USF Libraries
USF Digital Collections

The effects of physical distinctiveness and word commonness on brain waves and subsequent memory :

MISSING IMAGE

Material Information

Title:
The effects of physical distinctiveness and word commonness on brain waves and subsequent memory : an erp study
Physical Description:
Book
Language:
English
Creator:
Kamp, Siri-Maria
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Word Frequency
Event-Related Potential
P300
Frontal Slow Wave
Recall Memory
Dissertations, Academic -- Psychology -- Masters -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: Words that deviate in their physical characteristics from their surrounding lead to enhanced recall memory, a pattern known as the Von Restorff effect. Furthermore, common (high frequency; HF) words are more likely to be recalled than uncommon (low frequency; LF) words when they occur in pure lists, while this pattern is reversed in mixed lists of both HF and LF words. This study investigated whether the Von Restorff effect and the reversal of word frequency effects in mixed lists, which may both be explained by enhanced perceived distinctiveness, are associated with common underlying brain processes. Event-related potentials (ERPs) were recorded while participants studied and subsequently recalled 70 word lists using rote memorization strategies. The three list types included (1) 14 regular-sized and one larger word, (2) 14 HF words and one LF word, or (3) 14 LF words and one HF word. The behavioral data showed a typical Von Restorff effect, a word frequency effect, as well as a reversal of the word frequency effect for LF words isolated in HF word lists ("LF isolates"). Larger words and LF isolates elicited a P300, an ERP component associated with subjective distinctiveness, whose amplitude was correlated with subsequent recall for both word types. This indicates that LF isolates were perceived as distinctive, and that this perceived distinctiveness aided subsequent recall in a similar way as for physically deviant words. Both larger words and LF isolates also elicited a left-lateralized slow wave which was larger for subsequently recalled than for not recalled words. This ERP component supposedly reflects item-to-item elaborative processes, indicating that such elaborative processes are enhanced when LF words occur in HF word list. HF words isolated in lists of LF words did not elicit comparable ERP subsequent memory effects. Rather, for these "HF isolates", the N400 was negatively correlated with subsequent recall, an ERP component that reflects semantic integration processes. We conclude that the reversal of the word frequency effect in mixed lists can be explained by a combination of enhanced subjective distinctiveness and enhanced inter-item elaborative processes for LF words that occur in lists of HF words.
Thesis:
Thesis (M.A.)--University of South Florida, 2010.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Siri-Maria Kamp.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains X pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
usfldc doi - E14-SFE0003421
usfldc handle - e14.3421
System ID:
SFS0027736:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2010 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0003421
035
(OCoLC)
040
FHM
c FHM
049
FHMM
090
XX9999 (Online)
1 100
Kamp, Siri-Maria.
0 245
The effects of physical distinctiveness and word commonness on brain waves and subsequent memory :
b an erp study
h [electronic resource] /
by Siri-Maria Kamp.
260
[Tampa, Fla] :
University of South Florida,
2010.
500
Title from PDF of title page.
Document formatted into pages; contains X pages.
502
Thesis (M.A.)--University of South Florida, 2010.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
3 520
ABSTRACT: Words that deviate in their physical characteristics from their surrounding lead to enhanced recall memory, a pattern known as the Von Restorff effect. Furthermore, common (high frequency; HF) words are more likely to be recalled than uncommon (low frequency; LF) words when they occur in pure lists, while this pattern is reversed in mixed lists of both HF and LF words. This study investigated whether the Von Restorff effect and the reversal of word frequency effects in mixed lists, which may both be explained by enhanced perceived distinctiveness, are associated with common underlying brain processes. Event-related potentials (ERPs) were recorded while participants studied and subsequently recalled 70 word lists using rote memorization strategies. The three list types included (1) 14 regular-sized and one larger word, (2) 14 HF words and one LF word, or (3) 14 LF words and one HF word. The behavioral data showed a typical Von Restorff effect, a word frequency effect, as well as a reversal of the word frequency effect for LF words isolated in HF word lists ("LF isolates"). Larger words and LF isolates elicited a P300, an ERP component associated with subjective distinctiveness, whose amplitude was correlated with subsequent recall for both word types. This indicates that LF isolates were perceived as distinctive, and that this perceived distinctiveness aided subsequent recall in a similar way as for physically deviant words. Both larger words and LF isolates also elicited a left-lateralized slow wave which was larger for subsequently recalled than for not recalled words. This ERP component supposedly reflects item-to-item elaborative processes, indicating that such elaborative processes are enhanced when LF words occur in HF word list. HF words isolated in lists of LF words did not elicit comparable ERP subsequent memory effects. Rather, for these "HF isolates", the N400 was negatively correlated with subsequent recall, an ERP component that reflects semantic integration processes. We conclude that the reversal of the word frequency effect in mixed lists can be explained by a combination of enhanced subjective distinctiveness and enhanced inter-item elaborative processes for LF words that occur in lists of HF words.
590
Advisor: Emanuel Donchin, PhD
653
Word Frequency
Event-Related Potential
P300
Frontal Slow Wave
Recall Memory
690
Dissertations, Academic
z USF
x Psychology
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.3421



PAGE 1

TheEffectsofPhysicalDistinctivenessandWordCommonnessonBrainWavesandSubsequent Memory:AnERPStudy by Siri-MariaKamp Athesissubmittedinpartialfulllment oftherequirementsforthedegreeof MasterofArts DepartmentofPsychology CollegeofArtsandSciences UniversityofSouthFlorida MajorProfessor:EmanuelDonchin,Ph.D. GeoffreyF.Potts,Ph.D. KennethJ.Malmberg,Ph.D. DateofApproval: April14,2010 Keywords:WordFrequency,Event-RelatedPotential,P300,FrontalSlowWave,RecallMemory c Copyright2010,Siri-MariaKamp

PAGE 2

Acknowledgments IwouldliketothankmyadvisorEmanuelDonchinforallofhissupportthroughoutthisstudy. Michelle-ChinQueedeservesacknowledgementforhelpingwithdatacollectionandentry.Furthermore,IwouldliketothanktheothermembersoftheCognitivePsychophysiologylaboratory, especiallyTyBrumback,GeoffreyPottsandYaelArbel,aswellasKennethMalmbergandhislab, fortheirhelpfulcommentsandsuggestionsinlabmeetingdiscussions.

PAGE 3

TableofContents ListofFigures..........................................iii Abstract.............................................iv Chapter1Introduction....................................1 1.1EffectsofDistinctivenessonMemory........................1 1.2EffectsofWordFrequencyonMemory-AnotherTypeofDistinctiveness?....5 1.2.1OperationalDenitionofWordFrequency.................6 1.2.2EffectsofWordFrequencyonCognitionandMemory...........7 1.2.3AreLFWordsDistinctive?..........................8 1.2.4DoHFWordsEnhanceItem-to-ItemElaborativeProcesses?........10 1.3TheP300,theFrontalSlowWave,andtheN400..................11 1.3.1TheP300asaMeasureofDistinctiveness..................12 1.3.2SlowWavesasIndexforWorkingMemoryProcesses............13 1.3.3TheN400asanIndexofSemanticIntegration................15 1.4CorrelationsofERPAmplitudesandMemoryforDistinctiveStimuli........15 1.5ERP-andOtherNeuroimagingStudiesonWordFrequencyEffects.........20 1.6ObjectivesandHypothesesfortheCurrentStudy..................24 Chapter2Methods......................................27 2.1Participants......................................27 2.1.1IncentivesforGoodPerformance......................27 2.2Materials.......................................28 2.2.1Stimuli....................................28 2.2.2WordLists..................................29 2.2.3RecallSheets.................................30 2.2.4IndividualWordFamiliarityQuestionnaire.................30 2.2.5StimuliandDesignoftheOddballTask...................30 2.3TaskandProcedure..................................31 2.4EEGRecordingandAnalysis.............................32 2.5StatisticalAnalysis..................................33 Chapter3Results......................................35 3.1Debrieng.......................................35 3.2BehavioralData....................................35 3.2.1RecallRates.................................35 3.2.2AnalysisoftheSubjectiveFamiliarityQuestionnaire............36 i

PAGE 4

3.2.3AnalysisofOutputOrder...........................37 3.3Event-RelatedPotentials...............................38 3.3.1PCAResults.................................42 3.3.1.1FrontalFactor...........................43 3.3.1.2RightLateralizedSpatialFactor.................45 3.3.1.3LeftLateralizedSpatialFactor..................45 3.3.1.4ParietalSpatialFactor.......................48 3.4OddballTask.....................................50 Chapter4Discussion....................................54 4.1SummaryofResults..................................54 4.2CorrelationsofERPAmplitudesandRecallintheVonRestorffParadigm.....56 4.3WordFrequencyEffectsandTheirReversalduetoListComposition........59 4.4CommonProcessesoftheVonRestorffEffectandWordFrequencyEffects?....62 4.4.1ERPComponentsandCorrelationstoRecallforLFIsolates........63 4.4.1.1ComparisonofOutputOrderforLFIsolatesandSizeIsolates..66 4.4.2ERPComponentsandCorrelationstoRecallforHFandLFStandards...67 4.4.3CorrelationsofN400AmplitudesandRecallSuccessforHFIsolates...68 4.4.4SubsequentMemoryEffectsAcrossWordTypes..............70 4.4.5ImplicationsofourDataforWordFrequencyEffectsonRecall.......70 4.5ImplicationsoftheCurrentStudyfortheFrontalSlowWave............72 4.6AbnormalitiesintheCurrentStudy'sOddballParadigm...............74 Chapter5ConclusionsandFutureDirections........................76 References............................................78 ii

PAGE 5

ListofFigures Figure1Serialpositioncurves................................36 Figure2Subjectivefamiliarityratings.............................37 Figure3Outputorderanalysis.................................38 Figure4GrandaveragesforVonRestorfflists........................39 Figure5GrandaveragesforHFlistsincludingLFisolates..................40 Figure6GrandaveragesforLFlistsincludingHFisolates..................41 Figure7SpatialFactorLoadings...............................42 Figure8Spatialfactor1....................................44 Figure9Spatialfactor2....................................46 Figure10Spatialfactor3...................................47 Figure11Spatialfactor5...................................49 Figure12GrandaverageERPsfromtheoddballtask.....................51 Figure13Therstthreespatialfactorsfromtheoddballtask................53 iii

PAGE 6

TheEffectsofPhysicalDistinctivenessandWordCommonnessonBrainWavesand SubsequentMemory:AnERPStudy Siri-MariaKamp ABSTRACT Wordsthatdeviateintheirphysicalcharacteristicsfromtheirsurroundingleadtoenhancedrecall memory,apatternknownasthe VonRestorffeffect .Furthermore,commonhighfrequency;HF wordsaremorelikelytoberecalledthanuncommonlowfrequency;LFwordswhentheyoccur inpurelists,whilethispatternisreversedinmixedlistsofbothHFandLFwords.Thisstudy investigatedwhethertheVonRestorffeffectandthereversalofwordfrequencyeffectsinmixed lists,whichmaybothbeexplainedbyenhancedperceived distinctiveness, areassociatedwithcommonunderlyingbrainprocesses.Event-relatedpotentialsERPswererecordedwhileparticipants studiedandsubsequentlyrecalled70wordlistsusingrotememorizationstrategies.Thethreelist typesincluded14regular-sizedandonelargerword,14HFwordsandoneLFword,or 14LFwordsandoneHFword.ThebehavioraldatashowedatypicalVonRestorffeffect,aword frequencyeffect,aswellasareversalofthewordfrequencyeffectforLFwordsisolatedinHF wordlistsLFisolates.LargerwordsandLFisolateseliciteda P300 ,anERPcomponentassociatedwithsubjectivedistinctiveness,whoseamplitudewascorrelatedwithsubsequentrecallfor bothwordtypes.ThisindicatesthatLFisolateswereperceivedasdistinctive,andthatthisperceived distinctivenessaidedsubsequentrecallinasimilarwayasforphysicallydeviantwords.Bothlarger wordsandLFisolatesalsoelicitedaleft-lateralizedslowwavewhichwaslargerforsubsequently recalledthanfornotrecalledwords.ThisERPcomponentsupposedlyreectsitem-to-itemelaborativeprocesses,indicatingthatsuchelaborativeprocessesareenhancedwhenLFwordsoccurinHF wordlist.HFwordsisolatedinlistsofLFwordsdidnotelicitcomparableERPsubsequentmemory effects.Rather,fortheseHFisolates,the N400 wasnegativelycorrelatedwithsubsequentrecall, iv

PAGE 7

anERPcomponentthatreectssemanticintegrationprocesses.Weconcludethatthereversalof thewordfrequencyeffectinmixedlistscanbeexplainedbyacombinationofenhancedsubjective distinctivenessandenhancedinter-itemelaborativeprocessesforLFwordsthatoccurinlistsofHF words. v

PAGE 8

Chapter1 Introduction Thegoalofthisstudyistoinvestigatesimilaritiesanddifferencesbetweentheeffectsoftwo typesofdistinctivenessonmemory.First,itiswellknownthatitemsthatarephysicallydeviant fromtheirsurroundingaremorelikelytobecorrectlyrecalledonalatertest.Furthermore,how commonlyawordoccursinalanguagealsohasveryspeciceffectsonbothrecallandrecognition memory.Bothoftheseeffectsonrecallmemorycanbeexplainedintermsofdistinctivenesstheories.Thecurrentstudycomparestheneuro-cognitiveprocessesassociatedwiththesetwotypesof distinctivenesseffectstodeterminewhethercommonprocessesunderliethetwoeffectsonrecall, orwhetherthetermdistinctivenessisusedforentirelydifferentprocessesinthetwocases. Wemeasurebrainprocessesbythemeansofevent-relatedpotentialsERPsandattempttodeterminetheconditionsunderwhichtheycorrelatewithperformanceonsubsequentrecallmemory tests.WefocusontwoERPcomponentsknownasthe P300 andthe frontalslowwave .Previous studiesindicatethattheamplitudesofthesetwocomponentsareassociatedwithsubsequentrecall underdifferentconditionstheP300whenitemsarephysicallydistinctiveandwhenparticipantsdo notuseelaborativememorizationstrategies;andthefrontalslowwavewhentheydouseelaborative strategiesandfornon-distinctiveitems.Thecurrentstudyattemptstoclarifywhichofthetwocomponentsiscorrelatedwithsubsequentrecallwhenwordsexhibitdifferentlevelsofcommonness; thatis,for wordfrequencyeffects onrecallmemory. 1.1EffectsofDistinctivenessonMemory Thequestionwhyhumansremembersomefactsandexperiences,butnotothershaspuzzled manyresearchersforalongtime.Onerepeatedlyreportedndingisthatitemsthatdifferstrongly fromtheirtemporalorspatialneighborhoodinotherwords,itemsthatare distinctive exhibit anadvantageinlatermemorytests.TheGermanGestaltpsychologistHedwigVonRestorff 1

PAGE 9

systematicallymanipulateddistinctivenessinalistlearningparadigmandshowedthat,forexample, anumberthatisstudiedinasequenceofnonsensesyllablesisbetterrememberedthanthesyllables thatarestudiedinthesamesequence.Thesamememoryenhancingeffectisobservedforisolated syllablesthatarestudiedinasequenceofnumbers,aswellasforlistscomposedinasimilarwayof othertypesofmaterialsuchasgures,coloredsquaresandlettersVonRestorff,1933.Thiseffect, whichwaslaternamedthe VonRestorffeffect orisolationeffect,isusuallystrongwhenmemoryis retrievedinfreerecalltests,butisoftenreportedtobeabsentorweakforstandardrecognitiontests McLaughlin,1968;VonRestorff,1933.ItisworthnotingthatVonRestorff'sprimarygoalwasnot tostudymemory perse .Rather,sheinterpretedherndingsintermsofgure-groundseparation ideasoriginatingintheeldofGestaltpsychology.However,animportantquestionfollowingfrom thendingofenhancedrecallfordistinctivestimuliis why humanmemoryisbetterforeventsthat standoutfromtheirenvironment. Theoriesdifferintheirclaimsonthetimepointatwhichdistinctivenessaffectsmemory.Thatis, distinctiveitemsmaybedifferentiallywellencodedatstudytime,oralternatively,distinctivenessof studyitemsmayaffectretrievalprocessesatthetimeofthememorytestMcDaniel&Geraci,2006. Thehypothesisthattheeffectsofdistinctivenessoperateatthetimeofretrievalareinpartbasedon thendingthatmemoryfordistinctiveitemsisenhancedwhenoneoftherstitemsinthelististhe distinctiveone.Atthistime,theindividualcannotyetnoticetheitem'sdistinctivenessVonRestorff, 1933,leadingtotheconclusionthattheprocessesleadingtoenhancedmemoryoccuratalatertime. Incontrast,onewidelyacceptedandappliednotionisthatdistinctivenessoperatesatstudytime becausephysicallydistinctiveitemsattractmoreattentionthannon-distinctiveitems,whichleads tobetterencodingandthereforethememoryadvantageMcDaniel&Geraci,2006;Schmidt,1991. Aswillbeexplainedlaterinthischapter,strongevidencethat,atleastinpart,distinctivenesseffects onmemoryoperatealreadyatthetimeofstudycomesfromevent-relatedpotentialERPstudies. Thus,severalpreviousERPstudiesindicatethattheamplitudesofadistinctiveness-relatedERP component,theso-called P300 elicitedatstudytime,arecorrelatedtosubsequentrecallsuccess undercertainconditions.Itisworthnotingthatthemechanismssuggestedbydifferenttheories arenotnecessarilymutuallyexclusive.Therefore,althoughthefocusofthecurrentstudyison processesoperatingatthetimeofencoding,itseemslikelythatspecicprocessesatretrievaltime alsoprovideacontributiontodistinctivenesseffectsonmemory. 2

PAGE 10

Inordertoexplaineffectsofdistinctivenessonmemory,arelevantdistinctionisbetween itemspecic and relationalprocessing Hunt&Einstein,1981;Humphreys,1976.Itemspecicprocessingreferstotheencodingoffeaturesthatareuniquetoonestudyiteminagroup.Thesemay bephysicalorsemanticfeaturesthatcanbedetectedatthetimethestimulusisencountered.Relationalprocessing,inturn,involvesrelatingthestudyitemtootherinformationinmemory.This mayincludeformingassociationsbetweenstudyitemsorconnectingastudyitemtothecontextor informationretrievedfromlongtermmemory.HuntandEinsteinreportedthatthetwotypes ofprocessingelicitedatstudytimemakeseparablecontributionstosuccessfulmemory.According totheirndings,whetherintensiveinitem-specicorrelationalprocessingismostbenecialfor subsequentmemorydependsonthenatureoftheitemsinthestudylist,suchthatunrelatedstudy listsbenetmorefromrelationalprocessinginstructionsthansemanticallyinterrelatedstudylists. Furthermore,theHuntandEinsteinreportthatoptimalmemoryperformanceisachievedwhenboth item-specic and relationalprocessingareengagedattimeofencoding. Ofinterestforthecurrentdiscussionisthenotionthatthecognitiveprocessingofadistinctive iteminvolvesenhanceditem-specicprocessing,becausethisitemexhibitsuniquefeaturesthat distinguishitfromtheotheritemsinthegroupinwhichtheyarestudied.Thatis,ifanitemisdistinctivecomparedtoitsneighborhood,moreprocessingisdirectedtowardstheseuniquefeatures. This,inturn,leadstoadistinctivememorytracethatiswelldistinguishablefromothermemory traces.However,anitemcanonlystandoutfromitsstudygroupiftherestofthestudyitemsare processedassimilartoeachother.Forexample,asyllableinalistofnumberswillstandoutand thereforelaterberecalledmoreeasily,butasyllableinaheterogeneouslistofanumber,agure,a letterandacolorwillnotVonRestorff,1933.Therefore,distinctivenesscanbedenedasprocessingdifferenceinthecontextofsimilarity,whichimpliesthatbothstrongrelationalprocessing amongallsurrounding,similaritemsaswellasenhanceditem-specicprocessingofthedistinctive itemtakesplaceHunt,2006.Animportantimplicationofthisargumentisthattheconceptofdistinctivenessreferstoapsychologicalprocessthesubjectiveprocessingofdifferenceinthecontext ofsimilarityratherthantoacharacteristicthatisintrinsictothestudyitemHunt,2006;Donchin &Fabiani,1991.Atthispointitisworthnotingthattheso-called oddballparadigm ,whichisused tostudyonebrainwavecomponentofinterestforthecurrentstudy,the P300 ,canbethoughtofas manipulatingdistinctivenessaccordingtothedenitiongivenhere.Thus,thisparadigmcontainsin3

PAGE 11

frequent,distinctiveeventsonthebackgroundoffrequent,similarevents.Theoddballparadigm andthesuggestionthattheP300canhencebeconsideredameasureofsubjectivedistinctiveness willbediscussedindetaillaterinthischapter. Hunt'sdenitionofdistinctivenessseemsintuitiveifanitemisdistinctive,itmustbedistinctive relativetosomethingelse.Forthecurrentstudyitis,however,importanttonotethatdistinctiveness isnotalwaysdenedonthebasisoftheimmediatecontextofastudyitem,butcanbedened relativetopreviousexperiences.Oneexampleofthislattertypeofdistinctivenessarewordsthat occurveryuncommonlyinalanguage.Thesewordscanbethoughtofasdistinctivecompared tootherwordsinlexicalmemory,becausetheyhavebeenpreviouslyencounteredlessfrequently, becausetheyhavelesscommonfeaturesorfewerconnectionswithotheritemsinmemoryamore detaileddiscussionofwhatmakesuncommonwordsdistinctivefollowslaterinthischapter.That is,distinctivenessinthecaseofuncommonwordsisdenedonthebasisoffewerpreviousexperienceswiththesewordscomparedtoother,morecommonwords,ratherthanbasedontheimmediate context.Itisthistypeofdistinctivenessthathasbeennamed secondarydistinctiveness Schmidt, 1991.Arelevantquestioniswhetherwordsofdifferentfrequenciesbehavesimilarinan oddball paradigm comparedtowordsthathavedeviantphysicalproperties.Thisquestionwillbediscussed laterinthischapter. Theexampleofusingtheterm distinctiveness todescribewordsuncommonlyencounteredina languagedemonstratesthatdistinctivenessisnotaunitaryattribute.Differenttypesofdistinctivenesshavebeenidentied.Thus,Schmidtdistinguishesbetween emotionaldistinctiveness processingdistinctiveness,primarydistinctiveness and secondarydistinctiveness .Asnotedbefore, ofinterestforthecurrentstudyareprimaryandsecondarydistinctiveness.Schmidtdenesprimarydistinctivenessasdistinctivenesswithrespecttotheimmediatecontext.Effectsofprimary distinctivenesscanonlybeobservedinwithinsubjectmanipulations,whennon-distinctiveitems arepresent.AtypicalexampleofeffectsofprimarydistinctivenessonmemoryistheVonRestorff effectdiscussedabove.Secondarydistinctiveness,inturn,isdenedwithrespecttothepoolof allpriorexperiences,andoneinstancearewordsthatoccuruncommonlyinalanguage,asdescribedabove.AccordingtoSchmidt,secondarydistinctivenessalsocomprisesthebizarreness effect,whichreferstothephenomenonthatundercertainconditions,bizarrestudyitemsarebetterrecalledthanplausibleonesWorthen,2006;Davidson,2006.Furtherexamplesofsecondary 4

PAGE 12

distinctivenessareirregularlyspelledwordsorpicturesofunusualfaceswhichareoftenbetterrememberedthantheircounterpartsSchmidt,1991.Anopenquestioniswhetherthedifferenttypes ofdistinctivenessarerelatedinstancesofthesamephenomenon,orwhetherthesametermisused todescribeentirelydifferentprocessesthatarerelatedonlyonaconceptuallevel.Onewayof approachingthisissueisbyinvestigatingandcomparingthebrainprocessesassociatedwithboth typesofphenomena.Thisideaisreectedintheresearchquestionofthecurrentstudy:whetherthe effectsofsecondarydistinctivenessonrecallmemory,asmanipulatedbywordcommonness,are associatedundersomeconditionswithsimilarneuro-cognitiveprocessesastheeffectsofprimary distinctivenessonmemory.Alternatively,thetwomemoryphenomenacouldbebasedonentirely dissociableprocesses. Itisclearfromthislineofthoughtthatabrainwavesignaturethatcapturestheeffectsofprimary distinctivenessonmemory,asobservedintheVonRestorffeffect,couldhelpclarifyourquestion. Thus,ifthisbrainwavewasalsorelatedtomemoryprocessesforsecondarydistinctivenesseffects, onecouldconcludethatthetwotypesofdistinctivenessaresubservedbysharedbrainprocesses. Onebrainwavethathasbeenimplicatedasameasureofsubjectivedistinctiveness-the P300 -will bethemainfocusofthisstudy.Adiscussionofthecharacteristicsandfunctionalsignicanceofthis component,aswellasitsrelationshiptomemoryprocesses,followsafteramoredetailedreviewof theoriesandndingsonwordfrequencyeffects. 1.2EffectsofWordFrequencyonMemory-AnotherTypeofDistinctiveness? Howcommonlyawordoccursinalanguage,orhowfrequentlyitisused-inshortits word frequencyisapropertyoflinguisticmaterialthatisoftenstudiedinitseffectsoncognitiveprocesses.Whenstudyingtheeffectsofwordfrequencyoncognitionandmemory,itisrstnecessary toprovideanoperationaldenitionoftheconstruct.Differentpossibilitiesforaccomplishingthis arediscussedinthenextsubsection.Afterwards,wewillreviewtheeffectsofwordfrequencyon cognitionandmemoryaswellasprovidetwogroupsofpossibleexplanationsfortheseeffects. Thesetwogroupsofpossibleexplanationsformthehypothesesinthecurrentstudy. 5

PAGE 13

1.2.1OperationalDenitionofWordFrequency Theconstructofinterestformanystudiesonwordfrequencyeffectsishowfrequentlyanitem hasbeenpreviouslyencountered.Therearedifferentkindsofstimulusmaterialthatcanbeusedto operationalizethisconstruct.Probablythemostcommonlyusedstimulussetonwordfrequencyhas beendevelopedbyFrancisandKucera.Theseauthorscountedthenumberofoccurrencesof alargenumberofwordsinabigsampleofwrittenmaterialFrancis&Kucera,1982.Usingthese frequencynorms,onecanoperationallydeneLFwordsaswordsthatoccurlessthanacertain numberoftimesinamillion.HFwordsaredenedanalogouslyaswordsthatoccurmorethan somenumberinamillion.WhileFrancisandKucera'sfrequencynormsareaverycommonlyused meansofmeasuringwordfrequency,itisworthnotingthattheyhavebeencriticizedforapotential sampling-biasduetotheinclusionofonlywrittenmaterialBalota,Pilotti,&Cortese,2001. Otherresearchershavereliedonwordfamiliaritynormstooperationalizeparticipant'sprevious exposuretowords.Probablythemostcommonlyusednormsforthispurposeweredevelopedby TogliaandBattigToglia&Battig,1987.Theseauthorshadalargenumberofparticipantsrate theirperceivedfamiliarityofwordsamongothercharacteristicsonascalefrom1 veryunfamiliar to7 veryfamiliar .Someauthorssuggestthatwordfamiliarityisamoreaccuratemeasureof frequencyofusagerepresentedinthementallexiconthanwordfrequencynormsbasedononly writtenmaterialConnine,Mullennix,Shernoff,&Yelen,1990. Thereareadvantagesanddisadvantageswitheitheroneofthesemeasuresofpriorexposure.As notedabove,FrancisandKucera'snormsconsideronlywrittenmaterial.Thefamiliarityratingsby TogliaandBattig,inturn,maybeinuencedbysemantics,imageabilityandotherpropertiesofthe wordsbecauseparticipantsmayrelyonthesefeaturestoperformtheratingsBalotaetal.,2001. Becausepilotstudiessuggestedthatbothfamiliarityandwordfrequencynormsplayanimportant role,thechoiceofstimulusmaterialforthecurrentstudyisascombinationofsubjectivefamiliarity ratingsToglia&Battig,1987andwordfrequencynormsFrancis&Kucera,1982. Onecomplicationwithusinganyofthenormstoselectexperimentalstimuliisbetweensubject variabilityofpreviousexposure.Itisreadilyapparentthatnonormpreciselyreectseverysubject's priorexposuretoeachindividualword.Tocapturetheseindividualdifferences,thecurrentstudy measuresthedegreeoffamiliarityofalargesubsetoftheexperimentalstimulibyadministeringa subjectivefamiliarityquestionnaireattheendoftheexperiment. 6

PAGE 14

1.2.2EffectsofWordFrequencyonCognitionandMemory Awell-replicatedobservationisthatwordfrequencyaffectsthespeedofaword'srecognitionas awordvs.apronounceablenon-wordinlexicaldecisiontasksMason,1976;Polich&Donchin, 1988;Forster&Chambers,1973aswellaslatenciesinwordnamingtasksForster&Chambers, 1973;Gerhand&Barry,1999.Thisdifferenceinprocessingspeeddependentonwordfrequencyis commonlyattributedtoslowerlexicalaccessofuncommon,orlowfrequencyLFwordscompared tocommon,orhighfrequencyHFwordsForster&Chambers,1973;Connineetal.,1990. Beyondthiseffectofwordfrequencyonprocessingspeed,veryspecicinuencesofwordfrequencyonmemoryhavebeenreported.Thepreciseeffectsofwordfrequencyonmemorydepend onthetypeoftestusedtoassessmemory.Probablymostcommonlystudiedisthewordfrequency effectonrecognitionmemory,referringtothewell-replicatedndingthatLFwordsarerecognized moreaccuratelyinsubsequentmemoryteststhanHFwords.Morespecically,duringrecognition memorytests,participantsaremorelikelytocorrectlyjudgeLFwordsasoldpreviouslystudied, whileatthesametimeproducingfewerfalsealarmsjudgementsofoldtopreviouslynotstudied wordsforLFwords,comparedtoHFwordsMandler,Goodman,&Wilkes-Gibbs,1982.This patternisknownasthe mirroreffect ofwordfrequencyonrecognitionmemoryGlanzer&Adams, 1990,1985;Malmberg&Nelson,2003.Qualitatively,thewordfrequencyeffectonrecognition isusuallyobtainedregardlessoftheproportionofHFandLFwordsinastudylist,butstudylist compositiondoesappeartoaffectthe magnitude ofthedifferencesinrecognitionaccuracybetween LFandHFwordsMalmberg&Murnane,2002. Forfree-andserialrecalltests,reportsofwordfrequencyeffectsarelessconsistentacrossstudies. MostresearchersreportthatwhenstudiedinpureorblockedlistscontainingonlyHForonly LFwords,HFwordsleadtohigherrecalllevelsthanLFwordsMandleretal.,1982.Inmixedlists includingbothHFandLFwords,however,thetwowordtypesarerecalledequallywellWatkins, LeCompte,&Kim,2000;Fernndez,Klaver,Fell,Grunwald,&Elger,2002;Hulme,Stuart,Brown, &Morin,2003.Otherresearchersreportthatinmixedlists,uncommonwordsaremorelikelyto berecalled,whilethisdifferencedisappearsinblockedlistsMcDaniel&Geraci,2006.Although thedatafromthesedifferentstudiesdisagreeonthespecicpatternoftheeffectofwordfrequency onrecallmemory,theyareconsistentinthatLFwordsinmixedlistsarealwaysbetterremembered thanLFwordsinblockedlists,whilethereverseistrueforHFwords.Theconsistencyofthis 7

PAGE 15

ndingacrossstudiesindicatesthatthedifferentialeffectsofwordfrequencyonrecallmemory dependstronglyontheimmediatecontextatstudytime.Thecognitiveprocessesleadingtothis abolition,orreversal,oftheHFadvantageonrecallmemorywhenmixedlistsarestudied,arethe mainfocusofthisstudy. Thus,thequestionofinterestforthecurrentstudyiswhichcognitiveprocessesarequalitativelyor quantitativelydifferentbetweenmixedandpurelists,leadingtotheabolitionorreversaloftheword frequencyeffectonrecall.Therearedifferentpossibleexplanationsforthisreversaleffect,which canbeclassiedintotwobroadgroupsforthepurposeofthisstudy.Therstclassofexplanations statesthatLFwordsare distinctive comparedtoHFwords.Thesubjectiveperceptionoftheir distinctivenesswithinthelistisenhancedinlistswhereHFwordsthatis,lessdistinctiveitems arepresent.Accordingtothisideaitistheenhancementinperceiveddistinctivenessthat,similar totheVonRestorffeffect,thenfacilitatesmemoryretrievalforLFwordsembeddedinHFword lists.Althoughthetwoproposalsarenotmutuallyexclusive,thealternativegroupoftheoriesposes thatthepresenceofHFwordsinalistfacilitateselaborativeororganizationalprocesses.According tothisidea,eachword,regardlessofitsfrequency,equallybenetsfromthepresenceofother HFwordsinthelistbecausetheseprovidemoreopportunitiesforitem-to-itemelaboration.The outcomesoftheseelaborativeandorganizationalprocessescanthenbeusedtoguideandfacilitate memoryretrievalduringtherecalltest.Therefore,bothofthesegroupsofexplanationscanaccount forthendingthatHFwordsarerecalledbetterinlistsofonlyHFwords,whileLFwordsare recalledbetterinmixedlists,wheremanyHFwordsarepresent. Amoredetailedreviewofthetwogroupsofexplanationsforthewordfrequencyeffectonrecall anditsreversalformixedstudylistsfollowsnext. 1.2.3AreLFWordsDistinctive? TheideathatLFwordsaredistinctivecomparedtoHFwordsiswidelyacceptedandmost oftenusedtoexplainwordfrequencyeffectsonrecognitionmemorywithindifferenttheoretical frameworks.Thedenitionofthistypeofdistinctivenessis,however,notnecessarilyidenticalto whatwerefertowhendiscussingtheVonRestorffeffectandwhatwesuggestcanbemeasuredby theP300seenextchapter.Therefore,toavoidconfusion,thissectionwillrefertoLFwordsas more unique ,ratherthan distinctive ,comparedtoHFwords.Infact,whetherornottheuniqueness 8

PAGE 16

thatisoftenattributedtoLFwordsandthedistinctivenessofitemsthatphysicallystandoutfrom theircontextshareunderlyingneuro-cognitiveprocessesthatcorrelatetomemoryundercertain conditionsisthemainresearchquestionofthecurrentstudy. TheobviousreasonwhyLFwordscouldbeconsidereduniqueisthatpeopledonotfrequently encounterthemintheirdailylife.Thus,LFwordshavebeenpreviouslyexperiencedlessoftenand thereforearelikelytobelessfamiliarthanHFwords.Besidesthisreadilyapparentexplanationof LFword'suniqueness,thereareanumberofothersuggestionsforcharacteristicsthatco-varywith wordfrequencyandthatcauseLFwordstobemoreunique.Thus,thefrequencyofpriorexposure toawordmaynotbetheonlyexplanationforLFword'suniquenessandtherebyfortheeffectsof wordfrequencyoncognitiveprocessessuchasrecognitionandrecallmemory. AdifferentpossibilityisthatLFwordstendtohaveunusualfeaturesthatarenotsharedby manyotheritemsinmemory.Thus,theyaremorelikelytoexhibitirregularspelling,hencecontainingmoreunusuallettercombinationsCriss&Malmberg,2008,andmoreunusualphonemes Landauer&Streeter,1973thanHFwords.AdifferentpossiblesourceforaLFword'suniquenessisthatthereisatendencythatLFwordshavebeenpreviouslyencounteredinfewerdifferent contextsthanHFwordsi.e.,theyexhibitlower contextvariability Steywers&Malmberg,2003; Adelman,Brown,&Quesada,2006. YetanotherpossibilityisthatLFwordsattractmoreattention,orrequiremoreattentionalprocessingresourcestobeencodedepisodically.Severaltheoriesexplicitlyincorporatethatincreased attentionalresourcesareallocatedtoitem-specicinformationofLFwordscomparedtoHFwords, whichmaybeanimportantinuenceonwordfrequencyeffectsGlanzer&Adams,1990;Malmberg&Nelson,2003;Criss&Malmberg,2008;DeLosh&McDaniel,1996.Thisdifferential attentionatencodingtimeisthoughttomakethememorytracemoreeasilydistinguishablefrom othermemorytraces,thatis,thememorytracesbecomecomparablyuniqueDiana&Reder,2006; Criss&Malmberg,2008;Malmberg&Nelson,2003.InadditiontoevidencefrommemorystudieswhichlimittheattentionalresourcesavailableatstudytimeDiana&Reder,2006,theidea thatLFwordsrequiremoreattentionalresourcesatencodingissupportedbytheobservationthat lexicalaccessisslowerforLFwordscomparedtoHFwordsthisideawasdiscussedintheprevioussection.ThepossibilitythatattentioniselevatedtoLFwordsatstudytimeisnotmutually exclusivewiththepossiblesourcesforaLFword'suniquenesslistedbefore.Inthecontrary,itis 9

PAGE 17

wellconceivablethatLFwordsattractmoreattention because theyhavemoreunusualfeaturesor havebeenpreviouslyencounteredinalimitednumberofcontexts. Notethatmostofthetheoryandempiricalevidencereviewedhere,suggestingthatLFwords aremoreunique,stemsfromstudiesonrecognitionmemory.However,ifitisacharacteristicof thestimulusmaterialforLFwordstobeunique,thesameargumentcanbemadefortheword frequencyeffectinrecallanditsabolitioninmixedstudylists.Althoughmosttheoriesattempting toexplainwordfrequencyeffectsonrecallarecenteredarounddifferentialfacilitationofelaborative processesasdiscussedinthefollowingsubsection,someauthorsdosuggestthatenhancedperceived uniquenessofLFwordsinmixedlists,andthereforeenhanceditem-basedprocessingofLFwords, contributestothereversalofthewordfrequencyeffectonrecalle.g.DeLosh&McDaniel,1996. 1.2.4DoHFWordsEnhanceItem-to-ItemElaborativeProcesses? OnecommonlycitedobservationisthatHFwordstendtohavemoremeaningsGlanzer& Bowles,1976,andleadtomoreassociationsthanLFwordsCofer&Shevitz,1952.Arelated ndingsuggeststhatHFwordshavefewerpre-existingassociativelinks from butnot to other wordsinlexicalmemoryNelson&McEvoy,2000.Thus,purelistsofHFwordsareoftenbelievedtohavemorepre-existingassociationswitheachotherthanpurelistsofLFwords,andmixed listsscoreinbetween.Activatingtheseassociativelinksatstudytimecanthenenhancelaterretrievalsuccess.Therefore,inpurelists,HFwordsshouldfacilitateelaborativeprocessessuchas semanticelaborationorformationofitem-to-itemassociationsatstudytimeandtherebyleadto enhancedrecallcomparedtoLFlistsMandleretal.,1982.InmixedlistsofLFandHFwords, thelevelofitem-to-itemassociationsisintermediate,leadingtoenhancedrecallforLFwords,but attenuatedrecallforHFwords,comparedtopurelistsTse&Altarriba,2007;Hulmeetal.,2003. Itisreadilyapparentthatthisaccountimplicateselaborativeprocessesbetweenitemsinalist,rather thanindividualitem-basedprocesses,asthebasisforwordfrequencyeffectsonrecall. Anothertheorythatimplicateselaborative,ororganizational,processesasthebasisforwordfrequencyeffectsonrecall,butthatdifferswithrespecttothespecicprocesses,isthe orderencoding hypothesis .ThishypothesisstatesthatLFwordsrequiremoreattentionalresourcestobeencoded episodicallyseeabove,butthatthisenhanceditemencodingcomesatacostofreducedencoding ofpresentationorderDeLosh&McDaniel,1996;Merritt,DeLosh,&McDaniel,2006.HFwords, 10

PAGE 18

inturn,donotrequireasmuchattentiondirectedtowardtheitem-specicinformation,andthereforeleavemoreresourcesavailabletoenhanceencodingofpresentationorder,aprocessthatmay beattributedtoelaboratedprocessesbetweenlistitemsorbetweenthestudyitemanditstemporal context. Whenconsideringthesuggestionsofenhanceditem-specicprocessingofLFwordsinlistsof HFwordsduetotheirgreateruniquenessorenhancedelaborativeprocessesduetothepresence ofHFwordsasexplanationsforwordfrequencyeffectsonrecall,itisworthnotingthatseveral authorsexplicitlysuggestthat both oftheseprocessesshouldcontributetotheeffectsonrecall DeLosh&McDaniel,1996;Tse&Altarriba,2007.Hence,itshouldbekeptinmindthatalthough bothofthesehypothesesaretestedinthecurrentstudy,theyarenottested againsteachother ,but independentlyofeachother.Inotherwords,itispossiblethatbothhypotheseswillbesupportedby thisstudy. Asoutlinedabove,wewilltestthetwohypothesesbyinvestigatingwhichbrainpotentialsare pronouncedandcorrelatedwithrecallsuccessforlistsofHFandLFwordswithdifferentlistcompositions.Themostrelevantbrainpotentialsforthisstudyareintroducedinthenextsection. 1.3TheP300,theFrontalSlowWave,andtheN400 Wheninvestigatingbrainprocessesassociatedwithmemoryencodingofdiscreteevents,thetechniqueofevent-relatedpotentialsERPswithitsexcellenttemporalresolutioncanprovideuseful insights,asverysimilarbehavioralpatternscanbeassociatedwithquitedifferentbrainwavecharacteristicsOtten&Donchin,2000,forexample.TheERPistherecordoftheresponses,overa periodofseveralhundredmilliseconds,ofvariousbrainstructurestodiscreteevents.Becausethe ERPistimelockedtotheelicitingeventitcanbeextractedbymeansofsignalaveragingfromthe ongoingelectroencephalographicactivityLuck,2005. ThemaingoalofthecurrentstudyistoinvestigatetherelationshipofanERPcomponentknown asthe P300 tosubsequentmemoryforphysicallydistinctivewordsandforwordsofdifferentfrequenciesintheEnglishlanguage Theinterestinthiscomponentstemsfromalargenumberof previousstudiesindicatingitsclosecorrespondencetodistinctivenessprocessingforitemsthat standoutfromtheirimmediatecontext.Therefore,therstpartofthissectionwillreviewsome characteristicsandatheoryaboutthefunctionalsignicancetheP300. 11

PAGE 19

TwootherERPcomponents,the N400 andthe frontalslowwave ,areofinterestforthecurrent study.TheN400canbeconsideredanindexofadifferenttypeofdistinctiveness,namelydistinctivenesswithrespecttosemanticexpectationsbasedonthelinguisticcontext.Thefunctional signicanceofthefrontalslowwaveisratherunclear.Itisoftensuggestedthatitreectsworking memorybasedoperations.Thiscomponenthasbeenfoundtocorrelatewithmemoryinseveral previousERPsubsequentmemorystudies,andwillthereforebeinvestigatedinthecurrentstudyas well. 1.3.1TheP300asaMeasureofDistinctiveness Ofparticularinterestforthecurrentstudyisthe P300 -anERPcomponentthatiselicitedby eventsthatresolveuncertainty,asrstreportedbySuttonandcolleaguesSutton,Braren,Zubin, &John,1965.TheP300hasapositivepolarityandpeaksatalatencyofatleast300ms,with amaximumatparietalrecordingsites.TheP300ismostcommonlystudiedintheso-called oddballparadigm Donchin,1981.Inthisparadigm,twotypesofstimuliarepresentedinarandom sequence.Onetypeoccursfrequently,andtheothertypeoccursratherinfrequently.Thesubject isengagedinataskthatrequirestheclassicationofthestimuliaccordingtothetwotypes.Undertheseconditions,theinfrequentstimuluselicitstheP300.Notethattheoddballparadigmand theVonRestorffparadigmshareimportantcharacteristicssuchthatinbothtasks,aninfrequent oddballitemorisolateoccurs,thattheparticipantislikelyperceiveasdistinctive. TheamplitudeoftheP300isnegativelycorrelatedwiththefrequencyofoccurrenceoftherare stimulus.Thatis,themorerarelytheinfrequentstimulusoccurs,thelargeristheamplitudeof theP300Donchin,1981;K.C.Squires,Wickens,Squires,&Donchin,1976;Duncan-Johnson& Donchin,1977.Furthermore,theshapeoftheP300isnotnecessarilydeterminedonlybytheglobal frequencyoftherareevent,butratherbythesubject's perception ofthefrequencyofoccurrenceat thetimethestimulusisencounteredK.C.Squiresetal.,1976;Duncan-Johnson&Donchin,1977. Therefore,ithasbeenproposedthattheP300canbeusedasameasureofsubjectivedistinctiveness Donchin&Fabiani,1991. The contextupdatingtheory isoneinuentialhypothesisonthefunctionalsignicanceofthe P300thatwasdevelopedintheearly80'sDonchin,1981.Thishypothesisassumesthathumans maintainamentalmodelorschemaofallgoal-relevantinformationaboutthecurrentsituation 12

PAGE 20

andtheirenvironment.Wheneveraneventoccursthatisnovelorinconsistentwiththeschema, themodelneedstoberevised.Thenovelinformationhastobeintegratedintothemodelwhile takingintoaccountinformationstoredinlong-termmemory.Accordingtothetheory,wheneverthis contextupdatingprocessoccurs,aP300iselicited.Thesecognitiveprocessesthataresupposedly associatedwiththeP300canbedescribedas strategicprocessing Donchin,1981.Thatis,rather thanmerelyreactingtotheperceivedevent,theseprocessesaffectthewaythesubjectwillreactto futureevents.Thedevelopmentofthecontextupdatingtheoryledtospecicpredictionsaboutthe consequencesoftheP300.OneofthesepredictionsisthatifalargeP300iselicitedbyadistinctive eventindicatingthatastrongupdatingprocessoftheschemaoccursthisshouldincreasethe probabilityoflaterrecallsuccessDonchin,1981. Toexplainthispredictionmorespecically,itisknownthatitemsthataredistinctivefromtheir contextwillelicitaP300.TheamplitudeoftheP300,however,variesbetweentheencountersof differentdistinctiveitems,thatis,somedistinctiveitemswillelicitalargerP300thanothers.As describedabove,thepreciseamplitudeonagiventrialcanbeconsideredameasureofthestrength ofthecontextupdatingprocessneededtointegratethisnovelinformation,orameasureof subjective distinctiveness .Itisfurtherknownthatbehaviorally,distinctiveitemsaretypicallymorelikelyto berecalledtheVonRestorffeffect.Thepredictionarisingfromthecontextupdatinghypothesisis thattheamplitudeoftheP300elicitedbyspecicdistinctiveeventsasanindicationofhowmuch theschemahastobeupdatedtointegratethisnewinformationshouldbecorrelatedtowhetheror nottheeventwillbesuccessfullyrememberedinthefuture.Inotherwords,trial-by-trialvariability oftheP300asitisfoundinanoddballparadigmortheVonRestorffparadigm,shouldcorrelate totheprobabilityofsubsequentrecall.Evidenceforpreciselythispatternofresultsisreviewedin thesectionfollowingthereviewoftheslowwaveandtheN400. 1.3.2SlowWavesasIndexforWorkingMemoryProcesses SlowlyincreasingordecreasingpotentialsintheERPshaverstbeenobservedinthemid-70's. TheseearliestreportsindicatedthatapositiveslowwavepotentialfollowedtheP300indifferentinstancesoftheoddballtaskN.K.Squires,Squires,&Hillyard,1975;Duncan-Johnson&Donchin, 1977.Inthesestudies,theslowpositivewaveshowedaparietalmaximumandwaspresentonly forstimulithatalsoelicitaP300.ParietallydistributedpositiveBosch,Mecklinger,&Friederici, 13

PAGE 21

2001andnegativeRuchkin,Johnson,Canoune,&Ritter,1991;Deldin,Deveney,Kim,Casas,& Best,2001slowwaveshavesincebeenobservedinvariousfurtherstudies.Inadditiontothese parietallydistributedslowwaves,frontalnegativeBoschetal.,2001andpositivityRuchkinet al.,1991slowwaves,sometimeswithaleft-orrightlateralizedmaximum,havebeendiscoveredin differentexperimentalparadigms.Tothisdate,theprecisefunctionalsignicanceoftheslowwaves withdifferentspatialdistributionsremainscontroversial.Avarietyofcognitiveprocesseshavebeen proposed,suchaselaborativeprocessesperformedonstimuliDeldinetal.,2001,resourceallocationandmemory-relatedprocessesRuchkinetal.,1991orworkingmemoryrelatedprocesses Deldinetal.,2001;Boschetal.,2001.Itisnotclear,however,whetherslowwaveswithdifferent spatialdistributionselicitedindifferentexperimentalparadigmsservesimilarcognitivefunctions, orwhethertheyareentirelyseparablefromeachother.Therefore,thefollowingdiscussionwillfocusonlyonslowwavesthataremaximalinfrontalelectrodes,becausethisisthespatialdistribution relevantforthecurrentstudy. Thus,duetothendingsofprevioussubsequentmemorystudiesutilizingasimilarparadigmas thecurrentstudyseenextchapter,afrontallydistributedslowwaveisofbiggestinterest.Based onitsfrontalspatialdistribution,itislikelythatthisfrontalslowwavecomponentoriginatesin prefrontalareas.Theprefrontalcortexhasbeenrepeatedlyimplicatedincognitivecontrolprocesses, amongothersinworkingmemoryprocessesforareview,seeMiller,2000. Ruchkinandcolleaguesreportedprefrontallydistributedslowwavesinmentalarithmeticand mentalrotationtasks,whichincreasedinamplitudewhenthetaskdifcultyincreasedRuchkinet al.,1991.Boschandcolleaguesfoundthatasimilarfrontalslowwavevarieddependentonwhether verbal,object-basedorspatialinformationwastobeheldinworkingmemoryBoschetal.,2001. Furthermore,MuenteandcolleaguesconductedanERPstudyinwhichsentencesassociatedwith highandlowworkingmemorydemandwerepresentedMnte,Schiltz,&Kutas,1998.Aslowly increasingleftanteriornegativitywaslargerwhenworkingmemorydemandwashighcomparedto whenitwaslow.Thiseffectwasmodulatedbyindividualsubject'sworkingmemorycapacity. Theseandotherstudiesareconsistentwiththesuggestionfrontallydistributedslowwavesareassociatedwithsomeaspectofworkingmemoryoperations.Itisworthnoting,however,thatfrontal slowwaveshavealsobeensuggestedtoplayaroleinothercognitiveoperations,amongothers inretrievalofinformationfromlongtermmemoryMecklinger,inpressoremotionprocessing 14

PAGE 22

Diedrich,Naumann,Maier,&Becker,1997.Therefore,theprecisecognitiveprocessesmanifestedinthefrontalslowwaveremaintobeelucidated. 1.3.3TheN400asanIndexofSemanticIntegration AnERPcomponentthathasbeenlinkedtotheeffortnecessarytosemanticallyintegratenewly encounteredwordsintothecontextisthe N400 Kutas&Hillyard,1980 Interestinglyforthis study,thiscomponentcanalsobeconsideredasameasureofsemanticexpectancyviolationand maythereforebeconsideredasreectingadifferenttypeofdistinctivenessFabiani,2006.The N400hasanegativepolarity,andusuallypeaksaround400msafterthesemanticallyunexpected stimulus.Intheoriginalstudy,theN400waselicitedbythenalwordinasentence,andwas largeforsentenceendingsthatdidnotmeetsemanticexpectationsforexample,Hespreadthe warmbreadwithsocks.Kutas&Hillyard,1980.SubsequentstudiesindicatedthattheN400 issmallerforwordsthatareinthesamesemanticcategoryastheexpectedword,whileitislarger forwordsthatareofdifferentsemanticcategoriesFedermeier&Kutas,1999.Thisndingis consistentwiththenotionthattheN400isameasureoftheeffortneededtointegrateawordintoits semanticcontext.WhiletheoriginalstudiesontheN400usedentiresentences,laterstudiesfound thatanN400canbeelicitedintheabsenceofasententialcontext.Thatis,anN400isalsoelicited bythesecondwordinapairwhenthiswordissemanticallyunrelatedtotherstonecf.Kutas, VanPetten,&Kluender,2007. ItisinterestingandimportantforthecurrentstudythatafewstudiesindicatedthatwordfrequencyisnegativelycorrelatedwiththemagnitudeoftheN400.Thus,LFwordshavebeenreported byseveralauthorstoelicitlargerN400amplitudesthanHFwordsVanPetten&Kutas,1990;Barber,Vergara,&Carreiras,2004;Allen,Badecker,&Osterhout,2003;Rugg,1990.Thiseffect appearstobeespeciallypowerfulintheabsenceofstrongsemanticexpectations,developedfor examplefromthesententialcontext,whichmayconstrainsemanticexpectationstospecicsubsets ofwords. 1.4CorrelationsofERPAmplitudesandMemoryforDistinctiveStimuli AsequenceofERPstudiesutilizingtheVonRestorffparadigminvestigatedthepredictionthat P300amplitudesarecorrelatedwithsubsequentmemorysuccess.IntherststudyKaris,Fabiani, 15

PAGE 23

&Donchin,1984,participantsstudiedlistsof15wordsandimmediatelyaftereachlistcompleted afreerecalltest.Inmostlists,onewordwasprintedinalargerfontsizethanallotherwordssize isolate.Betweensubjects,differentbehavioralpatternswerefound,suchthatonegroupshowed aVonRestorffeffectbutaratherpooroverallmemoryperformance,whileanothergroupdidnot showaVonRestorffeffectbutoverallperformedmuchbetter,andathirdgroupscoredinbetween. Theauthorsfoundthatthegroupsdifferedwithrespecttothememorizationstrategiestheyhadused atstudytime.Therstgrouphadsilentlyrepeatedeachwordseveraltimesafteritspresentation rotememorization.Thehigh-performinggroupwithnoVonRestorffeffect,inturn,hadused elaborativestrategiessuchasconnectingthestudywordsintosentences.Forallgroups,aclearP300 waselicitedbytheisolates,butonlyfortherotememorizationgroupwasitsamplitudecorrelated withsubsequentrecall.Forthegroupthathadusedelaborativestrategies,anotherERPcomponent the frontalpositiveslowwave waspronounced,andthiscomponentwaslargerforsubsequently recalledthansubsequentlynotrecalledwords.KarisandcolleaguesconcludedthattheP300as anindexofearlycognitiveprocessingofastimulusisonlyrelatedtosubsequentrecallwhenno additional,elaborativeprocessingofthewordoccursafterwards. SubsequentstudiesfurtherinvestigatedwhetherthecorrelationbetweenP300amplitudeandsubsequentrecallistiedtotheabsenceofelaborativememorizationstrategies.InonestudyFabiani, Karis,&Donchin,1986,subjectsperformedanoddballtasksinwhichtheyclassiednamesaccordingtotheirgender.Atstudytime,participantswerenotinformedthatafreerecalltestwould follow,sothatparticipantswereunlikelytouseelaborativestrategies.Infrequentitemsassociated withalargeP300atstudytimeweremorelikelytoberecalledthatitemsassociatedwithasmaller P300.Thisndingcorroboratestheoriginalideathatwhenelaborativememorizationstrategiesare notused,P300amplitudeiscorrelatedtosubsequentrecall.Notethatinthisstudy,therewasno evidenceforafrontalslowwave,furthersupportingthenotionthatthisERPcomponentmaybe associatedwithelaborativeprocessesFabianietal.,1986. AnotherstudyusedasimilarparadigmasKarisandcolleagueson11-yearoldchildrenFabiani, Gratton,Chiarenza,&Donchin,1990.TheauthorsfoundthatchildrenshowedstrongVonRestorff effectsandastrongcorrelationbetweentheP300andsubsequentrecall.Aschildrenarenotlikely toengageinelaborativeprocessingstrategies,theauthorsreasoned,thisndingagainindicatesthat intheabsenceofelaborativestrategies,theP300iscorrelatedwithsubsequentrecall. 16

PAGE 24

ThegoalofafourthstudyinthisseriesFabiani,Karis,&Donchin,1990wastodirectlyinvestigate,withinsubjects,theinteractionsbetweenP300,subsequentmemoryandrehearsalstrategies. Foreachlist,participantswereinstructedtouseoneofthetwotypesofstrategies.Whenparticipants usedrotememorizationstrategies,therewasastrongVonRestorffeffectaswellasacorrelation betweenP300amplitudeelicitedbydistinctiveitemsandsubsequentrecall.Whenparticipantsused elaborativestrategies,afrontalpositiveslowwavewasprominentandcorrelatedtorecall.The resultsconrmthatwhenrotememorizationstrategiesareused,theP300isrelatedtosubsequent recall,butwhenelaborativestrategiesareused,arelationshiptorecallisfoundforthefrontalslow wave. ThenextstudyinvestigatedtheimpactofsemanticandphysicaldistinctivenessontherelationshipbetweenP300andrecallFabiani&Donchin,1995.Subjectswereinstructedtouserote memorizationstrategiestostudywordlists,whileatthesametimeeithercompletingaphysical taskjudgingthefontsizeofeachword,orasemantictaskjudgingwhetherornottheitemisan actualwordoranon-word.Ineachlistoneitemwasdistinctwithrespecttosemanticclasssemanticisolate,andonewordwasdistinctwithrespecttoitsfontsizesizeisolate.Forboth groups,theP300inresponsetophysicalisolateswascorrelatedtosubsequentsuccessofrecall.The semanticisolateelicitedasignicantP300onlyinthegroupcompletingthesemantictask,andthis P300wascorrelatedtosubsequentrecallaswell.Furthermore,theauthorsalsofoundthatsemantic isolateselicitedanN400.However,itsamplitudewasnotcorrelatedwithsubsequentrecallFabiani &Donchin,1995. Thelatterstudyalsousedarecognitiontest,whichindicatedthattherelationshipbetweenP300 andmemorycanbefoundinarecognitionparadigmifthememorytestusestheoriginalpresentation styleduringtestandtherebyreinstatesthestudycontextofstudy.FabianiandDonchinadditionally investigatedhowlikelyitisthatadistinctivestudyitemiswrittendownrstorlastintherecall phase,asopposedsomewheretowithinthebulkofnon-distinctivestudyitems.Theyfoundthat semanticisolatesweremostlikelytoberecalledrst,whereasphysicalisolatesaremostlikelytobe recalledlast.Thisndingindicatesthatthesetwotypesofdistinctivenessareprocesseddifferently, whilebothtypesofdistinctivenessseemtobeavailableasaidsfortheorganizationofthelistitems inmemoryFabiani&Donchin,1995. 17

PAGE 25

AnotherstudyinthisseriesinvestigatedhowtherelationshipbetweenP300andsubsequentrecallisaffectedbythemannerinwhichphysicaldistinctivenessisinducedOtten&Donchin,2000. Thus,thecorrelationmaybeduetoageneralprocessforalldistinctivenessattributes,oritmaybe onlypresentwhendistinctivenessisintegraltothestudyitem.Integraldistinctivenesswasmanipulatedbyfontsizesizeisolate,whilenon-integraldistinctivenesswasmanipulatedbydrawinga near,orafarframearoundtheword.Theresultsindicatedthatbehaviorally,integraldistinctiveness hadthesameenhancingeffectonrecallperformanceasnon-integraldistinctiveness.However,the ERPdataindicatedthatthesizeisolateelicitedlargeP300components,andonlyforthistypeof isolatewastheamplitudeoftheP300correlatedwithsubsequentrecall.Thewordssurroundedby afarframe,inturn,showedasmallerP300thatwasnotcorrelatedtosubsequentrecall.Instead, thefrontalslowwaveshowedacorrelationtomemory.Inanattempttoexplainthisnding,the authorsspeculatedthattheP300maybemorepronouncedandtheamountofcontextupdatingmay bemorestronglyrelatedtosubsequentrecall,whenthedistinctivenessfeaturehastobenecessarily processedduringencodingthatis,ithastobeintegraltothestudyitem. Oneshortcomingofallstudiesreviewedsofaristhattheyallusedonly10orlesselectrodesdue tothelimitedtechnologyatthetimetheywereconducted.TwosubsequentstudiesLian,Goldstein, Donchin,&He,2002;Ding,Goldstein,Lian,Donchin,&He,2002usedrecordingsfroma129sensordenseelectrodearrayinaparadigmanalogoustoKarisetal..Theyappliedacortical imagingtechniquetomaptheERPdataontolikelycorticalregionsoforigin.Theresultsindicated thatfortherotememorizationcondition,theERPsubsequentmemoryeffectwasmappedonto theleftinferiorprefrontalcortex,theleftmedialtemporallobeandtheleftparietallobe.Forthe elaborativestrategycondition,activationsintherightparietalandleftprefrontallobesdifferedin magnitudebetweensubsequentlyrecalledandsubsequentlynotrecalleditems. Thecombinedresultsofthesestudiesledtothesuggestionthatprocessingofdistinctiveitems inaVonRestorffparadigmcanbedescribedinthreephases.First,duringaninitial,feature-driven encodingofthedistinctiveitem,aP300iselicitedwhoseamplitudevariesaccordingtotheamount ofcontextupdatingthatiselicited.Thiscontextupdatingprocessinvolvesmarkingthedistinctive attributetowardsthememoryrepresentation,suchthatthestrongertheupdatingprocessandthereforethelargertheP300,thestrongerthedistinctivenessattributeismarked.Thisfeature-driven representationiscreatedindependentlyofmemorizationstrategy.Duringthesecondphase,theuse 18

PAGE 26

ofdifferentrehearsalstrategiesoperatesonthememoryrepresentation.Ifparticipantsusememorizationstrategieslikeformingsentenceswiththewords,anelaboratedrepresentationiscreated containingtheresultsofthisprocess.Thiselaboratedrepresentationinvolvesformingassociations betweenstudyitems.Thetheoryleavesopenwhethertheelaboratedrepresentationisindependentofthefeature-drivenrepresentationorwhethertheoriginalrepresentationisaltered.Thethird phaseistheretrievalphase,inwhichsubjectsareabletousedifferentkindsofretrievalaids.For rotememorizers,thedistinctiveattributeoftheisolatecanbeeffectivelyusedasacueforretrieval. Iftheoriginalrepresentationwaselaboratedoninthesecondphase,however,thedistinctivenessattributemaybelessbenecialasaretrievalcue,becausetheorganizationinducedbytheelaboration isstrongerandmoreefcientforretrieval.Itisforthisreasonthatelaborativerehearsalattimeof studyisassociatedwithareduction,orabsence,ofthecorrelationbetweentheP300andsubsequent recallFabiani&Donchin,1995;Donchin&Fabiani,1991. Relatedtothistheory,onepossibilityisthattheupdatingprocessassociatedwiththeP300is correlatedtorecallwhentheactualstudyepisodeisretrieved.Thefrontalslowwave,inturnmay berelatedtosubsequentmemorywhenworkingmemoryoperationsinthefrontallobeareactivated, whichenablerelationalprocessingprovidingeffectivecuesforretrievalDonchin&Fabiani,1991. Notethatotherauthorssuggesttheoppositepattern:thatthefrontalslowwaveiscorrelatedwith memorywhencontextualdetailisretrievedwhiletheP300showsacorrelationwhennocontextual detailisretrievedKim,Vallesi,Picton,&Tulving,2009.Therefore,thenatureofthedissociation betweencorrelationsofthetwobrainwavestomemoryremainstobeelucidated. Manystudiesinotherlaboratoriesprovideconvergingevidenceformorepositive-goingERPs forsubsequentlyrememberedcomparedtonotrememberedevents,eitherintheP300-orinthe slowwaverange.Thus,severalotherstudieshavefoundcorrelationsbetweenP300amplitudes Paller,Kutas,&Mayes,1987;Azizian&Polich,2007;Voss&Paller,2009,slowwaveamplitudesMangels,Picton,&Craik,2001;Otten,Sveen,&Quayle,2007,orbothunderdifferent conditionsKimetal.,2009;Wiswede,Rsseler,&Mnte,2007tosubsequentmemoryindifferentexperimentaldesigns,suchastraditionalstudyofsequentiallypresentedwordlists,orpaired associatestasks.Whilethesestudiesdidnotspecicallyinvestigatetheeffectsofphysicalisolation onmemory,theydoprovideconvergingevidencesupportingthegeneralizabilityoftheclaimsabout cognitiveprocessesassociatedwithP300andslowwaveandtheirrelationshiptosubsequentrecall. 19

PAGE 27

Asdiscussedabove,thetermdistinctivenessiscommonlyusedtoexplaindifferentphenomena. ThecurrentstudywillbuildonpreviousstudiesontheVonRestorffparadigmandinvestigatehow physicalisolationcomparestoisolationduetodeviantwordfrequency.Thatis,wewillinvestigate whetherwordsofdeviantfrequencyelicitaP300andafrontalslowwave,andfurthermorewhich oneoftheelicitedcomponentsiscorrelatedwithsubsequentrecall.Thereby,wewilldetermine whethertheeffectsofthetwotypesofdistinctivenessonmemoryaresupportedbysimilar,or separable,neuro-cognitiveprocesses. Therefore,ofinterestforthecurrentstudyarepreviousstudiesontheeffectsofwordfrequencyon recallmemory,aswellasthecorrespondingbrainwavecorrelates.AreviewoftheERPcorrelates ofwordfrequencyeffectsonmemoryfollowsnext. 1.5ERP-andOtherNeuroimagingStudiesonWordFrequencyEffects Thissectionwillprovideareviewofneuroimagingstudiesontheeffectsofwordfrequencyon neuro-cognitiveprocesses,aswellastheircorrelationstosubsequentrecall.First,itisrelevant whetherLFwordscanundercertaincircumstanceselicitaP300,whichshouldbethecaseifLF wordsareperceivedasdistinctiveinasimilarwayasitemsthatphysicallystandout. PolichandDonchindesignedanoddballparadigminwhichtheparticipantsjudgedeachstimulus inasequencebywhetheritisawordorapronounceablenon-word lexicaldecisiontask Polich& Donchin,1988.Theprobabilityofwordsandnon-wordsoccurringonanygiventrialwasmanipulatedbetweenexperimentalblocks.Ofthewords,50%wereLFwordsand50%wereHFwords ineveryblock.ThendingsindicatedthatbothLFandHFwordselicitedaP300,whichwasnegativelycorrelatedwiththeprobabilityofanactualwordoccurringonanygiventrial.Furthermore, theP300elicitedbyLFwordswassmallerinamplitudeandshowedalongerlatencycompared totheP300elicitedbyHFwords.Theauthorsinterpretedthisndingintermsofthecognitive capacityneededforprocessingcommonanduncommonwords:uncommonwordsrequiremore processingcapacityand-timefortheiridenticationleadingtothereductioninP300amplitude andtheincreaseinP300latencyKutas,McCarthy,&Donchin,1977.Notethatalthoughword frequencyhadanimpactoncharacteristicsoftheP300,theprobabilityofanactualwordtooccur withinablockwasamuchstrongerpredictorofP300amplitudePolich&Donchin,1988. 20

PAGE 28

ThisndingsuggeststhatastrongerP300isnotnecessarilyelicitedbyLFwordswhencompared toHFwords.Thus,consideringLFwordsasdistinctivewithoutregardtotheircontextdoesnot appearvalidwhensubjectivedistinctivenessisdenedastheprocessesmanifestedintheP300 althoughotherdenitionsofdistinctivenessmaystillholdforLFwords.Themoreimportant questionofthisstudy,however,iswhetherLFwordsareidentiedasdistinctivewhentheyoccur withalowerprobabilityinsequencesofHFwords.Thus,itispossiblethatthedirectcontrast withHFwordsbringsouttheLFwords'distinctiveness.Astudysystematicallyinvestigatingthis issueusingtheoddballparadigmorasimilardesignisnotknowntous.However,severalprevious studieshaveaddressedrelatedquestionsandthereforeareworthreviewing. AseriesofstudiesbyRuggandcolleaguesinvestigatedERPselicitedbyLFandHFwordsin lexicaldecisionparadigmsinwhichwordswererepeatedlypresentedRugg,1990,orinwhich arecognitiontestwasadministeredinasecondphaseRugg,Cox,Doyle,&Wells,1994;Rugg &Doyle,1992.TherststudysuggestedthatrepeatedpresentationsofLFwordselicitedlarger parietalpositivitiesintheP300latencyrangecomparedtorstpresentations,andingthatwasnot obtainedforHFwordsRugg,1990.Similarly,thesubsequentstudiessuggestedthattheold/new effecti.e.morepositivegoingERPsforpreviouslystudiedwordsthanfornewwordsislargefor LFwordsbutabsentforHFwordsRuggetal.,1994;Rugg&Doyle,1992.ThereisamethodologicaldifcultywheninterpretingthesendingssuchthatallthreestudiesmeasuredERPamplitudes byaveragingoverlargeintervalsofdurationsbetween100msand400ms.Withthistypeofanalysis,onecannotclearlydisentangledifferentERPcomponents,preventinganydeniteconclusions aboutthecomponentialnatureunderlyingtheirndings.Thesestudiesdo,however,providesome indicationsthatP300amplitudeselicitedbyrepeatedlypresentedwordsmaybelargerforLFwords thanHFwords.Asubsequentmemoryanalysiswasnotthegoalofanyofthesestudies,preventing anyconclusionsaboutpossiblesubsequentmemoryeffects. ExperimentsinvestigatingsubsequentmemoryeffectsforLFandHFwordsaremostinteresting forthecurrentstudy.ThemostimportantquestioniswhetheramplitudesoftheP300orthefrontal slowwavearecorrelatedwithsubsequentrecallandunderwhichconditions.However,sofar,only twostudieshavedirectlyinvestigatedERPsubsequentmemoryeffectselicitedbyLFandHFwords inarecallmemoryparadigmFernandezetal.,1998;Guo,Zhu,Ding,Fan,&Paller,2004. 21

PAGE 29

Theearlierstudy'sobjectivewastomanipulatedistinctivenessbywordfrequency,suchthatLF wordswerea-prioriconsidereddistinctiveFernandezetal.,1998.Underrote-memorizationinstructions,participantsstudiedlistsof15wordspresentedsequentially,whileERPswererecorded. ThewordlistseithercontainedeitheronlyLForonlyHFwordsblockedlists,orabouthalfof thewordsinalistwereLFandhalfofthewordswereHFwordsmixedlists.TheERPswere collapsedacrosslisttypes,becauseaccordingtotheauthorsthetwolisttypesdidnotelicitdifferencesinERPs.Interestingly,alatepositivecomplexthatis,aP300waselicitedbyLFwords only,whichwasusedasanindicationforthesuccessofthedistinctivenessmanipulation.TheamplitudeoftheP300was,however,notcorrelatedwithsubsequentrecall.Theexplanationforthe discrepancybetweenthendingsofthisstudyandtheseriesofstudiesontheP300andVonRestorff paradigmexperimentsisunclear.ItisconceivablethattheP300inthisstudywasdrivenbyonly asmallnumberofLFwordsthatthesubjectsdidnotknowandthereforestoodoutfromthelist. However,thesewordsthendidnotleadtoarecalladvantageduetothelackofsemanticcontent sinceunknownwordsmybeperceivedasnon-words.Fernandezandcolleaguesdidndtwosubsequentmemoryeffects,onethatwaspresentforbothHFandLFwords,andonethatwasonly presentforHFwords.Duetotheuseofsubtractiontechniquesandalackoffurtherinvestigating thecomponentialnatureoftheirndings,itisdifculttointerpretwhichcomponentswereinvolved inthesesubsequentmemoryeffects. Anotherstudyinvestigatingsubsequentmemoryeffectsforwordsofdifferentfrequenciesmanipulatedencodingtaskinastudy-testrecognitionmemoryparadigmGuoetal.,2004.Inthisstudy, subsequentlyrememberedLFwordswereassociatedwithalargercentrallydistributedpositivityin theP300timewindowcomparedtonotrecognizedLFwords.Suchasubsequentmemoryeffect wasnotpresentforHFwords.Again,onecomplicationofthisstudyistheusemeanamplitude measuresovertimewindowsover200ms,preventingdeniteconclusionsaboutthecomponential natureoftheseeffects.However,thisstudy'ssuggestion-thepresenceofacorrelationofP300to subsequentmemoryforLFwords-appearstocontradicttheresultsobtainedbyFernandezandcolleagues.Thereareafewdifferencesinmethodologybetweenthetwostudiespossiblycontributing tothedifferentresults,amongotherstheuseofarecognitiontestbyGuoandcolleaguesasopposed toarecalltestbyFernandezandcolleagues.Overall,theconictingresultsbetweenthetwostudies implythatthereisaneedforfurtherstudiesclarifyingunderwhichconditionsP300andfrontal 22

PAGE 30

slowwaveamplitudesarecorrelatedwithsubsequentrecallforwordfrequencyeffects.Another questionleftopenbybothofthesestudiesistheimpactofthelistcompositiononwordfrequency andsubsequentmemoryeffectsofERPs.Thelatterisoneofthegoalsofthecurrentstudy. Asdiscussedabove,anotherERPcomponentthatvarieswithwordfrequencyistheN400,whose amplitudeisinverselycorrelatedwithwordfrequencyVanPetten&Kutas,1990;Barberetal., 2004;Allenetal.,2003;Rugg,1990.Toourknowledge,nostudieshaveinvestigatedapotential correlationbetweenitsamplitudeandsubsequentmemoryinaparadigmcomparingLFandHF words.However,FabianiandDonchindidnotndacorrelationbetweenN400andsubsequent recallforwordsthatbelongedtoadeviantsemanticcategoryFabiani&Donchin,1995.Therefore, inourstudywedidnotexpectanycorrelationsbetweenN400andrecallwhenmanipulatingword frequency. Anothermethodofstudyingthebrain'sresponsetostimuli,whichisalsocommonlyusedin subsequentmemoryparadigms,isfunctionalmagneticresonanceimagingfMRI.Thetemporal resolutionofthismethodisratherpoorcomparedtoERPs,althoughitbearsagreatspatialresolution.AnotherpropertyworthnotingisthatfMRImeasuresbloodoxygenationlevelsindifferent partsofthebrain,whichisanindirectmeasureofbrainactivity.ERPs,inturn,measureelectrical potentialsemanatedfromneuronassembliesinthecortex.ItisimportanttokeepinmindthesedifferenceswhenintegratingndingsfromERPandfMRIstudies.Nevertheless,usefulinsightscan begainedifndingsfromfMRIandERPstudiesconvergeonthesame,oranalogous,conclusions. OnepreviousstudyinvestigatedsubsequentmemoryeffectsofLFandHFwordswitheventrelatedfMRIChee,Westphal,Goh,Graham,&Song,2003.Theresultsofthisstudyindicated thatLFwordsinanincidentallearningparadigmwereassociatedstrongerbloodoxygenationin leftprefrontalregionsthanHFwords.Furthermore,thisprefrontalactivationwasstrongerforLF wordsthatwerecorrectlyrecognizedinasubsequentmemorytestcomparedtosubsequentlynot recognizedLFwords.KeepinginmindtheprecautionsdiscussedaboveforcomparingERPand fMRIdata,onecanspeculatethatthesubsequentmemoryeffectinprefrontalareasreportedby CheeandcolleaguescorrespondstotheslowwaveactivityobservedinpreviousERPstudies.Itis importanttonote,however,thatthestudybyCheeandcolleaguesusedaverydifferentexperimental design,suchthatrecognitionmemorywastested,andthememorytestoccurredunexpectedly. 23

PAGE 31

Overall,previousresultsindicatethatLFwordsdonotnecessarilyelicitaP300component,but cansometimesdosodependingonthedetailsoftheexperimentaldesign,suchastheLFwords' immediatecontext.FewstudieshaveinvestigatedsubsequentmemoryeffectsofLFandHFwords, andthosethatdidprovidedconictingresults.Itisthereforeunclearuptothispoint,whether theP300orthefrontalslowwavearecorrelatedwithsubsequentrecallwhenwordfrequencyis manipulated,andhowthepreciseexperimentalconditionsaffectthesecorrelations. 1.6ObjectivesandHypothesesfortheCurrentStudy ThemaingoalofthisstudyistodeterminewhichERPcomponentsarecorrelatedwithsubsequentrecallforLFandHFwords,whentheyoccurinlistsofmostlysame-frequencywordsand whentheyoccurisolatedinlistsofopposite-frequencywords.Theobtainedcorrelationbetween ERPcomponentsandsubsequentrecallwillbecomparedtothesubsequentmemoryeffectsfor physicallydistinctivewordstoclarifywhethereffectsofprimarydistinctivenessandsecondary distinctivenessonrecallrefertosimilarneuro-cognitiveprocessesundercertainconditions.Specifically,thequestioniswhetherthereversalofwordfrequencyeffectsonrecallwhenmixedlistsare studiedisduetoenhancedperceiveddistinctivenessofLFwordsinlistsofHFwordscomparableto processeselicitedbyphysicaldevianceorwhetheritisduetoenhancedbetween-itemelaborative processes. ThestudyfocusesontwoERPcomponents:theP300andthefrontalslowwave.Previousstudies indicatethattheamplitudesofthesetwocomponentsareassociatedwithsubsequentrecallunder differentconditionstheP300whenitemsaredistinctiveandwhenparticipantsdonotuseelaborativememorizationstrategies;andthefrontalslowwavewhenelaborativestrategiesareusedand fornon-distinctiveitems.Thetheoreticalimplicationofthesepreviousstudiesisthatiftheoutcome ofamemoryupdatingprocesselicitedbysubjectivelydistinctiveitemsiseffectivelyutilizedforretrieval,theP300istocorrelatetosubsequentrecall.If,inturn,theoutcomesofelaborationbetween listitemsismosteffectivelyusedatretrieval,thefrontalslowwaveiscorrelatedwithrecall. Therefore,ifourresultsshowthatLFwordsisolatedinlistsofHFwordselicitaP300,andifits amplitudeiscorrelatedtosubsequentrecall,weconcludethattheeffectsofphysicaldistinctiveness anddistinctivenessduetoalowwordfrequencyonmemoryshareunderlyingcognitiveprocesses underthegivenconditions.Thatis,amemoryupdatingprocesselicitedbyphysicallydeviant 24

PAGE 32

wordsandbyunexpectedLFwordsiswhatcausestheenhancementinrecallrates.Thereversal ofwordfrequencyeffectsonrecall,inthiscase,couldinpartbeexplainedbyenhancedsubjective distinctivenessofLFwordswhenoccurringinlistsofHFwords. If,inturn,thefrontalslowwaveiscorrelatedtosubsequentrecallforisolatedLFwords,wewill concludethatoutcomesofenhancedelaborativeprocessesatstudyduetothepresenceofHFwords inthesamelistareusedasretrievalaidsatthetimeofthetest.Inthiscase,theconclusionwould bethatsubjectivedistinctivenessdoesnotplayamajorroleinthereversalofwordfrequencyeffect inmixedlists,butthatfacilitationofinter-itemelaborationduetothepresenceofHFwordsisthe cause. AnimportantcontrolconditioninthecurrentstudyiscomposedoftheHFwordsisolatedinlists ofLFwords.Asoutlinedabove,LFwordscanbeconsidereddistinctivewithrespecttoprevious experiences,whileHFwordsarenot.Ifourndingsareindeedduetothistypeofdistinctiveness denedbasedonpriorexperiencessecondarydistinctiveness,weshouldobtaineffectsforLF wordsisolatedinlistsofHFwords,butnotforHFwordsisolatedinlistsofLFwords.If,however, comparableeffectsareobtainedforisolatedLFwords and isolatedHFwords,thendingsmayin factbeanotherinstanceofprimarydistinctivenessthatisindependentofhowoftenawordhas beenexperiencedbeforebutonlydependentonwhetherthewordfrequency differs fromtheword frequenciesofallwordsinthesurroundingcontext. InadditiontotheP300andthefrontalslowwave,thisstudyalsoinvestigatestheN400.While previousresearchindicatedthatintheabsenceofaconstrainingsententialcontext,theN400is largerforLFthanforHFwords,toourknowledgetherearenostudiesthatresultedincorrelations betweenN400amplitudeandsubsequentrecall.Therefore,wedonotmakeanydirectedpredictions ofwhetherornotwewillobtainanN400foranyofthewordtypesandifwedo,whetherits amplitudewillbecorrelatedtorecall. Afurthergoalofthecurrentstudyistoreplicatepreviousndingswithadenseelectrodearray andadetailedinvestigationofthespatio-temporalcharacteristicsERPcomponentswithaspatiotemporalprincipalcomponentanalysisPCADien,Spencer,&Donchin,2003.Thus,oneshortcomingofmanyofthestudiesreviewedaboveisthelimitednumberofrecordingelectrodesused duetolimitedtechnologyavailableatthattime.OttenandDonchin'sstudyused10electrodes,whiletheevenearlierstudiesonlyusedonlyupto3electrodes.Theselownumbersof 25

PAGE 33

electrodesdonotallowforspecicconclusionsaboutspatialdistributionsofERPcomponents.In thecurrentstudywerecordERPswiththe128EGIelectrodenetnowavailableintheUSFPsychophysiologylab,whichincombinationwiththespatio-temporalPCAcanprovidemoredetailed informationofthespatio-temporalcharacteristicsofcomponentsofinterest. 26

PAGE 34

Chapter2 Methods 2.1Participants NineteenundergraduatestudentsattheUniversityofSouthFloridacompletedtwoexperimental sessionseach,foratotaldurationofapproximately5hours.DatafromveparticipantswereexcludedduetoexcessiveartifactsintheERPdata,anddatafromoneparticipantwereexcludeddue tosevereincompliancewiththeinstructions.Theremaining13participantswerebetween18and 45yearsoldmeanage:25.15,andallreportedEnglishtobetheirrstlanguage.Eightparticipants werefemaleandallbutoneparticipantwereright-handed.Participantsreceivedpartialcreditfor anintroductorypsychologycourseinreturnfortheirparticipation. 2.1.1IncentivesforGoodPerformance Toencouragegoodperformance,participantswereinformedbeforetherstsessionthattheir performanceinthememoryportionoftheexperimentwouldbescored.Theyweretoldthatthey wouldreceiveonepointforeachwordtheycorrectlyrecalled.Ifonagivenlisttheywrotedown morethatthreewordsthathadnotbeenstudiedonthepreviouslist,theywouldlosepoints.Each participantwasgiventheoptiontoreceiveanemailwiththeirownscoreaswellastheaveragescore ofallparticipantsoncetheirperformancewasevaluated. Furthermore,theparticipantwhoreachedthehighestscorewasdeterminedafterdatacollection wasnished,andthehighestscorewasrewardedwithabonusof$100.Itisworthnotingthatno participantwasinformedaboutnamesoranyotherpersonalinformationoftheotherparticipants whohadtakenpartinthestudy. 27

PAGE 35

2.2Materials 2.2.1Stimuli Thewordsinthememoryexperimentwereclassiedintothreegroups:high-,low-,ormedium frequencywords.Forthisgrouping,wecombinedthewordfrequencynormsbyFrancisandKucera Francis&Kucera,1982,withsubjectivefamiliaritynormsToglia&Battig,1987,becausepilot studiesindicatedthatbothnormedfrequencyandnormedfamiliarityaffectedtheresults.While mostauthorswhostudywordfrequencyeffectsuseonlyonenormintheiroperationalization,word frequencyandwordfamiliaritynormscanbethoughofasmeasuringthesameunderlyingphenomenonConnineetal.,1990,sothecombinationofthetwonormsseemedappropriate. Wecreatedthreegroupsofstimuli:Wordsofhighnormedfrequencymorethan50lemma occurrencesinamillionaswellashighnormedfamiliaritymeanratingofatleast6onascale of1to7;wordsoflowfrequencylessthan10occurrencespermillionandlowfamiliarity meanratingbetween3.5and5.5;andwordswithmiddlescoresofnormedfrequencybetween 10and50occurrencespermillionandnormedfamiliaritymeanratingbetween5.5and6.1.For thepurposeofsimplicity,therstwordtypewillinthefollowingbereferredtoashighfrequency HFwords,thesecondwordtypewillbereferredtoaslowfrequencyLFwords,andthethird wordtypeasmediumfrequencyMFwords.Weincludedonlyopen-classwordsnouns,verbsand adjectivesthatwerebetween3and8letterslong. NotethatthereisaslightoverlapinthenormedfamiliarityofMFwordsandHFwords,duetoa scarcityofstimulithatqualiedforeachgroup.Thecriticalwordfrequencymanipulation,however, isbetweenHFandLFwords,whiletheMFwordswereusedtocreateVonRestorfflistsinorderto replicatepreviousstudies.Therefore,wedidnotjudgethisslightoverlapinnormedfamiliarityas problematic. Allstimuliwerepresentedoneatatimeonacomputerscreeninawhitefontonablackbackgroundforadurationof250ms,withagapofof2000msbetweenstimuli,inwhichaxationcross wasshown.Wordsweredisplayedin16ptfontforregularsizedwordsand22ptfontforlarger wordssizeisolates,inArialUnicodestyle. 28

PAGE 36

2.2.2WordLists Weusedthreetypesofliststotestourexperimentalhypotheses,andinadditioncreatedasetof llerlists.Eachlist,regardlessoflisttype,consistedof15words. Thedesignoftherstlisttype,whichwewillcall VonRestorfflists, wasdesignedverysimilarly topreviousstudiesOtten&Donchin,2000.Ofthe15MFwordsusedineachVonRestorfflist, onewasdisplayedinalargerfontsizesizeisolatesthanalltheotherwordssizestandards. Thesizeisolateoccurredataserialpositionbetween6and10,determinedrandomlyforeachlist. Thesecondandthirdlisttype,respectively,consistedprimarilyofeitherHFwordsorLFwords HFstandardsandLFstandards,respectively.Onewordineachoftheselistsdifferedin frequencyfromtherestofthelist.Thatis,eachHFlistcontainedoneLFwordLFisolate, andeachLFlistcontainedoneHFwordHFisolate.Theisolateswererandomlyplacedata serialpositionbetween6and10.Eachparticipantencounteredatotalof25HFlistsand15LFlists throughouttheexperiment.Theunbalancednumberofwordlistsbetweenthelisttypesoccurred duetoascarcityofwordsfulllingtheLFwordcriteria. Thelastlisttypewasconstructedasllerliststhatwereintendedtokeepparticipantsfromdevelopingexpectationsforanisolateineachlist.Thus,thesetypesoflistsconsistedofamixtureof wordsfromdifferentfrequenciesandnormedfamiliaritiesanddidnotcontainanyisolates.There wereatotalof10suchllerliststhroughouttheexperiment. Foreachparticipant,thegroupingofwordsintolistsaswellastheorderofwordswithinthe listswasrandomized.Thus,eachwordwasdrawnrandomlyfromthepoolof300MFwords,235 LFwords,365HFwords,or150llerwords,respectively.Notethatthisrandomizationprocedure alsoinsuredthatawordthatoneparticipantviewedasanisolatewouldbeencounteredbyanother participantasanon-isolate.Eachparticipantstudiedeachwordonlyonceduringtheentirecourse oftheexperiment. Inadditiontothisrandomizationprocedureoftheindividualwords,thepresentationorderofthe listtypeswasdeterminedrandomlyaswell.Theonlyrestrictionwasthat14HFlists,8LFlists,12 VonRestorfflists,and6llerlistsatotalof40wordlistswerestudiedintherstsessionwhile 11HFlists,7LFlists,8VonRestorfflistsand4llerlistsatotalof30wordlistswerestudiedin thesecondsession. 29

PAGE 37

2.2.3RecallSheets Torecordtheparticipant'srecallmemory,onerecallsheetwasavailabletoeachparticipantfor eachstudylistonaclipboard.Eachrecallsheetincluded15lines,numberedfrom1to15,aswellas alittlebitofspaceforcommentsatthebottomofthepage.Thelistnumbereachsheetcorresponded towasprintedontopofthesheet.Apenholderwithseveralpensandpencilswaspositionedonthe tableintheexperimentroom,withwhichtheparticipantcouldlloutthesheetsduringeachrecall phase. 2.2.4IndividualWordFamiliarityQuestionnaire Awordfamiliarityquestionnairewasadministeredonapaper-and-pencilbase.Toinsurethat theratingsobtainedfromthisquestionnairewerecomparabletoTogliaandBattig'sratings,the instructionsandthedesignofthequestionnairewereconstructedtobeassimilaraspossibleto theirsToglia&Battig,1987.Becauseofthelargenumberofstimuliusedinthisexperimentand tocontrolforsequentialeffectsontheratings,weconstructedthreeversionsofthisquestionnaire thateachcontainedonlyasubsetsofthewords.EachofthethreeversionscontainedallLFwords sincethiswasthewordtypeforwhichsubjectivefamiliaritywasmostcritical,aboutonethirdof theHFwordsandaboutathirdoftheMFwords,resultingin300wordslistedinrandomorderfor eachversion.EachoftheHFwordsandMFwordswasrandomlyassignedtooccurinexactlyoneof thethreeversionsofthequestionnaire.Nexttoeachword,alikertscalefrom1veryunfamiliar to7veryfamiliarwasdisplayedinwhichtheparticipantcouldcircletheirfamiliarityratingfor therespectiveword.Eachparticipantcompletedonlyoneversionofthequestionnaire. 2.2.5StimuliandDesignoftheOddballTask InastandardoddballtaskparticipantsviewedarandomsequenceofanXandanOpresented onthescreen.TheprobabilityofanXoccurringonagiventrialwas0.8,andtheprobabilityof anOoccurringwas0.2.Duetotheseprobabilities,fortheXwewillinthefollowingusethe termstandardandfortheOwewillusethetermtarget.Overall,thesequencecontained200 stimuliabout40O'sand160X's.Theparticipant'staskwastorespondtoeachXbyaleft buttonpressandtotheObyarightbuttonpress.Thestimuliwerepresentedinblackfontona whitebackgroundin100ptCourierNewfont.Eachstimuluswasprecededbyaxationcrossfor 30

PAGE 38

1000msandremainedonthescreenuntiltheparticipanthadgivenaresponse,oratmostfor10 seconds. 2.3TaskandProcedure Foreachparticipant,eachofthetwoexperimentalsessionslastedatmost21/2hoursandbegan withtheusualpreparationsfortheEEGrecording.Then,participantswerecomfortablyseatedin theexperimentroom,atadistanceofabout3feetfromthecomputerscreen. Intherstsession,aftergivingwritteninformedconsentandreceivinginstructionsvisuallypresentedonthescreen,theparticipantcompletedtwopracticelists.Whenanyremainingquestions aboutthetaskandprocedurewereanswered,participantsstudied40wordlists,eachimmediately followedbyarecallphase.Thestartoftherecallphasewasindicatedbytheappearanceofalarge greytriangledisplayedonthescreen.Recallphaseslastedforatleast45seconds,buttherewasno uppertimelimitforcompletingthisphase.Theparticipantcouldnishtherecallphaseandstart thenextlistwithabuttonpresswhenevertheywereready.Theparticipant'staskwastostudyeach wordusingrotememorizationstrategiesthatis,repeatingeachwordsilentlyafteritspresentation, andintherecallphasetowritedowneverywordtheyrememberfromthepreviouslistinanyorder. Abreakwasallowedaftereachsetof5lists,inwhichtheparticipantcouldrelaxforaslongas theywished.Duringthebreaks,theparticipantwasremindedtouserotememorizationstrategies throughouttheexperiment. Thesecondsessiontookplacebetweenthreeandsevendaysaftertherstsession.First,participantsreceivedashortenedversionoftheinstructionsasareminder.Then,theystudiedandrecalled anothersetof30wordlistsunderthesameinstructionsasintherstsession.Afterthelaststudyrecallcyclewascompleted,theycompletedtheoddballtask.Inthistask,participantsindicatedthe detectionofanXbyaleftbuttonpressandthedetectionofaObyarightbuttonpressineach trial.After4practicetrials,participantscompleted200experimentaltrials,whichlastedupto5 minutes.Attheendofthesecondsession,aftertheelectrodenetwasremoved,eachparticipantwas givenunlimitedtimetolloutthesubjectivefamiliarityquestionnaire.Mostparticipantsneeded lessthan10minutestocompletethequestionnaire.Finally,subjectsweredebriefedandquestioned aboutthememorizationstrategiestheyhadused,aboutwhethertheyhadanydifcultiesseeingthe wordsonthescreen,andwhethertheynoticedthatsomewordsstoodoutfromthelists. 31

PAGE 39

2.4EEGRecordingandAnalysis TheEEGwasrecordedwitha128ElectricalGeodesicssystemEGI,Eugene,ORintheCognitivePsychophysiologylabattheUniversityofSouthFlorida.TheEEGwasdigitizedatasampling rateof250Hz,andreferencedtothecentralelectrodeCz.Inanofine-analysisusingNetstation software,theEEGwasdigitallylow-passlteredatacutofffrequencyof20Hzandbadchannels werereplacedbyamathematicalproceduretakingintoaccountthesignalsofthechannelsadjacent tothebadelectrode.Then,thecontinuousdataweresegmentedfrom400msbeforeto2000ms aftereachstimulus.Tocorrectforeyemovementartifactsandeyeblinks,weusedanindependent componentanalysisapproach,asprovidedinDien'sERPtoolkit,version1.3,basedonMATLAB Dien,2010.Notethatthiseyemovementcorrectionprocedureincludedabaselinecorrectionusingtheaverageamplitudeof400msprecedingthestimulus.Subsequently,eachtrialwasvisually inspectedforremainingartifactsandthecorrespondingtrialswereexcludedfromanyfurtheranalysis.Inanalpre-processingstepusingNetstationsoftware,theEEGwasre-referencedtoalinked mastoidreference. AllsubsequentanalysiswasconductedusingscriptsandfunctionswritteninMATLAB,while utilizingseveralfunctionsoftheEEGLABtoolboxDelorme&Makeig,2004.Toobtainthe subjectERPs,weaveragedoverallartifact-freetrialsseparatelyforeachwordtypeHF,LFand MF,isolationtypeisolateandnon-isolateandrecallsuccess.Weonlyincludedtrialsfromserial positions6to10topreventconfoundswithprimacyandrecencyeffects.Forillustrationpurposes, grandaverageERPswerecomputedbycombiningallsubjectaverages. Weconductedaspatio-temporalprincipalcomponentanalysisPCADienetal.,2003toidentifyandquantifyERPcomponents,usingversion1.23ofDien'sERPtoolkitDien,2010.Promax rotationswithouttheKaisernormalizationoptionwereappliedforboththespatialandthetemporal analysisstep.Inthespatialstep,ascreetestindicatedthat25factorsshouldberetainedandrotated. Foreachspatialfactor,thevirtualERPswerecomputedbyaveragingoverthefactorscoresof allparticipants,separatelyforwordtypesandsubsequentrecallsuccess.InthetemporalPCAstep, whichwasconductedseparatelyonthefactorscoresfromeachspatialfactor,8factorswereretained.Thenumberoftemporalfactorstoretainwasagaindeterminedbyascreetestandwasheld constantacrossspatialfactorsforthesakeofsimplicity.Thus,theinputtothespatio-temporalPCA consistedoftheERPsfrom129electrodesand600timepointsms,andfrom13participants 32

PAGE 40

and14conditionssizestandard,sizeisolate,HFstandard,LFisolate,LFstandard,HFisolate,and ller;eachwordtypeseparatelyforsubsequentlyrecalledandsubsequentlynotrecalleditems. NotethatweincludedthellerwordsintothePCAinordertoincreasethenumberofobservations. However,wedidnotanalyzetrialsfromthislisttypefurthersincethellerlistsdidnotcontrolfor normedwordfrequencyorfamiliarity. Therewereonly15HFisolatetrialsforeachparticipant,whichincludedbothrecalledandnotrecalleditems.Duetothislowtrialcount,plotsoftheHFisolateERPsmayappearrathernoisy. Moreimportantly,3participantshadnoartifactfreetrialsfortheHFisolate-recalledcondition, resultinginthreemissingvaluesforthiscondition.Asaconsequence,thegrandaverageERPs containdatafromonly10participantsforrecalledHFisolates,butdatafrom13participantsforthe notrecalledHFisolates.Furthermore,inordertobeabletoanalyzeasmuchoftheinformation availableforHFisolates,theywereincludedinthePCA,andamatrixcontainingonlyzeroswas submittedinplaceofanythemissingERPs.ThisproceduredoesnotaffecttheresultsofthePCA, sinceamatrixcontainingonlyzerosdoesnotinduceanyvarianceintothedata. 2.5StatisticalAnalysis Recallratestheproportionofwordscorrectlyrecalledintheimmediatelyfollowingrecallphase werecomputedforeachparticipant,separatelyforeachwordtypeHF,LFandMFwords,foreach isolationtypenon-isolateditemsandisolates,aswellasforeachserialpositiontoconstructserial positioncurves.VonRestorffeffectswereanalyzedbycomparingtherecallratesforsizeisolates totherecallratesofnon-isolateswithinthesamelistsatserialpositions6to10,usingtwo-sided pairedsamplest-tests.Wordfrequencyeffectsatserialpositions6to10wereinvestigatedwitha2 wordfrequencyx2isolatevs.standardrepeatedmeasuresANOVA.Forthebehavioralanalysis, allsignicancelevelswerechosenat0.05. AllstatisticalanalysisofERPcomponentswasconductedonthefactorscoresobtainedthrough thespatio-temporalPCA.Inarstanalysisstep,HFisolateswereexcludedfromthestatisticsdue tothethreemissingvaluesfortheHFisolaterecalledcondition.RepeatedmeasuresANOVAswith thewithinsubjectfactorswordtypelevels:sizeisolate,sizestandard,HFstandard,LFisolate, LFstandardandrecallsuccessrecalled,notrecalledwerethenconductedtotestformaineffects andinteractions.Signicantinteractionswereinvestigatedbypost-hocplannedcomparisons.For 33

PAGE 41

theoverallANOVA,weselectedandfurtheranalyzedmaineffectsandinteractionswithp-values smallerthan0.1,whileinpost-hocteststhechosensignicancelevelwas0.05.Furthermore,for afewspecicfactorsandconditionsofinterest,weconductedplannedcomparisonsforonegiven wordtype,comparingsubsequentlyrecalledtosubsequentlynotrecalledtrialsinatwo-sidedpaired samplest-test. SubsequentmemoryeffectsforHFisolateswereanalyzedseparatelyfromallotherwordtypes. Duetothemissingvaluesweconductedtwo-sidedindependentsamplest-testsforunequalsample sizes,including10observationsforHFisolatesthatwerelaterrecalledand13observationsforHF isolatesthatwerelaternotrecalled. 34

PAGE 42

Chapter3 Results 3.1Debrieng Whenaskedwhethertheyrealizedthatsomewordsstoodoutfromthelists,everyparticipant reportedtohavenoticedthatsomewordswereprintedinalargerfontthantheothers,indicating thatthemanipulationofphysicalisolationduetofontsizewassuccessful.Furthermore,mostparticipantsreportedthattheyhadtriedtouserotememorizationstrategies.Twoparticipantsadmitted tousingstrategiesotherthanrotememorizationoccasionallyduringtheexperiment,andanother twoparticipantsreportedthattheyhadusedelaborativestrategiesmostofthetimesuchasbuilding sentenceswiththewordsorrelatingthewordstoeachotherinsomeotherway. 3.2BehavioralData 3.2.1RecallRates TheserialpositioncurvefortheVonRestorfflistsispresentedingure1B.Thesizeisolates weresignicantlybetterrecalledthanthenon-isolatesatcomparableserialpositionspositions610, t =4.21, p <.01.Allbuttwoparticipantsshowedincreasedrecallratesforthesizeisolates thanforthestandards.Furthermore,byvisualinspectionoftheserialpositioncurvesoftheVon Restorfflists,itappearsthatforthestandards,clearprimacyandrecencyeffectsarepresent. Figure1AshowstheserialpositioncurveforHFandLFlists.Intheprimacyserialpositions serialpositions1-5,HFstandardswerebetterrecalledthanLFstandards, t =3.52, p <.01,which representsthetraditionalwordfrequencyeffectonrecallwhenpurelistsarestudied.Forthemiddle serialpositions-10,the2wordfrequencyby2isolatevs.non-isolateANOVArevealed asignicantinteraction, F ,12=10.93, p <.01.Post-hocplannedcomparisonsindicatedthatLF wordswererecalledbetterwhentheyoccurredisolatedinlistsofHFwordscomparedtoLFwords 35

PAGE 43

A.HFandLFlistsB.VonRestorffLists Figure1. :Serialpositioncurves.Probabilityofrecallasafunctionofpositioninthelist. inLFlists.ForHFwordsthereversewastruesuchthattheywererecalled worse whenoccurring inisolationinlistsofLFwords.Notealsothatwithineachlisttypetheisolateswererecalledat aboutthesamelevelasthenon-isolates.Thus,LFisolateswererecalledaboutequallywellas HFstandardsandHFisolateswererecalledaboutequallywellasLFstandards.Technically,this ndingdiffersfromtheVonRestorffeffect,inwhichtheisolatesarerecalledataclearly higher levelcomparedtothenon-isolateswithinthesamelist. Inanadditionalanalysis,wecomparedtherecallratesforthenon-isolateditemsintheVon RestorfflistscomposedofMFwords,LFlistsandHFlists.Aone-wayANOVAontherecall ratesatserialpositions6-10resultedinasignicantmaineffect, F ,24=10.01, p <.01.Posthoc testsindicatedthatnon-isolatedHFwordsweremorelikelytoberecalledthanbothnon-isolated MFwordsandnon-isolatedLFstandards,whileMFandLFstandardsdidnotdifferfromeachother. 3.2.2AnalysisoftheSubjectiveFamiliarityQuestionnaire Tofurtherinvestigatewhetherourmanipulationofpriorexposureusingwordfrequencyand familiaritynormswassuccessful,wecomputedtheaverageratingforeachparticipantforwordsin theLFgroup,wordsintheMFgroupandwordsintheHFgroup.Notethateachparticipantonly ratedasubsetoftheHFandMFwords,butweassumeherethatthoseratingsarerepresentative oftheentirewordtype.TheonewayANOVAonthesubjectivefamiliarityratingsrevealeda signicantmaineffectforwordtype, F ,24=21.79, p <.01.Ascanbeseeningure2A,HF 36

PAGE 44

A.RatingsforallWordTypesB.RatingsforLFIsolates Figure2. :Subjectivefamiliarityratings.A.Ratingsforallwordtypes,collapsedoverisolates andnon-isolates.B.RatingsonlyforLFisolatesasafunctionofsubsequentrecall.Notethatthe actualwordsincludedinthesecategoriesaredifferentforeachparticipant.Errorbarsrepresentthe standarderrorofthemean. wordswereratedasmorefamiliarthanMFwords,whichinturnwereratedasmorefamiliarthan LFwords.Thisndingindicatesthatourmanipulationofwordtypewassuccessfulsuchthatour manipulationofpriorexposureforourwordtypes,asagroup,matchedthesubjectiveexperience oftheparticipants.Itshouldbenoted,however,thatseveralparticipantsratedsomeLFwordswith a,indicatingthatthesewordswereveryunfamiliarorunknowntothem. Inaddition,weanalyzedthedifferenceinfamiliarityratingsbetweenLFwordsthatwerelater recalledandLFwordsthatwerelaternotrecalled.Ascanbeseeningure1B,therewereno signicantdifferenceswithrespecttosubsequentrecall, t =0.34, ns 3.2.3AnalysisofOutputOrder Thegraphingure3showstheprobabilitythataLFisolateorsizeisolatewaswrittendownat thebeginningoutputpositions1or2,themiddle,ortheendnexttolastandlastoutputpositions oftherecallsheet,giventhatthisisolatewassuccessfullyrecalled.Theprobabilitiesdonotaddup to1,sinceeachprobabilityiscorrectedforthenumberofwordswrittendownatthecorresponding positions.NotethatwedonotreportthecorrespondinganalysisforHFisolates,becausethenumber ofrecalledHFisolateswasgenerallyverylow,causinghighvariabilityintheaverageprobabilities foreachoutputposition. 37

PAGE 45

Figure3. :Outputorderanalysis.Probabilityofrecallatspecicoutputpositionsforsizeisolates andLFisolates,giventhattheseweresuccessfullyrecalled. Aone-wayANOVAontheoutputprobabilitiesforLFisolatesonserialpositions1and2,the middleoutputpositions,andnexttolastorlastoutputpositionsrevealedasignicantmaineffect, F ,24=3.52, p =.05,indicatingthatLFisolatesweremorelikelytobewrittendownononeofthe lasttwooutputpositions,comparedtomiddleorbeginningserialpositions.Asimilaranalysison thesizeisolatesdidnotrevealasignicantmaineffectforoutputposition, F ,24=.55, ns 3.3Event-RelatedPotentials GrandaverageERPsareshowningure4fortheVonRestorfflists,ingure5forHFlists withLFisolates,andingure6forLFlistswithHFisolates.VisualinspectionoftheERPsfor thesizeisolatesandtheLFisolatesindicatesthataparietallydistributedpositivity,peakingaround 700ms,islargerforsubsequentlyrecalledcomparedtonotrecalledwordsgures4and5,right panels.ThesameisnotapparentintheERPsforHFisolatesgure6,rightpaneloranytypeof non-isolatedwordgures4,5and6,leftpanels.Themorphologyandspatialdistributionofthis parietalpositivityindicatesthatitrepresentsaninstanceoftheP300.Inaddition,amorefrontally distributedsustainedpositivity,possiblyaninstanceofthefrontalslowwave,seemstodistinguish betweensubsequentlyrecalledandnotrecalledsizeisolates,LFisolates,andtoalesserdegree, sizestandardsgures4and5.Furthermore,fortheHFisolates,anegativecomponentmaximal at400msafterthestimulusseemstodistinguishbetweenrecalledandnotrecalledwordsgure6, 38

PAGE 46

FrontalElectrodeFz CentralElectrodeCz m ParietalElectrodePz Figure4. :GrandaveragesforVonRestorfflists.GrandaverageERPsforthreemidlineelectrodes asafunctionofsubsequentrecall. 39

PAGE 47

FrontalElectrodeFz CentralElectrodeCz ParietalElectrodePz Figure5. :GrandaveragesforHFlistsincludingLFisolates.GrandaverageERPsforthreemidline electrodesasafunctionofsubsequentrecall. 40

PAGE 48

FrontalElectrodeFz CentralElectrodeCz ParietalElectrodePz Figure6. :GrandaveragesforLFlistsincludingHFisolates.GrandaverageERPsforthreemidline electrodesasafunctionofsubsequentrecallforLFlistsincludingHFisolates. 41

PAGE 49

SF1SF2SF3SF4SF5 Figure7. :SpatialFactorLoadings.Theloadingsoftherstvespatialfactorsobtainedinthe spatialPCAstep. rightpanel.ItisconceivablethatthiscomponentisaninstanceoftheN400.Tomoreaccurately characterizeandquantifytheERPcomponents,thedataweresubmittedtoaspatio-temporalPCA. 3.3.1PCAResults AscreetestonthePCAresultsindicatedthat25spatialfactorsand8temporalfactorsforeach spatialfactorshouldberetainedandsubmittedtoaPromaxrotation.Notethatasimilaranalysis usingaVarimaxrotationinsteadofPromaxrevealedspatialandtemporalfactorsthatwereinclose correspondencetotheonesreportedhere.ThereasonthatPromaxwaschosenasthepreferred rotationforthisstudywasthat,possiblyduetothelargenumberofwordtypesandconditionsin thisstudywordtypesx2recallsuccesscategories=14conditions,theresultsofPROMAX appearedtobemoreclear-cutandeasiertointerpret.TheresultsoftheVarimaxrotationwere similar,butstatisticalcomparisonswereinmanycasesnon-signicant. Spatialfactorloadingsoftherstvespatialfactorsareshowningure7.Onlyspatialfactors one,two,threeandvewerefurtheranalyzed,sincethespatialdistributionsoffactorfourandall spatialfactorsbeyondfactor5didnotappeartocorrespondmeaningfullytoERPcomponents.The totalvarianceaccountedforinthespatialstepwas90.61%,andspatialfactorsone,two,threeand veexplained21.24%,8.99%,8.97%,and6.31%oftheoriginalvariance,respectively. Therstthreetemporalfactorsofspatialfactors1,2,3and5wereanalyzedina5wordtype: HFstandard,LFstandard,LFisolate,SizestandardandSizeisolateby2recallsuccessrepeated measuresANOVAandsignicantinteractionswerefollowedupbysubsequentplannedcomparisons.Furthermore,someadditionalplannedcomparisonswereconductedwhentherewerea-priori 42

PAGE 50

reasonstobelievethattheERPsmightdifferaccordingtosubsequentmemoryforaspecicword type. TheHFisolateswereanalyzedinindependentsamplest-testswithunevensamplesizes,using thefactorscoresfrom10participantsforthesubsequentlyrecalled,andfrom13participantsfor thenotrecalledHFisolates.Onlysignicantresults,resultsapproachingsignicance,orresultsof specicinterestwithrespecttoourapriorihypothesesarereportedhere. 3.3.1.1FrontalFactor Spatialfactoroneexhibitedaprefrontaldistribution,ascanbeseeningure8alongwithvirtual ERPsandtemporalfactors.Duetothefrontaldistribution,atemporalfactormostlikelycorrespondingtotheslowwavewasofspecialinterest,whichappearstobecapturedintemporalfactor 1gure8C.However,therewerenoindicationsforamaineffectofwordtype,recallsuccess,or aninteractionforthePCAscoresofthisfactorallp-valuesbiggerthan.47. BothinspectionoftheERPsoffrontalelectrodese.g.Fzingures4,5and6andthevirtual ERPsofthefrontalfactorgure8Bshowastrongpositive-goingpotentialpeakingaround700 msforallwordtypesandbothrecalledandnon-recalledwords.Formostwordtypes,thispositivity, whichiscapturedintemporalfactor2seegure8C,didnotappeartodifferbetweensubsequently recalledandsubsequentlynotrecalledwords.However,theoverallwordtypelevels:HFstandard,LFstandard,LFisolate,sizestandard,sizeisolatebyrecallsuccessinteractionapproached signicance, F ,48=2.3, p =.07.Asubsequentplannedcomparisonindicatedthatforsizeisolates, thepositivitywaslargerforsubsequentlyrecalledthannotrecalledwords, t =3.07, p <.01,while forLFstandardsthereversewastrue,suchthattheamplitudewassmallerforsubsequentlyrecalled comparedtonotrecalledwords, t =2.4, p =.03.Fortheremainingthreewordtypes,therewasno differencebetweenrecalledandnotrecalledwordsallp-valuesbiggerthan.23. Forthethirdtemporalfactor,peakingatabout1400ms,theoverallANOVAresultedinasignicantmaineffectforrecallsuccess, F ,12=6.67, p =.02.Thus,thefactorscoresforthisfactorwere morepositiveforsubsequentlyrecalledcomparedtonotrecalledwordsacrosswordtypes. NotethatnoneofthedifferencesduetorecallsuccessfortheHFisolatesapproachedsignicance, indicatingthatanyappearanceofsuchdifferencesinthevirtualERPsislikelyduetothelownumber oftrialsandhencethehighlevelofnoiseforthiswordtype. 43

PAGE 51

A.SpatialFactorLoadingsC.TemporalFactorLoadings B.VirtualERPs Figure8. :Spatialfactor1.A.Spatialfactorloadings.B.VirtualERPs:Averagedspatialfactor scores,plottedovertime.C.Temporalfactorsloadings. 44

PAGE 52

3.3.1.2RightLateralizedSpatialFactor Spatialfactortwowasrightlaterallydistributedgure9.Thenegativepeakaround400ms inthevirtualERPs,possiblyaninstanceoftheN400,seemedtocorrelatewithsubsequentrecall successforHFisolatesgure9B.However,neithervisualinspectionofgrandaveragewaveforms onrightelectrodes,norinspectionofthevirtualERPsforthisspatialfactorindicateanystrong differencesforsubsequentrecallsuccessforanyotherwordtype. Statisticalanalysisoftherstthreetemporalfactorsdidnotrevealanysignicantmaineffects orinteractionsinvolvingtherecallsuccesscomparison.However,theoverall5wordtypeby2 recallsuccessANOVAonthefactorscoresfortemporalfactor2didresultinasignicantmain effectforwordtype, F ,48=2.73, p =.04.Inasubsequentanalysiscollapsingacrossrecallsuccess, wefoundthattheamplitudeforthiscomponentwassignicantlymorepositiveforsizeisolatesthan forallotherwordtypesnote,however,thattheHFisolateswerenotincludedinthisanalysis.This temporalfactorpeakedaround600msgure9Canditisnotobviouswhichpreviouslyreported ERPcomponentitmightcorrespondto. Temporalfactors2and3overlappedthetimeperiodinwhichERPsappearedtodifferbetween subsequentlyrecalledandnotrecalledHFisolates.Indeed,temporalfactor2,peakingat600ms, wassignicantlymorepositiveforrecalledthannotrecalledwords, t =2.44, p =.02.Thedifferencefortemporalfactor3,peakingat400msandlikelyrepresentinganinstanceoftheN400, approachedsignicance, t =1.99, p =.06.ThisindicatesthattheN400factortendedtobemore negative-goingfornotrecalledthansuccessfullyrecalledwords.NotethatalthoughthisN400amplitudedifferencebetweenHFisolatesthatwereandthatwerenotsubsequentlyrecalledapproached signicance,furtherplannedcomparisonsrevealednooveralldifferencesinN400amplitudebetweenHFisolatesandanyotherwordtype. 3.3.1.3LeftLateralizedSpatialFactor ThethirdspatialfactorwasleftlateralizedanditsvirtualERPsappearedtomorphologicallyresembleslowwaveeffectsseegure10.Thus,thevirtualERPsexhibitedaslowpositivewavethat appearedtodifferentiatebetweenrecalledandnotrecalledwordsforsizeisolates,LFisolates,and, inthereversedirection,LFstandardsgure10B.Therstandsecondtemporalfactorspanned thetimeperiodcorrespondingtothispositive-goingslowwavegure10C.Fortemporalfactor 45

PAGE 53

A.SpatialFactorLoadingsC.TemporalFactorLoadings B.VirtualERPs Figure9. :Spatialfactor2.A.Spatialfactorloadings.B.VirtualERPs:Averagedspatialfactor scores,plottedovertime.C.Temporalfactorsloadings. 46

PAGE 54

A.SpatialFactorLoadingsC.TemporalFactorLoadings B.VirtualERPs Figure10. :Spatialfactor3.A.Spatialfactorloadings.B.VirtualERPs:Averagedspatialfactor scores,plottedovertime.C.Temporalfactorsloadings. 47

PAGE 55

1exhibitingaslowlyincreasingtemporaldistributionacrosstheepoch,theoverall5by2ANOVA didnotrevealanysignicanteffectsorinteractions.However,sincethevirtualERPsstronglysuggesteddifferencesduetorecallforLFandsizeisolates,weperformedfurtherplannedcomparisons foronlythesetwowordtypes.Forthesizeisolates,thedifferencebetweensubsequentlyrecalled andnotrecalledwordswassignicant, t =3.38, p <.01,whilefortheLFisolates,thedifference wasnon-signicant, t =1.5, p =.16. Thesecondtemporalfactoroverlappingthetimeperiodoftheslowwaveasvisibleinthevirtual ERPspeakedat650msgure10BandC.TheoverallANOVAonthescoresofthisfactor revealedasignicantwordtypebyrecallsuccessinteraction, F ,48=3.93, p <.01.ForLFisolates andsizeisolates,thisfactorwassignicantlymorepositiveforrecalledcomparedtonotrecalled words t =2.28, p =.04and t =3.14, p <.01,respectively.Unexpectedly,forLFstandardsthe relationshiptendedtobetheoppositeway.Thus,LFstandardsthatwerelaternotrecalledtended toelicitalargerpositivitythantheonesthatwerelaterrecalled, t =1.88, p =.08.Subsequent memoryeffectsfortheotherwordtypesdidnotapproachsignicance. NotethatalthoughthevirtualERPsforHFisolatesforspatialfactor3appeartobedifferent betweensubsequentlyrecalledandsubsequentlynotrecalledwords,noneofthetemporalfactor scoresobtainedinthePCAdifferedsignicantlybyrecallsuccess.Therefore,thevisualimpression ofanydifferenceduetorecallsuccessislikelyduetoahighlevelofnoisecausedbylowtrial numbers. 3.3.1.4ParietalSpatialFactor Thespatialdistributionofthefthspatialfactorisshowningure11,alongwithitsvirtual ERPsandtemporalfactorloadings.Spatialfactor5showedaparietaldistributioncharacteristic oftheP300gure11A.Furthermore,apositivepeakispresentinthevirtualERPsforthesize isolatesbutnotsizestandardsgure11B,supportingtheideathatthisfactorcorrespondsto theP300.Temporalfactors1and2overlapwiththetimeperiodoftheP300gure11Band C.Nosignicanteffectsorinteractionswereobtainedfortemporalfactor1,whichapparently correspondstoaposteriorslowwave.Temporalfactor2seemstomorecloselyresembleP300 morphology,althoughitpeakssomewhatlate,around750ms.Forthistemporalfactor,theoverall wordtypebyrecallsuccessinteractionapproachedsignicance, F ,48=2.39, p =.06.Subsequent 48

PAGE 56

A.SpatialFactorLoadingsC.TemporalFactorLoadings B.VirtualERPs Figure11. :Spatialfactor5.A.Spatialfactorloadings.B.VirtualERPs:Averagedspatialfactor scores,plottedovertime.C.Temporalfactorsloadings. 49

PAGE 57

plannedcomparisonsindicatedthatforLFisolates,factorscoresweresignicantlymorepositivefor subsequentlyrecalledcomparedtonotrecalledwords, t =2.56, p =.03.Whilethedifferencewas inthesamedirectionandappearedtobeofaboutequalmagnitudeforsizeisolates,thiscomparison didnotreachsignicance, t =1.72, p =0.11.Forallotherwordtypes,therewerenosignicant subsequentmemoryeffectsfortheP300factor. Fortemporalfactor3,showinganegativepeakaround400msafterthestimulus,themaineffect forsubsequentrecallsuccessapproachedsignicance, F ,12=3.23, p =.1.Thisfactortendedto bemorenegativeforsubsequentlynotrecalledcomparedtorecalledwords,whichisinlinewith theimpressiononemightgetfromvisuallyinspectingthevirtualERPsforsizeisolatesandLF isolates.Notethatalthoughasimilareffectseemstobepresentforthesamefactorelicitedbythe HFisolates,thisdifferencewasnotsignicant.Furthermore,noothercomparisonsforsubsequent recallforspatialfactor5forHFisolatesapproachedsignicance. 3.4OddballTask Intheoddballtask,allparticipantsreachedaresponseaccuracyofatleast96.8%mean:99.5% inrespondingtostandardsandatleast75%mean:90.7%whenrespondingtotargets,indicating thateveryparticipantperformedwellabovechance.Thedifferenceinresponseaccuracybetween standardsandtargetswassignicant, t =4.23, p <.01.Furthermore,reactiontimesweresignificantlylongertotargetsmean:451.6mscomparedtostandardsmean:399.76ms, t =3.15, p <.01.Figure12showsthegrandaverageERPsfromthreemidlineelectrodes.AP300isvisiblein theparietalelectrode,althoughitseemsasthoughthestandardselicitalargepositivityaswell. InthePCA,8spatialfactorsand5temporalfactorswereretainedandsubmittedtoaPromax rotation.Byvisualinspection,wedeterminedthatonlyspatialfactors1,2,and3corresponded meaningfullytodistributionsofERPcomponents,soallfurtheranalysisfocusedonthesethree spatialfactors.Spatialfactors1,2and3accountedfor38.65%,25.55%and9.5%oftheoriginal variance,respectively.Factorloadings,virtualERPsandtemporalfactorsfromspatialfactors1-3 areshowningure13.Pairedsamplest-testswereconductedonthescoresofthetemporalfactors fortherstthreespatialfactors.Onlystatisticallysignicantanalysesatalevelof.05arereported here. 50

PAGE 58

FrontalElectrodeFz CentralElectrodeCz ParietalElectrodePz Figure12. :GrandaverageERPsfromtheoddballtask. 51

PAGE 59

Ascanbeseeningure13AandB,spatialfactor1exhibitedafrontaldistributionanditsvirtual ERPswerecharacterizedbyaslownegativewaveprominentforthetargetERPs.Thersttemporal factorgure13C,whichappearedtocapturethisnegativewave,wassignicantlylargerfor targetsthanstandards, t =2.61, p =.02.Thesecondtemporalfactorexhibitedanuninterpretable distribution,butdifferedbetweentargetsandstandards, t =2.33, p =.04.Furthermore,thefourth temporalfactornotshowningure13,capturingthelargepositivitypeakingabout200msafter thestimulus,wassignicantlymorepositivefortargetsthanforstandards, t =4.17, p <.01. Spatialfactor2wascentrallydistributedandshowedapositivepeakaround200ms,followedby anegativepeakaround300ms,withbothpeaksapparentlymorepronouncedfortargets.Statistical analysisofthetemporalfactorcorrespondingtothepositivepeaktemporalfactor5,notshown ingure13, t =2.37, p =.04,aswellasthetemporalfactorcorrespondingtothenegativepeak temporalfactor3,seegure13C, t =2.42, p =.03conrmedthisimpression. Thethirdspatialfactorwasparietallydistributed,adistributionthatischaracteristicoftheP300. Notethatthespatialdistributionofthisfactorwasremarkablysimilartothedistributionofthe P300factorobtainedfromtheERPsinthememorytask.Onedifcultythatweencounteredin thestatisticalanalysisofthefactorscoresoftheP300componentintheoddballtaskwasthatthe standardsseemedtoelicitalargeP300aswell.However,thispositivityelicitedbythestandards peakedatashorterlatency,asvisibleinthevirtualERPsaswellastherawaveragedERPssee gure12,bottompanel,andgure13B,rightpanel.Thus,thetemporalfactorcorresponding totheP300elicitedbythetargetstemporalfactor1,gure13Cdidnotseemtobethesame temporalfactorthatcorrespondstotheparietalpositivityelicitedbythestandardstemporalfactor 2.Therefore,intherststepwecomparedthefactorscoresfortemporalfactors1and2between thetargetsandstandards,andinthesecondstepwecomparedthefactorscoresfortemporalfactor 1forthetargetswiththefactorscoresfortemporalfactor2forstandards,acomparisonthatseems tobemorevalidinreectingdifferencesinP300amplitude.Factorscoresoftemporalfactor1 weresignicantlylargerfortargetERPscomparedtostandardERPs, t =2.99, p =.01,while factorscoresfortemporalfactor2werelargerforthestandards, t =4.13, p <.01.However,factor scoresfortemporalfactor1forthetargetsdidnotsignicantlydifferfromfactorscoresoftemporal factor2forthestandards, t =0.01, p =.99.Noneoftheremainingtemporalfactorsforspatial factor5differedbetweentargetsandstandards. 52

PAGE 60

SpatialFactor1SpatialFactor2SpatialFactor3 A. B. C. msmsms Figure13. :Therstthreespatialfactorsfortheoddballtask.A.Spatialfactorloadings;B.Virtual ERPs;C.Temporalfactorloadingsfortherstthreetemporalfactors. 53

PAGE 61

Chapter4 Discussion 4.1SummaryofResults Behaviorally,weobtainedaclassicalVonRestorffeffectforphysicallyisolatedwords,atypical wordfrequencyeffectfornon-isolatedwords,aswellasareversalofthewordfrequencyeffect whenHFandLFwordsoccurredisolatedinlistsofopposite-frequencywords.ThePCAonthe subjectERPsrevealedafrontalfactor,arightlateralizedfactorpossiblyinpartcorrespondingto theN400,aleftlateralizedfactorthatmorphologicallyresembledthefrontalslowwave,aswellas aparietallydistributedfactorrepresentingtheP300. Inlinewithourexpectations,thesizeisolateselicitedaP300componentpeakingaround750ms afterthestimulus.ThedifferenceinP300amplitudebetweensubsequentlyrecalledandnotrecalled sizeisolates,asmeasuredbythecorrespondingPCAfactorscores,approachedsignicance.Size isolatesalsoelicitedaleftlateralizedslowwavewhoseamplitudeshowedaclearandsignicant subsequentmemoryeffect.Anotherpositivecomponentwaspresentinthefrontalspatialfactor, peakedaround700msafterthestimulusandwaslargerforsubsequentlyrecalledcomparedtonot recalledsizeisolates.Lastly,apositivityintherightlateralizedspatialfactor,alsopeaking700ms afterthestimulus,wassignicantlylargerforsizeisolatesthanforanyotherwordtype,butdidnot differwithrespecttosubsequentrecall. Similartothesizeisolates,LFisolatesalsoelicitedaP300thatwassignicantlymorepositivefor recalledcomparedtonotrecalledwords.Furthermore,theleftlateralizedslowwavealsoshoweda subsequentmemoryeffectsuchthatlaterrecalledLFisolateselicitedmorepositiveamplitudesthan notrecalledLFisolates.Thisdifference,however,wasonlystatisticallysignicantfortheearlier temporalfactoroverlappingtheslowwave.Thefrontalandrightlateralizedspatialfactorsdidnot containanytemporalfactorswithsignicantsubsequentmemoryeffectsforLFisolates. 54

PAGE 62

TheERPcomponentsandsubsequentmemoryeffectsweobtainedforHFisolateswereremarkablydifferentfromthoseforLFisolates.Thus,HFisolatesdidnotappeartoelicitastrongP300 oraleftlateralizedslowwavecomponent,norwereanysubsequentmemoryeffectssignicantfor thesefactors.Instead,therightlateralizednegativitypeakingat400msafterthestimulus,which ismostlikelyaninstanceoftheN400,wasnegativelycorrelatedwithsubsequentrecall.Thatis, HFisolatesthatwerenotrecalledelicitedalargerN400thanlaterrecalledwords.Furthermore, thepositivityfollowingtheN400waslargerforsubsequentlyrecalledcomparedtonotrecalledHF isolates. Fornon-isolatedLFwordsLFstandards,wealsoobtainedacoupleofinterestingndings. First,therewasasignicantsubsequentmemoryeffectforthefrontalpositivitypeakingat700ms, suchthatsubsequentlyrecalledLFstandardselicitedsmallerpositiveamplitudesthanLFstandards thatwerelaternotrecalled.Notethatthissubsequentmemoryeffectisintheoppositedirectionas theoneobtainedforsizeisolates.Secondly,theleftlateralizedslowwavecomponenttendedtobe morepositiveforlaterrecalledcomparedtolaternotrecalledLFstandards.Again,thissubsequent memoryeffectisinthereversedirectionthanforsizeisolatesandLFisolates. TheHFstandardsandsizestandardsdidnotelicitanysubsequentmemoryeffectsuniquetotheir wordtype.However,thereweretwomaineffectsforsubsequentrecallsuccessthatdidnotdepend onwordtype.First,inthefrontalfactorapositivitypeakingat1400msafterthestimulusdiffered betweensubsequentlyrecalledandnotrecalledwords.ItisnotobviouswhichERPcomponentthis effectcorrespondsto,butitseemsthattheeffectrepresentsthesustainedpositivitythatisapparentin thevirtualERPsforsubsequentlyrecalledsizestandards,sizeisolatesandHFstandardsseegure 8B.Secondly,theoverallsubsequentmemoryeffectintheamplitudesoftheparietalnegativepeak at400msapproachedsignicance. Fortheoddballtaskdata,thePCAresultedinthreeinterpretablespatialfactors.Therstone wasafrontalfactorcharacterizedbyanarrowpeakat200msfollowedbyaslownegativepotential. Bothofthecorrespondingtemporalfactorsdifferedsignicantlybetweentargetsandstandards. Secondly,acentralfactorshowedapositivepeakat200ms,followedbyanegativepeakat300ms andanotherpositivepeakat400ms.Therstpositivityaswellasthenegativitydifferedbetween targetsandstandards.Lastly,andofbiggestinterestforthepurposeofthecurrentstudy,thethird spatialfactorhadaparietaldistributionandastrongpositive-goingvirtualERPthatmostlikely 55

PAGE 63

correspondstotheP300.Thespatialdistributionofthisfactorwasinclosecorrespondenceto theP300obtainedinthememoryexperiment.Unexpectedly,theP300latencydifferedbetween targetsandstandards,andwhentheappropriatetemporalfactorscoreswereanalyzed,therewasno signicantamplitudedifferencebetweenthetwostimulustypes. Ourmainndingswillnowbediscussedinmoredetailinthelightofpreviousstudiesandtheoreticalframeworks. 4.2CorrelationsofERPAmplitudesandRecallintheVonRestorffParadigm OurbehavioralandERPdatafromtheVonRestorfflistsreplicateandextendndingsofpreviousstudiesonERPcorrelatesoftheVonRestorffparadigm.Sizeisolatesweremorelikelyto berecalled,resultinginastrongandsignicantVonRestorffeffect.Furthermore,wordsthatwere displayedinalargerfontthantheotherwordselicitedaP300.TheamplitudeoftheP300elicited bysizeisolateswascorrelatedwithsubsequentrecall,suchthatlaterrecalledwordselicitedlarger P300amplitudesthanlaternotrecalledwordsalthoughthiseffectonlyapproachedsignicance. TheP300factorobtainedinthePCAshowedaposteriordistributionthatwasinclosecorrespondencetotheP300elicitedinthestandardoddballparadigm.ThelatencyoftheP300inthecurrent studywassurprisinglylong-itpeakedbetween700and800msafterthedeviantstimulus.Inprevious,analogousVonRestorffmemoryexperiments,P300latencyrangedbetween500and600ms Fabiani&Donchin,1995;Otten&Donchin,2000;Karisetal.,1984.P300latencyisthoughtto reectthetimeittakesthesubjecttoevaluatethedistinctivepropertyofastimulusKutasetal., 1977.SincethedistinctivenessattributeintheVonRestorfflistswasaphysicalmanipulationof fontsize,however,itshouldhavebeendetectedandevaluatedrelativelyquickly.Furthermore,since themanipulationwasanalogoustootherstudies,theexpectedlatencyshouldhavebeensimilarto thesestudies.Therefore,thereasonforthelongP300latencyinthisstudyremainstobeelucidated. ThecorrelationbetweenP300amplitudeandsubsequentrecallsuccessunderrotememoryinstructionsreplicatesalargenumberofpreviousstudiesFabianietal.,1986;Fabiani,Karis,& Donchin,1990;Fabiani&Donchin,1995;Otten&Donchin,2000.However,inthecurrentstudy thissubsequentmemoryeffectonlyapproachedstatisticalsignicanceforsizeisolates.Thenonsignicantndingcouldhaveresultedfromlowstatisticalpower,orfromlargebetweensubject variability,forexampleduetotheuseofdifferentmemorizationstrategies. 56

PAGE 64

SincewereplicatedpreviousndingsoncorrelationsbetweenP300amplitudeandsubsequent recallwhenwordsarephysicallydistinctiveandrotememorizationisused,ourdataareinlinewith theprevioustheoreticalframeworkusedtoexplainthisnding.TheideaisthatP300amplitude correlateswithsubsequentmemorywhenadistinctiveattributeistaggedontothememorytrace andattestusedforsuccessfulretrieval.TherelationshipbetweenP300andrecallonlyholdstrue whenparticipantsuseanitem-basedmemorizationstrategy.Iftheparticipantengagesinelaborationofitem-to-itemrelationshipsatstudy,theoutcomesofthiselaborationbecomemoreeffective forretrievalsothatretrievaldoesnotnecessarilyrelyonthedistinctivenessattributeFabiani& Donchin,1995;Donchin&Fabiani,1991. Inourstudy,sizeisolatesalsoelicitedaleftlateralizedslowwavecomponentthatwasstrongly andsignicantlycorrelatedwithsubsequentrecall.Inpreviousstudies,afrontallydistributedslow wavewascorrelatedwithsubsequentrecallfornon-isolatedwordsandforisolatesonlywhenelaborativestrategieswereusedFabiani,Karis,&Donchin,1990,buttherewasusuallynosuchcorrelationwhenparticipantsusedrotememoryOtten&Donchin,2000.Similarly,Karis,Fabiani andDonchinfoundthatwhiletheslowwavewascorrelatedwithrecallacrossallmemorization strategies,theeffectwas strongest whenelaborativestrategieswereusedKarisetal.,1984.Our ndingsaresomewhatincontrastwiththesestudies,becausewefoundaverypronouncedcorrelationbetweenslowwaveamplitudesandrecallalthoughparticipantswereinstructedtouserote memory,aninstructionthatwasfollowedbymostparticipantsaccordingtoself-reports. TheslowwaveintheearlystudieswasreportedtohaveafrontalmaximumFabiani&Donchin, 1995;Otten&Donchin,2000,althoughlatercorticalimagingofthesubsequentmemoryERP effectsintheVonRestorffparadigmmappedtheminparttothe left prefrontalcortexDingetal., 2002;Lianetal.,2002.Theleftlateralizationoftheslowwavefactorinourstudyisconsistent withthesecorticalimagingstudies.However,thespatialfactorloadingsinourstudyweremore posteriori.e.lessfrontalthanexpected.Onecouldsuggestthattherotationmethodusedinthe PCAskewedtheactualspatialdistributionofthecomponentintherawdata.However,itshould benotedthattheuseofadifferentrotation,likeVarimax,resultedinaverysimilardistributionof thisslowwavefactor,indicatingthatthePCArotationmethodisunlikelytobetheonlyexplanation forournding.Therefore,itremainsunclearwhethertheleftlateralizedslowwaveobtainedinthe currentstudycorrespondstothefrontalslowwavereportedinpreviousstudies. 57

PAGE 65

Inconclusion,inourstudybothapositiveslowwaveandtheP300werecorrelatedwithsubsequentrecallforsizeisolatesunderthesamememorizationmethod,whichissomewhatincontrast withpreviousstudiesreportingadissociationbetweencorrelationsofthetwocomponents'correlationstorecalldependingonwordtypeandmemorizationstrategy.Itispossiblethatparticipants usedamixtureofmemorizationstrategieswithoutadmittingthisduringdebrieng,resultingin thecorrelationofboththeP300andthefrontalslowwavetosubsequentrecallinourstudy. Thecomponentialinvestigationofourndingswithaspatio-temporalPCAallowedustodiscover twootherpositivitieselicitedbysizeisolatesthattemporallyoverlappedwiththeP300butexhibited differentspatialdistributions.Notethatinmostpreviousstudies,smallernumbersofelectrodes limitedthepossibilitiestoseparateapartcomponentswithsimilartemporal,butdifferentspatial distributions.Therefore,thendingsdiscussednextarenovel.Therstoneofthesepositive componentswaspresentinthefrontalfactoranditsamplitudewassignicantlycorrelatedwith subsequentrecallforsizeisolates.Althoughitsdistributionmaybeslightlymorefrontalthan wouldbeexpectedbasedonpreviousPCAstudiesDienetal.,2003,thisfactormayrepresentan instanceoftheP3aN.K.Squiresetal.,1975,ornoveltyP3Courchesne,Hillyard,&Galambos, 1975.IntherststudyontheP3a,thiscomponentwaselicitedbyinfrequentstimuliinunattended oddballsequencesN.K.Squiresetal.,1975.ThenoveltyP3isacomponentwithacomparable spatio-temporaldistributiontotheP3aandiselicitedinmodiedoddballsequencesbyathirdtype ofstimulus,whichisnovelandtaskirrelevantCourchesneetal.,1975;Dienetal.,2003.While someauthorsrefertotheP3aandtheP300astwosub-componentsofthesameERPcomponent Polich,Howard,&Starr,1983,theyhavedifferentspatialdistributionsandresponddifferentlyto experimentalmanipulationsandshouldthereforeberegardedastwoseparatecomponents. BasedonpreviousknowledgeabouttheelicitingconditionsoftheP3aorthenoveltyP3,the elicitationofthiscomponentinthecurrentstudyisnotentirelysurprising.Thus,whileobviously eachwordinastudy-testmemoryparadigmisrelevantforthesubject'stask,thelargerfontsize oftheisolate perse isnottaskrelevant.Itisthereforewellconceivablethatthefrontalpositivity correlatedwithsubsequentrecallforsizeisolatesinourparadigmisaninstanceoftheP3a.The correlationofthiscomponenttosubsequentrecallisanovelndingandneedsfurtherexploration infuturestudies.Forexample,itremainstobeelucidatedwhetherthepresenceofthiscorrelation dependsonmemorizationstrategy. 58

PAGE 66

Anotheradditionalpositivityelicitedbysizeisolates,whichalsopeakedat700msandtherefore overlappedwiththeP300,wasfoundintherightlateralizedspatialfactor.Thispositivitywas largerforsizeisolatesthanforanyotherwordtypeandcouldalsopossiblybeclassiedasaP3a. However,thiscomponentwasnotcorrelatedwithsubsequentrecall. ToportrayacompletepictureofERPcorrelatestosubsequentrecallintheVonRestorffparadigm, itisimportanttoalsoconsidersubsequentmemoryeffectsforsizestandardsi.e.theregularsizedwordsthatwerestudiedVonRestorfflists.Previousstudieshaveindicatedthatforthese non-isolatedwords,thefrontalslowwaveiscorrelatedwithsubsequentrecallFabiani,Karis,& Donchin,1990.Inthecurrentstudywewereunabletoreplicatethisnding.Thatis,wedidnot obtainanycorrelationsbetweenERPamplitudeselicitedbysizestandardsandsubsequentrecall exceptfortheoveralleffectsforsubsequentrecallacrosswordtypes,thatwillbediscussedlater. Mostimportantly,theleftlateralizedslowwavethatshowedstrongcorrelationswithsubsequent recallforsizeisolatesandforLFisolates,aswillbediscussedlaterwasnotsignicantlycorrelated withrecallforsizestandards.Thediscrepancybetweenourresultsandpreviousndingsisunclear, sinceourexperimentalparadigmcloselyresembledpreviousstudies. 4.3WordFrequencyEffectsandTheirReversalduetoListComposition ThewordfrequencyeffectonrecallmemoryforpurelistsreferstothesuperiorrecallofHF comparedtoLFwordsMandleretal.,1982.Inourstudy,duetothepresenceofonlyoneword ofdeviatingfrequency,theLFandHFlistscouldbeconsideredclose-to-pure,resultinginthe predictionthatthiswordfrequencyeffectonrecallforpurelistsshouldbeobservedforLFandHF standards.Indeed,theHFstandardsweresignicantlybetterrecalledthanLFstandardsatprimacy andplateaupositions. Anadditionalpatternofinterestisthatwordfrequencyeffectsonrecalldependonlistcomposition,suchthatwhenmixedlistsarestudied,LFwordsarerecalledequallywellorbetterthanHF wordsWatkinsetal.,2000;Fernndezetal.,2002;Hulmeetal.,2003;McDaniel&Geraci,2006. Inthecurrentstudy,weinvestigatedpossibleexplanationsforthisabolitionofthewordfrequency effectinmixedlistsbyinsertinganindividualwordwithafrequencythatwasoppositeoftherest ofthelistLFandHFisolates.Weexpectedthattheseisolatedwordswouldbehavesimilarto wordsinmixedlistsofLFandHFwordsalthoughquantitativedifferencesarelikely,becausein 59

PAGE 67

mostpreviousstudiesmixedlistsconsistedof50%LFand50%HFwords,asopposedto14words ofonetypeandonlyonewordoftheothertypeinourstudy. Wefoundthattheisolationinlistsofopposite-frequencywordsincreasedtheprobabilityofrecall forLFwords,butdecreasedrecalllevelsforHFwords.Thatis,theprobabilityofrecallwashigher forLFisolatesthanforLFstandards,butlowerforHFisolatesthanforHFstandards.Thisnding isinlinewiththepredictionthatLFwordsleadtoequalorsuperiorrecallthanHFwordswhen studiedinmixedlists,whichisthetypicalreversalofthewordfrequencyeffectonrecallinmixed lists. ItisimportanttonotethateachLFwordandeachHFwordwasequallylikelytobeselected asanisolate,becauseisolatesandstandardswereselectedatrandomfromthesamepoolsofHF andLFwords.Furthermore,thewordsthatsomeparticipantsencounteredasLFandHFisolates wereencounteredbyotherparticipantsasLFandHFstandards.Therefore,differencesinlinguistic propertiesbetweenisolatesandstandardsarenotlikelytoexplaintheenhancementofrecallforLF wordsorreductionofrecallforHFwordsduetotheirisolation. Itisworthnotingthatourbehavioralndingsonwordfrequencyeffectsonrecallaredifferent fromtheVonRestorffeffect.VonRestorffemphasizedtheimportanceofcomparingeachisolated stimulustorecalllevelsofnon-isolateditemsinthesamelistthatwerehenceofadifferentstimulus type,aswellastorecalllevelsofthesameitemtypewhenthestimulusdidnotoccurinisolation VonRestorff,1933.Shefoundenhancedrecallfortheisolateditemcomparedto both non-isolated itemswithinthesamesequenceandnon-isolateditemsofthesamestimulustypewhenoccurring asanon-isolateinadifferentlist.Inourstudy,isolatedLFwordswererecalledatan equallyhigh levelofHFstandardsatcomparableserialpositions,whileisolatedHFwordswererecalledanan equallylow levelasLFstandards.Therefore,theLFisolateswerebetterrecalledthannon-isolates ofthesametypeindifferentlistsLFstandards,butnotcomparedtothenon-isolatedwordsinthe samelistHFstandards. WhileLFisolatesdidshowonecharacteristicoftheVonRestorffeffectenhancedrecallcomparedtonon-isolatedwordsofthesametype,HFisolatesdidnotshowanyindicationsforamemoryenhancingisolationeffect.Thus,theywererecalledatanequallevelasnon-isolatedwordsin thesamelistLFstandards,butataneven lower levelthannon-isolatedwordsofthesametype HFstandards. 60

PAGE 68

Howdoourbehavioraldataonthereversalofthewordfrequencyeffectonrecallcomparetoother studiesthatmanipulatedlistcompositionbeyondcomparingpurelistswithmixedlistsof50%HF and50%LFwords?Inaserialrecallexperiment,DeLoshandMcDanielmanipulatedlist compositionsuchthatlistscontained8LFwords,8HFwords,2LFand6HFwords,or 2HFand6LFwords.RecallwasbetterforpureHFliststhanforanyotherlist,butrecallwas onacomparablelevelforlisttypes2,3and4.Whilethisstudy'sparadigmwasquitedifferentfrom ours,theseresultsareconsistentwiththeabsenceofaVonRestorffeffectinourstudywhenwords wereisolatedduetotheirwordfrequency.AnotherstudybySaint-AubinandLeBlancinvestigated effectsoflistcompositiononserialrecallSaint-Aubin&LeBlanc,2005.Theseauthorshad participantsstudyeitherpurelistsorlistsofvewordsofthesamefrequencyandonewordofthe oppositefrequency.Inadditiontotheusualwordfrequencyeffectforpurelists,theyfoundthat anisolatedLFwordinalistofHFwordswasequallylikelytoberecalledastheHFwordsinthe samelist.Whiletheexperimentalparadigmwasagainverydifferentfromours,thendingsare consistentwiththelackofaVonRestorffeffectforLFisolatesinourstudy.However,incontrast toourndings,theseauthorsalsofoundanadvantagefortheisolatedHFwordcomparedtotheLF wordsinthesamelist.ThisndingisinconsistentwithourresultthatHFisolatesleadtoequally lowrecalllevelsasLFstandards. Ourbehavioralndingsappeartosuggestamajorroleforinter-itemelaborativeprocessesinthe reversalofthefrequencyeffect.Accordingtothisidea,LFwords'recalliselevatedtothesamelevel ofHFwords'recallwhentheyoccurinlistsofHFwords,becausethelatterfacilitateelaborative processesbetweenlistitems.Forthesamereason,HFwords'recallisreducedtothelevelofLF word'srecallwhentheseoccurisolatedinLFlists,becausetheLFlistsdonotfacilitateelaborative processesbetweenlistitemstoanequalextent.Thus,thetheorythattheoutcomesofitem-to-item associativeprocessesareusedasretrievalaidsforrecall,andthatthelevelofsuchassociations dependsontheratioofHFandLFwordsinalistHulmeetal.,2003seemstobealikelycandidate toexplainthecurrentstudy'sbehavioralndings. Areinuencesofdistinctivenessonrecallalsoapossibleexplanationforourbehavioralresults? Asdescribedabove,iftheHFandLFisolateswereperceivedasdistinctivecomparedtotherest ofthelistinasimilarwayasalargewordinalistofregular-sizedwords,theywouldbeexpected toproduce higher levelofrecallthanthenon-isolatedwordsinthesamelist.Wedidnotobtain 61

PAGE 69

suchaneffectinthecurrentstudy.However,thereisonepossibilitythatmayupholdaninuence ofdistinctivenessatencodingtimeonrecall:ThatthereissomeintrinsicpropertyofHFwordsthat makesthemmorelikelytoberecalledthanLFwords.IsolatedLFwords,however,areperceived asdistinctiveinHFlistsleadingtoanisolationeffect.ThegeneralmemoryadvantageforHF wordsandthedistinctivenessadvantagefortheLFisolatethenelevateretrievaltoanequallyhigh level,resultingintheapparentabsenceofaVonRestorffeffectobservedinourdata.Itisunclear, however,howthispossibleexplanationwouldaccountfortheimpairmentinrecallofHFwords whentheyoccurinisolation.ThisideawillbefurtherdiscussedaftertheoutlineofERPcorrelates associatedwithwordfrequencyeffectsonrecall.Overall,ifweonlyconsiderthebehavioralresults, theyappeartofavortheoriesbasedonelaborativeprocessesoverdistinctiveness-basedexplanations ofthewordfrequencyeffect'sreversalinmixedlists. Asnotedintheintroduction,aspecicpatternofbehavioralndingscanbeassociatedwithdifferentERPeffectsOtten&Donchin,2000.Therefore,inordertotestthedistinctivenessand inter-itemelaborationtheoriesforthereversalofthewordfrequencyeffectinmixedlists,weinvestigatedwhichERPcomponentswerepronouncedandcorrelatedwithsubsequentrecallforLF andHFstandardsandisolates.Specically,theseERPeffectswerecomparedtothesubsequent memoryeffectsfoundintheVonRestorfflistsdiscussedabovetoinvestigatewhethersubjective distinctivenessplaysaroleinwordfrequencyeffectsonrecallaswell.Thesedataarediscussed next. 4.4CommonProcessesoftheVonRestorffEffectandWordFrequencyEffects? Thecurrentstudywasdesignedtoinvestigatewhetherwordfrequencyeffectsonrecall,specicallythereversalofthewordfrequencyeffectinmixedlists,shareunderlyingcognitiveprocesses withtheVonRestorffeffectinrecall.WeapproachedthisquestionbycomparingtheERPsand theirsubsequentmemoryeffectselicitedbyphysicallydeviantwordstothoseelicitedbywords ofdifferentfrequencieswhentheyoccurredclusteredorinisolation.First,thediscussioninthis sectionwillfocusonERPselicitedbytheLFisolates.WedirectlycomparetheLFisolatestosize isolates,anditseemedthatthetwowordtypeselicitedverysimilarERPcomponentsandsubsequentmemoryeffects.Furthermore,wewillcomparetheanalysisofoutputorderforLFisolates andsizeisolatesduringtherecalltesttoinvestigatewhetherbothisolationtypesareavailableas 62

PAGE 70

organizationalaidsforretrieval.Afterwards,wewilldiscussthesubsequentmemoryeffectsforHF andLFstandards,followedbyasectiononthequitedifferentsubsequentmemoryeffectselicited byHFisolates.Nextfollowsasectiononthetwosubsequentmemoryeffectsthatwerefoundacross allwordtypesincludingsizeisolatesandstandards.Theconclusionofthissectionwillfocuson theoverallimplicationsofourndingsfortheoriesonthereversalofthewordfrequencyeffectin recall. 4.4.1ERPComponentsandCorrelationstoRecallforLFIsolates TheLFisolatesrepresentthewordtypethatismostcriticalfortestingourhypotheses,because thesearethewordsthatcouldbeeitherperceivedasdistinctive,orassociatedwithstrongerinteritemelaborativeprocessesthanLFstandards.Todistinguishbetweenthesepossibilitiesisthemain purposeforthecurrentstudy. OurdataindicatedthatLFwordsisolatedinlistsofHFwordswereassociatedwithverysimilar ERPsignaturesandsubsequentmemoryeffectscomparedtosizeisolates.First,LFisolateselicited aP300componentwhoseamplitudewasstronglyandsignicantlycorrelatedtosubsequentrecall. ThisndingseemstoindicatethatwhenLFwordsoccurinlistsofHFwords,theyareindeed perceivedasdistinctiveandthatthisdistinctivenessattributeofthememorytraceisusedeffectively asaretrievalaid.This,inturn,indicatesthatthereversalofthewordfrequencyeffectinmixed listscaninpartbeexplainedbyenhancedsubjectivedistinctivenessofLFwordswhenHFwords arepresent.Therefore,theERPdataofourstudyimplythatcommonprocessespartlyunderliethe effectsofphysicaldistinctiveness,andwordfrequencyonrecall. Thisimplicationsomewhatconictswiththeconclusionsonemaydrawfromthebehavioral results,asdiscussedintheprevioussection.Thus,ifLFisolatesareindeedperceivedasdistinctive, andifthisdistinctivenessattributeofthememorytraceiseffectivelyusedinretrievalprocesses asimpliedbythecorrelationofP300amplitudewithrecall,onecouldarguethatrecallshould beenhancedforLFisolatescomparedto both LFstandardsandHFstandards.Incontrasttothis expectation,therecallofLFisolateswasonanequallevelasforHFstandardsatthecorresponding serialpositions.Therearetwopossibleexplanationsforthisdiscrepancybetweenourbehavioral andERPndings. 63

PAGE 71

First,asoutlinedintheprevioussection,LFwordsmaybeinherentlymoredifculttorecallthan HFwords.RetrievalprocessesmaybemoreeffectiveforHFwordsingeneralbecausetheireasier lexicalaccessmayfacilitatethereconstructionoftheitemfromanincompleteornoisymemory trace-aprocessthathasbeennamedredintegrationHulmeetal.,1997.Atthesametime,recall ofLFwordsisolatedinHFlistsbenetsfromtheirsubjectivedistinctivenessinasimilarwayas physicallydistinctivewords,elevatingrecalllevelsforLFisolatesaswell.Theenhancingeffects offacilitatedredintegrationprocessesforHFwordsandtheenhancingeffectofdistinctivenessfor LFisolatesthenelevaterecalltocomparablelevels.Aproblemwiththisexplanationisthatit attributesthegeneralsuperiorityofHFwordsinrecalltoitem-basedeffectsatretrievalwithout attributinganyimportanceonstudycontext.Thisisinconsistentwiththereducedprobabilityof recallforHFisolatescomparedtoHFstandards.Thatis,ifrecallofHFwordsisenhancedbecause memorytracesofHFwordsareeasiertoreconstructatretrieval,thisshouldbetrueforallHF wordsregardlessofstudylistcomposition.Itappears,therefore,thattheenhancementofrecall levelsofHFstandardsandLFisolatestothesamelevelduetofacilitatedredintegrationprocesses anddistinctivenesseffects,respectively,isanunlikelyexplanationforourndings. AnotherpossibleexplanationforthediscrepantndingsbetweenbehavioralandERPresultsis thatduetovarianceincertaincharacteristics,someLFwordsstandoutfromlistsofHFwords whileotherLFwordsdonotstandouttoanequalextent.Inotherwords,someLFisolatesmaybe perceivedasdistinctive,elicitalargeP300andaresuccessfullyrecalled.OtherLFisolates,inturn, arenotperceivedasdistinctive,thatis,theyblendinwiththeHFwords,theydonotelicitmuchof aP300,anddonotleadtoenhancedrecall.ThismixtureofLFisolatesperceivedasdistinctiveand LFisolatesnotperceivedasdistinctivecomparedtotheHFstandards,then,ledtothecorrelation betweenP300andsubsequentrecallaswellasattenuatedlevelsofrecallforLFisolates.Inlinewith thisidea,itisapparentfromtheoutlineofdifferentpropertiesthatmaymakeLFwordsdistinctive seeintroduction,thatthereisahighvariabilityincharacteristicswithinthegroupofLFwords. WhichcharacteristicscouldcauseonlyasubsetofLFisolatestobesubjectivelyperceivedas distinctivewhenoccurringinlistsofHFwords?Giventhatwordfrequencyisusuallyconsidereda measureofpriorexposuretoaword,itappearsthattheindividual'ssubjectivefamiliaritywitheach wordmayplayarole.TheideaisthatifaLFisolateisveryunfamiliartotheparticipant,itislikely tostandoutfromalistofHFwords,whileLFisolatesthatareratherfamiliartotheparticipant 64

PAGE 72

maybeperceivedassimilartotherestofthelist.Ifthisistheexplanationforourndings,LF isolatesthatwerelaterrecalledshouldbeassociatedwithlowersubjectivefamiliarityratingsthan LFisolatesthatwerelaternotrecalled.Ananalysisofthesubjectivefamiliarityratingsgivenin thequestionnaireattheendoftheexperiment,however,didnotsupportthishypothesis.Thus, subsequentlyrecalledLFisolatesandsubsequentlynotrecalledLFisolatesreceivedaboutequal familiarityratingsthatwerenotsignicantlydifferentseegure2B. ThereareanumberofotherpropertiesofLFisolatesthatmayhaveledtoasubjectivedistinctivenessofonlyasubsetoftheLFisolates.Asoutlinedintheintroduction,LFwordstendtodifferfrom HFwordsintermsoforthographicorphonologicaluniquenessCriss&Malmberg,2008;Landauer &Streeter,1973,lowlevelsofcontextvariabilityAdelmanetal.,2006;Steywers&Malmberg, 2003,orlownumbersofconnectionstootherwordsinlexicalmemoryCofer&Shevitz,1952; Nelson&McEvoy,2000.DifferentlevelsoneormoreofthesecharacteristicsfordifferentLFisolatesmayhaveledtothesubjectiveperceptionofonlysomeLFisolatesasdistinctive.Ourcurrent analysisdoesnotallowforanystrongconclusionsaboutwhichcharacteristicsmayhaveplayeda crucialrole. RegardlessofwhichexplanationaccountsforthediscrepancybetweenourERPndingsandour behavioralndings,LFwordsinlistsofHFwordsdidelicitaP300whichwascorrelatedwith subsequentrecall.ThisistheERPpatternobtainedforsizeisolatesinourstudyandmanypreviousstudieswhenrotememorywasused.Conclusively,outcomesofcontextupdatingprocesses elicitedbyphysicallydeviantandbyLFwordsinlistsofHFwordsappeartobeeffectiveretrieval cuesattimeofrecall.ThereversalofwordfrequencyeffectsonrecallinmixedlistsmaythereforebeexplainedinpartbydistinctivenesseffectscomparabletothoseelicitedintheVonRestorff paradigm. InadditiontotheP300,LFisolatesalsoelicitedaleftlateralizedpositiveslowwave-thesame componentwealsofoundforsizeisolatesintheVonRestorfflists.Likeforsizeisolates,this componentwascorrelatedwithsubsequentrecallforLFisolates.Thedifferencebetweenrecalled andnotrecalledwordswas,however,onlystatisticallysignicantduringtheearliertemporalfactor overlappingtheslowwaveeffectabout700to1000msafterthestimulusforLFisolates. Asnotedabove,itisunclearwhetherthisleftlateralizedslowwavecomponentreectsthefrontal positiveslowwavereportedinpreviousstudies,whichhasbeenassociatedwithinter-itemelabo65

PAGE 73

rativememorizationprocesses.Iffutureresearchrevealsthatthetwocomponentsarethesame, ourndingscanbeconsideredevidencefortheinvolvementofenhancedelaborativeprocessesassociatedwithLFwordsoccurringinHFwordlistscomparedtobothHFandLFstandards.An additionalexplanationforthereversalofwordfrequencyeffectsonrecallinmixedlistsistherefore thatoutcomesofitem-to-itemelaborationLFwordsembeddedinlistsofHFwordsareenhanced andeffectivelyusedforretrieval. Thefrontalpositivitythatwascorrelatedwithsubsequentrecallforsizeisolates,didnotshow suchacorrelationforLFisolates.Aswesuggestedabove,thispositivitymaybeaninstanceof theP3a.Itappears,therefore,thattheamplitudeofthiscomponentispositivelyassociatedwith subsequentrecallforphysicaldeviantstimuli,butnotfordevianceintermsofwordfrequency. Insummary,LFisolateswereassociatedwithERPsubsequentmemoryeffectsthatshowedstrikingsimilaritiestosizeisolates.TheyelicitedaP300thatwascorrelatedwithsubsequentrecall. Furthermore,theleftlateralizedslowwavewasalsocorrelatedtosubsequentrecall.Thesendings indicatethatbothsubjectivedistinctivenessatstudytimeaswellasworkingmemoryprocessessupportingelaborativeprocessesbetweenitemsarethebasisforenhancedrecallofLFwordsinlistsof HFwords.Inotherwords,bothitem-specicandelaborativeprocessesappeartoplayaroleinthe reversalofwordfrequencyeffectsonrecall. 4.4.1.1ComparisonofOutputOrderforLFIsolatesandSizeIsolates InapreviouslypublishedERPstudyontheVonRestorffparadigm,FabianiandDonchininvestigatedtheeffectofisolationtype-comparingsemanticandphysicalisolates-ontheprobability thatthisisolatewillbewrittendownrstorlastontherecallsheetFabiani&Donchin,1995.The ideawasthatiftheisolationfeatureisavailableasanorganizationalaid,theisolateditemshould beretrievedsomewhatseparatelyfromtheotheritemsinthelist.Theauthorsfoundthatphysical isolatesweremostlikelytoberecalledlastwhilesemanticisolatesweremorelikelytoberecalled rst,comparedtoallotheroutputpositions.Theauthorsinterpretedthisndingsuchthatbothsemanticandphysicalisolationfeaturesareavailableasorganizationalaidsinmemoryretrieval,but thatthetwoisolationfeaturesareuseddifferentlyfromeachother. Inasimilarcomparison,wefoundthatbothsizeisolatesandLFisolatestendedtobewritten downattheendoftherecallsheet.However,thispatternwasonlysignicantforLFisolates.This 66

PAGE 74

ndingoncemorecorroboratesthatthedistinctivenessofLFisolatesandphysicalisolatesactupon memoryprocessesinasimilarwayatleastwhentheresultsforphysicalisolatesofFabianiand Donchin'sstudyaretakenintoaccountaswell.Thus,bothLFisolatesandsizeisolatesappearto besimilarlyavailableasorganizationalaidsinmemory. Notethatanalternativeinterpretationofourresultsonoutputorderisthattheyareonlyabyproductofthespecicretrievalstrategy.Thus,participantsmayrstrecallthelastwordsofthestudy lists,thencontinuewithwordsfromthebeginningofthelists,andnallyproceedtothemiddlelist positions.Sinceisolatesalwaysoccurredatserialpositions6-10,thesewordswouldautomatically tendtobewrittendownlastduringrecall.Ourcurrentanalysiscannotruleoutthispossibility,so furtheranalysisisneededtodrawmoredeniteconclusions. 4.4.2ERPComponentsandCorrelationstoRecallforHFandLFStandards Inordertostudyunderlyingneuro-cognitiveprocessesofwordfrequencyeffectsonmemory,not onlyERPcomponentselicitedbythewordsisolatedinlistsofopposite-frequencywords,butalso thoseelicitedbytheHFandLFstandardsareofinterest.Comparabletothenon-isolatedwords intheVonRestorfflists,theHFstandardsdidnotelicitsignicantsubsequentmemoryeffectsfor anyspatio-temporalfactorsthatwereuniquetothiswordtype.Theonlytwooverallsubsequent memoryeffectsthatappearedtobepresentacrosswordtypeswillbediscussedinalatersectionof thischapter. LFstandards,inturn,elicitedtwodistinctsubsequentmemoryeffectswhichareworthdiscussing. First,thefrontalpositivitythatwaspositivelycorrelatedwithsubsequentrecallforsizeisolates, wasnegativelycorrelatedwithrecallforLFstandards.Thatis,LFstandardsthatwerelaterrecalled elicitedsmallerfrontalpositivitiesthanLFstandardsthatwerelaternotrecalled.Asdiscussed above,thisERPcomponentmaybeaninstanceoftheP3a.Ifthecomponentinourstudyindeed reectsaP3a,thisndingcouldindicatethatphysicaldeviancethatmaybeinherenttosomeLF wordspossiblyduetounusuallettercombinationsactually harms theirprobabilityofsubsequent recallwhentheyoccurclusteredwithotherLFwords.Apossibleinterpretationisthatduetolarger difcultyforinter-itemelaborativeprocessesbetweenLFwords,thosewordsthataresomewhat deviantasreectedintheP3aareintegratedwiththeotherwordsevenmoreinefciently.Forthis reason,thepresenceofalargeP3amayharmrecallforLFstandards.Thisexplanationisrather 67

PAGE 75

speculative,anditisunclearwhysomeLFstandardsmayelicitalargerP3abutshownoevidence ofaP300.Therefore,moreresearchisneededtoclarifythenegativecorrelationoftheamplitudeof ourfrontalpositivecomponenttosubsequentrecallforLFstandards. ThesecondcomponentthattendedtodifferbetweenrecalledandnotrecalledLFstandardswas theleftlateralizedpositiveslowwave.Again,thecorrelationofthiscomponent'samplitudewith recallsuccesswasintheoppositedirectionthanforsizeisolates.Thatis,subsequentlyrecalledLF standardselicited smaller slowwaveamplitudesthansubsequentlynotrecalledLFstandards.This ndingisagaindifculttointerpret-ifthisslowwavecomponentindeedplaysaroleininter-item elaborativeencodingprocesses,onewouldexpectthattheLFstandardsthatelicitlargercomponent amplitudeswillbemorelikelytoberecalled.Itisunclearwhystrongerelaborativeprocesses shouldleadtoworserecallforLFstandardsandthereforetheexplanationforourndingneedsto beclariedinfurtherresearch. Conclusively,ouroverallndingsonsubsequentmemoryeffectsofLFstandardsdonotappear tohaveobviousinterpretations.Itisclear,however,thatbothsubsequentmemoryeffectsforLF standardsareintheoppositedirectionthanthesubsequentmemoryeffectsforthesamecomponents elicitedbysizeisolates.Thissuggeststhatdifferentneuro-cognitiveprocessesatencodingare effectiveforlaterretrievalofLFwordsthatoccurinclusterscomparedtophysicallydeviantwords. Itisthereforenotlikelythatwordfrequencyeffectsonrecallwhenpurelistsarestudiedcanbe attributedinanyextendtoprocessesthatalsoleadtoeffectsofsubjectivedistinctivenessonrecall asobservedintheVonRestorffparadigm. 4.4.3CorrelationsofN400AmplitudesandRecallSuccessforHFIsolates TheHFisolatesarethelastwordtypethatneedsdiscussiontogetacompletepictureofERP componentsandtheircorrelationstosubsequentrecallforwordfrequencyeffectsinmixedlists. Asoutlinedabove,theHFisolatesareanimportantcontroltotheLFisolates.Thus,anyeffects thatoccurfor both HFandLFisolateswouldneedtobeinterpretedsuchthattheircauseisnotthe frequencyoftheword perse ,butratherwhetherit deviates fromthefrequencyofthesurrounding words.IfthesameeffectswefoundforLFisolatesarenotobtainedforHFisolates,interpretations canincorporateideasofsecondarydistinctivenessaccountsthatimplythatLFwords,butnotHF wordsexhibitsomelevelofdistinctiveness. 68

PAGE 76

Itisclearinourdata,thatERPsubsequentmemoryeffectswerequalitativelyverydifferentfor HFisolatescomparedtoLFisolates.Therefore,thecorrelationstosubsequentrecallforP300and slowwaveappeartobespecictoLFisolatesandsizeisolates,andarenotsimplyduetothefactthat thefrequencyoftheisolatediffersfromtherestofthelist.Hence,itseemsjustiedtoconcludethat LFisolatesinlistsofHFwords,butnotHFisolatesinlistsofLFwordsaresubjectivelyperceived asdistinctivesimilarlytowordsthatdeviateduetofontsize. DespitethelackofP300andslowwavecorrelationstorecallforHFisolates,wefoundacorrelationbetweenN400amplitudeandsubsequentrecall.Thus,HFisolatesthatwerelaterrecalled elicitedasmallernegativeN400amplitudethanwordsthatwerelaternotrecalled.Furthermore,the positivityfollowingtheN400withthesamespatialdistributionwaslargerforsubsequentlyrecalled HFisolatesthanfornotrecalledHFisolates.Thisndingwasunexpected,sincepreviousstudies didnotndcorrelationsbetweenN400amplitudeandsubsequentrecallFabiani&Donchin,1995. Notethatinourstudy,overallN400amplitudedidnotdifferbetweenHFisolatescomparedtoany otherwordtype. Asoutlinedintheintroduction,theN400hasbeenimplicatedinsemanticintegrationprocesses Kutas&Hillyard,1980.Itsamplitudeislargerforunexpectedwordsthatareinadifferentsemanticcategorythantheexpectedwordcomparedtounexpectedwordsthatarewithintheexpected categoryFedermeier&Kutas,1999.Thisandotherndingsimplythatthemoreeffortisneeded tosuccessfullyintegrateawordintothesemanticcontext,thelargeristheN400amplitude.TheresultsofthecurrentstudyindicatedthatitismorelikelyforaHFisolatetoberecalledwhenitelicits asmallN400amplitude.CombinedwithpreviousknowledgeontheN400,ourndingssuggestthat thelesseffortisneededtosemanticallyintegrateaHFwordintothesemanticcontextestablished byalistofLFwords,themorelikelyitisthatthiswordwilllaterberecalled.Semanticdeviation, whichhasbeensuggestedasadifferenttypeofdistinctivenessFabiani,2006,isthereforenot benecialforencodingandretrievingHFwordsisolatedinlistsofLFwords. Itisunclearhowthisndingcanbeinterpretedwithrespecttowordfrequencyeffectsonrecall.Generally,thereshouldnotbesemanticdifferencesbetweenourpoolofHFandLFwords. However,iftherewereanysemanticdifferences,anN400effectshouldalsobeobtainedforLF isolates,whichwedidnotobserveinourstudy.Therefore,thecorrelationbetweenN400amplitude andsubsequentrecallforHFisolatesisprobablybasedonnaturaltrial-bytrialvariabilityinN400 69

PAGE 77

amplitudes.WhyN400relatedprocessesatencodingcouldharmsubsequentrecallforHFisolates remainsanopenquestion. Lastly,itisimportanttonotethatduetothesmalltrialnumbersofHFisolates,theERPdatawere quitenoisy.ItispossiblethatsomeadditionalsubsequenteffectsforHFisolateswereobscuredby thesehighlevelsofnoiseinthedata.However,visualinspectionoftheERPsdonotindicatethat thereisevenatendencyforaP300oraslowwaveeffectforHFisolates,soitappearssafeto concludethatthesewerespecictoLFandsizeisolates. 4.4.4SubsequentMemoryEffectsAcrossWordTypes Thereweretwomaineffectsforsubsequentrecallsuccessthatseemedtobepresentacrossword types.Bothoftheseeffectswereunexpectedandappeartoraisemorequestionsthantheyanswer. However,sincethesesubsequentmemoryeffectswerefoundforallwordtypesitisworthtobriey discusseachofthem.Itisimportanttokeepinmindthatduetothreemissingdatapoints,theHF isolateswerenotincludedintheoverallstatisticaltestthatresultedinthesemaineffects,sothe followingndingsmaynotgeneralizetothiswordtype. Therstoverallsubsequentmemoryeffectwasobtainedforapositivitythatpeakedat1400ms afterthestimulus,whichwaspresentinthefrontalfactor.Thisfactorappearstorepresentthe sustainedpositivitythatisapparentinthevirtualERPsforsubsequentlyrecalledsizestandards, sizeisolatesandHFstandardsseegure8B.Itmayberelatedtofrontalslowwaveeffects,but showsanunusualtemporaldistribution.Therefore,itisunclearwhichERPcomponentthisfactor maycorrespondtoandhowtointerpretthefunctionalsignicanceofitscorrelationtosubsequent recall. Secondly,theoverallsubsequentmemoryeffectintheamplitudeoftheparietalnegativepeak at400msapproachedsignicance.Thus,thiscomponentwaslargerforsubsequentlynotrecalled wordscomparedtorecalledwords.ItremainstobeinvestigatedwhetherthereisapossiblerelationshipbetweenthiscomponentandtherightlateralizedN400component. 4.4.5ImplicationsofourDataforWordFrequencyEffectsonRecall Overall,ourdataprovidestrongsupportforthedistinctivenesshypothesisofthereversalofword frequencyeffectsonrecallwhenmixedlistsarestudied.AlthoughthebehavioraldataonLFisolates 70

PAGE 78

donotexactlyresembletheVonRestorffeffect,aP300waselicitedandcorrelatedwithsubsequent recallforLFwordsthatoccurredasisolatesinlistsofHFwords.ThisisthesameERPeffectas observedforsizeisolatesinourstudyaswellasseveralpreviousstudies.Thisndingsupports theideathatsubjectsperceiveLFwordsasdistinctivewhentheyaresurroundedbyHFwords,and thatthisperceiveddistinctivenessaidsrecallinasimilarwayasaphysicaldistinctivenessattribute. HFisolates,inturn,donotappeartobearsubjectivedistinctiveness,indicatingthatthelowword frequencyisnecessaryfortheperceiveddistinctiveness,andnotonlywhetherthefrequencyof aword differs fromtherestofthelist.Conclusively,thememoryenhancingeffectofLFword's perceiveddistinctivenesswhentheyaresurroundedbyHFwordsmayplayalargeroleinthereversal ofwordfrequencyeffectsonrecallinmixedlists. AnopenquestioniswhatcharacteristicoftheLFwordsmakethemsubjectivelydistinctivecomparedtotheHFwords,elicitingtheP300andenhancingtheprobabilityofrecall.Thedistinctive attributedoesnotappeartobethesubjectivefamiliarity,sincewefoundnodifferenceinfamiliarity ratingsforrecalledandnotrecalledLFisolates.Onepossibilityisthatunusuallettercombinations intheLFwordsmakethesewordsdistinctivecomparedtootherwordsCriss&Malmberg,2008. Thisorthographicdistinctivenessexplanationwouldimplythatperceptual,orphysical,deviance playsaroleinourndings,comparabletotheVonRestorffeffect. AsecondquestionthatrequiresfurtherinvestigationishowtheprobabilityofaLFwordoccurringinalistaffectswhetheritisperceivedasdistinctiveasindexedbytheP300andwhetherthis distinctivenessattributeiseffectivelyusedatretrieval.Thus,wehaverepeatedlystatedthatour ndingsmayprovideexplanationsforreversalofwordfrequencyeffectsinmixedlists.However, typicallysuchndingsareobtainedinlistscontaining50%ofeachfrequencycategory.Inourstudy, onlyoneof15wordswasofoppositefrequency.Thereisstrongpriorevidencethatthehigherthe probabilityisthatadevianteventoccurs,thesmalleristheamplitudeoftheP300Duncan-Johnson &Donchin,1977,andtherebytheevent'sperceiveddistinctiveness.Therefore,itremainstobe investigatedwhetherLFwordsstillelicitaP300thatiscorrelatedtorecallwhentheprobabilityof aLFtooccuronagiventrialiscloserto50%.Inotherwords,beforeourdatacanbegeneralizedto otherlistcompositions,listscontainingalargerproportionofLFwordswillneedtobestudiedina similarparadigmtoours. 71

PAGE 79

Asdiscussedbefore,ourdataalsoprovidesupportforaninter-itemelaborationtheoryonthe wordfrequencyeffectonrecallaswellasitsreversalinmixedlists.Itislikelythattheleftslow waverepresentsthefrontalslowwavefoundinpreviousstudies.Thiscomponentwaspronounced andcorrelatedtorecallforbothLFisolatesandsizeisolates.Itappears,therefore,thatitem-toitemelaborativeprocessesatencodingareenhancedforLFwordsembeddedinHFlists,andthat theseareusedaseffectiveretrievalcuesatthetimeofthememorytest.Oneproblemwiththis interpretationisthatHFstandardsdidnotelicitsuchaslowwaveandacorrelationofitsamplitude tosubsequentrecall.Thisissuewillbediscussedinmoredetailinthenextsection. Notethatourdatacannotdistinguishbetweendifferenttypesofelaborativeprocessesthatmay bethebasisforthereversalofthewordfrequencyeffect.Notenoughisknownabouttheslowwave todrawanyconclusionsaboutwhichtypesofelaborativeprocessesitisassociatedwith.Therefore, whethertherelevantelaborativeprocessesaretheencodingoforderinformationDeLosh&McDaniel,1996;Merrittetal.,2006,theactivationofpre-existingassociationsHulmeetal.,2003, oryetothertypesofinter-itemencoding,remainsanopenquestion. Inconclusion,ourdataareinlinewiththeoriessuggestingthatbothitem-basedprocessesaswell asinter-itemelaborativeprocessesneedtobeconsideredtoexplainthereversedwordfrequencyeffectonrecallmemorywhenmixedlistsofbothHFandLFwordsarestudiedDeLosh&McDaniel, 1996;Tse&Altarriba,2007. 4.5ImplicationsoftheCurrentStudyfortheFrontalSlowWave Inadditiontotheimplicationsforprocessesunderlyingwordfrequencyeffectsonrecall,our dataalsoprovidenewinformationontheconditionsunderwhichtheslowwaveispronounced andtherebyforinterpretationsofitsfunctionalsignicance.Asdiscussedabove,onenewnding worthdiscussingisthatourspatio-temporalPCArevealedadifferentspatialdistributionoftheslow wavecomponentthanexpectedbasedonpreviousndings.Thus,theslowwavepreviouslyreported showedafrontalmaximum,butourslowwavewasleftlateralized.Whileitremainsunclearwhether theslowwavefoundinourstudycorrespondstotheslowwaveobservedinpreviousstudies,its temporalmorphologysuggestsso. IfourslowwaveindeedisaninstanceofthefrontalpositiveslowwavefoundinpreviousstudiesonasimilarparadigmFabiani,Karis,&Donchin,1990;Karisetal.,1984;Otten&Donchin, 72

PAGE 80

2000,ourndingsraisesomeadditionalquestions.Thus,purelistsofHFwordsarethoughtto facilitateitem-to-itemelaborativeprocesses,whilepurelistsofLFwordsdosotoalesserextent seeprevioussection.IfweassumethatHFwordselicithigherlevelsofinter-itemelaborative processes,andifwefurtherassumethattheslowwavemanifestssuchelaborativeprocesses,then thepredictionwouldbethatHFwordselicitlargerslowwaveamplitudesthanLFwordsinpure lists.Forexample,previousndingsbyFabiani,KarisandDonchinindicatedthatwhen subjectsusedelaborativestrategies,thefrontalslowwavewasmorepronouncedthanwhensubjects usedrotememorization.However,inourstudytherewasnosignicantdifferencebetweentheslow waveamplitudeelicitedbyHFstandardsandLFstandards. Furthermore,ifforHFwords,outcomesofitem-to-itemassociativeprocessesaremorelikelyto beusedasretrievalaidsthanforLFwords,weshouldalsoobserveastronger correlation between slowwaveamplitudesandsubsequentrecallforHFstandardsthanforLFstandards.Again,we didnotobservethispatternofresults.Inthecontrary,therewasnosignicantcorrelationbetween slowwaveamplitudesandrecallforHFstandards.ForLFstandards,therewasevenatendencyfor anegativecorrelationbetweenslowwaveamplitudesandrecall,butonlyfortheearliertemporal factoroverlappingtheslowwave. So,ifHFlistsfacilitateitem-to-itemprocessing,thenwhydidHFstandardsnotelicitlargerslow waveamplitudesandcorrelationstorecallthanLFstandards?Oneimportantpointtonoteisthatwe manipulatedcharacteristicsofthestimulusmaterial,notwhichencodingprocessestheparticipant wouldactivelyandconsciouslyengageintomemorizeit.Fabiani,KarisandDonchinhad participantsstudylistsinwhichsomewordsweresemanticallyrelatedusingtwodifferentitembasedprocessingtasks-eitherasemantictaskorastructuraltask.Onemaypredictthatinsucha design,thesemantictaskwoulddirecttheparticipanttoprocessthesemanticrelatednessbetween stimulimoreintensivelythanwiththestructuraltaskandthatduetotheseenhancedinter-item processes,thefrontalslowwaveshouldbemorepronouncedforthesemantictask.However,there werenoindicationsintheirdatatosupportthisprediction.Animportantparallelbetweenthe studybyFabianiandDonchinandthecurrentstudyisthatbothstudiesdidnotmanipulate whetherornottheparticipantwasactivelyengagedinelaborativeprocessingatstudytime.Intheir study,thesemanticjudgementwasmadeforeachitemindividually,andinourstudy,participants wereinstructedtouserotememory.Therefore,inbothstudies,ifstrongeritem-to-itemelaborative 73

PAGE 81

processestookplaceinoneoftheconditions,thesewereaby-productofthestimulusmaterialor theprocessingtask,respectively,andmaythereforehavebeenautomaticanduncontrolled. Inconclusion,ourdatacontributetopreviousknowledgeontheslowwaveindicatingthatitis pronouncedandcorrelatedwithsubsequentrecallfornon-isolatedwordsonlywhenparticipants areengagedinataskthatrequiresactive,controlledprocessingofstimulusmaterialinworking memory-aprocessthatlikelyreliesontheprefrontalcortex. Itremainstobeinvestigatedwhetherinstructionstouseitem-to-itemelaborativestrategies,such asformingsentenceswiththestudywordsasopposedtorotememorizationinstructionslikein thecurrentexperiment,havedifferenteffectsonslowwaveselicitedbyHFwordswhencompared toLFwords.Thatis,iftheseinter-itemelaborativetasksareeasiertoaccomplishforlistsofHF wordsthanforLFwords,andifthesetasksarefurthermoreassociatedwithincreasedslowwave amplitudes,onemayndadifferenceinslowwaveamplitudesandtheircorrelationtosubsequent recallforpureHFlistscompaeredtoLFlistswhensuchprocessingtasksareused. 4.6AbnormalitiesintheCurrentStudy'sOddballParadigm Instandardoddballparadigms,theinfrequent,taskrelevantstimulielicitaP300,butthefrequent stimulidonotelicitaP300Donchin,1981;Duncan-Johnson&Donchin,1977;Kutasetal.,1977. TheoddballparadigminourstudywastypicalinthattheOoccurredwithalowprobability%, whiletheX'soccurredwithahighprobability%.Thus,inthisparadigmweexpectedtond alargeP300amplitudeintheaverageERPselicitedbytheinfrequentO's,butnotforthefrequent X's.Incontrasttothisexpectation,wefoundlargeparietalpositivitiesforboththeinfrequent and thefrequentstimuli,whilethelatencywasshorterforthefrequents.Whentheappropriate temporalfactorswerestatisticallycompared,therewasnosignicantdifferencebetweentheP300 amplitudeselicitedbyfrequentsandinfrequents. Thisndingwasunexpectedandthereasonforthisunusualndingisunclear.Itisunlikelythat thendingisduetoanabnormalsample,sinceananalogousndingwasobtainedwiththeidentical paradigmusedonadifferentsampleofabout60studentsTyBrumback,personalcommunication. Onepotentialinuenceontheseunusualresultsmaystemfromtherequirementofeachparticipant torespondtoeachstimulus,includingfrequents and infrequents.However,previousstudiesfound thatindependentlyofwhichstimulusthesubjectrespondsto,atask-relevantinfrequentstimulus 74

PAGE 82

willelicitaP300whileatask-relevantfrequentstimuluswillnotK.C.Squires,Donchin,Herning, &McCarthy,1977.Therefore,requiringparticipantstorespondtobothstimuliisanunlikely explanationtheabsenceofamplitudedifferencesintheP300elicitedbystandardsandtargetsinour study. Anotherpossibleexplanationisthateachtrialendedatthetimetheparticipantgavetheirresponse.Therefore,theinter-trialintervalISIforanytwotrialsdependedonthereactiontimeto therstofthetwostimuli.WhiletheinuenceoftheISIonP300amplitudehasbeenstudiedbetweenblocksofoddballstimuliPolich,1990,toourknowledgetherehavenotbeenanystudies onvariableISI's within blocks. AfurtherpossibilityisthattheperceptualsalienceofourstimulicausedaP300tobeelicitedby everystimulus,includingstandardtrials.Thus,boththeXandtheOoccupiedalargepartof thescreenandwerethereforepossiblylargerthaninmostpreviousstudiesontheoddballparadigm. Perceptualsaliencedoesseemtoplayaroleinauditoryoddballparadigms,suchthattheP300is largestfortonesthataretask-relevant,loudandinfrequentK.C.Squiresetal.,1977. Finally,errorratesandreactiontimestotargetsweresignicantlylargerthanforstandards,suggestingthattherewasaresponsebiastowardresponsestothestandardstimuli.Thisresponsebias maybeanotherpossibleinuenceonourunusualndings. Therefore,furtheranalysiswillneedtofocusonwhythestandardsinouroddballparadigm appeartoelicitaP300thatisofshorterlatencythantheP300fortargets.Itispossiblethatonlya subsetofthestandardselicitedaP300,drivingtheappearanceofaP300intheoverallERPs.This possibilityissubjecttofurtherinvestigation. 75

PAGE 83

Chapter5 ConclusionsandFutureDirections Overall,ourstudysupportstheideathatLFwordsarerecalledequallywellorbetterthanHF wordswhentheyoccurinmixedlistsbecausetheyaresubjectivelyperceivedasdistinctiveinthese typesoflists.Similarlytotheprocessessuggestedforphysicalisolates,thisdistinctivenessattribute istaggedtothememorytraceandcanbeusedasaneffectiveretrievalcue.ThisperceiveddistinctivenessinmixedlistsisspecictoLFwords,anddoesnotoccurforHFwordsisolatedinlistsof LFwords.Conclusively,similarneuro-cognitiveprocessesappeartosupporttheVonRestorffeffect aswellasthereversalofthewordfrequencyeffectonrecallinmixedlists. ItstandstoreasonwhyourbehavioralresultsdidnotrevealaVonRestorffeffectforLFisolates althoughthecorrelationsbetweenP300amplitudeandsubsequentrecallsupportedthenotionthat distinctivenessaidedrecallforLFisolates.Itislikelythataparadigmusingarecognitionmemory testwillrevealbettermemoryforLFisolatesthanHFstandards.WhetheranalogousERPsubsequentmemoryeffectsfortheP300willbeobservedfortheLFisolatesinarecognitionparadigmis subjecttofurtherresearch. Ourstudyalsosupportsaninvolvementofenhancedinter-itemelaborativeprocessesforLFwords inlistsofHFwordsascomparedtoLFwordsinclustersasafactorinthereversalofthewordfrequencyeffectonrecall.Thus,correlationsbetweenERPamplitudesandrecallwerenotonlyfound fortheP300,butalsoforthepositiveslowwave,anERPcomponentthathasbeenimplicatedin elaborativeprocesses.Itappears,therefore,thatacombinationof both item-specicdistinctivenessprocessesaswellasbetween-itemelaborationcausesthereversalofwordfrequencyeffectson recallinmixedlists. Futurestudiesareneededtoinvestigatewhetherourconclusionsaregeneralizabletomixedlists composedofdifferentproportionsofHFandLFwordsthanusedinthecurrentstudy.Specically, animportantquestioniswhetherthesamepatternofresultswouldbeobtainedinamixedlist with50%HFand50%LFwords,sincethisisthelistcompositioninwhichthereversalofthe 76

PAGE 84

wordfrequencyeffectduetomixedlistsisusuallystudiedbehaviorally.Anotheropenquestionis howtheparticipant'sprocessingtaskatencodingaffectstheresults.CorrelationsbetweenP300 amplitudeandrecallforphysicalisolateshavebeenreportedonlywhenrotememoryisused.Itis possible,therefore,thatthecorrelationofP300amplitudetorecallforLFisolatesdependsonthe memorizationstrategyinasimilarway. Whileourndingsprovidesomenewinformationontheelicitingconditionsoftheslowwave,the dataalsoraisedsomenewquestions.Thatis,whyistheslowwaveleftlateralizedinourstudy,but showedafrontalmaximuminpreviousstudies?Thequestionwhetherthecomponentinourstudy thesameasinpreviousstudieswillneedfurtherinvestigation.Anotherissuethatneedsclarication infurtherstudiesiswhyHFstandardsdidnotelicitlargerslowwaveamplitudesthanLFstandards althoughtheyarethoughttofacilitateinter-itemassociativeprocesses. Itisunclearwhetherthecorrelationsbetweenthefrontalpositivityandsubsequentrecallfor sizeisolatesinourstudyactuallyreecttheP3a.ToclarifywhetherP3a-relatedprocessesmay aidsuccessfulmemoryencoding,futurestudiescouldinvestigatethisquestioninathree-stimulus oddballparadigmfollowedbyanincidentalmemorytestforthenovelitems.Itwouldbeagreat contributiontothecurrentknowledgeaboutthefunctionalsignicanceoftheP3atodetermine whetheritiscorrelatedtosubsequentmemorysuccessinsuchadesign. Finally,furtheranalysisisnecessarytoclarifythereasonsforourunusualndingsintheoddball task.Thatis,didonlyasubsetofthefrequentstimulielicitaP300likeERPcomponent?Ifso, whatarethecharacteristicsofthissubsetoffrequentstimuli,andwhydidtheyelicitaP300?A single-trialanalysisidentifyingwhichonesofthefrequentstimulimayhaveelicitedaP300will helpclarifytheseissues. 77

PAGE 85

References Adelman,J.S.,Brown,G.D.A.,&Quesada,J.F..Contextualdiversity,notwordfrequency, determinesword-namingandlexicaldecisiontimes. PsychologicalScience 34 ,495-513. Allen,M.,Badecker,W.,&Osterhout,L..Morphologicalanalysisinsentenceprocessing: Anerpstudy. LanguageandCognitiveProcesses 18 ,405-430. Azizian,A.,&Polich,J..Evidenceforattentionalgradientintheserialpositionmemory curvefromevent-relatedpotentials. JournalofCognitiveNeuroscience 19 ,2071-2081. Balota,D.A.,Pilotti,M.,&Cortese,M.J..Subjectivefrequencyestimatesfor2,938 monosyllabicwords. MemoryandCognition 29 ,639-647. Barber,H.,Vergara,M.,&Carreiras,M..Syllable-frequencyeffectsinvisualwordrecognition:evidencefromerps. NeuroReport 15 ,545-548. Bosch,V.,Mecklinger,A.,&Friederici,A.D..Slowcorticalpotentialsduringretentionof object,spatialandverbalinformation. CognitiveBrainResearch 10 ,219-237. Chee,M.W.L.,Westphal,C.,Goh,J.,Graham,S.,&Song,A.W..Wordfrequencyand subsequentmemoryeffectsstudiedusingevent-relatedfmri. NeuroImage 20 ,1042-1051. Cofer,C.N.,&Shevitz,R..Word-associationsasafunctionofword-frequency. American JournalofPsychology 65 ,75-79. Connine,C.M.,Mullennix,J.,Shernoff,E.,&Yelen,J..Wordfamiliarityandfrequency invisualandauditorywordrecognition. JournalofExperimentalPsychology:Learning, MemoryandCognition 16 ,1084-1096. Courchesne,E.,Hillyard,S.A.,&Galambos,R..Stimulusnovelty,taskrelevanceandthe visualevokedpotentialinman. ElectroencephalographyandclinicalNeurophysiology 39 131-143. 78

PAGE 86

Criss,A.H.,&Malmberg,K.J..Evidenceinfavoroftheearly-phaseelevated-attention hypothesis:Theeffectsofletterfrequencyandobjectfrequency. JournalofMemoryand Language 59 ,331-345. Davidson,D..Distinctivenessandmemory.InR.R.Hunt&J.B.WorthenEds.,p.157179.NewYork:OxfordUniversityPress. Deldin,P.,Deveney,C.,Kim,A.,Casas,B.,&Best,J..Aslowwaveinvestigationofworking memorybiasesinmooddisorders. JournalofAbnormalPsychology 110 ,267-281. Delorme,A.,&Makeig,S..Eeglab:anopensourcetoolboxforanalysisofsingle-trialeeg dynamics. JournalofNeuroscienceMethods 134 ,9-21. DeLosh,E.L.,&McDaniel,M.A..Theroleoforderinformationinfreerecall:Application totheword-frequencyeffect. JournalofExperimentalPsychology:Learning,Memoryand Cognition 22 ,1136-1146. Diana,R.A.,&Reder,L.M..Thelow-frequencyencodingdisadvantage:Wordfrequency affectsprocessingdemands. JournalofExperimentalPsychology:Learning,Memoryand Cognition 32 ,805-815. Diedrich,O.,Naumann,E.,Maier,S.,&Becker,G..Afrontalpositiveslowwaveintheerp associatedwithemotionalslides. JournalofPsychophysiology 11 ,71-84. Dien,J..Theerppcatoolkit:Anopensourceprogramforadvancedstatisticalanalysisof event-relatedpotentialdata. JournalofNeuroscienceMethods 187 ,138-145. Dien,J.,Spencer,K.M.,&Donchin,E..Localizationoftheevent-relatedpotentialnovelty responseasdenedbyprincipalcomponentanalysis. CognitiveBrainResearch 17 ,637650. Ding,L.,Goldstein,A.,Lian,J.,Donchin,E.,&He,B..Corticalimagingofp300component elicitedbyisolatewordsusingdifferentmemoryrehearsalstrategies. InternationalJournal ofBioelectromagnetism 4 ,269-270. Donchin,E..Surprise!...surprise? Psychophysiology 18 ,493-513. Donchin,E.,&Fabiani,M..Handbookofcognitivepsychophysiology.centralandautonomicnervoussystemapproaches.InJ.Jennings&M.ColesEds.,p.472-510.John WileyandSons. 79

PAGE 87

Duncan-Johnson,C.C.,&Donchin,E..Onquantifyingsurprise:Thevariationofeventrelatedpotentialswithsubjectiveprobability. Psychophysiology 14 ,456-467. Fabiani,M..Distinctivenessandmemory.InI.R.R.Hunt&J.B.WorthenEds.,p.339360.NewYork:OxfordUniversityPress. Fabiani,M.,&Donchin,E..Encodingprocessesandmemoryorganization:Amodelofthe vonrestorffeffect. JournalofExperimentalPsychology:Learning,MemoryandCognition 21 ,224-240. Fabiani,M.,Gratton,G.,Chiarenza,G.A.,&Donchin,E..Apsychophysiologicalinvestigationofthevonrestorffparadigminchildren. JournalofPsychophysiology 4 ,15-24. Fabiani,M.,Karis,D.,&Donchin,E..P300andrecallinanincidentialmemoryparadigm. Psychophysiology 23 ,298-308. Fabiani,M.,Karis,D.,&Donchin,E..Effectsofmnemonicstrategymanipulationinavon restorffparadigm. ElectroencephalographyandclinicalNeurophysiology 75 ,22-35. Federmeier,K.D.,&Kutas,M..Arosebyanyothername:Long-termmemorystructure andsentenceprocessing. JournalofMemoryandLanguage 41 ,469-495. Fernndez,G.,Klaver,P.,Fell,J.,Grunwald,T.,&Elger,C.E..Humandeclarativememory formation:Segregatingrhinalandhippocampalcontributions. Hippocampus 12 -519. Fernandez,G.,Weyerts,H.,Tendolkar,I.,Smid,H.G.O.M.,Scholz,M.,&Heinze,H.. Event-relatedpotentialsofverbalencodingintoepisodicmemory:Dissociationbetweenthe effectsofsubsequentmemoryperformanceanddistinctiveness. Psychophysiology 35 709-720. Forster,K.I.,&Chambers,S.M..Lexicalaccesssandnamingtimes. JournalofVerbal LearningandVerbalBehavior 12 ,627-635. Francis,W.,&Kucera,H.. Frequencyanalysisofenglishusage:Lexiconandgrammar Boston:HoughtonMifin. Gerhand,S.,&Barry,C..Age-of-acquisitionandfrequencyeffectsinspeededwordnaming. Cognition 73 ,B27-B36. Glanzer,M.,&Adams,J.K..Themirroreffectinrecognitionmemory. Memoryand Cognition 13 ,8-20. 80

PAGE 88

Glanzer,M.,&Adams,J.K..Themirroreffectinrecognitionmemory:Dataandtheory. JournalofExperimentalPsychology:Learning,MemoryandCognition 16 ,5-16. Glanzer,M.,&Bowles,N..Analysisoftheword-frequencyeffectinrecognitionmemory. JournalofExperimentalPsychology:HumanLearningandMemory 2 ,21-31. Guo,C.,Zhu,Y.,Ding,J.,Fan,S.,&Paller,K.A..Anelectrophysiologicalinvestigation ofmemoryencoding,depthofprocessing,andwordfrequencyinhumans. Neuroscience Letters 356 ,79-82. Hulme,C.,Roodenrys,S.,Schweickert,R.,Brown,G.,Martin,S.,&Stuart,G..Wordfrequencyeffectsonshort-termmemorytasks:Evidenceforaredintegrationprocessinimmediateserialrecall. JournalofExperimentalPsychology:Learning,Memory,andCognition 23 ,1217-1232. Hulme,C.,Stuart,G.,Brown,G.D.A.,&Morin,C..High-andlow-frequencywords arerecalledequallywellinalternatinglists:Evidenceforassociativeeffectsinserialrecall. JournalofMemoryandLanguag 49 ,500-518. Humphreys,M.S..Relationalinformationandthecontexteffectinrecognitionmemory. MemoryandCognition 4 ,221-232. Hunt,R.R..Distinctivenessandmemory.InR.R.Hunt&J.B.WorthenEds.,p.3-25. NewYork:OxfordUniverityPress. Hunt,R.R.,&Einstein,G.O..Relationalanditem-specicinformationinmemory. Journal ofVerbalLearningandVerbalBehavior 20 ,497-514. Karis,D.,Fabiani,M.,&Donchin,E.."p300"andmemory:Individualdifferencesinthe vonrestorffeffect. CognitivePsychology 16 ,177-216. Kim,A.S.N.,Vallesi,A.,Picton,T.W.,&Tulving,E..Cognitiveassociationformation inepisodicmemory:Evidencefromevent-relatedpotentials. Neuropsychologia 47 ,31623173. Kutas,M.,&Hillyard,S.A..Readingsenslesssentences:Brainpotentialsreectsemantic incongruity. Science 207 ,203-205. Kutas,M.,McCarthy,G.,&Donchin,E..Augmentingmentalchronometry:thep300asa measureofstimulusevaluationtime. Science 197 ,792-795. 81

PAGE 89

Kutas,M.,VanPetten,C.,&Kluender,R..Handbookofpsycholinguistics.InM.Gernsbacher&M.TraxlerEds.,nded.,pp.659.Elsevier. Landauer,T.K.,&Streeter,L.A..Structuraldifferencesbetweencommonandrarewords: Failureofequivalenceassumptionsfortheoriesofwordrecognition. JournalofVerbalLearningandVerbalBehavior 12 ,119-131. Lian,J.,Goldstein,A.,Donchin,E.,&He,B..Corticalpotentialimagingofepisodic memoryencoding. BrainTopography 15 ,29-36. Luck,S.J.. Anintroductiontotheevent-relatedpotentialtechnique .Cambridge,MA:MIT press. Malmberg,K.J.,&Murnane,K..Listcompositionandtheword-frequencyeffectforrecognitionmemory. JournalofExperimentalPsychology:Learning,Memory,andCognition 28 ,616-630. Malmberg,K.J.,&Nelson,T.O..Theword-frequencyeffectforrecognitionmemoryand theelevated-attentionhypothesis. MemoryandCognition 31 ,35-43. Mandler,G.,Goodman,G.O.,&Wilkes-Gibbs,D.L..Theword-frequencyparadoxin recognition. MemoryandCognition 10 ,33-42. Mangels,J.A.,Picton,T.W.,&Craik,F.I..Attentionandsuccessfulepisodicencoding: anevent-relatedpotentialstudy. CognitiveBrainResearch 11 ,77-95. Mason,J.M..Therolesofortographic,phonological,andwordfrequencyvariableson word-nonworddecisions. AmericanEducationalResearchJournal 13 ,199-206. McDaniel,M.A.,&Geraci,L..Distictivenessandmemory.InR.R.Hunt&J.B.Worthen Eds.,p.65-88.NewYork:OxfordUniversityPress. McLaughlin,J.P..Recallandrecognitionmeasuresofthevonrestorffeffectinserial learning. JournalofExperimentalPsychology 78 ,99-102. Mecklinger,A.inpress.Thecontroloflong-termmemory:Brainsystemsandcognitiveprocesses. NeuroscienceandBiobehavioralReviews Merritt,P.S.,DeLosh,E.L.,&McDaniel,M.A..Effectsofwordfrequencyonindividualitemandserialorderretention:Testsoftheorder-encodingview. MemoryandCognition 34 ,1615-1627. 82

PAGE 90

Mnte,T.F.,Schiltz,K.,&Kutas,M..Whentemporaltermsbelieconceptualorder. Nature 395 ,71-73. Nelson,D.L.,&McEvoy,C.L..Whatisthisthingcalledfrequency? Memoryand Cognition 28 ,509-522. Otten,L.J.,&Donchin,E..Relationshipbetweenp300amplitudeandsubsequentrecall fordistinctiveevents:Dependenceontypeofdistinctivenessattribute. Psychophysiology 37 644-661. Otten,L.J.,Sveen,J.,&Quayle,A.H..Distinctpatternsofneuralactivityduringmemory formationofnonwordsversuswords. JournalofCognitiveNeuroscience 19 ,1776-1789. Paller,K.A.,Kutas,M.,&Mayes,A.R..Neuralcorrelatesofencodinginanincidental learningparadigm. ElectroencephalographyandclinicalNeurophysiology 67 ,360-371. Polich,J..Probabilityandinter-stimulusintervaleffectsonthep300fromauditorystimuli. InternationalJournalofPsychophysiology 10 ,163-170. Polich,J.,&Donchin,E..P300andthewordfrequencyeffect. Electroencephalography andclinicalNeurophysiology 70 ,33-45. Polich,J.,Howard,L.,&Starr,A..P300latencycorrelateswithdigitspan. Psychophysiology 20 ,665-669. Ruchkin,D.S.,Johnson,R.,Canoune,H.,&Ritter,W..Event-relatedpotentialsduring arithmeticandmentalrotation. ElectroencephalographyandclinicalNeurophysiology 79 473-487. Rugg,M.D..Event-relatedbrainpotentialsdissociaterepetitioneffectsofhigh-andlowfrequencywords. MemoryandCognition 18 ,367-379. Rugg,M.D.,Cox,C.J.C.,Doyle,M.C.,&Wells,T..Event-relatedpotentialsandthe recollectionoflowandhighfrequencywords. Neuropsychologia 33 ,471-484. Rugg,M.D.,&Doyle,M.C..Event-relatedpotentialsandrecognitionmemoryforlowandhigh-frequencywords. JournalofCognitiveNeuroscience 4 ,69-79. Saint-Aubin,J.,&LeBlanc,J..Wordfrequencyeffectsinimmediateserialrecallofpure andmixedlists:Testsoftheassociativelinkhypothesis. CanadianJournalofExperimental Psychology 59 ,219-227. 83

PAGE 91

Schmidt,S.R..Canwehaveadistinctivetheoryofmemory? MemoryandCognition 19 523-542. Squires,K.C.,Donchin,E.,Herning,R.I.,&McCarthy,G..Ontheinuenceoftaskrelevanceandstimulusprobabilityonevent-relatedpotentialcomponents. ElectroencephalographyandclinicalNeurophysiology 42 ,1-14. Squires,K.C.,Wickens,C.,Squires,N.K.,&Donchin,E..Theeffectofstimulussequence onthewaveformofthecorticalevent-relatedpotential. Science 193 ,1142-1146. Squires,N.K.,Squires,K.C.,&Hillyard,S.A..Twovarietiesoflong-latencypositive wavesevokedbyunpredictableauditorystimuliinman. Electroencephalographyandclinical Neurophysiology 38 ,387-401. Steywers,M.,&Malmberg,K..Theeffectofnormativecontextvariabilityonrecognition memory. JournalofExperimentalPsychology:Learning,Memory,andCognition 29 760-766. Sutton,S.,Braren,M.,Zubin,J.,&John,E.R..Evoked-potentialcorrelatesofstimulus uncertainty. Science 150 ,1187-1188. Toglia,M.P.,&Battig,W.F.. Handbookofsemanticwordnorms. Hillsdale,NJ:Erlbaum. Tse,C.,&Altarriba,J..Testingtheassociative-linkhypothesisinimmediateserialrecall: Evidencefromwordfrequencyandwordimageabilityeffects. Memory 15 ,675-690. VanPetten,C.,&Kutas,M..Interactionsbetweensentencecontextandwordfrequencyin event-relatedbrainpotentials. MemoryandCognition 18 ,380-393. VonRestorff,H..berdiewirkungvonbereichsbildungenimspurenfeld. Psychologische Forschung 18 ,299-342. Voss,J.L.,&Paller,K.A..Rememberingandknowing:Electrophysiologicaldistinctions atencodingbutnotretrieval. NeuroImage 46 ,280-289. Watkins,M.J.,LeCompte,D.C.,&Kim,K..Roleofstudystrategyinrecallofmixedlists ofcommonandrarewords. JournalofExperimentalPsychology 26 ,239-245. Wiswede,D.,Rsseler,J.,&Mnte,T.F..Serialpositioneffectsinfreememoryrecallan erpstudy. BiologicalPsychology 75 ,185-193. Worthen,J.B..Distinctivenessandmemory.InR.R.Hunt&J.B.WorthenEds.,p.133156.NewYork:OxfordUniversityPress. 84