM.(1987).SignalingGamesandStableEquilibria,Quarterly
JournalofEconomics,102(1):179221.
Clemens,ChristianeandThomasRiechmann(2006).EvolutionaryDynamicsinPublicGoods
Games,ComputationalEconomics,28:399420.
Cowan,RobinA.andMiller,JohnH.(1990).EconomicLifeonaLattice:SomeGame
TheoreticResults,WorkingPaper90010,EconomicsResearchProgram,SantaFeInstitute,
NewMexico.
D'Arms,Justin,RobertBatterman,andKrzyzstofGrny(1998).GameTheoreticExplanations
andtheEvolutionofJustice,PhilosophyofScience,65:76102.
D'Arms,Justin(1996).Sex,Fairness,andtheTheoryofGames,JournalofPhilosophy,93
(12):615627.
(2000).WhenEvolutionaryGameTheoryExplainsMorality,WhatDoesItExplain?
JournalofConsciousnessStudies7(12):296299.
Danielson,P.(1992).ArtificialMorality:VirtuousRobotsforVirtualGames,London:
Routledge.
(1998).CriticalNotice:EvolutionoftheSocialContract,CanadianJournalof
Philosophy,28(4):627652.
Dekel,EddieandScotchmer,Suzanne(1992).OntheEvolutionofOptimizingBehavior,
JournalofEconomicTheory,57:392406.
Eaton,B.C.andSlade,M.E.(1990).EvolutionaryEquilibriuminMarketSupergames,
DiscussionPaper9030(November1989),DepartmentofEconomics,UniversityofBritish
Columbia.
Ellingsen,Tore(1997).TheEvolutionofBargainingBehavior,TheQuarterlyJournalof
Economics,112(1):581602.
Ellison,G.(1993).Learning,LocalInteractionandCoordination,Econometrica,61:1047
1071.
Enquist,MagnusandStefanoGhirlanda(2007).EvolutionofSocialLearningDoesNot
ExplaintheOriginofHumanCumulativeCulture,JournalofTheoreticalBiology,246:129
135.
Enquist,M.,Ghirlanda,S.,Jarrick,A.,andWachtmeister,C.A.(2008).WhyDoesHuman
CultureIncreaseExponentially?TheoreticalPopulationBiology,74:4655.
Epstein,JoshuaA.(1998).ZonesofCooperationinDemographicPrisoner'sDilemma,
Complexity,4(2):3648.
Eshel,Ilan,LarrySamuelson,andAvnerShaked(1998).Altruists,Egoists,andHooligansina
LocalInteractionModel,TheAmericanEconomicReview,88(1):157179.
Fishman,MichaelA.(2006).Involuntarydefectionandtheevolutionaryoriginsofempathy,
JournalofTheoreticalBiology,242:873879.
Fisher,R.A.(1930).TheGeneticTheoryofNaturalSelection,Oxford,ClarendonPress.
Fletcher,JeffreyA.andMartinZwick(2007).Theevolutionofaltruism:Gametheoryin
multilevelselectionandinclusivefitness,JournalofTheoreticalBiology,245:2636.
Fogel,DavidB.(1993).EvolvingBehavioursintheIteratedPrisoner'sDilemma,
EvolutionaryComputation,1(1):7797.
Forrest,StephanieandMayerKress,G.(1991).GeneticAlgorithms,NonlinearDynamical
Systems,andGlobalStabilityModels,inL.Davis,(ed.),TheHandbookofGenetic
Algorithms,NewYork:VanNostrandReinhold.
Foster,DeanandYoung,H.Peyton(1990).StochasticEvolutionaryGameDynamics,
JournalofTheoreticalBiology,38:219232.
Friedman,Daniel(1991).EvolutionaryGamesinEconomics,Econometrica,59(3):637

666.
Fudenberg,DrewandMaskin,Eric(1990).EvolutionandCooperationinNoisyRepeated
Games,AmericanEconomicReview(PapersandProceedings),80(2):274279.
Gintis,Herbert(2000).ClassicalVersusEvolutionaryGameTheory,Journalof
ConsciousnessStudies,7(12):300304.
(2007).Theevolutionofprivateproperty,JournalofEconomicBehavior&
Organization,64:116.
Gintis,Herbert,SamuelBowles,RobertBoydandErnstFehr(2003).Explainingaltruistic
behaviorinhumans,EvolutionandHumanBehavior,24:153172.
Guth,WernerandKliemt,Hartmut(1994).CompetitionorCooperationOnthe
EvolutionaryEconomicsofTrust,ExploitationandMoralAttitudes,Metroeconomica,45:
155187.
Guth,WernerandKliemt,Hartmut(1998).TheIndirectEvolutionaryApproach:Bridgingthe
GapBetweenRationalityandAdaptation,RationalityandSociety,10(3):377399.
Hamilton,W.D.(1963).TheEvolutionofAltruisticBehavior,TheAmericanNaturalist,97:
354356.
(1964).TheGeneticalEvolutionofSocialBehavior.I,JournalofTheoreticalBiology,7:
116.
(1964).TheGeneticalEvolutionofSocialBehavior.II,JournalofTheoreticalBiology,
7:1752.
Hammerstein,P.andSelten,R.(1994).GameTheoryandEvolutionaryBiology,inR.
AumanandS.Hart(eds.),HandbookofGameTheorywithEconomicApplications(Volume2),
Amsterdam:ElsevierScience,pp.931962.
Hansen,R.G.andSamuelson,W.F.(1988).EvolutioninEconomicGames,Journalof
EconomicBehaviorandOrganization,10(3):315338.
Harms,William(1997).EvolutionandUltimatumBargaining,TheoryandDecision,42:
147175.
(2000).TheEvolutionofCooperationinHostileEnvironments,Journalof
ConsciousnessStudies,7(12):308313.
Harms,WilliamandBrianSkyrms(2008).EvolutionofMoralNorms,inTheOxford
HandbookofPhilosophyofBiology,Oxford:OxfordUniversityPress.
Harsanyi,J.(1953).CardinalUtilityinWelfareEconomicsandtheTheoryofRiskTaking,
JournalofPoliticalEconomy,61:434435.
Harrald,PaulG.(inpress).EvolvingBehaviourinRepeatedGamesviaGeneticAlgorithms,
inP.Stampoultzsis(ed.),TheApplicationsHandbookofGeneticAlgorithms,BocaRaton,FA:
CRCPublishers.
Hassell,MichaelP.,HughN.Comins,andRobertM.May(1991).Spatialstructureandchaos
ininsectpopulationdynamics,Nature,353:255258.
Hauert,Christoph(2006).SpatialEffectsinSocialDilemmas,JournalofTheoretical
Biology,240:627636.
Hauert,Christoph,FranziskaMichor,MartinA.Nowak,andMichaelDoebeli(2006).Synergy
anddiscountingofcooperationinsocialdilemmas,JournalofTheoreticalBiology,239:195
202.
Hauert,Christoph,SilviaDeMonte,JosefHofbauerandKarlSigmund(2002).Replicator
DynamicsforOptionalPublicGoodsGames,JournalofTheoreticalBiology,218:187194.
Hausken,Kjell,andJackHirshleifer(2008).TruthfulSignalling,theHeritabilityParadox,and
theMalthusianEquiMarginalPrinciple,TheoreticalPopulationBiology,73:1123.
Hegselmann,Rainer(1996).SocialDilemmasinLinelandandFlatland,inLiebrandand
Messick(eds.),FrontiersinSocialDilemmasResearch,Berlin:Springer,pp.337361.
Hiebeler,David(1997).StochasticSpatialModels:FromSimulationstoMeanFieldandLocal
StructureApproximations,JournalofTheoreticalBiology,187:307319.
Hines,W.G.(1987).EvolutionaryStableStrategies:AReviewofBasicTheory,Theoretical
PopulationBiology,31:195272.
Hirshleifer,JackandMartinezColl,JuanCarlos(1988).WhatStrategiescanSupportthe

EvolutionaryEmergenceofCooperation?,JournalofConflictResolution,32(2):367398.
Hirshleifer,JackandMartinezColl,JuanCarlos(1992).Selection,Mutationandthe
PreservationofDiversityinEvolutionaryGames,PapersonEconomicsandEvolution,#9202,
editedbytheEuropeanStudyGroupforEvolutionaryEconomics.
Howard,J.V.(1988).CooperationinthePrisoner'sDilemma,TheoryandDecision,24:203
213.
Huberman,BernardoA.andGlance,NatalieS.(1993).EvolutionaryGamesandComputer
Simulations,ProceedingsoftheNationalAcademyofSciencesoftheUSA,90(16):7716
7718.
(1995).TheDynamicsofCollectiveAction,ComputationalEconomics,8:2746.
Hurd,PeterL.(1995).CommunicationinDiscreteActionResponseGames,Journalof
TheoreticalBiology,174:217222.
Ikegami,Takashi(1993).EcologyofEvolutionaryGameStrategies,inSelfOrganizationand
Life:FromSimpleRulestoGlobalComplexity(ProceedingsoftheSecondEuropean
ConferenceonArtificialLife,Brussels,Belgium2426May1993),Cambridge,MA:MIT
Press,pp.527536.
Jger,Gerhard(2008).EvolutionaryStabilityConditionsforSignalingGameswithCostly
Signals,JournalofTheoreticalBiology,253:131141.
Kameda,TatsuyaandDaisukeNakanishi(2003).Doessocial/culturallearningincreasehuman
adaptability?Rogers'squestionrevisited,EvolutionandHumanBehavior,24:242260.
Kandori,Michihiro,Mailath,GeorgeJ.andRob,Rafael(1993).Learning,Mutation,andLong
RunEquilibriainGames,Econometrica,61:2956.
Kendal,Jeremy,MarcusW.Feldman,andKenichiAoki(2006).Culturalcoevolutionofnorm
adoptionandenforcementwhenpunishersarerewardedornonpunishersarepunished,
TheoreticalPopulationBiology,70:1025.
Kreps,DavidM.(1990).GameTheoryandEconomicModelling,Oxford:ClarendonPress.
Kreps,DavidM.andFudenberg,Drew(1988).Learning,Experimentation,andEquilibriumin
Games,Cambridge,MA:MITPress.
Iwasa,Yoh,MayukoNakamaru,andSimonA.Levin(1998).Allelopathyofbacteriaina
latticepopulation:Competitionbetweencolicinsensitiveandcolicinproducingstrains,
EvolutionaryEcology,12:785802.
Kandori,Michihiro,GeorgeJ.Mailath,andRafaelRob(1993).Learning,Mutation,andLong
RunEquilibriainGames,Econometrica,61(1):2956.
Kaneko,KunihikoandJunjiSuzuki(1994).EvolutiontotheEdgeofChaosinanImitation
Game,inChristopherG.Langton(ed.),ArtificialLifeIII,Reading,MA:AddisonWesley,pp.
4353.
Kephart,JeffreyO.(1994).HowTopologyAffectsPopulationDynamics,inChristopherG.
Langton(ed.),ArtificialLifeIII,Reading,MA:AddisonWesley,pp.447463.
Kitcher,Philip(1999).GamesSocialAnimalsPlay:CommentaryonBrianSkyrms'Evolution
oftheSocialContract,PhilosophyandPhenomenologicalResearch,59(1):221228.
Krebs,Dennis(2000).EvolutionaryGamesandMorality,JournalofConsciousnessStudies,
7(12):313321.
Levin,B.R.(1988).Frequencydependentselectioninbacterialpopulations,Philosophical
TransactionsoftheRoyalSocietyofLondon(SeriesB),319:469472.
Lewontin,R.C.(1961).EvolutionandtheTheoryofGamesJournalofTheoreticalBiology,
1:382403.
Liebrand,WimB.G.andMessick,DavidM.(eds.)(1996).FrontiersinSocialDilemmas
Research,Berlin:SpringerVerlag.
Lindgren,Kristian(1990).EvolutioninaPopulationofMutatingStrategies,Preprint90/22S,
Copenhagen:NordicInstituteforTheoreticalPhysics.
Lindgren,KristianandNordahl,MatsG.(1993).EvolutionaryDynamicsofSpatialGames,
inSelfOrganizationandLife:FromSimpleRulestoGlobalComplexity(Proceedingsofthe
SecondEuropeanConferenceonArtificialLife,Brussels,Belgium2426May1993),
Cambridge,MA:MITPress,pp.604616.

Lindgren,KristianandMatsG.Nordahl(1994).Evolutionarydynamicsofspatialgames,
PhysicaD,75:292309.
Lindgren,K.(1991).Evolutionaryphenomenainsimpledynamics,inC.G.Langton,J.D.
Farmer,S.Rasmussen,andC.Taylor(eds.),ArtificialLifeII,RedwoodCity,CA:Addison
Wesley,pp.295312.
Lomborg,Bjorn(1992).CooperationintheIteratedPrisoner'sDilemma,Paperson
EconomicsandEvolution,#9302,editedbytheEuropeanStudyGroupforEvolutionary
Economics.
Lomborg,Bjorn(1996).NucleusandShield:TheEvolutionofSocialStructureintheInterated
Prisoner'sDilemma,AmericanSociologicalReview,61:278307.
Macy,Michael(1989).WalkingOutofSocialTraps:AStochasticLearningModelforthe
Prisoner'sDilemma,RationalityandSociety,1(2):197219.
Mailath,GeorgeJ.(1992).Introduction:SymposiumonEvolutionaryGameTheory,Journal
ofEconomicTheory,57:259277.
Mailath,GeorgeJ.,Samuelson,LarryandShaked,Avner(1992).EvolutionandEndogenous
Interaction,DraftPaper,DepartmentofEconomics,UniversityofPennsylvania,latestversion
24August1995.
Matsui,Akihiko(1993).EvolutionandRationalizability,WorkingPaper:9319(May1993),
CenterforAnalyticResearchinEconomicsandtheSocialSciences(CARESS),Universityof
Pennsylvania.
Mar,Gary(2000).EvolutionaryGameTheory,Morality,andDarwinismJournalof
ConsciousnessStudies,7(12):322326.
May,R.M.,Bohoeffer,S.andNowak,MartinA.(1995).SpatialGamesandtheEvolutionof
Cooperation,inF.Moran,A.Moreno,J.J.MereloandP.Chacon,P.(eds.),Advancesin
ArtificialLife:ProceedingsoftheThirdEuropeanConferenceonArtificialLife(ECAL95),
Berlin:SprngerVerlag,pp.749759.
MaynardSmith,John(1976).EvolutionandtheTheoryofGames,AmericanScientist,64
(1):4145.
MaynardSmith,John(1982).EvolutionandtheTheoryofGames,Cambridge:Cambridge
UniversityPress.
MaynardSmith,JohnandGeorgePrice(1973).TheLogicofAnimalConflictNature,146:
1518.
Miller,JohnH.(1988).TheEvolutionofAutomataintheRepeatedPrisoner'sDilemma,in
TwoEssaysontheEconomicsofImperfectInformation,Ph.D.Dissertation,Departmentof
Economics,UniversityofMichigan/AnnArbor.
(1996).TheCoevolutionofAutomataintheRepeatedPrisoner'sDilemma,Journalof
EconomicBehaviorandOrganization,29(1):87112.
Miller,JohnH.andShubik,Martin(1994).SomeDynamicsofaStrategicMarketGamewith
aLargeNumberofAgents,JournalofEconomics,60:128.
Miller,J.H.andJ.Andreoni(1991).CanEvolutionaryDynamicsExplainFreeRidingin
Experiments?EconomicLetters,36:915.
Nachbar,JohnH.(1990).'EvolutionarySelectionDynamicsinGames:Convergenceand
LimitProperties,InternationalJournalofGameTheory,19:5989.
Nachbar,JohnH.(1992).EvolutionintheFinitelyRepeatedPrisoner'sDilemma:A
MethodologicalCommentandSomeSimulations,JournalofEconomicBehaviourand
Organization,19(3):307326.
Nakahashi,Wataru(2007).TheEvolutionofConformistTransmissioninSocialLearning
whentheEnvironmentChangesPeriodically,TheoreticalPopulationBiology,72:5266.
Neyman,A.(1985).BoundedComplexityJustifiesCooperationintheFinitelyRepeated
Prisoner'sDilemma,EconomicsLetters,19:227229.
Nowak,MartinA.andMay,RobertM.(1992).EvolutionaryGamesandSpatialChaos,
Nature,359(6398):826829.
Nowak,MartinA.,JoshuaB.Plotkin,andDavidC.Krakauer(1999).TheEvolutionary
LanguageGame,JournalofTheoreticalBiology,200:147162.

Nowak,MartinA.andSigmund,K.(1992).TitForTatinHeterogenousPopulations,Nature,
359:250253.
Nowak,MartinA.andMay,RobertM.(1993).TheSpatialDilemmasofEvolution,
InternationalJournalofBifurcationandChaos,3:3578.
Nowak,MartinA.,SebastianBonhoeffer,andRobertM.May(1994).MoreSpatialGames,
InternationalJournalofBifurcationandChaos,4(1):3356.
Ockenfels,Peter(1993).CooperationinPrisoner'sDilemmaAnEvolutionaryApproach,
EuropeanJournalofPoliticalEconomy,9:567579.
Ostrom,Elinor(2000).CollectiveActionandtheEvolutionofSocialNorms,Journalof
EconomicPerspectives,14(3):137158.
Page,K.M.andM.A.Nowak(2002).Empathyleadstofairness,BulletinofMathematical
Biology,64:11011116.
Pawlowitsch,C.(2007).Finitepopulationschooseanoptimallanguage,Journalof
TheoreticalBiology,249:606616.
(2008).Whyevolutiondoesnotalwaysleadtoanoptimalsignalingsystem,Gamesand
EconomicBehavior,63:203226.
Reijnders,L.(1978).OntheApplicabilityofGameTheorytoEvolution,Journalof
TheoreticalBiology,75(1):245247.
Robles,J.(1998).EvolutionwithChangingMutationRates,JournalofEconomicTheory,79:
207223.
Robson,ArthurJ.(1990).EfficiencyinEvolutionaryGames:Darwin,NashandtheSecret
Handshake,JournalofTheoreticalBiology,144:379396.
Rogers,A.R.(1988).Doesbiologyconstrainculture?AmericanAnthropologist,90:819
831.
Samuelson,LarryandJ.Zhang(1992).EvolutionaryStabilityinAsymmetricGames,Journal
ofEconomicTheory,57:363391.
Samuelson,Larry(1993).DoesEvolutionEliminateDominatedStrategies?inKennethG.
Binmore,A.Kirman,andP.Tani(eds.),FrontiersofGameTheory,Cambridge,MA:MIT
Press,pp.213235.
(1997).EvolutionaryGamesandEquilibriumSelection.(Series:EconomicLearningand
SocialEvolution),Cambridge,Massachusetts:MITPress.
Snchez,AngelandJosA.Cuesta(2005).Altruismmayarisefromindividualselection,
JournalofTheoreticalBiology,235:233240.
Schlag,KarlH.(1998).WhyImitate,andIfSo,How?ABoundedlyRationalApproachto
MultiarmedBandits,JournalofEconomicTheory,78:130156.
Schuster,P.andSigmund,K.(1983).ReplicatorDynamics,JournalofTheoreticalBiology,
100(3):533538.
Selten,Reinhard(1983).EvolutionaryStabilityinExtensiveTwoPersonGames,
MathematicalSocialSciences,5:269363.
(1988).EvolutionaryStabilityinExtensiveTwoPersonGamesCorrectionandFurther
Development,MathematicalSocialSciences,16(3):223266.
Selten,Reinhard(ed.)(1991).GameEquilibriumModelsI:EvolutionandGameDynamics,
NewYork:SpringerVerlag.
Selten,Reinhard(1993).Evolution,Learning,andEconomicBehaviour,Gamesand
EconomicBehaviour,3(1):324.
Sinclair,P.J.N.(1990).TheEconomicsofImitation,ScottishJournalofPoliticalEconomy,
37(2):113144.
Skyrms,Brian(1992).ChaosinGameDynamics,JournalofLogic,Language,and
Information,1:111130.
(1994).ChaosandtheExplanatorySignificanceofEquilibrium:StrangeAttractorsin
EvolutionaryGameDynamics,inProceedingsofthe1992PSA(Volume2),Philosophyof
ScienceAssociation,pp.374394.
(1994a).DarwinMeetsTheLogicofDecision:CorrelationinEvolutionaryGame
Theory,PhilosophyofScience,61:503528.

(1994b).SexandJustice,JournalofPhilosophy,91:305320.
(1996).EvolutionoftheSocialContract,Cambridge:CambridgeUniversityPress.
(1997).GameTheory,RationalityandEvolution,inM.L.DallaChiaraetal.(eds.),
StructuresandNormsinScience,Dordrecht:KluwerAcademicPublishers,pp.7385.
(1998).Salienceandsymmetrybreakingintheevolutionofconvention,Lawand
Philosophy,17:411418.
(1999).PrcisofEvolutionoftheSocialContract,PhilosophyandPhenomenological
Research,59(1):217220.
(2000).GameTheory,RationalityandEvolutionoftheSocialContract,Journalof
ConsciousnessStudies,7(12):269284.
(2000).AdaptiveDynamicModelsandtheSocialContract,JournalofConsciousness
Studies,7(12):335339.
(2004).TheStagHuntandtheEvolutionofSocialStructure,Cambridge:Cambridge
UniversityPress.
(forthcoming).EvolutionofSignalingSystemswithMultipleSendersandReceivers,
PhilosophicalTransactionsoftheRoyalSocietyofLondon(SeriesB).
Smale,Steve(1980).ThePrisoner'sDilemmaandDynamicalSystemsAssociatedtoNon
cooperativeGames,Econometrica,48:16171634.
MaynardSmith,JohnandGeorgePrice(1973).TheLogicofAnimalConflict,Nature,246:
1518.
MaynardSmith,John(1982).EvolutionandtheTheoryofGames.Cambridge:Cambridge
UniversityPress.
Stanley,E.Ann,DanAshlock,andLeighTesfatsion(1994).IteratedPrisoner'sDilemmawith
ChoiceandRefusalofPartners,inChristopherG.Langton(ed.),ArtificialLifeIII
(ProceedingsoftheWorkshoponArtificialLife,heldJune1992inSantaFe,NewMexico),
Reading,MA:AddisonWesley,pp.131175.
Suleiman,RamziandIlanFischer(1996).TheEvolutionofCooperationinaSimulatedInter
GroupConflict,inLiebrandandMessick(eds.),FrontiersinSocialDilemmasResearch,
Berlin:Springer.
Taylor,PeterD.andLeoB.Jonker(1978).EvolutionaryStableStrategiesandGame
Dynamics,MathematicalBiosciences,40:145156.
Thomas,B.(1984).EvolutionaryStability:StatesandStrategies,TheoreticalPopulation
Biology,26:4967.
(1985a).EvolutionaryStableSetsinMixedStrategistModels,TheoreticalPopulation
Biology,28:332341.
(1985b).OnEvolutionaryStableSets,JournalofMathematicalBiology,22:105115.
Tomochi,MasakiandMitsuoKono(1998).SocialEvolutionBasedonPrisoner'sDilemma
withGenerationDependentPayoffMatrices,ResearchonPolicyStudies,3:7991.
Trivers,RobertL.(1971).Theevolutionofreciprocalaltruism,TheQuarterlyReviewof
Biology,46:3557.
Vanderschraaf,Peter(2000).GameTheory,Evolution,andJustice,PhilosophyandPublic
Affairs,28(4):325358.
VegaRedondo,Fernando(1996).Evolution,Games,andEconomicBehaviour,Oxford:Oxford
UniversityPress.
VegaRedondo,Fernando(1997).TheEvolutionofWalrasianBehavior,Econometrica,65
(2):375384.
Wakano,JoeYuichiro,KenichiAokiandMarcusW.Feldman(2004).Evolutionofsocial
learning:amathematicalanalysis,TheoreticalPopulationBiology,66:249258.
Wakano,JoeYuichiroandKenichiAoki(2006).Amixedstrategymodelfortheemergence
andintensificationofsociallearninginaperiodicallychangingnaturalenvironment,
TheoreticalPopulationBiology,70:486497.
Weibull,JuergenW.(1995).EvolutionaryGameTheory,Cambridge,MA:TheMITPress.
Witt,Ulrich(1989a).TheEvolutionofEconomicInstitutionsasaPropagationProcess,
PublicChoice,62(2):155172.

Young,H.Peyton.(1993).AnEvolutionaryModelofBargaining,JournalofEconomic
Theory,59:145168.
(1993).TheEvolutionofConventions,Econometrica,61(1):5784.
(2001).IndividualStrategyandSocialStrategy:AnEvolutionaryTheoryofInstitutions,
Princeton:PrincetonUniversityPress.
Zollman,Kevin(2005).TalkingtoNeighbors:TheEvolutionofRegionalMeaning,
PhilosophyofScience,72:6985.