The Spatiotemporal Dynamics of Autobiographical Memory Neural Correlates of Recall Emotional - PDF document

TheSpatiotemporalDynamicsofAutobiographicalMemory:NeuralCorrelatesofRecall,EmotionalIntensity,andRelivingSanderM.Daselaar,HeatherJ.Rice,DanielL.GreenbergRobertoCabeza,KevinS.LaBarandDavidC.RubinDepartmentofPsychologyandNeuroscience,DukeUniversity,Durham,NC27708,USACurrentaddress:SwammerdamInstituteforLifeSciences,UniversityofAmsterdam,Amsterdam,theNetherlandsCurrentaddress:DepartmentofPsychology,UCLA,LosAngeles,CA90095,USA periodsandsuchattemptsextendretrievalmode(Tulving2002;Velanovaetal.2003),whichisrightlateralized.TheSpatiotemporalDynamicsofAutobiographicalMemoryRetrievalMuchlessisknownaboutthetimecoursesofretrieval-relatedactivitythanaboutitslocation,althoughageneralanterior-to-posteriortrendexistsinelectrophysiologicaldata.Slowcorticalpotentialsduringtheretrievalandmaintenanceofautobio-graphicalmemoriesexhibitashiftfromprimarilyleftfrontalregionstoposteriortemporalandoccipitalregions(Conwayetal.2001,2003).Itisdif“culttodistinguishthetimecoursesoffMRIactivityindifferentmemory-relatedbrainregionsduringmanyepisodicmemorytasksbecausetheretrievalprocessoccurssoquickly.However,autobiographicalmemoriestakemuchlongertore-trievethanmemoriesforitemslearnedinthelaboratory.Inordertohaveenoughtimetoseparatetheprocessesofaccessorretrievalfromthoseofelaboratingandmaintainingmemories,andtoensurevariabilityinemotionalintensity,inthepresentstudy,autobiographicalmemorieswerecuedusingtheGalton/Crovitztechnique(Galton1879;CrovitzandShiffman1974;forreviewsseeRubin1982;RubinandSchulkind1997).Partic-ipantswereaskedtoprovideamemoryofaspeci“cpasteventforeachofaseriesofgenericcuewords(e.g.,tree).Incontrasttoprocedureswhereindividualscreatetheirownpersonallytailoredcuewords(Maguire2001;Greenbergetal.2005),theretrievalprocessusingtheGaltontechniquetakessigni“cantlylonger(typically10vs.3s)becausethememoriesarenotrehearsedwiththecuespriortomagneticresonanceimagingBymostbehavioral,neural,andcomputationalaccounts,retrievinganepisodicmemoryfromacueinordertoperformataskrequiresthatonememorybeselectedforretrievalandothercompetingmemoriesbeinhibited(i.e.,onesetofneuralactivationsdominates).Itisatthispointthattherecoveryofamemoryoccurs,accordingtoTulving(1983).Next,theselectedmemorymustbemaintainedandpossiblyelaboratedlongenoughtoperformtherequiredtask.Here,wecalltheearlierprocessesof“ndingonememorymemoryaccesstoavoidotherterms,whichentailstrongertheoreticalclaims.Wecallthelaterprocesseselaborationbecauseourparticipantswereaskedtomaintainandelaboratetheirmemories.Opera-tionally,wetakethepointwhenourparticipantsindicatetheyhaveaccessedamemory,viaabuttonpress,asthedividingpointbetweenprocessesthataremostlyinvolvedinaccessingthememoryandthosethataremostlyinvolvedinmaintenanceandelaboration.Thus,bycomparingactivitybeforeandafterthebuttonpressdirectlytoeachother,wecandistinguishbrainregionsprimarilyassociatedwithmemoryaccessversuselabo-ration.Wecheckthevalueofthisoperationalde“nitionbyexaminingtheactivityoffMRIsignalswithinregionsofinterest(ROIs)toseeiftheyfollowtheexpectedtimecourse.Accordingtocurrentmemorymodels,memoryaccessincludesthereactivationofdistributed,storedmemorytraces,aswellasthestrategicoperationssustainingretrievalsearch(Tulving1983;ConwayandPleydell-Pearce2000;RuggandWilding2000;Moscovitchetal.2005;Rubin2005).ThecandidateregionsfortheseprocessesaretheMTLandrightPFC.MTLishypothesizedtobecriticallyinvolvednotonlyintheformationbutalsointhereactivationofmemorytraces(Moscovitch1995;Eichenbaum2004;Squireetal.2004).RightPFChasbeenlinkedtoretrievaleffort(Kapuretal.1995;Wagner,Desmond,etal.1998)andtothementalsetthatguidestheretrievalofepisodicinformation„knownasretrievalmode(Velanovaetal.2003).Weexpectedthatfurtherelaborationofamemoryinvolvesimageryevokedbytheretrievedepisode,especiallyregionsinthevisualsystem(GreenbergandRubin2003;Rubin2006),aswellascontrolledattentionalandworkingmemoryoperationsrelatedtoselectingandkeepingretrievedinformationactiveinmind.Accordingly,wepredictedthattheperiodafteramemorywasinitiallyformedwouldbeassociatedwithactivityinvisualcortexregions,aswellastheprecuneus„aregionassociatedwithvisualimageryduringepisodicretrieval(Fletcheretal.1995).Inaddition,wepredictedactivityrelatedtocontrollingtheelaborationprocess,selectingrelevantinformation,andmaintainingthisinformationonline.Inparticular,theleftPFChasbeenassociatedwithcognitivecontrol(Wagneretal.2001;Buckner2003),top--downattentionalselectionofvisualin-formation(Hop“ngeretal.2000),andworkingmemorymaintenanceoperations(e.g.,WagerandSmith2003).TheSpatiotemporalDynamicsofRelivingandEmotionalIntensityTwophenomenologicalpropertieshaveplayedakeyroleindistinguishingepisodicmemoryretrievalfromotherformsofmemoryretrieval.The“rstpropertyistheextenttowhichamemoryisrecollected(relivedorreexperienced)asopposedtofeelingmerelyfamiliar(Tulving1983;Yonelinas2001;Dobbinsetal.2002;YonelinasandLevy2002).Recollectionisanecessary,de“ningfeatureofepisodicmemory,whichisoftenexpressedintermsoftravelingbackintimetorelivetheevent(Tulving2002).Whenrecallingautobiographicalmemoriesforspeci“ceventsratherthansemanticknowledge,itisthereforeassumedthatsomedegreeofrecollectionisalwaysinvolved.Todistinguishbetweendegreesofreexperiencinganevent,weevaluatedratingsofrelivingasacontinuousphenomenologicalpropertyofrecollection.Theabilitymentallytotimetraveltoreexperiencepasteventsisthoughttodependonfrontallobestructures(Wheeleretal.1997).Furthermore,activityinposteriorcortices,especiallyvisualareas(Kosslynetal.2001),shouldbeinvolvedinmemoriesassociatedwithahighdegreeofsubjectivereliving,asbehavioralstudiesshowthatthebestpredictorofthedegreeofexperiencedrelivingisvisualimagery(Rubinetal.2003;RubinandSiegler2004;Rubin2006).Asecondphenomenologicalpropertythatisimportantintheretrievalofpersonalmemoriesisemotionalintensity(Talaricoetal.2004).Akeyquestionhereishowtheneuralstructuressupportingemotionmodulatememoryretrieval.Severalneuro-imagingstudieshaveshownthattheextentofamygdalaactivationduringtheinitialencodingofemotionalpicturesor“lmclipspredictssubsequentretention(Cahilletal.1996;Canliandothers2000;KensingerandCorkin2004),whichisassociatedwithfunctionalinteractionsbetweentheamygdalaandthehippocampusduringencoding(Hamannetal.1999;Dolcosetal.2004).However,muchlessisknownabouttheamygdalasroleinmemoryretrieval.Someepisodicorautobio-graphicalmemorystudieshavereportedamygdalaactivationinhealthysubjects(Finketal.1996;Markowitschetal.2000,2003;MaguireandFrith2003;Greenbergetal.2005),butsomehavenot,evenwhenemotionswerespeci“callyprobed(Damasioetal.2000;Piefkeetal.2003;Keedwelletal.2005).Dolcosetal.(2005)foundgreateractivityinandinteractionsbetweenthe218SpatiotemporalDynamicsofAutobiographicalMemoryDaselaaretal. amygdalaandhippocampusforemotionalthanforneutralmemoryretrieval,especiallyforitemswhosememorywasaccompaniedbyasenseofrecollection(seealso,Sharotetal.2004).Addisetal.(2004)showedthatthehippocampusandamygdalaweremodulatedduringretrievalbyratingsofemo-tionalintensity.However,inthisstudy,theamygdalaeffectsweresubthresholdwhenconsideringautobiographicalmemo-riesthatoccurredonlyonce(asopposedtorepeatedevents).Thequestionofwhen,duringtheprocessofretrievingandmaintaininganepisodicmemory,relivingandemotionhavetheireffectsismuchlesswellunderstood.Behavioralstudieshaveindicatedthatthese2phenomenologicalaspectsofautobio-graphicalmemoriestendtobecorrelated(Talaricoetal.2004),buttheymaynotariseatthesametimenorinthesamebrainregionsduringrecall.Becausejudgmentsofrelivingaremeta-cognitivejudgments(Rubinetal.2003)thatoccurbehaviorallyafterthememoryhasbeenfullyformedbasedonthequalityofinformationretrieved,weexpectthedegreetowhichamemoryisreexperiencedtobeevidentinthebrainonlyafterthememoryisretrievedandbeingheldinmindtobejudged.Foremotionalintensity,2contrastingtheoreticalpredictionsexist.Similartotheargumentforreliving,subjectiveratingsofemotionalintensitymaybemadeonlyaboutafullyformedmemory,asemotionisanemergentpropertyofthatmemory.Thisargumentwouldfollowfromappraisalviewsofemotion(Lazarus1991),ifemotionalintensitywerenotaninherentpartofthememorybutajudgmentmadeonamemoryafteritwasrecovered(James1890).Thecontrastingview,whichhasmoresupportfromneurallybasedstudies,isthatemotionisanearlywarningsystem,awayofpreparingtheorganismforactionthatisgenerallyappropriateforthesituationathandevenbeforethesituationisfullycomprehended(LeventhalandScherer1987;LeDoux1996).Ifonepurposeofepisodicmemoryistoprovideadetailedrecordofeventsthatoccurundersimilarcuingsituationssoanorganismcanactappropriately,thenactivityinemotionalareasofthebraincouldoccurveryearlyintheretrievalprocesstosignalthatarelevantmemoryisbeingretrievedbeforethefullcomplexityofthememoryisassembledandinconsciousness.Bartlettpostulatedasimilarearlyroleofemotionduringtheactofremembering:Whenasubjectisbeingaskedtoremember,veryoftenthe“rstthingthatemergesissomethingofthenatureofanattitude.Therecallisthenaconstruction,madelargelyonthebasisofthisattitude,anditsgeneraleffectisthatofajusti“cationoftheattitude,whereforBartlettattitudeisverylargelyamatteroffeeling,oraffect.(Bartlett1932/1995).Thus,althoughlittleisknowndirectlyaboutthetimecourseofthephenomenologicalpropertiesassociatedwiththere-trievalofepisodicmemories,theexistingliteratureallowedustoformulateclearhypothesesaboutwhen,andinbroadtermswhere,theeffectsofthedegreeofrelivingandofemotionalintensityshouldoccur.Becausemetacognitivejudgmentsofrelivingdependonthereexperiencingofamemory,theyshouldoccurafterthebuttonpressthatindicatesthatthememoryisretrieved,andtheyshouldincludeactivityinfrontalareasrelatedtorecollectionandself-referenceandvisualareasneededtosupportrelivingjudgments.Thebest-supportedhypothesisfortheeffectofemotionalintensityisthatitwilloccurintheamygdalaandhippocampusasthememoryisbeingformedduringtheretrievalstage.However,thereisalsothepossibility,thoughlesswellsupported,oflateeffectsofemotionmodulationduringmemorymaintenanceandelaboration.Muchisknownaboutthelocalizationofthenetworkinvolvedinautobiographicalmemoryretrieval.Here,weaddtothisexistingknowledgebyexaminingthetimecourseofactivationinthatnetworkandhowitismodulatedbythesenseofrelivingandemotionalintensityoftheresultingmemories.WeexpectedmoreMTLandrightPFCactivitywhenthememorywasbeingaccessedandconstructed(i.e.,beforethekeypress),andmorevisualcortex,precuneus,andleftPFCactivityduringthesubsequentelaborationandmaintenanceofthememory(i.e.,afterthekeypress).Weexpectthatthedegreeofrelivingwillmodulatefrontalareasrelatedtorecollectionandself-referenceaswellasvisualareasandwilldosoafterthekeypress,andthatemotionalintensitywillmodulatetheamygdalabeforethekeypress,thoughthereissomesupportforeffectsafterthekeypress.MaterialsandMethodsParticipantsTheresultsof17healthyyoungadults(10females;agerange18--35)wereincludedintheanalyses.Allparticipantswererighthandedandwerenegativeforhistoriesofpsychiatricillness,neurologicaldisorder,anddrugabuse.TheInstitutionalReviewBoardatDukeUniversityMedicalCenterapprovedtheprotocolforethicalconsiderations.Participantsgavewritteninformedconsentpriortoparticipation.MaterialsWordcueswerechosenfromadatabaseof125wordcuespreviouslyshowntoelicitautobiographicalmemories(Rubin1980).The80wordcuesthatmostconsistentlyledtotheretrievalofanautobiographicalmemoryin24additionalparticipantsinabehavioralpilotstudywereselected.These80wordscomprisedthe“naltestlistusedinthescanner.Tenadditionalwordswerechosenfromthedatabasetocreateapracticelist.Auditorywordcueswererecorded.ProcedureThebasicprocedurewasadaptedfromGreenbergetal.(2005).Becauseoftheimportanceofvisualimageryinautobiographicalmemory,participantswerescannedwitheyesclosed,andallinstructionsweregivenauditorily(includingpresentationofthememorycuesandratingscales),sothatanypotentialeffectsofvisualimagerywerenotconfoundedbyrequirementstovisuallyattendtoexternaltask-relevantstimuli.Participantswerepresentedwith80auditorysingle-wordcues,viaheadphones,whileinthescanner.Withineachof8runs,10cuewordswerepresentedinapseudorandomorder,eachforadurationrangingfrom0.65to1.25s.Twenty-foursecondsfollowingstimulusonset,participantsheardaninstructiontoratetheemotionalintensity(thewordsrateemotion)andaninstructiontoratethedegreetowhichtheyfelttheywererelivingtheevents(thewordsratereliving)on4-pointscales(1low,4high).Theseinstructionswerepresentedinapseudorandomorder4.5safteroneanotherandwerefollowedbya4.5-or6-sbreakbetweentrials,resultinginsingle-trialdurationsof28.5and30s.Participantsmadeallresponsesviaabuttonboxheldintheirrighthand.Participantscompletedapracticerunduringacquisi-tionofanatomicalscanstoensuretaskcomprehension.Totalscantimeforeachsession,includingacquisitionofanatomicalscans,wasapproximately70min.Participantsweregiventhefollowinginstructions.First,afterhearingacueword,theyshouldthinkofamemoryofaspeci“ceventfromtheirliferelatedtothatword.Memoriesdidnothavetobedirectlyrelatedtothecueword;eventstangentiallyrelatedorretrievedthroughfreeassociationwiththewordwereconsideredvalid.Whentheyrecalledaspeci“cevent,theyshouldpushabuttonontheresponseboxandcontinuethinkingaboutthememoryfortherestofthetrial.Theyshouldnotpresstheresponsebuttonuntiltheyfelttheyhadfullyretrievedamemorybutadditionalinformationcouldcomeafterthebuttonpress.Second,whentheyheardthecuerateemotion,theyshouldratetheemotionalintensityofthememoryusingthe4buttonsCerebralCortexJanuary2008,V18N1219 ontheresponsebox.Theyshouldratethecurrentintensity,ratherthanratinghowintensetheemotionwasduringtheinitialevent.Third,whentheyheardthecueratereliving,theyshouldratehowmuchtheyfeeltheyarerelivingtheinitialeventagainusingthe4buttonsontheresponsebox.Fourth,participantswereinstructedtokeeptheireyesclosedforthedurationofeachrun.Attheendofeachrun,participantsheardaninstructiontorestandtoopentheireyes.ScanningParametersMagneticresonanceimageswereacquiredwitha1.5-TGeneralElectricSignaNviscanner(Milwaukee,WI)equippedwith41mT/mgradients.Theparticipantsheadwasimmobilizedusingavacuumcushionandtape.Theanteriorcommissure(AC)andposteriorcommissure(PC)wereidenti“edinthemidsagittalsliceofalocalizerseries.Thirty-fourcontiguoussliceswereprescribedparalleltotheAC--PCplaneforhigh--weightedstructuralimages(repetitiontime[TR]450ms,echotime[TE]20ms,“eldofview[FOV]24cm,matrix,slice3.75mm).Inversespiralimagessensitivetobloodoxygenlevel--dependentcontrastweresubsequentlyacquiredusingthesamesliceprescriptionasthe-weightedstructuralimages(TR1.5s,TE35ms,FOV24cm,matrix,”ipangle,slicethicknessmm,therebyproducing3.75mmisotropicvoxels).PreprocessingandDataAnalysisStatisticalParametricMappingsoftwareimplementedinMatlab(SPM2;WellcomeDepartmentofCognitiveNeurology,London,UK)wasused.Afterdiscardinginitialvolumestoallowforscannerstabilization,imageswereslice-timingcorrectedandmotion-corrected,thenspatiallynormal-izedtotheMontrealNeurologicalInstitutetemplate,andspatiallysmoothedusingaGaussiankernelof8mmfull-widthhalf-maximum.Finally,treatingthevolumesasatimeseries,thedataweretemporallysmoothedusing3-point“lteringandcubicdetrendingoptionsin-tegratedinMatlab6.5.StandardGLMAnalysisToaccountforthefactthatweusedaself-pacedparadigminwhichparticipantsindicatedbythemselveswhentheyrecalledaspeci“cevent,weimplementeda”exiblefMRIdesigninthecontextofthegenerallinearmodel(GLM).Thedesigndistinguished6componentsineachtrial„4transientand2sustainedcomponents.Transientcompo-nentsincludedthememorycue(immediatelyatonsetofthetrial),response-relateddecisionprocesses(750msbeforetheresponseindicatingamemorywasrecalled),andthe2ratings(secondandthirdresponse).Sustainedperiodsincludedthememoryaccessperiod(fromtrialonsettoresponse)andtheelaborationperiod(fromresponsetothe“rstrating).Thetransientperiodsweremodeledbyconvolvingacanonicalhemodynamicresponsefunctionwithavectorrepresentingperiodonsets,whereasthesustainedperiodsweremodeledwithaboxcarfunctionrepresentingbothperiodonsetsandoffsets(seeFig.1).Inordertoaccountfordifferencesinthetimingofactivationsduetotheself-paceddesign,theresponseindicatingamemorywasaccesseddeterminedthedurationofthememoryaccessperiodaswellastheonsetsoftheresponseandelaborationperiods.Toidentifyactivationdifferencesbetweenthememoryaccessandelaborationstages,theparameterestimates(betaweights)forthememoryaccessandelaborationregressorsweredirectlycomparedusingpairedatanuncorrectedthresholdof0.001andaminimumclustersizeof3adjacentvoxels.ParametricGLMAnalysis:RatingScalesModulationWithintheaforementionedGLManalysis,eachofthetransientandsustainedtrialcomponentswereadditionallymodulatedparametricallybythe2ratings,relivingandemotionalintensity,usingthe“rst-orderparametricmodulationoptionintegratedinSPM2.Toidentifybrainregionsthatshowmodulationofmemoryaccessorelaborationactivityassociatedwiththe2ratings,weappliedrandomeffectsanalysesontheparameterestimatesoftheparametricregressorsusingthesamethresholdasforthestandardGLManalysis(0.001,clustersizeTimeCourseAnalysis:Cue-lockedAnalysisToassessthevalidityofour”exiblefMRIdesign,weinvestigatedthetimecoursesforfastandslowresponseswithrespecttotheself-pacedbuttonpressindicatingthatamemorywasformed.Asa“rststep,weselectedasubgroupofparticipantsshowinganadequaterangeoffastandslowresponses.Employingaminimumcutoffof10trialswitha3-srangeintheresponsetimes,wefoundagroupof8participantsthathadsuf“cienttrialnumberstoallowadirectcomparisonbetweenbinsoffast(9--12s)andslow(12.01--15s)responses.Second,aGLMwascreatedinwhichtrialonsetstimelockedtothecueweremodeledwitha“niteimpulseresponsebasissetofperistimulustimebinsof1.5sduration(equaltotheTR).Theresultingparameterestimatesweresubsequentlyaveragedforeachperistimulustimebin,yieldingestimatesoffMRIsignalchangeacrossthewhole33strialperiodforbothfastandslowbins.Finally,fortheactualvalidation,weusedaROIapproach,focusingonperceptual/motorregions.Theclustersshowingsigni“cant0.001,clustersize3)cue-relatedactivityintheauditorycortexorresponse-relatedactivityinthemotorcortexbasedontheGLManalysiswerede“nedasROIs.FollowingthefMRImodeldepictedinFigure1wepredictedthatcue-relatedactivityshouldnotbemodulatedbyresponsetime,whereasthepeakofresponse-relatedactivityshouldshiftdependingonwhetheritwasafastoraslowresponse.TimeCourseAnalysis:Response-lockedAnalysisTofurtherevaluatetheresultsfromthestandardGLManalysis,weconductedasecondtimecourseanalysisusingthesameproceduresasforthecue-lockedanalysis,but,inthiscase,timelockedtotheresponseandincludinganadditionalperiod4.5sbeforethebuttonpressthatindicatedamemorywasaccessed.Thus,forthisparticularanalysis,onlytrialsinwhichthebuttonpressoccurredwithin4.5swereexcluded,sothatvirtuallyallthetrialsfromall17participantscouldbeincludedintheresponse-lockedanalysis.BasedonthemodeldepictedinFigure1wepredictedthatattheonsetoftheresponsecomponent,thetimecoursesofregionsassociatedwithmemoryaccessshouldbeonadownwardslope,whereasregionsassociatedwithelaborationshouldbeonanupwardslope.TimeCourseAnalysis:RatingScalesModulationTovalidatetheparametricmodulationanalysis,weconducted4additionaltimecourseanalysestimelockedtothecue.Foreachindividualsubject,weusedamediansplitoftheemotionalandrelivingratingsinordertogenerate4separatefMRItimecourses:highandlowemotionandhighandlowreliving.BehavioralDataParticipantswereabletorecallaneventmatchingthecueon90.5%(standarddeviation[SD]5.3)ofthetrials.Theaverageratingsofrelivingandemotionalintensityonascaleof1--4were2.50(SD0.43)and2.28(SD0.40),respectively.Themeanreactiontimefromthecuetothebuttonpressindicatingtheformationofamemorywas12.25s(SD3.68).Participantswerefasterformemoriesthatwerelaterratedashighinrelivingandhighinemotionalintensityasmeasuredusingamediansplitoftheratingsforeachparticipant,adivisionweuselaterinthefMRIanalyses.Themeantimeforhighandlowrelivingwere11.12s(SD0.95)versus10.32s(SD1.23);foremotionalintensity,theywere11.12s(SD0.89)versus10.14s(SD1.44).Thereactiontimesforthe“rstandsecondratingsoneachtrialwere2.24s(SD0.18)and2.53s(SD0.22),respectively.ModelValidation:Cue-lockedfMRITimeCourseAnalysisToassessthevalidityofour”exiblefMRIdesign,weperformedacue-lockedtimecourseanalysisinwhichfast(9--12s)andslow(12.01--15s)responsesforthebuttonpress(indicatingamemorywasformed)weremodeledseparately.BasedonthemodeldepictedinFigure1,wepredictedthatcue-relatedactivity(redline)intheauditorycortexshouldnotbemod-ulatedbyresponsetime,whereasthepeakofresponse-related220SpatiotemporalDynamicsofAutobiographicalMemoryDaselaaretal. activity(greenline)inthemotorcortexshouldshiftdependingonwhetheritwasafastoraslowresponse.Theresultscon“rmedourpredictions.AsshowninFigure1,thetimecoursesforfastandslowresponseswerealmostidenticalintheauditorycortex(redlines)butverydifferentinthemotorcortex(greenlines).Inparticular,therewasalargedifferenceinthepeaklatencycorrespondingto2TRs.Thislatencydifferencematchesthe3-sreactiontimedifferencethatexistedbetweenthefastandslowresponsebins.Yet,atthetimethattheratingcuesarepresented(24sfollowingtrialonset),thetimecoursesinauditoryandmotorcorticesjointogetherandshowadoublebumpre”ectingthe2auditorycuesandcorrespondingbuttonpresses.ThecorrespondenceofthetimecoursesinFigure1correspondswelltothemodeldepictedinFigure1,thusvalidatingourapproach.TimeCourseAnalysis:Response-lockedfMRIAnalysisTable1lists,andFigure2displays,regionsinvolvedinmemoryaccess--relatedandelaboration-relatedactivity.Consistentwithourpredictions,thememoryaccessperiodelicitedactivationsinMTLandrightPFCregions.TheMTLactivationwasfoundinthehippocampus,andtherightPFCactivationsincludedventrolateralanddorsolateralregions.Inadditiontothesepredictedregions,astrongactivationwasalsofoundintheretrosplenialareaoftheposteriorcingulatecortex.Turningtoregionsactivatedduringtheelaborationperiod,consistentwithourpredictions,theyincludedvisualcortex,precuneus,leftPFCregions,andauditorycortex.Theprecuneusregionwasclearlymoredorsalthantheretrosplenialregionassociatedwithmemoryaccess.LeftPFCactivationsincludedventrolateralandanteriorregions.Con“rmationofGLMResults:Response-lockedTimeCourseAnalysisTocon“rmtheresultsfromtheGLManalysis,weconductedatimecourseanalysisthatwastimelockedtotheresponseandincludedaperiodof4.5sbeforethebuttonpressthatindicatedamemorywasaccessed.BasedonthefMRImodelinFigure1wepredictedacrossoverdissociationbetweenmemoryaccess--relatedactivity,whichshouldbeonadownwardslopeatthetimeoftheresponse,andactivityassociatedwithsubsequentelaboration,whichshouldbeonanupwardslope.Foruseasareference,weextractedtheresponse-lockedtimecoursefromthemotorcortex(greenlineinthemiddlegraphofFig.).Again,theresultscon“rmedourpredictions.AsshowninthetoppartofFigure2,themeantimecoursesoftheregions Figure1.)Thedesigndistinguished6componentsineachtrial:theauditorymemorycue(leftpartofredline),thememoryaccessperiod(yellowline),theresponseindicatingamemorywasaccessed(greenline),theelaborationperiod(blueline),andthe2auditoryratingcues(rightpartofredline).Asindicatedbythearrows,theresponsetimedeterminedtheonsetoftheresponsecomponent,aswellastheoffsetoftheretrievalperiod,andtheonsetoftheelaborationperiod.()Timecoursesforfastandslowresponseswerealmostidenticalintheauditorycortex(audctx,redlines)butverydifferentinthemotorcortex(motctx,greenlines).Thedifferenceinpeaklatencyinthemotorcortex(whitearrows)matchesthe3-stimedifferencethatexistedbetweenthefastandslowresponses.CerebralCortexJanuary2008,V18N1221 thatshowedmemoryaccess--relatedactivity(i.e.,hippocampus,retrosplenialcortex,andrightPFCregions)areallonadescend-ingslopeattheonsetofthemotorresponse,whereasthetimecoursesoftheregionsassociatedwithelaboration(i.e.,visualcortex,precuneus,andleftPFC)areallonanascendingslope.Thein”ectionpointforthesecurvesisat1.5safterthebuttonpress;measuredintermsofhemodynamicresponse,thisis3sbeforethepeakofthemotorresponseforthebuttonpress.Hence,theresponse-lockedtimecourseanalysiscon“rmedtheresultsfromthestandardGLManalysis.RatingScalesModulationfMRIAnalysisThebrainregionsindicatedinTable2andFigure3wereidenti“edusingacombinedparametricmodelinwhichthe2ratings,relivingandemotionalintensity,wereenteredtogetherasparametricregressors.InTable2,wepresentallareasthathada-valuethatwassigni“cantatthesame0.001levelusedinTable1.Oneareainthefrontalpolewaspositivelycorrelatedwithemotionalintensityintheelaborationphaseandalsowithemotionalintensityatthe0.005levelinthememoryaccessphase.Withthisexception,allotherareasshowedexclusiveeffectstoonlyoneofthe2ratingsandinonlyoneperiod,with Figure2.)Hippocampus,retrosplenialcortex,andrightPFCshowedgreateractivityduringtheretrievalperiod,whereasvisualcortex,precuneus,andleftPFCshowedgreateractivityduringtheelaborationperiod.()Regionsassociatedwithretrievalshowedadescendingslopeatthetimethememorywasaccessed(toppanel),whereasregionsassociatedwithelaborationshowedanascendingslope(bottompanel).Allplotsaretheaverageofthecluster.Response-relatedactivityinthemotorcortexisdisplayedasareference(middlepanel). Table1BrainregionsassociatedwithmemoryaccessandelaborationContrastL/RBATalairachxyzzelaborationHippocampusR—23125.35RetrosplenialctxR29/30454208.49RightPFCR473829147.73R464920277.03R9860287.31R63415615.33MedialPFCR1044494.86AnteriorcingulatectxR8/32413456.42CerebellumL—266.05ElaborationonaccessVisualctxL18/1925.51R18/193065.12PrecuneusL752588.61LeftPFCL94521419.01L10424438.70L445315135.36ParietallobeL4053387.75R403849584.09AuditoryctxL41/222588.69R41/42601549.76Note:BABrodmannarea;ctx222SpatiotemporalDynamicsofAutobiographicalMemoryDaselaaretal. thelargestnonsigni“cantbeing1.95.Itisalsoofnotethat,exceptfortheconjunctionareainthefrontalpole,allsigni“cant-valuesthatcamefromthememoryaccessperiodweremodulatedonlybyemotionalintensity,andallthatcamefromtheelaborationperiodweremodulatedonlybyreliving.ThenegativeeffectsofrelivinginrightdorsolateralPFCandinsularegionsduringtheaccessmemoryphasemayre”ectgreaterretrievaleffortformemoriesassociatedwithlowrelivingscores.Thisinterpretation“tswithfMRIevidencelinkingtheseregionstolow-con“dencememorydecisions(e.g.,Flecketal.2006).However,giventhatthesenegativecorrelationswerenotpredicted,thisinterpretationisadhoc.Insummary,thesubjectiveratingsofrelivingandemotionalin-tensityofautobiographicalmemoryaretemporallydissociable.Emotionalintensityduringtheretrievalperiodwaspara-metricallyassociatedwithactivityinthetemporallobe(amyg-dala,hippocampus,andlateraltemporalcortex),putamen,PFC(superiorfrontalgyrusandfrontopolarcortex),andparietallobe(inferiorparietallobuleandsomatosensorycortex).Theseresultssupportthehypothesisofanearlycontributionofemotionduringthememoryretrievalprocess,includingtheengagementof3regionsimportantincurrentneurobiologicaltheoriesofemotion„theamygdala,somatosensorycortex,andfrontopolarcortex„aswellascanonicalmemory-relatedre-gions(e.g.,thehippocampus).Thesenseofrelivingduringonlinemaintenanceandelaborationofautobiographicalmem-orieswasparametricallyassociatedwithactivityintherightinferiorandmedialPFC,cingulategyrus(bothanteriorandposteriorsubdivisions),extrastriatecortex,cerebellum,thala-mus,andbrainstem.Activityintheextrastriatecortex(andpossiblyposteriorcingulatecortex)supportsourapriorihypothesisofaroleofvisualimageryduringthesubjectiverelivingofautobiographicalmemories.Becausethefrontopolarcortexwastheonlybrainregionequallysensitivetoemotional Table2BrainregionsassociatedwithparametricvariationinratingsofrelivingandemotionRegionBASideTalairachMemoryaccessElaborationRelivingEmotionRelivingEmotionMemoryaccessphaseonlyPositivecorrelationsAmygdala—L190.920.450.59Hippocampus—L—R260.630.03Somatosensoryctx3L38650.002.333R4130470.091.471.95Lateraltemporalctx22R561.560.40Putamen—R267111.221.960.77NegativecorrelationsDorsolateralPFC6R494491.881.070.32Insula—R3825120.730.120.25Superiortemporalctx42L10150.890.69ElaborationphaseonlyPositivecorrelationsVisualctx19R237770.541.183.95InferiorPFC47R522611.200.145.15Putamen—L15441.201.026.79Anteriorcingulate32—051110.030.386.3532R4331.191.725.29Posteriorcingulate31R1543271.500.206.77AccessandelaborationFrontalpole10R2360110.323.252.38Note:Allactivationswith0.001areshowninbold;bold;1.75,P\0.05;t[2.58,P\0.01;t[2.92,P\0.005;t[3.69,P\0.001.BABrodmannarea;ctx Figure3.Thetimecoursehighversuslowratingsofemotionalintensityandrelivingin3areas.Theplotsareforthemaximumvalue.Theverticaldottedlineshownat1.5smarkstheinectionpointofareasshowingincreasedanddecreasedactivityinFigure1.CerebralCortexJanuary2008,V18N1223 intensityduringboththememoryaccessandelaborationphases,itmayservetomaintainemotionalrepresentationsonlineasaformofaffectiveworkingmemoryduringautobiographicalAlthoughthetimecourseanalysisbasedonmodulationbytheratingscalesisabitnoisierthanthatbasedonactivity,Figure3supportstheclaimsoftheGLManalyses.Toproducethis“gure,wedidamediansplitforhighversuslowratingsforeachparticipantandplotted4TRsoneithersideofthe1.5-spostbutton-presstimethatwasthein”ectionpointforactivityinFigure2.TheplotsclearlysupporttheanalysisshowninTable2.Highemotionalintensityratingsproducedmoreactivityintheamygdalathandidlowratingsduringtheaccessphase,butnotduringtheelaborationphase.Intherightfrontalpolararea,theemotionalintensityeffectspersistedoverbothphases.Incon-trast,highrelivingratingsproducedmoreactivityinvisualcortexfromtheelaborationphasebutnotduringtheaccessphase.Takingadvantageoftheprotractedtimecourseofautobio-graphicalretrieval,wewereabletodissociatebrainregionsinvolvedintheearlyandlateaspectsofrememberingpersonallyexperiencedpastevents.Initialaccessingofmemoriesengagedhippocampal,retrosplenial,andrightPFCregions,whereaslaterelaborationrecruitedvisualcortex,precuneus,andleftPFCregions.Onlinereliving-modulatedactivitywasobservedduringtheelaborationphaseinvisualcortex,rightinferiorPFCandtheanteriorandposteriorcingulates.Emotion-modulatedactivitywasobservedduringtheinitialretrievalprocessintheamygdala,hippocampus,somatosensorycortex,andlateraltemporalcor-tex.Thus,our2ratingswereparametricallyassociatedwithactivationindifferentareasofthebrainatdifferenttimeperiods,providingbothspatialandtemporaldissociations.Wediscusstheseresultsinmoredetailinthefollowing3sections.AccessingMemories:Hippocampus,RetrosplenialCortex,andRightPFCDuringthe“rsthalfofeachtrial,whileparticipantsrecon-structedamemoryofapersonalevent,theactivatedbrainregionsincludedthehippocampus,theretrosplenialcortex,andtherightPFCandmedialPFC.Thehippocampusshowedgreateractivityduringthememoryaccessthanduringtheelaborationperiod(Fig.2top),anditstimecoursewasonadescendingslopeatthetimeparticipantsmadearesponseindicatingtheendofthememoryaccessperiod(Fig.2The“ndingthatthehippocampusisinvolvedinmemoryaccessoperations“tswellwithevidencefromensemblerecording(HoffmanandMcNaughton2002),event-relatedfMRIstudies(Daselaaretal.2004;Princeetal.2005),andneuropsychology(Squireetal.2004;Moscovitchetal.2005)indicatingthatMTLisinvolvedinnotonlytheformationbutalsothereactivationofmemorytraces.Furthermore,ourresultsprovidefurthersupportforthecriticalroleofthehippocampusinconsciousmemoryprocesses(Schacteretal.1996;Eldridgeetal.2000).Yet,atthesametime,our“ndingsshowthat,althoughthehippocampusisinvolvedinaccessingconsciousmemories,elaborationofthememoriesdependsonsensoryprocessingregionssuchasthevisualcortex.Likethehippocampus,theretrosplenialcortexalsoshowedgreateractivityduringmemoryaccessthanduringelaboration(Fig.2middle)andatimecoursewithadescendingslopeatthetimeofthetransitionfrommemoryaccesstoelaboration(Fig.2top).The“ndingthatboththehippocampusandtheretrosplenialcortexareassociatedwithmemoryaccessisinlinewithclinicalevidenceshowingthatdamagetotheretro-splenialcortexresultsinmemoryde“citssimilartothosefollowingdamagetotheMTL(Valensteinetal.1987).This“ndingisalsoinlinewiththestronganatomicalconnectionsthatexistbetweenthese2regions(KobayashiandAmaral2003).Moreover,inadditiontothehippocampus,theretro-splenialcortexalsosendsdenseprojectionstotheanteriorthalamicnuclei,whichinturnplaysacriticalroleinepisodicmemoryfunction.Damagetothisregionresultsindiencephalicamnesia,andtheanteriorthalamicnucleialsoreceivemassivehippocampalinput(ShibataandYukie2003;VanderWerfetal.2003).Inviewoftheseclinicalandanatomical“ndings,ithasbeensuggestedthattheretrosplenialcortexisanessentialnodeinthecommunicationbetweenthethalamusandthehippo-campus(KobayashiandAmaral2003;ShibataandYukie2003).However,theexactfunctionalcouplingbetweentheseregionsremainsaquestionforfutureresearch.MemoryaccesswasassociatedwithrightPFC.Thisregionshowedgreateractivityduringthememoryaccessthanduringelaboration(Fig.2bottom)andadescendingtimecourseatthetimeofthebuttonpress(Fig.2top).Earlypositronemissiontomography(PET)studiesalreadyhighlightedtheimportanceofrightPFCinepisodicretrievalprocesses.BasedontheirPET“ndings,severalresearchersproposedthatrightPFCisinvolvedinretrievalmode,thesustainedmentalset,whichisenteredintowhenepisodicmemoryretrievalisrequired(Tulving1983;Kapuretal.1995;Nybergetal.1995).ArecentfMRIstudythatdistinguishedbetweentransientandsustainedresponsesduringretrievalprovidedsupportfortheretrievalmodehypothesis.WhereasleftPFCregionsshowedatransientresponse”uctuatingonatrial-by-trialbasis,rightPFCshowedsustainedactivityovertrials(Velanovaetal.2003).Incontrast,otherstudieshavefoundtransientresponsesinrightPFCduringepisodicretrieval.Forinstance,rightPFCactivityhasbeenfoundtobemodulatedbytrial-by-trialchangesincon“denceratings(Hensonetal.2000;Flecketal.2006).Thecurrent“ndingofatransitionwithintrialsfromrighttoleftPFCalsodoesnot“twithasustainedroleofrightPFCinre-trievalmode,butrathersuggestsamoretransientroleinstra-tegicretrievaloperations.However,itshouldbenotedthatouranalysesdidnotspeci“callytargetsustainedactivityovertrials,leavingopenthepossibilitythatsustainedandtransientretrievalactivationsinvolvedifferentsubregionsoftherightPFC.Finally,memoryaccesswasassociatedwithmedialPFC,anareathatiscommonlyobservedintasksrequiringself-referentialprocessing(e.g.,Craiketal.1999;Gusnardetal.2001;Kelleyetal.2002).Forexample,onesuchstudyaskedparticipantstodecidewhetherascenewaspleasantversusunpleasant(self-referential)orthescenewasindoorversusoutdoor(nonself-referential),“ndingmoremedialprefrontalactivityduringthepleasant/unpleasanttask(Laneetal.1997).Morerecentstudieshavesuggesteditisnotreferencetoonesselfpersedrivingthisregion,butmorebroadly,referencetointernallygeneratedinformation,includingonesthoughtsorfeelingsatencoding(e.g.,Simonsetal.2005).Anotherexplanationisthatthisareaisinvolvedinacollectionofprocessesincluding:decisionmakingunderuncertainty,controlprocessesprovidingafeelingofrightnessandtheprocessingofself-referentialinformationthatcombinetomonitortheveracityofautobiographicalmemories224SpatiotemporalDynamicsofAutobiographicalMemoryDaselaaretal. (Gilboa2004).BothexplanationsareisconsistentwiththeresultsofapreviousfMRIstudyinwhichwefoundmedialprefrontalactivitytobegreaterforautobiographicalmemoryretrievalthanforlaboratorymemoryretrieval(Cabezaetal.ElaborationofRetrievedEvents:VisualCortex,Precuneus,andLeftPFCDuringthesecondhalfofeachtrial,whileparticipantsmain-tainedpersonalepisodesinworkingmemory,theregionsmostactivatedincludedvisualcortex,precuneus,andleftPFC.Thevisualcortexregionshowedhigheractivityduringtheelabora-tionperiodthanduringtheretrievalperiod(Fig.2top),anditstimecoursewasonanascendingslopeatthetimeofthebuttonpressindicatingthetransitionfromretrievaltoelaboration(Fig.bottom).The“ndingthatthevisualcortexisinvolvedintheelaborationofautobiographicalmemoriesisinlinewithbehavioral(Johnsonetal.1988;Rubinetal.2003),neuro-psychological(GreenbergandRubin2003),andneuroimagingdata(Cabezaetal.2004)indicatingthatvisualinformationconstitutesafundamentalcomponentofautobiographicalmemories.Furthermore,the“ndingthatthevisualcortexwasactiveduringretrievalintheabsenceofvisualstimulation(becauseourparticipantshadtheireyesclosed)isinlinewithfMRIstudiesshowingthatsensoryregionsthatareactiveduringencodingofperceptualinformationarereactivatedduringretrievalofthesameinformation(Wheeleretal.2000;Kahnetal.2004;Princeetal.2005).Ingeneral,our“ndingscon“rmtheviewthatthevisualcortexplaysakeyroleinautobiograph-icalmemoryretrieval(Rubin2005,2006).Elaborationalsoengagedtheprecuneus.Thisregionshowedgreateractivityduringtheelaborationperiodthanduringthememoryaccessperiod(Fig.2middle)andshowedanascendingtimecourseafterthememorywasaccessed(Fig.bottom).Theprecuneusisoneoftheregionsmostfrequentlyactivatedduringepisodicretrievaltasks(CabezaandNyberg2000a).Yet,theroleofthisregionremainsunclear(Wagneretal.2005).Accordingtoapopularviewbasedonneuroimaging“ndings,theprecuneusisinvolvedinvisualimageryprocessesduringepisodicretrieval(Fletcheretal.1995;CavannaandTrimble2006).Insupportofthisidea,activityinthisregionisgreaterduringretrievalofimageableversusnonimageablewords(Fletcheretal.1995)andduringretrievalofimaginedversusviewedpictures(Lundstrometal.2003).Thisideaisfurthersubstantiatedbythestronganatom-icalconnectivitythatexistsbetweentheprecuneusandlateralparietalregions,suggestingaroleinvisuospatialfunctions(KobayashiandAmaral2003;CavannaandTrimble2006).Thecurrent“ndingthattheprecuneus,togetherwiththevisualcortex,wasassociatedwithelaborationfurthersupportsaroleofthisregioninvisualimageryprocessesduringretrieval.Moreover,the“ndingofactivationsinperceptual,butnotMTL,regionsduringelaboration“tswithmemorytheoriespredictingthat,oncethememoryisselected,MTLbecomeslessimportantandcortical--corticalcoactivationbeginstodominate(Damasio1989;AlvarezandSquire1994).Finally,elaborationofpersonalmemoriesrecruitedleftPFC.Thisregionalsoshowedgreateractivityduringtheelaborationperiodthanduringthememoryaccessperiod(Fig.2andatimecoursewithanascendingslopeafterthememorywasaccessed(Fig.2bottom).LeftPFCiscommonlyactivatedduringsuccessfulitem(Konishietal.2000;Daselaaretal.2003)andassociativerecognition(Princeetal.2005)memorytasks.Giventhecurrentresults,thisactivitycouldwellre”ectelab-orationofaspectsofthestudyepisode.Inlinewiththisidea,sourcememorytasksthatrequireprocessingvisuospatialaspectsofrecoveredepisodicinformationtendtoengageleftPFCregions(Noldeetal.1998;Ranganathetal.2000;Rayeetal.2000;Kahnetal.2004).The“ndingthat,inthecurrentstudy,leftPFCwasassociatedwithelaborationalso“tswellwithfMRIevidenceindicatingaroleofthisregioninworkingmemoryoperations(WagerandSmith2003)andattentionalbiasingofsensoryregions(Hop“ngeretal.2000).Thus,basedon“ndingsfromtheepisodicmemory,workingmemory,andattentionalliterature,leftPFCseemsanobviouscandidateforthetop--downcontrolandworkingmemoryoperationsnecessaryfortheelaborationofrecalledmemories.TheSpatiotemporalDynamicsofRelivingParticipantsinthisstudyspentontheaverageabout12sretrievingautobiographicalmemoriestothepointwheretheythoughttheirmemorieswerefullyformed,followedbyabout12sofcontinuingtothinkaboutandelaboratethememories,asonewoulddowhenreminiscingorsharingmemorieswithothers,answeringquestionsaboutthem,orabstractingin-formationfromthem.Ourmain“ndingsarethatonlineratings(obtainedimmediatelyafterthis24speriod)ofthedegreetowhichthememoryhasbeenrelivedandthedegreeofemotionalintensitywereparametricallyassociatedwithactiva-tionindifferentareasofthebrainanddidsoatdifferenttimes.Asexpectedfromassumingthatjudgmentsofrelivingaremetacognitiveinnatureandmadeonthebasisoffullyformedmemories,allreliving-modulatedactivitywasobservedduringtheelaborationphase,intheperiodafterthememorywasformed.Incontrast,nearlyallemotion-modulatedactivitywasobservedduringtheinitialretrievalprocesspriortothefullformationofthememory.Judgmentsofrelivingwereparametricallyassociatedwithactivityinareasofthebraininvolvedinvisualimagery(extrastriatecortex)(Kosslynetal.2001)andaregionofrightinferiorfrontalcortexhypothesizedtobeessentialforrecol-lectionofepisodesfromthepersonalpast(Markowitsch1995).Theseregionshavebeenobservedacrossseveralstudiesofautobiographicalmemoryretrievalusingdissimilarcomparisontasks,suggestingtheirinvolvementisrelativelyubiquitousandnotsimplyafunctionofthechosensubtraction.Previousresearchsuggeststheidenti“edposteriorregions,whichareinvolvedininitialsensoryprocessingduringencoding,areoftenreactivatedduringretrievalparticularlywhensuccessfulre-trievalisbiasedtowardtheretrievalofvisualdetails(Wheeleretal.2000;Kahnetal.2004;Princeetal.2005;Gardinietal.2006).Becauseourparticipantswereinstructedtoclosetheireyesduringretrieval,activitywithintheseregionswasnotrelatedtosensoryprocessing,butrather,werespeci“ctovisualimageryassociatedwithautobiographicalmemoryretrieval.Rightinferiorfrontalgyrushasbeenidenti“edasacrucialregiontoautobiographicalmemoryretrieval,exhibitingactivityincomparisonsofautobiographicalmemoryretrievaltocate-goryexemplargeneration(Greenbergetal.2005)andto“c-titiousevents(Markowitschetal.2000).Furthermore,damagetothisregionleadstoretrogradeamnesia(Calabreseetal.1996;Levineetal.1998).Evidencefrommorestandardmemoryparadigmssuggestthisregionisassociatedwithdirectingattentiontooractiveselectionofperceptual,ratherthanCerebralCortexJanuary2008,V18N1225 conceptual,representationsduringretrieval(Wagner,Poldrack,etal.1998;Wagner1999;Cadoretetal.2001;DobbinsandWagner2005).Inthecurrentstudy,ratingsofrelivingmaybebasedupontheamountofvisualandperceptualdetailavailable,whoseretrievaldependsontherightinferiorfrontalgyrus.TheSpatiotemporalDynamicsofEmotionalIntensityAreasmodulatedbyjudgmentsofemotionalintensityincludedbrainregionsthoughttobecriticalforemotionalandsomaticsignaling(amygdala,frontopolarcortex,andsomatosensorycortex)aswellasotheronestypicallyengagedduringmemoryretrieval,suchasthehippocampus,lateraltemporalcortex,andposteriorparietalcortex.Interestingly,thesecanonicalmemoryareasweremodulatedbyemotionalintensitybeforethebuttonpressindicatingthattheparticipantthoughtthatthememorywasformed,suggestingthatearlydetectionofhigharousalduringretrievalmayspurandguideeffortsateventreconstruc-tion(Bartlett1932/1995;Reisbergetal.1988;Markowitschetal.2000).Oftheemotion-modulatedareas,onlythefronto-polarcortexcontinuedtobesensitivetoemotionalintensityratingsduringtheelaborationperiod,suggestingaroleinonlinemaintenanceandelaborationofemotionalrepresentations(DavidsonandIrwin1999).Thus,despitethebehavioralcorrespondencefoundbetweensubjectiveratingsofemotionandrelivingduringautobiographicalrecall(Talaricoetal.2004),thepresentfMRI“ndingsindicatethatthesephenomenologicalpropertiesaretemporallyandspatiallydissociable.Themodulationofbrainactivityofafullyformedmemorybyitsrateddegreeofrelivingisnotsurprising.Forinstance,themoreactivityinvisualcortexthemorevividonemightexpectthememorytobeandthusthemorestronglyitislikelytobejudgedashighlyrelived.Buthowcanaratingofemotionalintensitydoneafteramemoryisfullyformedhaveitseffectsduringthetimebefore,butnotafter,thememoryreachesconsciousness?Onecanconsiderwordsinalistasbeingstoredasunitaryitemsandretrievedbywhathavenowbecomefairlywellspeci“edneuralareasintheMTLandfrontallobe(CabezaandNyberg2000a,2000b;BucknerandWheeler2001).Therelativelyrapidreactiontimestoretrievesuchmemoriescomparedwiththetimesobservedhereisconsistentwiththisview.However,itislessplausibletoviewautobiographicalmemoriesinvolvingemotional,visual,auditory,spatial,andotherinformationasbeingstoredasunitaryitemsincontiguousneurallocations(RubinandGreenberg1998;GreenbergandRubin2003;Rubin2005).Rather,theyaremorelikelytohaveneuralrepresentationsthataremoredistributed,andthus,whichareassembledorcoactivatedatthetimeofretrievalusingthesamemedialtemporalandfrontalareasasareusedwithsimplermaterial,alongwithotherareas.Ifthisisthecaseandifemotionsfunctionasearlywarningsthroughretrievalmodulationthenemotionsmayhavetheirgreatesteffectearlyintheprocess,beforethememoryisassembled.Incontrast,visualandothersensoryareasmayhavetheirgreatesteffectlateintheprocess,afterthememoryisassembled.Proust(1928/1934)describedsomethingsimilarinhisclassicaccountofautobiographicalmemorywhereinthesmellandtasteofapetitemadeleine“rstproducesanexquisitepleasuredetached,withnosuggestionofitsoriginandthenafteratimethememoryreturnsatelytheoldgreyhouseuponthestreetroseuplikethesceneryofatheatreandwiththehousethetown.Inthisexample,aretrievalcueworksquicklytochangeemotion,butthememoryisassembledmoreslowly,andwhenitdoescome,itconsistsinlargepartofavisualimage.ThepresentstudyprovidessomecluesastohowthesesequencesofeventsduringrecallunfoldinthehumanbrainbycapitalizingonthedelayedtimecourseofautobiographicalmemoryretrievalusingtheGalton/Crovitztechnique.Previousneuroimagingstudieshavegenerallyfoundaweakerin”uenceofemotiononmemory-relatedbrainregionsduringautobiographicalrecall(MaguireandFrith2003;Addisetal.2004),althoughsomedifferencesinaffectivevalence(positivevs.negativeemotions)havebeenreportedinorbitofrontalcortexandvarioussectorsofthetemporallobe(Piefkeetal.2003).However,thesepriorreportsusedaprescanningcuegenerationmethodwherecuewordsforspeci“cmemorieswerecreatedafewdayspriortothestudy.Underthesecircumstances,emotionalintensitymaynotplayasimportantaroleduringthememoryreconstructionprocessbecausetheeventwasrecentlyretrievedinordertogeneratethecuewords.Usingtheprescanningcuegenerationmethod,wereportedsimilartemporalpro“lesandfunctionalconnectivitybetweenamygdalaandhippocampalactivationduringautobiographicalrecall,suggestingareciprocityofemotionalandmemoryfunc-tions(Greenbergetal.2005).Inthatstudy,though,onlineestimatesofemotionalintensitywerenotobtained,andthetimecourseofretrievalwasmuchfaster(ontheorderof3s)sothatthetemporalstageofmemoryprocessingrelatedtotheseactivationpatternscouldnotbespeci“ed.Onthesepoints,thepresentstudybuildsuponandextendspriorresearch,althoughfutureworkisneededtoexamineotheraspectsofemotionalprocessingduringrecall(e.g.,brainregionsthatparametricallycodethedegreeofphysiologicalarousal)aswellasothersubjectivefeaturesofthememoriesbesidesemotionalintensityandreliving.Coulddifferencesinthewordsusedascuesratherthanthememoriestheycuedbethesourceofourearlyresponsetoemotionratings?Thatis,couldtheearlyeffectsbecausedbyprocessingthewordsthemselvesandourratingsoftheemotionalintensityoftheautobiographicalmemoriestheycuefunctiononlythroughtheircorrelationwiththeemotionalpropertiesofthewords?Wedonotbelievethisisthecaseforseveralreasons.First,the80wordsweusedwerenotespeciallyhighlyemotional.Ifwerankthemintermsoftheemotionalintensityofthewordsthemselves,thetop10emotionalwords(BradleyandLang1999)areanger,ambulance,kiss,joy,“re,thief,trouble,party,love,andmother,andthewordsoccupyingthebottom10ranksareplant,orchestra,bowl,tree,cottage,street,butter,bird,pencil,andpaper.Becauseparametricscalingwasusedastheanalysistool(whichconsidersall4levelsofintensityratingsprovidedbytheparticipants),theeffectscouldnotbedrivenbyjustafewemotionallysalientitems.Second,thepatternofthelocationofactivityshowninTable1ismoreconsistentwiththeretrievalofautobiographicalmemoriesthanwiththesemanticprocessingofwords,withthepossibleexceptionoftheamygdalaactivity.Third,ifwecorrelatetheemotionalintensityratingsofthewords,drawnfromtheAffectiveNormsofEnglishWordsdatabase(BradleyandLang1999),withtheemotionalintensityratingsofthememoriesforeachparticipant,therelationshipisweak(meancorrelationof.19).Theseconsiderationsprovidesupportingevidencethatbrainareasmodulatedbyemotionalintensityratingsre”ecttheintensityofthespeci“cmemorybeing226SpatiotemporalDynamicsofAutobiographicalMemoryDaselaaretal. retrieved,nottheemotionalintensityofthereactiontothemeaningofthecueworditself.Thepresentstudyprovidesnovelneurobiologicalinsightsintosomeaspectsoftheretrievalandphenomenologyofautobio-graphicalmemory.Unlikelaboratory-basedmemoriesforsingleitems,autobiographicalmemoriesaremorecomplex,distrib-utedacrossabroadernetworkofbrainareas,andrequiremoretime-consumingreconstructiveprocesses.Employinganauto-biographicalmemoryparadigm,thepresentfMRIstudydisso-ciatedtheactivationtimecoursesofbrainregionsdifferentiallyinvolvedintheaccessingversuselaborationofpersonalmemories.Thememoryaccessperiodengagedregionsassoci-atedwithaccesstostoredmemorytraces(thehippocampusandtheretrosplenialcortex)andaregionassociatedwithmaintainingthementalsetofepisodicretrieval(rightPFC).Theelaborationperiodrecruitedregionsassociatedwithvisualimagery(extrastriatevisualcortexandprecuneus)andaregionassociatedwithworkingmemoryelaborationandtop--downattentionalcontroloverposteriorbrainregions(leftPFC).Inadditiontoperformingcognitiveoperationsinvolvedinmemoryaccessandelaboration,somebrainregionsweremodulatedbythesubjectivepropertiesofthememoryitself.Emotionalintensitywasparametricallyassociatedwithactivityintheamygdala,temporoparietalregions,andsectorsofPFCinthememoryaccessperiod.Representationofemotionalin-tensitywastemporallyextendedinthefrontopolarcortexafterthememorywasrecoveredandwasactivelymaintained.Thesenseofreliving,inwhichonetravelsbackintimetoreexperiencetheevent,wasrelatedtosubsequentprocessing(afterthememoryisaccessed)inbrainregionsthatincluded:ventralandmedialPFC,cingulategyrus,andextrastriatevisualcortex.TheseGLMpatternswerecon“rmedbytimecourseanalysesbasedontime-lockedaveragesofthefMRIsignal.Thecurrentstudyshowsthatautobiographicalmemoryparadigmsareideallysuitedtoinvestigatethecomponentprocessesunderlyingretrievalofcomplexevents.The“ndingsunderscorethediversityoftemporalsequencingandspatiallocalizationwithintheepisodicmemorynetworkfortheprocessingofsubjectivefeaturesthatarecentraltorecallingourpersonalpast.Indoingso,webelievethattheresultsarethe“rstfMRIstudytorevealthedynamicrecruitmentofdistributedbrainregionsduringdistinctperiodsofepisodicremembering.ThisworkwassupportedbytheNationalInstituteofHealthgrantsR01-AG023123toDCR,R01-DA14094toKSL,andRO1-AG23770andRO1-AG19731toRC.WewishtothankPeggyStJacquesforhercomments.Con”ictofInterest:Nonedeclared.AddresscorrespondencetoDavidC.Rubin,PhD,DepartmentofPsychologyandNeuroscience,POBox90086,DukeUniversity,Durham,NC27708-0086,USA.Email:david.rubin@duke.edu.AddisDR,MoscovitchM,CrawleyAP,McAndrewsMP.2004.Recol-lectivequalitiesmodulatehippocampalactivationduringautobio-graphicalmemoryretrieval.Hippocampus.14:752--762.AlvarezP,SquireLR.1994.Memoryconsolidationandthemedialtemporallobe:asimplenetworkmodel.ProcNatlAcadSciUSA.91:7041--7045.AndreasenNC,OLearyDS,ParadisoS,CizadloT,ArndtS,WatkinsGL,PontoLL,HichwaRD.1999.Thecerebellumplaysaroleinconsciousepisodicmemoryretrieval.HumBrainMapp.8:226--234.BartlettFC.1932.Remembering:astudyinexperimentalandsocialpsychology.Cambridge(UK):CambridgeUniversityPress.BradleyMM,LangPJ.1999.AffectivenormsforEnglishwords(ANEW):stimuli,instructionmanualandaffectiveratings.Gainesville(FL):TheCenterforResearchinPsychophysiology,UniversityofFlorida.TechnicalReportC-1.BucknerRL.2003.Functional-anatomiccorrelatesofcontrolprocessesinmemory.JNeurosci.23:3999--4004.BucknerRL,KoutstaalW,SchacterDL,WagnerAD,RosenBR.1998.Functional-anatomicstudyofepisodicretrievalusingfMRI.I.Re-trievaleffortversusretrievalsuccess.Neuroimage.7:151--162.BucknerRL,WheelerME.2001.Thecognitiveneuroscienceofre-membering.NatRevNeurosci.2:624--634.CabezaR,NybergL.2000a.ImagingcognitionII:anempiricalreviewof275PETandfMRIstudies.JCognNeurosci.12:1--47.CabezaR,NybergL.2000b.Neuralbasesoflearningandmemory:functionalneuroimagingevidence.CurrOpinNeurol.13:415--421.CabezaR,PrinceSE,DaselaarSM,GreenbergDL,BuddeM,DolcosF,LaBarKS,RubinDC.2004.Brainactivityduringepisodicretrievalofautobiographicalandlaboratoryevents:anfMRIstudyusinganovelphotoparadigm.JCognNeurosci.16:1583--1594.CadoretG,PikeGB,PetridesM.2001.Selectiveactivationoftheventrolateralprefrontalcortexinthehumanbrainduringactiveretrievalprocessing.EurJNeurosci.14:1164--1170.CahillL,HaierRJ,FallonJ,AlkireMT,TangC,KeatorD,WuJ,McGaughJL.1996.Amygdalaactivityatencodingcorrelatedwithlong-term,freerecallofemotionalinformation.ProcNatlAcadSciUSA.93:8016--8021.CalabreseP,MarkowitschHJ,DurwenHF,WidlitzekH,HauptsM,HolinkaB,GehlenW.1996.Righttemporofrontalcortexascriticallocusfortheecphoryofoldepisodicmemories.JNeurolNeurosurgPsychiatry.61:304--310.CanliT,ZhaoZ,BrewerJ,GabrieliJD,CahillL.2000.Event-relatedactivationinthehumanamygdalaassociateswithlatermemoryforindividualemotionalexperience.JNeurosci.20:RC99.CavannaAE,TrimbleMR.2006.Theprecuneus:areviewofitsfunctionalanatomyandbehaviouralcorrelates.Brain.129:564--583.ConwayMA,Pleydell-PearceCW.2000.Theconstructionofautobio-graphicalmemoriesintheself-memorysystem.PsycholRev.107:261--288.ConwayMA,Pleydell-PearceCW,WhitecrossSE.2001.Theneuroanat-omyofautobiographicalmemory:aslowwavecorticalpotentialstudyofautobiographicalmemory.JMemLang.45:493--524.ConwayMA,Pleydell-PearceCW,WhitecrossSE,SharpeH.2003.Neurophysiologicalcorrelatesofmemoryforexperiencedandimaginedevents.Neuropsychologia.41:334--340.ConwayMA,TurkDJ,MillerSL,LoganJ,NebesRD,MeltzerCC,BeckerJT.1999.Apositronemissiontomography(PET)studyofautobio-graphicalmemoryretrieval.Memory.7:679--702.CraikFIM,MorozTM,MoscovitschM,StussDT,WinocurG,TulvingE,KapurS.1999.Insearchofself:apositronemissiontomographystudy.PsycholSci.10:26--34.CrovitzHF,ShiffmanH.1974.Frequencyofepisodicmemoriesasafunctionoftheirage.BullPsychonSoc.4:517--518.DamasioAR.1989.Time-lockedmultiregionalretroactivation„asystems-levelproposalfortheneuralsubstratesofrecallandrec-ognition.Cognition.33:25--62.DamasioAR,GrabowskiTJ,BecharaA,DamasioH,PontoLL,ParviziJ,HichwaRD.2000.Subcorticalandcorticalbrainactivityduringthefeelingofself-generatedemotions.NatNeurosci.3:1049--1056.DaselaarSM,RomboutsSA,VeltmanDJ,RaaijmakersJG,LazeronRH,JonkerC.2001.Parahippocampalactivationduringsuccessfulrecognitionofwords:aself-pacedevent-relatedfMRIstudy.Neuro-image.13:1113--1120.DaselaarSM,VeltmanDJ,RomboutsSARB,RaaijmakersJGW,JonkerC.2003.Neuroanatomicalcorrelatesofepisodicencodingandretrievalinyoungandelderlysubjects.Brain.126:43--56.CerebralCortexJanuary2008,V18N1227 DaselaarSM,VeltmanDJ,WitterMP.2004.Commonpathwayinthemedialtemporallobeforstorageandrecoveryofwordsasrevealedbyevent-relatedfunctionalMRI.Hippocampus.14:163--169.DavidsonRJ,IrwinW.1999.Thefunctionalneuroanatomyofemotionandaffectivestyle.TrendsCognSci.3:11--21.DobbinsIG,FoleyH,SchacterDL,WagnerAD.2002.Executivecontrolduringepisodicretrieval:multipleprefrontalprocessessubservesourcememory.Neuron.35:989--996.DobbinsIG,WagnerAD.2005.Domain-generalanddomain-sensitiveprefrontalmechanismsforrecollectingeventsanddetectingnov-elty.CerebCortex.15:1768--1778.DolcosF,LaBarKS,CabezaR.2004.Interactionbetweentheamygdalaandthemedialtemporallobememorysystempredictsbettermemoryforemotionalevents.Neuron.42:855--863.DolcosF,LaBarKS,CabezaR.2005.Rememberingoneyearlater:roleoftheamygdalaandthemedialtemporallobememorysysteminretrievingemotionalmemories.ProcNatlAcadSciUSA.102:2626--2631.EichenbaumH.2004.Hippocampus:cognitiveprocessesandneuralrepresentationsthatunderliedeclarativememory.Neuron.44:109--120.EldridgeLL,KnowltonBJ,FurmanskiCS,BookheimerSY,EngelSA.2000.Rememberingepisodes:aselectiveroleforthehippocampusduringretrieval.NatNeurosci.3:1149--1152.FinkGR,MarkowitschHJ,ReinkemeierM,BruckbauerT,KesslerJ,HeissWD.1996.Cerebralrepresentationofonesownpast:neuralnetworksinvolvedinautobiographicalmemory.JNeurosci.16:4275--4282.FitzgeraldDA,PosseS,MooreGJ,TancerME,NathanPJ,PhanKL.2004.Neuralcorrelatesofinternally-generateddisgustviaautobiographi-calrecall:afunctionalmagneticresonanceimaginginvestigation.NeurosciLett.370:91--96.FleckMS,DaselaarSM,DobbinsIG,CabezaR.2006.Roleofprefrontalandanteriorcingulateregionsindecision-makingprocessessharedbymemoryandnon-memorytasks.CerebCortex.16:1623--1630.FletcherPC,FrithCD,BakerSC,ShalliceT,FrackowiakRS,DolanRJ.1995.Themindseye„precuneusactivationinmemory-relatedimagery.Neuroimage.2:195--200.GaltonF.1879.Psychometricexperiments.Brain.2:149--162.GardiniS,CornoldiC,DeBeniR,VenneriA.2006.Leftmediotemporalstructuresmediatetheretrievalofepisodicautobiographicalmentalimages.Neuroimage.30:645--655.GilboaA.2004.Autobiographicalandepisodicmemory„oneandthesame?evidencefromprefrontalactivationinneuroimagingstudies.Neuropsychologia.42:1336--1349.GilboaA,WinocurG,GradyCL,HevenorSJ,MoscovitchM.2004.Rememberingourpast:functionalneuroanatomyofrecollectionofrecentandveryremotepersonalevents.CerebCortex.14:1214--1225.GreenbergDL,RiceHJ,CooperJJ,CabezaR,RubinDC,LaBarKS.2005.Co-activationoftheamygdala,hippocampusandinferiorfrontalgyrusduringautobiographicalmemoryretrieval.Neuropsychologia.43:659--674.GreenbergDL,RubinDC.2003.Theneuropsychologyofautobiograph-icalmemory.Cortex.39:687--728.GusnardDA,AkbudakE,ShulmanGL,RaichleME.2001.Medialprefrontalcortexandself-referentialmentalactivity:relationtoadefaultmodeofbrainfunction.ProcNatlAcadSciUSA.98:4259--4264.HamannSB,ElyTD,GraftonST,KiltsCD.1999.Amygdalaactivityrelatedtoenhancedmemoryforpleasantandaversivestimuli.NatNeurosci.2:289--293.HensonRN,RuggMD,ShalliceT,DolanRJ.2000.Con“denceinrecognitionmemoryforwords:dissociatingrightprefrontalrolesinepisodicretrieval.JCognNeurosci.12:913--923.HensonRN,RuggMD,ShalliceT,JosephsO,DolanRJ.1999.Recollec-tionandfamiliarityinrecognitionmemory:anevent-relatedfunc-tionalmagneticresonanceimagingstudy.JNeurosci.19:3962--3972.HoffmanKL,McNaughtonBL.2002.Coordinatedreactivationofdistrib-utedmemorytracesinprimateneocortex.Science.297:2070--2073.Hop“ngerJB,BuonocoreMH,MangunGR.2000.Theneuralmecha-nismsoftop-downattentionalcontrol.NatNeurosci.3:284--291.JamesW.1890.Theprinciplesofpsychology.NewYork:H.HoltandJohnsonMK,FoleyMA,SuengasAG,RayeCL.1988.Phenomenalcharacteristicsofmemoriesforperceivedandimaginedautobio-graphicalevents.JExpPsycholGen.117:371--376.KahnI,DavachiL,WagnerAD.2004.Functional-neuroanatomiccorrelatesofrecollection:implicationsformodelsofrecognitionmemory.JNeurosci.24:4172--4180.KapurS,CraikFM,JonesC,BrownGM,HouleS,TulvingE.1995.Functionalroleoftheprefrontalcortexinretrievalofmemories:aPETstudy.Neuroreport.6:1880--1884.KeedwellPA,AndrewC,WilliamsSC,BrammerMJ,PhillipsML.2005.Adoubledissociationofventromedialprefrontalcorticalresponsestosadandhappystimuliindepressedandhealthyindividuals.BiolPsychiatry.58:495--503.KelleyWM,MacraeCN,WylandCL,CaglarS,InatiS,HeathertonTF.2002.Findingtheself?anevent-relatedfMRIstudy.JCognNeurosci.14:785--794.KensingerEA,CorkinS.2004.Tworoutestoemotionalmemory:distinctneuralprocessesforvalenceandarousal.ProcNatlAcadSciUSA.101:3310--3315.KobayashiY,AmaralDG.2003.Macaquemonkeyretrosplenialcortex:II.Corticalafferents.JCompNeurol.466:48--79.KonishiS,WheelerME,DonaldsonDI,BucknerRL.2000.Neuralcorrelatesofepisodicretrievalsuccess.Neuroimage.12:276--286.KopelmanMD,KapurN.2001.Thelossofepisodicmemoriesinretrogradeamnesia:single-caseandgroupstudies.PhilosTransRSocLondBBiolSci.356:1409--1421.KosslynSM,GanisG,ThompsonWL.2001.Neuralfoundationsofimagery.NatRevNeurosci.2:635--642.LaneRD,FinkGR,ChauPM,DolanRJ.1997.Neuralactivationduringselectiveattentiontosubjectiveemotionalresponses.Neuroreport.8:3969--3972.LazarusRS.1991.Emotionandadaptation.NewYork(NY):OxfordUniversityPress.p.557.LeDouxJ.1996.Emotionalnetworksandmotorcontrol:afearfulview.ProgBrainRes.107:437--446.LeventhalH,SchererK.1987.Therelationshipofemotiontocognition:afunctionalapproachtoasemanticcontroversy.CognEmot.1:3--28.LevineB,BlackSE,CabezaR,SindenM,McIntoshAR,TothJP,TulvingE,StussDT.1998.Episodicmemoryandtheselfinacaseofisolatedretrogradeamnesia.Brain.121(Pt10):1951--1973.LundstromBN,PeterssonKM,AnderssonJ,JohanssonM,FranssonP,IngvarM.2003.Isolatingtheretrievalofimaginedpicturesduringepisodicmemory:activationoftheleftprecuneusandleftprefrontalcortex.Neuroimage.20:1934--1943.MaguireEA.2001.Neuroimagingstudiesofautobiographicaleventmemory.PhilosTransRSocLondBBiolSci.356:1441--1451.MaguireEA,FrithCD.2003.Lateralasymmetryinthehippocampalresponsetotheremotenessofautobiographicalmemories.JNeuro-sci.23:5302--5307.MarkowitschHJ.1995.Whichbrainregionsarecriticallyinvolvedintheretrievalofoldepisodicmemory?BrainResBrainResRev.21:117--127.MarkowitschHJ,ThielA,ReinkemeierM,KesslerJ,KoyuncuA,HeissWD.2000.Rightamygdalarandtemporofrontalactivationduringautobiographic,butnotduring“ctitiousmemoryretrieval.BehavNeurol.12:181--190.MarkowitschHJ,VandekerckhovelMM,LanfermannH,RussMO.2003.Engagementoflateralandmedialprefrontalareasintheecphoryofsadandhappyautobiographicalmemories.Cortex.39:643--665.McDermottKB,JonesTC,PetersenSE,LagemanSK,RoedigerHL3rd.2000.Retrievalsuccessisaccompaniedbyenhancedactivationinanteriorprefrontalcortexduringrecognitionmemory:anevent-relatedfMRIstudy.JCognNeurosci.12:965--976.MoscovitchM.1995.Recoveredconsciousness:ahypothesisconcern-ingmodularityandepisodicmemory.JClinExpNeuropsychol.17:276--290.MoscovitchM,MeloB.1997.Strategicretrievalandthefrontallobes:evidencefromconfabulationandamnesia.Neuropsychologia.35:1017--1034.228SpatiotemporalDynamicsofAutobiographicalMemoryDaselaaretal. MoscovitchM,RosenbaumRS,GilboaA,AddisDR,WestmacottR,GradyC,McAndrewsMP,LevineB,BlackS,WinocurG,etal.2005.Functionalneuroanatomyofremoteepisodic,semanticandspatialmemory:auni“edaccountbasedonmultipletracetheory.JAnat.207:35--66.NoldeSF,JohnsonMK,DEspositoM.1998.Leftprefrontalactivationduringepisodicremembering:anevent-relatedfMRIstudy.Neuro-report.9:3509--3514.NybergL,TulvingE,HabibR,NilssonLG,KapurS,HouleS,CabezaR,McintoshAR.1995.Functionalbrainmapsofretrievalmodeandrecoveryofepisodicinformation.Neuroreport.7:249--252.PiefkeM,WeissPH,ZillesK,MarkowitschHJ,FinkGR.2003.Differentialremotenessandemotionaltonemodulatetheneuralcorrelatesofautobiographicalmemory.Brain.126:650--668.PrinceSE,DaselaarSM,CabezaR.2005.Neuralcorrelatesofrelationalmemory:successfulencodingandretrievalofsemanticandpercep-tualassociations.JNeurosci.25:1203--1210.ProustM.1928.Swannsway.NewYork:RandomHouse.RanganathC,JohnsonMK,DEspositoM.2000.Leftanteriorprefrontalactivationincreaseswithdemandstorecallspeci“cperceptualinformation.JNeurosci.20(RC108):1--5.RayeCL,JohnsonMK,MitchellKJ,NoldeSF.2000.fMRIinvestigationsofleftandrightPFCcontributionstoepisodicremembering.Psychobiology.28:197--206.ReisbergD,HeuerF,McLeanJ,OShaughnessyM.1988.Thequantity,notthequality,ofaffectpredictsmemoryvividness.BullPsychonSoc.26:100--103.RubinDC.1980.51propertiesof125words:aunitanalysisofverbalbehavior.JVerbLearnVerbBehav.19:736--755.RubinDC.1982.Ontheretentionfunctionforautobiographicalmemory.JVerbLearnVerbBehav.21:21--38.RubinDC.2005.Abasic-systemsapproachtoautobiographicalmemory.CurrDirPsycholSci.14:79--83.RubinDC.2006.Thebasic-systemsmodelofepisodicmemory.PerspectPsycholSci.1:277--311.RubinDC,GreenbergDL.1998.Visualmemory-de“citamnesia:adistinctamnesicpresentationandetiology.ProcNatlAcadSciUSA.95:5413--5416.RubinDC,SchraufRW,GreenbergDL.2003.Beliefandrecollectionofautobiographicalmemories.MemCognit.31:887--901.RubinDC,SchulkindMD.1997.Thedistributionofautobiographicalmemoriesacrossthelifespan.MemCognit.25:859--866.RubinDC,SieglerIC.2004.Facetsofpersonalityandthephenomenologyofautobiographicalmemory.ApplCognPsychol.18:913--930.RuggMD,WildingEL.2000.Retrievalprocessingandepisodicmemory.TrendsCognSci.4:108--115.SchacterDL,AlpertNM,SavageCR,RauchSL,AlbertMS.1996.Consciousrecollectionandthehumanhippocampalformation:evidencefrompositronemissiontomography.ProcNatlAcadSciUSA.93:321--325.ShannonBJ,BucknerRL.2004.Functional-anatomiccorrelatesofmemoryretrievalthatsuggestnontraditionalprocessingrolesformultipledistinctregionswithinposteriorparietalcortex.JNeurosci.24:10084--10092.SharotT,DelgadoMR,PhelpsEA.2004.Howemotionenhancesthefeelingofremembering.NatNeurosci.7:1376--1380.ShibataH,YukieM.2003.Differentialthalamicconnectionsoftheposteroventralanddorsalposteriorcingulategyrusinthemonkey.EurJNeurosci.18:1615--1626.SimonsJS,GilbertSJ,OwenAM,FletcherPC,BurgessPW.2005.Distinctrolesforlateralandmedialanteriorprefrontalcortexincontextualrecollection.JNeurophysiol.94:813--820.SquireLR,StarkCE,ClarkRE.2004.Themedialtemporallobe.AnnuRevNeurosci.27:279--306.StussDT,LevineB.2002.Adultclinicalneuropsychology:lessonsfromstudiesofthefrontallobes.AnnuRevPsychol.53:401--433.SvobodaE,McKinnonMC,LevineB.2006.Thefunctionalneuroanatomyofautobiographicalmemory:ameta-analysis.Neuropsychologia.44:2189--2208.TalaricoJM,LaBarKS,RubinDC.2004.Emotionalintensitypredictsautobiographicalmemoryexperience.MemCognit.32:1118--1132.TulvingE.1983.Elementsofepisodicmemory.Oxford:ClarendonPress.TulvingE.2002.Episodicmemory:frommindtobrain.AnnuRevPsychol.53:1--25.ValensteinE,BowersD,VerfaellieM,HeilmanKM,DayA,WatsonRT.1987.Retrosplenialamnesia.Brain.110(Pt6):1631--1646.VanderWerfYD,JollesJ,WitterMP,UylingsHB.2003.Contributionsofthalamicnucleitodeclarativememoryfunctioning.Cortex.39:1047--1062.VelanovaK,JacobyLL,WheelerME,McAvoyMP,PetersenSE,BucknerRL.2003.Functional-anatomiccorrelatesofsustainedandtransientprocessingcomponentsengagedduringcontrolledretrieval.JNeurosci.23:8460--8470.WagerTD,SmithEE.2003.Neuroimagingstudiesofworkingmemory:ameta-analysis.CognAffectBehavNeurosci.3:255--274.WagnerAD.1999.Workingmemorycontributionstohumanlearningandremembering.Neuron.22:19--22.WagnerAD,DesmondJE,GloverGH,GabrieliJD.1998.Prefrontalcor-texandrecognitionmemory.Functional-MRIevidenceforcontext-dependentretrievalprocesses.Brain.121(Pt10):1985--2002.WagnerAD,MarilA,BjorkRA,SchacterDL.2001.Prefrontalcontribu-tionstoexecutivecontrol:fMRIevidenceforfunctionaldistinctionswithinlateralprefrontalcortex.Neuroimage.14:1337--1347.WagnerAD,PoldrackRA,EldridgeLL,DesmondJE,GloverGH,GabrieliJD.1998.Material-speci“clateralizationofprefrontalactivationduringepisodicencodingandretrieval.Neuroreport.9:3711--3717.WagnerAD,ShannonBJ,KahnI,BucknerRL.2005.Parietallobecontributionstoepisodicmemoryretrieval.TrendsCognSci.9:445--453.WheelerMA,StussDT,TulvingE.1997.Towardatheoryofepisodicmemory:thefrontallobesandautonoeticconsciousness.PsycholBull.121:331--354.WheelerME,BucknerRL.2004.Functional-anatomiccorrelatesofrememberingandknowing.Neuroimage.21:1337--1349.WheelerME,PetersenSE,BucknerRL.2000.Memorysecho:vividrememberingreactivatessensory-speci“ccortex.ProcNatlAcadSciUSA.20:11125--11129.YonelinasAP.2001.Componentsofepisodicmemory:thecontributionofrecollectionandfamiliarity.PhilosTransRSocLondBBiolSci.356:1363--1374.YonelinasAP,LevyBJ.2002.Dissociatingfamiliarityfromrecollectioninhumanrecognitionmemory:differentratesofforgettingovershortretentionintervals.PsychonBullRev.9:575--582.CerebralCortexJanuary2008,V18N1229