Game Theory of Mind 英文参考文献.docVIP

GameTheoryofMind WakoYoshida*,RayJ.Dolan,KarlJ.Friston TheWellcomeTrustCentreforNeuroimaging,UniversityCollegeLondon,UnitedKingdom Abstract Thispaperintroducesamodelof‘theoryofmind’,namely,howwerepresenttheintentionsandgoalsofotherstooptimise ourmutualinteractions.Wedrawonideasfromoptimumcontrolandgametheorytoprovidea‘gametheoryofmind’. First,weconsidertherepresentationsofgoalsintermsofvaluefunctionsthatareprescribedbyutilityorrewards.Critically, thejointvaluefunctionsandensuingbehaviourareoptimisedrecursively,undertheassumptionthatIrepresentyourvalue function,yourrepresentationofmine,yourrepresentationofmyrepresentationofyours,andsoonadinfinitum.However, ifweassumethatthedegreeofrecursionisbounded,thenplayersneedtoestimatetheopponent’sdegreeofrecursion (i.e., sophistication) to respond optimally. This induces a problem of inferring the opponent’s sophistication, given behaviouralexchanges.Weshowitispossibletodeducewhetherplayersmakeinferencesabouteachotherandquantify theirsophisticationonthebasisofchoicesinsequentialgames.Thisrestsoncomparinggenerativemodelsofchoiceswith, and without, inference. Model comparison is demonstratedusing simulated and real datafrom a‘stag-hunt’. Finally, we note that exactly the same sophisticated behaviour can be achieved by optimising the utility function itself (through prosocialutility),producingunsophisticatedbutapparentlyaltruisticagents.Thismayberelevantethologicallyinhierarchal gametheoryandcoevolution. Citation:YoshidaW,DolanRJ,FristonKJ(2008)GameTheoryofMind.PLoSComputBiol4(12):e1000254.doi:10.1371/journal.pcbi.1000254 Editor:TimBehrens,JohnRadcliffeHospital,UnitedKingdom ReceivedJuly2,2008;AcceptedNovember13,2008;PublishedDecember26,2008 Copyright: ? 2008 Yoshida et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalauthorandsourcearecredited. Funding:ThisworkwassupportedbyWellcom