Stimulus–response bindings in priming

HAL Id: hal-02386864

https://hal.archives-ouvertes.fr/hal-02386864

Submitted on 29 Nov 2019

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of

sci-entific research documents, whether they are

pub-lished or not. The documents may come from

teaching and research institutions in France or

abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est

destinée au dépôt et à la diffusion de documents

scientifiques de niveau recherche, publiés ou non,

émanant des établissements d’enseignement et de

recherche français ou étrangers, des laboratoires

publics ou privés.

Stimulus–response bindings in priming

Richard Henson, Doris Eckstein, Florian Waszak, Christian Frings, Aidan

Horner

To cite this version:

Richard Henson, Doris Eckstein, Florian Waszak, Christian Frings, Aidan Horner.

Stimulus–

response bindings in priming.

Trends in Cognitive Sciences, Elsevier, 2014, 18 (7), pp.376-384.

�10.1016/j.tics.2014.03.004�. �hal-02386864�

Stimulus–response

bindings

in

priming

Richard

N.

Henson

,

Doris

Eckstein

,

Florian

Waszak

,

Christian

Frings

,

and

Aidan

J.

Horner

1

MRCCognitionandBrainSciencesUnit,Cambridge,UK

2

Institutfu¨rPsychologie,Universita¨tBern,Bern,Switzerland

3_{CenterforCognition,Learning,andMemory,Universita¨tBern,Bern,Switzerland} 4

InstitutNeurosciencesCognition,Universite´ParisDescartes,Paris,France

5

CNRSLaboratoirePsychologiedelaPerceptionUMR8242,Universite´ParisDescartes,Paris,France

6_{AllgemeinePsychologieundMethodenlehre,Universta¨tTrier,Trier,Germany} 7_{InstituteofCognitiveNeuroscience,UniversityCollegeLondon,London,UK} 8

InstituteofNeurology,UniversityCollegeLondon,London,UK

Peoplecanrapidlyformarbitraryassociationsbetween

stimuliandtheresponsestheymakeinthepresenceof

those stimuli. Suchstimulus–response (S–R) bindings,

when retrieved, affecttheway that peoplerespond to

thesame,orrelated,stimuli.Onlyrecently,however,has

the flexibilityand ubiquityof these S–Rbindingsbeen

appreciated,particularlyinthecontextofpriming para-digms.Thisisimportantforthemanycognitivetheories thatappealtoevidencefrompriming.Itisalsoimportant

for the control of action generally. An S–R binding is

more than a gradually learned association between a

specific stimulus and a specific response; instead, it

captures the full, context-dependentbehavioral

poten-tialof astimulus. Introduction

Ourdailylivesinvolveinteractingwithalargenumberof

stimuli. Many of these stimuli occur not only once, but

recur at differenttimescales. Wetherefore needtolearn

rapidly how to process these recurring stimuli, without

necessarily intentionallyrecallingpriorexperiences with thosestimuli.Oneexampleofthisrapidlearningis

prim-ing, in which a change in the mental processing of a

stimulus is normally measured by an overt behavioral

response cuedby thatstimulus (Box1). Primingisoften interpretedasfacilitationofoneormoreofthe computa-tions, or ‘componentprocesses’[1], that arenecessaryto generatethatresponse.Inatypicallaboratoryexperiment, forexample,participantsmightpressoneoftwobuttons

depending on whether they judge a visually presented

object tobelivingor nonliving,forwhich primingwould be apparent if their average reaction time (RT) for this judgmentisshorterforrepeatedthanforinitial presenta-tions oftheobjects. Inthis example,the component

pro-cesses that are facilitated might include perceptual

identification ofthe object depicted(perceptualpriming)

and/orretrievalofsemanticinformationaboutthatobject (conceptualpriming).

However,ithaslongbeensuspectedthat primingcan

alsoresultfromdirectlyassociating,orbinding,astimulus andresponse.IftheseS–Rbindingsareretrievedwhenthe stimulusisrepeated,theresponsecanbeproducedwithout

necessarily recapitulating the component processes that

wereinitiallyusedtogeneratethatresponse(Figure1A).

Evidence for S–R bindings has been found in all major

priming paradigms (Box 1 and Figure 1B): repetition

priming[2],negativepriming[3],andmaskedpriming[4]. ThestudyofS–Rbindingsisimportantfortworeasons. First,thepresenceofS–Rbindingspotentiallyconfounds

interpretation of many priming effects; for example,

whether unattended items are truly inhibited [5,6] or

whethersemanticrepresentationscanreallybeaccessed

unconsciously [4,7].More importantly,S–R bindingsare nowrecognizedtoplayavitalroleinthecontrolofaction;a rolethat goes beyond themere acceleration of stimulus-drivenaction.Forexample,researchreviewedbelow

indi-cates that S–R bindings encode information at multiple

levels of abstraction, furnishing considerable flexibility

and context sensitivity in their deployment. However,

despitetherecentinterestinS–Rbindings,manycrucial

questionsremainunanswered.

UbiquityandflexibilityofS–Rbindings

AlthoughtheconceptofS–Rbindingsisnotnew(Box2),a recentresurgenceininteresthasbeendrivenbyevidence fortheirgreaterprevalenceandflexibilitythanpreviously

thought. For example, S–R bindings are far from being

simpleassociationsbetweenaspecificstimulusand specif-icresponse;rather,theyappeartobestructuredbindings involving multiple levels of representation of responses, stimuli,andtasks(Figure2).Moreover,thesebindingsdo notneedtobegraduallylearned;theycanbeformedfroma singlepairingofastimulusandresponse.

Thenatureof‘R’

Despite evidence for effector-specific response codes [8],

there is general agreement that responses can also be

representedby theirgoal,ratherthanjustspecificmotor programs[9,10].Primingeffectsthatsurviveaswitchin 1364-6613/

ß2014TheAuthors.PublishedbyElsevierLtd.Thisisanopenaccessarticleunder theCCBYlicense(http://creativecommons.org/licenses/by/3.0/).http://dx.doi.org/ 10.1016/j.tics.2014.03.004

Correspondingauthor:Henson,R.N. (rik.henson@mrc-cbu.cam.ac.uk).

Keywords: S–R bindings; repetitionsuppression; automaticity; masked priming; subliminalpriming;negativepriming.

effectorbetweenprimeandprobeareconsistentwiththis claim(e.g.,[11]inthecaseofnegativepriming).

However, priming that is invariant to a change in

effector isnot necessarily evidence forabstract response

representations in S–R bindings, because the residual

primingcouldreflectotherfactors,likefacilitationof com-ponent processes(orinhibition ofastimulus representa-tioninthecaseof[11]). Toidentify thelevelofresponse

representationwithinanS–Rbinding,onemustmeasure

the difference inpriming betweena congruent condition

(wherethesameresponseisgiveneachtimeastimulusis presented)andanincongruentcondition(wherethe oppo-siteresponseisgiven)oranunrelatedcondition(wherea differenttypeofresponseisgiven).Inthecaseofrepetition

priming,forexample,Horner andHenson[12]showed a

reduction in priming when the response to a repeated

stimulus switched from a vocal yes/no to a manual key

press (relative to the congruent condition of a manual

response to both prime and probe). This suggests that

the specific motor action is indeed encoded in the S–R

binding. However, primingfrom avocal yes/no response

tomanualresponsewasstillgreaterthanwhentheprime objectwassimplynamed.ThissuggeststhatS–Rbindings

additionally encode more abstract response

representa-tions,suchas ayes/nodecision(seealso[13]).

FurtherresearchhassuggestedthatS–Rbindingscan

encodeevenmoreabstractresponserepresentations,such astheparticularclassificationmade(suchaslivingversus nonliving)[14–18].Bychangingthereferenceobjectduring arelative-sizejudgmenttask(anexperimental

manipula-tion introduced by [19]), Horner and Henson [12] found

greaterrepetitionprimingforobjectsthatmaintainedthe same bigger/smaller classification despite this reference changethan forobjectsforwhichtheclassification chan-ged.In thelattercase,note thatthe yes/nodecisionand motoractionwereunchanged;onlytheclassificationlabel changed.ThesedatasuggestthatS–Rbindingscan simul-taneously encode at least three levelsof response repre-sentation: action, decision, and classification (Figure 2).

Furthermore, response representations like these have

beenshowntohaveindependent[20,21]effectson behav-ior.

Ithasalsobeensuggestedthatstimulicanbeboundto

nonspecific ‘stop codes’ that, when retrieved, inhibit

responses in any ongoing task [22,23]. It has even been proposed that stimuli canbebound toattentional filters that have been previously applied to those stimuli [24].

Box1.Majortypesofpriming

Primingreferstoachangeinaccuracy,bias,orreactiontimetorespond

toastimulus(‘probe’)owingtopriorpresentationofthesameorasimilar

stimulus(‘prime’).Itisindexedasthedifferencebetweentheresponseto

theprimeandprobeorbetweentheprobeandacomparablestimulus

notpresentedbefore(‘unprimed’).Thereareatleastthreemainpriming

paradigms,whichdifferinwhetheraresponseismadeto:(i)theprime;

and/or(ii)theprobe(seeFigure1Binmaintext).

Inrepetitionpriming,aresponseismadetobothprimeandprobe.

Theprimeandprobetypicallyoccurinseparatetrialsseparatedby

multipleinterveningtrials(sothatprimingdoesnotsimplyoweto

repetition of the same response across successive trials; that is,

‘responsepriming’).Notethatthestimulusmaynot berepeatedin

exactlythesamephysicalform(e.g.,itmayswitchfromawrittentoa

spokenwordacrosstrials).

Innegative priming, the prime (andsometimes the probe)is a

distractor; that is, it is not the focus of attention, such that the

response isgenerated insteadby a different, concurrent stimulus.

Responsesaretypicallyslowedwhentheprimeisthenpresentedasa

probeinasubsequenttrial(althoughnotethatprimingisnotalways

‘negative’inthesenseofaslowing;positiveprimingcanoccurifthe

response previously pairedwith the prime is congruent with that

requiredtotheprobe).

Inmaskedpriming,theprime ismasked torenderitsubliminal,

suchthatnoresponsecanbemadetoit.Thesubsequentsupraliminal

probetypicallyoccurswithinafewhundredmilliseconds(i.e.,within

thesametrial).TheS–Rbindingthatispotentiallyretrievedbythe

maskedprimethenusuallycomesfrompreviouspresentationsofthe

samestimulusasaprobeinothertrials.

Theseparadigmsdifferinotherwaystoo.Forexample,repetition

primingcansurvivelagsofmanyinterveningtrialsandsometimes

last days, whereas negative priming typically occurs only across

successivetrialsandmaskedprimingrarelylastsbeyondonetrial.

There are also other priming paradigms (e.g., semantic/affective

priming,whereprimeandprobearedifferentstimulibutsemantically

oraffectivelyrelated),butherewefocusonparadigmsinwhichS–R

bindings have been shown to play a dominant role, and these

normally involve repeating a stimulus to cue retrieval ofan S–R

binding(althoughseetextaboutpossibleF–Rbindings).

Box2.S–Rtheories

TherearemanytheoriesrelatingtoS–Rbindings;wefocusonone

example pertinent to each of the three priming paradigms

consideredhere(Box1).

Loganproposedan‘instancetheory’[64] ofautomaticitythathas

been mainly applied to repetitionpriming. Responses are

deter-mined by a race between an algorithmic route (comprising

computationsusedtoproducearesponsethefirsttimeastimulus

isencountered)andretrievalofanyrelevantS–Rbindings(witha

separatesuch‘instance’storedeachtimearesponseisgiventoa

stimulus).AssumingthatallS–Rbindingsraceindependently,RTs

willdecreasewiththenumberofinstancesaccording toapower

law, with the mean and standard deviation having the same

exponent.Alaterversionofthetheory[74]includedtwoadditional

decisionrules(acounterandarandom-walkmodel)thatprioritize

responseaccuracyatthecostofincreasedRTsinconflictsituations.

With this extension, instance theory can alsopotentially explain

negativepriming.

Hommel [31] hypothesized the existence of‘episodic records’.

Theseencodefeaturesofstimuliandresponses,witheachrecord

binding only twofeatures (whichcan be fromdifferent stimuli),

althoughthesamefeaturecanoccurinmultiplerecords.Hommel’s

theorywasdevelopedprimarilytoexplain‘carry-over’costsfrom

onetrialtothenext(suchasnegativepriming),ratherthan

longer-livedfacilitatoryeffectsseeninrepetitionprimingparadigms.

Kundeetal.[75] proposedan‘actiontrigger’theory(seealso[76]),

whichismostoftenappliedtomaskedpriming.Thebasicideais

thatrepeatedpairingsofastimulusandresponseestablishatrigger

that releases an action when a related stimulus reappears. The

generalizationtorelatedstimulicomesfrombeingabletospecify

thetriggerconditionsinbroadterms.Forexample,evidencethat

the masked digit ‘3’ can prime a relative size judgment to a

subsequentprobe‘4’,evenifthe3wasnotpresentedpreviouslyin

theexperiment[60],canbeexplainedbyagenerictriggerofthesort

‘smallnumbersshouldproducea‘‘no’’response’.

Although theseS–R theories have proved helpful, each needs

further development toexplainthe preciserole ofattention and

awarenessduringencodingandretrievalofS–Rbindings(seetext)

andtheirinteractionswithothercomponentprocesses.Moreover,

fewifanytheoriesspecifyneuralmechanismsthatcanbecompared

withrecentbraindata(Box3).

Review

Thishasbeenusedtoexplainstimulus-specific

congruen-cy-proportion effects in Stroop tasks, where congruency

effectsarelargerforstimulithathavebeenpresentedin contexts withahigher proportion ofothercongruent sti-muli.

Multiplesimultaneouslevelsofresponserepresentation potentially allow rapid execution of a specific action, as appropriate, for example, if the context (e.g., task) is

unchanged, as well as allowing more flexible response

options if the context changes. The downside of this

1st presentaon (A) (B) 2nd presentaon Repeon priming Negave priming Masked priming Ta sk e.

g. “is it a living thing

”? “Yes” _{?? (“Yes”)} “Yes” “Yes” ?? (“No”) ?? (“No”)

Perceptual Conceptual Response

selecon

Perceptual Conceptual Response

selecon

Perceptual Conceptual Response

selecon

TRENDS in Cognitive Sciences

Figure1.Schematicofcomponentprocesses,stimulus–response(S–R)bindings,andprimingparadigms.(A)Whensomeoneisaskedtomakeadecisionaboutastimulus (e.g.,whethertheobjectdepictedbyanimageislivingornonliving),severalcomponentprocessesarerequiredto,forexample,identifyperceptuallytheobject(here,a lion)andretrieveconceptualinformationaboutit(thatalionisalivingentity)(toprow).Whenthatstimulusispresentedasecondtime,thereactiontime(RT)tomakethe samejudgmentisnormallyfaster,aphenomenoncalledpriming.Thiscouldreflectfacilitationofoneormoreofthecomponentprocessesengagedoninitialpresentation (secondrow)oritcouldreflectretrievalofanS–Rbindingthatencodesthestimulusandresponsemadeontheinitialpresentation,withoutneedingtore-engagethe originalcomponentprocesses(thirdrow).(B)Thethreemaintypesofprimingparadigmconsideredherearerepetitionpriming(toprow),negativepriming(middlerow), andmaskedpriming(bottomrow).Theinitialpresentationisshownontheleftandtherepeatpresentationontheright.Inthecaseofnegativepriming,theredsquare indicatesthetargetstimulustowhichparticipantsattendtodeterminetheirresponse(otherconcurrentstimuliaredistractors).Inthemaskedprimingcase,theprimeis oftenpresentedforlessthan50msandfollowedbyabackwardmask(illustratedbyasquareofpixelnoisehere)torenderitsubliminal.Thebrokenlinesindicatepotential encodingorretrievalofanS–Rbinding.

flexibilityforprimingexperimentsisthatsimplychanging theeffector between prime andprobeis notsufficient to ‘controlfor’S–Rbindings.Likewise,thepresenceofmore specificresponsecodesmeansthatchangingtasksbetween primeandprobe isnotasufficientcontroleither,ifboth tasksrequireayes/nodecision or thesamemotor action

[25]. Toproperly investigate priming irrespective of the influenceofS–Rbindings,alllevelsofresponse represen-tationmustbecontrolledsimultaneously[26].

Thenatureof‘S’

Similarquestionsrelate tothe natureofstimulus repre-sentationswithinS–Rbindings.Priming,andits

modula-tionbyresponsecongruency,hasbeenshowntodecrease

withdecreasingperceptualsimilaritybetweenprimeand

probe [19,26,27], consistent with S–R bindings encoding

relatively form-specific representations. However,

re-sponse congruency effects have also been found despite

switchingfromobjectpicturestowrittenobjectnames[26]

or from object picturesto object sounds [28].This again suggeststhat S–Rbindingscanencodemultiple levelsof stimulusrepresentation,includingattheabstractlevelof stimulus‘identity’(Figure2).

Response congruency effectshave beenshown for

se-manticallyrelatedstimuli[29,30].Thisraisesthequestion

ofwhetherbindingscanbeformedbetweenresponsesand

thefeaturesthatcomprisestimuli(see[31]).Inthecaseof maskedpriming,forexample,such‘feature–response’(F–

R)bindingsmayexplainprimingfromstimuli thatoccur

onlyoncein anexperiment: so-called‘novel’ primes[32].

This finding hasbeen assumed toexclude S–R bindings

(althoughseeBox2).Ifrelatedstimulihavebeenseen(as probes)andpairedwitharesponse,suchthatfeaturesof

those stimuli become bound with that response, later

repetitionofsomeofthosefeaturesinanovel(butrelated) primestimulusmaybesufficienttoretrievetheresponse andhenceprimethesubsequentprobe.Thishypothesisis

consistent with priming by novel words that comprise

parts of words seen previously as probes [33] and with

claimsthatmaskedprimingfromnovelstimulioccursonly whenstimulicomefromasmallandtightlyrelated stimu-lusset[32,34,35].

S–Rbindingsmayalsoincluderepresentationsofmore

than oneconcurrentstimulus.In negativepriming

para-digms,forexample,thereisevidencethatthetargetand distractorstimulusalsobecomeboundtogether, indepen-dentoftheirbindingtotheresponse[23].Such‘S–S bind-ings’seemtobedeterminedbytheprinciplesofperceptual grouping [36]. There is alsoevidence of S–Sbindings in

associative priming tasks where the response requires

comparing two or more concurrent stimuli [37]. Again,

thesedataimplyamorecomplexpictureofS–Rbindings

than is normally conceived, including multiple levels of

stimulus as well as response representation, bindings

between stimulus features and responses, and bindings

betweenmultiplestimuli.Thiscomplexityaffordsyet fur-therflexibilityin,forexample,allowinglearnedresponses tobetriggerednotonlybythesamestimulus,butalsoby similarstimuli.

Contextualbindings

Aspectsoftheconcurrentcontextmightalsobeboundwith thestimulusandresponse.Oneexampleisthetasksetin whichastimulusisencountered.Ithasbeenshownthatthe typical‘task-switchcost’,whichreflectsslowerRTsfortrials precededbyadifferentrelativetosametask,isincreasedif stimuliarerepeatedacrossthetasks[38].Thissuggeststhat the repetition of a stimulus automatically retrieves the

previous task set associated with that stimulus, which

can interfere with any new task set (also see [14,39]).

Importantly, Waszak and colleagues [17,40] argued that

S–Rbindings aremorelikelytoberetrievedifthey were compiled underatasksetthatremainsactiveduringthe probetrial(giventhataprevioustasksetremainsactivefor acertaintimeafterataskswitch:so-called‘task-setinertia’

[41]).Taskset-dependentretrievalclearlymakesadaptive sense,inthatonewouldnotwantallpreviousresponsesthat havebeenassociatedwithastimulusconstantlytocompete with current behavioral goals (cf. ‘utilization’ behavior

[42,43]). Other types ofspatial or temporal context (e.g., laboratorysetting)mayalsomediateS–Rretrieval.Overlap inthislevelofcontextmayexplainwhypriorresponsescan stillbecuedbyarepeatedstimulusdespiteaswitchintask

[12], at least when specific response options are shared betweenthetasks(asinFigure2).

Morerecently,anewlineofresearchhasexploredhow S–Rbindingsmightbeformedsimplybyverbalinstruction

[44–48].Forexample,Wenke andcollaborators[49]

pre-sented participants with a set of S–R mappings (e.g.,

N=leftkey, K=right key) foronetask (TaskA).Before Size task Living task Spaal/ temporal context Classificaon: “Bigger” Decision: “Yes” [Lion] identy Lion Acon: Right finger-press Smulus Classificaon: “Living”

TRENDS in Cognitive Sciences

Figure 2.Possible stimulus–response (S–R) binding. Schematic of a possible structuredS–Rbindingformedbygivingaresponsetoapictureofalionduringa binary ‘bigger than shoebox?’ categorization task, where red lines indicate bindings.Stimulusrepresentationsincludeavisualimageofthepictureanda moreabstractrepresentationoftheidentityofthatstimulus,suchthatiftheword ‘lion’islaterpresented,itcanalsocueresponsesviathebindingsbetweenthe identityrepresentationandresponserepresentations.Responserepresentations includeaspecificmotor action(e.g.,rightindex finger depression),abinary decision(e.g.,‘yes’)andaparticularclassification(e.g.,‘bigger’inthesizetask). Thismeansthatretrievalofanactionordecisioncaninfluenceresponsesevenif thetaskischanged;forexample,toan‘istheobjectliving?’categorizationinstead (asshown).Similarly,retrievalofadecisioncaninfluenceresponsesevenifthe effector(action)ischangedandretrievalofaclassificationcaninfluenceresponses evenifthetask(andhencedecisionandaction)isreversed(e.g.,toa‘smallerthan shoebox?’task).RetrievaloftheS–Rbindingmayalsobemediatedbythespatial/ temporalcontext(e.g.,laboratorysetting).

Review

attemptingTaskA,participants performedanothertask, TaskB.AlthoughtheinstructedmappingsforTaskAwere irrelevanttoTaskB,theyinterferedwiththeperformance ofTaskBwhenthestimuliinTaskBoverlappedwiththose

instructed. Instruction-based S–R bindings might allow

people quicklytoimplement anyarbitraryS–Rmapping

and to use this mapping to guide behavior early in the

learning of complex skills. However, the nature of

instructed S–R mappings is not yet well known. One

possibility is that instructions result in covertexecution oftheinstructedmapping,withresultssimilartotheovert applicationofthemapping.

Roleofattentionandawarenessatencodingand

retrieval

Althoughattentionandawarenessareintimatelyrelated, onecanbeawareofastimulusdespiteitnotbeingthefocus ofattention,asinnegativepriming,oronecanattendtoa

specific point in time and space but not be aware of a

stimulus presented atthat point,as inmasked priming.

Towhatextentareattentionandawarenessimportantfor theencodingand/orretrievalofS–Rbindings?

Thenegativeprimingparadigmhasshownthat

atten-tionisnotnecessaryforencodingS–Rbindings.For exam-ple,Rothermundetal.[3]presentedstringsoffiveletters

(e.g.,BFBFB), inwhichonlythesecondandfourth were

taskrelevant.Theyfoundthestandardnegativepriming

effect whenthe distractorletters (intheother positions) became taskrelevant (i.e.,targets)in asubsequent trial andthecorrectresponsewasincongruentwiththatgiven on the original trial, but positive priming when the re-sponsewascongruent.Indeed,itmaymakeadaptivesense tobindallstimulitoresponseswhenencodingnew

experi-ences,becauseonedoesnotalwaysknowwhichstimulus

will berelevantinthefuture. Fringsandcolleagues[50]

also found a response congruencyeffect when distractor

stimuli were repeated as distractors,suggestingthat at-tentionisnotnecessaryforretrievalofS–Rbindingseither (see also [51–53] for evidence from repetition priming).

Nonetheless, negative priming experiments using other

stimulus configurations [54,55] or longer lags between

repetitions (A. Horner, PhD thesis, University of

Cam-bridge, 2010) suggest that attention may sometimes be

necessary.Onepossibilityisthatbindingsinitiallyoccur between allstimuli, attended or unattended, ina short-lived ‘event file’ [31], but only the bindings to attended stimulilastlonger.

Althoughattentionappearstobenecessaryformasked

priming [56,57], response congruency effects in masked

priming suggest that awareness is not necessary for

re-trieving S–R bindings [58]. Eckstein and Henson [58],

however,foundnoevidenceofresponsecongruencyeffects

formaskedprimesthat wereneverseenunmasked,

sug-gesting that awareness is necessary for encoding such

bindings.Althoughotherstudieshavefoundmaineffects

ofprimingfromprimesneverseenunmasked[32,59–61],

thisresidualprimingcouldreflectunconsciousfacilitation

of component processes [59,60] rather than subliminal

encoding of S–R (or F–R) bindings. Again, to establish

the role of a factor like awareness or attention in the

encoding or retrieval of S–R bindings per se, one needs

to find an interaction between that factor and response

congruency.

InteractionsbetweenS–Rbindingsandcomponent

processesinresponseselection

SeveralquestionsremainaboutthenatureofS–Rbindings

andhowtheyinteract withother processestodetermine

behavior.Forexample,doeseachpairingofastimulusand

response produce a new S–R binding or progressively

strengthen an association between an existingstimulus

and response representation? Either possibility can

ex-plain why response congruency effects tend to increase

with the number of stimulus–response pairings

[4,12,62,63]. The finding that the standard deviation as wellasthemeanofRTsscaleswiththenumberofpairings hasbeenusedtoargueforseparateS–Rbindingsthatrace

independently to produce the response [64]. However,

Box3.S–Rbindingsinthebrain

Another reason for the resurgence of interest in S–R bindings

concerns recent neuroimaging and neuropsychological data. In

particular, the phenomenon of ‘repetition suppression’ has been

assumed to reflect the facilitation of component processes and

therefore used to map neural representations in different brain

regions[77,78].AninfluentialfunctionalMRI(fMRI)studybyDobbins

et al. [63], however, suggested that repetition suppression in

ventrotemporal regions (associated with visual object perception)

reflectsinsteadthebypassingofsuchprocessesowingtoretrievalof

S–Rbindings (see Figure3A in maintext). Although laterstudies

suggestedthatS–Rbindingscannotexplainallrepetitionsuppression

inperceptualregions,retrievalofS–Rbindingsclearlyhasimportant

effectsonneuroimagingdata[15,79].Recentworkhasfocusedon

relatingtheeffectsofresponsecongruencyinprefrontalcortextothe

integration of: (i) responses retrieved from S–R bindings; and (ii)

responsesgeneratedfromcomponentprocesses[15,80].

Effects of response congruency have been found in

response-lockedevent-relatedpotentials(ERPs)overfrontalelectrodesafew

hundred milliseconds before the response occurs [15,80] (see

Figure3Dinmaintext).ERPslockedtostimulusonset,bycontrast,

appear less affected by response congruency, suggesting that

stimulusrepetitioneffectsontheseERPsmayreflectfacilitationof

componentprocesses. Stimulus repetition alsomodulates 5–15-Hz

powerinventrotemporalregionsasmeasuredby

magnetoencepha-lography(MEG)[81]andincreasesthesynchronyofthisoscillatory

activitybetweenprefrontalandventrotemporalregions[82](cf.[83]).

Although these MEG studies did not manipulate response

con-gruency,theyraisethepotentialimportanceofchangesin

commu-nication between brain regions [81]. The importance of such

interactions was reinforced by a study showing that transcranial

magnetic stimulation ofthe prefrontal cortex abolished repetition

suppressioninventrotemporalregions[84].

Although amnesic patientswith damagetothemedial temporal

lobes(MTLs)havelongbeen claimedtoshowintact priming[85],

Schnyeretal.[86]foundnoeffectofresponsecongruencyinsuch

patients. This is consistent with hippocampal lesions in animals,

whichtypicallydisruptlearningofarbitraryvisuomotorassociations

[87].OnetentativepossibilityisthatS–Rbindingsarestoredinthe

MTLs (even if not necessarily in a conscious manner) that, when

retrieved,interactintheprefrontalcortexwithresponsesgenerated

bycomponentprocessesintheventrotemporalcortex.Finally,there

are also relevant data from single-cell recording in animals, and

neurallyplausiblecomputationalmodelsareclearlyvitaltointegrate

allthesetypesofdata[88,89].

when responses from these two routes are incongruent,

extra time seems necessary to resolve this discrepancy,

slowingRTs relativeto unprimedtrials [26], which sug-geststhatretrievalofS–Rbindingsinteractswith compo-nentprocessesduringthefinalstagesofresponseselection (Box2).

The idea that potentialresponsesretrieved from S–R

bindingsarevettedbyafinal stageofresponseselection affordsanextralayerofcognitivecontrolthatislikelytobe

important.For example,in situations where strong

top-downcontrolisrequired, itmaybepossibletobias selec-tionagainsttheresponsesretrievedfromS–Rbindingsand

in favor of responses generated by component processes

(for example, when accuracy is emphasized over speed).

Thus,althoughretrieval ofS–Rbindingsmaynot

neces-sarily require awareness or attention, retrieval is not

‘automatic’,inthesensethatitismodulatedbycontextual factorsliketaskset(reviewedabove),Gestaltmechanisms

[36], semantic matching [51,65], and feedback [66]. The

cognitive control of response selection then provides an

extra level of flexibility, which means that, even when

retrieved, S–Rbindings do notnecessarily dominate our

behaviorinaninflexiblemanner.However,thedetailsof thisresponseselectionstageremaintobeestablished,and wouldcertainlybenefitfromcomputationalmodelling(Box 3)and possibly convergent evidence from neuroscientific data(Figure3).

Although we have focused on RTs, incongruent S–R

bindingsmayalsoleadtoincreasederrorrates[4,60,67– 69].In thecaseofnegativepriming,multinomial proces-singmodelshavebeenusedtoanalyzetheprobabilityof

erroneous probe responses due toretrieval ofthe prime

response [68,70].If astimulusfromthe primeepisode is repeatedintheprobe,theprobabilityofresponding erro-neouslywiththeprimeresponseissignificantlyincreased

comparedwith whennostimulusisrepeated.Errorscan

thereforebeunderstoodasfailuresofcomponentprocesses toovercomeretrievalofS–Rbindingsinincongruenttrials.

2 P e rc en ta g e signal chang e 0.00 0.10 0.20 0.30 0.00 0.10 0.20 0.30 4 Fusiform PFC

Start Switch Return

(B)

(A) (C)

(D)

6 8 10

Post-smulus onset mes (s) Milliseconds

Response-locked Milliseconds Micr ov olts Smulus-locked Repeon congruent Repeon incongruent Novel incongruent Novel congruent Micr ov olts 2 4 6 8 10 2 4 6 8 10 –300 –100 0 Key: 100 200 300 400 500 600 700 800 –3 –2 –1 0 1 2 3 –6 –4 –2 0 2 4 6 –15 –200 –100 0 100 200

TRENDS in Cognitive Sciences

Figure3.Neuralcorrelatesofstimulus–response(S–R)retrieval.(A,B)DatafromthefunctionalMRI(fMRI)studyofDobbinsetal.[63]inwhichsimplyreversingthetaskina repetitionprimingparadigmreducedrepetitionsuppression(RS).Inthestartphase,participantsjudgedwhethervisualobjectswerebiggerthanashoebox;intheswitch phase,thesameobjects(togetherwithnew,unprimedobjects)werejudgedastowhethertheyweresmallerthanashoebox(inthereturnphase,theoriginal‘bigger’task wasreinstated).Theredpatchesinthethreeviewsofacanonicalbrainin(A)indicateregionsshowingsmallerresponsestoprimedthanunprimedtrialsinthestartphase (i.e.,RS).Theplotsin(B)showtheaveragefMRIevokedresponsetounprimed(darkblue)andprimed(lightblue)trialsfromtworepresentativesuchregions:theprefrontal cortex(PFC)andtheventrotemporalcortex(fusiform).NotethatRSinthefusiformisabolishedwhenthetaskisreversed.Dobbinsetal.suggestedthattheRSinthestart phasereflectedbypassingofcomponentprocesseswhenS–Rbindingsareretrieved,whereasthelackofRSintheswitchphaseariseswhenS–Rbindingsarenolonger used.Reproduced,withpermission,from[63].(C,D)Datafromtheevent-relatedpotential(ERP)studyofHornerandHenson[80].Participantsperformedthesame size-judgmenttaskasinDobbinsetal.,exceptthatthereferentobject(e.g.,ashoebox)wasswitchedbetweenprimeandprobetorenderthepreviousresponsecongruentor incongruent.(C)Atimewindow(greybox)withinastimulus-lockedERPoverparietalelectrodesduringwhichaneffectofstimulusrepetitionwasseenthatwasnot modulatedbywhethertheresponsewasrepeatedorreversedbetweenpresentations(atleastuntillater).(D)Aneffectoverfrontalelectrodesshowedaresponse congruencyeffectforprimed(repeated)stimuli,butfornotunprimed(novel)stimuli,afewhundredmillisecondsbeforeakeywaspressed(i.e.,response-locked).Whereas thestimulus-lockedeffectwashypothesizedtoreflectthefacilitationof(perceptual)componentprocesses,theresponse-lockedeffectwashypothesizedtoreflectdecision processesthatresolvetheconflictwhenresponsesretrievedfromS–Rbindingsandresponsesgeneratedbycomponentprocessesareincongruent.Reproduced,with permission,from[80].

Review

LimitationsofS–Rbindings

Althoughwehaveemphasizedthepervasivenessand

flex-ibility of S–R bindings in priming, we should note that

thereareprimingeffectsthatcannoteasilybeexplainedby S–Rbindings.Oneexampleisresidual(positive)priming

when all obvious levels of response code are reversed

between prime and probe, or at least when there is no

obvious overlap in response codes between prime and

probe [12]. Such cases arise when taskslike naming or

perceptual identificationareperformed,onthe primefor example,togetherwithadifferent(e.g.,classification)task performedontheprobe(orviceversa).Insuchcases,each stimuluswouldbeassociatedwithauniqueresponsethat

is not repeated in the probe task so could not modulate

priming. More generally, there is little doubt that prior processingofanintactvisualobjectcanmodifysubsequent perception ofadegradedversion (e.g.,abinarizedimage, suchasthefamousDalmatiandog[1]),suchthattheobject

is clearly seen when primed but not when unprimed,

without any overt behavioral response being made. For

furtherargumentsaboutprimingeffectsthatare

indepen-dent of S–R binding, see [12,60,71–73]. Moreover,

researchers should be wary of automatically appealing

toS–Rbindingstoexplainprimingunlessthere isdirect evidencefortheirexistence,suchasmodulationsofthesize oftheprimingeffectsbyresponsecongruency,asdescribed above.Finally,becausewehavealsoraisedthepossibility

ofF–R andS–Sbindings,itmayseemthatbindingscan

explainjustaboutanyaspectofhumanbehavior

(render-ing them somewhat vacuous as explanatory concepts).

However, we emphasize that S–R bindings are only

as-sumedheretoinfluencebehaviorintaskswhere

stimulus-cuedresponses overlapwith previousresponses tothose

stimuli;thatis,insituationswherethereareafter-effects ofpriorexperience.

Concludingremarks

Althoughthe cognitiverevolutiondispensedwith the

be-haviorist claim that all behavior can be understood in

termsofS–Rlearning,suchassociationsundoubtedlyplay aroleinmanyofourbehaviors.Importantly,S–Rbindings aremorethansimpleassociationsbetweenaspecific

per-cept and motor act; theyare complex, structured

repre-sentations that simultaneously bind multiple levels of

stimulus,response,andtaskrepresentation.Furthermore, S–Rbindingscan,undercertainexperimentalconditions,

be encoded and retrieved in the absence of attention or

awareness.Thiscomplexityandubiquitymakeitdifficult tocontrolforS–Rbindingswhenusingprimingto investi-gateothertheoreticalquestions.Moreover, S–Rbindings arenolongerviewedonlyasaconfound;theyhavebecome interestingintheirownright(Box4).Indeed,theirability toallowustointeractwith ourenvironmentrapidly,yet alsoflexibly,suggeststhattheyconstituteafundamental aspectofhumancognition.

Acknowledgments

R.N.H.andA.J.H.weresupportedbyUKMedicalResearchCouncilgrant

MC_US_A060_5PR10.D.E.wassupportedbySwissNationalFoundation

Fellowship PA001–113106/1. F.H. was supported by French Agence

Nationale de la Recherche grant ANR-09-BLAN-0318. C.F. was

sup-portedbyGermanResearchFoundationgrantDFGFR2133/1-2.

References

1Roediger, H.L., III et al. (1989) Explaining dissociations between implicitandexplicitmeasuresofretention:aprocessingaccount.In Varieties of Memoryand Consciousness:Essays in Honorof Endel Tulving(Roediger,H.L.,III,Craik,F.I.M.,eds),pp.67–84,Erlbaum

2Logan, G.D. (1990) Repetition priming and automaticity: common underlyingmechanisms?Cogn.Psychol.22,1–35

3Rothermund,K.etal.(2005)Retrievalofincidentalstimulus–response associationsasasourceofnegativepriming.J.Exp.Psychol.Learn.31, 482–495

4Damian, M.F. (2001) Congruity effects evoked by subliminally presentedprimes: automaticityratherthansemanticprocessing.J. Exp.Psychol.Hum.Percept.Perform.27,154–165

5Tipper, S.P. (2001) Does negative priming reflect inhibitory mechanisms? A review and integration of conflicting views. Q. J. Exp.Psychol.(Colchester)54A,321–343

6Wu¨hr,P.andFrings,C.(2008)Acaseforinhibition:visualattention suppressestheprocessingofirrelevantobjects.J.Exp.Psychol.Gen. 137,116–130

7Marcel, A.L. (1983) Conscious and unconscious perception: experiments on visual masking and word recognition. Cognit. Psychol.15,197–237

8Eimer,M.etal.(2002)Locusofinhibitioninthemaskedprimingof responsealternatives.J.Mot.Behav.34,3–10

9Eder, A.B. et al. (2012) The structure of affective action representations: temporal binding of affective response codes. Psychol.Res.76,111–118

10 Prinz,W.(1997)Perceptionandactionplanning.Eur.J.Cogn.Psychol. 9,129–154

11 Mayr,S.etal. (2011)Evidenceof vocaland manualevent filesin auditorynegativepriming.Exp.Psychol.58,353–360

12 Horner,A.J.andHenson,R.N.(2009)Bindingsbetweenstimuliand multiple response codes dominate long-lag repetition priming in speededclassificationtasks.J.Exp.Psychol.Learn.35,757–779 Box4.Outstandingquestions

How do S–R bindings and component processes interact; for

example, to affect decision processes that select the final

behavioralresponse?

How areS–R bindings structured?For example, doesa single

bindingcontainmultiplestimulusandresponserepresentations

(asinFigure2inmaintext)ordoeseachstimulusandresponse

representationformaseparate‘binary’S–Rbinding?Underwhat

conditionsdocomplexversusspecificS–Rbindings form?Are

simpleS–Rbindingsformedfromonelearningepisode,whereas

more complex S–R bindings require more, and more varied,

learningepisodes?

HowareS–Rbindingsretrieved?Iftheycontainmultiplestimulus

andresponserepresentations(asinFigure2 inmaintext),howdo

matching and mismatching stimulusrepresentations affect the

probabilityofretrievingabinding?Iftherearemultiple,binaryS–

R bindings, how do they compete for retrieval – is a single

winning binding retrieved or domultiple bindings feed intoa

decisionprocess?

WhatarethelimitsofstimulusandresponserepresentationsinS–

Rbindings?Forexample,areindividualstimulusfeaturesbound

toresponses?Howarecontextsliketasksetrepresented:aspart

of the same binding or as some kind of index that selects

(activates)thoseS–Rbindingsthatarecurrentlyrelevant?

Do different types of S–R binding have different lifetimes

(potentiallyaccountingfordifferencesacrossrepetition,negative,

and masked priming paradigms) and how long do they last

relativetofacilitationofcomponentprocesses?

Howexactlydoattentionandawarenessmodulatetheencoding

andretrievalofS–Rbindings?

How many previous claims for component processes (e.g., in

fMRI)reflectS–Rbindingsinstead?

Which brain regions enable S–Rbindings and by what neural

mechanismsdotheyinteractwithcomponentprocesses?

13 Dennis, I.and Perfect,T.J.(2013)Dostimulus–actionassociations contributetorepetitionpriming?J.Exp.Psychol.Learn.39,85–95

14 Koch,I.andAllport,A.(2006)Cue-basedpreparationand stimulus-basedprimingoftasksintaskswitching.Mem.Cognit.34,433–444

15 Race, E.A. et al. (2009) Neural priming in human frontal cortex: multiple forms of learning reduce demands on the prefrontal executivesystem.J.Cogn.Neurosci.21,1766–1781

16 Race,E.A.etal.(2010)Multipleformsoflearningyieldtemporally distinctelectrophysiologicalrepetitioneffects.Cereb.Cortex20,1726– 1738

17 Waszak,F.andHommel,B.(2007)Thecostsandbenefitsofcross-task priming.Mem.Cognit.35,1175–1186

18 Wylie,G.andAllport,A.(2000)Taskswitchingandthemeasurement of‘‘switchcosts’’.Psychol.Res.63,212–233

19 Denkinger,B.andKoutstaal,W.(2009)Perceive–decide–act,perceive– decide–act: howabstract isrepetition-related decision learning? J. Exp.Psychol.Learn.35,742–756

20 Moutsopoulou,K.andWaszak,F.(2012)Across-taskprimingrevisited: responseandtaskconflictsdisentangledusingex-Gaussiandistribution analysis.J.Exp.Psychol.Hum.Percept.Perform.38,367–374

21 Moutsopoulou,K.andWaszak,F.(2013)Durabilityofclassification and action learning: differences revealed using ex-Gaussian distributionanalysis.Exp.BrainRes.226,373–382

22 Verbruggen, F. and Logan, G.D. (2008) Long-term aftereffects of response inhibition: memory retrieval, task goals, and cognitive control.J.Exp.Psychol.Hum.Percept.Perform.34,1229–1235

23 Giesen,C.andRothermund,K.(2013)Youbetterstop!Binding‘‘stop’’ tagstoirrelevantstimulusfeatures.Q.J.Exp.Psychol.67,1–24

24 Crump,M.J.C.etal.(2008)Context-specificlearningandcontrol:the rolesofawareness,taskrelevance,andrelativesalience.Conscious. Cogn.17,22–36

25 Vriezen,E.R.etal.(1995)Primingeffectsinsemanticclassification tasks.J.Exp.Psychol.Learn.21,933–946

26 Horner,A.J. and Henson, R.N. (2011)Stimulus–response bindings codebothabstractand specificrepresentationsofstimuli:evidence fromaclassificationprimingdesignthatreversesmultiplelevelsof responserepresentation.Mem.Cognit.39,1457–1471

27 Schnyer,D.M.etal.(2007)Itemtodecisionmappinginrapidresponse learning.Mem.Cognit.35,1472–1482

28 Frings,C.etal.(2013)Retrievalofevent filescanbeconceptually mediated.Atten.Percept.Psychophys.75,700–709

29 Perry,J.R.andLupker,S.J.(2012)Aninvestigationofthetimecourse of category congruence and priming distance effects in number classificationtasks.Can.J.Exp.Psychol.66,193–203

30 Waszak, F. et al. (2004)Semantic generalization of stimulus–task bindings.Psychon.Bull.Rev.11,1027–1033

31 Hommel,B.(1998)Eventfiles:evidenceforautomaticintegrationof stimulus–responseepisodes.Vis.Cogn.5,183–216

32 Quinn,W.M.andKinoshita,S.(2007)Congruenceeffectinsemantic categorizationwithmaskedprimeswithnarrowandbroadcategories. J.Mem.Lang.58,286–306

33 Abrams,R.L.andGreenwald,A.G.(2000)Partsoutweighthewhole (word)inunconsiousanalysisofmeaning.Psychol.Sci.11,118–124

34 Abrams,R.L.(2008)Influenceofcategorysizeandtargetsetsizeon unconsciousprimingbynovelwords.Exp.Psychol.55,189–194

35 Pohl, C. et al. (2010) Early and late selection in unconscious informationprocessing.J.Exp.Psychol.Hum.Percept.Perform.36, 268–285

36 Frings,C.andRothermund,K.(2011)Tobeornottobe.includedinan eventfile:integrationandretrievalofdistractorsinstimulus–response episodesisinfluencedbyperceptualgrouping.J.Exp.Psychol.Learn. 37,1209–1227

37 Dew,I.T.Z.andGiovanello,K.S.(2010)Thestatusofrapidresponse learninginaging.Psychol.Aging25,898–910

38 Waszak,F.etal.(2003)Taskswitchingandlong-termpriming:roleof episodicstimulus–taskbindingsintask-shiftcosts.Cogn.Psychol.46, 361–433

39 Kiesel,A.etal.(2007)UnconsciousprimingaccordingtomultipleS–R rules.Cognition104,89–105

40 Waszak,F.etal.(2013)Top-downversusbottom-up:wheninstructions overcomeautomaticretrieval.Psychol.Res.77,611–617

41 Allport,A.etal.(1994)Shiftingintentionalset–exploringthedynamic controloftasks.In AttentionandPerformance XV–Consciousand

NonconsciousInformationProcessing(Umilta,C.andMoscovitch,M., eds),pp.421–452,MITPress

42 Lhermitte,F.(1983)Utilizationbehavioranditsrelationtolesionsof thefrontallobes.Brain106,237–255

43 Norman,D.A.andShallice,T.(1986)Attentiontoaction:willedand automaticcontrolofbehaviour.InConsciousnessandSelf-regulation (Davidson,R.J.etal.,eds),pp.1–18,Plenum

44 Brass,M.etal.(2009)Neuralcorrelatesofovercominginterference frominstructedandimplementedstimulus–responseassociations.J. Neurosci.29,1766–1772

45 Cohen-Kdoshay, O. and Meiran, N. (2007) The representation of instructions in working memory leads to autonomous response activation:evidencefromthefirsttrialsintheflankerparadigm.Q. J.Exp.Psychol.60,1140–1154

46 Waszak,F.etal.(2008)Cross-talkofinstructedandappliedarbitrary visuomotormappings.ActaPsychol.(Amst.)127,30–35

47 Ruge,H.andWolfensteller,U.(2010)Rapidformationofpragmatic rule representations in the human brain during instruction-based learning.Cereb.Cortex20,1656–1667

48 Cole,M.W.etal.(2010)Prefrontaldynamicsunderlyingrapidinstructed tasklearningreversewithpractice.J.Neurosci.30,14245–14254

49 Wenke,D.etal.(2007)Instruction-inducedfeaturebinding.Psychol. Res.71,92–106

50 Frings, C. et al. (2007) Distractor repetitions retrieve previous responsestotargets.Q.J.Exp.Psychol.60,1367–1377

51 Hommel,B.(2005)Howmuchattentiondoesaneventfileneed?J.Exp. Psychol.Hum.Percept.Perform.31,1067–1082

52 Hommel,B.andColzato,L.(2004)Visualattentionandthetemporal dynamicsoffeatureintegration.Vis.Cogn.11,483–521

53 Hommel,B.(2007)Featureintegrationacrossperceptionandaction: eventfilesaffectresponsechoice.Psychol.Res.71,42–63

54 Ihrke, M. et al. (2011) Response-retrieval and negative priming encoding-andretrieval-specificeffects.Exp.Psychol.58,154–161

55 Moeller, B. and Frings, C. (2014) Attention meets binding: only

attendeddistractorsareusedfortheretrievalofevent files.Atten.

Percept.Psychophys. http://dx.doi.org/10.3758/s13414-014-0648-9

56 Naccache,L.etal.(2002)Unconsciousmaskedprimingdependson temporalattention.Psychol.Sci.13,416–424

57 Schubert,T.etal.(2013)Thetimecourseoftemporalattentioneffects on nonconscious prime processing. Atten. Percept. Psychophys. 75, 1667–1686

58 Eckstein, D. and Henson, R.N. (2012) Stimulus/response learning masked congruency priming of faces: evidence for covert mental categorisations?Q.J.Exp.Psychol.65,92–120

59 VandenBussche,E.etal.(2009)Maskedprimescanbegenuinely semanticallyprocessed:apictureprimestudy.Exp.Psychol.56,295–300

60 Naccache,L.andDehaene,S.(2001)Unconscioussemanticpriming extendstonovelunseenstimuli.Cognition80,223–237

61 Finkbeiner, M. and Friedman, J. (2011) The flexibility of nonconsciouslydeployed cognitiveprocesses: evidencefrommasked congruencepriming.PLoSONE6,e17095

62 Lowe, D.(1998)Long-term positiveand negativeidentity priming: evidenceforepisodicretrieval.Mem.Cogn.26,435–443

63 Dobbins,I.G.etal.(2004)Corticalactivityreductionsduringrepetition primingcanresultfromrapidresponselearning.Nature428,316–319

64 Logan, G.D. (1988) Toward an instance theory of automatization. Psychol.Rev.95,492–527

65 Giesen,C.andRothermund,K.(2011)Affectivematchingmoderates S–Rbinding.Cogn.Emot.25,342–350

66 Waszak,F.andPholulamdeth,V.(2009)EpisodicS–Rbindingsand emotion:abouttheinfluenceofpositiveandnegativeactioneffectson stimulus–responseassociations.Exp.BrainRes.194,489–494

67 Soldan,A.etal.(2012)Primingandstimulus–response learningin perceptualclassificationtasks.Memory20,400–413

68 Mayr,S.and Buchner,A. (2006)Evidenceforepisodic retrieval of inadequate prime responses in auditorynegativepriming. J. Exp. Psychol.Hum.Percept.Perform.32,932–943

69 Wiswede,D.etal.(2013)Notallerrorsarecreatedequally:specific ERNresponsesforerrorsoriginatingfromdistractor-basedresponse retrieval.Eur.J.Neurosci.38,3496–3506

70 Frings,C.etal.(2014)Auditorydistractorprocessinginsequential

selection tasks. Psychol. Res.

http://dx.doi.org/10.1007/s00426-013-0527-3

Review

71 Giesen, C. et al. (2012) Differences in the strength of distractor inhibition donotaffect distractor–response bindings.Mem. Cognit. 40,373–387

72 Hydock,C.etal.(2013)Distinctresponsecomponentsindicatethat bindingistheprimarycauseofresponse repetitioneffects.J.Exp. Psychol.Hum.Percept.Perform.39,1598–1611

73 VandenBussche,E.etal.(2009)Mechanismsofmaskedpriming:a meta-analysis.Psychol.Bull.135,452–477

74 Logan, G.D. (2002) An instance theory of attention and memory. Psychol.Rev.109,376–400

75 Kunde, W. et al. (2003) Conscious control over the content of unconsciouscognition.Cognition88,223–242

76 Neumann,O.(1990)Directparameterspecificationandtheconceptof perception.Psychol.Res.52,207–215

77 Henson,R.N.A.andRugg,M.D.(2003)Neuralresponsesuppression, haemodynamic repetition effects, and behavioural priming. Neuropsychologia41,263–270

78 Dehaene,S.etal.(2001)Cerebralmechanismsofwordmaskingand unconsciousrepetitionpriming.Nat.Neurosci.4,752–758

79 Horner, A.J. and Henson, R.N. (2008) Priming, response learning and repetition suppression. Neuropsychologia 46, 1979–1991

80 Horner,A.J.andHenson,R.N.(2012)Incongruentabstractstimulus– response bindings result in response interference: fMRI and EEG

evidencefromvisualobjectclassificationpriming.J.Cogn.Neurosci. 24,760–773

81 Gotts,S.J.etal.(2012)Repetitionprimingandrepetitionsuppression: acaseforenhancedefficiencythroughneuralsynchronization.Cogn. Neurosci.3,250–259

82 Ghuman,A.S.etal.(2008)Theeffectsofprimingonfrontal–temporal communication.Proc.Natl.Acad.Sci.U.S.A.105,8405–8409

83 Hsu,Y.F. and Waszak, F. (2012)Stimulus-classificationtraces are dominantinresponselearning.Int.J.Psychophysiol.86,262–268

84 Wig,G.S.etal.(2005)Reductionsinneuralactivityunderliebehavioral componentsofrepetitionpriming.Nat.Neurosci.8,1228–1233

85 Squire,L.R.etal.(1993)Thestructureandorganizationofmemory. Annu.Rev.Psychol.44,453–495

86 Schnyer, D.M. et al. (2006) Rapid response learning in amnesia: delineating associative learning components in repetitionpriming. Neuropsychologia44,140–149

87 Wise,S.P.andMurray,E.A.(1999)Roleofthehippocampalsystemin conditional motor learning: mapping antecedents to action. Hippocampus9,101–117

88 Ashby,F.G.etal.(2007)Aneurobiologicaltheoryofautomaticityin perceptualcategorization.Psychol.Rev.114,632–656

89 Saggar,M.etal.(2010)Behavioral,neuroimaging,andcomputational evidenceforperceptualcachinginrepetitionpriming.BrainRes.1315, 75–91