Table Of ContentRecognitionMemory 1
RunningHead: RECOGNITIONMEMORY
RecognitionMemoryforValencedandArousing
MaterialsUnderConditionsofDividedAttention
ArloClark-Foos
and
RichardL.Marsh
UniversityofGeorgia
Addresscorrespondence:
RichardL.Marsh
DepartmentofPsychology
UniversityofGeorgia
Athens,GA30602-3013
email: [email protected]
706-542-0058
RecognitionMemory 2
Abstract
Twoexperimentsarereportedthatexaminedtheeffectofbothvalenceandarousalon
recognitionmemoryperformance. Eachexperimentusedtwoclassesofnegativeitemsthat
differedinarousal,aswellasaneutalandnonarousingsetofitems. InExperiment1a
difficultdividedattentiontaskwascrossedwiththelearningandtestphasesofthe
experiment. InExperiment2encodingtimewasmanipulatedandremember-know
judgmentswerecollected. Theemotionalenhancementeffectoftenfoundwithverbal
materialssurvivedthedepletionofcognitiveresources,asdidtheextrabenefitacruingfrom
higharousal. Althoughwefoundthatarousalleadtomorerecollection,thegeneral
conclusionthatwedrawisthattheeffectofemotiononrecognitionmemorycanbe
attributedtorelativelyautomaticsourcessuchasthedegreeoffamiliarityengenderedbya
classofmaterials.
RecognitionMemory 3
RecognitionMemoryforValencedandArousing
MaterialsUnderConditionsofDividedAttention
Everyoneknowsthatlifebringswithitavarietyofexperiences,someofwhichare
pleasantandothersareonesthatwewouldratheravoid. Intermsoftheirmemorability,
valencedexperiencestendtobebetterrememberedthanotherwisecomparableneutralones
(forreviewsseeBuchanan&Adolphs,2002;Dolan,2002;Hamann,2001). Thisemotional
enhancementeffectisarathergeneralpropertyofourmemorysystembecauseitoccurswith
anymannerofstimulustypes(e.g.,words,pictures,stories,etc.). Italsogenerallyoccurs
whenmemoryistestedusingawiderangeofassessmenttechniques(e.g.,recognition,free
recall,cuedrecall,etc.). Theexactmechanismsunderlyingthememorialbenefittoemotional
materialshavebeendebatedovertheyears. Forexample,somearguethatemotionalitems
receivemorerehearsalormoreelaborateprocessingatthetimethattheyareencountered
(e.g.,Christianson&Engelberg,1999). Theseaccountsusuallypositfurtherthatsuch
elaborationcanresultinidiosyncraticpersonalrelevanceofthematerialwhichfurther
increasesmeasuresofmemory(inparticular,recollection;Kensinger&Corkin,2003). Others
arguethatthecognitivesystemiswiredinafashiontopreattentivelydetectvalenced
materialsagainstthefabricofeverydayexperiences,andthismaybeespeciallytruefor
negativeoraversivematerial(e.g.,Anderson,2005;Christianson,1992;Williams,Matthews,
&MacLeod,1996). Ofcourse,therearemanybiologicalandpharmacologicaltheoriesforthe
emotionalenhancementeffect(Anderson&Phelps,2001;Cain,Kapp,&Puryear,2002).
Thesetheoriesalignthemselveswitheitherthemorecontrolledprocessingexplanationsor
theonespositingmoreautomaticbenefitstomemory.
Inrecentyears,investigatorshavebecomeincreasinglyawareofthefactthatemotional
informationdoesnotvaryalongoneunderlyingdimension(Bradley,Greenwald,Petry,&
Lang,1992). Rather,emotionalmaterialhastwounderlyingdimensions: valenceandarousal.
Valencereferstothedimensionofhowastimulusevokespositiveversusnegativeaffect. By
RecognitionMemory 4
contrast,arousalisthedegreetowhichastimulusevokesanexcitingversusacalming
reaction. Thepsychophysiologicalevidencesuggeststhatvalenceexertsitsinfluenceonmore
controlledprocessessuchaselaborationandrehearsal. Consequently,whenvalenceditems
areexperiencedthatarenotarousingthenprefrontalcorticalareasaremoreactivatedas
comparedwithwithneutralmaterialsthatlackavalencedcomponent(e.g.,Kensinger&
Corkin,2004). But,whenitemsarepresentedthatarebothvalencedandarousingthen
amygdalarandhippocampalregionsbecomeactivatedtherebyimplicatingtheeffectof
arousalasamoreautomaticinfluenceonmemory(Cahill&McGaugh,1990;Kensinger&
Corkin,2004;McGaugh,Introini-Collison,Cahill,Kim,&Liang,1992). Thus,thereappearsto
betworoutesbywhichtheemotionalenhancementeffectcanoccur: onethroughcontrolled
processingdistinguishingbetweenneutralversusvalenceditemsandonethroughmore
automaticprocessingdistinguishingbetweenarousingvalencedmaterialsandnonarousing
valencedmaterials.
Supportingthisviewofthedualityinherentintheemotionalenhancementeffect,
KensingerandCorkin(2004)conductedabehavioralstudyinwhichtheypresentedneutral
andtwoclassesofnegativeitemsthatdifferedintheirlevelofarousal(arousingand
nonarousing). Underconditionsoffullattentionatencoding,bothclassesofnegativeitems
wererecognizedbetterthanneutralitems. However,underdividedattentionduring
learning,thenegativeandarousingitemscontinuedtoberecognizedmorefrequently,but
thenegativeandnonarousingitemsnolongerheldanyadvantageoverneutralwordsin
theirrateofsubsequentrecognition. Therefore,theeffectofdividedattentionduringlearning
eliminatedtheemotionalenhancementeffectduetovalencebecauseitpresumablyinterfered
withmorecontrolledprocesses. However,thearousingmaterialscontinuedtodisplaytheir
advantageimplyingthattheeffectofarousalonmemoryisarelativelyautomaticprocess
thatsurvivesdivisionofattention.
Thepurposeofthepresentstudywastoexploretheseeffectsfurther. Morespecifically,
ourgoalwastocomparetheeffectsofdividingattentionatbothencodingandretrieval.
RecognitionMemory 5
Obviously,withdivisionofattentionatencodingwewouldexpecttofindthatnegativeand
arousingitemsarebetterrecognizedtherebyreplicatingKensingerandCorkin(2004). Hicks
andMarsh(2000)haveclaimedthatdivisionofattentionduringarecognitiontestreduces
memoryforwell-learned,recollectiveinformationbutleavesintactinformationthatwould
havebeenrecognizedbasedonvaguerfeelingsoffamiliarity(cf. Jacoby,1999;Jennings&
Jacoby,1993,1997). Therefore,ifthedifferencebetweenarousingandnonarousingitems
persistsunderdividedattentionattest,thenthedifferenceislikelyowingtoaverage
differencesinvaguerfeelingsoffamiliaritybetweenthetwoclassesofnegativeinformation.
Ifthedifferencedoesnotpersist,thenthatoutcomewouldsuggestthatarousaland
amygdalarprojectionsconfertheirmemorialadvantagethroughmoredisruption-prone
recollectivedetails. Inaddition,thedivisionofattentionatbothencodingandretrievalforces
participantstorelysolelyonfamiliarityfortheirrecognitionjudgmentsandhasbeenknown
tohavetheironiceffectofresultinginbettermemory,thansay,dividingattentiononly
duringencoding(Jacoby,1991;Hicks&Marsh,2000). Toourknowledge,whethervalence
andarousalstillconfermemorialadvantageshasnotbeentestedunderdividedattentionat
bothstudyandtest. Theoretically,findingeitheravalenceadvantageand/oranadditional
arousaladvantagewouldhelptospecifythetypeofinformationconferringthatadvantage
(i.e.,recollectionorfamiliarity).
Experiment1
ThegoalofExperiment1wastoorthogonallycrossdivisionofattentionwiththephase
(encodingvs. test)inwhichitwasapplied. Theresultwasfourconditions: onewithfull
attentionatbothencodingandretrieval,anotherwithdividedattentionduringbothphases,
andtwoadditionalconditionswithdividedattentiononlyduringlearningortestbutnot
both.
RecognitionMemory 6
Method
Participants. UndergraduatestudentsfromtheUniversityofGeorgiavolunteeredin
exchangeforcredittowardaresearchappreciationrequirement. Eachparticipantwastested
individuallyinsessionsthatlastedapproximately30min. Wetested25volunteersineachof
thefourbetween-subjectsconditions. Twopeoplewithat-chanceperformancewerereplaced
undertheassumptionthattheywerenotcognitivelyengagedinthetasksthatweaskedthem
toperform.
MaterialsandProcedure. ThematerialswerewordsselectedfromtheAffectiveNorms
forEnglishWords(ANEW;Bradley&Lang,1999). Eachofthethreeclassesofitemswas
comprisedof48wordsthatwereequatedonbothANEWwordfrequencyandKucˇeraand
Francis(1967)measures. Obviously,theneutralandbothnegativeclassesitemsdiffered
significantlyinvalence(5.78vs. 2.43,respectively,inANEWvalues),buttheneutraland
negativenonarousingitemswereofthesamelowlevelofarousal(4.38inANEWvalues).
Onlythetwonegativeclassesofitemsdifferedinarousal(4.38vs. 6.84inANEWvalues). For
agivenparticipant,thesoftwarecontrollingtheexperimentrandomlychoseanew24items
fromeachofthethreeclassestobestudiedandreservedtheremaining24itemsineachclass
toserveasdistractorsontherecognitiontest. Inthisway,theencodingphasewas72trials
longandthetestphasewas144itemslong. Eachitemwasstudiedfor2sinthecenterofthe
computermonitorduringencoding;andthetestphasewasselfpaced. Allinstructionsto
participantswerefirstreadfromthecomputermonitorandthentheywerereiteratedinthe
experimenter’sownwords. Noparticipantwasallowedtocontinuewithanytaskunlessthe
experimenterwasconfidentthattheparticipanthadcompleteunderstandingofwhatwas
beingrequested. Thepresentationofallstimuliwereprecededbya250mswarningtoneand
fixationpointatbothstudyandtest. Theintertrialintervalduringtestingwas700ms.
Participantswhowereaskedtoperformrandomnumbergeneration(RNG)asthe
dividedattentiontaskweregivendetailedinstructionsonhowtoberandom(seeHicks&
RecognitionMemory 7
Marsh,2000). Theywereaskedtocalloutanumberbetween1and10(inclusive)forevery
beepofanelectronicmetronomeprogrammedtoissueabeepat1sintervals. Thispacingis
quitefastandthetaskisquitedifficult. Participantsweregivenapproximately1.5minto
practicethetaskunencumberedbyeitherlearningortestingduringwhichtheexperimenter
recordedthedigits,missedbeats,etc. toserveasabaselinemeasureofrandomnessto
comparewithperformancewhenthetaskwasperformedconcurrently. Afterpractice,either
thelearningphaseortestphasewasadministered. Inallcaseswherethebaselinedidnot
needtobetaken(oralreadyhadbeentakenearlierinthecaseofdividingattentionatboth
studyandtest)allotherparticipantsweregivenamathematicaldistractortasktoequatethe
timingandretentionintervalsacrossallfourbetween-subjectsconditions.
ResultsandDiscussion
p
Inthisandthefollowingexperiment,unlessa valueisreportedwithastatisticaltest,
theprobabilityofaTypeIerrordoesnotexceedtheconventional5%. Webeginbyreporting
theresultsoftheRNGtask. Ameasureofrandomnesswascomputedusingthefirst100
digitsatbothbaselineandwhenthetaskwasperformedconcurrently. Numericallylower
valuesofRNGindicatemorerandomsequencesandhighervaluesindicatelessrandom
performance. Fortheconditionswherethetaskwasperformedwhilestudyingwords,
baselineperformancewas0.29anditwas0.33whenperformedconcurrentlyduring
t
encoding, (59)=5.65. Fortheconditionswherethetaskwasperformedduringtesting,
baselineperformancewas0.30anditwas0.34whenperformedconcurrentlyduringtesting,
t
(59)=4.57. Consequently,therewerenodualtasktradeoffsinwhichparticipantssacrificed
performanceononetasktoperformwellontheotherbecause,aswewillshowshortly,
memoryperformancealsosufferedwhenperformedconcurrentlyaswell. Inshort,
performancedeficitswereshownonbothtasks(cf.,Marsh&Hicks,1998).
ThehitratesandfalsealarmratesarereportedinTable1foreachofthethreeclassesof
itemsacrossthefourbetween-subjectconditions. Thesemetricshavealsobeencombinedby
RecognitionMemory 8
subtractingthefalsealarmratefromthehitratewithinagivenclassofitemstoarriveat
measuresofcorrectedrecognition. OnlythelastthreecolumnsofTable1wereanalyzed
×
statistically. Weconducteda2(fullvs. dividedattentionatencoding) 2(fullvs. divided
×
attentionattest) 3(itemtype)AnalysisofVariance(ANOVA).Asthereadercansee,
F
negativeitemswererecognizedmoreoftenthantheirneutralcounterparts, (2,192)=27.81,
η2
p =.23. Thus,theemotionalenhancementeffectwasobtained. Insimpleeffectsteststhat
contrastedthenegativearousingitemsagainstthenegativenonarousingitems,therewere
significantdifferenceswhenattentionwasdividedatstudyaloneandtestalone,smallerof
t
thetwo (24)’s=2.15. However,therewasonlyanominaleffectfavoringthearousing
negativeitemsoverthenonarousingintheremainingtwoconditions. Mostcritically,the
evidenceappearstoconvergeontheideathattheemotionalenhancementeffectfornegative
itemsoverneutralitemsisnotnecessarilyeliminatedbydivisionofattentionateitherstudy
ortest(cf. Kensinger&Corkin,2004). Inaddition,thebenefitofarousingovernonarousing
itemswasnumericallypresentintwoofthefourconditions,andstatisticallypresentinthe
othertwoconditionssoittooappearstogenerallysurvivethedepletionofcognitive
resources. ThisclaimcanbefurtherbolsteredbyrunningareducedANOVAmodelwithout
thefullattentionconditionandwithouttheneutralitemsthatshowsthedifferencebetween
F η2
thetwoclassesofnegativeitemsisstatisticallysignificant, (1,72)=8.26, p =.10.
Notsurprisingly,performancewasmuchlowerwhenattentionwasdividedduring
F η2
studyascomparedwithfullattention, (1,96)=25.03, p =.21. Thatoutcomereplicatesthe
extensiveworkthatNaveh-Benjaminandhiscolleagueshaveperformedtoshowthatalmost
inevitablydepletionofresourcesduringstudyhasadeleteriouseffectonmemory(e.g.,
Naveh-Benjamin,Kilb,&Fisher,2006;Naveh-Benjamin,Guez,&Marom,2003). Inthe
omnibustest,themaineffectofdivided(versusfull)attentionattestdidnotreach
significanceandnoneoftheinteractionsweresignificant. Theabsenceofthistesteffectin
contrasttoHicksandMarsh’s(2000)significanteffectislikelyowingtothefactthat
two-thirdsoftheentiuresetofmaterialswereprotectedbytheemotionalenhancementeffect.
RecognitionMemory 9
Inaddition,participantsassignedtotheconditionwithbothdividedattentionatstudyand
testperformednumericallybetterthantheconditionwithonlydividedattentionduring
study. Participantsintheformerconditionareforcedtorelyalmostexclusivelyonfamiliarity
andindoingsoadjusttheirdecisioncriteriatomaximallycapturestudieditems.
Insum,althoughthedepletionofresourcesatstudyandtestdidnegativelyimpact
memory,itgenerallyleftboththeemotionalenhancementeffectandtheadvantageto
arousingovernonarousingitemsintact. OnecannotclaimthattheRNGtaskwastooeasy
becauseHicksandMarsh(2000)usedtheverysametaskataslowerpaceandfounda
significantreductioninrecognitionmemoryforneutralmaterials. Therefore,theseresults
suggestthattherearetasksandconditionswherethebenefitstoemotionalmaterialsurvive
evenaseveredepletionofresources.
Experiment2
Ourgoalinthisnextexperimentwastoascertainjusthowearlyintheprocessing
sequencetheemotionaladvantagetoarousingitemsoccurs. Althoughdividingattention
duringstudyhurtperformanceinExperiment1,participantsstillhadafull2satencodingto
processtheitems. Becauseshorteningthestudytimeactslikedividedattention,weexpected
worseperformancewhenitemswerestudiedfor250msascomparedwith2000msof
encodingtime. Thequestionwaswhethertheemotionalenhancementeffectwouldbe
observedwithsuchshortprocessingtime,andoverandabovethat,whetherthearousing
itemswerestillbetterrecognized. IfourconclusionsfromExperiment1areaccuratethat
theseeffectsarelargelyduetoautomaticprocesses,thentheeffectmayindeedsurvivevery
fastpresentationrates. Toprovidealargerwindowintomemoryperformance,wealsotook
remember-knowjudgmentsinthisexperiment. Valencedmaterialswilloftenberecognized
togreaterdegreewithrememberresponsesthanmoreneutralmaterial(e.g.,Kensinger&
Corkin,2003;Ochsner,2000).
RecognitionMemory 10
Method
Participants. UndergraduatesfromtheUniversityofGeorgiavolunteeredinexchange
forpartialcredittowardaresearchappreciationrequirement. Eachofthe30participantswas
testedindividuallyinsessionsthatlastedapproximately30min.
MaterialsandProcedure. ThematerialsusedinExperiment1wereusedagain,aswasthe
coreoftheprocedures. Themaindifferencewasthathalfofeachclassofitemswerestudied
for250mswhereastheotherhalfwerestudiedfor2000ms. Studydurationinthestudy
sequenceappearedrandomtotheparticipant,buttheywereinformedthatdifferentstudy
timeswerebeingmanipulated. Attest,detailedinstructionsonthedistinctionbetween
rememberingandknowingwerereadinitiallyfromthecomputermonitor. Thesewere
followedbyadetailedrestatementfromtheexperimenterinhisorherownwords. Finally,
theparticipanthadtorepeatbacktotheexperimentertheirunderstandingofthisdistinction.
Becausethiswasafairlyinvolvedprocessrequiringapproximately6min,nodistractortask
wasadministered. Duringthetestitself,participantsdecidedforeachitemwhetherthey
recollectedit,knewthatithadbeenstudied,oritwasbrandnew(i.e.,threeresponseoptions).
ResultsandDiscussion
Weanalyzedtherememberresponsesseparatelyfromtheknowresponses;andthese
×
proportionscanfoundinTable2. Intheomnibus2(encodingtime) 3(itemtype)ANOVA,
F η2
therewasamaineffectofencodingtimeonrememberresponses, (1,29)=38.31, p =.57.
Greaterencodingtimeresultedinmorerecollection. Therewasalsoamaineffectofitem
F η2
type, (2,58)=7.08, p =.20,butnointeraction. Asthereadercansee,arousingitems
receivedmorerememberresponsesthannegativenonarousingorneutralitemsdid. The
specificcontrastbetweenthetwonegativeclassesofitemswasstatisticallysignificantatboth
t t p
short, (29)=2.09,andlong, (29)=1.98, =.05,encodingdurations. Thus,inrecollectionwe
didnotfindageneralemotionalenhancementeffect,butratherfoundonlyaneffectof
arousalthatwaspresentevenattheveryshortencodingduration.
Description:Richard L. Marsh difficult divided attention task was crossed with the learning and test phases of the materials survived the depletion of cognitive resources, as did the extra For example, some argue that emotional items.