UM AN N IEN REVIEWARTICLE published August doi - PDF document

REVIEWARTICLEpublished:19August2014doi:10.3389/fnhum.2014.00562 WhatcanindividualdifferencesrevealaboutfaceGalitYovelJeremyB.WilmerBradDuchaineSchoolofPsychologicalSciences,SagolSchoolofNeuroscience,TelAvivUniversity,TelAviv,IsraelDepartmentofPsychology,WellesleyCollege,Wellesley,MA,USADepartmentofPsychologicalandBrainSciences,DartmouthCollege,Hanover,NH,USAEditedby:MarkA.Williams,MacquarieUniversity,AustraliaReviewedby: FrontiersinHumanNeurosciencewww.frontiersin.orgAugust2014|Volume8|Article562 Yoveletal.Individualdifferencesinfaceprocessing FIGURE1|(A)Groupmeanndingsoftheclassicalholisticprocessingmeasures.Thepart-wholetaskshowsbetterrecognitionoffacepartswhenpresentedwithinthewholeface(i.e.,wholecondition)thanwhenpresentedinisolation(i.e.,partcondition).Thecompositefacetaskshowsbetterrecognitionofupperorlowerpartofthefacewhenitismisalignedthanwhenitisalignedwithananotherfaceofadifferentidentity,indicatinginteractions(whichleadstointerference)betweenthetwofacehalvesinthealignedcondition(seealsofootnote1).(B)Correlationalanalysesbetweenthetwomeasuresofholisticprocessingandtheirrelationshipwithfacerecognitionabilitiesrevealmoderatecorrelations(Degutisetal.,2013,seealsofootnote2). Therststudythatdirectlyassessedthecorrelationsbetweenthesetasks(Konaretal.,2010)measuredthecompositeeffectbysubtractingperformanceonaligned(i.e.,wholecondition)andmisaligned(i.e.,partcondition)faces.Surprisinglythismeasureofholisticprocessingfailedtocorrelatewiththeirmeasureoffacerecognitionabilities.Threesubsequentstudies,however,havefoundsignicantpositiverelationshipsbetweenfacerecognitionandboththecompositeeffect(Richleretal.,2011;Wangetal.,2012;Degutisetal.,2013)andthepart-wholeeffect(Wangetal.,2012;Degutisetal.,2013)butnotwithperformanceonanon-facegloballocaltask(Wangetal.,2012)suggestingthatthesecor-relationsarenotmediatedbygeneralvisualprocessingabilities.Twoofthesestudies(Richleretal.,2011;Degutisetal.,2013)usedthereliableandwell-validatedCambridgeFaceMemoryTest(CFMT;Wilmeretal.,2012),whichmaypartiallyaccountforthehighercorrelationstheyrevealed.Thesestudiesfurtherfoundthatthesizeofthecorrelationsdependsontheanalyticmethodusedtomeasureholisticprocessing(regressionvs.subtractionscores,Degutisetal.,2013)andthedesignofthecompositetaskused(congruency/interferencevs.standarddesign,RichlerandGautheir,2013;Rossion,2013).1 1Thecompositefaceeffecthasbeenmeasuredinthesestudieswithtwodifferentparadigms,thecongruency/interferenceparadigmorthestandard FrontiersinHumanNeurosciencewww.frontiersin.orgAugust2014|Volume8|Article562|2 Yoveletal.IndividualdifferencesinfaceprocessingTakenasawhole,thesestudiessuggestthatholisticprocessingandfacerecognitionareindeedlinked,butnotasstronglyashadbeenwidelyassumed;therefore,otherfactorsmustalsocontributetofacerecognitionabilities.Onesuchfactorwassuggestedbyarecentstudyshowingasmallbutsignicantcorrelationacrossindividualsbetweenfacerecognitionabilityandfaceaftereffects(Dennettetal.,2012).Thefaceaftereffectisusedasatooltoassessfacespacecodingoffaceidentity.Themagnitudeofthefaceaftereffectreectsthesteepnessoftheresponsefunctionalongagivendimension(i.e.,facialfeature).Dennettetal.(2012)haverevealedthatsteeperfunctionsareassociatedwithbetterfacerecognitionabilities.Futurestudiesareneededtoassesstherelativecontributionsofholisticpro-cessing,facespacecoding,andotherfactorstofacerecognitionabilities.DODIFFERENTMEASURESOFHOLISTICFACEPROCESSINGTAPTHESAMEMECHANISMS?Givenevidencethatfacesareprocessedbyholisticmechanisms,anotherbasicquestionthathasbeenoverlookedformanyyearsiswhetherdifferentmeasuresofholisticprocessingreectthesameholisticprocessingmechanism.Thepart-wholeandthecompositefaceeffecthaveoftenbeenconsideredmeasuresofthesameprocessandhavebeenusedinterchangeablyinthefacepro-cessingliterature(Richleretal.,2012).Itwasthereforepuzzlingwhenstudiesrevealedlowcorrelationsbetweenthesetwoholisticprocessingmeasures(r=0.23inDegutisetal.,2013;r=0.03inWangetal.,2012).Degutisetal.(2013)howeverfoundthatthecorrelationbetweenthetwoholisticmeasureswassubstantial(r=0.44)whenholisticprocessingscoreswerecomputedviaaregression-basedmethod(Figure1B)andwerehigherthanwhentheywerecomputedviathesubtraction-basedmethodusedingroup-meanstudies;thisndinghasgenerateddiscussionofhowtotranslatemeasuresusedingroup-meanstudiesintoaformthatcanvalidlycapturebothclinicalandnon-clinicalhumanvariation(foranextensivediscussionofthisquestionseeDegutisetal.,2013)2.Moregenerally,suchindividualdifferencesbasedstudiesforceustocriticallyexaminecommonlyusedmeasuresandbetterdeterminewhattheymeasure.AREFACEPARTSANDTHEIRSPACINGREPRESENTEDBYTHESAMEMECHANISMORDIFFERENTMECHANISMS?Astudythatexaminedindividualdifferencesindiscriminationoffacepartsandwithspacingbetweenpartsprovidesanother paradigm.AcomprehensivediscussionaboutthetwotypesofparadigmsandtheextenttowhichtheyprovideavalidmeasureofholisticprocessingarediscussedinRichlerandGautheir(2013)andRossion(2013).Briey,thestandardparadigmprovidesameasureofholisticprocessingbyassessingtheinterferenceofirrelevantdifferentfacehalvesontheprocessingoftheotherfacehalves.Thecongruency/interferencedesignprovidesageneralstroop-likemeasureinwhichcongruentandincongruenttrialsarecompared.2Regressionscoresarecomputedbyregressing,ratherthansubtracting,thepart-basedconditionfromthewholefacecondition.Inthepart-wholetask,regressionscoresarecomputedbyregressingthepartconditionfromthewholecondition.Inthecompositefacetask,regressionscoresarecomputedbyregressingthemisalignedtaskfromthealignedtask.AmoredetaileddiscussionaboutthesubtractionandregressionapproachesisfoundinDegutisetal.(2013).exampleofhowindividualdifferencesndingsmaynotonlycomplementdatafromgroup-meansstudies,butalsoaidindeningourmeasuresofinterest(YovelandKanwisher,2008).Thetermconguralprocessinghasbeenfrequentlyusedtodescribehowfacesarerepresented(Maureretal.,2002).Onewayinwhichconguralprocessinghasbeenmeasuredisbyexaminingsensitivitytothedistanceamongfaceparts(e.g.,distanceamongthetwoeyes),whichhasbeenclaimedtobecriticalforfacerecognition(LeGrandetal.,2001).However,morerecentpapershighlightedtheroleofboththespacingandtheshapeoftheirpartsinfaceprocessingandsuggestedthattheyarebothmediatedbythesamefaceprocessingmechanism(YovelandDuchaine,2006;McKoneandYovel,2009;AmishavandKimchi,2010).Anindividualdifferencesstudystronglysupportedthelatterclaim,byshowingahighcorrelationbetweendiscriminationoffacesthatdifferinspacingamongpartsandfacesthatdifferintheshapeofpartsforuprightfaces.Incontrast,thesamecorrela-tionwaseffectivelyzeroforinvertedfacesorhouses(YovelandKanwisher,2008).Thesendingshaveledtothesuggestionthatthedenitionofholistic/conguralprocessingshouldincludetheprocessingofboththeshapesofpartsandthespacingamongthem(McKoneandYovel,2009).Consistentwiththesendings,YovelandDuchaine(2006)havereportedthatprosopagnosicindividualsshowsimilarlypoordiscriminationoffacepartsandthespacingamongthem,whichsuggestsbothtypesofinfor-mationareimpairedinindividualswithpoorfacerecognitionabilities.DOFACEEXPRESSIONANDFACEIDENTITYPROCESSINGRELYONDIFFERENTMECHANISMS?Thequestionofwhetherfaceexpressionandidentityarepro-cessedbyacommonmechanismorseparatemechanismshasbeendebatedinthecognitive(GanelandGoshen-Gottstein,2004),neuropsychological(BruceandYoung,1986)andneu-roimagingliterature(CalderandYoung,2005;GobbiniandHaxby,2007).Anindividualdifferencesapproachcanaddressthisquestionbyassessingthecorrelationsamongtestsofexpres-sionandidentityprocessing.Acriticalprerequisiteforsuchanapproach,however,isreliableandvalidtestsofexpressionprocessing.InarecentstudyPalermoetal.(2011)arguedthatnoexistingexpressionprocessingtestmetthehighstandardsnec-essarytoenablesuchanapproach.Theythendevelopedtwonewtests,oneexpressionmatchingtestandoneexpressionlabelingtest.Bothtestsefcientlycapturedexpressionprocessingabilities,demonstratingstrongreliabilityandvaliditydespitetheirbrevity;moreover,thesetestsdemonstratedsuitabilityforcapturingabroadrangeofperformance,avoidingtheceilingeffectsfoundinthemajorityofexistingexpressionprocessingtests(Palermoetal.,2011).Interestingly,inan80-personsample,thevariationsharedbetweenthesetwoexpressionrecognitiontests(r=0.45)wasvirtuallyindependentofperformanceontheCFMT(DuchaineandNakayama,2006),evidenceforexpression-specicmechanisms.Atthesametime,theexpressionmatchingtestcorrelatedrobustlywiththeCFMT(r=0.40),evidenceforface-generalmechanisms.Furtherstudiesarenowneededtodeter-minewhichparticularexpressionprocessingmechanismsaresharedwith,andwhichareindependentof,identityprocessing. FrontiersinHumanNeurosciencewww.frontiersin.orgAugust2014|Volume8|Article562|3 Yoveletal.IndividualdifferencesinfaceprocessingAREFAMILIARANDUNFAMILIARFACESPROCESSEDBYDIFFERENTMECHANISMS?Facerecognitionabilitieshavebeenmeasuredbothwithfamousorpersonallyfamiliarfacesandwithunfamiliarfaces.Insev-eralstudiesBurtonetal.providedevidencethatfamiliarfacesmaybeprocessedqualitativelydifferentlyfromunfamiliarfaces(JenkinsandBurton,2011).Theythenusedanindividualdif-ferencesapproachtoprovideanindependenttestoftheirtheory,examiningthecorrelationsbetweenperformanceonseveraltasksthatexaminedmatchingofuprightandinvertedfamiliarandunfamiliarfaces(MegreyaandBurton,2006).Whereascorre-lationsbetweentasksthatmeasuredunfamiliarfacematchingabilitieswerehigh,relativelylowcorrelationswerefoundbetweentasksthatexaminedmatchingofunfamiliarandfamiliarfaces.Interestingly,thecorrelationbetweenmatchingofunfamiliaruprightfaceswashighlycorrelatedwithmatchingofinvertedfamiliarfaces.Basedonthenotionthatfaceprocessingmech-anismsarespecializedfortheprocessingofuprightbutnotinvertedfacestheauthorsinterpretthesendingsasstrongsupportfortheirtheoryofqualitativelydifferentprocessingoffamiliarvs.unfamiliarfaces,goingsofarastosuggestthat“unfamiliarfacesarenotfaces”.Thesendingsillustratehowindividualdifferencescanprovideanindependenttestofatheoryderivedfromgroup-meanstudies.Itisnoteworthythatunlikethelackofcorrelationfoundinmatchingtasks,correlationsbetweenfamousandunfamiliarfacesarefoundinamem-orytask(Wilmeretal.,2012).Furthermore,mostprosopag-nosicindividualsareimpairedonbothfamiliarandunfamiliarfacerecognitiontasks(Duchaineetal.,2007;Dalrympleetal.,2011).Futurestudiesarenowneededtodeterminewhetherthesendings,bothgroup-meanbasedandindividualdiffer-encesbased,holdacrossavarietyoffacematchingandfacememorytasks.DOCOGNITIVEANDNEURALMEASURESOFFACEPROCESSINGREFLECTTHESAMEMECHANISMS?FacesareknowntoelicitrobustanddistinctneuralresponseswithbothfunctionalMRIandelectrophysiologicalmeasures(Figure2).Tobetterunderstandwhattypeofprocessingtheseneuralmeasuresreect,itisimportanttodeterminetowhatextenttheyareassociatedwithcognitivemeasuresoffacepro-cessingaswellastheextenttowhichdifferentneuralmeasuresarecorrelatedamongthemselves.Oneofthemostwell-establishedndingsinthefaceprocess-ingliteratureisthefaceinversioneffect—thatisthesubstantialdropinperformancefoundforinvertedrelativetouprightfaces(Figure2A,Yin,1969).Adifferencebetweenthegroupmeanresponsetouprightandinvertedfaceswasfoundintwofaceareas,thefusiformfacearea(FFA)andthesuperiortemporalsulcusfacearea(STS-FA)responsewashigherforuprightthaninvertedfaces,whereasinthelateraloccipitalcomplex(LOC)objectareatheresponsewashigherforinvertedthanuprightfaces(YovelandKanwisher,2005).However,acorrelationbetweenthebehavioralandfMRImeasuresofthefaceinversioneffectwasfoundonlywiththeFFA(Figure2D).ThesendingssuggesttheFFAasaneurallocusofthefaceinversioneffectandhighlighttheimportanceofassessingcorrelationsaswellasdifferencesinmeanresponses,becausegroupmeansmaybeconsistentwiththebehavioraleffectbutnotassociatedwithit.Therelationshipbetweencognitiveandneuralmeasuresoffaceprocessinghasbeenalsoexaminedinastudythatexamineddifferentcognitivemeasuresoffaceperceptionandmemoryandvariousface-relatedevent-relatedpotential(ERP)components(Herzmannetal.,2010).Thisstudyrevealedmoderatecorre-lationsbetweenacognitivemeasureoffaceprocessing(i.e.,acombinedperformancescoreonvariousperceptionandmemorytasks)andthelatencyoftheN170,anERPcomponentthatismuchstrongertofacesthanotherstimuli(Figure2B),butnocorrelationwithanearliercomponent,theP100.ModeratecorrelationswerealsofoundwithlaterERPcomponentsrelatedtofacememoryorpersonrecognition.SuchstudiesareimportantindeterminingwhichaspectsoffaceprocessingarereectedbydifferentERPcomponentsandprovideconvergingevidencetothemajorityofERPstudiesthatemploythemorecom-mongroup-basedanalysisapproach(e.g.,Schweinbergeretal.,2004).DOEEGANDfMRIMEASURESOFFACEPROCESSINGREFLECTTHESAMEMECHANISMS?Face-selectiveneuralresponses(i.e.,highergroup-meanresponsetofacesthannon-faces)havebeenreportedinhundredsoffMRIandEEGstudies.However,onlyonestudyhasexaminedthecorrelationacrossindividualsbetweentheEEGandfMRIface-selectivemeasures.Thisstudyrevealedthatthemagnitudeofface-selectivityinboththeFFAandtheSTS-FAwereassociatedwiththeface-selectivityoftheEEGresponseapproximately170msafterstimulusonset(N170)(Sadehetal.,2010;Figure2E).Theface-selectivityoftheoccipitalfacearea(OFA)wasnotcorrelatedwiththeface-selectivityoftheN170butwascorrelatedwithERPface-selectivityat100–110msafterstimulusonset,consistentwithtranscranialmagneticstimulationstudiesthatvariedpulsetiming(Pitcheretal.,2007,2012).ThesestudiesnicelydemonstratehowcorrelationalanalysisofEEGandfMRIcanrevealtemporaldissociationsamongdifferentbrainregionsandlinkdifferentbrainareastothetimecourseofdifferentstagesoffaceprocessing.Importantly,thesecorrelationsextendgroup-meansndingsbyshowingwhichoftheseneuralface-selectivemeasures,whichhavebeenprimarilystudiedseparately,arestronglylinkedandthereforereectthesameunderlyingneuralmechanismsoffaceprocessing.AsimilarapproachhasbeenusedtoinvestigatethefaceinversioneffectpresentinERPandfMRImeasures.Themid-temporalface-selectiveareas,theFFAandSTS-FAshowahigherresponsetouprightthaninvertedfaces.Incontrast,objectgeneralareas(LOC)showahigherresponsetoinvertedthanuprightfaces(YovelandKanwisher,2005).TheN170showsincreasedandslightlydelayedamplitudetoinvertedthanuprightfaces.TwomechanismshavebeensuggestedtoaccountfortheincreasedN170amplitudetoinvertedthanuprightfaces.Accordingtothequalitativehypothesisincreasedamplitudeforinvertedfacesreectstherecruitmentofadditionalnon-facemechanismsthatarenotusedfortheprocessingofuprightfaces.Thus,theincreasedresponsetoinvertedfacesintheobjectareamaycon-tributetotheincreasedN170amplitudetoinvertedfaces.In FrontiersinHumanNeurosciencewww.frontiersin.orgAugust2014|Volume8|Article562|4 Yoveletal.Individualdifferencesinfaceprocessing FIGURE2|(A)Thefaceinversioneffectreferstothebetterrecognitionoffacespresentedinuprightorientationthaninverted(i.e.,upsidedown).(B)ThefaceN170:event-relatedpotentials(ERPs)showhigheramplitudetofacesthannon-facestimuli(i.e.,face-selectivity)at170msafterstimulusonset.(C)FunctionalMRIstudiesrevealthreeface-selective(i.e.,higherresponsetofacesthannon-faceobjects)regionsintheoccipital-temporalcortex:theoccipitalfacearea(OFA)inthelateraloccipitalcortex,thefusiformfacearea(FFA)inthefusiformgyrusandthesuperiortemporalsulcusfacearea(STS-FA)intheposteriorpartoftheSTS.(D)Theassociationbetweenthebehavioralfaceinversioneffect(i.e.,differenceinperformancelevelforuprightthaninvertedfaces)andthefMRIfaceinversioneffect(i.e.,differenceinfMRIresponsetouprightthaninvertedfaces)wasfoundonlywiththeFFA(YovelandKanwisher,2005).(E)Correlationsamongthemagnitudeoffaceselectivity(i.e.,differenceinERPorfMRIresponsetofacesthannon-faces)wasfoundat170mswiththeFFAandSTS-FA,butnotwiththeOFA(Sadehetal.,2010). FrontiersinHumanNeurosciencewww.frontiersin.orgAugust2014|Volume8|Article562|5 Yoveletal.Individualdifferencesinfaceprocessingcontrast,thequantitativehypothesissuggeststhesameprocessesgeneratetheN170responsetouprightandinvertedfacesbutthattheincreasedamplitudeforinvertedfacesreectsthegreaterdemandsthatinvertedfaceprocessingplacesonfacemechanisms.Todirectlytestthesetwohypothesis,theN170andfMRIfaceinversioneffectsweremeasuredinasimultaneousEEG-fMRIstudy.TheN170faceinversioneffectwascalculatedforeachsubjectasthenormalizeddifferencebetweentheresponsetouprightandinvertedfaces(Sadehetal.,2011).Inaddition,face-selectiveandobjectgeneralareaswerelocalizedandthedifferenceintheirresponsetouprightandinvertedfaceswasmeasured.AcorrelationalanalysisbetweenthefMR-faceinversioneffect(i.e.,thedifferencebetweentheresponsetoinvertedthanuprightfaces)intheobjectandface-selectiveareasandtheN170faceinversioneffectrevealedaverystrongcorrelationwiththeobjectareas(r=0.8)butnotwiththefaceareas.Thesendingsfurthersupportthequalitativehypothesis,whichsuggeststhatinvertedfacesengageobjectmechanismsthatarenotusedfortheprocess-ingofuprightfaces(seealso,Moscovitchetal.,1997;Pitcheretal.,2011).ThesesimultaneousfMRI-EEGstudiesnicelydemonstratehowcombiningthetwomethodscanprovideinsightintothetemporalcharacteristicsofbrainareasandthepossibleneuralgeneratorsofERPsignals.Inparticular,thecorrelationsbetweentheface-selectivemeasuresindicateanearlierlatencyfortheOFAthanforthemid-temporalfaceareas.ThefaceinversioneffectstudyattributedtheincreasedamplitudeoftheN170toinvertedfaces,toobjectareasratherthantothenearbyface-area,andingthatcannotbeobtainedfromsourcelocalizationanalysisofEEGdataalone.ThesendingsthereforedonotonlyfurtherestablishthelinkbetweentheERPandfMRIfacemarkersbutalsoenhanceourunderstandingofthespatial-temporalarchitectureofthefacesystem.HOWSPECIFICISFACERECOGNITIONABILITY?Inthesectionsabove,wehaveexploredhowindividualdifferencescanlinkanddissociatemechanismswithinthedomainoffaceprocessing.Individualdifferencescanalsoreveallinksanddis-sociationsbetweenfaceprocessingandothercognitiveabilities.Anactivelineofresearchhasrecentlyrevealedthatfacerecog-nitionisanuncommonlyspecicability.Inthepsychometricliterature,thetermspecicistypicallyappliedtoanabilitythatshowssomedegreeofindependencefromgeneralintelligence(Waietal.,2009).Bythisdenition,facerecognitionisexceptionallyspecic.Todate,itsmeanreportedcorrelationwithmeasuresofgeneralintelligence,weightedforsamplesizeandcorrectedforrangerestrictionintheIQmeasures,is0.01(Davisetal.,2011;PetersonandMiller,2012;Palermoetal.,2013). FIGURE3|Specicityoffacerecognitionability(Wilmeretal.,2012).Facerecognitionperformance(xaxis,CambridgeFaceMemoryTest(CFMT))isplottedagainstfamousfacerecognitionperformance(yaxis,graphA,FamousFacesMemoryTest(FFMT))andverbalrecognitionperformance(yaxis,graphB,VerbalPaired-associatesMemoryTest(VPMT)).FFMTandCFMTareverydifferenttests.FFMTmeasurestheabilitytonamefacesstoredincidentallyinmemoryovermonthsoryearsofculturalexposure.CFMT,ontheotherhand,measurestheabilitytoidentifyfacesstoredintentionallyinmemoryshortlybeforebeingtested.CFMTandFFMTneverthelessshowahighcorrelation,demonstratingthatCFMTcapturesageneralfacerecognitioncapacity.CFMTdissociatesstrongly,however,fromVPMT,whichmeasurestheabilitytoidentifyword-pairsstoredintentionallyinmemoryshortlybeforebeingtested.ThisdissociationdemonstratesthatCFMTcapturesaspecicrecognitioncapacity. FrontiersinHumanNeurosciencewww.frontiersin.orgAugust2014|Volume8|Article562|6 Yoveletal.IndividualdifferencesinfaceprocessingMoreover,facerecognitiondissociatesstronglyevenfromothertypesofrecognitionmemory.Forexample,indiversesamplestotalingover4000participants,well-validatedtestsofverbalandnon-facevisualrecognitionability,respectively,explainedonlyabout3%and7%ofthevarianceinfacerecognitionability(Wechsler,1997;Wilmeretal.,2010,2012;Figure3).Thesendingsareconsistentwithseveralreportsthatshowthatneuropsychologicalcasessometimesshowselectiveimpairmentorsparingoffacerecognition(Moscovitchetal.,1997;Duchaineetal.,2006;Rezlescuetal.,2012;Busignyetal.,2014).Ironically,thehistoryofpsychometricabilitytestinghasseenfacerecognitionabilitydroppedatleasttwicefromprominenttestdevelopmenteffortswhenitspervasivedissociationsfromothersocialandmemoryabilitiesweremistakenforlackofvalidmeasurement(Thorndike,1936;KihlstromandCantor,2000;HoldnackandDelis,2004;Wilmeretal.,2012).Onlyinrecentyearshasitbeendiscoveredthatfacerecognition'sdissociationsfromotherabilitiesreectavalid,uniquedimensionofhumanability(Wilhelmetal.,2010;Wilmeretal.,2010,2012;Hilde-brandtetal.,2011;McGuginetal.,2012;Figure3).Guidedbytheexampleoffacerecognition,wesuggestthatanindividualdifferencesapproachbeusedtofurtherdenethecognitiveandneuralboundariesoffaceprocessing,aswellastosearchforotheruniquesocialandcognitiveabilitydimensions.HOWISFACERECOGNITIONABILITYSHAPEDBYGENESANDENVIRONMENT?Individualdifferencesinfaceprocessingabilitiesprovideapowerfulvehicleforexploringthecontributionsofgenesandenvironmentstofaceprocessingviatwinandfamilystudies.Arecenttwinstudyshowedthatfacerecognitioninadultsishighlyheritable(Wilmeretal.,2010).CorrelationsbetweenidenticaltwinsontheCFMT(0.70)weretwicethoseoffraternaltwins(0.29),evidencethatthestrongfamilyresemblanceforfacerecog-nitionabilityresultedfromgeneticfactorsratherthancommonenvironmentalfactors(Figure4).Thecombinationoffacerecognition'suncommonspecicityandhighheritabilityrunscountertoaclassicndinginbehavioralgeneticsthatmorespecicabilitiesarelessheritable(Plominetal.,2013;seesectionaboveonfacerecognition'sspecicity).Thatclassicndinginspiredaprominenttheorythatthemajorityofgeneticvarianceinanycognitiveabilityisattributabletogeneralintelligence(KovasandPlomin,2006).Facerecognitionpresentsaclearexceptiontothattheory.Further,facerecog-nition'sheritabilitysuggeststhatitcouldbeusedasamodelsystemtostudycognitiveandneuralresiliencetoenvironmentalvariation.Despiteitsstrongdissociationfromgeneralintelli-gence,facerecognitionmaybesimilartogeneralintelligenceinshowingincreasingheritabilitywithage(Wilmeretal.,2010;Zhuetal.,2010).Ifso,thenincreasingheritabilitymaybeamoregeneralizedprincipleofdevelopmentthanpreviouslyrecognized.Futureworkisneededtodeterminewhetheradultfacerecog-nitioncanbeparsedintoheritablesubcomponents.Onetwinstudyfoundanon-zerogeneticcontributiontothecompositefaceeffect,butnottothepart-wholeeffect,suggestingarelativelyconstrainedroleofholisticfaceprocessingmechanismsinfacerecognition'sheritability(Zhuetal.,2010).Futureworkcould FIGURE4|Heritabilityoffacerecognitionability(Wilmeretal.,2010).Inthisstudyoftwins,thesecond-borntwin'sCFMTscore(yaxis)isplottedagainsttherst-borntwin'sscore(xaxis)formonozygotic(MZ,n=164)twins(A)anddizygotic(DZ,n=125)twins(B).MZcorrelationisrMZ(162)=0.70,andDZcorrelationisrDZ(123)=0.29.ThehighrMZindicateshighfamilyresemblanceforCFMTperformance,whereastherDZoflessthanhalftherMZindicatesthatmostorallofthisfamilyresemblancecanbeattributedtofamilygenesratherthanfamilyenvironments. FrontiersinHumanNeurosciencewww.frontiersin.orgAugust2014|Volume8|Article562|7 Yoveletal.Individualdifferencesinfaceprocessingalsoexplorethespecicgeneticandenvironmentalmechanismsbywhichabroadnaturaltendencyforrelativelygoodorpoorfacerecognitionabilityisexpressed.Aricherunderstandingofsuchmechanismsmightinspirenovelinterventionstoenhance,oraccommodationstosupport,theimportantsocialtaskofrec-ognizingothersinoureverydaylives.Moreover,amoredetailedunderstandingofthereasonsforfacerecognition'shighdegreeofresiliencetoenvironmentalvariationmightfueleffortstomaximizeneuralandcognitiveresilienceinotherdomainsaswell.CONCLUSIONThisreviewdemonstrateshowassessmentofassociationsanddissociationsamongmeasuresoffaceprocessingbyanindividualdifferencesapproachcanprovideanswerstobasicquestionsaboutfaceprocessingmechanisms.Thequestionstackledbytheexamplesinourreviewaddressthenatureofvariousfaceprocessingmechanisms,theirrelationshiptoeachother,andtheirrelationshiptobroaderaspectsofcognition.Manyquestionsstillawaitsuchinvestigations.Thesequestionsinclude:whatassociationsanddissociationsexistbetweenadditionalfaceprocessingmechanisms,includingthoseusedtogleanage,gender,andattractiveness?Whatarethedetailedneuralandgeneticmechanismsofeachaspectoffaceprocessing?Whatplasticityexists,atwhatages,andwhatarethepracticalcorrelatesgoodorbadatfaceprocessing?Howdoaspectsoffaceprocessingbeyondfacerecognitionrelatetoabroaderarrayofhumancapacities?Mostexistingworkonindividualdifferencesinfaceprocessinghasaimedtoisolatebroadpatternsofassociationanddissoci-ationamongabilities,orbetweenabilitiesandtheirunderlyingmechanisms.Muchworkremainstobedoneatthisrelativelycoarselevelofanalysis.Atthesametime,thereisaneedtobegindiggingdeeper,makingincreasinglyne-grainedtheo-reticaldistinctionsaboutthespecicneural,cognitive,genetic,andenvironmentalmechanismsthatshapeandconstitutesuchbroadassociationsanddissociations.Fine-grainedworkofthissortisbothpromisingandchallenging;itrequires(a)agreaternumberofhigh-qualitytests;(b)morehighlymultivariatestatis-ticalmodels;and(c)alargernumberofparticipant-hours.Eachsuchrequirementcomeswithitsowncostsandcomplications,however,allcanbeovercomeforne-grainedquestionsofsuf-cienttheoreticalorpracticalimport.Asthisreviewindicates,correlationalanalysesnotonlyexpandourmethodologicalandstatisticalarmorybutalsoforcecon-siderationofthetheoreticalmeaningofourmeasuresinawaythatagroup-meanapproachmaynot.Theindividualdifferencesapproachcanthereforeprovidevaluableinformationthatcom-plementsandextendstheinferencessupportedbythecommonlyusedgroup-meanapproach.Weanticipatethisapproachwillbeasfruitfulinotherdomainsofcognitivescienceasithasbeen,andwilllikelycontinuetobe,inthestudyoffaceprocessing.REFERENCESAmishav,R.,andKimchi,R.(2010).Perceptualintegralityofcomponentialandconguralinformationinfaces.Psychon.Bull.Rev.17,743–748.doi:10.3758/pbr.17.5.743Avidan,G.,Tanzer,M.,andBehrmann,M.(2011).Impairedholisticprocessingincongenitalprosopagnosia.Neuropsychologia49,2541–2552.doi:10.1016/j.neuropsychologia.2011.05.002Bruce,V.,andYoung,A.(1986).Understandingfacerecognition.Br.J.Psychol.77,305–327.doi:10.1111/j.2044-8295.1986.tb02199.xBusigny,T.,VanBelle,G.,Jemel,B.,Hosein,A.,Joubert,S.,andRossion,B.(2014).Face-specicimpairmentinholisticperceptionfollowingfocallesionoftherightanteriortemporallobe.Neuropsychologia56,312–333.doi:10.1016/j.neuropsychologia.2014.01.018Calder,A.J.,andYoung,A.W.(2005).Understandingtherecognitionoffacialiden-tityandfacialexpression.Nat.Rev.Neurosci.6,641–651.doi:10.1038/nrn1724Dalrymple,K.A.,Oruc,I.,Duchaine,B.,Pancaroglu,R.,Fox,C.J.,Iaria,G.,etal.(2011).Theanatomicbasisoftherightface-selectiveN170INacquiredprosopagnosia:acombinedERP/fMRIstudy.Neuropsychologia49,2553–2563.doi:10.1016/j.neuropsychologia.2011.05.003Davis,J.M.,McKone,E.,Dennett,H.,O'Connor,K.B.,O'Kearney,R.,andPalermo,R.(2011).Individualdifferencesintheabilitytorecognisefacialiden-tityareassociatedwithsocialanxiety.PLoSOne6:e28800.doi:10.1371/journal.pone.0028800Degutis,J.,Mercado,R.J.,Wilmer,J.,andRosenblatt,A.(2013).Individualdifferencesinholisticprocessingpredicttheown-raceadvantageinrecognitionmemory.PLoSOne8:e58253.doi:10.1371/journal.pone.0058253Dennett,H.W.,McKone,E.,Edwards,M.,andSusilo,T.(2012).Faceafteref-fectspredictindividualdifferencesinfacerecognitionability.Psychol.Sci.23,1279–1287.doi:10.1177/0956797612446350Duchaine,B.,andNakayama,K.(2006).Thecambridgefacememorytest:resultsforneurologicallyintactindividualsandaninvestigationofitsvalidityusinginvertedfacestimuliandprosopagnosicparticipants.Neuropsychologia44,576–585.doi:10.1016/j.neuropsychologia.2005.07.001Duchaine,B.C.,Yovel,G.,Butterworth,E.J.,andNakayama,K.(2006).Prosopag-nosiaasanimpairmenttoface-specicmechanisms:eliminationofthealter-nativehypothesesinadevelopmentalcase.Cogn.Neuropsychol.23,714–747.doi:10.1080/02643290500441296Duchaine,B.,Yovel,G.,andNakayama,K.(2007).Noglobalprocessingdecitinthenavontaskin14developmentalprosopagnosics.Soc.Cogn.Affect.Neurosci.2,104–113.doi:10.1093/scan/nsm003Ganel,T.,andGoshen-Gottstein,Y.(2004).Effectsoffamiliarityontheperceptualintegralityoftheidentityandexpressionoffaces:theparallel-routehypothesisrevisited.J.Exp.Psychol.Hum.Percept.Perform.30,583–597.doi:10.1037/0096-1523.30.3.583Gobbini,M.I.,andHaxby,J.V.(2007).Neuralsystemsforrecognitionoffamiliarfaces.Neuropsychologia45,32–41.doi:10.1016/j.neuropsychologia.2006.04.015Herzmann,G.,Kunina,O.,Sommer,W.,andWilhelm,O.(2010).Individualdifferencesinfacecognition:brain-behaviorrelationships.J.Cogn.Neurosci.22,571–589.doi:10.1162/jocn.2009.21249Hildebrandt,A.,Wilhelm,O.,Schmiedek,F.,Herzmann,G.,andSommer,W.(2011).Onthespecicityoffacecognitioncomparedwithgeneralcognitivefunctioningacrossadultage.Psychol.Aging26,701–715.doi:10.1037/a0023056Holdnack,J.A.,andDelis,D.C.(2004).Parsingtherecognitionmemorycom-ponentsoftheWMS-IIIfacememorysubtest:normativedataandclinicalndingsindementiagroups.J.Clin.Exp.Neuropsychol.26,459–483.doi:10.1080/13803390490496687Jenkins,R.,andBurton,A.M.(2011).Stablefacerepresentations.Philos.Trans.R.Soc.Lond.BBiol.Sci.366,1671–1683.doi:10.1098/rstb.2010.0379Kihlstrom,J.F.,andCantor,N.(2000).Socialintelligence.Handb.Intell.2,359–379.doi:10.1017/CBO9780511807947.017Konar,Y.,Bennett,P.J.,andSekuler,A.B.(2010).Holisticprocessingisnotcorrelatedwithface-identicationaccuracy.Psychol.Sci.21,38–43.doi:10.1177/0956797609356508Kovas,Y.,andPlomin,R.(2006).Generalistgenes:implicationsforthecognitivesciences.TrendsCogn.Sci.10,198–203.doi:10.1016/j.tics.2006.03.001LeGrand,R.,Cooper,P.A.,Mondloch,C.J.,Lewis,T.L.,Sagiv,N.,deGelder,B.,etal.(2006).Whataspectsoffaceprocessingareimpairedindevelopmentalprosopagnosia?BrainCogn.61,139–158.doi:10.1016/j.bandc.2005.11.005LeGrand,R.,Mondloch,C.J.,Maurer,D.,andBrent,H.P.(2001).Earlyvisualexperienceandfaceprocessing.Nature412,786.doi:10.1038/35090636LeGrand,R.,Mondloch,C.J.,Maurer,D.,andBrent,H.P.(2004).Impairmentinholisticfaceprocessingfollowingearlyvisualdeprivation.Psychol.Sci.15,762–768.doi:10.1111/j.0956-7976.2004.00753.xMaurer,D.,LeGrand,R.,andMondloch,C.J.(2002).Themanyfacesofcon-guralprocessing.TrendsCogn.Sci.6,255–260.doi:10.1016/s1364-6613(02)01903-4 FrontiersinHumanNeurosciencewww.frontiersin.orgAugust2014|Volume8|Article562|8 Yoveletal.IndividualdifferencesinfaceprocessingMcGugin,R.W.,Richler,J.J.,Herzmann,G.,Speegle,M.,andGauthier,I.(2012).Thevanderbiltexpertisetestrevealsdomain-generalanddomain-specicsexeffectsinobjectrecognition.VisionRes.69,10–22.doi:10.1016/j.visres.2012.07.014McKone,E.,andYovel,G.(2009).Whydoespicture-planeinversionsometimesdissociateperceptionoffeaturesandspacinginfacesandsometimesnot?Towardanewtheoryofholisticprocessing.Psychon.Bull.Rev.16,778–797.doi:10.3758/pbr.16.5.778Megreya,A.M.,andBurton,A.M.(2006).Unfamiliarfacesarenotfaces:evidencefromamatchingtask.Mem.Cognit.34,865–876.doi:10.3758/bf03193433Moscovitch,M.,Winocur,G.,andBehrmann,M.(1997).Whatisspecialaboutfacerecognition?Nineteenexperimentsonapersonwithvisualobjectagnosiaanddyslexiabutnormalfacerecognition.J.Cogn.Neurosci.9,555–604.doi:10.1162/jocn.1997.9.5.555Palermo,R.,O'Connor,K.B.,Davis,J.M.,Irons,J.,andMcKone,M.(2013).Newteststomeasureindividualdifferencesinmatchingandlabellingfacialexpressionsofemotionandtheirassociationwithabilitytorecognisevocalemotionsandfacialidentity.PLoSOne8:e68126.doi:10.1371/journal.pone.0068126Palermo,R.,Willis,M.L.,Rivolta,D.,McKone,E.,Wilson,C.E.,andCalder,A.J.(2011).Impairedholisticcodingoffacialexpressionandfacialidentityincongenitalprosopagnosia.Neuropsychologia49,1226–1235.doi:10.1016/j.neuropsychologia.2011.02.021Peterson,E.,andMiller,S.F.(2012).Theeyestestasameasureofindividualdifferences:howmuchofthevariancereectsverbalIQ?.Front.Psychol.3:220.doi:10.3389/fpsyg.2012.00220Pitcher,D.,Duchaine,B.,Walsh,V.,Yovel,G.,andKanwisher,N.(2011).Theroleoflateraloccipitalfaceandobjectareasinthefaceinversioneffect.Neuropsychologia49,3448–3453.doi:10.1016/j.neuropsychologia.2011.08.020Pitcher,D.,Goldhaber,T.,Duchaine,B.,Walsh,V.,andKanwisher,N.(2012).Twocriticalandfunctionallydistinctstagesoffaceandbodyperception.J.Neurosci.32,15877–15885.doi:10.1523/jneurosci.2624-12.2012Pitcher,D.,Walsh,V.,Yovel,G.,andDuchaine,B.(2007).TMSevidencefortheinvolvementoftherightoccipitalfaceareainearlyfaceprocessing.Curr.Biol.17,1568–1573.doi:10.1016/j.cub.2007.07.063Plomin,R.,DeFries,J.C.,Knopik,V.S.,andNeiderhiser,J.M.(2013).BehavioralGenetics.6thEdn.(NewYork:WorthPublishers).Rezlescu,C.,Pitcher,D.,andDuchaine,B.(2012).Acquiredprosopagnosiawithsparedwithin-classobjectrecognitionbutimpairedrecognitionofdegradedbasic-levelobjects.Cogn.Neuropsychol.29,325–347.doi:10.1080/02643294.2012.749223Richler,J.J.,Cheung,O.S.,andGauthier,I.(2011).Holisticprocessingpre-dictsfacerecognition.Psychol.Sci.22,464–471.doi:10.1177/0956797611401753Richler,J.J.,andGautheir,I.(2013).Whenintuitionfailstoalignwithdata:areplytoRossion(2013).Vis.Cogn.21,254–276.doi:10.1080/13506285.2013.796035Richler,J.J.,Palmeri,T.J.,andGauthier,I.(2012).Meanings,mechanismsandmeasuresofholisticprocessing.Front.Psychol.3:553.doi:10.3389/fpsyg.2012.00553Rossion,B.(2013).Thecompositefaceillusion:awholewindowintoourunderstandingofholisticfaceperception.Vis.Cogn.21,139–253.doi:10.1080/13506285.2013.772929Sadeh,B.,Pitcher,D.,Brandman,T.,Eisen,A.,Thaler,A.,andYovel,G.(2011).Stimulationofcategory-selectivebrainareasmodulatesERPtotheirpreferredcategories.Curr.Biol.21,1894–1899.doi:10.1016/j.cub.2011.09.030Sadeh,B.,Podlipsky,I.,Zhdanov,A.,andYovel,G.(2010).Event-relatedpotentialandfunctionalMRImeasuresofface-selectivityarehighlycorrelated:asimul-taneousERP-fMRIinvestigation.Hum.BrainMapp.31,1490–1501.doi:10.1002/hbm.20952Schweinberger,S.R.,Huddy,V.,andBurton,A.M.(2004).N250r:aface-selectivebrainresponsetostimulusrepetitions.Neuroreport15,1501–1505.doi:10.1097/01.wnr.0000131675.00319.42Tanaka,J.W.,andFarah,M.J.(1993).Partsandwholesinfacerecognition.Q.J.Exp.Psychol.A46,225–245.doi:10.1080/14640749308401045sThorndike,R.L.(1936).Factoranalysisofsocialandabstractintelligence.J.Educ.Psychol.27,231–233.doi:10.1037/h0059840Wai,J.,Lubinski,D.,andBenbow,C.P.(2009).SpatialabilityforSTEMdomains:aligningover50yearsofcumulativepsychologicalknowledgesolidiesitsimportance.J.Educ.Psychol.101,817–835.doi:10.1037/a0016127Wang,R.,Li,J.,Tian,M.Q.,andLiu,J.(2012).Individualdifferencesinholisticprocessingpredictfacerecognitionability.Psychol.Sci.23,169–177.doi:10.1177/0956797611420575Wechsler,D.(1997).WechslerMemoryScale.3rdEdn.SanAntonio,TX:ThePsychologicalCorporation.Wilhelm,O.,Herzmann,G.,Kunina,O.,Danthiir,V.,Schacht,A.,andSommer,W.(2010).Individualdifferencesinperceivingandrecognizingfaces-oneelementofsocialcognition.J.Pers.Soc.Psychol.99,530–548.doi:10.1037/a0019972Wilmer,J.B.(2008).Howtouseindividualdifferencestoisolatefunctionalorganization,biologyandutilityofvisualperception;withillustrativeproposalsforstereopsis.Spat.Vis.21,561–579.doi:10.1163/156856808786451408Wilmer,J.B.,Germine,L.,Chabris,C.F.,Chatterjee,G.,Gerbasi,M.,andNakayama,K.(2012).Capturingspecicabilitiesasawindowintohumanindividuality:theexampleoffacerecognition.Cogn.Neuropsychol.29,360–392.doi:10.1080/02643294.2012.753433Wilmer,J.B.,Germine,L.,Chabris,C.F.,Chatterjee,G.,Williams,M.,Loken,E.,etal.(2010).Humanfacerecognitionabilityisspecicandhighlyheritable.Proc.Natl.Acad.Sci.USA107,5238–5241.doi:10.1073/pnas.0913053107Yin,R.(1969).Lookingatupside-downfaces.J.Exp.Psychol.81,141–145.doi:10.1037/h0027474Young,A.W.,Hellawell,D.,andHay,D.C.(1987).Congurationalinformationinfaceperception.Perception16,747–759.doi:10.1068/p160747Yovel,G.,andDuchaine,B.(2006).Specializedfaceperceptionmechanismsextractbothpartandspacinginformation:evidencefromdevelopmentalProsopag-nosia.J.Cogn.Neurosci.18,580–593.doi:10.1162/jocn.2006.18.4.580Yovel,G.,andKanwisher,N.(2005).Theneuralbasisofthebehavioralface-inversioneffect.Curr.Biol.15,2256–2262.doi:10.1016/j.cub.2005.10.072Yovel,G.,andKanwisher,N.(2008).Therepresentationsofspacingandpart-basedinformationareassociatedforuprightfacesbutdissociatedforobjects:evidencefromindividualdifferences.Psychon.Bull.Rev.15,933–939.doi:10.3758/pbr.15.5.933Zhu,Q.,Song,Y.Y.,Hu,S.Y.,Li,X.B.,Tian,M.Q.,Zhen,Z.L.,etal.(2010).Heritabilityofthespeciccognitiveabilityoffaceperception.Curr.Biol.20,137–142.doi:10.1016/j.cub.2009.11.067ConictofInterestStatement:Theauthorsdeclarethattheresearchwasconductedintheabsenceofanycommercialornancialrelationshipsthatcouldbeconstruedasapotentialconictofinterest.Received:27April2014;accepted:10July2014;publishedonline:19August2014.Citation:YovelG,WilmerJBandDuchaineB(2014)Whatcanindividualdifferencesrevealaboutfaceprocessing?Front.Hum.Neurosci.8:562.doi:10.3389/fnhum.2014.00562ThisarticlewassubmittedtothejournalFrontiersinHumanNeuroscience.Copyright©2014Yovel,WilmerandDuchaine.Thisisanopen-accessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense(CCBY).Theuse,distributionorreproductioninotherforumsispermitted,providedtheoriginalauthor(s)orlicensorarecreditedandthattheoriginalpublicationinthisjournaliscited,inaccordancewithacceptedacademicpractice.Nouse,distributionorreproductionispermittedwhichdoesnotcomplywiththeseterms. FrontiersinHumanNeurosciencewww.frontiersin.orgAugust2014|Volume8|Article562|9