Keepon interactionw/normallydevelopingchildren• 0–1-year-olds:Interactionwasdominatedbytactileexplorationusingthehandsand/ormouth.Thechildrendidnotpaymuchattentiontotheattentiveexpressionsoftherobot,buttheyexhibitedpositiveresponses(suchaslaughingorbobbingtheirbodies)toitsemotiveexpressions.• 1–2-year-olds:Thechildrendemonstratednotonlytactileexplorationbutalsoawarenessoftherobot’sattentivestate,sometimesfollowingitsattention.Somemimickeditsemotiveexpressionsbyrockingandbobbingtheirownbodies.• 2+-year-olds:Thesechildrenfirstcarefullyobservedtherobot’sbehaviorandhowcaregiversinteractedwithit.Soontheyinitiatedsocialexplorationbyshowingittoys,soothingit(bystrokingitshead),orverballyinteractingwithit(suchasaskingquestions).
Preschoolers
• Studywith27normallydeveloping3-4yearoldsovermultipleweeks
• Violentvs.protectivebehavior:InS3,aboyTM(hereafterTM/m)beatKeepon severaltimes,andagirlSR(hereafterSR/f)stoppedhim,“No!No!”InS9,whenNR/mhitKeepon’s headseveraltimes,HN/fstoppedhimbysaying,“Ithurts!Ithurts!”InS13,FS/mandTA/mstronglyhitKeepon’s headacoupleoftimes,asifdemonstratingtheirbravenesstoeachother.YT/fandIR/f,observingthis,approachedKeepon andcheckedifithadbeeninjured,thenYT/fsaidtoKeeponandIR/f,“Boysareallalike.TheyallhitKeepon,”strokingitsheadgently.
• Demonstrativebehavior:InS6,KT/fplayedwithKeepon intheoutdoorplayground;aboyinthe4-year-oldclassapproachedKeeponandsaidtoKT/f(referringtoKeepon),“Thisisacamera.Thisisamachine,”butKT/finsisted,“No,theseareKeepon’s eyes!”• InS8,pointingtoaninsectcage,SR/fguidedKeepon’s attentiontoit.DuringreadingtimeinS11,NK/fandTM/mapproachedandshowedtheirpicturebookstoKeepon.• InS17,YT/ftaughtKeepon somewords—showingitthecap,shesaid,“say,Bo-shi,”thenswitchedtoKeepon’s knittedcapandsaid,“ThisisaknitBo-shi,thatyouwearinwinter”(towhichKeepon couldonlyrespondbybobbing).• InS25,NK/fgaveatoysledtoKeepon.Keepon showedapreferencetoanothertoyNK/fwasholding.Aftersomenegotiation,NK/fbroughtoveranothersledandpersuadedKeepon,“Nowyouhavethesamethingasme.”
• Self-consciousbehavior:InS22,afterallthechildrenpracticedasongwiththeteachers,severalofthemrantoKeepon andaskedonebyone,“Wasitgood?”,towhichKeepon respondedbynoddingandbobbingtogivepraise.InS24,NZ/msangasongquietlyforawhile;whenhenoticedKeepon besidehim,hebecamesurprisedandashamed.
Longitudinalautismstudy
• 1.5yearstudy,>500childinteractionsessions• 2-4yearoldchildrenw/autism,Down’ssymdrome,Asperger’sandsimilarconditions• teleoperated Keepon robot
MostcommonchallengesassociatedwithAutism• Socialinteraction:Difficultyinunderstandingothers‘intentionandemotion,andinsharinginterestsandactivitieswithothers;difficultyinformingsocial,cooperativerelationships.• Communication:Difficultyinverbalandnon-verbalcommunication,especiallyofeverydaypragmaticuse;delayorlackoflanguagedevelopment;stereotypedandrepetitivespeech.• Imagination:Stereotypedandrestrictedpatternofinterestandbehavior;adherencetospecificthingsandaimlessroutines;difficultyincopingwithnovelsituations(e.g.,unknownplacesoractivities).
ParticipantM(15sessions,3months)
• MshowedstronginterestfromSession1(hereafterS1),butdidnotgetclosetoKeepon.ThroughS1toS7,MavoidedbeinglookedstraightatbyKeepon (i.e.,aversiontoeye-contact);howeverMoftenlookedintoKeepon's prfiole.
ParticipantM(15sessions,3months)
• InS5,afterobservinganotherchildRputapapercylinderonKeepon's head,Mdraggedthenurse'sarm,askinghertodothesamethingtoKeepon.Whenthenursecompletedthetask,MlookedsatisfiedandleftfromKeepon.ThroughS5toS10,thedistancetoKeepon graduallygotshorter(to40.50cm).
ParticipantM(15sessions,3months)
• InS11,MtouchedtheheadofKeepon withaxylophonestick,thenMdirectlytouchedwithherhandasifMhadexaminedthetextureandsmellofKeepon.
• Afterthisfirsttouch,Mstartedsocialinteractionsincludingeye-contactandvocalization(S12.),puttingacaponKeepon,askinghermothertodothesame(S13.),andhuggingandkissingKeepon(S14).
ParticipantN(39sessions,18months)
• InS1,NgazedatKeepon foralongtime.AfterobservinganotherchildWplayedwithKeepon usingatoy,NwasencouragedtoplaywithKeepon usingthesametoy,butNdidnotshowinterestindoingthat.
• ThroughS2toS14,NdidnotpayattentiontoKeepon,evenwhenshewasnexttoit.However,NoftenglancedatKeepon,whenshenoticeditssound,suchas“Pongpong pong”.
ParticipantN(39sessions,18months)
• InS15,afterobservinganotherchildRputacaponKeepon's head,NtouchedKeepon withherfinger.• InS16(after3-monthbreakfromS15),NcameclosetoKeepon andlookedintoitsmovement.Inthesnacktime,NcametoKeepon andpokeditsnose,towhichKeepon respondedbybobbing,andNshowedsurpriseandsmile.Themothersandnursesburstintolaughter.Duringthisplay,Noftenmadereferentiallookswithsmiletohermotherandthenurse.
ParticipantN(39sessions,18months)
• FromS17,NoftensatinfrontofKeepon withhermother;sometimesshetouchedKeepon toderivesomeresponse.FromS20,NstartedexploringKeepon's abilitybywalkingaroundittoseeifitcouldfollowher.
ParticipantN(39sessions,18months)
• InthesnacktimeofS33,NcametoKeepon andstartedimitationplay,whenNperformedonemovement(bobbing,rocking,orbowing),soonKeeponmimicked;thenNdidanother,andKeepon didthesame.ThroughS33toS39,Noftenplayedthis“imitationgame”withKeepon,duringwhichsheoftenmadereferentiallookstohermotherandthenurse.
Dottedlinerepresentsfirsttouch.
Conclusions(frompaper)
• evenautisticchildrenpossessthemotivationforsharingandexchangingmentalstateswithothers,andthatthechallengefortherapistsandparentsistoelicitthismotivation.
• Althoughitiswidelybelievedthatthismotivationisimpairedinautism,wehaveobservedinanumberofcasesthatautisticchildrenestablishedsocialrelationshipswiththesimplerobot,whichwascarefullydesignedtoexpressitsmentalstatescomprehensibly.
Conclusions(frompaper)
• Whilewehavepresentedthreecasesthatarerepresentativeofthesuccessfulelicitationofsocialbehaviorfromchildrenwithautism,itshouldbenotedthattheseresultsarenottobeconsidered“generalizable.”Rather,theyaremeanttoillustratethepotentialforanappropriatelydesignedrobottoevokerarebutpositiveresponses.
• Therobotthusservesasatoolduringthetherapeuticsessionsbyenablingtherapiststoconductanovelformofmediatedinteractionwiththechildrenaswellasafterthesessionsbyprovidingarecordedbodyofdatathatcanbeusedbyparents,therapists,andresearchersinstudyingautismandinevaluatingortailoringindividualchildren’stherapeutictreatments.
Althoughalotofworkwentintothisstudy,anditprovidesaninterestingperspective,noneofthesefindingsareacceptable
byclinicalstandards.
SingleSubjectStudyDesign
• usefulwhentheresearcherisattemptingtochangethebehaviorofanindividualorasmallgroupofindividuals• theparticipantservesasboththecontrolandtreatmentgroup• onlyonevariableischangedatatime• Singlesubjectresearchdesignsare“weak”whenitcomestoexternalvalidity
Example– EffectofPraise
• investigatetheeffectofpraiseonreducingdisruptivebehaviorovermanydays.• First,establishabaselineofhowfrequentlythedisruptionsoccurred(measurehowmanydisruptionsoccurredeachdayforseveraldays)
Example– EffectofPraise
• baselineofbehaviorhasbeenestablishedonceaconsistentpatternemergeswithatleastthreedatapoints• interventioncanbegin-- researchercontinuestoplotthefrequencyofbehaviorwhileimplementingtheinterventionofpraise.
GraphLabels
A-BDesign
A-B-ADesign
• involvesdiscontinuingtheinterventionandreturningtoanontreatment condition.
B-A-BDesign
• Usedwhenanindividual’sbehaviorissoseverethattheresearchercannotwaittoestablishabaselineandmustbeginwithanintervention.
MultipleBaselineDesign
• Usedtoaddressseveralissuesforonestudentorasingleissueforseveralstudents.• Interventionintroducedatdifferenttimestoshowmoreclearlythateffectscanmorelikelybecreditedtotheinterventionitselfasopposedtoothervariables
MeasuringtheEfficacyofRobotsinAutismTherapy:HowInformativeareStandardHRIMetrics?[Begumetal2015]
• Single-subjectdesignexperimentinaclinicalsettingwith3individualswithsevereautism• Oneofthevery fewroboticsstudieswhichshowstransferofaskilllearnedwithassistanceofarobottohuman-humaninteraction
Metrics
EfficacyMetrics:skillexecutionandpromptdependencyHRIMetrics:Gaze,Communication,Affect
Findings
• HRImetricspredict‘