FACILITATINGANALOGICALTRANSFER INAHYPOTHETICO-DEDUCTIVEREASONINGTASK: THECONTRIBUTIONSOFSURFACEANDSTRUCTURALSIMILARITY I CYNTHIASUMMERSKOENIG ADISSERTATIONPRESENTEDTOTHEGRADUATESCHOOL OFTHEUNIVERSITYOFFLORIDAINPARTIALFULFILLMENT OFTHEREQUIREMENTSFORTHEDEGREEOF DOCTOROFPHILOSOPHY UNIVERSITYOFFLORIDA 2002 ACKNOWLEDGMENTS Thisjourneythroughgraduateschoolhasbeenlongandarduous,yetextremely fulfilling.Throughouttheexperience,therehavebeenmanypeoplewhosecontinued supportwasthesourceofmystrengthandendurance.Iwouldnowliketoacknowledge theircontributions. First,andmostimportantofall,Iwouldliketothankthethreepeoplewhohave madethemostsacrificesinhelpingmetoachievemydreams-mychildren,Russand Emily,andmyhusband,Dan.Theywillinglyrelocatedtoanewcity,gaveupthe "luxury"ofregularlyhome-cookedmeals,andtoleratedlessfamilytimesothatIcould workonmyeducation.Inparticular,Danspentuntoldhours/days/weeksofpersonaltime commutingtowork-achoreheseemswillingtocontinueasIbeginanewacademic career.Ithasbeenablessinginmylifetohavethreepeoplewhoseunconditional support,comicrelief,andsincereencouragementhavetrulybeenthe"windbeneathmy wings." Myextendedfamilyhasalsobeenanimportantresourcethroughoutmypursuits. Iappreciatemysiblingsforcontinuallyaskingaboutmyprogressandencouragingmeto keepworkinghard:Mark,Lucy,Greg,Kathleen,Chip,Amy,Dave,Cindy,and Lawrence.Ithankmyparents,DonandSaraSummers,forinstillinginmeaworkethic thathasbeeninstrumentaltomyaccomplishmentsingraduateschool.Finally,Ithankmy ii in-laws,HarryandBettyKoenig,notonlyforbeingsupportiveovertheyearsbut especiallyforraisingsuchafinesonwhohasbeenanessentialelementinmysuccess. IappreciatethecontributionsofthefacultymembersattheUniversityofFlorida whoprovidedadvice,expertise,andsupportduringthedevelopment,datacollection,and writingofthisdissertation:myfacultyadvisorandcommitteechair.Dr.RichardGriggs, andfourcommitteemembers.Dr.fraFischler,Dr.ShariEllis,Dr.JeffFarrar,andDr. JamesAlgina. Ithasbeenaprivilegetoworkwithindividualswhoseuniquetalentshave contributedsignificantlytomyprofessionaldevelopmentduringgraduateschool. Havingfellowgraduatestudentsandfacultywhoprovideencouragement, fiiendship,andstressreliefenhancessuccessingraduateschool.Ihavebeenfortunateto haveanumberofstudentsandfacultymemberswhoprovidedmentoring,advice,and emotionalsupportthroughoutgraduateschool.IthankLisaMaagforhercomicrelief, DuanaWelchformentoringmethroughmyfirstyearofgraduateschool,Jennifer Blessingforherstatisticalwizardry.Dr.ShariEllisforbeingbothagreatfiiendanda rolemodel,andDr.JeffFarrarforhisnever-endingsupport. Finally,IwouldliketoexpressmysincereappreciationtoDr.RichardGriggs.At atimewhenmygraduatecareerwaslookingratherbleak,hetookmeunderhiswingand helpedto"polish"myedgessothatImightbecomeabetterresearcher,teacher,and colleague.Ithasbeenaprivilegetoworkwithsomeonewhonotonlysharesasimilar teachingphilosophyandworkethicbutwhohasbeentirelessinhissupportofmy academicpursuits.Iwillbeeternallygratefulforhismanycontributionstobothmy graduatecareerattheUniversityofFloridaandmyupcomingacademiccareeratSt. Mary'sCollegeofMaryland. iii ;^^^ TABLEOFCONTENTS \ ' ' ^ , ~ page .. ACKNOWLEDGMENTS ii ABSTRACT vi CHAPTER 1 INTRODUCTION 1 TheoreticalExplanationsofAnalogicalTransfer 4 TheTHOGTask 9 TheoriesofReasoningandtheTHOGTask 10 FacilitationontheTHOGTask 13 AnalogicalTransferandtheTHOGTask 18 Summary 24 2 OVERVIEWOFDISSERTATIONEXPERIMENTS 26 3 DESIGNANDMETHODOLOGYOVERVIEW 30 Participants 30 Procedure 30 Materials 30 SimilarityQuestionnaire 31 ErrorPatterns 31 4 EXPERIMENT1-TESTINGFORTRANSFER 33 Introduction 33 DesignandMethodology 33 Results 35 Experiment1Conclusions 40 5 EXPERIMENT2-CHANGINGTHEVALUESOFTHEDIMENSIONS 43 Introduction 43 DesignandMethodology 43 Results 45 iv Experiment2Conclusions 51 6 EXPERIMENT3-CHANGINGTHEDIMENSIONS 54 Introduction 54 DesignandMethodology 55 Results 57 Experiment3Conclusions 60 7 EXPERIMENT4-EXAMININGSTRUCTURALFEATURES 63 Introduction 63 DesignandMethodology 64 Results 65 Experiment4Conclusions 70 .1] 8 SUMMARY 72 SurfaceandStructuralSimilarity 72 SeparationandHypothesisGeneration 73 FacilitatingAnalogicalTransfer 74 ReconsideringtheTHOGTask 76 REFERENCES 80 APPENDIX A STANDARDTHOGPROBLEM 84 B VARIANTSOFTHETHOGPROBLEM 85 C ADDITIONALEXPERIMENTALMATERIALS 92 D ERRORPATTERNS 96 BIOGRAPHICALSKETCH 99 V AbstractofDissertationPresentedtotheGraduateSchool oftheUniversityofFloridainPartialFulfillmentofthe RequirementsfortheDegreeofDoctorofPhilosophy FACILITATINGANALOGICALTRANSFER INAHYPOTHETICO-DEDUCTFVEREASONINGTASK: THECONTRIBUTIONSOFSURFACEANDSTRUCTURALSIMILARITY CynthiaSummersKoenig , August2002 , Chair:RichardA.Griggs MajorDepartment:Psychology TheBlackboardandPythagorasversionsoftheTHOGproblemwereutilizedto investigatehowsurfaceandstructuralsimilaritycontributetoanalogicaltransferinthis hypothetico-deductivereasoningtask.Inexperimentalgroupsof12orfewer,466 UniversityofFloridaintroductorypsychologystudentsseparatelysolvedsourceand targetproblems,andthendescribedanyproblemsimilaritiesandtheirfacilitatingeffects. InExperiment1,significanttransferwasobservedforbothTHOGproblemswhen surfacesimilaritybetweensourceandtargetproblemswasmaintained.InExperiment2, changingthedimensionalvaluesofthegeometricfiguresinthetargetproblemdecreased surfacesimilarity.ThePythagorasproducedanalogicaltransferwhereastheBlackboard didnotandwasdroppedfromfurtherexperimentalinvestigation.InExperiment3,the targetproblem'sgeometricdesignswerereplacedwiththedimensionsoflettersand vi numbers,andthePythagorasagainyieldedtransfer.InExperiment4,threeversionsof thePythagorasproblemwerecreated,andahypothesisgenerationrequestincombination witheitherfigurallabelingornarrativestructureproducedanalogicaltransfer. Theseexperimentsrevealedthatparticipantscouldsolvethedifficultabstract THOGtask,intheabsenceofsurfacesimilarity,iftheyabstracteditssolutionprinciple fromasourceproblem.Whensurfacesimilaritywasreduced,itwasnecessaryforthe sourceproblemtoentailfeaturesthatseparatedthevaluesoftheexemplarfromthe possiblewrittendowncombinationsandrequestedparticipantstogeneratehypothesesfor thesecombinations.Thesefindingscontributetoourunderstandingofbothhypothetico- deductivereasoningandanalogicaltransfer.First,confiisiontheorywassupportedbythe findingthatseparationandreducedworkingmemoryloadyieldedbothfacilitationand fransfer.Second,indirectexperimentalhelp(i.e.,ahypothesisgenerationrequest)was necessarytoobtaintransfer.Finally,theprimacyofstructuralsimilaritytoanalogical transferwasinterpretedassupportingpragmaticschematheoryandstructuremapping theory.Theseresultssuggestthemajorityofadultscanreasonhypotheticallyand deductivelyontheTHOGtaskiftheyareledtoabstractitsunderlyingsolutionprinciple. ' - . * ' ' vii CHAPTER 1 INTRODUCTION Thedoctrineof"formaldiscipline,"datingbacktotheteachingsofPlatoand presentedinitsclassicformbyJohnLocke(Dewey,1916),inmanyrespectsliesatthe heartofspontaneousanalogicaltransfer.Accordingtoformaldiscipline,themindislike a"muscle"thatmustbetrainedincertainfaculties,suchasobservation,attention,and reasoning,inordertostrengthentheseprocesses(Singley&Anderson,1989).Locke believedthecontentstudiedinorderto"exercise"thesefacultiesv^^assomewhat irrelevant.Rather,heproposeditwastheamountofexertionrequiredtomasteragiven abilitythatwasmostimportant(Dewey,1916).Theprimarypurposeofformaleducation then,asproposedbyformaldiscipline,shouldbetofocusonthosecontentareaslikelyto trainspecificcognitiveprocessesthatcanthenbeapplied(transferred)acrossmany disciplines. ThefirsttoempiricallytestthetenetsofformaldisciplinewasE.L.Thomdike (1906,1913).Inaresearchendeavorspanningoveraquartercenturyusingavarietyof tasks,Thomdikedeterminedthatthetypeofspontaneoustransferproposedtoresultfrom formaldisciplinetrainingwasentirelytoowideinscope.Hesuggestedthatratherthan entailingparticularfaculties,themindiscomposedofspecificassociationsandhabits thatnarrowdramaticallythetypesofresponsesthatcanbeexpectedinanygiven situation.Asaconsequence,transferbetweenactivitiesshouldonlybeexpectedifthere isahighdegreeofsimilarity(i.e.,sharedcommonstimulus-responseelements). 1 2 Thomdike'stheoryoftransfer,calledthetheoryofidenticalelements,proposedthat: Onementalfunctionoractivityimprovesothersinsofarasandbecausetheyare inpartidenticalwithit,becauseitcontainselementscommontothem.Addition improvesmultiplicationbecausemultiplicationislargelyaddition;knowledgeof LatingivesincreasedabilitytolearnFrenchbecausemanyofthefactslearnedin theonecaseareneededintheother...[T]hemindissospecializedintoa multitudeofindependentcapacitiesthatwealterhumannatureonlyinsmall spots,andanyspecialschooltraininghasamuchnarrowerinfluenceuponthe mindasawholethanhascommonlybeensupposed.(Thomdike,1906,pp.243- 246) Thus,althoughThomdikerejectedthestrongtheoryoftransferproposedbyformal discipline,hedidacceptaweakerversionsuggestingthattraininginagiven disciplineisonlylikelytotransfertoothersituationsifthereareidentical"elements" thatcanbemappedontoeachother. Theabilitytotransferknowledgefromoneexperiencetoanotherhasbroad implicationsacrossavarietyoftasksinaneducationalsetting(e.g.,working mathematicsproblems,learningforeignlanguages,andapplyingtherulesof science).Transferalsoplaysanimportantroleineverydayreasoninganddecision- makingbecausepreviousexperiencescanbeutilizedtoelucidateandbenefit subsequentsimilarexperiences.Perhapsbecausetheeducationenvironmentisonein whichsuccessful/unsuccessfiiltransferismoreeasilymeasuredquantifiably,much oftheresearchonanalogicaltransferhasfocusedonproblemsolvingtasks. Inthetypicaltransfertask,problemsolversarepresentedwitheitherasolvedor to-be-solvedproblem(thesource)thatisexpectedtofacilitateperformancewhenan analogicallysimilarunsolvedproblem(thetarget)islaterpresented.Thereisconsensus amonganumberofresearchers(e.g.,Holyoak&Koh,1987;Mandler&Orlich,1993; Needham&Begg,1991;Novick,1995;Reeves&Weisberg,1994)thatinorderfor 3 successfulanalogicaltransfertooccur,theproblemsolvermustbeableto(a)constructa mentalrepresentationofthesourceandtargetproblems,(b)selectthesourceproblemas arelevantanalogforsolvingthetargetproblem,(c)maptherelevantcomponents betweenthesourceandtarget,and(d)extendthismappingtogenerateasolutiontothe targetproblem. • Inordertomentallyrepresentthesourceandtargetproblems,theproblemsolver needstoinitiallyencodevariousproblemcomponents,eachofwhichmayhavevarying degreesofinfluenceonthetransferprocess.Someresearchers(e.g.,Gick&Holyoak, 1980,1983;Centner,1983)arguethatmostrelevanttosuccessfultransferisabstracting thedeepstructureofaproblem,whichmaintainsits"essence"irrespectiveofitscontent. Thisissometimescalledtheproblem's"solutionprinciple"andcanbeaformula(asin algebra,physics,orlogic),aproposition(e.g.,"Alargeforcemustbebrokenintosmaller elementsandsummateonatargetinordertoachieveagoal"),orarule(e.g.,"Afigure willbeclassifiedasaTHOGif,andonlyif,itcontainseitherthecolororshapeIhave writtendown,butnotboth")(Reeves&Weisberg,1994).Infact,ifthesolutionprinciple isnotencoded,transferisnotexpectedtooccur. Thesurfacecontentsofthesourceandtargetproblemsshouldalsobeencodedin suchamaimerthattherelationsbetweenthemareeitherirrelevanttoorinstantiatethe solutionprinciple.Forexample,intheTHOGruleexampleabove,thespecific dimensionsofshapeandcolorareessentiallyirrelevanttosolvingthetask,whereasthe relationshipsamongthedimensionalvalueswrittendown,theTHOGrule,andagiven exampleofaTHOGcomprisetheessenceofthesolutionprincipleandthusneedtobe encodedtofacilitatetransfer.