Who knows what's good to eat? Infants fail to match the food preferences of antisocial others

13 Pages·2012·0.51 MB·English

by  J. Kiley Hamlin

CognitiveDevelopment27 (2012) 227–239

ContentslistsavailableatSciVerseScienceDirect

Cognitive Development

Who knows what’s good to eat? Infants fail to match the

food preferences of antisocial others

J.KileyHamlina,∗,KarenWynnb

aDepartmentofPsychology,TheUniversityofBritishColumbia,2136WestMall,Vancouver,BCV6T1Z4,Canada
bDepartment of Psychology, Yale University ,2 Hillho useAve,Ne wHa ven,C T065 11,UnitedS tat es

a r t i c l e i n f o a b s t r a c t

Keywords: Humansgathermostoftheirknowledgeabouttheworld,includ-
Socialevaluation ingobjec tivelyt ruefa cts and specificcult uralno rm s,byob serving
Social learning/imitation and beingtaug htb yoth ers. Somein dividual sarew ort hyteach-
Negat ivity bias ersa ndobj ectsof imi tation,h aving knowledgeo fcu lturalpr actices

andpositiveintentionstoinform.Othersarebetterignoredbecause

theyareignorant,becausetheymeanusharm,orsimplybecause

wedonotwishtobe“likethem.”Thisstudyexamineswhether

16-month-oldsaresensitivetothepro-orantisocialbehaviorofa

sourcethatdemonstratespreferencefortwonovelfoods.Infants

took the emotional reactions displayed by novel and previously

prosocial sources, but not antisocial sources, into account when

deciding what to eat. These results suggest that others’ social

behaviorinfluencesinfants’likelihoodtomatchtheirpreferences,

illustratingtheinfluenceofsocialevaluationonsociallearning.

© 2012 Elsevier Inc. All rights reserved.
Most of what we know comes from others. By observing and interacting with others socially,

wegainawealthofinformationrequiredforsuccessfulexistence,fromthelanguageandcultural

practicesofoursocialgrouptotheworkingsofaparticularpieceofcomplexmachinery.Humans,

even in infancy, appear uniquely adapted to using others as sources of information, through such

mechanismsasthereproductionofnovelbehaviors(Bandura,1977;Barr,Dowden,&Hayne,1996;

Meltzoff,2007;Tomasello,1999;Tomasello,Kruger,&Ratner,1993),theuseofemotionalreactionsto

appraiseunfamiliarobjectsandevents(Campos&Stenberg,1981;Feinman,1982;Klinnert,Campos,

Sorce,Emde,&Svejda,1983),andsensitivitytodirectedpedagogicalinteractions(Csibra&Gergely,

2006;Gergely&Csibra,2005,2006).Usingtheseandothersociallearningmechanisms,infantsrapidly

∗
Correspondingauthor.Tel.:[phone];fax:[phone].

E-mailaddress:[email protected](J.K.Hamlin).

0885-2014/$–seefrontmatter© 2012 Elsevier Inc. All rights reserved.

http://dx.doi.org/10.1016/j.cogdev.2012.05.005
228 J.K.Hamlin,K.Wynn/CognitiveDevelopment27 (2012) 227–239

acquireinformationandskillsthatwouldbedifficultorimpossibletolearnviafirstpersontrial-and-

error.
Despitethepowerfulinfluenceofsocialsourcesinhumandevelopment,researchoverthepast

twodecadessuggeststhatchildrendonottakeinalltheinformationasourceprovides.Instead,they

selectivelyincorporatesomepiecesofinformationandignoreothers.Inimitativeinteractionsearlyin

thesecondyearoflife,infantsdonotsimplyreproduceeveryactionamodelperformsbutseemingly

incorporatetheirownanalysisoftheintentionalandcausalstructureofactionintotheirimitative

acts.Forexample,youngtoddlersfailtoimitatebehaviorsmarkedasaccidental(Carpenter,Akhtar,

&Tomasello,1998)andreproduceonlythosepiecesofactionsthatarecausallyrelatedtoamodel’s

overarchinggoal,leavingoutunnecessaryorsubordinatecomponents(Brugger,Lariviere,Mumme,

& Bushnell, 2007; Carpenter, Call, & Tomasello, 2005; Gergely, Bekkering, & Király, 2002; Nielsen,

2006;butseeHorner&Whiten,2005;Lyons,Young,&Keil,2007,forevidenceoffaithfulimitation

inchildren).Additionally,infantssometimesgo“belowthesurface”ofobservedbehaviors,effecting

end-statesofgoal-directedactionsthatwereneveractuallycompletedandthereforemusthavebeen

inferred(Bellagamba&Tomasello,1999;Johnson,Booth,&O’Hearn,2001;Legerstee&Markova,2008;

Meltzoff,1995).Theseresultssuggestprivilegedrolesforthementalandcausalstructuresdrivinga

source’sbehavior,overthephysicalnatureoftheactionsthemselves,asinfantsgatherinformation

fromsocialothers.

Inadditiontoselectingwhattolearnfromothers,itisalsoimportanttoselectfromwhomtolearn.

Someindividualshaveinformationusefultoinfants,whereasothersmaynot(Feinman,1982).There

areavarietyofnon-mutually-exclusivedimensionsonwhichonemightdiscriminatethevalueof

informationfrompotentialsources.Perhapsthemostbasicisaccuracy:whetherasource’sinforma-

tionisobjectivelytrueorfalse.Infantsinthesecondyeararesensitivetoothers’knowledgestates

(Liszkowski,Carpenter,&Tomasello,2008;Onishi&Baillargeon,2005;Tomasello&Haberl,2003),

and can identify (and sometimes attempt to correct) false statements (e.g., when a speaker labels

acupa“dog;”Koenig&Echols,2003;Pea,1982).Agrowingbodyofresearchshowingimpressive

selectivityinpreschoolchildren(Birch&Bloom,2002;Birch,Vauthier,&Bloom,2008;Corriveau&

Harris,2009a;Jaswal&Neely,2007;Koenig&Harris,2005;Rakoczy,Warneken,&Tomasello,2009;

Sabbagh&Baldwin,2001)suggeststhatsensitivitytoepistemicstatesplaysaroleinhowinfants

useothersassourcesofnewinformation.Forexample,16-month-oldsfailtolearnnewwordsfrom

previouslyinaccuratelabelers(Koenig&Woodward,2010),14-month-oldsarelesslikelytofollow

thegazeofsomeonewhowaspreviouslyunreliable(e.g.,whoexpressedexcitementwhilelooking

intoanemptycontainer;Chow,Poulin-Dubois,&Lewis,2008),and14-month-oldstendnottoimitate

thenovelactionsofthosewhoincompetentlyperformedfamiliaracts(e.g.,putshoesontheirhands;

Zmyj,Buttelmann,Carpenter,&Daum,2010).Theseresultssuggestthatevenininfancy,childrenare

sensitivetopastaccuracyanddonotlearnfrominaccuratesources.

However,pastaccuracyisnotalwaysdirectlyavailableinalearningsituation.Therefore,itwould

be beneficial to be sensitive to a source’s potential for accuracy, or, in other words, that source’s

general expertise (Henrich & Gil-White, 2001; Lutz & Keil, 2002). Expertise may be evident from

one’sconfidenceand/orfamiliaritywithasituation.Themoreconfidenceorfamiliarityanindividual

displays,themorelikelyalearnershouldbetoselectthatindividualtolearnfrom.Inanoveltoy

situation,infantsasyoungas12monthsselectivelyreferenceanexperimenterknowntobefamiliar

withalaboratoryratherthantheirmother(unfamiliarwiththelab),presumablyduetoanassumption

thattheexperimenter’sfamiliaritymakeshermoreknowledgeable(Stenberg,2003,2009;Walden&

Kim,2005).

Further,somesourcesshouldbeselectedoverothersduenottotheirobjectiveaccuracyorexper-

tise,butbecauseofthepotentialrelevanceoftheirinformationtothelearner.Thingslikelinguistic

forms,ritualbehaviors,andfoodandobjectusemaydifferwidelyacrossculturesandgroupsand

thereforeshouldonlybeacquiredfromsourceswhoknowtheformsrelevanttoaparticularlearner.

Determiningwhohasculturallyrelevantinformationisnotnecessarilyeasy,butcouldbefacilitated

byselectivelyattendingtosourceswhoarefamiliar,similartothelearner,orknowntobewithinthe

learner’ssocialgroup.Infants’tendencytosociallyreferencehighlyfamiliarindividuals(likemothers)

versusunfamiliarindividuals(likeresearchers)isambiguous.Somestudieshavefoundthatinfants

shownopreferenceforcaregiversoverexperimenters(Devouche,2004;Klinnert,Emde,Butterfield,
J.K.Hamlin,K.Wynn/CognitiveDevelopment27 (2012) 227–239 229

& Campos, 1986; McCabe & Uzgiris, 1983; Meltzoff & Moore, 1992), others show a preference for

experimenters(Stenberg,2003,2009;Walden&Kim,2005),andstillothershaveshownacaregiver

advantage(Zarbatany&Lamb,1985).Despitethisambiguity,preschoolersdoselectivelylearnfrom

familiaroverunfamiliarteachers(Corriveau&Harris,2009b),infantsselectivelychooseobjectsand

foodsprovidedbyindividualswhospeaktheirnativelanguage(Kinzler,Corriveau,&Harris,2011;

Kinzler, Dupoux, & Spelke, 2007; Shutts, Kinzler, McKee, & Spelke, 2009), and 14–18-month-olds

morecloselyimitatetheactionsofmodelsthatarenearertotheminage,evenaweekaftertheinitial

modeling(Ryalls,Gul,&Ryalls,2000).Together,theseresultssuggestthatfamiliarity,similarity,and
groupme mbershi ppla y arolei ninfan ts’source select ionand informa tion retention.1

Alsopotentiallyrelevantiswhetherasourceisliked.Likingisoftenrelatedtofamiliarityandsim-

ilaritybutcanalsoexistindependently.Selectivityforpositivelyand/ornegativelyevaluatedsources

mightserveavarietyofinter-relatedfunctions.First,ifadislikedsourceinturndislikesthelearner,

orisanantisocialindividualingeneral,heorshemaymeanthelearnerharm.Ifso,thesource’sinfor-

mationmaybeintentionallymisleadingandshouldbeignored(Mascaro&Sperber,2009;Vanderbilt,

Liu,&Heyman,2011).Second,evenifadislikedsource’sinformationisobjectivelytrue,alearnermay

stillchoosetoignoreittoavoid“beinglike”or“affirmingasharedstatewith”thesource.Achieving

asenseofsharedlikenesshasbeentheorizedtobeasocialfunctionofsociallearningmechanisms,

inparticularimitation(Hobson&Lee,1999;Lakin&Chartrand,2003;Meltzoff&Moore,1995;Nadel,

Guérini,Pezé,&Rivet,1999;Over&Carpenter,2009;Over&Carpenter,2012;Uzgiris,1981).Ifso,

individuals should avoid accepting information from those they have negatively evaluated. In one

seriesofstudies(Nielsen,2006;Nielsen,Simcock,&Jenkins,2008),infantsinthesecondyearoflife

morecloselyimitatetheactionsofavailable,“sociallyengaged”individualsthansocially-unavailable,

“aloof” individuals, potentially indicating a role for social evaluation in imitative learning. Shutts,

Kinzler,etal.(2009)foundthatinfants’tendencytomatchthefoodpreferencesofnativeoverforeign

languagespeakerswasenhancedwhenthenativelanguagespeakerdisplayedpositiveaffect(pre-

sumablylikedbytheinfants)andtheforeignlanguagespeakerdisplayednegativeaffect(presumably

disliked).Theseresultssuggestaroleforliking/dislikinginsourceselectivityininfancy;however,they

donotprovideadirecttestofthishypothesis.

Thepresentstudydirectlyexaminesthelinkbetweensocialevaluationandthetendencytogain

anduseinformationfromsocialsourcesininfancy.LikeShutts,Kinzler,etal.(2009),wedidthisin

thedomainoffood,askingwhetherinfants’tendencytoeatthefoodsasourcelikesoverthefoodsthe

samesourcedislikesisinfluencedbythesource’spreviousthird-partyprosocialorantisocialbehavior.

Sixteen-month-oldinfantsobservedasinglepuppetsourcetastetwosimilar,novelfoods;thesource

expressedlikingforonefoodanddislikingfortheother.Infantswerethengiventheopportunitytoeat

thefoodsthemselves.Weobservedtheextenttowhichinfantsmatchedthesource’spreference.Prior

tothefooddemonstrationphase,allinfantswatchedthesamepuppetshow,inwhichapuppettried

butfailedtoopenaboxcontainingatoy.Theunsuccessfulpuppetwasthenhelpedinhisgoaltoopen

theboxbyasecond“prosocial”characterandhinderedinhisgoalbyathird“antisocial”character.

Thus,intheProsocialSourcecondition,thesourceoffoodliking/dislikinginformationhadpreviously

helped a third-party to obtain a desired object; in the Antisocial Source condition, the source had

hinderedthethird-party.Finally,inathird,NovelSourcecondition,althoughinfantsbeganbywatching

aprosocial/antisocialpuppetshow,thesourcewhosubsequentlyexpresseditsfoodpreferencewas

notapartoftheshowandwasthereforeunfamiliartotheinfants.

Throughoutthefirsttwoyearsoflife,infantspositivelyevaluatetheprosocialandnegativelyeval-

uatetheantisocialcharactersinthisexactscenario.Theseevaluationsareevidentininfants’affiliative

behavior(e.g.,reaching;Hamlin&Wynn,2011),intheirsocialresponses(e.g.,whotheygivetreatsto

andtaketreatsawayfrom;Hamlin,Wynn,Bloom,&Mahajan,2011),andintheirjudgmentsofothers

whodirectpositiveandnegativebehaviorstotheprosocialandantisocialcharacters(Hamlinetal.,

1 Whiletoourknowledgetherearenoadditionalstudiesreportingselectivelearningfromsimilarindividualsininfancy,

alargebodyofresearchsuggeststhatpreschoolchildrengaininformationfromsimilaroverdissimilarothersalongmultiple

dimensions,preferringthesameobjects,foods,andactivitiesasthosewhoaresimilartotheminage(ShuttsBanaji,&Spelke,

2010;c.f.VanderBorght&Jaswal,2009),andwhoaretheirsamegender(Bradbard&Endsley,1983;Bradbard,Martin,Endsley,

&Halverson,1986;Martin,Eisenbud,&Rose,1995;Martin&Little,1990;Shuttsetal.,2010).
230 J.K.Hamlin,K.Wynn/CognitiveDevelopment27 (2012) 227–239

2011).Thisbreadthofsocialresponses,aswellastheuseofpuppetsinmanyotherstudiesofyoung

children’ssocialjudgmentandinteraction(Olson&Spelke,2008;Rakoczyetal.,2009;Vaish,Missana,

&Tomasello,2011),suggeststhatinfantstreatthesepuppetsassocialotherswhoareappropriate

tointeractwith,toevaluate,and(potentially)touseasasourceofinformation.Hereweexamine

thisfinalsuggestion,byaskingwhetherasource’spreviousprosocialorantisocialbehaviorinfluences

infants’tendencytomatchthatsource’sfoodpreference.

1. Method

1.1. Participants

Previousstudiesonselectivelearningininfancyhaveinvolvedinfantsinthefirsthalfofthesecond

yearoflife(Chowetal.,2008;Koenig&Woodward,2010;Shutts,Condry,etal.,2009;Shutts,Kinzler,

etal.,2009;Zmyjetal.,2010).Asourprocedurerequiredgraspingsmallitemsoutofsmallbowlsand

eatingfingerfoods,wechosetostudy16-month-olds.Forty-eightinfantswererandomlyassignedto

threeconditions:ProsocialSource(sevenboys;meanage16months,3days,range15months,15days

to16months,25days),NovelSource(eightboys;meanage16months,0days,range15months,14

daysto16months,16days),andAntisocialSource(sevenboys;meanage15months,26days,range15

months,12daysto16months,17days).Nineadditionalinfantswereexcludedduetoparentalinter-

ference(3),proceduralerror(2),orrefusaltoparticipate(4).Demographicinformationwascollected

fromonethirdofparticipants;infantswere71%Caucasian/Non-Hispanic,11%Caucasian/Hispanic,6%

Black/Non-Hispanic,6%NativeAmerican/Hispanic,and6%NativeAmerican/Non-Hispanic.

1.2. Stimuliandprocedure

1.2.1. Eatingwarm-up

Infantssatonaparent’slapintheofficewaitingroom.Anexperimenterholdingawhitebowl

containingyellowFruityCheerioskneeledinfrontoftheinfant,said“Look!”andateaCheerio,saying

“Mmm!”Shethenofferedthebowltotheinfant,encouragingthechildtoeatuntiltheinfantwillingly

tastedCheeriostwice.

1.2.2. Prosocial/antisocialpuppetshows(viewableonlineat

www.yale.edu/infantlab/socialevaluationpreferencematching)

Regardlessofeventualsourcecondition,allinfantswatchedthesameprosocial/antisocialpuppet

show.Infantssatintheparent’slapbeforeatablewithacurtainatitsfarend(165cmfromtheinfants)

thatcouldbeloweredtooccludetheshow.Puppetswereapproximately10in.high.Theprotagonist

puppetwasalwaysabrowndog,theprosocialandantisocialpuppetswere(counterbalancedacross

infants in each condition) a grey and an orange cat (Prosocial/Antisocial Source conditions) and a

black&whiteandagreycat(NovelSourcecondition).Parentswereinstructedtositquietlywiththeir

infantsandnotattempttoinfluencetheminanyway.

PuppetshowswereidenticaltothoseinExperiment1byHamlinandWynn(2011).Duringall

trials,thecurtainraisedtorevealastagesurroundedonthesidesandbackbyblackcurtains;theback

curtainoccludedapuppeteer.Theprosocialandantisocialpuppetssatattherearcornersofthestage,

andaclearboxcontainingarattlerestedinthemiddle.Theprotagonistenteredfromthebackofthe

stage,movedtowardonesideofthebox,andleaneditsheaddowntolookthroughthesideofthe

boxtwice.Itthenattemptedtoliftthelidoftheboxfourtimes,liftingslightlyanddroppingthelid

twice,andthenliftingslightlyandloweringthelidtwice.Duringthefifthattempt,thepuppetonthe

oppositesideofthestagefromtheprotagonist(whoenteredonalternatingsidesoftheboxduring

eachtrial)intervened(seeFig.1).

Duringprosocialevents,theprosocialpuppetrantothefrontofthebox,grabbedthecornerofthe

lid,andopenedittogetherwiththeprotagonist.Theprotagonistdovedowninsidethebox,grabbing

therattle.Theprosocialpuppetthenranoff-stage,andtheprotagonistliftedtherattlefromthebox.

Allactionceased.
J.K.Hamlin,K.Wynn/CognitiveDevelopment27 (2012) 227–239 231

Fig.1. Stimulipresentedtoinfants:rows1and2representtheprosocialandantisocialpuppetshows,inwhichadogpuppet

triedbutfailedtoopenaclearboxcontainingarattle.Onprosocialevents(row1)theprosocialpuppethelpedopentheboxand

thedogobtainedtherattle.Onantisocialevents(row2)theantisocialpuppetjumpedontheboxandthedogdidnotobtainthe

rattle.Row3representsthefoodpreferencepresentationbyasourcewhowaspreviouslyeitherprosocial,novel,orantisocial.

Duringantisocialevents,theantisocialpuppetrantothefrontoftheboxandjumpedsideways

ontothelid,slammingitshut.Theprotagonistdovedownnexttothebox.Theantisocialpuppetthen

ranoff-stage,andtheprotagonistsatupnexttothebox.Allactionceased.

Bothprosocialandantisocialeventslastedapproximately15s.Anindependentcoderrecorded

infants’lookstothedisplaythroughaholeinthecurtainfromthetimeactionceaseduntiltheinfant

lookedawayfortwoconsecutiveseconds,oruntil30selapsed,asinpreviousstudies(Hamlin,Wynn,

&Bloom,2007;Hamlin,Wynn,&Bloom,2010;Hamlinetal.,2011).Infantswerefamiliarizedtothree

prosocialandthreeantisocialeventspresentedinalternation,forsixtotalevents.Thefollowingwere

counterbalancedacrossinfantswithineachcondition:colorofprosocial/antisocialpuppets(black&

whiteorgreyfortheNovelSourcecondition,andgreyororangefortheProsocial/AntisocialSource

conditions),sideofprosocialpuppetduringevents(rightorleftsideofstage),andorderofprosocial

events(firstorsecond).

1.2.3. Preferencedisplay(viewableonlineat

www.yale.edu/infantlab/socialevaluationpreferencematching;seeFig.1)

Infantssatontheparent’slapatatable,approximately105cmfromtheexperimenterwhohad

performedtheeatingwarm-up.(Thisexperimenterhadnotobservedtheprosocial/antisocialpuppet

show and was therefore blind to the identity of the source puppet in the Prosocial and Antisocial

Sourceconditions.)Parentswereaskedtoclosetheireyes.Theexperimenterplacedtwoclearbowls

onthetableinfrontofher,approximately30cmapart,onecontainingredFruityCheeriosandone

containingpurpleFruityCheerios.

Theexperimenterbroughtoutanorangecatpuppet(theProsocial,Novel,orAntisocialSource)on

herrighthandandspoketoit,saying“HiKitty!”andmakingitwavetotheinfant.Shethenaskedthe
232 J.K.Hamlin,K.Wynn/CognitiveDevelopment27 (2012) 227–239

sourcepuppet:“Doyouwanttotrysome?”andmadeitnod“yes”towardtheinfant.Theexperimenter

thenmovedthesourcepuppettowardoneofthefoodbowls,‘tasting’theCheeriosinitbyleaning

intoandoutofthebowlandmakingeatingsounds.Aftertasting,thesourcepuppetexpressedeither

apositive(“Mmmm!Yum!Ilikethat!”inahigh-pitched,positivevoice)ornegative(“Blech!Yuck!I

don’tlikethat!”inalow-pitched,negativevoice)opinionoftheCheeriosinside.Afterexpressingits

preference,thesourcepuppetmovedbacktowardtheexperimenter,whoasked:“Doyouwanttotry

theotherone?”andmadeitnod“yes”towardtheinfant.Thesourcethentastedfromthesecondbowl

andexpressedtheoppositeopinion.

EachinfantsawthesourcetasteboththeredandpurpleCheerios,expressinglikingforoneand

dislikingfortheother.Afterthetastepreferencesweremodeled,theexperimentersaid“ThanksKitty!

Byebye!”andthesourcepuppetwavedtotheinfantandleft.

Becausetheywereaskedtoclosetheireyesforthepreferencedisplay,parentsknewneitherthe

identityofthesourcepuppetnorwhichfooditpreferred.Theexperimenterwasblindtotheprevious

behaviorofthesourcepuppetintheProsocialandAntisocialSourceconditions.Thefollowingwere

counterbalancedacrossinfants:locationofredCheerios,whetherredCheerioswerelikedordisliked,

andwhetherlikingordislikingwasexpressedfirst.

1.2.4. Foodchoice

Theexperimentermadeeyecontactwiththeinfant,pushedthetwofoodbowlswithinhisorher

reach,andsaid,“Doyouwanttotrysome?”Shethenwaiteduntiltheinfanthadmadefourconsecutive

foodchoices.Theexperimenterusedavarietyofwaystoencourageinfantstoeat.Encouragement

waspermissibleaslongasaninfantwasnotcurrentlyholdinganyCheerios,norhadhisorherhands

ineitherbowl,toavoidpromptingtheinfanttoeatfromaparticularbowl.Iftheinfantwasdistracted,

theexperimentershookbothbowlsorknockedthemtogetherlightlyandrepeatedtheinvitationto

trysome.IftheinfantseemedreluctanttotryanyCheerios,theexperimenterencouragedeatingby

saying,“It’sOK,youcantrysome,”or“Youcaneatsomeifyouwant.”Althoughparentswereinstructed

beforethestudynottotalktotheirinfantsduringthetask,ifinfantswereveryreluctanttoeat,the

experimenterpromptedparentstotelltheinfant,“It’sOK.”Eatingandencouragingcontinueduntil

infantshadmadefourclearfoodchoices.

1.3. Coding

1.3.1. Attentionfollowingpuppetshows

Infants’attentionwascodedbyanonlinecoderfollowingeachpuppetshow(fromthetimeaction

ceased, as described above). We calculated how long infants watched following prosocial puppet

eventsandfollowingantisocialpuppeteventstoexaminewhetherthisinfluencedinfants’laterfood

choices.Asecondindependentcodercodedarandom25%ofinfants’attentionineachcondition.The

inter-ratercorrelationswere.995,.98and.97fortheProsocial,Novel,andAntisocialSourceconditions,

respectively.
1.3.2. Attentioncodingtofoodpreferencedisplay

To examine the effect of infants’ attention to the sources on eating behaviors, an independent

coder(blindtocondition)codedeachinfant’sattentiontothefoodpreferencedisplayfromvideotape.

Becausepreferencewasdisplayedliveandtherewerethereforesmalldifferencesinthetotallengthof

eachpresentation,wealsocodedthetotaldisplaylengthforeachinfantanddeterminedtheproportion

ofthetotaldisplaytimethateachinfantattendedto.Asecondindependentcodercodedarandom

25%ofinfants’attentionineachcondition.Theinter-ratercorrelationcoefficientwas.77.

1.3.3. Foodchoicecoding

The experimenter coded infants’ choices as the first four times an infant grasped one or more

Cheeriosfromabowlandatethemwithoutinterruption.“Interruptions”werecodedasanybehavior
J.K.Hamlin,K.Wynn/CognitiveDevelopment27 (2012) 227–239 233

thatcouldhaveledtoaninfantnotknowingwhichbowlaCheeriohadcomefrom2andtookavariety

offorms,includingdroppingCheeriosontoclothingorthetableandlaterfindingandeatingthem,

picking up multiple different-colored Cheerios with both hands and moving them back and forth

betweentheirhands.Theexperimenterdeterminedwhetherachoicehadbeeninterrupted.Eating

continueduntiltheinfantmadefournon-interruptedchoices.Eachinfantreceivedascoreindicating

thenumberoftrials(offour)inwhichtheyatethefoodthepuppetmodelhadliked.

Anindependentcoderblindtothemodel’sidentityrecodedarandomlychosen25%ofinfants’

choicesandagreedwiththeoriginalexperimenteron100%oftrials.

2. Results

2.1. Attentiontopuppetevents

Collapsedacrosscondition,infantslookedlongeratprosocialpuppetshowevents(mean=32.58s

of total looking over three trials, SEM=3.35) than antisocial puppet show events (mean=26.25s,
SE M=2 .69); F(1 ,47)= 10.08 , p<.0 05, (cid:2) 2 =.1 8. Th is tendenc y did not di ffer ac ross co n dition s,
p
F(2,45)=.75,p=.48,(cid:2)2=.03,nordidinfants’relativeattentiontoprosocialversusantisocialevents
p
(adding this as a covariate in the choice analyses) influence their food choices across condition,
F(1,46)= .004 ,p =. 95,(cid:2)2=.0 0, orw ithinan yconditio nalone(a llps> .28).
p
2.2. Attentiontothepreferencedisplay

InfantsintheProsocialSourceconditionlookedtothepresentationanaverageof95%ofthetime

(SEM=1%);infantsintheNovelSourceconditionlookedanaverageof95%ofthetime(SEM=1%),and
infant s inth eAntis oc ialS ource conditio nlooked anaver age of96%o ft het im e(S EM= 1%).3 Th ism ea-

suredidnotdifferacrossconditionaccordingtoaunivariateanalysisofvariance(ANOVA),F(2,40)=.32,
p=.7 3,(cid:2) 2= .02), andwh enadded asacova ria te didnotsi gnifican tly influenc einfants’ foodch o ices
p
acrosscondition,F(1,39)=.76,p=.40,(cid:2)2=.02,orwithinanycondition(ps>.35).
p
2.3. Foodchoice

Infants’foodchoicesbyconditionarerepresentedinFigs.2and3.AunivariateANOVAinclud-

ingcondition,sexofinfant,orderofprosocialactionduringfamiliarization,coloroflikedCheerios,

and order of liking event as between-subject factors, and including age as a covariate revealed
only aneff ect ofcon dition on infants’foodcho ices,F(2, 45)= 6.094,p< .05, (cid:2)2 = .484,andn osignif-
p
icantinteractions.Plannedcontrastsusingone-samplet-testsrevealedthatinfantsintheProsocial

Sourceconditionsignificantlyatethesource’spreferredoverdispreferredfood,mean(SEM)=3.0(.27);

t(15)=3.65,p<.005,asdidinfantsintheNovelSourcecondition,mean(SEM)=2.88(.26),t(15)=3.42,

p<.005. In contrast, infants who saw the antisocial puppet’s food preference ate the two foods
eq u ally, me an(SEM) =1.69(.2 5);t( 15)= −1.2 3,p=.24.T hissignifi cant tendencyto ma tcht heso urce’s

preference in the Prosocial and Novel Source conditions but not in the Antisocial Source condi-

tionwasevidentinnon-parametricKolmogorov–Smirnovone-sampletests,Dmax(Prosocial)=.437,

p<.01,Dmax(Novel)=.375,p<.05,Dmax(Antisocial)=.187,p>.05,andinindividualinfants’patternsof

response.Thatis,12of16infantschosetheProsocialSource’spreferredfoodmorethanhalfthetime

(onthreeorfouroffourchoices,chance=31.5%,binomialtest,p<.001),11of16infantsintheNovel

Sourceconditiondidso(binomialtest,p<.005),andonlytwoof16infantsintheAntisocialSource

2 Whileourtwofoodsweredistinguishablebybothcolorandlocation,itisunclearwhichofthesecuesinfantsusedto

distinguishthem.Recentworksuggeststhatyounginfantsmaynotyetcategorizefoodtypesasadultsdo(i.e.,uniquelyby

substance;Birch,1980;Brody&Stoneman,1981;c.f.VanderBorght&Jaswal,2009;Hendy&Raudenbush,2000;Jaswal&

Neely,2007;Shutts,Condry,Santos,&Spelke,2009).
3 Du etoeq uipmen terror, videoda ta werelo stfromtwoinfantsintheprosocialandthreeinfantsintheantisocialcondition;

thisanalysisthereforeincludesdatafromtheremaining14and13infantsinthosetwoconditions,respectively.
234 J.K.Hamlin,K.Wynn/CognitiveDevelopment27 (2012) 227–239

Fig.2. Results:numberoftimes(offour)infantsatelikedversusdislikedfoodfollowingpreferencemodelingbyProsocial,

Novel,orAntisocialSource.

conditiondidso.Furtheranalysesrevealedthatinfantsweresignificantlymorelikelytomatchthe

preferenceoftheProsocialSourcethantheAntisocialSource,t(30)=3.52,p<.005;Fisher’sexacttest

p<.001.TheywerealsomorelikelytomatchthepreferenceoftheNovelSourcethantheAntisocial

Source,t(30)=3.29,p<.005,Fisher’sexacttest,p<.005,butequallylikelytomatchthepreferenceof

theProsocialandNovelSource,t(30)=.333,p=.71,Fisher’sexacttest,p=.99.

3. Discussion

Resultssuggestthatinfantsgaininformationfromcertainindividualsandignoreinformationfrom

others.Whenpresentedwiththefoodpreferencesofapuppetwhohadpreviouslybeenprosocial

towardathirdparty,infantschosethefoodthepuppethadexpressedlikingoverthefooditexpressed

dislikingfor.Incontrast,whenapreviouslyantisocialpuppetpresenteditspreferences,infantsdid

not appear to take this information into account in deciding what to eat, choosing both the liked

anddislikedfoodsequally.ComparingtheseresultswiththoseintheNovelSourceconditionsug-

geststhatthiseffectisdrivenbyarelativefailuretomatchthefoodpreferencesofantisocialothers:

infantswereequallylikelytomatchthefoodpreferencesofanunknownandaprosocialsource.Taken

together,theseresultssupporttheclaimthatone’sevaluationofapotentialsourceofinformationmay

contributetoselectivelearningininfancy.

Fig.3. Results:numberofinfantswhomatchedthesource’spreferencezero,one,two,threeorfouroffourtimesfollowing

preferencemodelingbyProsocial,Novel,orAntisocialSource.
J.K.Hamlin,K.Wynn/CognitiveDevelopment27 (2012) 227–239 235

Note,however,thatthecurrentstudydidnotdirectlytestwhetherinfantspositivelyevaluated

theprosocialcharacterandnegativelyevaluatedtheantisocialcharacter.Wechosenottoexamine

infants’ preferences in order to avoid influencing their subsequent interactions with the puppets.

Instead,infants’evaluationswereinferredfrompreviousstudies–allusingtheboxscenarioutilized

here–whichhavedemonstratedthat(1)infantsprefertheprosocialtotheantisocialcharacter(Hamlin

&Wynn,2011),(2)infantspreferthosewhohelptheprosocialcharacterandthosewhohinderthe

antisocialcharacter(Hamlinetal.,2011),and(3)toddlersthemselvesgivetreatstotheprosocialchar-

acterandtaketreatsawayfromtheantisocialcharacter(Hamlinetal.,2011).Together,theseresults

stronglysuggestsocialevaluationistakingplaceinthepresentstudy;however,weacknowledgethat

inferredevaluationisapotentiallimitationoftheresults.

Importantly,theobservedeffectscannotbeaccountedforbyattentionaldifferences:infantswere

aslikelytoattendtothepreferencesofProsocial,Antisocial,andNovelSources,andtheirattention

followingprosocialandantisocialeventsdidnotinfluencetheirfoodchoices.Itisadditionallydifficult

toexplainthispatternofresultsasstemmingfrompurelyassociativemechanisms.Becauseallinfants

sawbothprosocialandantisocialpuppetevents,wecanbesurethattheydidnot,forexample,simply

feellesssocialafterviewingantisocialeventsandsubsequentlyfailtotakeinanysource’sinformation.

Inaddition,becauseallsourcesexpressedbothpositiveandnegativefoodpreferences,infantscould

nothavesimplytaggedfooditemswithpositiveversusnegativeemotionalinformationwithoutalso

consideringthesourceofthatinformation,norcouldtheyhavemerelyrespondedtothepresenceof

positiveversusnegativesourceswithoutconsideringthespecificpreferenceinformation.Finally,asall

infantswereprovidedwithonlyonepotentialsourceofinformation,theycouldnothavebasedtheir

responsesonwhetheronesourcewasrelativelypreferabletoanothersource,butonlyonwhether

theonlysourceavailablehadinformationworthusing.

Thereareseveralwaystointerprettheseresults.Onepossibilityisthatinfantsconsideredasource’s

emotionalinformationtoreflectitsindividualpreference.Previousresearchhasshownthatby13

monthsofage,infantsexpectsomekindsofinformation(e.g.,language)tobesharedacrossindividuals

butappreciatethatotherkindsofinformation(e.g.,objectpreferences)maydifferfromoneindividual

toanother(Buresh&Wooward,2007;Repacholi&Gopnik,1997).Ifinfantsinterpretedthepuppet’s

expressions to reflect personal attitudes about the foods, our results may reflect infants’ desire to

affiliatewithand/or“belike”theNovelandProsocialSources,butnottheantisocialsource.Whilethe

puppetmodelitselfwasnotpresentduringinfants’eatingtest(andwasthereforeperhapsunavailable

foraffiliation),thepuppeteerexperimenterwaspresent,whichmayhaveexertedasocialinfluence

oninfants’choice.Researchwithchildrenandadultssuggeststhatimitativebehaviorisbothcause

andconsequenceofadrivetoaffiliate(Chartrand&Bargh,1999;Lakin&Chartrand,2003;Over&

Carpenter,2009;Over&Carpenter,2012;Uzgiris,1981).Ourresultssuggestthatthislinkbetween

imitation/emulationandlikingmaybepresentininfancy.

Alternatively,infantsmayhavetakenthesources’expressedpreferenceinformationtobeabout

theactualstateoftheworld(i.e.,thataparticularfoodisgoodandanotherfoodisbad).Ifso,infants’

selectivitymayreflectlowerlevelsoftrustintheaccuracyofinformationprovidedbyanAntisocial

Source,similarlyto howolderchildrenbelievethata previouslyinaccuratelabelermay also have

incorrectknowledgeofotherobjectnames(Birch&Bloom,2002;Sabbagh&Baldwin,2001).Onthis

explanation,wewouldexpectinfantstoavoidtheProsocialorNovelSource’sdislikedCheeriocolor

innewsituationsinwhichneitherpuppetnorpuppeteerispresent.

Finally,thisstudyislimitedinthatthepuppets’behaviorwastheonlyinformationinfantshad

abouttheirpotentialvalueasknowledgesources.Intherealworld,learnersoftenknowmuchmore

inadditiontoasource’simmediatebehavior,suchastheirpriorbehavioralhistory,skillsindifferent

areas,andgroupmemberships,Wemayalsoknowsomethingaboutthetargetofone’sbehaviors,

suchaswhetherthetargetsthemselvesaregoodorbad,orwhetherornottheyareinourgroup.

Infants’evaluationsofhelpfulandunhelpfulcharactersareknowntodifferdependingonthepast

actsofthetargetofthosecharacters’behaviors(Hamlinetal.,2011);perhapsinfants’judgmentsof

theAntisocialSourcemighthavedifferedhadtheyknownmoreaboutthesourceitself(forinstance,if

itwereanin-groupmember)ormoreaboutthetargetoftheAntisocialSource’sbehavior(forinstance,

ifthetargetwerebadoranout-groupmember).Furtherstudyshouldattempttoteaseapartthese

possibilitiestoshedlightontheroleofaffiliationandlikinginsociallearning.
236 J.K.Hamlin,K.Wynn/CognitiveDevelopment27 (2012) 227–239

Interestingly,infantsinthepresentstudieswereequallylikelytomatchthefoodpreferencesof

theProsocialandNovelSources,relativetotheirreluctancetomatchtheAntisocialSource,whom

theyneithermatchednormismatched.Whileourresultsmayrepresentaceilingeffectonpreference

matchingoftheprosocialandneutralsources,andfurtherstudycouldshowthatprosocialsources

areprivilegedoverneutralones,infantsinboththeProsocialandNovelSourceconditionsdidsome-

timeseatthefoodstheirsourcehaddisliked(onaverageonceoffourtimes),makingthispossibility

somewhatlessplausible.Ourresultssuggestthatthebaselineresponsemaybetoacceptothers’infor-

mationunlesstheypresentsomereasonnotto(inthiscase,beingantisocial).Thisresponsepattern

wouldbeadaptive(morelearningmaybebetter)andisreminiscentofthenegativitybias,inwhich

negative(relativetopositiveorneutral)individualsandactionsmaybeparticularlysalient,influen-

tial,andmemorabletoadultandchildobservers(Abelson&Kanouse,1966;Aloise,1993;Kanouse

&Hanson,1972;Knobe,2003a;Leslie,Knobe,&Cohen,2006).Theseresultssupportclaimsthatthe

negativitybiasisearly-developing(Hamlinetal.,2010;Vaish,Grossman,&Woodward,2008),andis

alsohighlyconsistentwithrecentreportsofa“pitchforkeffect”inpreschoolers’selectivelearning:

Beingpreviouslyinaccuratehasarelativelygreaterinfluenceonchildren’stendencytogaininforma-

tionfromasourcethandoesbeingpreviouslyaccurate(Corriveau,Meints,&Harris,2009;Koenig&

Jaswal,2011).

Overall,theseresultsindicatethatinfantsareselectivenotonlyinwhatkindsofinformationthey

gainfromsocialsources,butalsoregardingwhichsourcestheyreceiveinformationfrom.Whilewe

havenotyetaccountedforthespecificmechanismleadingtoinfants’responses(accuracy,familiar-

ity/similarity,liking),ourfindingsaddtothesmallbutexpandingliteratureonselectivesourcelearning

andimitationininfancy(Chowetal.,2008;Koenig&Woodward,2010;Nielsen,2006;Nielsenetal.,

2008;Ryallsetal.,2000;Shutts,Condry,etal.,2009;Shutts,Kinzler,etal.,2009;Zarbatany&Lamb,

1985;Zmyjetal.,2010),aswellastorecentevidencethatinfantsevaluateothersbasedontheirpro-

andanti-socialbehaviors(Dunfield&Kuhlmeier,2010;Geraci&Surian,2011;Hamlin&Wynn,2011;

Hamlinetal.,2007;Hamlinetal.,2010;Schmidt&Sommerville,2011),andcallsforfurthereluci-

dation.Ifsimplyseeinganindividualbehaveantisociallytowardanunknownthird-partydecreases

infants’tendencynotonlytointeractwiththatindividual,butalsotousethatindividualasasource

ofnewinformation,earlydevelopingmechanismsforsocialevaluationmay,overtime,havelasting

effectsondevelopment.

Acknowledgements
WethankNehaMahajan,AnnieSpokes,andothermembersoftheInfantCognitionCenteratYale

University,andtheparentsandinfantswhoparticipated.

References
Abelson,R.P.,&Kanouse,D.M.(1966).Subjectiveacceptanceofverbgeneralizations.InS.Feldman(Ed.),Cognitiveconsistency
(pp.1 7 3–1 9 9).SanDi ego ,C A:Acad emicPress .
Aloise,P .A.(1993). Trai tconfir mat ionanddi sconfirmation:Thedevelopmentofattributionbiases.JournalofExperimentalChild
Psy ch olo gy,55, 177– 193.
Bandura,A.(19 77) .Sociallearningtheory.NewYork:GeneralLearningPress.
Barr,R.,D ow den,A .,&Ha yne,H. (1996). Deve lopme ntalcha ngesind eferredimitationby6-to24-month-oldinfants.Infant
B eha viorandD ev elo pment, 19 ,159–17 1.
Bellagamba, F.,& Tomasello,M .( 1999).Re-enactingintendedacts:Comparing12-and18-month-olds.InfantBehaviorand
Developm en t, 22,277–282 .
Birch,L.L.(1980). Effe ctsofpeermodels’foodchoicesandeatingbehaviorsonpreschoolers’foodpreferences.ChildDevelopment,
51 ,4 8 9–496.
Birch,S .,Vauthier,S.,&Bloom,P.(2008).Three-andfour-year-oldsspontaneouslyuseothers’pastperformancetoguidetheir
lea rn ing.Cogni tio n, 107(3), 10 18–103 4.
Birch,S.A.J. ,&Bloom, P.(200 2).Preschoolersaresensitivetothespeaker’sknowledgewhenlearningpropernames.Child
De ve lop m en t,73,43 4– 444.
Bradbard,M.R.,& E ndsley,R.C.(1983).Theeffectsofsex-typedlabelingonpreschoolchildren’sinformation-seekingand
retent ion .Se xR oles,9,24 7– 26 0.
Bradbard,M.R .,M artin, C. L.,Endsley,R.C.,&Halverson,C.F.(1986).Influenceofsexstereotypesonchildren’sexplorationand
memo ry: A compete nc e versuspe rfo rm a ncedistinc tio n. Develop mentalPs ych olo gy,22,481–4 86 .