Resolving dual-task interference an fMRI study

1Resolvingdual-taskinterference:AnfMRIstudyYuhongJiangHarvardUniversityCorrespondenceshouldbeaddressedtoYuhongJiangDepartmentofPsychologyHarvardUniversity33KirklandStreet,Room820Cambridge,MA02138E-mail:[e-mail]:(617)-496-4486Fax:[phone](“Room820”)Acknowledgement.ThisresearchissupportedbyaHelenHayWhitneyResearchFellowshipandaMiltonFundtoY.J,andaHumanFrontiersGranttoNancyKanwisher.ItisalsosupportedinpartbytheNationalCenterforResearchResources(P41RR14075)andtheMentalIllnessandNeuroscienceDiscovery(MIND)Institute.IthankNancyKanwisherforherintellectualgenerosityandRebeccaSaxeandLauraC.Wagnerforcomments.2AbstractThehumancognitivesystemisseverelylimitedintheamountofinformationitcanprocesssimultaneously.Whentwotasksarepresentedwithinashortstimulus-onset-asynchrony(SOA),reactiontimeofeachtask,especiallytask2,isdramaticallydelayed.Previousstudieshaveshownthatsuchdelayisaccompaniedbyincreasedactivationinrightinferiorfrontalgyrus(GFi).Inthisstudy,weaddresstheroleofrightGFiinresolvingdual-taskinterferenceattwodifferentstages:allocationofperceptualattentionandresponseselection.Wescan12subjectsusingfunctionalMRIwhiletheyconducttwotasks–shapediscriminationintask1andcolordiscriminationintask2–andvarytheSOAbetweentasksas100msor1500ms.Thetargetsarelocatedatthecenterorattheperiphery.Whenbothareatthecenter,theycompeteprimarilyforresponseselection.Whenbothareattheperiphery,theyadditionallycompetefortheallocationofperceptualattention.ResultsshowthattherightGFiandfrontaloperculumregionsaresignificantlymoreactiveintheshortSOAthanthelongSOAcondition,butonlywhensubjectsattendtotheperipheryinbothtasks.Weconcludethattherightlateralfrontalregionsareimportantforresolvingdual-taskinterferenceattheperceptualattentionstage.3IntroductionOneofthemostfascinatingaspectsofhumancognitionisourineffectivenessinconductingtwotasksatthesametime.Despitefrequentwishestodivideattentionbetweentwotasks–suchasturningleftintrafficwhilecarryingoutacoherentconversation–mostofuschoosetostopperformingonetaskuntiltheotheriscomplete.Butwhatpreventsonefromrespondingtothetrafficlightatthesametimeaschoosingthenextsentencetosay?Researchincognitivepsychologyhasprovidedextensiveevidencethathumanattentionislimitedinatleasttworespects:perceptualattentionandresponseselection(Jiang&Kanwisher,2003a).Perceptualattentionisusedprimarilytoselecttargetsandfilteroutdistractorswhenmultipleobjectsarepresented.Itcanbesimultaneouslydirectedtoapproximatelyfourvisualobjects.Forexample,inmultiple-objecttracking,subjectsfirstsee10dots,afewofthemblinkedinitially,andthenalldotsmoverandomlyonthescreen.Subjects’abilitytotrackthepreviouslyblinkeddotsisseverelyimpairedifmorethan4dotsaretobetracked(Pylyshyn&Storm,1988).Responseselectionalsoposesacognitivelimitation.Pashlerandcolleaguesproposethatthereisacentralcognitivebottleneck,involvedwheneverastimulushastobemappedontoaresponseonthebasisofanarbitraryrule,suchasdecidingtosteponthebrakewhenthetrafficlightturnsred(Pashler,1984,1994).Mappingastimulustoaresponse–adecisionprocess–precedestheexecutionofmotorresponses.Thus,whilewecanpressakeyandsayawordatthesametime(Pashler,1993),wecannotsimultaneouslydecidewhichkeytopressonthebasisofashapeanddecidewhichwordtosayonthebasisofatone.4Responseselectionisnotonlydifferentfrommotorexecution,butalsoseparatefromtheallocationofperceptualattention.Thisisrevealedinthatthecentralbottleneckcannotbedividedbetweentworesponseselectionprocesses,butperceptualattentioncanbedividedbetweenafewperceptualobjects.Thus,whenpeopleencounteracompetitionforresponseselection,theyholdthesecondtaskinaqueue;butwhentheyencounteracompetitionforperceptualattention,theydivideaproportionofperceptualattentiontoeachobject.Therefore,perceptualattentioncanbesimultaneouslydividedbetweenuptofourperceptualobjects(LuckandVogel,1997;Pylyshyn&Storm,1988),butresponseselectioncanonlybesequentiallyallocatedtoonetaskatatime(Pashler,1989,1991).Inthisstudy,weusefunctionalMRItoinvestigatehowthebrainresolvesdual-taskinterferenceatthestageofperceptualattentionandthestageofresponseselection.Competitionatathirdstage–themotorresponsestage–hasbeenaddressedbyaseparatestudybyHerathetal.(2001)andwillbediscussedlater.Weusethe“overlappingtask”design,adoptedfrombehavioralstudiesonthecentralbottleneck(Welford,1952;Pashler,1984).Inthisdesign,twotasksarepresented,separatedbyavariablestimulus-onset-asynchrony(SOA).WhentheSOAisshort(e.g.,100ms),thetwotasksoverlapintimeandcompeteforperceptualattentionandresponseselection.Inthiscase,subjectshavetoperformtask1ANDtask2.AstheSOAincreases(e.g.,to1500ms),thetwotasksnolongeroverlap,sosubjectsonlyneedtoperformtask1ORtask2.ThedifferencebetweenashortSOAcondition(“AND”)andalongSOAcondition(“OR”)reflectshowthebraincopeswithoverlappingstagesofprocessingbetweentwodifferenttasks.5Severalpreviousneuroimagingstudieshavestudiedtheeffectofdual-taskprocessing(e.g.,Adcocketal.,2000;Bungeetal.,2000;D’Espositoetal.,1995;Klingberg,1998;Koechlinetal.,1999).Thesestudiescompareblocksofconcurrentprocessingoftwotasks(e.g.,sen