Age and gender dependent development of Theory of Mind in 6

... which allows us to meaningfully test ToM development on children 6 to 8-years old. We find that kids this age are still not entirely ... Skiptomaincontent DownloadArticle DownloadPDF ReadCube EPUB XML(NLM) totalviews ViewArticleImpact SHAREON Abstract Introduction MaterialsandMethods Results Discussion ConflictofInterestStatement Acknowledgments References Exportcitation EndNote ReferenceManager SimpleTEXTfile BibTex Peoplealsolookedat ORIGINALRESEARCHarticle Front.Hum.Neurosci.,17June2013Sec.CognitiveNeuroscience https://doi.org/10.3389/fnhum.2013.00281 AgeandgenderdependentdevelopmentofTheoryofMindin6-to8-yearsoldchildren CeciliaI.Calero1,2*†,AlejoSalles1,2†,MarianoSemelman1,2andMarianoSigman1,2,3 1LaboratoriodeNeurocienciaIntegrativa,DepartamentodeFísica,FacultaddeCienciasExactasyNaturales,UniversidaddeBuenosAires,BuenosAires,Argentina 2IFIBA,CONICET,BuenosAires,Argentina 3UniversidadTorcuatoDiTella,BuenosAires,Argentina Theabilitytoattributedifferentmentalstatestodistinctindividuals,orTheoryofMind(ToM),iswidelybelievedtobedevelopedmostlyduringpreschoolyears.Howdifferentfactorssuchasgender,numberofsiblings,orcoarsepersonalitytraitsaffectthisdevelopmentisnotentirelyagreedupon.Here,weintroduceacomputerizedversionofthescaledToMsuiteoftasksintroducedbyWellmanandLiu(2004),whichallowsustomeaningfullytestToMdevelopmentonchildren6to8-yearsold.WefindthatkidsthisagearestillnotentirelyproficientinallToMtasks,andcontinuetoshowaprogressionofperformancewithage.Bytestingthisnewagerange,too,weareabletoobserveasignificantadvantageofgirlsoverboysinToMperformance.Otherfactorssuchasnumberofsiblings,birthorder,andcoarsepersonalitytraitsshownosignificantrelationwiththeToMtaskresults.Finally,weintroduceanovelwaytoquantifythescalingpropertyofthesuiteinvolvingasequenceofsetinclusionsononehandandacomparisonbetweenspeciallytailoredsetsoflogisticmodelsontheother.Thesemeasuresconfirmthevalidityofthescaleinthe6-to8-yearsoldrange. Introduction TheoryofMind(ToM)isanimportantcognitiveskillthatrefersbroadlytoourcapacitytounderstandothers'mentalstatesincludingbeliefs,desires,andknowledge,andtheabilitytocomprehendthatthesemaydifferfromourown(PremackandWoodruff,1978).Aparadigmaticexampleofataskrequiringawell-developedToMisthatoffalsebelief,whichinvolvesunderstandingthatanagentmighthaveawrongrepresentationofexternalreality,andactaccordingtothatrepresentation(Baron-Cohenetal.,1985).ToMisdeeplyintegratedwithothercognitivedomainsandabilities.Amongotherfactors,ToMdevelopmenthasbeenrelatedtothesuccessonexecutivefunctiontests,andvariousassessmentsoflanguageandsocialexperience(Dunnetal.,1991;Dunn,1995;AstingtonandJenkins,1999;CarlsonandMoses,2001;Apperly,2012). Overthepasttwodecades,developmentalchangesinchildren'sunderstandingofothers'mindshavebeenthefocusofintenseresearch(WellmanandWoolley,1990;Dunnetal.,1991;GopnikandSlaughter,1991;Flavell,1999;AstingtonandJenkins,1999;Meltzoff,1999;Wellmanetal.,2001).Mostoftheworkperformedtodatewasdoneinpreschoolchildren(3to5-yearsold)andcoarselyagreesinthatthebasicaspectsofToMaremostlydevelopedwithinthisagerange[BartschandWellman,1995;seeFlavell(1999),forareview].However,somestudiesarguethatToMcontinuestodevelopandchangethroughoutlife(BosackiandAstington,1999;Apperly,2012;DevineandHughes,2012;Moran,2013).RaiandMitchell's(2004)studyhasshownthatthereisstillconsiderableinstabilityinunderstandingfalsebeliefsin5-yearsold,especiallywhenthefalsebeliefscenarioisframedinrelationtoaperson'sconsciouschoiceordecision,ratherthanaphysicalobject.Furthermore,Dumontheiletal.'(2010)resultssuggestthatToMimprovesbetweenlateadolescenceandadulthoodandevenifToMtasksarepassedbyage4;theirdataindicatethattheinteractionbetweenunderstandingothers'mindandexecutivefunctionscontinuestodevelopinlateadolescence(Dumontheiletal.,2010).BosackiandAstington(1999)usedambiguoussocialvignettesfollowedbyquestionstoassesstheunderstandingofparticularaspectsofother'smentalstatesinastudywithpreadolescentchildren.TheirstudywasconceivedfromaToMperspectiveinordertoquantifypreadolescents'mentalizingabilitiesandtheirresultspartiallysupportthat,individualdifferencesinpreadolescents'abilitytounderstandthethoughtsandemotionsofotherswouldberelatedtotheirsocialcompetence(BosackiandAstington,1999). MoststudiesonToMhavenotaddressedissuesofgender,familyenvironment,andmeasuresoftemperament.Aslightadvantageofpreschoolgirlsonemotionunderstandingandfalsebelieftaskperformancehasbeenobservedbefore(Banerjee,1997;CharmanandClements,2002;Walker,2005),nevertheless,mostpreviousstudieshavefoundnosignificantgenderdifferencesonToMdevelopment(HughesandDunn,1998;CharmanandClements,2002;Walker,2005;MathiesonandBanerjee,2011;DevineandHughes,2012).Ontheotherhand,instudiescarriedoutinpreadolescence,girlsperformedsignificantlyhigheronthesocialunderstandingtask(ToM)thanboysindependentofvocabularyability(BosackiandAstington,1999).TheseresultssupportHatcheretal.'(1990)findings,inwhichgirlsscoredhigherthanboysonsocialunderstandingtasksacrossgrades4through12.Further,therecentstudyinadolescentsbyIbanezetal.(2013)presentsamodelthatshowsthedirecteffectofempathy,sex,andfluidintelligenceonToM.Onlyrecentstudieshavefoundsomerelationsbetweenscoresonfalsebelieftasksandpreschoolchildren'sfamilyenvironments(Perneretal.,1994;Farhadianetal.,2010).However,thereisnoconsensusonwhethertheamountofsiblingsorthebirthorderinfluencesthedevelopmentofToM(AzmitiaandHesser,1993;Lewisetal.,1996;Ruffmanetal.,1998;CuttingandDunn,1999;Farhadianetal.,2010). AcomplementaryaimofmostdevelopmentalstudiesofToMconsistsinunderstandingthesequentialunfoldingofabilitiesunderlyingafullToM.Todatethereisconsensusinthenotionthatachildwillcorrectlyjudgeaperson'sdesiresbeforeshecancorrectlyjudgeherbeliefs,andthatshewillbeabletograspthatanagentmighthaveabeliefdifferentfromherownfirstifshedoesn'tknowthetruestateofaffairs,andonlylaterifshedoesknowwhatrealityreallyislike(falsebelief).CuttingandDunn(1999),WellmanandLiu(2004),Wellmanetal.(2001)and,morerecently,Wellmanetal.(2011)haveinvestigatedthisprogressionofabilities.Inparticular,WellmanandLiu(2004)proposedasuiteofToMtasks,baseduponameta-analysisoftheliteratureofToMdevelopmentalstudies,andtesteditinchildrenfrom3to5-yearsofage.TheirresultssuggestthattheabilitiesunderlyingToMareattainedprogressively,andcanthusbetestedindividuallybyanorderedsuitesuchthatachildcapableofcorrectlyperformingacertaintaskinthesuiteshouldalsobeabletocorrectlyperformallprecedingtasks.WellmanandLiu'sscaledsuiteoftaskshasbeensubsequentlyemployedtopinpointculturaldifferencesinthedevelopmentofToM(Wellmanetal.,2006,2011;Shahaeianetal.,2011). Inthiswork,weimplementandtestacomputerversionofWellmanandLiu's(2004)ToMsuiteinchildreninthe6to8-yearsoldrange.Theaimofthisworkisthreefold:first,wetestwhetheritispossibletousethesuitetotestToMdevelopmentinolderkidsandcheckifthescalingseeninpreschoolersisstillvalidinourversionofthetest.Theagerangechoseninthisworkissometimesoverlookedintheliterature,evenifithasbeenshownthatchildrendonotunderstandmetaphororironybeforetheageofsixtoseven(Ackerman,1981)—-twobehaviorsthatentailthecapacitytogobeyondtheliteralmeaningofastatement—-andthattheycannotreliablydistinguishjokesfromliesbeforeage6to7-years(Sullivanetal.,1995).Inaccordance,ourfirsthypothesisisthatToMprogressionofWellmanandLiu'ssuitewillstillbepresentinchildren6to8-yearsold. Second,giventhatgenderdifferencesmightbeexpectedinToM,andinlinewiththegenderintensificationhypothesis(HillandLynch,1983),whichestablishesthatgenderdifferencesincreaseintimebecauseofgrowingpressuretoconformtotraditionalgender-rolestereotypes,wehypothesizedthatinaslightlyoldergroupofchildren—-inrelationtotheusuallyagerangeexploreintheliterature(3to5-yearsold)—-gendermayhaveanappreciableeffectonToMperformance. Thethirdandfinalaimofthisworkistodevelopanovelanalysistoquantifythevalidityofthescalinginthesuite.Themethodweusehastwoparts:oneusessetinclusionstoquantifytheextenttowhichthedatadifferfromaperfectscaling,whiletheotherinvolvesthecomparisonbetweenspeciallytailoredsetsoflogisticmodels.Thedifferenceinpredictionpoweramongthesesetsofmodelsgivesanothermeasureofthescalingquality. MaterialsandMethods Participants Seventy-sixfirst(36)andsecond(40)graders[meanage:7-yearsand3months(86.5months);rangefrom6-yearsand1monthto8-yearsand7months]participatedinthestudy.Therewere42boysand34girlsinthesample,allofahighsocioeconomicstatusandattendingawellreputedprivatebilingualschoolinBuenosAires.Theschoolinwhichthestudywasperformedapprovedtheresearchandallchildren'sparentsorlegalguardiansgavesignedvoluntaryconsent.Theconsentform,presentedtothecaregiverssupplementedwithanotewhichexplainedtheprocedure,waspreviouslyauthorizedbytheCentrodeEducaciónMédicaeInvestigacionesClínicas“NorbertoQuirno”(CEMIC)'sEthicalCommittee. ToMSuite WellmanandLiu'sToMsuiteoftasksisthoroughlydescribedintheappendixoftheoriginalpaper(WellmanandLiu,2004).Briefly,thetasksinvolvedinourversionare:(1,DD)DiverseDesires:thechildjudgesthattwopersons(hervs.someoneelse)havedifferentdesiresaboutthesameobjects;(2,DB)DiverseBeliefs:thechildjudgesthattwopersons(hervs.someoneelse)havedifferentbeliefsaboutthesameobject,whileshedoesnotknowwhichbeliefistherightone;(3,KA)KnowledgeAccess:thechildseeswhatisinaboxandjudgestheknowledgeofanotherpersonwhodoesnotseewhatisinit;(4,FB)ContentsFalseBelief:thechildjudgesanotherperson'sfalsebeliefaboutwhatisinadistinctivecontainerwhileshe(thechild)knowswhatactuallyisinsidethecontainer;(5,EFB)ExplicitFalseBelief:thechildjudgeshowsomeonewillsearch,giventhatperson'smistakenbelief,and(6,BE)Beliefvs.Emotion:thechildjudgeshowapersonwillfeelgivenabeliefthatismistaken.Alltasksinvolveacontrolquestionwhichisusedtomakesurethatthechildunderstoodthetask,andatargetquestion,whichevaluatestheirperformance.AlthoughallsixtaskswereusedforstudyingtheinfluenceofdiversefactorsinToMdevelopment,wenotethatonlythefirstfour(DD,DB,KA,andFB)areinvolvedintheprogressivesuite,andthusallscalingtestswereperformedonlyonthese.Finally,wenotethatWellmanandLiu'soriginalversionofthesuiteincludedalsoahiddenemotiontask.Asthistaskinvolvestwotargetquestionsinsteadofonetargetandonecontrol,wechosetoleaveitoutinordertofacilitatedirectcomparison(therandomchoiceperformancebaselineforthistaskis33%insteadoftherestofthetasks'twochoice50%).TheimplementationofthesuiteisdepictedinFigure1. FIGURE1 Figure1.Descriptionofthetest.Thetestwasdeliveredtoeachchildinanovelcomputerformat.Theexperimenterfirstsaidtoeachchild:“Therearesixlevels;youwillhavetoanswercorrectlyineachoneofthemtowinthegame.Let'sgo!”Then,foreachtask,theexperimenterexplainedthetasktothechild,whocouldchooseanyoftheoptionsgiven.Therewerenowronganswers,suchthat,independentlyoftheanswer,a“CORRECT”screenappearednext.Threeofthetasksareshowninthefigure. InWellmanandLiu'sversionofthesuite,sometaskswerepresentedusingtoyfigurineswhileothersinvolveddrawings.Thismakesitdifficultbothtopreservetheparallelsamongthedifferenttasksandtocomparetheresultswiththoseinvolvingothermethodsfoundintheliterature.Further,followingouraimofcarryingthescaledToMsuitetoanolderagerange,weformulateditasanengagingcomputergame,whichnotonlyunifiesthepresentationformatacrossalltasks,butalsoreducestheexperimenter'sinvolvement. Procedure Childrenweretestedinaquietroomintheschoolbyoneoftwoadultexperimenters.ThesixtaskswerepresentedintheincreasingdifficultyorderproposedbyWellmanandLiu(2004).Allthechildrenthatparticipatedinthestudycorrectlyansweredthecontrolquestionineachtask. TeacherscompletedtheshortformoftheChildBehavioralQuestionnaire(CBQ)foralltheirstudents.Thisallowedustomeasurechild'stemperamentalongthreebroaddimensions:(1)Extraversion/Surgency,(2)NegativeAffectivity,and(3)EffortfulControl(Rothbartetal.,2001;PutnamandRothbart,2006).Theyalsofilledasecondformthatincludedthefamilybackgrounddata(birthorderandnumberofsiblings),age,andgenderofeachchild. DataAnalysis Inordertosummarizethechildren'sperformanceintheToMsuitewecomputeforeachchildherz-score,definedastheamountofcorrecttargetanswers.Alternativemeasureswerealsotested,inwhichthecontributionofeachtasktothetotalscorewasweighedeitherprogressivelywiththetasknumberorwithafactorequaltooneminustheobservedmeanperformanceforthattask,andthenallcontributionssummedtogetherinafinalscore.Theresultsobtainedareinsensitivetothescoringschemechosen;wehencesticktothez-scoreabove. Throughouttheanalysis,weusednon-parametricpermutationtestsinordertoassessthesignificanceofresults.Ineachcase,werandomlyshuffledtheassignmentsbetweenpredictoranddependentvariables,toproduceasurrogateversionofthedata.Therelevantquantity(forinstancecorrelation)wasthenevaluatedforthissurrogatedata.Byiteratingthisproceduremanytimes(typicallyathousand),weobtainedthesignificanceleveloftheresult. Apartfromstudyingthecorrelationsinthedata,webuiltalogisticmodeltoassesskids'responsesforalltasks.Withthismodel,wecanstudytheeffectofthedifferentfactorsinthewholesetofresponses,withoutlimitingtheanalysistoaparticulardefinitionofscore.Nevertheless,acertainamountofindependenceamongthefactorsenteringthemodelisrequiredinordertocorrectlyinterprettheresults,andhencewerestrictthemodeltoage,gender,andnumberofsiblingsandbirthorder(consideringtheselasttwoasasinglefactorinordertoaccountfortheirinterdependence).CBQscores,ontheotherhand,wereleftout,sincetheycorrelatemildlywithgender(ex.0.4linearcorrelationbetweensurgencyandgender). Apartfromgender,age,numberofsiblings,andbirthorderaspredictors,themodelincludesdummyvariablesfordiscerningamongthesixtasksinthesuite.Sinceeachkidrespondstoallsixtasks,wealsoneedtoindexthesubjects.Asitturnsout,age,sex,siblingamount,andsiblingordertakentogetherarealmostenoughtoidentifyallsubjects.Therearehowever,fivecasesinwhichtheserepeat,soweresolvethembyaddinganextraindicatorvariable(i.e.,orderinwhichtheytooktheexperiment). Tomeasuretheimportanceofeachfactorinpredictingtheresults,wecomparethefullmodelwiththatwiththefactorinquestionremoved.Thedifferencebetweentheloglikelihoodsofbothmodelsfollowsaχ2distribution[apartfromafactorof2,seeStevenson(2004)],andwecanhenceevaluatethecorrespondingp-value.Thep-valuesthusobtainedareinagreementwiththosecomputedwiththefullmodelundertheassumptionofnormallydistributederrors. Results ToMPerformancein6to8-YearsOldChildren DespitethefactthatsomestudiesindicatethatthedevelopmentofafullToMcontinuesallthroughoutlife(DevineandHughes,2012;Moran,2013),almostallresearchhasfocusedin3to5-yearsoldchildren.Wetookastepfurthertocontributetoelucidatethesenotionsandweexaminedage,gender,andfamilybackgroundinfluenceonToMdevelopmentintheagerangeof6to8yearsold.Figure2showstheaverageperformanceforthefourtasksinvolvedinthescalingforallchildren(thickblackbars).TheaverageperformanceforourfirstandsecondgradersisinthesamerangeasthatofthepreschoolersstudiedbyWellmanandLiu,suggestingthatthesmallerinterventionoftheexperimenterenabledbythecomputerplatformallowsforthetestingofolderkidswithoutsaturatingthesuite.AverageperformancefortasksEFBandBEwas0.64and0.6,respectively,alsosimilartothatofWellmanandLiu'spreschoolers.Thesetaskswerenotincludedinthegraphtoemphasizetheprogressioneffectinthefirstfour. FIGURE2 Figure2.Kids'averageperformance.Fractioncorrect,mean,thickblackbars;boys,leftthinbluebars;girls,rightthinredbars. InaccordancewithourfirsthypothesisthatToMprogressionofWellmanandLiu'ssuitewillstillbepresentin6to8-yearsold,weobservedanincreaseinToMproficiencywithage,withz-scoressignificantlycorrelatedwithage(r=0.334,p=0.003permutationtest),Thisresultstressesevenfurtherthefactthatthetestiseffectiveinthenewagerange,andshowsthatthegeneralperformancedropisnotmerelyduetoastatisticalfluctuation. AverageperformancedividedbygenderisalsoshowninFigure2(boysinleftbluethinbarsandgirlsinrightredthinbars).FortaskEFBwehaveaverageperformancesof0.7forgirlsand0.6forboys,whilefortheBEtaskweobtain0.76forgirlsand0.48forboys.WeobserveacleareffectofgenderinToMperformance,withgirlsperformingsignificantlybetterthanboysinalltasks(p=0.037permutationtest,groupedtasks).Thisgendereffectistask-independent,asweshowusingthelogisticmodeldescribedinthefollowingsection. Wealsostudiedtheeffectoffamilybackgroundandchildtemperament,factorswhosepotentialinfluenceinToMperformancehasbeenpreviouslydiscussedintheliterature(Dunnetal.,1991;Perneretal.,1994;Dunn,1995;Farhadianetal.,2010).Contrarytoageandgenderfindingsandinagreementwithpreviousdata(Lewisetal.,1996;CuttingandDunn,1999)wefindnoeffectofsiblingamount(r=0.034,p=0.766,permutationtest)orbirthorder(r=−0.043,p=0.729,permutationtest)inToMperformance. Similarly,coarsepersonalitytraitsasevaluatedbytheCBQdonotcorrelatesignificantlywiththez-score(Surgency:r=−0.169,p=0.149;NegativeAffect:r=0.197,p=0.093;EffortfulControl:r=0.118,p=0.315;permutationtests). Thelogisticmodelanalysisfullysupportstheresultsdiscussedabove.Bytakingonefactoratatimeoutofthemodel,wecanseehowrelevanteachfactorisinexplainingthekids'responses(seeMaterialsandMethods).Thecomputedp-valuesare:age,p=0.002;sex,p=0.015;siblings(amountandorder),p=0.978.Bytestinganexpandedmodelincludingajointsex-taskfactor,weseethatthesetwovariablesdonotinteract(p=0.47).Hence,inaccordancewithoursecondhypothesisweobservedagendereffectinToMwhichwasnotobservedforthesametaskinpreschoolers.Theotherdemographicorindividualvariableshadnoeffectinperformance. ToMSuiteProgression WenowturntoquantifytheextenttowhichtheToMsuiteembodiesaprogressivetest,thatis,oneinwhichinordertocorrectlyresolveagivenstep,onemusthavetheabilitiesrequiredtosolveallprevioussteps.Aswementionedbefore,weonlyconsidertasks1–4(DD,DB,KA,andFB)forthispartoftheanalysis,sinceitisonlythesethatareinvolvedinthehierarchy. TheprogressionindifficultyofToMperformanceisevidentfromFigure2,whichshowsadiminishingperformancewithincreasingtasknumber.Anaveragedecreaseinperformance,however,doesnotnecessarilyimplyasequentialprocess.ItisstillpossiblethatagroupofchildreniscapableofsuccessfullyperformingTask1(G1)andasmallergroupiscapableofsuccessfullyperformingTask2(G2)butthatthesegroupshavenointrinsicrelationotherthantheirdifferenceinsize,i.e.,|G2|