Autism and the development of face processing

www.elsevier.com/locate/clires

Autismandthedevelopmentoffaceprocessing

GolijehGolaraia,*,KalanitGrill-Spectora,b,AllanL.Reissa,caDepartmentofPsychiatryandBehavioralSciences,StanfordUniversitySchoolofMedicine,Stanford,CA94305,USA

bDepartmentofPsychology,StanfordUniversity,CA94305,USAcPrograminNeuroscience,StanfordUniversity,CA94305,USA

Abstract

Autismisapervasivedevelopmentalcondition,characterizedbyimpairmentsinnon-verbalcommunication,socialrelationshipsandstereotypicalpatternsofbehavior.Alargebodyofevidencesuggeststhatseveralaspectsoffaceprocessingareimpairedinautism,includinganomaliesingazeprocessing,memoryforfacialidentityandrecognitionoffacialexpressionsofemotion.Insearchofneuralmarkersofanomalousfaceprocessinginautism,muchinteresthasfocusedonanetworkofbrainregionsthatareimplicatedinsocialcognitionandfaceprocessing.Inthisreview,wewillfocusonthreesuchregions,namelytheSTSforitsroleinprocessinggazeandfacialmovements,theFFAinfacedetectionandidentificationandtheamygdalainprocessingfacialexpressionsofemotion.Muchevidencesuggeststhatabetterunderstandingofthenormaldevelopmentofthesespecializedregionsisessentialfordiscoveringtheneuralbasesoffaceprocessinganomaliesinautism.Thus,wewillalsoexaminetheavailableliteratureonthenormaldevelopmentoffaceprocessing.Keyunknownsinthisresearchareaaretheneuro-developmentalprocesses,theroleofexperienceandtheinteractionsamongcompo-nentsofthefaceprocessingsysteminshapingeachofthespecializedregionsforprocessingfacesduringnormaldevelopmentandinautism.

PublishedbyElsevierB.V.onbehalfofAssociationforResearchinNervousandMentalDisease.

1.Introduction

Facesarerichconduitsofpersonalinformation.Duringabriefencounter,healthyadultsoftenautomaticallyattendtoandquicklyperceivethecomplexsetofinformationcon-tainedinaface,recognizingtheemotionalstateandsocialcontext,andoftenrememberingtheindividualfacelater.Thiscomplextaskoffaceprocessinginnormaladultsinvolvesadistributedneuralsysteminwhichspecificlociareimplicatedinprocessing-specificfacialinformation.Forexample,aregionalongthesuperiortemporalsulcus(STS)isinvolvedindetectingfacialmovementsassociatedwitheyegaze,speech,andemotionalexpressionandinten-tion[1–3].Theamygdalarespondstofaces,especiallyfear-fulfaces[4–6].Andaregionintheventral-occipitalcortex,the‘‘fusiformfacearea’’(FFA)[7]isimplicatedinfacedetection,categorizationandidentityrecognition[7–12].

*Correspondingauthor.Tel.:+16507250797;fax:+16507244794.E-mailaddress:ggolarai@psych.stanford.edu(G.Golarai).

Somecomponentsofthefaceprocessingsystemmayexistatbirth,whileotherscontinuetodevelopduringchildhoodandadolescencebeforereachingtheadultlevel[13,14].Somebasicgoalsoffaceprocessingresearcharetodeter-minethespecificfunctionofeachofthecomponentsofthefaceprocessingsystem,thetimecourseandmechanismofdevelopmentofeachcomponent,andtheinteractionsamongthevariouscomponentssubservingnormalfaceprocessing(Fig.1).

Interestintheneuralmechanismsoffaceprocessinganditsdevelopmentispartlyfueledbyseveraldevelopmentalconditionssuchasautism,whichareassociatedwithanom-alousfaceprocessing.Faceprocessingimpairmentsinaut-ismarethefocusofintenseinvestigation,giventheimportanceoffacesinconveyingsocialandemotionalinformationstartingsoonafterbirth.Althoughthereisageneralagreementthatautisminvolvesdeficitsinfacepro-cessing,severalquestionsremain.Forexample,theprecisenatureofthesedeficits,theunderlyingmechanismsandtherelationshipsbetweenanomalousfaceprocessingand

1566-2772/$-seefrontmatterPublishedbyElsevierB.V.onbehalfofAssociationforResearchinNervousandMentalDisease.doi:10.1016/j.cnr.2006.08.001

146G.Golaraietal./ClinicalNeuroscienceResearch6(2006)145–160

Fig.1.(fromKalanitGrill-Spector)Top:Posteriorviewofaninflatedbrainshowregionsthatrespondmoretofacesthantoothervisualcategories(abstractobject,scenes,andscrambledimages).Severalfaceselectiveregionsshowninshadesofyellowandorangeareshownintheoccipitalcortex,superiortemporalsulcus(STS)andfusiformgyrus(i.e.,‘‘fusiformfacearea’’,FFA).Amotionsenstiveregion(MT)andretinotopicvisualareasarealsoshownintheoccipitalcortex.Bottom:VentralviewoftheFFAinrelationtothecollateralsulcus(CoS).

atypicalsocio-emotionalfunctioninautismremainunclear.Thislackofclarityislikelyduetoanumberoffac-tors,althoughitisprobablethattheheterogeneityofriskfactorsandvarietyofpathogeneticmechanismsassociatedwithautismareparticularlyimportant[15–17].Thus,giventhecomplexityofthefaceprocessingsystem,theremayexistsubpopulationsamongindividualswithidiopathicautismshowingdeficitsindifferentcomponentsofthisessentialneurocognitivesystem.Alsocontributingtouncertaintyinunderstandingfaceprocessingdeficitsinaut-ismisarelativelackofknowledgeconcerningthelongitu-dinaldevelopmentofneuralsystemsunderlyingfaceprocessinginhealthychildren,adolescentsandadults.Thus,researchoncorecomponentsoffaceprocessingandtheirneuromaturationaltimecourseinnormaldevel-opmentmayaugmentunderstandingoffaceprocessingdeficitsinautismandtheirrelationtosocialandemotionalmaturation.Ultimately,suchworkmayprovideusefulclin-icaltoolsforearlydiagnosisandremediation.Here,we

reviewresearchrelevanttoseveralcomponentsoffacepro-cessinginpersonswithautismandhealthycontrols,andtheirdifferentialdevelopmentduringinfancyandchild-hood,pointingtogapsinknowledgeandpotentiallyfruit-fulareasforfutureresearch.2.WhatisAutism?

Autismisapervasivedevelopmentaldisorderwhosecausesorunderlyingbiologicalmechanismsarenotwellunderstood.Thisbehaviorallyanddevelopmentallydefinedsyndromeischaracterizedbyimpairmentsinnon-verbalcommunication,socialrelationshipsandstereotypedpat-ternsofbehavior(DSM-IV,AmericanPsychiatricAssoci-ation,1994).Autismisconsideredasevereformofautismspectrumdisorder(ASD),whichincludesmilderformssuchasAsperger’ssyndrome[18].Currently,therearenogeneticorneurologicalmarkersforthemajorityofindividualswithautism,andthediagnosisofASDistypicallydeterminedbyexpertcliniciansusingstandard-izedassessments,suchastheAutismDiagnosticObserva-tionSchedule[19,20].Althoughthecauseofautismisunknown,thevariousformsofASDarelikelytoresultfromcombinationsofgeneticandenvironmentalfactors[15–17].Despitethisheterogeneity,problemswithsocialandemotionalreciprocityareconsideredahallmarkofASD.Preliminaryindicationsmayappearasearlyasthefirstyearoflife,includinganomaliesinmutualgaze,alackofinterestinthehumanface,andpreferenceforinanimateobjects[18,21–24].

Inoneview,earlyspecificimpairmentsinfaceprocess-ingeventuallygeneralizetoalackofsocialinterest.Sup-portingthismodelisthecentralimportanceoffacialcommunicationbetweeninfantsandprimarycaregiversduringearlydevelopment,especiallyinattributingmeaningandemotionalsignificancetohumaninteractions.Suchearlyimpairmentsinfaceprocessingmaybea‘‘bottom-up’’resultofanomaliesinlow-levelvisualprocesses,suchasmotionprocessing(reviewedin[25]).Alternatively,deficitsinfaceprocessingmaybeamongtheconsequencesofa‘‘top-down’’disinterestinhumaninteractioninASD.SupportingthelattermodelarethevariedandpervasivenatureofASDsocialdeficits,includingnon-visualmodal-ities,andtheexistenceofindividualswithASDwhopro-cessfaceswithinthenormalrange.However,theabsenceofdetectabledeficitscouldvariouslyresultfromnormalfaceprocessing,successfuluseofalternativestrategiesorlackofsensitivityofcurrentexperimentalmethods.Adetailedexaminationofnormalfaceprocessingwillsetthestagefordiscriminatingbetweensomeofthesepossibilities.

2.1.Viewingbehaviorandgazeprocessing

Thehumanfaceisafocusofvisualattentioninmosthealthyindividualsstartingsoonafterbirth.Newborns(9-minold)typicallylookmoreataschematicdrawing

G.Golaraietal./ClinicalNeuroscienceResearch6(2006)145–160147

ofafacethanatothervisualpatterns,includingafaceoutlinewithreconfiguredinternalparts[26–28]andalsopreferdirectgazeoveravertedorclosedeyes[29,30].Themechanismsunderlyingthesepreferencesarenotwellunderstood.Onemodelsuggeststhatthesepreferencessignifyaninnatefacedetectingmechanismthatdrivesattentiontofacesearlyduringdevelopmentandmaybemediatedbysubcorticalneuralsystems[31,32].Alterna-tively,theseearlyfacepreferencesmayreflectthebasicpropertiesoftheimmaturevisualsystematbirththatarebeststimulatedbythephysicalpropertiesoffaces,suchasthehighcontrastbetweenthepupilandsclera[27,33].

Ineithercase,infants’attentiontofacesisthoughttobeanimportantcomponentofnormaldevelopment,wherebyfacesbecomeafrequentandsalientvisualstimulusearlyafterbirth,alikelytimeforactivity-dependentplasticityinthevisualsystem.Indeed,somedevelopmentalmodelsofvisualcortexhypothesizethathabitualpatternsoffixa-tionshapethelong-termorganizationofface-processingcorticalregions[34].Consistentwithitsearlyonsetininfancy,eyecontactisalsoanimportantmeansofcommu-nicationthroughthelifespan.Childrenandadultsarepro-ficientinrapidlydiscriminatingthedirectionofgazeandreflexivelyorientingtowardsthecorrespondingdirection[35],althoughitisnotwellknowniforhowproficiencyingazediscriminationmaychangeduringthecourseofdevelopment.

2.2.Theneuralbasisofgazeprocessing:TheroleofSTSImagingstudiesinadultssuggestthatgazeprocessinginvolvesanetworkofbrainregionsincludingacorticalregionintheposteriorsuperiortemporalsulcus(STS).RegionsintheposteriorSTShavebeenshowntorespondtomovingandstationaryeyesandmouth,butnottomovingcheckerboardsorcontractingcircles[1].TheSTSismoreactivatedwhensubjectsselectivelyattendtoeyegazethantofaceidentity[1,2,9].RegionsintheSTSalsorespondtoarangeofvisualsignalssalientforsocialinteraction,suchasmutualgaze,emotionalexpres-sion,speech,intentionallimbmovements,andbiologicalmotioningeneral[1,2,9,36–39].Grossmanetal.(2005)suggestedthatnormalSTSfunctioningisrequiredforper-ceptionofbiologicalmotion,asrepeatedtranscranialmagneticstimulation(rTMS)disruptedcorticalactivityintheposteriorSTSandreducedperceptualsensitivitytopoint-lightanimationsinhealthyadults[36].TheSTShasalsobeenimplicatedinsocialcognitionandattribu-tionofmentalstatesandintentions[40–44].Inonestudy,theSTSwasactivatedbysimplegeometricshapes,whichhadnoresemblancetofacesorbodypartsexceptthattheirpatternsofmotionconveyed‘‘intention’’[43].Thus,STSactivationshavebeenassociatedwithattributionofintention,evenintheabsenceofeyes,facesorotherbio-logicalforms.Giventhatattributionofmentalstatesinvolvesactivationofmanybrainregions(e.g.,fusiform

gyrus,amygdala,andprefrontalcortex),determiningthespecificcontributionofSTStothesetaskswilllikelyrequirefurtherrTMS.

FewdevelopmentalstudiesoftheSTSexist.OnefMRIstudyfoundsimilarresponsesintheSTSofadultsandchil-dren(ages7–10)duringprocessingofavertedgaze[45],consistentproficientgazeprocessinginchildren.However,aMEGstudyfoundgazemodulationofanearlymagneticfieldactivity(Plm,around140ms)inchildrenages8–12,butnotadults,thatwaslocalizedtotheinferiortemporalsulcusandposterioroccipitalcortex[46].Overall,thissparseliteraturesuggeststhatsomematurationofthetim-ingandfunctionalorganizationofgazeprocessingmaycontinuethroughoutchildhood,butfuturestudiesareneededtoexaminechildren’sperceptualsensitivitytofine-grainvariationsingaze,andtheeffectofthesevaria-tionsonresponsesofSTSandotherrelatedfunctionalregions.AlsounknownarethetimecourseofdevelopmentofSTS,andtheextenttowhichitsneuralorganizationandfunctiondependongeneticsand/orexperiencedependentmechanisms.

2.3.Anomalousviewingbehaviorinautism

Lackofreciprocaleyecontactisanearlyandstrikingmanifestationofautism[47–50].Retrospectivereviewsoffamilyhomemoviessuggestsignsofatypicalsocialbehaviorinchildrenwhoarelaterdiagnosedasautistic[48–50].Theseanomaliesincludepooreyecontact,andsloworabsentmutualgazeininfantsandchildren,buttheunderlyingmechanismsarenotwellunderstood.Inaddition,recentevidenceforanomalousgazebehav-iorinparentsofsomeautisticchildrenraisesthepossibilityofcomplexinteractionsamonggenesandenvironmentalfactorsaffectingdevelopmentofeyecon-tact[51].

Littleisknownabouttherelationshipbetweenearlyanomalousgazeandlong-termfaceprocessingdeficitsinASD,butthereisevidencethatotherdisruptionsinearlyvisualexperiencemayleadtolong-termdeficitsinfacepro-cessing.Forexample,otherwisehealthyinfantswhoaretemporarilydeprivedofhigh-resolutionvisualinputduetocongenitalcataractforaperiodof45to863days(mean199days)showspecificdeficitsinrecognitionoffacialiden-tities8–29yearslater(mean17years)[52].Byanalogy,itmaybethatearlyanomaliesingazebehaviorinASDreducevisualexposuretointernalfacefeatures,andcon-tributetolong-termfaceprocessingdeficits.Conversely,lessfrequenteyecontactinASDmayresultfromdifficul-tiesinprocessingfacialinformation.Longitudinalstudiesofface-processingdevelopmentandtheroleofexperience,inASDandnormalchildren,areneededtoevaluatethesedivergentpossibilities.Suchstudiesmaydeterminewhetherthereisindeedacorrelationbetweenearlygazebehavior(e.g.,amongoffspringandcaregivers),andsubsequentcompetenceinfaceprocessing,aswellasemotionalandsocialskills.

148G.Golaraietal./ClinicalNeuroscienceResearch6(2006)145–160

2.4.Anomalousgazeprocessingamongchildrenandadultswithautism

AnomalousgazeprocessingremainsahallmarkofASDthroughoutchildhoodandintoadulthood,althoughsub-stantialvariabilityisreported.ChildrenwithASD(ages9–14)aresloweratdetectingdirectgazerelativetocontrols[53]eventhoughtheydecodeandorienttothedirectionofavertedgazeaccurately[53,54].Consistentwithdeficitsinreciprocalgazebehavior,theresultsofarecentstudyshowedthatadultswithASDtypicallyspendlesstimelookingattheinnerfeaturesoftheface,particularlytheeyes,incontrasttohealthyorIQmatchedcontrols[55].Inthisstudy,controlsviewedfacesinastereotypicalpat-tern,generallytracingatrianglethatsubtendedtheeyes,noseandmouth.Incontrast,thepatternsoffaceviewinginASDappearederraticandlesspredictable,variablyfocusedonfacialfeaturessuchastheear,chinorhair-line.Inanotherstudyofadults,viewingnaturalisticsocialscenes,reducedvisualfixationontheeyeregionandincreasedfixationonobjectspredictedtheASDdiagnosis[56].Inapparentcontradiction,vanderGeestetal.(2002)foundnodifferencesinhowhighfunctioning10-year-oldchildrenwithASDandnormalcontrolsvieweduprightfaces,althoughgroupdifferencesinstrategyemergedforupside-downfaces[57].Similarly,othershavereportedthatindividualswithASDlookwithnormalfre-quencytowardcaregivers[58],orataperson’sfacewhenattentionisdrawn[59,60].ThedivergenceamongthesedatamayreflectacombinationofageortaskspecificityofgazeanomaliesinASD,and/oradiversityofASDphe-notypes.Toresolvethiscontroversy,itwillbeimportanttodetermineifthereareanyagedependentdifferencesinfaceviewingstrategiesamonghealthyorindividualswithASD.AlsoimportantistodeterminewhethertheviewingdeficitsinASDarespecifictofacesasopposedtootherobjects,similarforfamiliarandunfamiliarfaces,andreproduciblewithinthesamesubjects(Fig.2).

2.5.TheneuralbasisofanomalousgazeprocessinginASD:AroleforSTS?

AlthoughtheneuralbasisofaberrantviewingstrategiesinASDisnotwellunderstood,thereisevidencethattheSTSandtheamygdalamaybeinvolvedintheatypicalpro-cessingoffacialinformation.Forexample,anMRIstudyrevealedanatomicaldisplacementofmajorsulciinfrontalandtemporalbrainregionsinchildrenwithASD(meanage10.7±3.1std)relativetohealthychildren(meanage11.3±2.9std),includinganteriorandsuperiordisplace-mentsofthesuperiortemporalsulcusbilaterally[61].Avoxel-basedmorphometryanalysisofthebrainsofchil-drenwithASD(meanage15.4±2.2std)comparedtoIQmatchedcontrols(meanage15.5±1.6std)foundtotalbraingraymattervolumeincreasesandlocalizedincreasesthatincludedtherightsuperiortemporalgyrus[62].Alsousingvoxel-basedmorphometry,Boddaertetal.(2004)

Fig.2.(fromPelphreyetal.2002)Patternsofvisualscanningofaffectivefacesfollowedastereotypicalpatternincontrols,outliningatrianglebetweentheeyesandmouth.Incontrast,scanpathsinparticipantswithASDweremoreerraticandoftenexcludedtheinternalfeaturesoftheface.

foundcontraryresultsinchildrenwithASD(meanage9.3±2.2std),withdecreasesofgreymattervolumelocal-izedtotheSTScomparedtohealthycontrols(meanage10.8±2.7std)[63].Thisdiscrepancybetweenstudiesmayreflectdifferencesintheagerangeorchoiceofcontrols.Alternatively,giventheevidenceforanatomicaldisplace-mentofSTSpositioninASD[61],divergentresultsfromvoxel-basedmorphometrystudiesmayreflectmethodolog-icalproblemsarisingfromimagenormalizationproce-dures,thus,callingforfurtheranalysiswithindividuallydefinedanatomicalboundaries.Also,itwillbeimportanttodeterminethedevelopmentaltimecoursewhenanatom-icalmeasuresofSTSdivergeamongnormalsandindividu-alswithASD.

FMRIofgazeprocessingsuggestsabnormalSTSresponsesamongindividualswithASD.Inanevent-relatedfMRIstudy,subjectswithASD(meanage9.3±2.2std)andcontrols(meanage10.8±2.7std)watchedavirtualactorwhoseavertedgazewaseitherdirectedtowardsacheckerboardtarget(‘‘congruent’’condition)orawayfromthetargetandtowardsemptyspace(‘‘incongruent’’condi-tion)[64].Subjectswereinstructedtopressabutton,when-everthe‘‘eyesmoved’’.Bothgroupswereequally(andhighly)accurateindetectingchangesineyegaze.However,innormalsubjects,theincongruentconditionevoked

G.Golaraietal./ClinicalNeuroscienceResearch6(2006)145–160149

greaterSTSactivationthandidthecongruentcondition,aresultthatwasnotseeninsubjectswithASD.ThislackofmodulationintheSTSoftheASDgroupwasinterpretedtoreflectdeficitsininterpretingotherpeople’sintentionsthatarenormallyconveyedbygazeshifts.However,thisstudydidnotcontrolforthepossibilitythattheabsenceofSTSmodulationmaybeassociatedwithamoregeneralanomalyinmotionprocessing,whichhasbeenfoundinseveralbehavioralstudies(seereview[25]).Also,giventheevidenceforanatomicaldisplacementofSTSpositionamongsubjectswithASDrelativetocontrols[61],individ-ualROIanalysesoffMRIdatawouldbeprudentinfuturework.Despitetheseissues,fMRIevidencefromotherclin-icalpopulationssupportthenotionthatanomalousgazeprocessingisassociatedwithaberrantSTSactivations.Forexample,individualsdiagnosedwithfragileXsyn-dromewhosharemanyofthesocialandemotionalpheno-typesofautism,includinganomalousfaceandgazeprocessing,alsoshoweddecreasedSTSactivationduringagazeprocessingtask[65].Itremainstobedeterminedifamongtheseclinicalpopulations,STSfunctionisalsoabnormalduringprocessingofbiologicalandnon-biologi-calmotionandinterpretationofothers’mentalstatebasedonnon-gazebehavior.

2.6.DevelopmentoffacialidentityrecognitioninchildrenNewborninfantsat3–4daysshowsignsoffacerecogni-tion[66].At6monthsinfantsareabletodiscriminatefacialidentitiesofhumanandofotherspeciesofprimatesequallywell,suggestingalackofbiasforhumanfaces[67].How-ever,by9monthsinfantslosetheabilitytodiscriminateamongotherspeciesfaces,suggestingthattheirfacerecog-nitionsystemismorenarrowlytunedtohumanfaces[67],unlesstheyaretrainedonotherspecies’faces[68].Byage2childrenareabletorecognizemanyhumanfaces.

Despitetheseearlycapabilities,facerecognitionunder-goesprolongeddevelopment,extendingintoadolescence[13,14,69–74].Recognitionperformancefornewlylearnedfacesimprovessignificantlyduringchildhood.Althoughthedetailsvaryamongreports,performancerangesfrom50%to70%oftheadultlevelfromage6to14,withadipbeforepubertyandslowergainsafterage16[13,69,71–74].Theunderlyingmechanismsarenotfullyunderstood,butfromages6–14yearsseveralaspectsoffacerecognitionchangeinsuchawaythatitseemsunlikelythatagerelatedimprovementsinfacerecognitionaresolelyduetomaturationofdomaingeneralmnemonicprocesses.ThefollowingthreeexampleswillinformandmotivateourreviewofliteraturerelatedtothedevelopmentoffacerecognitioninASDandnormals.

2.7.Threedevelopmentalaspectsofidentityrecognitioninhealthychildren

Threephenomenahavebeenidentifiedinthedevelop-mentofidentityrecognitioninhealthychildren.First,

comparedtoadults,childrentypicallyshowlessdecrementinrecognitionperformanceforfacesofotherraces[74,Chance,1982#52];eventhoughthis‘‘raceeffect’’maystartearlyduringinfancy[75,76].Inadults,thisraceeffectisthoughttoreflectaccumulatedexperiencewithownracefacescomparedtootherracefaces[74,77].Consistentwiththisidea,childrenseemtobemoreequallyproficientatrec-ognizingallfaces,albeitbelowtheadultlevel.Thisearlybreadthoffacerecognitioncapacitythatnarrowsasfacerecognitionskillsimprovewithageisanalogoustochil-dren’swiderangeofphonemesduringearlylanguageacquisition,whichnarrowswithincreasingskillandspe-cializationinagivenlanguage.Ourrecentfindingssuggestthatthedevelopmentofthisraceeffectinchildrenisasso-ciatedwithmaturationofafaceselectivecorticalregion,namelythe‘‘fusiformfacearea’’[78].Second,comparedtoadults,childrenbetween3and11makemoreerrorsinmatchingfacesofthesamepersonaftertransformations,suchasaging,facialexpression,viewpoint,oradditionofparaphernalia[79,80].Thesestudiessuggestweakertransformationinvarianceinprepubescentchildrencom-paredtoadults,perhapsinpartduetoheavierrelianceonindividualfeatures.Third,unlikeadults,childrenaged6–8yearsolddonotshowpreferentialrecognitionmemoryfor‘‘distinct’’facescomparedtotypicalfaces,andevenatage13thisaspectofchildren’sperformanceisfarfromadult-like[80].Thisfindingsuggeststhattypicalityislessusedorwell-definedinchildren,comparedtoadults.2.8.Developmentalmodelsofnormalfacialidentityrecognition

Resultsfrombehavioralstudiesoffaceprocessinginhealthychildrenareconsistentwiththehypothesisthatnormalfacerecognitionrequiresperceptualskillsgainedbyaccumulatedexperiencewithmanyfacesoveryears[13,81].Severalmodelshaveattemptedtoexplainhowvisualexperiencewithfacesmayimproveface-recognitionskills.Thesemodelsprovideaframeworkforthenormaldevelopmentoffaceprocessingandhowareducedpropen-sitytoviewfaces(asinthecaseofautism)mightleadtoatypicaldevelopmentoffaceprocessingskills.

Valentine’sinfluentialmodeloffacerecognitionproposesamultidimensionalarrayforstorageoflearnedfaces[81].Inthisarray,‘‘typical’’facesformacluster(duetotheirsimilarities)and‘‘distinct’’facesaredistributedmoresparselyintheperiphery.Afteraccumulatedexperiencewithfaces,fine-graindifferencesamongthe‘‘typical’’facescanbeefficientlydetectedduetothemanyexemplars,andthelargedifferencesbetween‘‘typical’’and‘‘distinct’’facesarereadilydetected.Incontrast,itwouldbedifficulttomakefinegraindistinctionsamongatypicalfaces(e.g.,facesbelongingtoagivenotherrace)duetotheavailabilityoffewerexemplars.Extrapolating,othershaveinterpretedchildren’slackofbiasforsameraceordistinctfacestomeanthatyoungchildrendonothaveawell-definedpro-totype,duetofewerexperienceswith‘‘typical’’faces[80].

150G.Golaraietal./ClinicalNeuroscienceResearch6(2006)145–160

Theneuralsubstratesforsuchanexperience-dependent‘‘tuning’’offacerecognitionarenotwelldefined.However,severallinesofevidencesuggestthatfaceselectiveregionsintheoccipito-temporalcortex,namelythe‘‘fusiformfacearea’’(FFA)maybeinvolved[82–84].

Inanothermodel,DiamondandCareyviewthedevel-opmentofexpertiseinfacerecognitionasatransitionfromaninefficientpiecemealor’’featural’’representationtoamoreefficientprocessingofthewholeface(’’holistic’’representation)orofthespatialrelationsamongthefacialfeatures(‘‘configural’’representation)[85,86].Duetodiffi-cultyinrelatingtheseconceptspreciselytoexperimentalstimuli[87,88]supportforthismodelhasbeencontrover-sial[85,89–93].Despitethesechallenges,thenotionsoffeaturalandconfiguralprocessingremainattractivecon-ceptsinunderstandingvariationsinfaceprocessingperfor-manceamonghealthyandclinicalpopulations.

Morerecently,psychophysicsstudiesofadultfacerec-ognitionsuggestthatcertainconstellationsoffacialfeaturesareespeciallyinformativeforfacerecognition[94–96]raisingthequestionofwhetherchildrenaresimilarinthisregard.Otherpsychophysicsstudiesfoundthatear-lierstagesoffaceprocessing,suchascategorization(e.g.,facevs.object)arefollowedbyalaterstageoffaceidenti-fication(e.g.,Billvs.George)[97–99].Itisnotknownifmaturationoffacerecognitioninvolveschangesinspeedofrecognition,orcorrelateswithchangesintheaccuracyortimingofearlierstagesoffaceprocessing,suchascate-gorization.Althoughfaceselectiveevent-related-potentials(namelytheN170)havebeenreportedtobeslowerinchildrenthaninadults[100],supportingbothofthesepos-sibilities,morestudiesareneededtodeterminethematura-tionaltimecourseandtherelationshipbetweenspecificstagesoffaceprocessingduringnormaldevelopment.2.9.AnomalousidentityrecognitioninASD

DeficitsinrecognitionmemoryfornewlyviewedfaceshavebeenreportedinchildrendiagnosedwithASD,asyoungas2yearsoldandthroughadulthood.Forexample,Klinetal.(1999)foundthatidentityrecognitionfornewlyviewedfaceswasmorevulnerabletochangesinposeorexpressioninchildrenwithASD(ages2–10)comparedtocontrols,matchedforchronologicalageorabilitiesinclud-ingshort-termvisualmemory[101].Similarly,BoucherandLewis(1992)reporteddeficitsinsubsequentrecognitionofnewlyviewedfacesinchildrenwithautism(8to17yearsold)comparedtoagematchednormals,orlearningdis-abledcontrols[102].Theseresultscouldnotbeexplainedbydeficitsinvisualdiscriminationamongfaces,asthesamesubjectswithautismaccuratelymatchedfacestocon-currentsamples.Furthermore,childrenwithASD(ages3–11)wereabletoaccuratelysort,basedonidentity,picturesoffacesthatvariedacrossidentityandexpression[103].ThesefindingssuggestthatfacerecognitionmemoryisimpairedinASDevenasvisualdiscriminationamongfacesmaybeintact(butsee[104]).

Apotentialmechanismforpoorfacerecognitionmem-oryinASDisanomalousfaceviewingstrategies.Forexample,inalandmarkstudybyLangdell(1978),childrenwithASDandcontrols(matchedformentalageandchro-nologicalage)wereaskedtoidentifyfacesoftheirpeersthatwerepartiallymasked[105].TheoverallperformanceofparticipantsdiagnosedwithASDandcontrolsweresim-ilar.However,betweengroupdifferencesemergeddepend-ingonmaskplacement,suggestingthatchildrenwithASDreliedmoreonfacialinformationinthemouthareaincon-trasttocontrolswhoreliedmoreontheeyeregion.OthershavefoundevidenceforunusualrelianceonfacialfeaturesasapposedtoconfiguralfaceinformationinASD[106,107].Furtherstudiesareneededtodeterminethecor-relationbetweenfaceviewingstrategiesandsubsequentidentityrecognitionperformanceamongchildrenwithASDandtypicallydevelopingcontrols.

PoorfacerecognitioninASDmaynotbelimitedtoexperimentalsettings.Inonestudy,recognitionmemoryforfacesoffrequentlypresentindividualsinaschoolenvi-ronmentwasloweramongchildrenwithASD(ages7–11),comparedtocontrolsmatchedforageandverbalability[108].Thus,accordingtoValentine’smodel(discussedabove),autisticchildrenmayencodefewerfaceexemplarsthannormal,anddevelopless‘‘expertise’’infacerecogni-tion.NotethatthisperiodofpossibleinattentiontofacescoincideswithsubstantialmaturationoftheFFA,accord-ingtorecentfMRIstudies[82–84].Longitudinalstudieswouldbevaluableinaddressingthispossibility.

EvidenceforthedomainspecificityoffacerecognitiondeficitsinASDisnotuniform.Insupport,Boucheretal.(1992)foundthatchildrenwithASDandpoorfacerecog-nitionweresimilartonormals(matchedforverbalability)forvisualmemoryoftheirschoolbuildingrelativetocon-trols[108].Similarly,Blairetal.2002foundthatcomparedtoageandverbalIQmatchedcontrols,agroupofadultswithASDshowedpoorfacerecognitionmemory,butnor-malrecognitionmemoryforbuildingsandleaves[109].Incontrast,anumberofotherstudieshavefoundvisualper-formancedeficitsinASDthatweregeneralizedtonon-facestimuli,includingsortingofgeometricshapes[110]andmemoryfornon-socialstimuli,whencomparedtocontrolswhoarematchedforothernon-verbalabilities[111,112].ThesevariedfindingsmayreflecttheheterogeneityofASDorofthechoiceofcontrols,non-faceobjectsortasks.FuturestudiesareneededtotestrecognitionmemoryinASDacrossarangeofabilityandage,includingavarietyofnon-facestimuli.

2.10.ERPcorrelatesoffaceprocessinginnormaladultsandchildren

ScalpERPsandintracranialrecordingsfromotherwisehealthyepilepticpatientshaverevealedareliableandselec-tiveindexoffaceprocessingat150–200mslatency,namelythescalpN170[113–115].Similarly,theintracranialN200,hasbeenrecordedatdiscretelocationsoverthefusiform

G.Golaraietal./ClinicalNeuroscienceResearch6(2006)145–160151

gyrus,theinferiortemporalsulcus(ITS)andthesuperiortemporalsulcus(STS)[116].Possiblymarkingthecatego-ry-recognitionstage,theN170issensitivetofeaturesusu-allypredictiveoffaces,suchasisolatedeyes,noseormouth[117,118],orbackviewsofthehead,orfaceoutlineswithoutfeatures.Incontrast,theN170appearsindepen-dentoffaceidentity[113,119,120]asitisinsensitivetofacerepetition[119,121],learning[122],familiarity[123,124],orattention[125].AlthoughtheN170isnotdynamicallyresponsivetofaceidentityormemory,comparisonamongadultsubjectssuggeststhatitsamplitudeandlatencymaypredictgeneralproficiencyatfacerecognition[114].

Inchildrenages5–14yearsold,theN170wasdelayedinonset,longerindurationandlessnegativecomparedtoadults,andbecamemoreadult-likewithage[100,117].Fol-lowingasimilartimecourse,identityrecognitionprogres-sivelyimprovedintolateadolescencebeforereachingtheadultlevel[73].Thusfacerecognitionmaybeamulti-stageprocess,wheretheearlystageoffacecategorizationisimportantforlaterstagesofmnemonicprocessingthataresensitivetofaceidentity[113,120,123].

Consistentwithamulti-stagemodeloffaceprocessing,theN170/N200isfollowedat300–700msbymnemonicallysensitivepotentialsthatrapidlyhabituatewithrepetitionandaremodulatedbysemanticpriming,face-namelearn-ingandidentification[121].TheN700issignificantlylargerforfacesthanforotherstimuli[116],suggestinglocalgen-erationbyface-selectivemodulesintheoccipito-temporalcortex.Thesefindingsareconsistentwithidentity-sensitive,late,unitresponsesfromface-selectivecellsintheprimatetemporalcortex[126].

2.11.ERPevidenceforanomalousidentityrecognitioninASD

RecentstudieshavefoundanomaliesinfaceselectiveERPsintheoccipito-temporalcortexofindividualswithASD.Inonestudy,theN170wasdelayed,broadened,andinsensitivetofaceinversioninASDsubjects(ages15–42)comparedtonormalcontrols(ages16–37),parallel-ingdifferencesfoundbetweenhealthychildrenandadults(discussedabove).AnotherstudyfoundslowerandloweramplitudeN170amongadultswithAsperger’ssyndromecomparedtocontrols,butsurprisinglynobetweengroupdifferencesamongchildrenwithAsperger’sandcontrols(meanage11.2±1.89;ASDmeanage11.6±1.9)[127].TheseresultssuggestthattheanomalousN170inASDrep-resentsaprogressivedivergencefromnormaldevelopment,ratherthanadevelopmentaldelay.Totestthishypothesisandexploremechanisms,longitudinalstudiesareneededtodeterminethetimecourseofdevelopmentofN170anomaliesinASD,thelocationofN170generatorsinchildrenwithandwithoutASD,andtherelationshipofN170timingandamplitudewithinstantaneousandhabitualpatternsoffaceviewing.

ConsistentwiththeideathattheN170representsacru-cialearlystageoffaceprocessing,McPartlandetal.(2004)

foundapositivecorrelationbetweenN170latencyandfaceidentificationdeficitsamongindividualswithASD[128].InaseparatestudyofchildrenwithASD(ages3–4)comparedtocontrols,Dawsonetal.(2002)foundatypicalmodula-tionoflatepotentialsinresponsetorepeatedpresentationofafamiliarversusanunfamiliarfaceinchildrenwithASD[129].AdditionalERPandMEGstudiesspecifyingtherelativedevelopmentaltimecourseandgeneratorsoftheanomalousN170andthelatefamiliarity-sensitivepotentialswouldhelpclarifythepotentialroleofmnemonicprocessing(i.e.,storageofmanyfaceexemplars)inshapingtherelativelyearlystagesoffaceperception,bothinnormalsandinASD.Onekeyquestioniswhetherthefamiliaritysensitivepotentialsareassociatedwiththefusi-formfacearea.

2.12.Neuralsubstratesofidentityrecognitioninhealthyadults:TheroleofFFA

Facerecognitioninvolvesadistributednetworkofbrainareas,includingregionsinthemedialtemporallobeandprefrontalcortex(see[130,131])aswellasinthefusi-formgyrus[132].Here,wefocusonafaceselectiveregioninthefusiformgyrus,namelythe‘‘fusiformfacearea’’(FFA)[7]whichhasreceivedparticularattentionforitsroleinfacerecognitioninadultsforseveralreasons.First,somepatientswhohavesufferedfocalinjurytothetem-poralcortexareselectivelyimpairedinfacerecognition(prosopagnosia),whileothersareselectivelyimpairedintherecognitionofnon-faceobjects(objectagnosia)[133–135].Second,neuroimagingmethodshaverevealed-specificface-selectiveregionsinthefusiformgyrus,namelythe‘‘FFA’’[7].Third,FFAresponsestovaguelyface-likestimulicorrelatewiththesubjectiveexperienceoffaceperception[8,136–138],suggestingthattheFFArespondstothefacegestalt.Fourth,FFAresponsesaremodulatedbychangesinfacialfeaturesandconfiguration[139].Fifth,theadultFFAiscorrelatedwithdetection,categorizationandidentificationoffaces,butnotnon-faceobjects[10,11,136,138].Incontrast,activationsofface-selectiveregionsintheSTSdonotcorrelatewithrec-ognitionperformance[11].Sixth,theFFArapidlyhabitu-atestorepeatedpresentationofaface,suggestingthatitisinvolvedinfine-grainprocessingoffaces,notjustcate-gorydetection[12,134].Seventh,theFFAisactivatedmorebyfaceidentitythanvariationsineyegazeorexpression[9,12],whiletheSTSshowedtheoppositeeffects[9].Eighth,theFFA’sactivityismodulatedduringworkingmemory,encodingandrecognitiontasksinvolv-ingfaces[140–144]andpredictedsubsequentrecognitionmemoryforfaces[143].Takentogether,thesefindingssupportakeyrolefortheFFAinfacerecognition.How-ever,littleisknownaboutthespecificvisualpropertiesoffacesthatactivatetheFFA,whatneuralinteractionswithintheFFAunderliefaceprocessing,andhowaccu-mulatedexperiencemayshapetheFFA’sdevelopmentandfunction.

152G.Golaraietal./ClinicalNeuroscienceResearch6(2006)145–160

2.13.Competingmodelsofthefunctionalorganizationoftheoccipito-temporalcortex

ThediscoveryofcategoryselectiveregionssuchastheFFAledtoanintensedebateregardingtheorganizationoftheoccipito-temporalcortex.Adomain-specificmodelsuggeststhattheFFAisacorticalmodulededicatedtoprocessingfacialinformation[145,146].However,somestudieshavefoundthattheFFAisactivatedduringview-ingnon-faceobjectswithwhichsubjectshaveexpertise[147,148],suggestingaprocessmapmodelinwhichtheFFAisanexpertprocessingregion[149,150].Adistribut-ed-representationmodelquestionsthemodularityoftheoccipito-temporalcortexandsuggeststhatprocessingofallvisualformsisdistributedwidely[130,151].Althoughnoneofthesemodelsprovideafullaccountofhowhighlyselectivefunctionalregionsdevelop,allthreeareconsistentwiththehypothesisthatlowerfrequencyoffaceviewingorinattentiontofacesinautismhaslong-termeffectontheneuralsubstratesoffaceprocessingintheventralstream.ThishypothesisandtheneurallocusofASDface-pro-cessingabnormalitieswouldbebestaddressedbyexperi-mentsthatalsodiscriminatebetweenthesemodels.Thedomain-specificmodelwouldpredictthatanydeficitsinfacedetectionoridentityrecognitioninautismwouldbespecif-icallyassociatedwithfunctionalabnormalitiesintheFFA,whichwouldnotsubserveotherexpertprocessing.Thepro-cessmapmodelwouldpredictthatanyFFAfunctionalabnormalitiesinautismwouldbespecifictofacesbutnotothernon-faceobjectsofexpertise.Thedistributed-repre-sentationmodelwouldpredictthatthefunctionalabnor-malitiesassociatedwithidentityrecognitionwouldbeapparentacrossadistributednetworkintheventralstream,andnotconfinedtotheFFA.Thusprogressinunderstandingtheneuralbasisofidentityrecognitiondeficitsinautismmayprovideabetterunderstandingofthenormalorganizationoftheventralstream.2.14.EvidenceformaturationoftheFFAduringnormaldevelopment

Recentimagingstudieshaveexaminedface-specificacti-vationsintheoccipito-temporalcortexinhealthychildren,findingvariedevidenceforamaturationprocess.Forexample,aPETstudyfoundgreaterresponsestofacesthantogeometricshapesintheventral-occipito-temporalcortexoftwo-montholdinfants[152],whileothersreportedtheabsenceoffaceselectiveresponsesinthefusiformgyrusrelativetoobjectsorplaces(i.e.,theFFA)in5–8[82]and8–10yearolds[83]basedongroupanalyses.Thus,itwasunclearifthereportedreductionorabsenceoftheFFAactivationinchildrenunderagetenreflectslesserrespon-sivenesstofacesorgreaterresponsivenesstoobjectsinthefusiformgyrus,asmalleranatomicalsizeofthefusi-formgyrus,ornon-specificartifactssuchashigherlevelsoffMRIrelatednoiseinchildren.Furthermore,itwasnotknownifFFAmaturationspecificallyrelatesto

developmentoffacerecognitionperformanceorifitisuniquelyprolongedcomparedtootherhigher-levelvisualregionsspecializedforfaces(e.g.,STS).

ToaddressthesequestionsweexaminedthematurationoftheFFAafterageseven,usingastandard‘‘localizer’’tofindface-selectiveregionswithinindividualsubjects[84].WefoundstrikingevidenceforselectiveFFAmaturationduringthisperiod.First,therightFFAinchildrenages7–11wasabouthalfthesizeofthatofolderchildren(12–16yearolds),andaboutone-thirdthatofadults,evenaftermatchingagegroupsonseveralmeasuresoffMRIsignal-to-noise-ratio.Second,thesmallerFFAsizeinchil-drenwasstronglycorrelatedwithfacerecognitionmemory.Third,thedevelopmentalchangeswefoundweredomainspecific.Forexample,therewerenodifferencesamongagegroupsinobjectrecognition,orthesizeofobject-selec-tivecortexorSTS.

Thesefindingssupportanexperiencedependentmodel,inwhichtheFFAemergesduringaprolongeddevelop-mentalprocessinvolvingaccumulatedexperiencewithfaces[153–156].Thesefindingsalsoraiseanumberofquestionsthatarecurrentfociofresearch,namely:doesFFAmaturationinvolveincreasingfaceselectivityamongbroadlyfacetunedneuralelementsorincreasingnumbersoffaceselectiveelements?IsasmallerFFAinchildrenassociatedwithdifferencesinfaceviewingstrategiesorthespecifictypesofinformationthatchildrenextractfromfaces?WhatarethespecificfunctionsofthedevelopingFFAinchildren,especiallyduringthestagesoffacepro-cessingsuchascategorizationandidentification?2.15.FFAandfaceprocessinginASD

GiventheapparentroleoftheFFAincategorizationandrecognitionoffaces,severalstudieshaveexaminedtheFFAinASD.EarlystudiesofadultswithASDcom-paredtocontrolsfoundevidenceforlesserFFAactivationandgreaterobject-areaactivationbyfaces,eventhoughmeasuresofattentiontothefacesweresimilar[157–159].Fromtheprocess-mapperspective,thisapparentfailureofindividualswithASDtodevelopnormalcorticalfacespecializationintheFFAand‘‘expertise’’infacerecogni-tionmaybetheaccumulatedeffectofreducedsocialinter-estandlackofmotivationtoviewfaces[160,161].However,otherstudiesvariouslyfoundFFAactivationbyfacesthatwasmodulatedbypersonalfamiliarityinASD[162],orFFAactivationthatwasnotdistinguishablefromcontrols[163]orFFAactivationthatwasdifferentiallymodulatedbytaskamongASDsubjects,comparedtocontrols[164].

Usingeye-tracking,arecentstudysuggeststhatthisvarietyoffindingsmaybeexplainedbydifferencesinfixa-tionbehavior[165].AmongpersonswithASDbutnotcon-trols,activationofthefusiformgyrusappearstobestronglyandpositivelycorrelatedwiththetimespentfixat-ingontheeyeregionofthefacestimuli.Whethertheabsenceofthiscorrelationincontrolsisduetoceiling

G.Golaraietal./ClinicalNeuroscienceResearch6(2006)145–160153

effectsorbetweengroupdifferencesinFFA’ssensitivitytoviewingbehaviorisunknown.Inanyevent,thesestudiestogethersuggestthattheFFAdoesengageinfaceprocess-inginASDandinamannermodulatedbyfixationupontheeyes(mutualgaze),withimplicationsforthepossibledevelopmentalconsequencesofhabitualgazeavoidanceinASD.Theseimplicationsareconsideredbelow.

Consistentwiththedomain-specificmodeloftheventralstream,andexperiencedependentmodelsofitsdevelop-ment,onepossibilityisthathabitualgazeavoidance(andensuingchronichypo-activationoftheFFAduringdevel-opment)wouldresultinlong-termanomaliesinFFAfunc-tions,suchasfacedetectionandidentification,evenwhensubjectsdoviewtheeyesinfacestimuli.Consistentwiththeprocess-mapmodel,asecondpossibilityisthattheFFAinASDsubjectswouldbepredictedtoactivatetoadegreecorrelatingwithmeasuresofexpertisewhenprocess-ingfacesandotherobjects.Consistentwiththedistributedmapmodel,athirdpossibilityisthatfunctionalface-pro-cessingabnormalitiesinautismarenotconfinedtotheFFA,butaredistributedmorewidelyacrosstheventralstream.Inviewofthesevariousmodelpredictions,adetailedanalysisofthefunctionalpropertiesoftheFFAandtheventralstreaminASDmayprovidenewinsightsintotheroleofexperienceinthenormaldevelopmentandorganizationoftheoccipito-temporalcortex.2.16.DevelopmentoffacialemotionrecognitioninchildrenTherecognitionoffacialexpressionsofemotionisthoughttodevelopslowlyduringthefirsttwoyearsoflifeandcontinuetomatureintoadolescence.Somereportsindicatethatinthefirstseveralmonthsoflifeinfantscandiscriminatebetweenavarietyofemotionalexpressions[166–170].However,therearedifficultiesinassessingvisualdiscriminationinyounginfants.Forexample,infants’abil-itytodiscriminateamongsomeexpressionsissensitivetotheorderofpresentation,andmaybeaffectedbydifferen-tialratesofhabituationtospecificexpressions[166,167,171].Otherdatasuggestthataninfant’sdiscrimi-nationoffacialexpressionsisbasedonsimplefeaturaldif-ferences[172](butsee[173]).Forexample4to8-month-oldinfantsdiscriminated‘‘toothy’’similesfromclosed-mouthsmilesandclosed-mouthanger,butnotfromtoothyanger[172].Otherfindingssuggestthatdiscriminationofemo-tionalexpressionsmaybesensitivetoearlycontextualinformation.Forexample,inonestudy,3.5-month-oldinfantsdiscriminatedbetweenhappyorsadfacialexpressions(accompaniedbyaffectivelymatchingvocalexpressions)onlyiftheemotionalexpressionsweredis-playedbytheirownmotherintheexperimentalsettings[174](alsosee[175]).Takentogetherthesefindingssuggestthatduringthefirstmonthsoflife,processingoffacialexpressionsofemotionisatarudimentarylevel,andmaybesensitivetofamiliaritywiththefacialidentity.

Recognitionofemotionexpressionsimprovesbyasmuchas40%betweenages2and5,towithin10%ofadult

performanceinsomestudies,albeitnon-uniformlyfordif-ferentemotions.Children,ages4and5,recognizefacialexpressionsofhappiness,sadness,andangerinorderofdescendingaccuracy,andarelessaccurateinrecognizingsurprise,fearandneutralexpression[176–184].Also,chil-dren’smisjudgmentsoffacialexpressionsfollowsystemat-icpatterns.Forexample,childrenoftenconfusefacesdepictingsadnessandanger,orangeranddisgustandmis-judgeneutralfacesassad[178,179Reichenbach,1983#4630,182].Inonestudyofchildrenbetween2and5yearsold,BullockandRussell(1985)plottedmisjudg-mentsoffacialexpressionsonacontinuumoffacialexpressionsarrangedalongtwoaxesofpleasureandarousal,foundanarrowingofthedistributionoferrorsaroundthecorrectexpressionwithincreasingage[178].Theseauthorsinferredanarrowingofemotioncategoriesduringdevelopment[178],perhapsduetosocializationandexperiencewithfaces[185].

2.17.Theneuralsubstratesforrecognitionofemotionexpressions:Roleofamygdala

Althoughprocessingoffacialexpressionsofemotionisthoughttoinvolvenumerousbrainregionswithvaryingspecificity(reviewedin[186]),muchresearchhasfocusedontheroleofamygdala,duetoitsinvolvementinemo-tionallearning(e.g.,fearconditioning),emotionalmemo-ry(e.g.,‘‘flashbulb’’memories),andprocessingofsocialinformationinvolvingvariouscuesincludingvisualinputsofobjectsandfaces(reviewedin[187]).Forexample,amygdaladamageinhumansmayimpairtheabilitytoanthropomorphizemovinggeometricshapesthatnormalsubjectscaninterpretascharacterswithmotivesinteract-inginacomplexsocialsituation[188].Amygdaladamageisalsoassociatedwithimpairmentsinsocialjudgment,suchasoverestimationoftrustworthinessandapproach-abilityofpeoplebasedontheirfacesinrecognizingneg-ativeexpressions,suchasfear,anger,surprise,andsadness[189–197],butnotrecognizingfacialidentity[190].

Supportingaroleforamygdalainfaceprocessingandsocialjudgment,earlyneuroimagingstudiesfoundamygdalaactivationinresponsetofacialexpressionsofemotion,particularlyfear,evenwhensubjectswerenotinstructedtojudgeemotion[6,198].Indeed,amyg-dalaisactivatedbysubliminalpresentationsoffacialexpressionstimuli[199].Incontrast,verballylabelingfacialexpressionstimulimayreduceamygdalaactiva-tion[200].Ingeneral,theamygdalaisthoughttoenga-geinrapid,automaticprocessingoffacialexpressionsaspartofitsroleindetectingpotentialdanger[189,194].

Theeyesmaybeaparticularlysalientstimulusfortheamygdala.Forexample,Kawashimaetal.(1999)reportedleftamygdalaactivationwhensubjectsinterpretedgazedirection,whereastherightamygdalawasactivatedduringeye-to-eyecontact[201].Inanotherstudy,theeyeregion

154G.Golaraietal./ClinicalNeuroscienceResearch6(2006)145–160

wassufficienttoelicitamygdalaresponsesduringfMRI[202].Morespecifically,thewhitesclerasurroundingthedarkpupilinfearfuleyeswasanecessarycomponentofthestimulus.Consistentwiththesefindings,apatientwithbilateraldamagetotheamygdalafailedtouseinformationfromtheeyeregionwhenviewingfaces[191].Specifically,duringexposuretoisolatedsamples(‘‘bubbles’’)offacesoffearandhappinessthissubjectusedfacialinformationfromthemouth,butnottheeyes.Eyetrackingmethodssuggestedthatthisinabilitytouseeyeinformationwasrelatedtoalackoffixationontheeyeregion.Incontrast,normalsubjectsspontaneouslyfixatedupontheeyes,pref-erentiallyusingeyeinformationtorecognizefear.Whentheamygdaladamagedsubjectfollowedinstructionstofix-ateoneyes,theirabilitytorecognizefearfulfacesimprovedtoanormallevel.Thesefindingssuggestthatamydalafunctionmaybeimportantindirectinggazeandattentionontoeyes,asasourceofsocialandaffectiveinformation(Fig3).

Thesestudiesalsoraiseseveralquestions.Forexample,isthiscasestudyofeffectsofamygdaladamageonfearconditioningrelevanttoothercasesofamygdaladamageanddysfunction?Couldothersocialprocessingdeficitsafteramygdaladamage,suchasjudgingtrustworthinessoranthropomorphizingmovinggeometricshapes,arisefromacorrectablefailuretoattendtocriticalinformationinstimuli?Towhatextentcouldtrainingremediateineffec-tiveviewingstrategies?Howwouldageofonsetofamyg-daladamageordysfunction,anditsspecificcharacter,affecttreatmentstrategies?Thesequestionshavesignificantimplicationsforconditionssuchasautismthatinvolvefaceprocessingdeficitsandperhapsabnormalitiesofamygdalafunction.

2.18.Emotionexpressionsandamygdaladevelopment:Neuroimagingevidence

AfewfMRIstudiesofamygdalafunctionduringchild-hoodandadolescencehavefocusedonitsresponsestofearfulfacialexpressions[203–205].Bairdetal.(1999)examinedamygdalaactivationstofearfulfacesin12–17-year-oldsandreportednoeffectsofageorsex[203].Inanotherstudy,Thomasetal.(2001)comparedamygdalaactivationstofearandneutralfacesamongchildren(meanage11yearsold)andadults[205].Inadults,theamygdalarespondedmoretofearfacesthantoneutralfaces,asexpected,whereasinchildrentheamygdalarespondedmoretoneutralfaces.Althoughsubjectiveratingsoffacialexpressionwerenotobtained,theseresultswereinterpretedtoreflecttheambiguityofneutralfacesforchildren,consis-tentwithbehavioralstudiesfindingthatchildrenclassifyneutralfacesasexpressingnegativeemotions[179,181].Inaseparatestudy,Killgoreetal.(2001)examineddevel-opmentalchangesinneuralresponsetofearfulfacesinchildrenandadolescents,findingthatleftamygdalaresponsestofearfacesdecreasedduringadolescenceingirlsbutnotboys[204].Girlsalsodisplayedincreasingactiva-tionofthedorsolateralprefrontalcortexbyfearfacesdur-ingthisperiod,whereasboysshowedtheoppositepattern[204].Thesefindingssuggestthatamygdalamaturationmaybeaccompaniedbyageandsexdependentchangesoutsidetheamygdala,inneuralsystemsthatexertaregu-latoryinfluenceonitsfunctioninadulthood[206].Futurestudiesareneededtocharacterizethedevelopmentaltrajec-toryofamygdalaresponsestoawiderangeofemotionallysalientstimuli,includingfaces.Salientquestionsincludewhethertheadultpatternofpreferentialamygdala

Fig.3.(fromAdolphsetal.2005)Whenexposedtoisolatedfragmentsoffacesoffearandhappiness,normalcontrolsusedinformationmostlyfromtheeyeregion.Incontrast,anamygdaladamagedsubject(SM)usedinformationmostlyformthemouthregion.

G.Golaraietal./ClinicalNeuroscienceResearch6(2006)145–160155

responsestofearrelativetootheremotionalexpressionsdevelopsovertimeduringchildhood,andwhetherchildrenandadultsusesimilarviewingstrategiesinextractingfea-turalinformationwhenmakingemotionalorsocialjudg-mentsaboutfaces.

2.19.ImpairmentsoffacialexpressionrecognitioninASD:Arolefortheamygdala?

AnumberofreportssuggestdeficitsinrecognitionoffacialexpressionsamongpersonswithASD,startingasyoungas5yearsold.Forexample,whenchildrenwithASDandverbal-abilitymatchedcontrolssortedpicturesofpeople,whodifferedalongdimensionsincludingemo-tionalexpressionandthetypeofhat,mostchildrenwithautismfirstsortedpicturesaccordingtothetypeofhat,whereasmostcontrolssortedpicturesbasedonemotionexpressionsbeforesortingbasedonhats[207,208].AnotherstudyreportedthatchildrenwithASDwerelessableinmatchingaffectivefacesthanobjectscomparedtocontrols,andthedegreeofdeficitpredictedsocialimpairmentsamongtheASDgroup[209].ThesefindingssuggestthatemotionexpressionsarelesssalientforchildrenwithASDthanforcontrolsandthatthisdeficitmayberelatedtosocialdeficits.However,theseresultsmightbesensitivetochoiceofcontrols,asanotherstudyfoundthatchildrenwithASDshoweddeficitsinsortingemotionexpressionscomparedtocontrols,whenmatchedfornon-verbalmen-talagebutnotwhenmatchedforverbalmentalage[103].Finally,astudycomparedhighfunctioningchildrenwithASD(meanage10.6±2.1)toageandIQmatchedcontrols,andfoundnodifferencesinnamingofexpressions(angry,happy,neutral,andsurprise)orinfoveationduringfaceviewing[57].Thus,itremainsunclearhowtointerpretreportedemotion-expressionprocessingdeficitsinASD.Longitudinalstudiesofemotion-expressionrecognitionandintensitydiscriminationinchildrenwithASDandwell-chosencontrolgroupsareneededtoaddresstheseissuesandexpandourunderstandingofnormaldevelopment.

Nonetheless,thepredominanceofdatasuggeststhatrecognitionofemotionexpressionsisimpairedinadultswithASD.Forexample,highfunctioningASDadultswerereportedtoratefacesasmoretrustworthyandapproach-ablethandidnormalcontrols[210].Also,Pelphreyetal.(2002)foundthatrelativetoageandgendermatchedcon-trols,highfunctioningadultmaleswithASDmademoreerrorsinidentifyingemotionexpressions,beingmorelikelytomisidentifyfearasanger,surpriseordisgust,whileiden-tifyinghappiness,sadnessandsurpriseatcontrollevels[55].IndividualswithASDalsolookedlessattheeyes.Aseyeinformationhasbeenidentifiedascriticalforrecog-nitionoffearfaces[191]andactivationofamygdala(incaseoffearfaces)[202]adifferenceinfixationbehaviormayaccountfortheobservedfear-specificemotion-expres-sionprocessingdeficitamongsubjectswithASD.However,itremainsunclearwhetheramygdalahypo-activation

mightbethecauseoraresultoftheabnormalfixationbehavior[165].

SupportingthehypothesisthatamygdaladysfunctioncontributestofaceprocessingabnormalitiesinASDaresomecommonalitiesbetweenASDandpatientswithbilat-eralamygdalalesions(reviewedin[211,212]),includingdeficitsinfear-facerecognition,biastowardsratingfacesasmoretrustworthyandapproachable[210],atypicalpat-ternsoffaceviewing[55,56](butsee[57])andfailuretoattributesocialintentiontomovinggeometricshapes[213].Otherlinesofevidenceimplicatingamygdaladys-functioninASDincludeevidenceforamygdalaneuropa-thologyfrompost-mortemstudies[214],andautistic-likesocialandemotionalbehavioramongnon-humanprimateswithamygdalalesionsearlyduringinfancy[215](butsee[216]).Also,structuralMRIstudiessuggestabnormalitiesinthedevelopmentoftheamygdalainASD.Initialreportswerecontradictory,showingevidenceforsmaller[217],larger[218,219]andequivalent[220]amygdalavolumesinsubjectswithASDcomparedtocontrols.However,morerecentdatasuggestanunusualdevelopmentaltimecoursefortheamydalainASD,inwhichtheamydalaislargerinchildrenwithASD(ages7.5to12.5yearsold)comparedtoseveralcontrolgroups,whereasnodifferencesaredetect-ablebetweenadolescentswithASDversuscontrols[221].Itremainstobedeterminedifthisanomalousearlydevelop-mentleadstopersistentcompensatoryabnormalitiesinbrainregionsconnectedtotheamygdalaoraberrantintrin-sicconnectivityandfunctionoftheamygdala.2.20.NeuroimagingofamygdalafunctioninASD

AnumberoffMRIstudiessuggestanomalousamygdalaactivationinchildrenandadultswithASD.Forexample,Wangetal.(2004)examinedmodulationofamydalaacti-vationduringmatchingversuslabelingofemotionalfacesinchildrenandadolescentswithASD(ages8–16),com-paredtohealthycontrols(ages6.9to19.8)[222].Bothgroupsengagedsimilarneuralnetworksduringfacialemo-tionprocessing.However,incontrolsbutnottheASDgroup,amygdalaactivationwashigherduringemotionalfacematchingrelativetolabeling[222].Also,acasestudyofa12-year-oldboydiagnosedwithASDsuggeststhatamygdalaactivationinASDmaybesensitivetoexpertise,asamygdalahypo-activationwasfoundwhenviewingpic-turesofrealisticfacesbutnormallevelsofactivationwerefoundwhenviewingafamiliarandwelllikedcartoonchar-acter[161].

FMRIstudiesinadultshavefoundabnormallylowacti-vationintheamygdalaamongpersonswithASDrelativetocontrolsfortasksinvolvingemotionrecognitionfromeithertheeyeregionalone[223]orthefaceasawhole[224].Incontrast,anotherfMRIstudyreportedthatlevelsofleftamygdalaactivationinsubjectswithASDweregreaterthanincontrolsandpositivelycorrelatedwiththeaveragetimeoffixationupontheeyeregion,duringbothemotionandfamiliarityjudgmentsonfaces[165].Dalton

156G.Golaraietal./ClinicalNeuroscienceResearch6(2006)145–160

etal.(2005)interpretedtheseresultstosuggestthatviewingfaces,particularlyeyes,maybeassociatedwithheightenedarousalamongpersonswithASD,andthatamygdalahypo-activationmaybearesultofviewingstrategiesinthisgroup[165].Futurestudiesareneededtotestthishypoth-esis.Takentogether,thesefindingsemphasizetheimpor-tanceofcontrollingforviewingstrategiesandarousallevelswhenstudyingamygdalafunctioninfaceprocessing.3.Conclusions

Autismisapervasivedevelopmentalcondition,charac-terizedbyimpairmentsinnon-verbalcommunication,socialrelationshipsandstereotypicalpatternsofbehavior.Currently,theunderlyingcausesofautismarenotknown,althoughtheyarelikelytobequiteheterogenousandinvolvecombinationsofgeneticandenvironmentalfactors.Inthesearchforneuralmarkersofautism,muchinteresthasfocusedonanetworkofbrainregionsthatareimpli-catedinsocialcognitionandfaceprocessing.Inthisreview,wefocusedonthreesuchregions,namelytheSTSforitsroleinprocessinggazeandfacialmovements,theFFAinfacedetectionandidentificationandtheamygdalainprocessingfacialexpressionsofemotion.Alargebodyofevidencesuggeststhataspectsoffaceprocessingassociatedwiththeseregionsareimpairedinautism,includinganom-aliesingazeprocessing,memoryforfacialidentityandrec-ognitionofemotionexpressions.ThereisalsofMRIevidenceforabnormalactivationinbrainregionsunderly-ingeachoftheseneurocognitivefunctionsinautism.Recentdatasuggestthatsomeofthesefunctionalanoma-liesmayreflectatypicalfaceviewingstrategies,asmanyindividualswithautismdonotfixatenormallyupontheeyes,startingearlyinlife.Thesefindingsraisenewques-tionsregardingthedevelopmentofanomalousfacepro-cessinginautism.Forexample,couldatypicalgazebehaviorstartingearlyinlifeadverselyaffectthedevelop-mentofallorspecificcomponentsofthefaceprocessingnetwork?Recentfindingsonthenormaldevelopmentoffaceprocessingduringinfancy,childhoodandadolescencesuggestthatwhilesomeaspectsoffaceprocessingsuchasautomaticorientingtowardsfacesmayexistshortlyafterbirth,others,suchasidentityandemotionprocessingundergoprolongeddevelopment.Openquestionsinthisresearchareaarethebrainbasisofthesedevelopmentalprocesses,theroleofexperienceinshapingeachofthespe-cializedregionsforfaceprocessing,andtheimpactofeachregion’sfunctionuponthedevelopmentofothercompo-nentsofthefaceprocessingnetwork.Progressinunder-standingthenormaldevelopmentoffaceprocessingwillbeessentialforabetterunderstandingoffaceprocessinganomaliesinautism,andthepotentialtimingandeffectofearlyremediationfocusedonfaces.Inthiscontext,lon-gitudinalstudiesofcomponentsoffaceprocessingusingcombinationsofbehavioralandimagingmethodsduringtypicalandatypicaldevelopmentwillbeespeciallyvaluable.

Acknowledgement

ThankstoDr.A.C.Greenwoodforeditsandvaluablesuggestions.References

[1]PuceAetal.Temporalcortexactivationinhumansviewingeyeand

mouthmovements.JNeurosci1998;18(6):2188–99.

[2]AllisonT,PuceA,McCarthyG.Socialperceptionfromvisualcues:

roleoftheSTSregion.TrendsCognSci2000;4(7):267–78.

[3]WinstonJSetal.Automaticandintentionalbrainresponsesduring

evaluationoftrustworthinessoffaces.NatNeurosci2002;5(3):277–83.

[4]YoungAWetal.Facialexpressionprocessingafteramygdalotomy.

Neuropsychologia1996;34(1):31–9.

[5]AdolphsRetal.Fearandthehumanamygdala.JNeurosci

1995;15(9):5879–91.

[6]MorrisJSetal.Adifferentialneuralresponseinthehuman

amygdalatofearfulandhappyfacialexpressions.Nature1996;383(6603):812–5.

[7]KanwisherN,McDermottJ,ChunMM.Thefusiformfacearea:a

moduleinhumanextrastriatecortexspecializedforfaceperception.JNeurosci1997;17:4302–11.

[8]GeorgeNetal.Contrastpolarityandfacerecognitioninthehuman

fusiformgyrus.NatNeurosci1999;2(6):574–80.

[9]HoffmanEA,HaxbyJV.Distinctrepresentationsofeyegazeand

identityinthedistributedhumanneuralsystemforfaceperception.NatNeurosci2000;3(1):80–4.

[10]Grill-SpectorKetal.Thedynamicsofobject-selectiveactivation

correlatewithrecognitionperformanceinhumans.NatNeurosci2000;3(8):837–43.

[11]Grill-SpectorK,KnoufN,KanwisherN.Thefusiformfacearea

subservesfaceperception,notgenericwithin-categoryidentification.NatNeurosci2004;7(5):555–62.

[12]WinstonJSetal.fMRI-adaptationrevealsdissociableneural

representationsofidentityandexpressioninfaceperception.JNeurophysiol2004;92(3):1830–9.

[13]CareyS.Becomingafaceexpert.PhilosTransRSocLond

1992;335:95–103.

[14]ChanceJE,GoldsteinAG.Face-recognitionmemory:implications

forchildren’seyewitnesstestimony.JSocialIssues1984;40(2):69–85.[15]FomboneeE.Theepidemiologyofautism:areview.PsycholMed

1999;29:769–86.

[16]International,MolecularGeneticsStudyofAutismConsortium,A

fullgenomescreenforautismwithevidenceforlinagetoaregiononchromosome7q.HumMolGenet1998;7:571–578.

[17]LombrosoPJ,PaulsDL,LeckmanJF.Geneticmechanismsin

childhoodpsychiatricdisorders.JAmAcadChildAdolescPsychi-atry1994;33:921–38.

[18]CohenDJ,VolkmarFR.Handbookofautismandpervasive

developmentaldisorders.2nded.NewYork:Wiley;1997.

[19]LordCetal.Autismdiagnosticobservationschedule:astandard-izedobservationofcommunicativeandsocialbehavior.JAutismDevDisord1989;19:185–212.

[20]Willemsen-SwinkelsSH,BuitelaarJK.Theautisticspectrum:

subgroups,boundaries,andtreatment.PsychiatrClinNorthAm2002;25(4):811–36.

[21]Baron-CohenSetal.EarlyidentificationofautismbytheCHeck-listforAutisminToddlers(CHAT).JRSocMed2000;93(10):521–5.

[22]BairdGetal.Ascreeninginstrumentforautismat18monthsof

age:a6-yearfollow-upstudy.JAmAcadChildAdolescPsychiatry2000;39(6):694–702.

[23]OsterlingJ,DawsonG.Earlyrecognitionofchildrenwithautism:a

studyoffirstbirthdayhomevideotapes.JAutismDevDisord1994;24(3):247–57.

G.Golaraietal./ClinicalNeuroscienceResearch6(2006)145–160

157

[24]DawsonGetal.Childrenwithautismfailtoorienttonaturally

occurringsocialstimuli.JAutismDevDisord1998;28(6):479–85.[25]DakinS,FrithU.Vagariesofvisualperceptioninautism.Neuron

2005;48(3):497–507.

[26]GorenCC,SartyM,WuPY.Visualfollowingandpattern

discriminationofface-likestimulibynewborninfants.Pediatrics1975;56(4):544–9.

[27]KleinerKA,BanksMS.Stimulusenergydoesnotaccountfor2-month-olds’facepreferences.JExpPsycholHumPerceptPerform1987;13(4):594–600.

[28]JohnsonMHetal.Newborns’preferentialtrackingofface-like

stimulianditssubsequentdecline.Cognition1991;40(1-2):1–19.[29]FarroniTetal.Eyecontactdetectioninhumansfrombirth.Proc

NatlAcadSciUSA2002;99(14):9602–5.

[30]BatkiAetal.Isthereaninatemodule?Evidencefromhuman

neonates.InfantBehavDev2000;23:223–9.

[31]JohnsonMH.Subcorticalfaceprocessing.NatRevNeurosci

2005;6(10):766–74.

[32]MortonJ,JohnsonMH.CONSPECandCONLERN:atwo-process

theoryofinfantfacerecognition.PsycholRev1991;98(2):164–81.[33]CassiaVM,TuratiC,SimionF.Cananonspecificbiastowardtop-heavypatternsexplainnewborns’facepreference?PsycholSci2004;15(6):379–83.

[34]LevyIetal.Center-peripheryorganizationofhumanobjectareas.

NatNeurosci2001;4(5):533–9.

[35]GibsonJJ,PickAD.Perceptionofanotherperson’slooking

behavior.AmJPsychol1963;76:386–94.

[36]GrossmanED,BattelliL,Pascual-LeoneA.RepetitiveTMSover

posteriorSTSdisruptsperceptionofbiologicalmotion.VisionRes2005;45(22):2847–53.

[37]MichelsL,LappeM,VainaLM.Visualareasinvolvedinthe

perceptionofhumanmovementfromdynamicformanalysis.Neuroreport2005;16(10):1037–41.

[38]VainaLMetal.Functionalneuroanatomyofbiologicalmotion

perceptioninhumans.ProcNatlAcadSciUSA2001;98(20):11656–61.

[39]ServosPetal.Theneuralsubstratesofbiologicalmotionpercep-tion:anfMRIstudy.CerebCortex2002;12(7):772–82.

[40]CastelliFetal.Movementandmind:afunctionalimagingstudyof

perceptionandinterpretationofcomplexintentionalmovementpatterns.Neuroimage2000;12(3):314–25.

[41]FrithU,FrithCD.Developmentandneurophysiologyof

mentalizing.PhilosTransRSocLondBBiolSci2003;358(1431):459–73.

[42]HarrisLT,TodorovA,FiskeST.Attributionsonthebrain:neuro-imagingdispositionalinferences,beyondtheoryofmind.Neuroim-age2005;28(4):763–9.

[43]SchultzJetal.Activationofthehumansuperiortemporalgyrus

duringobservationofgoalattributionbyintentionalobjects.JCognNeurosci2004;16(10):1695–705.

[44]VollmBAetal.Neuronalcorrelatesoftheoryofmindand

empathy:afunctionalmagneticresonanceimagingstudyinanonverbaltask.Neuroimage2006;29(1):90–8.

[45]MosconiMWetal.Takingan‘‘intentionalstance’’oneye-gaze

shifts:afunctionalneuroimagingstudyofsocialperceptioninchildren.Neuroimage2005;27(1):247–52.

[46]KimuraIetal.Childrenaresensitivetoavertedeyesattheearliest

stageofgazeprocessing.Neuroreport2004;15(8):1345–8.

[47]RutterM,SchoplerE.Autismandpervasivedevelopmental

disorders:conceptsanddiagnosticissues.JAutismDevDisord1987;17(2):159–86.

[48]AdrienJLetal.Autismandfamilyhomemovies:preliminary

findings.JAutismDevDisord1991;21(1):43–9.

[49]VolkmarF,ChawarskaK,KlinA.Autismininfancyandearly

childhood.AnnuRevPsychol2005;56:315–36.

[50]OsterlingJA,DawsonG,MunsonJA.Earlyrecognitionof1-year-oldinfantswithautismspectrumdisorderversusmentalretardation.DevPsychopathol2002;14(2):239–51.

[51]DawsonGetal.Neurocognitiveandelectrophysiologicalevidence

ofalteredfaceprocessinginparentsofchildrenwithautism:implicationsforamodelofabnormaldevelopmentofsocialbraincircuitryinautism.DevPsychopathol2005;17(3):679–97.

[52]LeGrandRetal.Expertfaceprocessingrequiresvisualinputtothe

righthemisphereduringinfancy.NatNeurosci2003;6(10):1108–12.[53]SenjuAetal.Eyecontactdoesnotfacilitatedetectioninchildren

withautism.Cognition2003;89(1):B43–51.

[54]SenjuAetal.Reflexiveorientinginresponsetoeyegazeandan

arrowinchildrenwithandwithoutautism.JChildPsycholPsychiatry2004;45(3):445–58.

[55]PelphreyKAetal.Visualscanningoffacesinautism.JAutismDev

Disord2002;32(4):249–61.

[56]KlinAetal.Visualfixationpatternsduringviewingofnaturalistic

socialsituationsaspredictorsofsocialcompetenceinindividualswithautism.ArchGenPsychiatry2002;59(9):809–16.

[57]vanderGeestJNetal.Gazebehaviorofchildrenwithpervasive

developmentaldisordertowardhumanfaces:afixationtimestudy.JPsycholPsychiatry2002;43(5):669–78.

[58]DawsonGetal.Affectiveexchangesbetweenyoungautistic

childrenandtheirmothers.JAbnormChildPsychol1990;18(3):335–45.

[59]CohenILetal.Parent-childdyadicgazepatternsinfragileXmales

andinnon-fragileXmaleswithautisticdisorder.JChildPsycholPsychiatry1989;30(6):845–56.

[60]Willemsen-SwinkelsSHetal.Timingofsocialgazebehaviorin

childrenwithapervasivedevelopmentaldisorder.JAutismDevDisord1998;28(3):199–210.

[61]LevittJGetal.Protonmagneticresonancespectroscopicimaging

ofthebraininchildhoodautism.BiolPsychiatry2003;54(12):1355–66.

[62]WaiterGDetal.Avoxel-basedinvestigationofbrainstructurein

maleadolescentswithautisticspectrumdisorder.Neuroimage2004;22(2):619–25.

[63]BoddaertNetal.Superiortemporalsulcusanatomicalabnormal-itiesinchildhoodautism:avoxel-basedmorphometryMRIstudy.Neuroimage2004;23(1):364–9.

[64]PelphreyKA,MorrisJP,McCarthyG.Neuralbasisofeyegaze

processingdeficitsinautism.Brain2005;128(Pt5):1038–48.

[65]GarrettASetal.Here’slookingatyou,kid:neuralsystems

underlyingfaceandgazeprocessinginfragileXsyndrome.ArchGenPsychiatry2004;61(3):281–8.

[66]PascalisO,deSchonenS.Recognitionmemoryin3-to4-day-old

humanneonates.Neuroreport1994;5(14):1721–4.

[67]PascalisO,deHaanM,NelsonCA.Isfaceprocessingspecies-specificduringthefirstyearoflife?Science2002;296(5571):1321–3.[68]PascalisOetal.Plasticityoffaceprocessingininfancy.ProcNatl

AcadSciUSA2005;102(14):5297–300.

[69]CrossJF,CrossJ,DalyJ.Sex,race,age,andbeautyasfactorsin

recognitionoffaces.PerceptPsychophys1971;10(6):393–6.

[70]CrossJF,CrossJ.Age,sex,race,andtheperceptionoffacial

beauty.DevPsychol1971;5(3):433–9.

[71]CareyS.Developmentoffaceperception.In:DaviesG,EllisH,

ShephardJ,editors.Perceivingandrememberingfaces.NewYork:Academicpress;1981.p.9–38.

[72]ChanceJE,TurnerAL,GoldsteinAG.Developmentofdifferential

recognitionforown-andother-racefaces.JPsychol1982;112(1):29–37.

[73]ChungMS,ThomsonDM.Developmentoffacerecognition.BrJ

Psychol1995;86(Pt1):55–87.

[74]GoldsteinAG,ChanceJE.Memoryforfacesandschematheory.J

Psychol1980;105(1):47–59.

[75]KellyDJetal.Three-month-olds,butnotnewborns,preferown-racefaces.DevSci2005;8(6):F31–6.

[76]SangrigoliS,DeSchonenS.Recognitionofown-raceandother-race

facesbythree-month-oldinfants.JChildPsycholPsychiatry2004;45(7):1219–27.

[77]Valentine,T.,Face-spacemodelsoffacerecognition.2001.

158G.Golaraietal./ClinicalNeuroscienceResearch6(2006)145–160

[78]GolaraiGetal.Racialbiasinfaceprocessingcontinuestodevelop

inchildrenages7–11yearsold.SocCognNeurosciAbstr2006.[79]EllisHD,YoungAW.Facesintheirsocialandbiologicalcontext.

In:DaviesMetal.,editors.Faceandmind.Oxford:OxfordUniversityPress;1998.

[80]EllisHD.Thedevelopmentoffaceprocessingskills.PhilosTransR

SocLond1992;335:105–11.

[81]ValentineT.Aunifiedaccountoftheeffectsofdistinctiveness,

inversion,andraceinfacerecognition.QJExpPsycholHumExpPsychol1991;2:161–204.

[82]GathersADetal.Developmentalshiftsincorticallociforfaceand

objectrecognition.Neuroreport2004;15(10):1549–53.

[83]AylwardEHetal.Brainactivationduringfaceperception:evidence

ofadevelopmentalchange.JCognNeurosci2005;17(2):308–19.[84]GolaraiG,GhahremaniDG,Grill-SpectorK,GabrieliJD.

Evidenceformaturationofthe‘‘fusiformfacearea’’inchildrenages7–16.VisionSciSoc2005.

[85]CareyS,DiamondR,WoodsB.Thedevelopmentofface

recognition–amaturationcomponent?DevPsychol1980;16:257–69.

[86]CareyS,DiamondR.Developmentalchangesintherepresentation

offaces.JExpChildPsychol1977;23:1–22.

[87]FarahMJ,TanakaJW,DrainHM.Whatcausesthefaceinversion

effect?JExpPsycholHumPerceptPerform1995;21(3):628–34.[88]FarahMJetal.Whatis‘‘special’’aboutfaceperception?Psychol

Rev1998;105(3):482–98.

[89]GoldsteinAG,MackenbergEJ.Recognitionofhumanfacesfrom

isolatedfacialfeatures:adevelopmentalstudy.PsychonSci1966;6(4):149–50.

[90]GoldsteinAG.Learningofinvertedandnormallyorientedfacesin

childrenandadults.PsychonSci1965;3(10):447–8.

[91]Carey.Perceptualclassificationandexpertise.In:Perceptualand

cognitivedevelopment.AcademicPress;1996.p.49–69.

[92]TanakaJWetal.Facerecognitioninyoungchildren:whenthe

wholeisgreaterthanthesumofitsparts.VisCogn1998;5(4):479–96.

[93]BaenningerMA.Thedevelopmentoffacerecognition:featuralor

configurationalprocessing?JExpChildPsychol1994;57:377–96.[94]SchynsPG,BonnarL,GosselinF.Showmethefeatures!Under-standingrecognitionfromtheuseofvisualinformation.PsycholSci2002;13(5):402–9.

[95]SchynsPGetal.Aprincipledmethodfordeterminingthe

functionalityofbrainresponses.Neuroreport2003;14(13):1665–9.[96]GosselinF,SchynsPG.Bubbles:atechniquetorevealtheuseof

informationinrecognitiontasks.VisionRes2001;41(17):2261–71.[97]PurcellDG,StewartAL.Theface-detectioneffect:configuration

enhancesdetection.PerceptPsychophys1988;43(4):355–66.

[98]RollsETetal.Theresponsesofneuronsinthetemporalcortexof

primates,andfaceidentificationanddetection.ExpBrainRes1994;101(3):473–84.

[99]Grill-SpectorK,KanwisherN.VisualRecognition:assoonasyou

knowitisthere,youknowwhatitis.PsycholSci2005;16:152–60.[100]TaylorMJetal.ERPevidenceofdevelopmentalchangesin

processingoffaces.ClinNeurophysiol1999;110(5):910–5.

[101]KlinAetal.Anormedstudyoffacerecognitioninautismand

relateddisorders.JAutismDevDisord1999;29(6):499–508.

[102]BoucherJ,LewisV.Unfamiliarfacerecognitioninrelativelyable

autisticchildren.JChildPsycholPsychiatry1992;33(5):843–59.[103]OzonoffS,PenningtonBF,RogersSJ.Arethereemotionperception

deficitsinyoungautisticchildren?JChildPsycholPsychiatry1990;31(3):343–61.

[104]RobelLetal.Discriminationoffaceidentitiesandexpressionsin

childrenwithautism:sameordifferent?EurChildAdolescPsychiatry2004;13(4):227–33.

[105]LangdellT.Recognitionoffaces:anapproachtothestudyof

autism.JChildPsycholPsychiatry1978;19(3):255–68.

[106]HobsonRP,OustonJ,LeeA.What’sinaface?Thecaseofautism.

BrJPsychol1988;79(Pt4):441–53.

[107]TantamDetal.Autisticchildren’sabilitytointerpretfaces:a

researchnote.JChildPsycholPsychiatry1989;30(4):623–30.

[108]BoucherJ,LewisV,CollisG.Familiarfaceandvoicematchingand

recognitioninchildrenwithautism.JChildPsycholPsychiatry1998;39(2):171–81.

[109]BlairRJetal.Fractionationofvisualmemory:agencydetection

anditsimpairmentinautism.Neuropsychologia2002;40(1):108–18.

[110]DaviesSetal.FaceperceptioninchildrenwithautismandAsperger’s

syndrome.JChildPsycholPsychiatry1994;35(6):1033–57.

[111]BoucherJ,WarringtonEK.Memorydeficitsinearlyinfantile

autism:somesimilaritiestotheamnesicsyndrome.BrJPsychol1976;67(1):73–87.

[112]BoucherJ,LewisV.Memoryimpairmentsandcommunicationin

relativelyableautisticchildren.JChildPsycholPsychiatry1989;30(1):99–122.

[113]EimerM.Theface-specificN170componentreflectslatestages

inthestructuralencodingoffaces.Neuroreport2000;11(10):2319–24.

[114]EimerM,McCarthyRA.Prosopagnosiaandstructuralencodingof

faces:evidencefromevent-relatedpotentials.Neruoreport1999;10:255–9.

[115]BentinS,DeouellLY,SorokerN.Selectivevisualstreaminginface

recognition:evidencefromdevelopmentalprosopagnosia.Neurore-port1999;10(4):823–7.

[116]AllisonTetal.Electrophysiologicalstudiesofhumanfacepercep-tion.I:Potentialsgeneratedinoccipitotemporalcortexbyfaceandnon-facestimuli.CerebCortex1999;9(5):415–30.

[117]TaylorMJetal.Eyesfirst!Eyeprocessingdevelopsbeforeface

processinginchildren.Neuroreport2001;12(8):1671–6.

[118]McCarthyGetal.Electrophysiologicalstudiesofhumanface

perception.II:Responsepropertiesofface-specificpotentialsgeneratedinoccipitotemporalcortex.CerebCortex1999;9(5):431–44.

[119]MunteTFetal.Event-relatedbrainpotentialstounfamiliarfaces

inexplicitandimplicitmemorytasks.NeurosciRes1997;28:223–33.[120]BruceV,YoungA.Understandingfacerecognition.BrJPsychol

1986;77:305–27.

[121]PuceA,AllisonT,McCarthyG.Electrophysiologicalstudiesof

humanfaceperception.III:Effectsoftop-downprocessingonface-specificpotentials.CerebCortex1999;9(5):445–58.

[122]SommerW,SchweinbergerSR,MattJ.Humanbrainpotential

correlatesoffaceencodingintomemory.ElectroencephalogrClinNeurophysiol1991;79(6):457–63.

[123]EimerM.Event-relatedbrainpotentialsdistinguishprocessing

stagesinvolvedinfaceperceptionandrecognition.ClinNeuro-physiol2000;111:694–705.

[124]SommerW,KomossE,SchweinbergerSR.Differentiallocalization

ofbrainsystemssubservingmemoryfornamesandfacesinnormalsubjectswithevent-relatedpotentials.ElectroencephalogrClinNeurophysiol1997;102:192–9.

[125]CauquilAS,EdmondsGE,TaylorMJ.Istheface-sensitiveN170

theonlyERPnotaffectedbyselectiveattention?Neuroreport2000;11(10):2167–71.

[126]SugaseYetal.Globalandfineinformationcodedbysingleneurons

inthetemporalvisualcortex.Nature1999;400:869–73.

[127]O’ConnorK,HammJP,KirkIJ.Theneurophysiologicalcorrelates

offaceprocessinginadultsandchildrenwithAsperger’ssyndrome.BrainCogn2005;59(1):82–95.

[128]McPartlandJetal.Event-relatedbrainpotentialsrevealanomalies

intemporalprocessingoffacesinautismspectrumdisorder.JChildPsycholPsychiatry2004;45(7):1235–45.

[129]DawsonGetal.Neuralcorrelatesoffaceandobjectrecognitionin

youngchildrenwithautismspectrumdisorder,developmentaldelay,andtypicaldevelopment.ChildDev2002;73(3):700–17.

[130]HaxbyJVetal.Distributedandoverlappingrepresentationsof

facesandobjectsinventraltemporalcortex.Science2001;293(5539):2425–30.

G.Golaraietal./ClinicalNeuroscienceResearch6(2006)145–160

159

[131]HaxbyJV,HoffmanEA,GobbiniMI.Humanneuralsystemsfor

facerecognitionandsocialcommunication.BiolPsychiatry2002;51(1):59–67.

[132]PuceAetal.Face-sensitiveregionsinhumanextrastriatecortex

studiedbyfunctionalMRI.JNeurophysiol1995;74(3):1192–9.[133]IshaiAetal.Therepresentationofobjectsinthehumanoccipital

andtemporalcortex.JCognNeurosci2000;12(Suppl2):35–51.[134]Grill-SpectorKetal.Differentialprocessingofobjectsunder

variousviewingconditionsinthehumanlateraloccipitalcomplex.Neuron1999;24(1):187–203.

[135]FarahMJ,LevinsonKL,KleinKL.Faceperceptionandwithin-categorydiscriminationinprosopagnosia.Neuropsychologia1995;33(6):661–74.

[136]HassonUetal.Vaseorface?Aneuralcorrelateofshape-selective

groupingprocessesinthehumanbrain.JCognNeurosci2001;13(6):744–53.

[137]TongFetal.Binocularrivalryandvisualawarenessinhuman

extrastriatecortex.Neuron1998;21(4):753–9.

[138]RotshteinPetal.MorphingMarilynintoMaggiedissociates

physicalandidentityfacerepresentationsinthebrain.NatNeurosci2005;8(1):107–13.

[139]YovelG,KanwisherN.Faceperception:domainspecific,not

processspecific.Neuron2004;44(5):889–98.

[140]DruzgalTJ,D’EspositoM.Activityinfusiformfaceareamodulated

asafunctionofworkingmemoryload.BrainResCognBrainRes2001;10(3):355–64.

[141]HaxbyJVetal.Faceencodingandrecognitioninthehumanbrain.

ProcNatlAcadSciUSA1996;93(2):922–7.

[142]KimJJetal.Directcomparisonoftheneuralsubstratesof

recognitionmemoryforwordsandfaces.Brain1999;122(Pt6):1069–83.

[143]GolbyAJetal.Differentialresponsesinthefusiformregionto

same-raceandother-racefaces.NatNeurosci2001;4(8):845–50.[144]WiserAKetal.Novelvs.well-learnedmemoryforfaces:apositron

emissiontomographystudy.JCognNeurosci2000;12(2):255–66.[145]KanwisherN.Domainspecificityinfaceperception.NatNeurosci

2000;3(8):759763.

[146]DowningPEetal.Domainspecificityinvisualcortex.CerebCortex

2005.

[147]GauthierIetal.Activationofthemiddlefusiform‘facearea’

increaseswithexpertiseinrecognizingnovelobjects.NatNeurosci1999;2(6):568–73.

[148]GauthierIetal.Expertiseforcarsandbirdsrecruitsbrainareas

involvedinfacerecognition.NatNeurosci2000;3(2):191–7.

[149]GauthierIetal.Thefusiform‘‘facearea’’ispartofanetworkthat

processesfacesattheindividuallevel.JCognNeurosci2000;12(3):495–504.

[150]GauthierII.Whatconstrainstheorganizationoftheventral

temporalcortex?TrendsCognSci2000;4(1):1–2.

[151]IshaiAetal.Distributedrepresentationofobjectsinthehuman

ventralvisualpathway.ProcNatlAcadSciUSA1999;96(16):9379–84.

[152]Tzourio-MazoyerNetal.Neuralcorrelatesofwomanface

processingby2-month-oldinfants.Neuroimage2002;15(2):454–61.[153]deHaanM,PascalisO,JohnsonMH.Specializationofneural

mechanismsunderlyingfacerecognitioninhumaninfants.JCognNeurosci2002;14(2):199–209.

[154]JohnsonMH.Ontogeneticconstraintsonneuralandbehavioral

plasticity:evidencefromimprintingandfaceprocessing.CanJExpPsychol1999;53(1):77–91.

[155]JohnsonMHetal.Theemergenceofthesocialbrainnetwork:

evidencefromtypicalandatypicaldevelopment.DevPsychopathol2005;17(3):599–619.

[156]GauthierI,NelsonCA.Thedevelopmentoffaceexpertise.Curr

OpinNeurobiol2001;11(2):219–24.

[157]SchultzRTetal.Abnormalventraltemporalcorticalactivity

duringfacediscriminationamongindividualswithautismandAspergersyndrome.ArchGenPsychiatry2000;57(4):331–40.

[158]PierceKetal.Faceprocessingoccursoutsidethefusiform’face

area’inautism:evidencefromfunctionalMRI.Brain2001;124(Pt10):2059–73.

[159]HublDetal.Functionalimbalanceofvisualpathwaysindicates

alternativefaceprocessingstrategiesinautism.Neurology2003;61(9):1232–7.

[160]GrelottiDJ,GauthierI,SchultzRT.Socialinterestandthe

developmentofcorticalfacespecialization:whatautismteachesusaboutfaceprocessing.DevPsychobiol2002;40(3):213–25.

[161]GrelottiDJetal.fMRIactivationofthefusiformgyrusand

amygdalatocartooncharactersbutnottofacesinaboywithautism.Neuropsychologia2005;43(3):373–85.

[162]PierceKetal.Thebrainresponsetopersonallyfamiliarfacesin

autism:findingsoffusiformactivityandbeyond.Brain2004;127(Pt12):2703–16.

[163]HadjikhaniNetal.Activationofthefusiformgyruswhen

individualswithautismspectrumdisorderviewfaces.Neuroimage2004;22(3):1141–50.

[164]PiggotJetal.Emotionalattributioninhigh-functioningindividuals

withautisticspectrumdisorder:afunctionalimagingstudy.JAmAcadChildAdolescPsychiatry2004;43(4):473–80.

[165]DaltonKMetal.Gazefixationandtheneuralcircuitryofface

processinginautism.NatNeurosci2005;8(4):519–26.

[166]Young-BrowneG,RosenfeldHM,HorowitzFD.Infantdiscrimi-nationoffacialexpressions.ChildDev1977;48:555–62.

[167]SerranoJM,IglesiasJ,LoechesA.Visualdiscriminationand

recognitionoffacialexpressionsofanger,fear,andsurprisein4-to6-month-oldinfants.DevPsychobiol1992;25(6):411–25.

[168]NelsonCA,MorsePA,LeavittLA.Recognitionoffacialexpres-sionsbyseven-month-oldinfants.ChildDev1979;50(4):1239–42.[169]NelsonCA,DolginKG.Thegeneralizeddiscriminationoffacial

expressionsbyseven-month-oldinfants.ChildDev1985;56(1):58–61.

[170]NelsonCA,HorowitzFD.Theperceptionoffacialexpressionsand

stimulusmotionbytwo-andfive-month-oldinfantsusingholo-graphicstimuli.ChildDev1983;54(4):868–77.

[171]CaronAJ,CaronRF,MacLeanDJ.Infantdiscriminationof

naturalisticemotionalexpressions:theroleoffaceandvoice.ChildDev1988;59(3):604–16.

[172]CaronRF,CaronAJ,MyersRS.Doinfantsseeemotional

expressionsinstaticfaces?ChildDev1985;56(6):1552–60.

[173]KestenbaumR,NelsonCA.Neuralandbehavioralcorrelatesof

emotionrecognitioninchildrenandadults.JExpChildPsychol1992;54(1):1–18.

[174]Kahana-KalmanR,Walker-AndrewsAS.Theroleofperson

familiarityinyounginfants’perceptionofemotionalexpressions.ChildDev2001;72(2):352–69.

[175]BarreraME,MaurerD.Theperceptionoffacialexpressionsbythe

three-month-old.ChildDev1981;52(1):203–6.

[176]GrossAL,BallifB.Children’sunderstandingofemotionfromfacial

expressionsandsituations:areview.DevRev1991;11:368–98.

[177]BullockM,RussellJA.Preschoolchildren’sinterpretationoffacial

expressionsofemotion.IntJBehavDev1984;7:193–214.

[178]BullockM,RussellJA.Furtherevidenceonpreschoolers’interpre-tationoffacialexpressions.IntJBehavDev1985;8:15–38.

[179]FellemanEetal.Children’sandadults’recognitionofspontaneous

andposedemotionalexpressionsinyoungchildren.DevPsychol1983;19:405–13.

[180]CamrasL,AllisonAK.Children’sunderstandingofemotional

facialexpressionsandverballabels.JNonverbalBehav1985;9:84–94.

[181]ReichenbachL,MastersJ.Children’suseofexpressiveand

contextualcuesinjudgmentsofemotion.ChildDev1983;54:993–1004.

[182]WaldenT,FieldT.Discriminationoffacialexpressionsby

preschoolchildren.ChildDevelopment1982;53:1312–9.

[183]StifterC,FoxN.Preschoolchildren’sabilitytoidentifyandlabel

emotions.JNonverbalBehav1986;10:255–66.

160G.Golaraietal./ClinicalNeuroscienceResearch6(2006)145–160

[184]RussellJA,BullockM.Onthedimensionspreschoolersusetointer-pretfacialexpressionsofemotion.DevPsychol1986;22:97–102.[185]PollakSD,KistlerDJ.Earlyexperienceisassociatedwiththe

developmentofcategoricalrepresentationsforfacialexpressionsofemotion.ProcNatlAcadSciUSA2002;99(13):9072–6.

[186]AdolphsR.Neuralsystemsforrecognizingemotion.CurrOpin

Neurobiol2002;12(2):169–77.

[187]PhelpsEA.Emotionandcognition:insightsfromstudiesofthe

humanamygdala.AnnuRevPsychol2006;57:27–53.

[188]HeberleinAS,AdolphsR.Impairedspontaneousanthropomor-phizingdespiteintactperceptionandsocialknowledge.ProcNatlAcadSciUSA2004;101(19):7487–91.

[189]AdolphsRetal.Recognitionoffacialemotioninnineindividuals

withbilateralamygdaladamage.Neuropsychologia1999;37(10):1111–7.

[190]AdolphsRetal.Impairedrecognitionofemotioninfacial

expressionsfollowingbilateraldamagetothehumanamygdala.Nature1994;372(6507):669–72.

[191]AdolphsRetal.Amechanismforimpairedfearrecognitionafter

amygdaladamage.Nature2005;433(7021):68–72.

[192]AdolphsR,TranelD,DamasioAR.Thehumanamygdalainsocial

judgment.Nature1998;393(6684):470–4.

[193]AdolphsR,Baron-CohenS,TranelD.Impairedrecognitionof

socialemotionsfollowingamygdaladamage.JCognNeurosci2002;14(8):1264–74.

[194]AdolphsR,TranelD.Impairedjudgmentsofsadnessbutnot

happinessfollowingbilateralamygdaladamage.JCognNeurosci2004;16(3):453–62.

[195]SchmolckH,SquireLR.Impairedperceptionoffacialemotions

followingbilateraldamagetotheanteriortemporallobe.Neuro-psychology2001;15(1):30–8.

[196]AndersonAK,PhelpsEA.Perceivingemotion:there’smorethan

meetstheeye.CurrBiol2000;10(15):R551–4.

[197]AndersonAK,PhelpsEA.Expressionwithoutrecognition:contri-butionsofthehumanamygdalatoemotionalcommunication.PsycholSci2000;11(2):106–11.

[198]BreiterHCetal.Responseandhabituationofthehumanamygdala

duringvisualprocessingoffacialexpression.Neuron1996;17(5):875–87.

[199]WhalenPJetal.Maskedpresentationsofemotionalfacialexpres-sionsmodulateamygdalaactivitywithoutexplicitknowledge.JNeurosci1998;18(1):411–8.

[200]HaririAR,BookheimerSY,MazziottaJC.Modulatingemotional

responses:effectsofaneocorticalnetworkonthelimbicsystem.Neuroreport2000;11(1):43–8.

[201]KawashimaRetal.Thehumanamygdalaplaysanimportant

roleingazemonitoring.APETstudy.Brain1999;122(Pt4):779–83.[202]WhalenPJetal.Humanamygdalaresponsivitytomaskedfearful

eyewhites.Science2004;306(5704):2061.

[203]BairdAAetal.Functionalmagneticresonanceimagingoffacial

affectrecognitioninchildrenandadolescents.JAmAcadChildAdolescPsychiatry1999;38(2):195–9.

[204]KillgoreWD,Yurgelun-ToddDA.Sexdifferencesinamygdala

activationduringtheperceptionoffacialaffect.Neuroreport2001;12(11):2543–7.

[205]ThomasKMetal.Amygdalaresponsetofacialexpressionsin

childrenandadults.BiolPsychiatry2001;49(4):309–16.

[206]OchsnerKN,GrossJJ.Thecognitivecontrolofemotion.Trends

CognSci2005;9(5):242–9.

[207]JenningsWB.Astudyofthepreferencesforaffectivecuesin

autisticchildren.Memphis,TN:MemphisStateUniversity;1973,p.92-96.

[208]WeeksSJ,HobsonRP.Thesalienceoffacialexpressionforautistic

children.JChildPsycholPsychiatry1987;28(1):137–51.

[209]BravermanMetal.Affectcomprehensioninchildrenwithperva-sivedevelopmentaldisorders.JAutismDevDisord1989;19(2):301–16.

[210]AdolphsR,SearsL,PivenJ.Abnormalprocessingofsocial

informationfromfacesinautism.JCognNeurosci2001;13(2):232–40.

[211]PelphreyK,AdolphsR,MorrisJP.Neuroanatomicalsubstratesof

socialcognitiondysfunctioninautism.MentRetardDevDisabilResRev2004;10(4):259–71.

[212]Baron-CohenSetal.Theamygdalatheoryofautism.Neurosci

BiobehavRev2000;24(3):355–64.

[213]KlinA.Attributingsocialmeaningtoambiguousvisualstimuliin

higher-functioningautismandAspergersyndrome:thesocialattributiontask.JChildPsycholPsychiatry2000;41(7):831–46.[214]KemperTL,BaumanML.Thecontributionofneuropathologic

studiestotheunderstandingofautism.NeurolClin1993;11(1):175–87.

[215]MachadoCJ,BachevalierJ.Non-humanprimatemodelsof

childhoodpsychopathology:thepromiseandthelimitations.JChildPsycholPsychiatry2003;44(1):64–87.

[216]AmaralDG,BaumanMD,SchumannCM.Theamygdalaand

autism:implicationsfromnon-humanprimatestudies.GenesBrainBehav2003;2(5):295–302.

[217]AylwardEHetal.MRIvolumesofamygdalaandhippocampusin

non-mentallyretardedautisticadolescentsandadults.Neurology1999;53(9):2145–50.

[218]HowardMAetal.Convergentneuroanatomicalandbehavioural

evidenceofanamygdalahypothesisofautism.Neuroreport2000;11(13):2931–5.

[219]SparksBFetal.Brainstructuralabnormalitiesinyoungchildren

withautismspectrumdisorder.Neurology2002;59(2):184–92.

[220]HaznedarMMetal.Limbiccircuitryinpatientswithautism

spectrumdisordersstudiedwithpositronemissiontomographyandmagneticresonanceimaging.AmJPsychiatry2000;157(12):1994–2001.

[221]SchumannCMetal.Theamygdalaisenlargedinchildrenbutnot

adolescentswithautism;thehippocampusisenlargedatallages.JNeurosci2004;24(28):6392–401.

[222]WangATetal.Neuralcorrelatesoffacialaffectprocessingin

childrenandadolescentswithautismspectrumdisorder.JAmAcadChildAdolescPsychiatry2004;43(4):481–90.

[223]Baron-CohenSetal.Socialintelligenceinthenormalandautistic

brain:anfMRIstudy.EurJNeurosci1999;11(6):1891–8.

[224]CritchleyHDetal.Thefunctionalneuroanatomyofsocialbehav-iour:changesincerebralbloodflowwhenpeoplewithautisticdisorderprocessfacialexpressions.Brain2000;123(Pt11):2203–12.