0% found this document useful (0 votes)
151 views

Game Theory Through Examples

This document provides examples to illustrate concepts from game theory, including: 1) Games against nature involving a single agent making decisions under uncertainty. 2) Non-cooperative games involving two or more autonomous agents with conflicting interests. Examples include the Prisoner's Dilemma and the Battle of the Sexes. 3) Equilibrium concepts in game theory including dominant strategies, Nash equilibria, mixed strategies, and Pareto optimal outcomes. The examples demonstrate how rational self-interested agents may arrive at different outcomes.

Uploaded by

trillion5
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
151 views

Game Theory Through Examples

This document provides examples to illustrate concepts from game theory, including: 1) Games against nature involving a single agent making decisions under uncertainty. 2) Non-cooperative games involving two or more autonomous agents with conflicting interests. Examples include the Prisoner's Dilemma and the Battle of the Sexes. 3) Equilibrium concepts in game theory including dominant strategies, Nash equilibria, mixed strategies, and Pareto optimal outcomes. The examples demonstrate how rational self-interested agents may arrive at different outcomes.

Uploaded by

trillion5
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
You are on page 1/ 4

SYMBOLICSYSTEMS202:TheRationalityDebate(3units)WinterQuarter20032004,Stanford UniversityInstructor:ToddDavies GameTheoryThroughExamples(2/11/04) Gamesagainstnaturedecisiontheoryforasingleagent Expectedutilitytheoryforasingleagentissometimescalledthetheoryof"gamesagainstnature". Considerthisexample. Example1:Planningaparty Ouragentisplanningaparty,andisworriedaboutwhetheritwillrainornot.

Theutilitiesand probabilitiesforeachstateandactioncanberepresentedasfollows: Nature'sstates: Rain (p=1/3) Partyplanner'spossibleactions: Outside 1 Inside 2 TheexpectedutilityofanactionAgivenuncertaintyaboutastateS=Probability(S|A)*Utility(S|A)+ Probability(notS|A)Utility(notS|A)NotethatactionAcanbeviewedasacompoundgambleoroutcome. Also,notethattheprobabilityofastatecandependontheagent'schoiceofaction,although,intheabove example,itdoesnot. Forthepartyproblem:EU(Outside)=(1/3)(1)+(2/3)(3)=2.67;EU(Inside)=(1/3)(2)+(2/3)(2)=2 Therefore,chooseOutside,theactionwiththehigherexpectedutility (Noncooperative)gametheorydecisiontheoryformorethanoneagent,eachactingautonomously(no bindingagreements) Intheexamplesbelow,we'llassumetwoselfutilitymaximizingagents(orplayers),eachofwhomhas completeinformationabouttheoptionsavailabletothemselvesandtheotherplayeraswellastheirown andtheother'spayoffs(utilities)undereachoption. Example2Friendshopingtoseeeachother Considertwopeople,ChrisandKim.Theybothenjoyeachother'scompany,butneithercan communicatewiththeotherbeforedecidingwhethertostayathome(wheretheywouldnotseeeach other)orgotothebeachthisafternoon(wheretheycouldseeeachother).Eachprefersgoingtothe beachtobeingathome,andprefersbeingwiththeotherpersonratherthanbeingapart.Thisgamecanbe representedbythefollowingnormal(ormatrix)form:

Kim Home Chris Home (0,0) Beach (1,0) Eachplayerhasasetofstrategies(={Home,Beach}forbothplayersinthisexample).Specifyingone strategyifortherowplayer(Chris)andonestrategyjforthecolumnplayer(Kim)yieldsanoutcome, whichisrepresentedasapairofpayoffs(Rij,Cij),whereRijistheutilitytherowplayerreceives,andCij istheutilitythecolumnplayerreceives. Inthisexample,goingtothebeachisa(strictly)dominantstrategyforeachplayer,becauseitalways yieldsthebestoutcome,nomatterwhattheotherplayerdoes.Thus,iftheplayersarebothmaximizing theirindividualexpectedutilities,eachwillgotothebeach.SoBeachBeachisadominantstrategy equilibriumforthisgame.Becauseofthis,KimandChris,iftheyarerational,donotneedtocooperate (makeanagreement)aheadoftime.Eachcanjustpursuetheirowninterest,andthebestoutcomewill occurforboth. Example3"Friends"withasymmetricpreferences NowconsiderBettyandJohn.JohnlikesBetty,butBettydoesn'tlikeJohnthatmuch.Eachknowsthis, andneitherwantstocalltheotherbeforedecidingwhattodothisafternoon:stayattheirrespective homesorgototheneighborhoodswimmingpool.Hereisthenormalform: John Home Betty Home (2,0) Pool (3,0) Inthiscase,Betty'sbeststrategydependsonwhatJohndoes.ButifsheassumesJohnisrational,shewill reasonthathewillnotstayhome,becausegoingtothepoolisadominantstrategyforhim.Knowing this,shecandecidetostayhome(because2>1).Thisiscallediterateddominance.Inthisexample, BettygetshigherutilitythanJohnbecauseoftheirrelativepreferences,andJohngetslessutilitythanhe wouldhaveifBettywantedtobewithhim. Inthisexample,PoolHome(3,0),HomePool(2,1),andPoolPool(1,2)areallParetooptimal outcomes.AnoutcomeisParetooptimal(orefficient)ifnoagentcanbemadebetteroffthanthat outcomewithoutmakinganotheragentworseoff.Theequilibriumoutcomesinboththisexampleand thepreviousoneareParetooptimal. Example4Prisoners'dilemma ConsiderStanandLeland,twoprisonerswhohaveeachbeenofferedadealtoturnstate'switness(defect) againsttheother.Theycan'tcommunicate.Theyhadorginallyagreedtoremaininsolidarity,i.e.not testifyagainsteachother,butsincetheagreementcannotbeenforced,eachmustchoosewhethertohonor it.Ifbothremaininsolidarity,thentheywilleachonlybeconvictedofaminorchage.Ifonlyone defects,thenthestatewillthrowthebookattheotherandletthedefectorgo.Iftheybothdefect,each willgetconvictedofaseriouscharge.Thepayoffmatrix(higherpositiveutilityimpliesabetter outcome)isasfollows:

Leland

Solidarity Stan Solidarity (3,3) Defection (4,1) Inthisgame,thestrategyofdefectionisweaklydominantforeachplayer,meaningthatwhateverthe otherplayerdoes,defectingyieldsanoutcomeatleastasgoodandpossiblybetterthanremainingin solidaritywould.Notethatifthebottomrightcellpayoffswere(2,2)insteadof(1,1),thendefecting wouldbestrictlydominantforeachplayer.Eitherway,DefectionDefectionisadominantstrategy equilibrium.However,itisnotParetooptimal.Bothplayerscouldbemadebetteroffifneitherdefected againsttheother. Thisisanexampleofasocialdilemma:asituationinwhicheachagent'sautonomousmaximizationof selfutilityleadstoaninefficientoutcome.Suchasituationcanoccurforanynumberofpeople,notjust two.Anagreementbytwopeopletotradewitheachother(involvinggoods,services,and/ormoney) set'supaprisoners'dilemmatypegamewhenevertheagreementcannotbeenforced. Example5Coordination Let'sgobacktoChrisandKim.Theyaregoingtothesameconference,andeachisexpectingtheother tobethere,buttheyhaven'tseeneachotheryet.Theconfereeshavetheirchoiceoftwoactivitiesonthe firstafternoon:swimmingorhiking.Theybothhopetoseeeachotheriftheydon'ttheywillhaveno fun,andeachprefersswimmingoverhiking.Theymusteachdecidewhattodobeforeknowingwhere theotherisgoing.Hereisthenormalform: Kim Swim Chris Swim (2,2) Hike (0,0) ThebestoutcomeisobviouslySwimSwim,butgoingswimmingisnotdominantforeitherplayer.Both SwimSwimandHikeHikehavethepropertythateachplayer'sstrategyisthebest(ortiedforthebest) responsetotheotherplayer'sstrategyinthatpairing.Thisdefinesamoregeneralequilibriumnotion calledtheNashequilibrium.Thedominanceequilibriaofexamples13areallNashequilibriaaswell. Athirdequilibriumexistsinthisgameinvolvingwhatarecalledmixedstrategies.Amixedstrategyisa probabilitydistributionoverthepurestrategies(whichareSwimandHikeforeachplayerinthis example).(Notethattheplayersdonothavetohavethesamesetofstrategiesavailabletothem,even thoughthathasbeenthecaseinallourexamples.)Inthisexample,ifeachplayerindividuallythrowsa dieandgoesswimmingifthediecomesup1or2,andgoeshikingifthediecomesup3,4,5,or6,the resultingexpectedutility(2/3foreachplayer)cannotbeimproveduponforeitherplayergiventhatthe otherplayerusesthisstrategy. In1950,JohnNash(depictedsomewhatfictitiouslyinthefilmABeautifulMindthebookismore accurate!)provedthateveryfinitegame,involvinganynumberofplayers,hasatleastone(Nash) equilibrium,thoughtheremightnotbeanythatinvolveonlypurestrategiesforallplayers.Inthis example,therearethreeequilibria:themixedstrategyequilibrium(Swim,1/3;Hike,2/3) (Swim,1/3;Hike,2/3),andtwopurestrategyequilibriaSwimSwimandHikeHike.

Whenthereismorethanoneequilibrium,andplayerscannotmakebindingagreements,theymusttryto coordinatetoarriveatanequilibriumoutcome.WhenonlyoneequilibriumisalsoParetooptimal,as SwimSwimisinthiscase,thatfactshouldsuggesttorationalplayersthatitwillbetheonearoundwhich theycoordinate.Manyothercriteriaforequilibriumselectionhavebeenstudied(e.g.focalpoints, subgameperfection,stabilityseethereadingongametheory). Example6"Battleofthesexes" Finally,let'sconsiderRoyandJen.TheyaregoingtothesameconferenceasKimandChrisinexample 5.Theyeachwouldprefertobeinthesameplace(theswimorthehike),buttheirpreferencesdiffer aboutwhichitshouldbe.Roywouldrathergoswimming,andJenwouldrathergohiking.Hereisthe matrixform: Jen Swim Roy Swim (3,2) Hike (1,1) ThisgamehasthreeNashequilibria:SwimSwim,HikeHike,and(Swim,2/3;Hike,1/3) (Swim,1/3;Hike,2/3).Notethatthemixedstrategiesdifferforeachplayerinthethirdequilibrium:each goestotheirpreferredactivitywith2/3probability. AlloftheequilibriaareParetooptimalthistime,sothatdoesnothelpforselection.Onlythemixed strategyequilibriumresultsinequalexpectedutilitiesforthetwoplayers,soifbothvalueequalityor symmetry,thismightbethefocalpoint.ButofcourseitwillbedifficultforRoyandJentoseethat unlesstheyhavestudiedgametheory!

You might also like