1

Articletype:FocusArticle

Articletitle:Guidelinesforstudyingdevelopmentalprosopagnosiainadultsandchildren

Firstauthor:Fullnameandaffiliation;plusemailaddressifcorrespondingauthorKirstenA.Dalrymple*InstituteofChildDevelopment,UniversityofMinnesota,Minneapolis,USAkad@umn.eduSecondauthor:Fullnameandaffiliation;plusemailaddressifcorrespondingauthorRominaPalermo*SchoolofPsychology,andARCCentreofExcellenceinCognitionanditsDisordersUniversityofWesternAustralia,Crawley,Australiaromina.palermo@uwa.edu.au

Pleasenotethatbothauthorswouldliketobelistedas“correspondingauthors”.

2

Abstract

Developmental prosopagnosia (DP) is a neurodevelopmental condition characterized by severe face

identity recognition problems that results from a failure to develop the mechanisms necessary for

adequate face processing1. It occurs in children and adults with normal visual acuity, and without

intellectualimpairmentsorknownbraininjuries.Giventheimportanceoffacerecognitionindailylife,

and the detrimental effects of impaired face recognition, DP is an important area of study. Yet

conventionsforclassifyingindividualsasDPforresearchpurposesarepoorlydefined.Inthisfocuspaper

wediscuss:1)criteriaforanoperationaldefinitionofDP;2)testsoffacerecognition,andconventions

for classifying individuals asDP; and 3) important considerations regarding common associations and

dissociations, and cognitive heterogeneity in DP. We also highlight issues unique to studying DP in

children,arelativelynewendeavourthat isprovingtobean importantcomplementtotheworkwith

adults. Ultimately we hope to identify challenges researchers face when studying DP, and offer

guidelinesforotherstoconsiderwhenembarkingontheirownresearchpursuitsonthetopic.

3

Developmental prosopagnosia (DP) is a neurodevelopmental condition characterized by severe face

identityrecognitiondifficultiesthatoccurwhenfaceprocessingmechanismsfailtoadequatelydevelop1.

DPoccursintheabsenceofbraininjury,inpeoplewithoutintellectualimpairmentandwithotherwise

typicalvision.Sometimesreferredtoascongenitalprosopagnosia[1]3,DPhasbeenestimatedtoaffect2-

2.9%ofthepopulation4,5.FacerecognitiondeficitsinDPcanbeassevereasthosereportedinacquired

prosopagnosia,which results frombraindamage1. Studieson thepsychosocial consequencesofDP in

adultsandchildrenindicatethatDPcanhaveaprofoundimpactonsocialandpsychologicalfunctioning,

creating difficulties making and maintaining relationships, participating in social activities, as well as

increasedlevelsofanxiety6-8.Giventheimportanceoffacerecognitionindailylifeandthedetrimental

effects of DP, it is particularly important to discover what underlies face recognition deficits and, in

doing so, inform potential treatment strategies. Studying DP also has the potential to provide

informationaboutthedevelopmentofhigh-levelvision,andotherdevelopmentaldisorders.

In recent years therehasbeen a surgeof interest in developmental prosopagnosia.A Scopus

search for articles or reviews that make reference to developmental (or congenital) prosopagnosia

reveals216publicationssincetheyear2000,aspecialissueinCognitiveNeuropsychologyin2012,anda

forthcomingspecialissueintheQuarterlyJournalofExperimentalPsychology.Thiscomparestoonly13

publicationsbetween1960and1999.Despitethisinterest,conventionsforclassifyingindividualsasDP

forresearchpurposesarestillpoorlydefined,potentially limitingprogressinunderstandingDP.Inthis

focus paperwewill discuss how to best define DP and how to determinewhether an adult or child

should be classified as DP in research studies. We examine commonly used tests that tap different

aspects of the face recognition process, and discuss which of these seem most suitable for the

measurementofabilitiesinadults,andalsoimportantly,inchildren(seeTable1).Wealsodiscusstests

and questionnaires that measure associated abilities and disabilities (e.g., object processing, autism)

which areuseful indeterminingwhether the childor adult is likely tohaveabroaderdevelopmental

conditionratherthanamore“pure”caseofDP(Table1).Wearguethatitisimportantfortheprogress

of the field to study those with relatively circumscribed face deficits, and also those with broader

deficits,buttobetransparentastowhichindividualsand/orgroupsareincludedinthesample.

[1] “Congenital” prosopagnosia implies that face recognition difficulties were present at birth, while“developmental” prosopagnosia is agnostic as to when in the developmental process the difficultiesemerge. As such, some caseswhere brain injury occurred early in development and resulted in facerecognitiondifficultieshavebeenreferredtoasdevelopmentalprosopagnosia(e.g.,2).However,theseareexcludedfromourdefinition,whichfocusesonthosewithnoknownbraininjury.

4

OPERATIONALDEFINITIONOFDP

To develop a coherent body of research on DP, it would be useful to have relatively standard

diagnosticcriteriaacrossstudies.Onecriticalaspectofdefiningadisorderoffacerecognitionisdefining

face recognition itself.Face recognition involvesperceivinga face,encoding it intomemory,and then

subsequentlyretrievingthatmemorytodeterminewhetherthefaceisfamiliarandwhoitmightbe9.An

impairmentoffacerecognitioncouldresultfromafailureofanyoneoftheseprocesses.Therefore,like

other developmental disorders (e.g. autism10), there are likely to be subtypes of DP, rather than one

singleform.Inadults,someindividualswithDPhavenormalfaceperceptiondespitetheirfacememory

deficits11-15.Thismayalsobethecaseforchildren,althougharecentstudysuggeststhatdeficitsinboth

face perception and memory are more common in children than adults12. To accommodate this

heterogeneity,DPisbestdefinedasadeficitoffacememory,withnormalorabnormalfaceperception.

Thereverse,impairedfaceperceptionwithpreservedfacememory,hasbeenreportedintheliterature

onacquiredprosopagnosiaforfacesthatwereseenpriortobraindamage16.However, inDP, itseems

unlikelythatanindividualwhosefaceperceptionhasalwaysbeenimpairedwouldhavetheopportunity

toacquireaccuratefacememories,makingfacememoryandfaceperceptiondifficultdisentangleinDP.

Facememorydeficitscanbemeasuredusingtestsoffacerecognition(seebelow;Table1),but

DPshouldalsobecharacterisedbytheexperienceoffacerecognitiondifficultiesindailylife.Forsome

individualswith DP these experiences include failures identifying very familiar people, such as family

members,whereasotherindividualswithDPprimarilyreportfailuresrecognisingfamiliarbutlessoften

seenpeople,suchasneighbours,whenseenoutofcontext (e.g.,at themall). IndividualswithDPare

also less likely to report a sense of familiarity for unrecognised faces17. However, self-report is

subjective,both in termsof thereportandthe interpretationof thereport,and individualsmayhave

onlyminimalinsightintotheirfacerecognitionabilities18(butsee19).Forexample,failingtorecognizea

familiarpersonoutofcontextisarelativelyubiquitousexperienceandthereisnostandardforwhenan

anecdotalreportcrossesintoevidenceofpathologyinfacerecognition.Furthermore,someindividuals

withDPwereunawarethatthefacerecognitiondifficultiestheyexperiencedwerenotsharedbyothers

until theyparticipated in testing sessions or learnedaboutDP and realized that it explainedmanyof

their lifeexperiences4,20.Unlikeacquiredprosopagnosia, inwhichan individual is typicallyaware that

they have lost an ability they formerly possessed, individuals with DP do not have experience with

normalfacerecognitiontoprovideapointofcomparison1.Children,inparticular,maylackawarenessof

their difficulties hence it is often parents who provide the key anecdotal evidence of their child’s

everyday face difficulties based on observations they have made6. Thus in children, identifying DP

5

requireskeenobservationbyparentsorteachers,combinedwiththeknowledgeofthedisorder,and/or

thetime,resources,andskillsettoresearchthesignsandsymptoms.Recently,severalresearchershave

attempted to improve the assessment of everyday face difficulties by designing self-report

questionnaires.A15-itemquestionnaire19containsquestionsonfaceidentityrecognition,aswellothers

onotheraspectsof faces suchas attractivenessandemotions.Themore recentProsopagnosia Index

(PI20)consistsof20 itemsspecifictoface identityrecognition,andmayproveuseful inscreeningand

standardisingtheassessmentofeverydayfacerecognitionproblems21.Nosuchscalesareyetavailable

forchildren.

We recommend that adefinitionofDP includeobjectivelypoorperformanceon testsof face

memory (just ‘how poor’ is discussed below), along with a subjective feeling of repeated face

recognitionfailuresindailylife.Testsareneededbecauseevenifpeopleknowthattheyarepoorerthan

averageatfacerecognition,theymaynothaveinsightintothedegreeoftheirdeficit18(i.e.,whetherthe

classificationwouldbeasa“poorrecogniser”ratherthanthemoresevereDP).Alternatively,ifaperson

performs at DP levels onmultiples tests of face recognition but reports no everyday difficulties (via

interview and/or questionnaire) then they would seem qualitatively different to people who are

classifiedasDP.Assuch,peoplewhoareidentifiedvialarge-scalescreeningexercisesthatonlyinclude

face recognition tests and nomeasure of subjective difficulty should only be classified as “potential”

DPs.Thereisalsotheinterestingcaseofpeoplewhoreporteverydaydifficultiesbutperformattypical

levelsonmultipletestsoffacerecognition.TheseindividualsaretypicallyexcludedfromDPsamplesbut

itwouldbeinformativetodeterminewhythisoccurs.

An additional definitional factor that is often raised regarding DP is whether the recognition

deficitisspecifictofacesoralsoaffectstheprocessingofotherobjects.Weexpectthatindividualswith

DPwouldbeabletorecogniseobjectsatthe“basic”level(e.g.,bicycle,car,etc.)andthisistypicallythe

case22 when they are asked to name pictures of line drawn objects in the Birmingham Object

RecognitionBattery (BORB)23 (althoughnote that these typesof tests can suffer fromceilingeffects).

However, it is less clear whether they should be proficient at distinguishing similar non-face objects

(e.g., different bicycles). There are cases of “pure”DPwhere the individual does not appear to have

difficulty differentiating between non-face similar objects24-26, and the double dissociation of an

individualabletorecognisefacesbutnotobjects27.However,somepeoplehavedifficultywithfacesand

objects24,28,29.WearguethatthesepeopleshouldstillbeconsideredDP.However, inthesecases it is

importanttodemonstratethatthe individualcanperformnormallyonsomecognitivetaskstoensure

that poor scores cannot be explained by general deficits of cognitive functioning or a failure to

6

understandtaskinstructions,andthattheyhaveadequatevisualacuity(seeTable1forexampletasks).

Thisisparticularlyimportantwhenworkingwithchildren,withwhomattentionandeffortlevelsshould

alsobecarefullymonitored.Moreover,itwouldbeusefulforstudiestotryanddifferentiatethosewith

more face-specific problems from those with more general visual memory impairments, as the

aetiologiesofthesemayvary.

TESTSOFFACERECOGNITION

One difficultywith studying DP is the lack of well-designed clinical tests that can be used to

reliablyevaluatefacerecognition4,20.Togetacompletepictureoffacialidentityrecognitionabilities,it

is importanttousemultiplemeasuresoffacememory.Inaddition,faceperceptiontasksareusefulto

helpqualifythenatureoftheidentityrecognitiondeficits.

Facememory

Assessmentsoffamiliarfacememoryareanobviouschoiceoftaskbecauseoftheirecological

validity.Photosofpersonallyfamiliarindividualsprovidedbyafamilymemberortakenforthepurpose

of the task are ideal, particularlywhen controlled for extra facial cues like hair, jewelry, glasses, and

clothing.Drawbacksofthistypeoftestarethatitcanbedifficulttoacquirealargenumberofcarefully

controlledphotographs,theindividualcanoftendeducewhowillbeincludedinthetest,andthetime

andeffortrequiredtomakecustomizedtestsforeachindividual.Additionally,itisdifficulttocontrolthe

leveloffamiliarity;someimagesmaybeofclosefriendsandfamily,whileothersmaybeofco-workers,

teachers,or acquaintances. Familiar face tasksalso requirea special control groupof individualswho

are similarly familiar with the individuals featured in the test, making it difficult to get an adequate

controlsampletocomparetotheDP’sperformance.

Famousfacesprovideanalternativemethodthatcircumventsmanyoftheseissuesandcanbe

usedformultipleparticipantstoenablecomparisonacrossDPsandwithacontrolgroup.However,the

issueofdegreeoffamiliaritymuststillbetakenintoaccountandparticipantsshouldbetestedlaterwith

thenamesofthecelebritiesinthetasktodeterminewhethertheyhavehadexposuretothefaceofthe

individual and could reasonably be expected to recognize them. Many famous people (especially

politiciansandsportspeople)aretypicallyonlyfamousinoneortwocountries,whichmeansthatthese

facescannotbeusedacrosscountries(althoughthetestformatscanbethesame).Itisalsoimportant

tonote that someDPs report that theyhave little interest in celebritiesorwatchingTVbecause they

find it difficult to recognize faces. These issues are exacerbated when designing tasks for children:

7

finding famous individualswhoare familiar tochildrenofawiderangeofages ischallengingbecause

differentcohortsareoftenexposedtodifferentcelebrities.Anotherissuewithfamousfacetestsisthat

participantsmayperformwellbyusingcompensatorystrategies,suchasidentifyingdistinctivefeatures

(e.g.,abeautyspotonMarilynMonroe),whichmaybemorecommononveryfamousfaces.Asaresult,

apoorscoreisoftenmoreinformativethanahighscore:ifanindividualperformspoorlyatidentifying

thefacesinafamiliarfacetaskdespitedemonstratingfamiliaritywiththeidentitiesinthetestbasedon

names,thisallowsconfidencethatthepersonis impaired.However,ahighscoreonthetaskcouldbe

explainedbyalternativerecognitionstrategies(e.g.memoryfordistinctivetraitsoffamiliarfaces).

Unfamiliar face memory tests can supplement tests of familiar face memory, and provide a

common starting point for all participants because level of familiarity is controlled4. Some existing

clinical tasks for adults such as theWarrington RecognitionMemory for Faces Test (RMF)30 and the

BentonFacialRecognitionTest(BFRT)31havebeenformallyevaluatedanddeemedtobepoormeasures

of true face recognition32,33. Individualswith impaired face recognition candowell on these tasksby

usingextra-face cues (e.g. theRMF includeshair and clothing), or bymatching images (e.g. theBFRT

includesdistinctivehairlinesandusessimultaneouspresentationwithunlimitedresponsetimes).Newer

tasks,suchastheCambridgeFaceMemoryTest(CFMT)34werecarefullydesignedtoreduceextra-face

cuesaswellasother imagecuessuchas lightingandlow-levelvisualproperties. IntheoriginalCFMT,

participantslearnsixfacesofadultmalesandaretestedwithathree-alternativeforcedchoiceformat.

There are three parts that increase in difficulty as the test progresses: the introduction involves

immediaterecallofindividualfaceswhilelatersectionsinvolverecallofanyofthesixtargetfacesand

introduce changes in lighting and viewpoint and finally added visual noise (Figure 1a). Bowles and

colleagues4foundasignificantdeclineinperformanceontheCFMTaftertheageof50,soperformance

onthesetasksshouldbecomparedtoage-matchedcontrols.Alternatively,age-adjustedz-scorescanbe

calculatedfromtheformulascontainedinBowlesetal.4TheCFMTisfreelyavailabletoresearchersand

iscommonlyuseddueto itsvalidityandreliability,whichmakes ituseful inscreeningforDP14,34,35. In

addition,severalversionsexisttoallowfollow-uptestingandtestingofdifferentraces(CFMT-Chinese36)

and ethnicities (CFMT-Australian14). Another advantage of using the CFMT is that there arematched

tests for within-category object recognition – the Cambridge Car Memory Test (CCMT)37 and the

Cambridge Bicycle Memory Test (CBMT)38 (Figure 1b) although note that other tests of object

recognitionareavailablee.g.,TheVanderbiltExpertiseTest(VET)39.

Likealltests,theCFMTdoeshavelimitations.Somepeopleperformpoorlyontestsoffamous

face recognition, but not on the CFMT, perhaps because they have used effective compensatory

8

strategies, leading to an ambiguous classification14, 40. In these cases, administering two different

versionsoftheCFMTcanhelpreachaconsensus13.Somepeoplewillperformpoorlyononeofthetwo

versions,increasingthelikelihoodofacorrectdiagnosisofDP(e.g.,someAustraliansperformpoorlyon

theCFMT-AustralianbutnottheoriginalCFMT,possiblybecausethefacesintheformertestmatchthe

ethnicity of the faces most commonly experienced in Australia and make it more difficult to match

distinctive features)14. Alternatively, others will perform at “poor”, but not DP levels on either test,

excluding a diagnosis of DP14. For some of these cases, the short delays betweenmemorization and

recognition in theCFMTmaymaskdifficulties retaining faces over longdelays (i.e., they shouldhave

difficulty with famous faces but not other tasks)14, 41. The CFMT-Australian includes a 20 min delay

sectionthatmayproveusefulindisambiguatingthesecases13.Asnotedearlier,thefacesusedinCFMT

tasksarecarefullycontrolled,andithasbeenarguedthatthesekindsofexperimentalmanipulationsare

likelytoaltertheprocessingofunfamiliarfacesandthattestsshouldalsobedesignedwithmore“real”

faces (e.g., that includehair)42. It is of interest here to note thatwedesigned a versionof theCFMT

(CFMT-Films)where the study stage involveswatching short film clips of people interacting, and this

correlateshighlywiththeoriginalCFMT17.

Forchildren,somegeneralneuropsychologicalassessmentbatteriesdoincludefacerecognition

subtests43,44,however,liketheRMTandtheBFRT,theyincludesuperficialcuesforrecognition(e.g.,hair

andfacesofmultipleraces)allowingchildrentoinferthecorrectanswersusingextra-facialinformation

ratherthannormalfacerecognitionmechanisms.NewtasksoffacerecognitionsuchastheCambridge

Face Memory Test for Children (CFMT for Children)45 and the Cambridge Face Memory Test – Kids

(CFMT-Kids)20(Figure1c)havebeendesigned.Althoughtheybothfollowthesamegeneralprocedures

astheoriginalCFMTforadults,theCFMTforChildrenusesadultfaces,fivetargets,andtwo-alternative

forcedchoiceformat,whiletheCFMT-Kidsuseschildren’sfaces(meanage=9.3years),sixtargets,and

three-alternativeforcedchoiceformat.Theexistenceofown-agebiasesinrecognition46suggeststhatit

maybepreferabletousechildren’sfacesintestsforchildren.However,giventhatthesebiasesappear

tobedrivenby experience47, the facesof young adultswould alsobe suitable. The advantageof the

formeristhatitiseasierforchildren,buttheadvantageofthelatteristhatchancelevelperformanceis

33%,allowingalargerrangeofscoresabovefloor(i.e.scorestwostandarddeviationsbelowthecontrol

meanyetabovechancearepossible,seeFigure2).Othersimpletestsofunfamiliarfacememoryexist,

such asOld/NewFaces48. This task involvesmemorizing ten target faces and thendeterminingwhich

fromasetoftwoisatargetface(i.e.the“old”face).Thetaskiseasytocreateandquicktoadminister.

Itisalsoeasytocreatematchedtasksusingdifferentclassesofobjects26(e.g.,horses,cars,guns,tools,

9

houses,glasses,etc.)thoughoneweaknessofthesetestsisthattheycurrentlyusethesameimagesfor

studyandtest.

Faceperception

TheCambridgeFacePerceptionTest(CFPT)49isausefultoolforassessingfaceperceptionability

inadultparticipants(Figure1d).Itinvolvessortingadultmorphfacesonacontinuumfrommostliketo

leastlikeatarget.Themorphfaceshelpavoiddirectcomparisonormatchingoffeaturesfromthetarget

facetothesortingfaces.Additionally,thetargetfaceispresentedat3/4-profileviewwhilethesorting

facesarepresentedinfrontalviews.AlimitationoftheCFPTisthat itrequiresusingthemousetore-

order thecontinuum,sopoor scorescould reflectpoormotivation to shift the faces ina time-limited

(oneminutepertrial)task.PerformanceontheCFPTdeclinesbymiddleage,andpastyoungadulthood

femalesperformbetter thanmales4.Therefore,both sexandage shouldbe taken intoaccountwhen

administering the CFPT (age- and sex-adjusted z-score calculations for the CFPT are available4). Even

whenusingage-matchedcontrols,standardtaskssuchastheCFPTshouldnotbeusedtoevaluateface

perceptioninindividualsover80becauseofflooreffects4.Forchildren,theDartmouthFacePerception

Test(DFPT)12involvesdecidingwhichofthreechildfaceslooksthemostlikeatargetfacethatremains

at thetopof thescreen(Figure1e).LooselybasedontheCFPT, thetarget face ispresentedata3/4-

profileviewandthechoicefacesarefrontalviewstakenfromamorphcontinuumbetweenthetarget

faceandadistractorface.Themorphstrengthsofthechoicefacesweretitratedtosystematicallyvary

taskdifficulty.

Oneconcernwithfaceperceptionteststhatusemorphsisthatthefacesareunnaturallysimilar

andcontainelementsofthetargetface,reducingtheecologicalvalidityofthetask. Incontrasttothe

CFPT andDFPT, theGlasgowFaceMatching Test (GFMT)50 doesnot usemorphs and thus avoids this

issue. In this task, participants are asked to determine whether face pairs are the same individual

(match), or different individuals (mismatch), with no time limit. Faces are frontal views with neutral

expressions. Matched faces are images of the same individual, taken with different cameras several

minutes apart.Mismatched face pairs were chosen based on similaritymeasures generated through

pilotwork.Inadditiontoretainingecologicalvaliditythatmaybeabsentintasksthatusemorphs,one

advantageoftheGFMTisthatitprovidesaformatthatcouldbeeasilymatchedtoanobject-matching

task(tasksinvolvingmorphsaremoredifficulttoconstructusingobjects,thoughithasbeendone51).

Specialconsiderationsforthedesignoftasksforchildren

10

One major challenge for designing face recognition tasks for children is achieving a level of

difficulty thatwill be suitable for all ages: above floor for young children, but below ceiling for older

children.These issuesaremorepronouncedfor testsof facememorybecauseunlike faceperception,

whichdevelopsatthesamerateasperceptionforotherobjects,facememoryappearstodevelopmore

slowlythanmemoryforotherobjects51.Thisdifferencemakesitdifficulttodesignmatchedtestsofface

andobjectmemory thatwill helpevaluate face-specificity: anobjectmemory task that ismatched in

difficultytoafacememorytaskforolderchildrenmaybepoorlymatchedforyoungerchildrenandvice

versa.However,creatingmultipleversionsofatask,forexampleaneasierversionofthefacememory

taskforyoungerchildren,leadstootherdifficulties:1)determiningwhichtaskshouldbeadministered

toanindividualwhoisatanintermediateage,ordeterminingafixedagerangeforeachtest;2)trying

toestablishadevelopmental trajectory for the task (i.e. comparingkidsacrossawide rangeofages);

and3)comparinganindividualchild’sscoreovertime:iftheindividualwasoriginallyevaluatedonatest

foryoungerchildren,whichtestshouldbeadministeredwhenthechildhasagedoutsidethecut-offfor

the original test? Amove to amore difficult test couldmask improvement, yet re-administering the

originaltestforyoungchildrencouldleadtoinflatedscores.

We have been confronted with the above difficulties when designing our tests of face and

objectprocessingforchildren.Wehavedesignedmultipletestsoffacememory(CFMT-Kids20,Old/New

Faces),matchedobjectmemorytasks(CBMT38,Old/NewFlowers12),andatestoffaceperception12(see

Table1).TheCFMT-Kids20seemstobeausefultoolforassessingfacememoryinchildren10andabove,

butsuffers fromflooreffects inyoungerchildren.Wedesignedanewversionof thetask foryounger

childrenbyreducingthenumberoftargetsfromsixtofour,but it isstillclosetofloor inyoungerage

groups. To address this difficulty,we designed a CFMT-style test that involves immediate recall only.

Althoughthisnewtestimprovesscoresacrosstheboard,itleadstoceilingeffectsinolderchildrenand

doesnotevaluatetheabilitytomaintainmultipleidentitiesinmemoryatonetime(Figure2).

Incontrasttothefacememorytasks,ourothertasks,suchastheDartmouthFacePerception

Test (DFPT)12,havebeenuseful fortestingchildrenbetweentheagesof7to12years.Controlmeans

areareasonabledistancefromfloorandceiling,andstandarddeviationsaresmallenoughthatthetask

maybeusefulforalargeragerange.Asaresult,wehavebeenplacingparticularemphasisonimpaired

face perception, combined with poor face memory scores and evidence of difficulties with face

recognitionindailylife,whenidentifyingprosopagnosiainyoungchildren6.

STATISTICALGUIDELINES

11

One popular criterion for neuropsychological impairment is a score that fallsmore than two

standarddeviationsbelowthemeanforage-matchedcontrols34.Intermsoffacerecognition,scoreson

the CFMT and CFPT that are greater than two standard deviations below the control mean tend to

correspondwithclinicallysignificantcomplaintsof facerecognitiondifficulties4,52.While twostandard

deviationsbelowthemeanisausefulguideline,statisticaltechniqueshavebeendevelopedtoformally

testtheprobabilitythatasingle individual’sscore istakenfromthecontrolgroup53. If the individual’s

scoreisstatisticallyunlikelytocomefromthegroup(i.e.p<0.05),theindividualisconsideredimpaired

for the ability that is being measured. Software to perform these calculations using the group size,

mean, standard deviation, and individual’s test score can be downloaded for free (SINGLIMS)54,55. To

corroborate case-by-case diagnoses of DP, we suggest reporting whether scores fall more than two

standard deviations below themean andwhether theymeet statistical significance12,56. Similarly,we

suggestusingbothstatisticalmethodswhenreportingperformanceonother tasks (e.g.objects, facial

expression,gender).

As discussed above, identifying cases of DP in children can be particularly challenging due to

floor effects. Children produce variable results both between subjects, and even trial-to-trial within

subjects57,58. If thecontrolmeanonagiventask is lowandthestandarddeviation ishigh, itbecomes

difficult to detect a score that ismore than two standard deviations below themean (see Figure 2).

These factors similarly affect the outcome of statistical tests. It is difficult to advise how to classify

children who are on the cusp or who have ambiguous results, but it is recommended that case

descriptionsprovideasmuchinformationaspossibletojustifyinclusionorexclusionofaparticularcase

from theDP group.A combinationof anecdotal reports of failures of face recognition in daily life, in

addition to multiple test scores of face memory and face perception compared to a two standard

deviationfromthemeancut-offandformalstatisticalanalysessuchastheCrawford,Garthwaite,and

Howell53modifiedstatisticsisappropriate.Thesestatisticaltestscanbeusedtotestforadeficitonone

test relative to a control group (SINGLIMS)54 55 or for dissociations in performance on two tests

(DiffBayes_ES_CP.EXE software)59-61. Ideally, when drawing conclusions with regards to a given

hypothesis,somecaseswillprovideclearresults.However,casesthataremoreambiguousshouldalso

beincludedtopresentamorecompletepictureofthebehaviorsbeingstudied.

ASSOCIATIONSANDDISSOCIATIONS

Formostresearchstudies,theidealparticipantswouldhavepureprosopagnosia(i.e.difficulties

withfacialidentityrecognitionalone).However,researchersdonotappeartoroutinelyreportwhether

12

the people in their DP sample show impairments with other aspects of face processing, such as

expressionorgenderrecognition.Wesuggestthatthismightbeduetothetime-consumingnatureof

administering a battery of tests to examine various aspects of face processing (e.g., expressions, eye

gaze, gender, sex etc.), coupled with a paucity of brief, widely accepted, reliable tests to diagnose

deficitsintheseprocesses.

Similarly,therearecurrentlynoformallyestablishedexclusioncriteriaforindividualswhoshow

impairmentsdistinguishingbetweennon-faceobjects.Moreover,somehavesuggestedthatpurelyface-

specificdeficitscannotoccurandthatallprosopagnosicsmusthavecorresponding,thoughoftenmilder,

recognitiondeficits, suchasobject agnosiaorwordagnosia, that canbe revealed through theuseof

carefullydesigned tasks62,63.Althoughcomorbidobjectprocessingdeficits are common24,28, thereare

individualswhoonlyhavedifficultywithfaces24,26.Navigationaldifficultiesmayalsobeassociatedwith

DP, with approximately 15% of adults in the faceblind.org database reporting severe navigational

difficulties64,althoughitisnotyetknownwhetherthisratewouldbehigherthanthatreportedbythe

general population. These comorbiditiesmaybeexplainedby theproximitybetween face,place, and

object processing areas in the brain64. At least for objects, tests are available to enable comparisons

between face and object processing abilities (see Table 1) andwe recommend that researchers note

whetherthosetheyclassifyasDPappeartohaveface-selectiveormorewidespreaddeficits.

Faceidentityrecognitiondeficitsalsoexistinotherdevelopmentaldisorders,suchasautism65-68

andTurner syndrome69,but face recognitiondeficits cananddoexist in individualswhohavenormal

socialfunctioning67.Itisunclearatthispointwhetherthefacerecognitiondeficitsthataccompanyother

developmental disorders should be classified as DP, but given themuch broader cognitive effects in

disorders like autism spectrumdisorders (ASD), it is possible that face recognitiondeficits outsideDP

maybequalitativelydifferent(see70forexample).Thusitisadvisabletoexcludeindividualswithsocial

developmentaldisordersfromstudiesofDP,unlesstheyareaparticularpopulationofinterest.

When working with children with no formal diagnosis of autism, we recommend including

measuresofautistictendenciestoconfirmthatthechilddoesnothaveautism.TheAutismDiagnostic

ObservationSchedule(ADOS)71isonecomponentofthegoldstandardassessmentbatteryforASD,but

trained personnel must administer it, and research reliability is difficult to achieve. It is also time

consuming to administer (30minutes to > 1 hour, plus scoring), and additional assessments are still

required for diagnosis. Alternative, more practical, measures for researchers are parent report

questionnairessuchastheSocialResponsivenessScale–2(SRS-2)72andtheRepetitiveBehaviorScale–

Revised73. While these questionnaires are not sufficient for diagnosing ASD, they can be used as

13

guidelinessuchthatindividualswithhighscores(indicatingautistictendencies)canbefurthertestedor

excluded from the study as a precaution. The Autism Quotient (AQ)74, 75 is commonly used among

researchers as a quick assessment of autistic tendencies, but parents complain that the items are

antiquated (e.g. “S/he isnotverygoodat rememberingphonenumbers.”)andmay thereforebe less

current than the SRS-2 and the RBS-R. For adults, some previous studies15 have tended to exclude

individualswhoscored32oraboveon theAQ,whichwasa range indicativeofapossibleASD74.One

drawbackofusingtheAQwithDPs is that it includesquestionsrelatedto facerecognitionabilityand

relatedsocialinteractionandcouldthereforeresultininflatedscores(e.g.,“WhenI’mreadingastory,I

caneasilyimaginewhatthecharactersmightlooklike”,“Ifindsocialsituationseasy”,“Ienjoymeeting

newpeople”).Additionally,wenotethattheAQisdesignedtobedescriptiveratherthandiagnosticand

thatthereareupdatedmeanAQscoresfortypicalandASDpopulations76.Otherquestionnairesarealso

availabletomeasureautistictraitsinadults(see77foracomparison).

In terms of strictly visual comorbidities, we have noticed a recent increase in the number of

children requesting assessment for face recognition deficits who already have a diagnosis of Cortical

Visual Impairment(CVI).CVI isdefinedasvisual impairmentaffectingbothvisualfieldsdespitenormal

oculomotor systems. It can be related to abnormal development of the visual cortex78, 79 due to

congenitalbrainmalformation78,withthelocationandextentofthecorticalabnormalitiesdetermining

thefunctionaloutcomes80.SomeindividualswithCVIhaveprosopagnosia80makingitanopenquestion

whether a diagnosis of CVI, in some cases,would be better classified as DP, orwhether some or all

individualswithDPmayhaveadditionalsymptomsthatwouldqualifyforadiagnosisofCVI.

Another consideration when identifying DP is whether the individual had early visual

disturbancessuchascongenitalcataracts,amblyopiaorstrabismus.Thereisevidencethatnormalvisual

experiencewithfacesfromanearlyage iscritical forthedevelopmentofnormalfacerecognition81-85.

Forexample,some individualswhohadcongenitalcataractshave long-lasting faceprocessingdeficits,

evenifthecataractswereremovedatanearlyage(e.g.withinthefirst8monthsof life)andacuity is

restored to normal81-83, 86. This is especially true for individuals with bilateral or left eye cataracts84,

possibly because in infancy information from the left eye projects almost exclusively to the right

hemisphere87, which is thought to be particularly important for face processing88-91. Thus one issue

concernswhether an individualwith face recognitiondeficitswhohad congenital cataracts shouldbe

excludedfromresearchonDP.

Onemight predict that early amblyopia, a conditionwhere the vision in one eye is reduced,

could similarly affect face processing and could therefore raise similar flags for participant inclusion.

14

However,atpresentthere isnoempiricalevidencethatfacerecognitiondeficitsaremorecommonin

individuals with amblyopia than the general public. While some researchers might advocate for

excludingthesecases,iflow-levelvisionisnormalorcorrected-to-normal,andtheintricaciesofthecase

are described in detail, cases with suspected aetiologies may help elucidate the developmental

processesleadingtoDP.ThesecasesmayalsorepresentaninterestingsampletocomparetoDPswith

nohistoryofissueswithlow-levelvision.

COGNITIVEHETEROGENEITY

Different phenotypes ofDPhavebeen identified in adults: some individualswithDP are only

impairedwithfacialidentityrecognition15,56,92-94,whileothersmayhaveproblemswithfacedetection56,

94, expression recognition26, or gender discrimination26. These dissociations can provide information

about which abilities are mediated by distinct neural mechanisms. Addressing heterogeneity,

associations, and dissociations in children with DP is particularly important: dissociations between

different types of face processing abilities in children with DP can provide clues as to when in

development theseabilitiesbegin to relyon separatemechanisms. Identifyingandunderstanding the

different subtypes of DP in childhood and adulthood will help inform choices for therapeutic

interventions.

While individual differences canprovideunique information aboutDP, distinct phenotypesof

DPsuggesttheneedforcautionwhenplanninggroupstudies(c.f.95).Therearenoclearguidelinesfor

whichindividualscanbegroupedtogether,andhowonedeterminesinclusioncriteriamaydependon

theprimaryresearchquestion.Thus,werecommendthatwhenpresentinggroupdataonDP,individual

databeincludedtoprovideamorecompletepictureofthedistributionofscores.Onedifficultyhereis

thatmanyof themeasuressuitable forassessinggroupdifferences in faceprocessingmechanismsdo

nothavetherequiredpsychometricpropertiestoreliablydeterminewhetheranindividual is impaired

or not. As an example, faces are represented as an integratedwholemore thanmost other types of

objects,aprocesstermedconfiguralorholisticprocessing96,97.Holisticcodingcanbemeasuredwiththe

composite effect96 and part-whole effect98, and recent studies have typically shown impairments in

groups of DPs15, 99, 100. However, current versions of holistic processing tasks are not appropriate for

evaluating holistic processing at an individual level. One way that researchers have attempted to

compensateforthelowtomoderatereliabilityofeachtaskhasbeenthecognitiveneuropsychological

approachofusingmore thanoneversionofaholisticcoding task to try toobtainconsensus foreach

individual101. Interestingly, individuals with DP can show typical holistic coding101, 102, indicating that

15

whilemeasuringholistic codingmaybeuseful for thedifferentiationofpotential subtypes, it isnota

definingfeatureofDP.

With sucha complex,heterogeneousdisorder, it is likely thataportionof thepopulationwill

produceborderlineorambiguousresults.Especiallywithchildren,itispossibleforanindividualtoscore

inthe impairedrangeforonetestof facememory,butthenscore inthenormalrangeforadifferent

testof thesameability.While it isperhapssimplest todisregard these individualsasuninterpretable,

thesevariableprofilesmayprovideamorecompleteaccountofthedisorderanditsrelationshiptothe

hypothesis. It isthereforerecommendedtoincludetheseambiguouscases inreports,buttointerpret

them with caution, or to only draw firm conclusions from the clear cases. This again speaks to the

drawbacksofgroupstudies,whereborderlineorinconsistentcasesmaydilutegroupeffects.

SUMMARYOFGUIDELINESANDFUTUREDIRECTIONS

Table1providesexamplesoftasksthatareavailablefortheassessmentofDPinchildrenand

adults.Itwouldbeidealtohavevalid,reliable,andbriefteststomeasureeachoftheprocessesinvolved

infacerecognitiontodeterminethelocusoftheimpairment(e.g., intactfaceperceptionbutimpaired

facememory).Inaddition,itwouldbepreferabletohavemorethanonetestavailabletomeasureeach

oftheseprocesses,thoughatpresent,onlyCFMTtasksareavailableinmorethanoneversion.Itisalso

desirabletouseagreatermixtureofhighlycontrolledandecologicallyvalidtasks.Thetestsneedtobe

sensitive to diet of faces experienced by individuals and so will need to vary by race, and perhaps

ethnicity14. Finally, each of the tests needs to be age-appropriate – easy enough for children and

sensitivetodeclinesinperformancethatbeginfrommiddle-age.Thedesignandvalidationofadditional

testscouldbeguidedbytheprocessoftestdesignforotherconditions,suchasdevelopmentaldyslexia,

where a number of appropriate tests are available to distinguish between different subtypes of the

disorder103. We note that most recently devised tests of face recognition are freely available to

researchersviacontactingtheauthorsofthetest.However,thefieldmaywishtoconsiderestablishing

a central website for tests. We know of one that hosts numerous tests of language processing

(http://www.motif.org.au),fromwhichtestscanbeaccessedbythosewithappropriatecredentialsand

whichprovidesupdatestoresearchers.

Giventheavailable tests,weproposethe followingcertainminimumguidelines forevaluating

face recognition in adults. First, the person should report repeated instances of difficulty recognizing

faces(notnames)ineverydaylife,eitherviaquestionnaireorinterview,andnotreportanybraininjury

16

or disorder that would be expected to result in face recognition impairments (e.g. temporal lobe

epilepsy). Second, theperson shoulddemonstrate impaired face recognitionmemory (orperception),

ideally on more than one test (e.g., famous faces and CFMT; CFMT and CFMT-Australian; CFPT and

CFMT). To differentiate poor recognisers from DPs, we suggest examining z-scores and significance

levels on statistical tests such as Crawford, Garthwaite, & Howell’s (2009) modified t-tests. While a

stringent criteria would advocate that performance should be significantly impaired and below two

standarddeviationsonall testsof facememory, inpractice it is typicallymore instructive toexamine

performanceontherangeoftests includedandbeclearaboutwhichcasesmaybeambiguous.These

criteriawouldenableadiagnosisofDP,butwouldnotbe informativeas towhether the impairments

extendedtootherobjectsandifso,whetherthisisindicativeofageneralcognitiveimpairment.Forthe

former,weadvocate includingatestofobjectmemory;forthelatterweadvocateassessingcognitive

functioning with brief measures. We also recommend asking whether the individual knows of any

currentorpreviousvisual impairment,andtomeasurevisualacuity,whichif impairedcouldsuggesta

low-leveloriginforDP.Finally,ameasureofsocialfunctioningand/orpresenceofautistictraitscanhelp

todifferentiateadiagnosisofDPfromabroaderdevelopmentaldisorderofsocialfunction.

Similarguidelinesshouldbefollowedforchildren,butspecialconsiderationsmustbemade.For

instance, reportsof realworld facerecognitionproblemsarestill important,butwill likelycomefrom

parentsratherthanthechild.DiagnosisofDPinchildrencanbehinderedbyflooreffectsintestsofface

memory, particularly in younger children. As such, it is recommended to evaluate face perception in

childrenwithsuspectedDPasawayofprovidingconvergentevidenceforfaceprocessingdeficits.Given

thevariability in test taking inchildhood (bothwithin,andbetween individuals),ambiguouscasesare

likely to arise. These cases may be informative, and should be included in scientific reports, but

appropriatecautionshouldbeusedwhenofferinginterpretationsofthedata.

CONCLUSIONS

IndividualswithDPhave severe face recognitiondeficits in the absenceof brain damage and

despite normal low-level vision and intellect. The heterogeneity of the disorder, and the variety of

comorbidities that can accompany it, lead to challenges for researchers.We provided guidelines for

studyingDPbasedonourexperiences.AlthoughfacememoryimpairmentiscriticalforidentifyingDPin

adults, faceperceptionmightbethemost reliablewayto identifyDP inkids.Real lifedifficultieswith

facerecognitionalsoneedtobeconsidered.Theuseofsinglecasemodifiedt-testscansupplementthe

17

traditional approach of using two standard deviations below the mean as a cut-off for impairment.

Overallitisessentialtobetransparentandincludeasmuchinformationaspossibleforallcases.

18

Notes

ThewritingofthisarticlewassupportedbytheAustralianResearchCouncil(ARC)CentreofExcellence

in Cognition and its Disorders (CCD) [grant number CE110001021] http://www.ccd.edu.au. KAD was

supportedbyaBantingPostdoctoralFellowshipawardedbytheCanadianInstitutesofHealthResearch.

RPwas supportedby theCCDandARCDiscoveryProjectgrants (DP110100850andDP1401743).KAD

andRPwouldliketothankthemanychildrenandadultswhosegenerousparticipationinourresearch

projectshashelpedustoestablishtheseguidelines.SpecialthankstoBradDuchaineandLindaJeffery

fortheirinsightfulcommentsonpreviousdrafts.

Therearenoconflictsofinterest.

19

Table1.Someavailablemeasuresthatcanbeuseddeterminewhetherchildrenandadultshavesignificantfacerecognitiondifficultiescommensuratewithdevelopmentalprosopagnosia(DP),andwhethertheyalsopresentwithotherassociateddifficulties.

Abilities Children(ages6+)

Adults(ages18+)

Facerecognitiontests

Faceperception DartmouthFacePerceptionTest(DFPT)12

CambridgeFacePerceptionTest(CFPT)49

GlasgowFaceMatchingTest(GFMT)50

Facememory–unfamiliarfaces

CambridgeFaceMemoryTest–Kids(CFMT-Kids)20

CambridgeFaceMemoryTest–Children(CFMT-Children)45

CambridgeFaceMemoryTest(CFMT)34

Old-NewFaceRecognitiontest48

Facememory–familiarfaces

Familiarfacesprovidedbyfamilies Familiarfacesprovidedbyfamilies

FamousfaceTests17

Subjectivereportsofeverydayfacerecognitiondifficulties

Parentandchildreportsoffacerecognitiondifficulties

Anecdotalreportsoffacerecognitiondifficulties

ProsopagnosiaIndex(PI-20)21

Othertestsandmeasures

Within-categoryobjectmemory(usedtodeterminewhetherfaceandobjectmemoryimpaired)

CambridgeBicycleMemoryTest(CBMT)38

Old-NewFlowers12

CambridgeCarMemoryTest37

VanderbiltExpertiseTest39

Old-NewTests26

Selectivesocialdeficits(usedtodeterminewhethersocialdeficitsevident)

SocialResponsivenessScale-272

RepetitiveBehaviorScale-Revised73

AutismQuotient74

20

VisualAcuity(usedtodetermineadequatevisualacuity)

Parentalreportofvisualimpairments

HOTVDistancePediatricEyeChart

Self-reportofvisualimpairments

VisualacuitytestusingSloanfont

Generalfunctioning

Parentreportofanybraininjury

BirminghamObjectRecognitionBattery23

WechslerAbbreviatedScaleofIntelligence-II104

Self-reportofanybraininjury,disorder,educationallevelandoccupation.

BirminghamObjectRecognitionBattery23

DigitSpan(backwards)fromtheWechslerAdultIntelligenceScale-IV105

Raven’sProgressiveMatrices106

21

Figures

Figure1.a)FormatofCambridgeFaceMemoryTest(CFMT,Duchaine&Nakayama,2006a)foradults.

Thetargetfaceisforillustrativepurposesandisnotusedintheactualtest.b)Sampletrialfrom

CambridgeBicycleMemoryTest(CBMT,Dalrymple&Duchaine,2013)c)SampletrialfromCambridge

FaceMemoryTest–Kids(CFMT-Kids,Dalrymple,etal,2012)d)CambridgeFacePerceptionTest(CFPT,

Duchaine,Yovel,&Nakayama,2007)e)DartmouthFacePerceptionTest(forchildren)(DFPT,Dalrymple,

Garrido,&Duchaine,2014).

22

Figure2.MeanscoresbyagefordifferentversionsofCambridgeFaceMemoryTest–Kids(CFMT-Kids,

Dalrympleetal.,2012).Six-targetversionfollowstheformatoftheCFMTforadults,with72trialsin

total.Four-targetversionfollowsthesameformatastheCFMTforadults,butuses4targetsinsteadof

6,with48trialsintotal.TheImmediaterecallversionpresentstheIntroductorysectionoftheCFMT-

Kidssix-targetversiontwice,foratotalof12targets,with36trialsintotal.Specifically,childrenlearna

targetandthenpickitoutfromachoiceofthree,butareneverrequiredtoholdmultipletargetsin

memory.Errorbarsrepresentstandarddeviation.Twostandarddeviationsbelowthemeanisone

conventionforcalculatingacut-offforimpairment.Dottedlinerepresentschancelevelperformance.

23

References

1. DuchaineBC,NakayamaK.Developmentalprosopagnosia:awindowtocontent-specificface

processing.CurrOpinNeurobiol2006,16:166-173.2. BartonJJS,CherkasovaMV,PressDZ,IntriligatorJM,O’ConnorM.Developmental

prosopagnosia:Astudyofthreepatients.BrainandCognition2003,51:12-30.3. BehrmannM,AvidanG.Congenitalprosopagnosia:face-blindfrombirth.TrendsCognSci2005,

9:180-187.4. BowlesDC,McKoneE,DawelA,DuchaineB,PalermoR,SchmalzlL,RivoltaD,WilsonCE,Yovel

G.Diagnosingprosopagnosia:effectsofageing,sex,andparticipant-stimulusethnicmatchontheCambridgeFaceMemoryTestandCambridgeFacePerceptionTest.CognNeuropsychol2009,26:423-455.

5. KennerknechtI,GrueterT,WellingB,WentzekS,HorstJ,EdwardsS,GrueterM.Firstreportofprevalenceofnon-syndromichereditaryprosopagnosia(HPA).AmJMedGenetA2006,140:1617-1622.

6. DalrympleKA,FletcherK,CorrowS,dasNairR,BartonJJS,YonasA,DuchaineB."Aroomfullofstrangerseveryday":Thepsychosocialimpactofdevelopmentalprosopagnosiaonchildrenandtheirfamilies.JournalofPsychosomaticResearch2014,77:144-150.

7. DiazA.DoIknowyou?Acasestudyofprosopagnosia(faceblindness).TheJournalofSchoolNursing2008,24:284-289.

8. YardleyL,McDermottL,PisarskiS,DuchaineB,NakayamaK.Psychosocialconsequencesofdevelopmentalprosopagnosia:aproblemofrecognition.JPsychosomRes2008,65:445-451.

9. BruceV,YoungA.Understandingfacerecognition.BrJPsychol1986,77(Pt3):305-327.10. HappeF,RonaldA,PlominR.Timetogiveuponasingleexplanationforautism.NatNeurosci

2006,9:1218-1220.11. ChatterjeeG,NakayamaK.Normalfacialageandgenderperceptionindevelopmental

prosopagnosia.CognNeuropsychol2012,29:482-502.12. DalrympleK,GarridoL,DuchaineB.Dissociationbetweenfaceperceptionandfacememoryin

adults,butnotchildren,withdevelopmentalprosopagnosia.DevelopmentalCognitiveNeuroscience2014,10:10-20.

13. HumphreysK,AvidanG,BehrmannM.Adetailedinvestigationoffacialexpressionprocessingincongenitalprosopagnosiaascomparedtoacquiredprosopagnosia.ExpBrainRes2007,176:356-373.

14. McKoneE,HallA,PidcockM,PalermoR,WilkinsonRB,RivoltaD,YovelG,DavisJM,O'ConnorKB.Faceethnicityandmeasurementreliabilityaffectfacerecognitionperformanceindevelopmentalprosopagnosia:evidencefromtheCambridgeFaceMemoryTest-Australian.CognNeuropsychol2011,28:109-146.

15. PalermoR,WillisML,RivoltaD,McKoneE,WilsonCE,CalderAJ.Impairedholisticcodingoffacialexpressionandfacialidentityincongenitalprosopagnosia.Neuropsychologia2011,49:1226-1235.

16. BartonJJ,CherkasovaM.Faceimageryanditsrelationtoperceptionandcovertrecognitioninprosopagnosia.Neurology2003,61:220-225.

17. PalermoR,RivoltaD,WilsonCE,JefferyL.Adaptivefacespacecodingincongenitalprosopagnosia:typicalfiguralaftereffectsbutabnormalidentityaftereffects.Neuropsychologia2011,49:3801-3812.

18. PalermoR,RossionB,RhodesG,LaguesseR,TezT,HallB,AlbonicoA,MalaspinaM,DainiR,IronsJ,etal.Dopeoplehaveinsightintotheirfacerecognitionabilities?.underreview.

24

19. KennerknechtI,HoNY,WongVC.Prevalenceofhereditaryprosopagnosia(HPA)inHongKongChinesepopulation.AmJMedGenetA2008,146A:2863-2870.

20. DalrympleKA,CorrowS,YonasA,DuchaineB.Developmentalprosopagnosiainchildhood.CognNeuropsychol2012,29:393-418.

21. ShahP,GauleA,SowdenS,BirdG,CookR.The20-itemprosopagnosiaindex(PI20):aself-reportinstrumentforidentifyingdevelopmentalprosopagnosia.RoyalSocietyOpenScience2015,2:1-11.

22. RivoltaD,PalermoR,SchmalzlL,ColtheartM.Covertfacerecognitionincongenitalprosopagnosia:agroupstudy.Cortex2012,48:344-352.

23. RiddochJ,HumphreysGW.BORB:Birminghamobjectrecognitionbattery.Hove,UK:Erlbaum;1993.

24. DuchaineB,NakayamaK.Dissociationsoffaceandobjectrecognitionindevelopmentalprosopagnosia.JCognNeurosci2005,17:249-261.

25. NunnJA,PostmaPFau-PearsonR,PearsonR.Developmentalprosopagnosia:shoulditbetakenatfacevalue?

26. DuchaineBC,YovelG,ButterworthEJ,NakayamaK.Prosopagnosiaasanimpairmenttoface-specificmechanisms:Eliminationofthealternativehypothesesinadevelopmentalcase.CognNeuropsychol2006,23:714-747.

27. GermineL,CashdollarN,DüzelE,DuchaineBC.Anewselectivedevelopmentaldeficit:Impairedobjectrecognitionwithnormalfacerecognition.Cortex;ajournaldevotedtothestudyofthenervoussystemandbehavior2010.

28. BehrmannM,AvidanG,MarottaJJ,KimchiR.Detailedexplorationofface-relatedprocessingincongenitalprosopagnosia:1.Behavioralfindings.JCognNeurosci2005,17:1130-1149.

29. RighartR,deGelderB.Impairedfaceandbodyperceptionindevelopmentalprosopagnosia.ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica2007,104:17234-17238.

30. WarringtonE.RecognitionMemoryTest.Windsor,UK:NFER-Nelson;1984.31. BentonA,SivanA,HamsherK,VarneyN,SpreenO.Contributiontoneuropsychological

assessment.NewYork:OxfordUniversityPress;1983.32. DuchaineBC,NakayamaK.DevelopmentalprosopagnosiaandtheBentonFacialRecognition

Test.Neurology2004,62:1219-1220.33. DuchaineBC,WeidenfeldA.Anevaluationoftwocommonlyusedtestsofunfamiliarface

recognition.Neuropsychologia2003,41:713-720.34. DuchaineB,NakayamaK.TheCambridgeFaceMemoryTest:resultsforneurologicallyintact

individualsandaninvestigationofitsvalidityusinginvertedfacestimuliandprosopagnosicparticipants.Neuropsychologia2006,44:576-585.

35. WilmerJB,GermineL,ChabrisCF,ChatterjeeG,GerbasiM,NakayamaK.Capturingspecificabilitiesasawindowintohumanindividuality:theexampleoffacerecognition.CognNeuropsychol2012,29:360-392.

36. McKoneE,StokesS,LiuJ,CohanS,FiorentiniC,PidcockM,YovelG,BroughtonM,PellegM.Arobustmethodofmeasuringother-raceandother-ethnicityeffects:theCambridgeFaceMemoryTestformat.PLoSOne2012,7:e47956.

37. DennettHW,McKoneE,TavashmiR,HallA,PidcockM,EdwardsM,DuchaineB.TheCambridgeCarMemoryTest:ataskmatchedinformattotheCambridgeFaceMemoryTest,withnorms,reliability,sexdifferences,dissociationsfromfacememory,andexpertiseeffects.BehavResMethods2012,44:587-605.

25

38. DalrympleK,DuchaineB.Variedbehavioralprofilesinchildrenwithdevelopmentalprosopagnosiarevealdissociationsinthedevelopingfaceprocessingsystem.JournalofVision2013,13:994.

39. McGuginRW,RichlerJJ,HerzmannG,SpeegleM,GauthierI.TheVanderbiltExpertiseTestrevealsdomain-generalanddomain-specificsexeffectsinobjectrecognition.VisionRes2012,69:10-22.

40. BateS,CookSJ,DuchaineB,TreeJJ,BurnsEJ,HodgsonTL.Intranasalinhalationofoxytocinimprovesfaceprocessingindevelopmentalprosopagnosia.Cortex2014,50:55-63.

41. StollhoffR,JostJ,ElzeT,KennerknechtI.DeficitsinLong-TermRecognitionMemoryRevealDissociatedSubtypesinCongenitalProsopagnosia.PLoSONE2011,6:e15702.

42. MikeBurtonA.Whyhasresearchinfacerecognitionprogressedsoslowly?Theimportanceofvariability.TheQuarterlyJournalofExperimentalPsychology2013,66:1467-1485.

43. KorkmanM,KirkU,FellmanV.NEPSY:Adevelopmentalneuropsychologicalassessment..SanAntonio,Texas:ThePsychologicalCorporation;1998.

44. KaufmanA,KaufmanN.KaufmanAssessmentBatteryforChildren.CirclePines,MN:AmericanGuidanceService;1983.

45. CroydonA,PimpertonH,EwingL,DuchaineBC,PellicanoE.TheCambridgeFaceMemoryTestforChildren(CFMT-C):anewtoolformeasuringfacerecognitionskillsinchildhood.Neuropsychologia2014,62:60-67.

46. WieseH,KomesJ,SchweinbergerSR.Ageingfacesinageingminds:Areviewontheown-agebiasinfacerecognition.VisualCognition2013,21:1337-1363.

47. MacchiCassiaV.Agebiasesinfaceprocessing:Theeffectsofexperienceacrossdevelopment.BritishJournalofPsychology2011,102:816-829.

48. DuchaineBC,Nieminen-vonWendtT,NewJ,KulomakiT.Dissociationsofvisualrecognitioninadevelopmentalagnosic:evidenceforseparatedevelopmentalprocesses.Neurocase2003,9:380-389.

49. DuchaineB,GermineL,NakayamaK.Familyresemblance:tenfamilymemberswithprosopagnosiaandwithin-classobjectagnosia.CognNeuropsychol2007,24:419-430.

50. BurtonAM,WhiteD,McNeillA.TheGlasgowFaceMatchingTest.BehavResMethods2010,42:286-291.

51. WeigeltS,KoldewynK,DilksD,BalasB,McKoneE,KanwisherN.Domain-specificdevelopmentoffacememorybutnotfaceperception.DevelopmentalScience2014,17:47-58.

52. YovelG,DuchaineB.Specializedfaceperceptionmechanismsextractbothpartandspacinginformation:EvidencefromdevelopmentalProsopagnosia.JournalofCognitiveNeuroscience2006,18:580-593.

53. CrawfordJR,GarthwaitePH,HowellDC.Oncomparingasinglecasewithacontrolsample:analternativeperspective.Neuropsychologia2009,47:2690-2695.

54. CrawfordJR,GarthwaitePH.Investigationofthesinglecaseinneuropsychology:confidencelimitsontheabnormalityoftestscoresandtestscoredifferences.Neuropsychologia2002,40:1196-1208.

55. CrawfordJR,HowellDC.Comparinganindividual'stestscoreagainstnormsderivedfromsmallsamples.ClinicalNeuropsychologist1998,12:482-486.

56. DalrympleK,DuchaineB.Impairedfacedetectionmayexplainsomebutnotallcasesofdevelopmentalprosopagnosia.DevelopmentalScience.inpress.

57. SieglerRS.CognitiveVariability-aKeytoUnderstandingCognitive-Development.CurrentDirectionsinPsychologicalScience1994,3:1-5.

58. SieglerRS.Cognitivevariability.DevSci2007,10:104-109.

26

59. CrawfordJR,GarthwaitePH.Comparisonofasinglecasetoacontrolornormativesampleinneuropsychology:developmentofaBayesianapproach.CognNeuropsychol2007,24:343-372.

60. CrawfordJR,GarthwaitePH,PorterS.Pointandintervalestimatesofeffectsizesforthecase-controlsdesigninneuropsychology:rationale,methods,implementations,andproposedreportingstandards.CognNeuropsychol2010,27:245-260.

61. CrawfordJ,GarthwaiteP,RyanK.Comparingasinglecasetoacontrolsample:Testingforneuropsychologicaldeficitsanddissociatiosinthepresenceofcovariates.Cortex2011,47:1166-1178.

62. BehrmannM,PlautDC.Bilateralhemisphericprocessingofwordsandfaces:evidencefromwordimpairmentsinprosopagnosiaandfaceimpairmentsinpurealexia.CerebCortex2014,24:1102-1118.

63. GauthierI,BerhmannM,TarrMJ.Canfacerecognitionreallybedissociatedfromobjectrecognition?JournalofCognitiveNeuropsychology1999,11.

64. DuchaineB,YovelG.Facerecognition.In:A.Basbaum,A.Kaneko,ShepherdG,WestheimerG,eds.TheSenses:AComprehensiveReference.Vol.2.SanDiego:AcademicPress.;2008,329-358.

65. BartonJJ,CherkasovaMV,HefterR,CoxTA,O'ConnorM,ManoachDS.Arepatientswithsocialdevelopmentaldisordersprosopagnosic?PerceptualheterogeneityintheAspergerandsocio-emotionalprocessingdisorders.Brain2004,127:1706-1716.

66. WeigeltS,KoldewynK,KanwisherN.Faceidentityrecognitioninautismspectrumdisorders:areviewofbehavioralstudies.NeurosciBiobehavRev2012,36:1060-1084.

67. DuchaineB,MurrayH,TurnerM,WhiteS,GarridoL.Normalsocialcognitionindevelopmentalprosopagnosia.CognNeuropsychol2009,26:620-634.

68. WilsonCE,PalermoR,SchmalzlL,BrockJ.Specificityofimpairedfacialidentityrecognitioninchildrenwithsuspecteddevelopmentalprosopagnosia.CognNeuropsychol2010,27:30-45.

69. HongD,ScalettaKentJ,KeslerS.CognitiveprofileofTurnersyndrome.DevDisabilResRev2009,15:270-278.

70. DaltonKM,NacewiczBM,JohnstoneT,SchaeferHS,GernsbacherMA,GoldsmithHH,AlexanderAL,DavidsonRJ.Gazefixationandtheneuralcircuitryoffaceprocessinginautism.NatNeurosci2005,8:519-526.

71. LordC,RutterM,DiLavoreP,RisiS.AutismDiagnosticObservationSchedule-WPS(ADOS-WPS).LosAngeles,CA:WesternPsychologicalSciences;1999.

72. ConstantinoJ,GruberC.SocialResponsivenessScale.SecondEdition.ed.LosAngeles,CA:WesternPsychologicalServices.;2012.

73. LamKS,AmanMG.TheRepetitiveBehaviorScale-Revised:independentvalidationinindividualswithautismspectrumdisorders.JAutismDevDisord2007,37:855-866.

74. Baron-CohenS,WheelwrightS,SkinnerR,MartinJ,ClubleyE.Theautism-spectrumquotient(AQ):Evidencefromaspergersyndrome/high-functioningautism,malesandfemales,scientistsandmathematicians.JournalofAutismandDevelopmentalDisorders2001,31:5-17.

75. Baron-CohenS,HoekstraRA,KnickmeyerR,WheelwrightS.TheAutism-SpectrumQuotient(AQ)--adolescentversion.JAutismDevDisord2006,36:343-350.

76. RuzichE,AllisonC,SmithP,WatsonP,AuyeungB,RingH,Baron-CohenS.Measuringautistictraitsinthegeneralpopulation:asystematicreviewoftheAutism-SpectrumQuotient(AQ)inanonclinicalpopulationsampleof6,900typicaladultmalesandfemales.MolAutism2015,6:2.

77. NishiyamaT,SuzukiM,AdachiK,SumiS,OkadaK,KishinoH,SakaiS,KamioY,KojimaM,SuzukiS,etal.Comprehensivecomparisonsofself-administeredquestionnairesformeasuringquantitativeautistictraitsinadults.JournalofAutismandDevelopmentalDisorders2014,44:993-1007.

27

78. CardenS,GoodW.Corticalvisualimpairment.In:Handbookofpediatricneuro-ophthalmology.NewYork:Springer.;2006,247-254.

79. GoodWV,JanJE,DeSaL,BarkovichAJ,GroenveldM,HoytCS.Corticalvisualimpairmentinchildren.SurvOphthalmol1994,38:351-364.

80. MorseM.Anotherviewofcorticalvisualimpairment:Issuesrelatedtofacialrecognition.ProceedingsoftheSummitonCerebral/CorticalVisualImpairment:Educational,Family,andMedicalPerspectives2006.Vol.April30,2005,Pages131-136.

81. deHeeringA,MaurerD.Facememorydeficitsinpatientsdeprivedofearlyvisualinputbybilateralcongenitalcataracts.DevelopmentalPsychobiology2014,56:96-108.

82. LeGrandR,MondlochCJ,MaurerD,BrentHP.Impairmentinholisticfaceprocessingfollowingearlyvisualdeprivation.PsycholSci2004,15:762-768.

83. LeGrandR,MondlochCJ,MaurerD,BrentHP.Neuroperception.Earlyvisualexperienceandfaceprocessing.Nature2001,410:890.

84. LeGrandR,MondlochCJ,MaurerD,BrentHP.Expertfaceprocessingrequiresvisualinputtotherighthemisphereduringinfancy.NatNeurosci2003,6:1108-1112.

85. PollakS,NelsonC,SchlaakM,RoeberB,WewerkaS,WiikK,FrennK,LomanM,GunnarM.Neurodevelopmentaleffectsofearlydeprivationinpost-institutionalizedchildren.ChildDevelopment2010,81:224-236.

86. OstrovskyY,AndalmanA,SinhaP.Visionfollowingextendedcongenitalblindness.PsycholSci2006,17:1009-1014.

87. LewisTL,MaurerD.Thedevelopmentofthetemporalandnasalvisualfieldsduringinfancy.VisionRes1992,32:903-911.

88. BentonA.Facialrecognition.Cortex1990,26:491-499.89. KanwisherN,McDermottJ,ChunMM.Thefusiformfacearea:amoduleinhumanextrastriate

cortexspecializedforfaceperception.JNeurosci1997,17:4302-4311.90. LandisT,CummingsJ,ChristenL,BogenJ.Areunilateralrightposteriorcerebrallesions

sufficienttocuaseprosopagnosia?Clinicalandradiologicalfindingsinsixaddtionalpatients.Cortex1986,22:243-252.

91. McCarthyG,PuceA,GoreJC,AllisonT.Face-specificprocessinginthehumanfusiformgyrus.JCognNeurosci1997,9:605-610.

92. DuchaineBC,ParkerH,NakayamaK.Normalrecognitionofemotioninaprosopagnosic.Perception2003,32:827-838.

93. GarridoL,FurlN,DraganskiB,WeiskopfN,StevensJ,TanGC,DriverJ,DolanRJ,DuchaineB.Voxel-basedmorphometryrevealsreducedgreymattervolumeinthetemporalcortexofdevelopmentalprosopagnosics.Brain2009,132:3443-3455.

94. GarridoL,DuchaineB,NakayamaK.Facedetectioninnormalandprosopagnosicindividuals.JNeuropsychol2008,2:119-140.

95. McCloskeyM,CaramazzaA.TheoryandMethodologyinCognitiveNeuropsychology-aResponsetoOurCritics.CognitiveNeuropsychology1988,5:583-623.

96. RossionB.Thecompositefaceillusion:Awholewindowintoourunderstandingofholisticfaceperception.VisualCognition2013,21:139-253.

97. RobbinsR,McKoneE.Canholisticprocessingbelearnedforinvertedfaces?COGNITION2003,88:79-107.

98. TanakaJW,FarahMJ.Partsandwholesinfacerecognition.TheQuarterlyjournalofexperimentalpsychologyA,Humanexperimentalpsychology1993,46:225-245.

99. AvidanG,TanzerM,BehrmannM.Impairedholisticprocessingincongenitalprosopagnosia.Neuropsychologia2011.

28

100. DeGutisJ,CohanS,MercadoRJ,WilmerJ,NakayamaK.Holisticprocessingofthemouthbutnottheeyesindevelopmentalprosopagnosia.CognitiveNeuropsychology2012,29:419-446.

101. SusiloT,McKoneE,DennettH,DarkeH,PalermoR,HallA,PidcockM,DawelA,JefferyL,WilsonCE,etal.Facerecognitionimpairmentsdespitenormalholisticprocessingandfacespacecoding:evidencefromacaseofdevelopmentalprosopagnosia.CognNeuropsychol2010,27:636-664.

102. SchmalzlL,PalermoR,ColtheartM.Cognitiveheterogeneityingeneticallybasedprosopagnosia:Afamilystudy.Journalofneuropsychology2008,2:99-117.

103. ZoubrinetzkyR,BielleF,ValdoisS.NewInsightsonDevelopmentalDyslexiaSubtypes:HeterogeneityofMixedReadingProfiles.PLoSONE2014,9:e99337.

104. WechslerD.WechslerAbbreviatedScaleofIntelligence(WASI-II).USA:Pearson;2011.105. WechslerD.WechslerAdultIntelligenceScale-IV.SanAntonio,Texas:PearsonAssessment;

2008.106. RavenJC.Raven'sStandardProgressiveMatrices.USA:Pearson;2008.