An associative analysis of object memory

Full text

(1)

ContentslistsavailableatScienceDirect

Behavioural

Brain

Research

j o ur na l h o me p a g e :w w w . e l s e v i e r . c o m / l o c a t e / b b r

Research

report

An

associative

analysis

of

object

memory

Jasper

Robinson

,

Charlotte

Bonardi

SchoolofPsychology,TheUniversityofNottingham,NG72RD,England,UK

h

i

g

h

l

i

g

h

t

s

•SynopsisgivenofmodelofassociativememorywhoseoriginisasatheoryofPavlovianconditioning.

•Objectrecognitiontasksareexplicableintermsoftheassociativememorymodel.

•Experimentalpredictionsfromtheassociativemodelarefoundtobesupported.

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received30June2014

Receivedinrevisedform27October2014 Accepted29October2014

Availableonline13November2014

Keywords:

Associativelearning Discrimination Objectrecognition Pavlovianconditioning Priming

Recognitionmemory

a

b

s

t

r

a

c

t

Differentaspectsofrecognitionmemoryinrodentsarecommonlyassessedusingvariantsofthe sponta-neousobjectrecognitionprocedureinwhichanimalsexploreobjectsthatdifferintermsoftheirnovelty, recency,orwheretheyhavepreviouslybeenpresented.Thepresentarticledescribesthreestandard variantsofthisprocedure,andoutlinesatheoryofassociativelearning,SOP[1]whichcanofferan expla-nationofperformanceonallthreetypesoftask.Theimplicationsofthisfortheoreticalinterpretations ofrecognitionmemoryandtheproceduresusedtoexploreitarediscussed.

©2014TheAuthors.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/3.0/).

1. Introduction

Wemaytakerecognitionmemoryasbeingthediscrimination betweennovelandfamiliarevents[2–4].Ithasbeenpopularto examinerecognitionmemoryinrodentsbycapitalisingontheir tendencytoapproachandexplorenovelorunexpectedobjectsin preferencetofamiliarorexpectedobjects[5,6].Systematic varia-tionsinthisprocedurehaveencouragedparticularaccountsofthe psychologicalprocessesunderlyingobjectrecognition:Ennaceur &Delacour[7](seealso[8,9])parsimoniouslydescribememoryas thehabituationofanapproachresponsetowardsanobjectwith increasedexposure;AggletonandBrown[10]maintainthatitis

Abbreviations:A1,primarystateofstimuluselementactivityinSOP;A2,primed orsecondarystateofactivityinSOP;CS,conditionedstimulusorsignal;US, uncon-ditionedstimulusoroutcome.

Correspondingauthor.Tel.:+441159515322;fax:+441159515324. E-mailaddress:jasper.robinson@nottingham.ac.uk(J.Robinson).

URL:http://www.jasperrobinson.wordpress.com/(J.Robinson).

basedonasenseoffamiliaritytowardsanobjectthatisbuiltup throughexposuretoit(seealso[2,11]);Cowell,Bussey,and Sak-sida[12]describeamulti-layerconnectionistnetworkthatallows recognitiontooccurasin thenatural courseofthebuildingof stimulusrepresentationsandseveralauthorshavesuggestedthat rodents’recognitionmemoryrepresentsarelativelysophisticated formofmnemonicprocess[13,14].

Inthisreportwedescribeanalternativepsychologicalaccount ofrecognitionmemory,whichisbasedonBrandon,Vogel,& Wag-ner’smodel[15],SOP(StandardOperatingProcedures;[16–18]). SOPisprimarilyamodelofassociativelearning,butunlikesome others[19]incorporatesaconceptualisationofmemory.Weargue herethatthisfeatureofSOPallowsittoaccommodatemuchofwhat isknownaboutrodents’objectrecognitionandshowthatithas generatedpredictionsthathavebeenexperimentallysupported. SOPhasproducedsuccessfulaccountsofmanyphenomena(e.g., habituation[3,20,21],associativelearning,discriminationlearning) andifitsaccountofobject recognitionwereaccepted,itwould strengthen its positionasa general-purposemodel of adaptive behaviour.

http://dx.doi.org/10.1016/j.bbr.2014.10.046

(2)

Fig.1. Objectrecognitionexperimentaldesigns.Experimentaldesignsforthree, keytypesofobjectrecognitionprocedure:spontaneousobjectrecognition,relative recencyandobject-in-context.PandQrefertodistinctlydifferentobjects;xand yrefertodistinctlydifferentcontexts.Theinequalitysignsindicatetheobjectthat elicitsthemostapproachbehaviour.

2. Objectrecognitioninnon-humananimals

2.1. Generalproceduralfeaturesandexperimentalconsiderations

Spontaneous object recognition [7,22–25], relative recency

[13,26–31]andobject-in-context[32–38]aretermsgiventothree ofthemostpopularproceduresforlookingatobjectmemoryin rodents(seeFig.1).Wedescribeeachinturn,beforediscussingan associativemodelofmemory,SOP,[15]thatcanprovidea satis-factoryexplanationofperformanceonthesethreetasks,andalso generatenovel,testablepredictions.

Allthreeproceduresinvolvetheplacementofarodentinoneor moreexperimentalcontextsalongwithoneormoreobjects;unlike matching-andnon-matching-to-sampleproceduresforstudying memory[39–41],foodreinforcementisnotusedinthesetasks. Experimental contexts vary in complexity, from being a single undecoratedwoodenboxtopairsofcontextsdifferentiatedintheir floors and/orwalls. Theexperimentalobjects areoften domes-ticitems(bottles,vases,etc.)thatareselectedtobedifficultfor rodentstoknock over, chewor situpon,behaviours thatcould easilycontaminatethemeasuredbehaviour.Thethreetasksdiffer principallyinthepreexposurethatrodentsreceive(whichisfully describedbelow).Forexample,inspontaneousobjectrecognition oneobject,P,ispreexposed.Inallthreetasks,testperformance involvesthesimultaneouspresentationofapairofdifferentobjects (PandQ),whichhavebeentreateddifferentlyearlierinthe pro-cedure(butsee[42]).Theexperimenterwillmeasuredifferences intheamounts ofbehaviourdirected towardsobjectsPand Q. Thismayeitherbespecificexploratoryresponses[32,33]orentry intonotionalzonesthatsurroundeachobject[31,43,44].In spon-taneousobjectrecognition,approachtoPisdepressedrelativeto Qontest[26,33]anddecreasesoverthecourseofpre-exposure

[20,45].

InterpretationofadifferenceinbehaviourtowardsobjectsPand Qrequiresappropriateexperimentalmanagement.Forexample,it isessentialtocounterbalancetheobjectsservingasPandQ(e.g., halfoftherodentshaveabottleasPandavaseasQ;theremaining rodentsreceivethereversearrangement)aswellastheirpositions withinthecontext(e.g.,halfoftherodentsreceiveobjectPonthe leftofobjectQ;theremainingrodentsreceivethereverse arrange-ment).Itisalsonecessarytoensurethatthetwoobjectsdonot differintheirmarking(e.g.,byodourcues),whichmaybeachieved eitherbycleaningorreplacingthemafteruse.

2.2. Threetypesofobjectrecognitiontask

In a spontaneous object recognition experiment [7,22–25], rodentsreceiveapre-exposuretrialwithobjectPbeforetesting withPandanovelobject,Q.QisapproachedinpreferencetoPon

testing.Typicallytwoversionsofasingleobject(here,P)are pre-exposedsothatthearrangementofobjectsinthecontextduring preexposurematchesthatduringtesting.Justasindelayed non-matchingtosample [46–49],in spontaneousobject recognition theintervalbetweenpre-exposureandtestingcanbeextendedto reducetestdiscrimination[50–52].

Thistemporal dynamic ofthe intervalbetween preexposure andtestisexplicitlymanipulatedintherelativerecencyprocedure

[13,26–31].Heretwopre-exposuretrialsprecedetesting:thefirst iswithP; thesecondwithQ.Ontest,rodentsapproachPmore thanQ,demonstratingbettermemoryofthemorerecently pre-sentedQ.Theinterval betweenthetwopre-exposuretrialsand theintervalbetweenthesecondpre-exposuretrialandtestwill emergelaterasimportantconsiderationsintheevaluationofthe associativeanalysisofobjectmemory.

Liketherelativerecencyprocedure,theobject-in-context pro-cedureinvolvestheseparatepre-exposureofbothobjectsPandQ beforethetest.Thedifferenceisthatpre-exposuretoeachobject occursinadistinctlydifferentcontext:ObjectPincontextxand objectQincontexty.GreaterapproachtowardsQthanPisfound whentestingoccursincontextx[32–38];and,ofcourse,testing maybegivenincontextyinwhichapproach isbiasedtowards P.Theparticularsequenceofpre-exposuretrials(xPbeforeyQor thereverse)mayconstitutearelativerecencymanipulationand itsimplicationsfor ourevaluation ofSOP areconsideredbelow (Section3.4.3).

2.3. Brainareasassociatedwiththesethreetypesofobject recognition

Althoughourreportis intendedtoconsidera psychological-levelofanalysisofrecognitionmemoryitisworthwhiletoconsider brain areas that are associated with the three forms of object recognition.Evidenceoftheimportanceoftheperirhinalcortex inspontaneous objectrecognitionis excellentin thatthere are manyexamplesofitsinvolvement(primarilyfromlesionstudies inrats)fromavarietyoflabs(e.g.,[22,23,28,37,45,49,53,54])and, asfarasweareaware,therearenoclaimsofnullfindings.There isalsosomeevidenceofinvolvementofprefrontalcorticalregions

[28]butmixedfindingsofinvolvementofthehippocampus(e.g., withdeficitsreportedinsomelabs,e.g.,[23,50],butnotinothers, e.g.,[35,38,55]).Fewerbrainregionshavebeenexaminedin rel-ativerecencyproceduresbutthereisabundantevidenceforthe involvementofspecificregionsoftheprefrontalcortexinrats(e.g.,

[27,28,56,13,57])and,againoftheperirhinalcortexandits connec-tionswiththeprefrontalcortex(e.g.,[28,13]).Object-in-context learninghasbeendemonstratedtobeaffectedbymanipulations ofthehippocampus(e.g.,[35,38,55],seealso[58]forevidenceof hippocampalinvolvementinanaudio-visualanalogueof object-in-contextlearning)andtheprefrontalcortex(e.g.,[28,36]).

3. TheSOPmodelofassociativememory

3.1. Stimulusrepresentation

(3)

Fig.2. Activitytransitionsofstimuluselements.

ThisfigureisastillfromamoviethatisembeddedinthePDFversionofthis manuscript(seeAppendixA).Themovierepresentsthesequenceofactivity tran-sitionstatesinBrandonetal.’sSOPmodel.Threestimuluselementsareinitiallyin aninactivestate.Clickingthe“Activate”buttonisanalogoustothepresentationof astimulustoanorganism,suchasthepresentationofatonetoarat.Theratwill typicallyshowastrongorientingresponsetothesourceofthetone(representedas, “?!”),correspondingtothestimuluselements’promotiontotheirA1statesof activ-ity.Theelementsarenext“primed”(i.e.,theyenteraninferiorA2stateofactivity) wheretheyprovokeweakerorientingresponses.Finally,theelementsre-entertheir inactivestateswheretheydonotprovokestimulus-specificbehaviour.For compu-tationalreasons,Brandonetal.conceiveofstimulusrepresentationsascomprising muchlargerpopulationsofelementsthanthis;nonethelessthesequenceinwhich thestimuluselementspassthroughthevariousactivationstatesisidentical,and ele-mentscannotpassdirectlyfromA2backtoA1.Thuspresentationofthetone(i.e., clicking“Activate”)whenitselementsareintheirA2statedoesnotpromotethem intotheirA1state:EntryintoA1statemustcomedirectlyfromtheinactivestate.

tonerepresentationwouldbecomeassociatedwiththe represen-tationalelementsthatcodeforthefood(withtheirowndistinctive codingpropertiesrelatingtoitssmell,texture,etc.)Presentation ofthetonecouldproducefood-relatedrespondingbyactivationof thefoodstimuluselementsviatheseinter-elementassociations. Moreover,presentationofaphysicallysimilar,butdiscriminably different,tone(e.g.,onehavinga1.8Hzpitchwithanamplitude, again,of 65dB SPL) couldalso produceresponding, albeit at a reducedlevel,totheextentthatitsharedoneormorestimulus elementswiththetoneusedintraining[59–61].

3.2. Activitystates,theirtransitionsandbehaviour

Hebb[62]describedlearningprinciplesinwhichcellassemblies (whicharenotionallyidenticaltostimuluselementsinSOP)are ineitheractiveorinactivestates.Coincidentperiodsofactivityin cellassemblieswouldpromotetheformationofexcitatory associa-tions.However,Hebb’ssystemfailstopredicthallmarkassociative learningphenomenasuchasblockingandrelativevalidity[63,64]. LikeHebb’smodel,SOPsupposesthatconcurrentlyactivestimulus elementswillsupportassociationformation(thiswillbeoutlined belowinSection3.3)butsupposestheexistenceoftwoqualitatively differentformsofactivity.

Arepresentationofabrief,2.0-kHztone’sstimuluselements andtheirtransitionthroughSOP’sactivitystatesissummarisedin theHTMLmovieinFig.2.Thetone’selementsbegininthe inac-tivestateandtheywillnotelicitanybehaviour.ClickingtheAction buttoncorrespondstothepresentationofthetoneandthiswill promotethetone’selementsintotheirprimaryformofactivity(A1 state)inwhichtheywillelicitbehaviour.Oninitialpresentationof atone,theratwillelicitastrongorientingresponse[9],forexample towardstheloudspeakerandthisisrepresentedbythelargehazard sign,markedwith“?!”.Withsustainedorrepeatedpresentation, thetone’selementswillfadeintotheirsecondarystateofactivity (A2state)wheretheywillelicitweakerorienting,whichis repre-sentedbyasmallerhazardsignmarkedwith“?”.Elementsintheir A2statesarereferredtoasprimed.AftertheirperiodofA2activity, orpriming,theelementsdecaytotheirinactivestatesandnolonger elicitorienting.Threepointsarenotablehere:1.Themomentary

probabilityofeachelement’stransitionfromonestatetothenext neednotbeuniform.Thisallowselementsofasinglestimulusto inhabitdifferentactivitystatesatthesametime;2.Astimulus ele-mentisassumedtobeindivisible–itcannotinhabitmorethanone activitystatesimultaneously;3.ElementsmustpassintotheirA1 statesviatheirinactivestates,aswesawinthemovie:Theycannot movedirectlyfromA2toA1.Thus,forexample,asecond presenta-tionofthetonewhiletheelementsareintheirA2statesdoesnot returnthemtotheirA1states.

Brandonetal.describeasecond,associativerouteofprimingin SOP.Followingsuccessfulassociativelearning,forexampleinan appetitiveconditioningprocedure,atonesignalwillgenerateA2 activityinthefoodoutcome.Thatis,thepresentationofthetone generatesA1activityinitscentralrepresentation,whichprimesthe food’srepresentationdirectlyintotheA2stateviatheexcitatory associationbetweenthem.Onceprimed,thefoodrepresentation provokesconditionedresponding.SOP’sassociativeprimingallows ittoaccommodatemanylong-standingphenomenainanimal dis-criminationlearning,suchasblocking[64,65]andrelativevalidity

[63]–inkeepingwithothermodels[19,60,66].Associativepriming alsoallowsSOPtocorrectlyanticipatethatassociativelearningwill reachanasymptote,afeaturelackinginHebb’s[62]model.

3.3. SOP’srulesforassociationformation

Having outlined SOP’s transition characteristics for stimulus elementswe arein thepositiontounderstanditsfourrulesfor associativelearning.Therulesarestraightforwardandrepresented inFig.3:Excitatoryassociationswillbeformedbetween stimu-luselementswhoseelementsaresimultaneouslyintheirA1states (firstrule);inhibitoryassociationswillbeformedfromA1-state

(4)

elementstoA2-stateelements(secondrule)butnoassociations willbereciprocated(thirdrule).Noassociativechangeswilloccur amongA2-stateelements(fourthrule).Futurepresentationsofa stimuluswillactivatetherepresentationofassociatedelements viaexcitatoryassociationsintotheirA2state;butinhibitory asso-ciationswillpreventthis,leavingtheelementsinactive.Wenext outlinecircumstancesinPavlovianconditionedexperimentswhere SOPassumestheseassociativechangestakeplace.

In standard Pavlovian conditioning in which, for example, rodentsreceiveabrieftonewhoseterminationiscoincidentalwith deliveryofafoodpellet(Tone→Food;e.g.,[67,68]),thepairingof thetoneandfoodwilltendtopromotetheirstimuluselementsinto theirA1statesandthiswillsupportexcitatoryconditioning.SOP predictsthatthebestconditionforsupportingassociativechange iswhenstimuluselementsarepairedsimultaneouslybutPavlovian conditioningprocedurestypicallyinvolvestheserialpairingofthe signal(orconditionedstimulus–CS)andoutcome(or uncondi-tionedstimulus–US).Thereasonsforthisaremerelypractical: Withserialsignal→outcome pairings,measurement of respon-dingtotheCSisfree ofcontaminationfromtheunconditioned responsestotheoutcome.However,speciallydesignedanalytical experimentshavefoundlearningaboutsimultaneouslypresented stimulitobesuperiortolearningaboutseriallypresentedstimuli

[69],thefindinganticipatedbySOP.

Aninhibitorysignalforanoutcome(aninhibitor)willnot pro-ducedistinctbehaviourwhentesteddirectlyinthewaythatan excitatorysignalwill.Instead,inhibitionisinferredfromtheresults ofapairofindirecttestsforinhibition[70].In“retardationtests”, theinhibitor servesas a standard Pavloviansignal forthe out-comefollowinginhibitiontraining.Theacquisitionof(excitatory) conditioned responding to the inhibitor is slow relative to an appropriatecontrolcondition.The“summationtest”involvesthe superimpositionoftheinhibitoronapreviouslyestablished exci-torforthesameoutcome.Theinhibitorreducestheconditioned responsethatisotherwiseelicitedbytheexcitor.Retardationand summationtestscomplementeachotherandmanyarguethatboth mustbepassedtoclaimthatastimulusisatrueinhibitor[71].

“Backwardconditioning”and“featurenegativetraining” proce-dureshavebothbeenfoundtoproduceinhibitionbyretardation andsummationtesting.A1-signalactivitycanbepairedwith A2-outcomeactivityinPavlovianconditioningwhenthe“outcome” precedes,ratherthanfollows,the“signal”(i.e.,outcome→signal pairings,or “backward conditioning”). Whentested after train-ingthesignal mayproducenegligible performance [72], which SOPwouldassumetobetheresultofamixtureofexcitatoryand inhibitorysignal-outcomelearningcausedbytheA1-signal activ-ity’sco-occurrencewiththeoutcome’selementsastheydecayfrom A1toA2activitystates.Withalterationsinexperimental param-eters,itispossibleforoutcome→signalpairingstoproducefull inhibitorylearninginwhichthesignalisfoundontesttosuppress activityintheoutcomerepresentation[72].HereSOPwillassume thatthereisgreaterco-occurrenceoftheoutcome’sA2stateand thesignal’sA1statethantheoutcome’sA1stateandthesignal’sA1 state.AcommonermethodforarrangingthepairingofA1-signal andA2-outcomeactivityisfeature-negativediscrimination train-ing[70,71,73].Heresomethirdstimulus,e.g.,alight,isestablished asasignalforfoodandthetoneandlightarepresentedasa com-pound,whichisnotreinforcedbyfooddelivery(i.e.Light→Food, [Tone+Light]→nothing).

There are important parallels between SOP and its better-knownrelative,theRescorlaandWagnermodel[19].TheRescorla and Wagner model describes the notional changes in associa-tive strength between the representations of a pair of events on any particular trial, by the expression, V=˛ˇ(V). In particular,thechangeinassociativestrength(V)isproportional tothedifference in themaximum level of associative strength

(asymptote, ) and the sum total of each previous associative change(V)multipliedbytheproductofapairoflearning-rate parameters(˛andˇ)thatareassociatedwiththetwoeventsand where1≥˛>0and1≥ˇ>0.DuringPavlovianconditioning,each signalled outcomepresentationwill beless surprisingthanthe last;thisiscapturedbyincreasesinV,producingadeclineinthe term(−V),whichresultsinincreasinglysmallincrementsin V,yieldinganegativelyacceleratinglearningcurve.SOPshares this property,which is achieveddifferently through associative priming(Section3.3).Onthefirsttrialthesignalandoutcome’s conjointA1activitywillbegoodandthiswillincreaseexcitatory strengthbetweenthem.On thenexttrialthesignal’sexcitatory strengthwillprimetheoutcome.Theassociativeprimingwillbe weakonthesecondtrialbutitwillallowsomeoverlapofthe sig-nal’sA1elementsandtheoutcome’sA2elements(Fig.3).Outcome elementsthatarenotprimedwillbefreetoentertheirA1stateson thesecondtrialandthiswillproducesomeexcitation.Theratioof excitationtoinhibitionwilldecreaseasVincreasesoneachtrial andVwillreachasymptote()andatthispointchangesin exci-tationandinhibitionareequaloneachtrial.ThustheRescorlaand WagnermodelandSOPaccountforcompetitionforafixedquota ofassociativestrengthinthesamewayandthisprincipleextends tootherkeyassociativephenomenasuchasblocking[64,65],super conditioning[74]andrelativelyvalidity[63].SOPmaybeseenas a morecomplete modelthan Rescorlaand Wagner’sbecauseit operatesinrealtime[75,76]andanticipateslatentinhibitionand itscontextspecificity[77],someperceptuallearningphenomena

[78],andmodulationofUCSprocessing[76,79].

3.4. Decayprimingandobjectrecognition

Theconclusionsofthisandthefollowingsectionare summari-sedinFig.4.

Itmayalreadybeobviousthatthelearningprocessdescribed by SOP for Pavlovian conditioning (Sections 3.2 and 3.3) may applytootherstimuli,includingthoseusedinobjectrecognition experimentswithrodents.Althoughwemightconceiveofthe rep-resentationsofobjectsasbeingmorecomplexthanthoseofapure

(5)

tone,andcomposedofelementscorrespondingtomultiplesensory domains,thesameprocessescanbeapplied.

3.4.1. Spontaneousobjectrecognition

Therodent’sinitialencounterwithobjectPduringspontaneous objectrecognitionwillgenerateA1activityintherepresentational elementsthatprovokestrongapproachbehaviour.Aftersometime this behaviour willweaken, correspondingto theobject-P ele-ments’entryintotheirA2state.Testpresentationduringthistime wouldresultinthestandardbiastowardsexplorationofthenovel object,Qbecauseitsrepresentationalelementscanpassfreelyfrom theirinactivestatestotheirA1stateswhereastheelementsofPthat arestillintheA2-statecannot.

3.4.2. Relativerecency

Wenotedabove(Section2.2)thatspontaneousobject recogni-tionperformanceworsenswhenaretentionintervalisinterpolated betweenthepre-exposureandtesting[50–52].OneofSOP’s expla-nationsof thisisthat longerretentionintervalsallowagreater proportionofP’selementstoreturnfromtheirA2totheir inac-tivestates.Thus,ontest,bothP’sandQ’sinactiveelementscan passdirectlytotheirA1stateswheretheycanelicitsimilar,strong levelsofapproach.Demonstrationsofrelativerecency[13,26–31]

receiveanidenticalexplanationfromSOP:Thepre-exposureofP andthenofQwillelicitA1activityintheirelementsthatwill,over time,decaytoA2andtheninactivestates.BecausePispre-exposed beforeQitispossibletoarrangefortestingwithPandQatatime whenP’selementstendtobeinactivebutQ’selementsremainin theirA2states.HereP’selementsarefreetoentertheirA1states wheretheyelicitstrongapproach;butmanyofQ’selementsmay remainintheirA2statesandonlyweakapproachiselicited.

3.4.3. Object-in-context

ThemostnaturalSOPaccountofobject-in-contextlearningis describedfullybelow(Section3.5.1)andreliesontheoperation ofassociationsbetweeneachcontextandtheobjectthathasbeen pre-exposedinit(i.e.,x→Pandy→Q).Testingincontextxwill primeobjectP’selementsbutnotobjectQ’s.Thusthepresenceof PandQontestinxwillallowQ’selementstoentertheirA1states fromtheirinactivestates,wheretheycanelicitrelativelystrong approach;butP’sprimedelementswillremainintheirA2state, unabletoelicitsuchresponding.

Althoughsuchassociative-primingprocessesseemplausible,it isimportantalsotoconsiderpossibledecay-primingprocessesin object-in-contextlearning.Thetemporalasymmetriesdescribed forrelativerecencyarealsopossibleinobject-in-contextlearning whenasingletrialofeachtypeisused:Heretherearetwo possi-blesequencesfortheobjectandcontextpre-exposures(i.e.,either xP,yQoryQ,xP).Ofcourse,thisispreciselythearrangementof objectpre-exposureinarelativerecencyexperiment,andsoSOP anticipatesthatthefirst-pre-exposedobjectwillprovokestrongest approach,basedondecaypriming.Theassociative-priming mech-anismpredictsgreaterapproachofQthanofPontestincontext x,inwhichitisunexpected.Thus,decay-andassociative-priming willopposeoneanother(i.e.,weakeningdiscrimination)butpullin thesamedirection(i.e.,enhancingdiscrimination)withtheyQ,xP pre-exposuresequence.Wehaverecentlyfoundevidenceofsuch interactionsbetweenassociativeanddecaypriming(Tam,Bonardi, &Robinson,reportunderreview).

Althoughdecayprimingappearstoinfluenceperformance in object-in-context learning, it cannot adequately account for it. Therearetwo reasonsfor this statement: First,in experiments inwhich onlya singlexPandyQtrialisgiven [33–38], object-in-contextlearning isdetectedwhen performanceoverthetwo possiblepre-exposuresequencesiscollapsed.Thatis,although dif-ferencesindecayprimingappeartooperate,theywillassistand

hinderdiscriminationperformanceinequalmeasureoverboth pre-exposuresequences.Thefactthattestdiscriminationissuccessful impliestheoperationofasecondprocessoverandabovedecay priming;second,itispossibletoarrangemultiplepre-exposure trialssothatforeachrat,onaverage,objectsPandQhavebeen presentedequallyrecentlybeforetesting.Thisisachievedby pre-sentingxPandyQpre-exposuresmultipletimes,e.g.,xP,yQ,yQ,xP forhalfofthesubjectsandyQ,xP,xP,yQfortheremainderandalso resultsinobject-in-contextlearning[32].

Thus,itappearsthatdecayprimingisrequiredtoexplain rela-tiverecencybutcannotaccommodateobject-in-contextlearning. Decayprimingcouldalsoexplainspontaneousobjectrecognition.

3.5. Associativeprimingandobjectrecognition

3.5.1. Spontaneousobjectrecognition

Brandonet al. anticipate a second, association-based source ofspontaneousobjectrecognition:Contextcuesenterinto excit-atoryassociationswithobjectP’srepresentationalelementsduring objectP’spre-exposure(see,e.g.,[18,21,43]).Thus,contextcues willgenerateA2activity(butnotA1activity)inobjectP’s repre-sentationalelementsontesting.Aswiththedecay-basedsource ofperformance,theassociativesourceisbasedontheelements forobjectAtendingtobeintheirA2states,whilethoseforobject QtendtobeintheirA1stateatthepointoftesting.Thusitseems thatbothassociativeanddecayprimingcouldsupportspontaneous objectrecognition.

3.5.2. Relativerecency

Inthecaseofrelativerecency,anysuchcontext→Passociation willbematchedbyacontext→Qassociationandsotest discrim-inationcannotobviouslybeaccountedforbyassociativepriming. However,non-reinforcementofasignalmayextinguishexcitatory learning[80,81]andfollowsSOP’srules(i.e.,thenon-reinforced signalwillelicitA2activityintheoutcome’srepresentationthat willgenerateinhibitoryassociativelearning).Thusanycontext→P associationcouldundergoextinctionduringthesubsequentpairing ofthecontext withQ,duringwhichP isabsent.Thus test per-formancecouldbebasedentirelyonassociativeprimingbecause, duringtesting,thecontextwillbebetterassociatedwithQthan withP.ThiswillproducebetterprimingofQthanofP,leadingtothe observeddifference:greaterapproachofPthanQ.However,studies showthatmultiplenon-reinforcedtrialsarenecessarytoachieve appreciableextinction[80,81]–unlikethesingletrialnecessaryfor relativerecency.Thus,althoughanassociativeprimingaccountof relativerecencyislogicallypossibleitisempiricallyimprobable, andthisindicatesdecayprimingasthesolesourceofperformance.

3.5.3. Object-in-context

Themostdirectevidencefortheoperationofcontext→object associativeprimingcomesfromobject-in-contexttasks[32–38]. Aswenotedabove(Section3.4.3),accountsofobject-in-context learningbasedondecayprimingareimplausibleandso,by elim-ination,theassociativeprimingaccountseemsthemostrealistic componentoftheSOPmodelavailabletoexplainperformance.

3.6. Conclusionsabouttherolesofdecayandassociativepriming inobjectrecognition

(6)

parameters(timing,stimulusduration,etc.),wecouldmanipulate themixofdecayandassociativepriminginspontaneousobject recognition.Buttheunavoidableconclusionisthat,accordingto SOP’sanalysis,spontaneousobjectrecognitionisanimprecisetool andisunsuitableforunderstandingseparablepsychological pro-cesses.Themixedfindingsontheeffectsofhippocampuslesions onrats’spontaneousobjectrecognition(cf.[23,40,50]and[22,39]) maybeexplicableifitisassumed,say,thatassociativepriming isdependentonthehippocampusandismoreprominentinthe proceduresofthosefindinglesioneffectsthanthosewhohavenot (cf.[18]).Thislogicappliestoothermanipulations,ofcourse:For example,extensionoftheretentionintervalbetweenpre-exposure andtestwillhaveadevastatingeffectonspontaneousobject recog-nitionthatislargelydependentondecaypriming,butonebased largelyonassociativeprimingwillbeunaffected.Butthediscussion hereimpliesthatrelativerecencyandobject-in-contextlearning arerelativelyselectivetoolsforexaminingdecayandassociative primingrespectively.

Spontaneousobjectrecognitionhasbeendescribedasaformof habituation(7],seealso,[8,9])–thatis,adeclineinthebehaviour thatiselicitedbyastimuluswhenitispresentedrepeatedlyorover alongperiodoftime.Inthecaseofspontaneousobjectrecognition thestimulusisobject,Pandthemeasuredresponseisapproach orexploratorybehaviour.Thedeclinein thisbehaviourmaybe assumedtooccurduringpreexposurebutismeasuredinthetest relativetoobjectQ.Thestimulusspecificityshownheremeetsakey featureofhabituation.Severalauthorshavenotedthathabituation canbeunderstoodastheresultofadeclineintheefficacyofthe pathwaybetweenastimulusandresponse[82].Itisquitepossible thatsimple,stimulus-responsebasedadaptiveprocessmayoccur alongsidethoseanticipatedbySOP.Oneimplicationofthisisthat neuralmanipulationsmayaffectSOP-basedprocessesandspare stimulus-responsebasedprocesses(e.g.,[8,9,58]),orviceversa.

Sokolov([83],seealso[84,85])described amodel of habitu-ationin which thepresentationof a newstimuluswill leadto thedevelopmentofarepresentationofit.Prolongedorrepeated presentationofthestimuluswillfosterincreasingdetail ofthat representationuntilitisentirelyaccurate.Theprocessof repre-sentationalmodificationisassumedtobetheresultofamismatch inmemory–thatisthestimulusiscomparedwithsetsof stimu-lusrepresentationsinmemory.Whencomparisonfindsamatch, norepresentationalmodificationis required;when comparison failstofindamatchanewrepresentationisassembled(oraclose matchisimproved).Inanhabituationprocedure,improvementsin stimulusrepresentationwillbeassociatedwithdeclinesin uncon-ditionedresponse(i.e.,habituation).LikeSokolov’smodel,theSOP model[15]allowsfortheassemblyofstimulusrepresentations– whentheirelementsareconcurrentlyintheirA1states–butno additionalprocessofcomparisontakesplace.Rather,changesin responsivenesstorepeatedlypresentedstimuliarebasedsolelyon theprimingprinciplesoutlinedabove.

4. Evidenceofassociativeprocessesinobjectrecognition

WehaveseenthatSOPcanbeappliedtostandard demonstra-tionsofobjectrecognition.Ofcourse,themerefactthatthismodel fitsthedatadoesnotdemandthatthemodelshouldbeaccepted. Wenextdescribesomeevidencefromanalyticalobjectrecognition experimentsthatmakeastrongercaseforthisposition.

4.1. Enhancedspontaneousobjectrecognitionwhenmultiple pre-exposuresarespaced

We saw that spontaneous object recognition test perfor-mancecouldbetheresultofobjectP’srepresentationalelements

Fig.5.Spacedandmassedpre-exposureillumination.SchematicofWhittand Robinson’s[44]multi-trialpre-exposureinspontaneousobjectrecognition.Rats werepresentedwithobjectPduringpre-exposure,whichwasonlyvisibleduring eight30-speriodsofilluminations–contextswereotherwisedarkened.Theperiods ofdarknessdifferedacrosstwoconditions,beingeither4-min(spacedcondition) or30-s(massedcondition).Subsequentdiscrimination(greaterapproachtoobject QthanobjectP)wassuperiorinthespacedconditionthaninthemassedcondition.

becomingprimedthroughamixofdecayandassociativepriming, whichmayvaryaccordingtothepreciseexperimental parame-ters(e.g.,thepre-exposure-testretentioninterval).Totheextent thatassociativeprimingisinvolvedintestperformance,itwillbe importanttomaximisethestrengthofthecontext→object associ-ationduringpre-exposure.Aswesawabove(Section3.3),learning aboutthecontext→objectrelationshipwilloccuronlywhenthe context’srepresentationalelementsareintheirA1states,andthe valenceofthislearning(i.e.,whetheritisexcitatoryorinhibitory) willdependontheactivitystateoftheobject’srepresentational elements:WhentheyareintheirA1statestheassociationwillbe strengthenedthroughadditionalexcitation,buttheassociationwill beweakenedbyinhibitionwhentheyareintheirA2states.

Suchweakeningofthecontext→objectassociationisexpected whentheobject,Phasbeenrecentlypresented,asaresultof decay-basedpriming.Severalauthorshavereportedthatwhenmultiple pre-exposurestoobjectParegiven,testperformanceisbetterwhen thesepre-exposures are relatively well-spaced(e.g., [20,44,86]) andwehaveunpublisheddemonstrationsofanalogouseffectsin humanparticipants’performanceonanauditoryrecognition mem-orytask.Thepre-exposurescheduleusedbyWhittandRobinsonis representedinFig.5.Toensurefullexperimentalcontrolover stim-ulusscheduling,objectsPandQwerepresentedinglasscylinders, renderingthementirelyvisualstimuli.Allratsreceivedeight,30-s periodsofilluminationduringasinglepre-exposuresession, dur-ingwhichtimetheyremainedinthecontext.ObjectPwouldonly havebeenvisibleduringtheseperiodsofillumination.Rats’test performancefollowing spacedpre-exposure,inwhich the inter-valbetweeneachilluminationperiodwas4min,wassuperiorto thatfollowingmassedpre-exposure,inwhichitwasonly30s.Our interpretationisthatanappreciableproportionoftheobjects’ rep-resentationalelementsreturnedtoinactivityduringthefour-min inter-illuminationperiodandwereavailabletoentertheirA1states onthesubsequentperiod,thussupportingtheformationofthe context→objectassociation.However,themassedpre-exposure trialsoccurredwhenP’selementsweremorelikelytobeprimed. Thiswillhaveproducedsomeinhibitoryx-Plearningthatwould acttooffsettheexcitatoryx-Plearninguponwhichperformance partlyrelies.

4.2. Decaypriminginrelativerecency

(7)

Fig.6. Temporalmanipulationsinrelativerecency.Schematicofthetemporal manipulationsinpre-exposure-trialandretention-intervalspacingintwo relative-recencyexperiments.Ratsinbothexperimentsreceivedsequentialpre-exposure trialswithobjectsPandQbeforetestingwiththetwoobjects.Tametal.used afive-minretentionintervalbetweenpre-exposuretoobjectQandtesting.The durationoftheintervalbetweenthetwopre-exposuretrialswaseither5minor 2h;discriminationwasbetterwiththelongerinterval.MitchellandLaiaconafixed thepre-exposureinter-trialinterval(at1h)andmanipulatedtheretentioninterval. Testdiscriminationwasgoodatretentionintervalsof24horless,butwasabsent at72-hintervals.Note:MitchellandLaiaconaincludedadditionalintervals,which arenotsummarisedhere.

duration of the inter-trial intervals. For example, with a suffi-cientlyshortretention interval (i.e.,thetime betweenthefinal pre-exposureandtesting),theintervalbetweenpre-exposureto PandQbecomescrucialinensuringthatP’selementsarelargely inactivebytest.Tam,Robinson,Jennings,andBonardi[87]have confirmedthispredictionofSOPinadesignthatissummarisedin

Fig.6.Theypresentedratswithpairsofobjects,PandQ,witha five-minintervalbetweenQandthetest.Inoneconditiontheinterval betweenPandQwasalso5min,andinanother2h.The2-hP-Q intervalproducedmarkedlystrongerdiscriminationthanthe five-minP-Qinterval.Tametal.demonstratedthatthistime-related reductioninperformance wasnot ageneraldeficit,byshowing avariantofobject-in-contextlearningtoremainunaffectedbya similarmanipulationoftheintervalbetweentrainingandtest.Itis alsonotablethattheanalysisofobject-in-contextlearning(Section

3.5.1)anticipatesthisinsensitivitybecausetheassociativepriming onwhichitrelieswillbeunaffectedbychangesinthepassageof time.

Arelatedpredictionisthatrelativerecencyshouldalsobe abol-ishedbyanextensionoftheretention interval(i.e.,theinterval betweenthesecondpre-exposuretrialandtest).Whensufficiently long,theretentionintervalwillallowtheelementsofbothPand Qtobecomeinactivesothat,ontest,A1activityinbothmaybe generatedandsimilarlevelsofapproachelicited.Mitchelland Laia-cona[29]reportjustsuchafinding.Ratsinthreegroupsreceived pre-exposuretoobjectPbeforeobjectQ,whichwasseparatedby a1-hinterval.TestapproachofQwasgreaterthanPwhen test-ingoccurredimmediatelyafterthepre-exposuretoP,but with theinterpolationofa72-hretentionintervaltestperformancewas attenuated.However,onefeatureofMitchellandLaiacona’sresults isproblematicforSOP:Relativerecencydiscriminationwasseen alsoinagroupwitha24-hretentioninterval.Inordertoexplain thispatternofresultstheanalysisofrelativerecencyoutlinedin Section3.4.2needstoassumethatP’selementshavedecayedto inactivityat25h,butthatat24hQ’sremainlargelyprimedintheir A2states.Whilethisisnotimpossible,itcouldbeseenasa sur-prisingcoincidence;giventheirpotentialtheoreticalsignificance, MitchellandLaiacona’sfindingswouldbenefitfromconfirmation andfurtheranalysis.

Fig.7.Schematicofprimedobject-in-context.Schematicofanobject-in-context experiment(e.g.,Whitt,Haselgrove,&Robinson)inwhichthethreestagesofthe experimentoccurfromlefttoright.Theprocedurebeganwithratsreceiving pre-exposuretoobjectPincontextxandwithpre-exposuretoobjectQincontexty, asinthestandardobject-in-contexttask.Nextratswereplacedincontextx;no objectwaspresentbutthistreatmentshouldbesufficienttoallowtheassociative primingofobjectP.Inthesubsequenttest,ratswerepresentedwithbothobjects, PandQ,butinathirdcontext,z,whichwasassociatedwithneitherobject.Ontest, ratsapproachedQinpreferencetoPdespitetestingnotoccurringinP’scontext,x. Performanceisconsistentwiththeprimingmechanisms.

4.3. Associativeprimingofobject-in-contextlearning

Thecurrentanalysisofobject-in-contextlearning(Section3.5.3; seeFig.3)reliesontheassociativeprimingofobjectPbythetest context,x.Thisprimingof Pusuallyoccurs duringtherodents’ placementincontextxattest.However,accordingtothe analy-sissuggestedhere,thisisnotnecessary:Theonlyrequirementis thatxisprimedatthetime thatxandyaretested.Inorderto examinethissuggestionWhittHaselgroveandRobinson[43]used theproceduresummarisedinFig.7.Exactlyasinastandard object-in-contextlearningtask,ratswereinitiallypresentedwithobject PincontextxandobjectQincontexty,toallowtheestablishment ofx→Pandy→Qassociations.Inasubsequentstageratswere placedincontextx,butthiscontainednoobjects.Thisstagewas intendedtoassociativelyactivateobjectP’srepresentation,andthis primingwasassumedtoremainappreciableovertheshorttime beforetesting.Unlikestandardobject-in-contextlearning,testing withobjectsPandQoccurredinathirdcontext(e.g.,z)inwhich noobjectshadbeenexperienced.Despitethischangeinprocedure, andconsistentwithSOP,ratsapproachwasbiasedtowardsobject QrelativetoobjectP.

5. Discussion

Wehaveattemptedtoapplytheassociativemodelofmemory, SOP [16–18] to three representative forms of object recogni-tion that are popular procedures in experimental psychology and behavioural neuroscience. In considering its dynamic pro-cesses(decayandassociativepriming)anditsrulesforassociation formation,SOP produced asatisfactory accountofobject mem-ory. Experimental evidence [20,43,44,88] largely, though not universally (Section 4.2) supported SOP’s account’s of object recognition.

(8)

ahigherformoftemporalmemory(e.g.,[13]).Thereisnothingin theprecedingdiscussiontoinformonthestatusoftheseintriguing andimportantpossibilitiesbuttheyareprobablymorecomplex explanationsthatarerequiredtoexplainavailablefindings.More troublingforsuchaccountsisthattheevidencedescribedhere sup-portsdirectpredictionsfromSOP.Itispossibleofcoursethatthese findingscouldalsobeinterpretedintermsofmoresophisticated memorysystems,thoughitisuncleartoushowthiscouldbedone anditis probablyunnecessarytodo so.Beingbasedonalarge bodyofexperimentalevidencefromthestudyofhabituationand Pavlovianconditioning,SOPcanbeusedtodeliverseveraltestable areasofexperimentationthatcouldfurtherevaluatethisaccount ofrecognitionmemory.

Disclosurestatement

Neitherauthorisawareofanyconflictofinterestcreated asso-ciatedwiththisreport.

Acknowledgements

TheauthorsthankDavidSandersonandEricTamforhelpful discussion.

AppendixA. Supplementarydata

Supplementarymaterialrelatedtothisarticlecanbefound,in theonlineversion,athttp://dx.doi.org/10.1016/j.bbr.2014.10.046.

References

[1]BrandonSE,VogelEH,WagnerAR.StimulusrepresentationinSOP:I. The-oreticalrationalizationandsomeimplications.BehavProcess2003;62:5–25, http://dx.doi.org/10.1016/S0376-6357(03)00016-0.

[2]Aggleton JP, Brown MW. Interleaving brain systems for episodic and recognition memory. Trends Cogn Sci 2006;10:455–63, http://dx.doi.org/10.1016/J.Tics.2006.08.003.

[3]MackintoshNJ.Neurobiology,psychologyandhabituation.BehavResTher 1987;2:81–97.

[4]MandlerG.Recognizingthejudgementofpreviousoccurrence.PsycholRev 1980;87:252–712.

[5]DereE,HustonJP,DeSouzaSilvaMa.Thepharmacology,neuroanatomyand neurogeneticsofone-trialobjectrecognitioninrodents.NeurosciBiobehavRev 2007;31:673–704,http://dx.doi.org/10.1016/j.neubiorev.2007.01.005. [6]EacottMJ,GaffanEa.Therolesofperirhinalcortex,postrhinalcortex,andthe

fornixinmemoryforobjects,place,andeventsintherat.QJExpPsycholB 2005;58:202–17,http://dx.doi.org/10.1080/02724990444000203.

[7]EnnaceurA,DelacourJ.Anewone-trialtestforneurobiologicalstudiesof memoryinrats.I.Behaviouraldata.BehavBrainRes1988;31:47–59. [8]JonesPM,WhittEJ,RobinsonJ. Excitotoxicperirhinalcortexlesionsleave

stimulus-specifichabituationofsuppressiontolightsintact.BehavBrainRes 2012;229:365–71,http://dx.doi.org/10.1016/j.bbr.2012.01.033.

[9]RobinsonJ,SandersonDJ,AggletonJP,JenkinsTA.Suppressiontovisual, audi-tory,andgustatorystimulihabituatesnormallyinratswithexcitotoxiclesions oftheperirhinalcortex.BehavNeurosci2009;123:1238–50.

[10]BrownMW,AggletonJP.Recognitionmemory:whataretherolesofthe perirhi-nalcortexandhippocampus?NatRevNeurosci2001;2:51–61.

[11]RobinsonJ,WhittEJ,HorsleyRR,JonesPM.Familiarity-basedstimulus general-izationofconditionedsuppressioninratsisdependentontheperirhinalcortex. BehavNeurosci2010;124:587–99,http://dx.doi.org/10.1037/a0020900. [12]Cowell Ra, Bussey TJ, Saksida LM. Why does brain damage

impair memory? A connectionist model of object recognition memory in perirhinal cortex. J Neurosci 2006;26:12186–97, http://dx.doi.org/10.1523/JNEUROSCI.2818-06.2006.

[13]Hannesson DK, Howland JG, Phillips AG. Interaction between perirhi-nal andmedial prefrontal cortexis required fortemporalorder but not recognition memory for objects in rats. J Neurosci 2004;24:4596–604, http://dx.doi.org/10.1523/JNEUROSCI.5517-03.2004.

[14]EastonA,ZinkivskayA,EacottMJ.Recollectionisimpaired,butfamiliarity remainsintactinratswithlesionsofthefornix.Hippocampus2009;19:837–43, http://dx.doi.org/10.1002/hipo.20567.

[15]BrandonSE,VogelEH,WagnerAR.Acomponentialviewofconfiguralcuesin generalizationanddiscriminationinPavlovianconditioning.BehavBrainRes 2000;110:67–72.

[16]WagnerAR.SOP:amodelofautomaticmemoryprocessinginanimalbehavior. In:SpearNE,MillerRR,editors.Inf.process.anim.mem.mech.Hillsdale,New Jersey:Erlbaum;1981.p.5–47.

[17]WagnerAR,BrandonSE.Evolutionofastructuredconnectionistmodelof Pavlovianconditioning(AESOP).In:KleinSB,MowrerRR,editors.Contemp. learn.theor.Pavlov.cond.statustradit.learn.theory.Hillsdale,NJ:Erlbaum; 1989.

[18]HoneyRC,GoodM.Associativecomponentsofrecognitionmemory.CurrOpin Neurobiol2000;10:200–4.

[19]RescorlaRA,WagnerAR.AtheoryofPavlovianconditioning:variationsinthe effectivenessofreinforcementandnonreinforcement.In:BlackAH,Prokasy WF,editors.Class.Cond.IICurr.Res.Theory.NewYork: Appleton-Century-Crofts;1972.p.64–99.

[20]SandersonDJ,BannermanDM.Competitiveshort-termandlong-term mem-ory processes inspatial habituation. J Exp Psychol AnimBehav Process 2011;37:189–99,http://dx.doi.org/10.1037/a0021461.

[21]DavisM.Effectsofinterstimulusintervallengthandvariabilityon startle-response habituationintherat.JCompPhysiolPsychol1970;72:177–92, http://dx.doi.org/10.1037/h0037208.

[22]WintersBD,ForwoodSE,CowellRA,SaksidaLM,BusseyTJ.Doubledissociation betweentheeffectsofperi-postrhinalcortexandhippocampallesionsontests ofobjectrecognitionandspatialmemory:hetrogeneityoffunctionwithinthe temporallobe.JNeurosci2004;24:5901–8.

[23]LiuP,BilkeyDK.Theeffectofexcitotoxiclesionscenteredonthe hippocam-pusorperirhinalcortexinobjectrecognitionandspatialmemorytasks.Behav Neurosci2001;115:94–111.

[24]KivyPN,EarlRW,WalkerEL.Stimuluscontextandsatiation.JCompPhysiol Psychol1956;49:90–2.

[25]EnnaceurA,NeaveN,AggletonJP.Neurotoxiclesionsoftheperirhinalcortext donotmimicthebehaviouraleffectsoffornixtransectionintherat.Behav BrainRes1996;80:9–25.

[26]Good MA, Barnes P, Staal V, McGregor A, Honey RC. Context but not familiarity-dependent formsof objectrecognition are impairedfollowing excitotoxichippocampallesionsinrats.BehavNeurosci2007;121:218–23, http://dx.doi.org/10.1037/0735-7044.121.1.218.

[27]NelsonAJD,CooperMT,ThurKE,MarsdenCA,CassadayHJ.Theeffectof cate-cholaminergicdepletionwithintheprelimbicandinfralimbicmedialprefrontal cortexonrecognitionmemoryforrecency,location,andobjects.Behav Neu-rosci2011;125:396–403,http://dx.doi.org/10.1037/a0023337.

[28]Barker GRI,BirdF, AlexanderV,WarburtonEC.Recognitionmemoryfor objects,place,andtemporalorder:adisconnectionanalysisoftheroleofthe medialprefrontalcortexandperirhinalcortex.JNeurosci2007;27:2948–57, http://dx.doi.org/10.1523/JNEUROSCI.5289-06.2007.

[29]MitchellJB, LaiaconaJ.Themedial frontalcortexandtemporalmemory: testsusingspontaneousexploratorybehaviourintherat.BehavBrainRes 1998;97:107–13.

[30]ChibaAA,KesnerRP,GibsonCJ.Memoryfortemporalorderofnewandfamiliar spatiallocationsequences:roleofthemedialprefrontalcortex.LearnMem 1997;4:311–7,http://dx.doi.org/10.1101/lm.4.4.311.

[31]Tam SKE,RobinsonJ, Jennings DJ,Bonardi C. Dissociations inthe effect of delay on object recognition: evidence for an associative model of recognition memory. J Exp Psychol Anim Behav Process 2013, http://dx.doi.org/10.1037/xan0000003.

[32]DixS,AggletonJA.Extendingthespontaneouspreferencetestofrecognition: evidenceofobject-locationandobject-contextrecognition.BehavBrainRes n.d.;99:191–200.

[33]NormanG,EacottMJ.Dissociableeffectsoflesionstotheperirhinalcortexand thepostrhinalcortexonmemoryforcontextandobjectsinrats.BehavNeurosci 2005;119:557–66.

[34]WilsonDIG,LangstonRF,SchlesigerMI,WagnerM,WatanabeS,AingeJA. Lateralentorhinalcortexiscriticalfornovelobject-contextrecognition. Hip-pocampus2013;23:352–66,http://dx.doi.org/10.1002/hipo.22095.

[35]SpanswickSC,SutherlandRJ.Object/context-specificmemorydeficits associ-atedwithlossofhippocampalgranulecellsafteradrenalectomyinrats.Learn Mem2010;17:241–5,http://dx.doi.org/10.1101/lm.1746710.

[36]Spanswick SC,Dyck RH. Object/contextspecific memory deficits follow-ing medial frontal cortex damage in mice. PLOS ONE 2012;7:e43698, http://dx.doi.org/10.1371/journal.pone.0043698.

[37]MumbyDG,PiterkinP,LecluseV,LehmannH.Perirhinalcortexdamageand anterogradeobject-recognitioninratsafterlongretentionintervals.Behav BrainRes2007;185:82–7.

[38]Langston RF, Wood ER. Associative recognition and the hippocam-pus: differential effects of hippocampal lesions on object-place, object-contextandobject-place-contextmemory.Hippocampus2010;20:1139–53, http://dx.doi.org/10.1002/hipo.20714.

[39]DuvaCA,FlorescoSB,WunderlichGR,LaoTL,PinelJP,PhillipsAG.Disruptionof spatialbutnotobject-recognitionmemorybyneurotoxiclesionsofthedorsal hippocampusinrats.BehavNeurosci1997;111:1184–96.

[40]ClarkRE,WestAN,ZolaSM,SquireLR.Ratswithlesionsofthe hippocam-pusareimpairedonthedelayednonmatching-to-sampletask.Hippocampus 2001;11:176–86,http://dx.doi.org/10.1002/hipo.1035.

(9)

[42]McTighe SM,Cowell RA,WintersBD, BusseyTJ, SaksidaLM.Paradoxical false memoryforobjectsafterbraindamage. Science2010;330:1408–10, http://dx.doi.org/10.1126/science.1194780.

[43]WhittE,HaselgroveM,RobinsonJ.Indirectobjectrecognition:evidencefor associativeprocessesinrecognitionmemory.JExpPsycholAnimBehavProcess 2012;38:74–83,http://dx.doi.org/10.1037/a0025886.

[44]Whitt E, Robinson J. Improved spontaneous object recognition follow-ing spaced preexposure trials: evidence for an associative account of recognition memory. J Exp PsycholAnim Behav Process 2013;39:174–9, http://dx.doi.org/10.1037/a0031344.

[45]AlbasserMM,DaviesM,FutterJE,AggletonJP.Magnitudeoftheobject recogni-tiondeficitassociatedwithperirhinalcortexdamageinrats:effectsofvarying thelesion extentandthedurationofthesampleperiod.BehavNeurosci 2009;123:115–24.

[46]MeunierM,BachevalierJ,MishkinM,MurrayEA.Effectsonvisualrecognition ofcombinedandseparateablationsoftheentorhinalandperirhinalcortexin rhesusmonkeys.JNeurosci1993;13:5418–32.

[47]Baxter MG, Murray Ea. Opposite relationship of hippocampal and rhi-nal cortex damage to delayed nonmatching-to-sample deficits in monkeys. Hippocampus 2001;11:61–71, http://dx.doi.org/10.1002/ 1098-1063(2001)11:1<61::AID-HIPO1021>3.0.CO;2-Z.

[48]Zola-MorganS,SquireLR,AmaralDG,SuzukiWa.Lesionsofperirhinaland parahippocampalcortexthatsparetheamygdalaandhippocampalformation produceseverememoryimpairment.JNeurosci1989;9:4355–70.

[49]MumbyDG,PinelJP.Rhinalcortexlesionsandobjectrecognitioninrats.Behav Neurosci1994;108:11–8.

[50]ClarkRE,ZolaSM,SquireLR.Impairedrecognitionmemoryinratsafterdamage tothehippocampus.JNeurosci2000;20:8853–60.

[51]WintersBD,BusseyTJ.Transientinactivationofperirhinalcortexdisrupts encoding,retrieval,andconsolidationofobjectrecognitionmemory.JNeurosci 2005;25:52–61,http://dx.doi.org/10.1523/JNEUROSCI.3827-04.2005. [52]NormanG,EacottMJ.Impairedobjectrecognitionwithincreasinglevelsof

featureambiguityinrats withperirhinal cortexlesions.BehavBrainRes 2004;148:79–91.

[53]Warburton EC,KoderT, ChoK,MasseyPV,Duguid G,Barker GRI,et al. Cholinergicneurotransmissionisessentialforperirhinalcorticalplasticityand recognitionmemory.Neuron2003;38:987–96.

[54]EastonA,EacottMJ.Recollectionofepisodicmemorywithinthemedial tem-porallobe:behaviouraldissociationsfromothertypesofmemory.BehavBrain Res2010;215:310–7,http://dx.doi.org/10.1016/j.bbr.2009.10.019.

[55]MumbyDG, GaskinS,GlennMJ,SchramekTE, LehmannH.Hippocampal damageandexploratorypreferencesinrats:memoryforobjects,places,and contexts.LearnMem2002;9:49–57,http://dx.doi.org/10.1101/lm.41302. [56]MitchellJB, LaiaconaJ.Themedial frontalcortexandtemporalmemory:

testsusingspontaneousexploratorybehaviourintherat.BehavBrainRes 1998;97:107–13,http://dx.doi.org/10.1016/S0166-4328(98)00032-1. [57]HannessonDK,VaccaG,HowlandJG,PhillipsAG.Medialprefrontalcortex

is involvedinspatialtemporalordermemorybutnotspatial recognition memoryintestsrelyingonspontaneousexplorationinrats.BehavBrainRes 2004;153:273–85,http://dx.doi.org/10.1016/j.bbr.2003.12.004.

[58]HoneyRC,WattA,GoodM.Hippocampallesionsdisruptanassociative mis-matchprocess.JNeurosci1998;18:2226–30.

[59]RescorlaRA.Stimulusgeneralization:Somepredictionsfromamodelof Pavlo-vianconditioning.JExpPsycholAnimBehavProcess1976;2:88–96. [60]PearceJM.AmodelforstimulusgeneralizationinPavlovianconditioning.

PsycholRev1987;94:61–73.

[61]HarrisJA.Elementalrepresentationsofstimuliinassociativelearning.Psychol Rev2006;113:584–605,http://dx.doi.org/10.1037/0033-295x.113.3.584. [62]HebbDO.Theorganizationofbehavior.NewYork:JohnWiley&Sons,Inc.;

1949.

[63]WagnerAR,LoganFA,HaberlandtK,PriceT.Stimulusselectioninanimal dis-criminationlearning.JExpPsychol1968;76:171–80.

[64]KaminLJ.Predictability,surprise,attention,andconditioning.In:CampbellBA, ChurchRM,editors.Punishm.AversiveBehav.NewYork: Appleton-Century-Crofts;1969.p.279–96.

[65]Bonardi C, Ward-Robinson J. Occasion setters: specificity to the US and the CS-US association. Learn Motiv 2001;32:349–66, http://dx.doi.org/10.1006/Lmot.2001.1089.

[66]Pearce JM, Hall G. A model for Pavlovian learning – variations in the effectivenessofconditionedbutnotofunconditionedstimuli.PsycholRev 1980;87:532–52.

[67]Haselgrove M, Robinson J, Nelson A, Pearce JM. Analysis of an ambiguous-feature discrimination. Q J Exp Psychol 2008;61:1710–25, http://dx.doi.org/10.1080/17470210701680746.

[68]AllmanMJ,Ward-RobinsonJ,HoneyRC.Associativechangeinthe represen-tationsacquiredduringconditionaldiscriminations:furtheranalysisofthe natureofconditionallearning.JExpPsycholBehavProcess2004;30:118–28, http://dx.doi.org/10.1037/0097-7403.30.2.118.

[69]RescorlaRA.Simultaneousandsuccessiveassociationsinsensory precondi-tioning.JExpPsycholBehavProcess1980;6:207–16.

[70]RescorlaRA.Pavlovianconditionedinhibition.PsycholBull1969;72:77–94, http://dx.doi.org/10.1037/h0027760.

[71]PapiniMR,BittermanME.Thetwo-teststrategyinthestudyofinhibitory con-ditioning.JExpPsycholAnimBehavProcess1993;19:342–52.

[72]HallJF.BackwardconditioninginPavloviantypestudies.Reevaluationand presentstatus.PavlovJBiolSci1981;19:163–8.

[73]Holland PC, Lamarre J. Transfer of inhibition after serial and simul-taneous feature negative discrimination training. Learn Motiv 1984;15: 219–43.

[74]RescorlaRa.Superconditioningfromareducedreinforcer.QJExpPsycholB 2004;57:133–52,http://dx.doi.org/10.1080/02724990344000051.

[75]JenningsDJ,AlonsoE,MondragónE,FranssenM,BonardiC.Theeffectof stim-ulusdistributionformontheacquisitionandrateofconditionedresponding: implicationsfortheory.JExpPsycholAnimBehavProcess2013;39:233–48, http://dx.doi.org/10.1037/a0032151.

[76]Vogel EH, Brandon SE, Wagner AR. Stimulus representation in SOP: II. Anapplication to inhibition of delay. Behav Process 2003;62:27–48, http://dx.doi.org/10.1016/S0376-6357(03)00050-0.

[77]HallG,HoneyRC.Contextualeffectsinconditioning,latentinhibition,and habituation:associativeand retrievalfunctionsof contextualcues.JExp PsycholAnimBehavProcess1989;15:232–41.

[78]Blair CAJ, Hall G. Perceptual learning in flavor aversion: evidence for learnedchangesinstimuluseffectiveness.JExpPsycholAnimBehavProcess 2003;29:39–48.

[79]HoneyRC,GoodMA,ManserKL.Negativepriminginassociativelearning: evi-dencefromaserial-habituationprocedure.JExpPsycholAnimBehavProcess 1998;24:229–37.

[80]PavlovIP.Conditionedreflexes.NewYork:DoverPublications,Inc.;1927. [81]RescorlaRA.Experimentalextinction.IntJPsychol2000;35:109.

[82]GrovesPM,ThompsonRF.Habituation:adual-processtheory.PsycholRev 1970;77:419–50,http://dx.doi.org/10.1037/h0029810.

[83]SokolovYN.Perceptionandtheconditionedreflex.Oxford:PergamonPress, Ltd.;1963.

[84]KonorskiJ.Integrativeactivityofthebrain.Chicago:UniversityofChicago Press;1967.

[85]VinogradovaOS.Hippocampusascomparator:roleofthetwoinputandtwo outputsystemsofthehippocampusinselectionandregistrationof informa-tion.Hippocampus2001;11:578–98.

[86]AndersonMJ,JablonskiSA,KlimasDB.Spacedinitialstimulusfamiliarization enhancesnoveltypreferenceinLong-Evansrats.BehavProcess2008;78:481–6, http://dx.doi.org/10.1016/j.beproc.2008.02.005.

[87]Tam SKE, Jennings DJ, Bonardi C. Dorsal hippocampal involvement in conditioned-response timing and maintenance of temporal infor-mation in the absence of the CS. Exp Brain Res 2013;227:547–59, http://dx.doi.org/10.1007/s00221-013-3530-4.

Figure

Updating...