Вы находитесь на странице: 1из 264

Cover

Pagei

RewardsandIntrinsicMotivation
Pageii

Thispageintentionallyleftblank.
Pageiii

REWARDSANDINTRINSICMOTIVATION
ResolvingtheControversy

JudyCameronandW.DavidPierce


Bergin&Garvey
Westport,ConnecticutLondon
Pageiv

LibraryofCongressCataloginginPublicationData
Cameron,Judy,1953
Rewardsandintrinsicmotivation:resolvingthecontroversy/JudyCameronandW.David
Pierce.
p.cm.
Includesbibliographicalreferencesandindex.
ISBN0897896777(alk.paper)
1.Employeemotivation.2.Achievementmotivation.3.Organizationalbehavior.4.Psychology,
Industrial.5.Interpersonalrelations.I.Pierce,W.David.II.Title.
HF5549.5.M63C352002
158.7dc212001043794
BritishLibraryCataloguinginPublicationDataisavailable.
Copyright2002byJudyCameronandW.DavidPierce
Allrightsreserved.Noportionofthisbookmaybe
reproduced,byanyprocessortechnique,without
theexpresswrittenconsentofthepublisher.
LibraryofCongressCatalogCardNumber:2001043794

ISBN:0897896777

Firstpublishedin2002
Bergin&Garvey,88PostRoadWest,Westport,CT06881

AnimprintofGreenwoodPublishingGroup,Inc.
www.greenwood.com

PrintedintheUnitedStatesofAmerica

Thepaperusedinthisbookcomplieswiththe
PermanentPaperStandardissuedbytheNational
InformationStandardsOrganization(Z39.481984).
10987654321
Acknowledgment
OurresearchwassupportedbytheSocialScienceandHumanitiesCouncil(SSHRC)ofCanada.
Pagev

Contents
PartI:
Introduction

1 AnIntroductiontotheRewardsandIntrinsicMotivationControversy 3

PartII:
RewardsandIntrinsicMotivation:ALookattheEarlyResearch

2 HowRewardsGotaBadReputation 11
3 WhyRewardsDoNotDeserveaBadReputation 23

PartIII:
TheoreticalDisputesOverRewardsandIntrinsicMotivation

4 TheoreticalPerspectivesonRewardsasHarmful 37
5 TheoreticalPerspectivesonRewardsasHelpful 59

PartIV:
TheEmpiricalEvidencefortheImpactofRewardsonIntrinsicMotivation

6 AnOverviewoftheExperimentsonRewardsandIntrinsicMotivation 87
7 ACritiqueofMetaanalysesoftheEffectsofRewardsonIntrinsicMotivation 101
Pagevi

8 AMetaanalysisoftheEffectsofRewardsonIntrinsicMotivation 115
9 DiscussionandImplicationsofOurMetaanalyticFindings 155

PartV:
RewardsandIntrinsicMotivation:ASociohistoricalPerspective

10 ASociohistoricalAnalysisoftheLiteratureonRewardsandIntrinsicMotivation 177

PartVI:
PracticalApplicationsofRewards

11 TheEffectiveUseofRewardsinEverydayLife 201

PartVII:
Conclusion

12 ResolvingtheControversyoverRewardsandIntrinsicMotivation 231
References 235
Index 251
Page1

PARTI
INTRODUCTION
Page2

Thispageintentionallyleftblank.
Page3

Chapter1
AnIntroductiontotheRewardsandIntrinsicMotivationControversy
Mostparentsandeducatorswouldagreethattheidealstudentisonewhoperformsacademictasksatahighlevel,showshighinterestandinvolvementinschool
activities,iswillingtotakeonchallengingassignments,andisanoverallselfmotivatedlearner.Mostofuswouldagreethatinanidealworkplace,peopleare
productive,happy,andproudoftheiraccomplishments.Moregenerally,mostofuswanttoliveinasocietywherepeoplecontributetothewelfareofothers,arefree
tochooseamongavarietyofactivities,haveasenseofcompetenceandachievement,andfeelcontentmentinmostaspectsoftheirlife.Weagreewiththesevalues.
However,howtoattaintheseidealshasarousedfarlessconsensus.
Onewayinwhichsocietyattemptstoachieveitsgoalsistopunishundesirablebehavior.Teachersgivedetentionsforincompleteschoolassignmentsinthehopethat
studentswilldevelopasenseofresponsibilityfortheirwork.Parentscriticizetheirchildrensworkhabitstoinstillstandardsofhighachievement.Employersthreaten
lossofjobstomotivateworkerstobemoreproductive.Thelegalsystempunishesthosewhobreakthelawsinordertoteachresponsibility,maintainorder,and
protectthefreedomofothers.
Systemsofpunishmenthavebeendesignedandmaintainedaswaystoachievetheaimsofhighproductivity,satisfactionwithachievement,caringforothers,andso
on.Thefactis,however,thatpunitivesystemshaveneverattainedthesegoals.Instead,theuseofpunishmenthasresultedinresentment,discontent,feelingsoflow
selfworth,andattemptstooverthrowthosewhowieldpower.Inotherwords,punishmentbackfires.1
Page4

PUNISHMENTSYSTEMS
MEANS ENDS/GOALS EFFECTS
Detention Senseofresponsibility Resentment
Criticism Highstandardsofachievement Discontent
LossofJob Productivity Lowselfworth
Legalpenalties Responsibility,order,andfreedom Attackorsubvertthoseinpower

Punishmentisamajorsourceofexternalcontroloverhumanbehavior.Thefactthatpunitivesystemsresultinseriousnegativeeffectshasledsomesocialadvocatesto
condemntheuseofanyformofexternalinfluence,includingreward.Forexample,aprominentspokespersonineducationandbusiness,AlfieKohn,rejectsboth
punishmentandrewardasmeansofsocialinfluence.2Kohnspremiseisthatbecausepunishmentinvolvesexternalcontrolandhasmanynegativedetrimental
consequences,positiveexternalinfluencebasedontheuseofrewardswillalsoleadtopervasivenegativeeffects.Kohnechoesthesentimentsofmanyresearchersand
writersinsocialpsychology,education,business,andthemedia.

THECLAIMTHATREWARDSAREHARMFULISBASEDONTHEFOLLOWINGLOGIC:
Punishmentproducesnegativeeffects.
Punishmentisanexternalinfluence.
Rewardisanexternalinfluence.
Therefore,rewardsmustproducenegativeeffects.

Intheearly1970s,EdwardDeci,asocialpsychologistandresearcher,claimedthatexternalrewardsshouldnotbeusedbecausetheytakeawaypeoplesinterestand
enjoymentofactivities.Decisetouttoshowscientificallythatrewardsandpositivereinforcementcouldharmpeoplesintrinsicmotivation.3FromthetimeofDecis
initialstudy,literallyhundredsofexperimentshavebeencarriedouttoinvestigatethenegativeeffectsofreward.Thegeneralconclusionfromthisliteratureisthat
rewardsystemsandpositivereinforcementhaveextensivenegativeeffects.Thisconclusionhasfilteredintothepopularmediaandiscontinuallyreiteratedin
introductorytextbooksonpsychology,education,andbusinessmanagement.
Althoughtheclaimthatrewards(andpositivereinforcement)havegeneralnegativeeffectsispopular,itisalsowrong.Itiswrongbecauseitisbasedonanide
Page5

alisticandfaultyviewofhumannature,becausetheevidencedoesnotsupporttheclaim,becauseofseriousoverstatementanddistortionoftheresearchfindings,and
becauseofconceptualconfusion,faultyreasoning,andmisunderstandings.Finally,itiswrongbecausethereisnoacknowledgmentofthevastliteratureshowingthat
rewardscanbeusedeffectivelywithoutnegativebyproducts.4
Theclaimthatrewardshavegeneralnegativeeffectsisnotonlywrongitisharmful.Itisharmfulbecauseteachersandeducatorswhoacceptthisviewresistusing
rewardswhenthereisnoneedtodoso.Studentsenjoyschoolandworkhardwhenapositivereinforcementsystemisproperlyarranged.Employeesstrivetoattain
highstandardsintheworkplacewhentheyarerewardedfortheiraccomplishments.Thesepositivemeansofinfluencearebeingdeniedtoindividualsbasedonthe
claimthatrewardsystemsaredetrimental.Theviewthatrewardsareharmfulispreventingusfromdesigningbetterworkandlearningenvironments.Theresultisthat
manypeoplenolongerhavethemeanstoachievethepersonalandsocialgoalsthatweallstriveforinoursociety.
Thereisanevenmoreinsidiousaspecttotheclaimthatrewardsarenegative:theclaimiselitist.Wealladmirechildrenwhoenjoylearningandwhodosowithout
anyobviousrewards.Thechildwhoisapparentlyselfmotivatedtolearnisvaluedinoursocietybutwhoisthischild?Usually,heorshecomesfromawelleducated
family,whichteachesthechildtheskillsandvaluesoflearning.Childrenfromsuchfamiliesgooneducationalexcursions,travelwiththeirparentstootherpartsofthe
world,andhavethemeansandresourcestoinquireabouttheworldinwhichtheylive.Thesechildrenreceivepraise,recognition,andprivilegesfortheir
accomplishments.Inaschoolsystembasedonselfmotivation,thesechildrenexcel.

THECLAIMTHATREWARDSARENEGATIVEISWRONGANDHARMFUL
Whyisitwrong? Whyisitharmful?
Basedonanidealisticandfaultyviewofhumannature Preventsusfromdesigningbetterworkandlearningenvironments
Evidencedoesnotsupporttheclaim Deniespeoplethemeanstoachievepersonalandsocialgoals
Basedonseriousoverstatementandgeneralizationofresearchfindings Denieslessprivilegedindividualsopportunitiesandaccesstoresources
Basedonconceptualconfusion,faultyreasoning,andmisunderstandings
Researchshowsthatrewardscanbeusedeffectively

Butwhatisthefateoflessprivilegedchildren?Inaschoolsystemthathasrejectedtheuseofpositivereinforcementandrewardprograms,lessprivilegedchildrenare
stillexpectedtoinquireanddiscoverontheirown.Theproblemis,however,thattheymaynothavethedesire,theskills,ortheresourcestosucceedwithouthelp.
Suchchildrencomefromfamilieswhohavebeenunabletoprovidethemwitheducationalopportunitiesandpositiveconsequencesprior
Page6

toenteringschool.Withoutpositivereinforcementforprogressinlearning,lessprivilegedchildrenfalltothewaysideintheschoolsetting.Theybecomeclassroom
managementproblemsandeventuallyencounterthepunitiveaspectsoftheeducationalsystemandsocietyingeneral.Inotherwords,ourreluctancetousepositive
incentivesresultsinamajorinjusticetoalargesegmentofchildreninoursociety.
Weagreewiththosewhoseektoreducetheuseofsocialpunishment.Thenegativesideeffectsofpunishmentarewellestablished.5Itisalsotruethatrewardscan
beusedinawaytobenefitthoseinpowerandcontrolthosesubjectedtotherewardsystem.Sucharewardsystemwouldbenegative.Butletusbeclear.Thereisno
needtoequatepunishmentwithreward.Rewardsandpositivereinforcementcanbeusedeffectivelytoenhancepeoplesperformance,interest,andenjoyment.The
viewthatrewardsareinherentlydetrimentalisawrongheadednotion,whichmustberectifiedifwearetoseeimprovementsineducation,work,andotherapplied
settings.
Basicprinciplesofpositivereinforcementarenowwellestablished.6Inaddition,thereisalargebodyofresearchonhowtoimplementtheseprinciplesineducation
andbusiness.7Designingsocialenvironmentsthroughtheuseofrewardsandpositivereinforcementisnotaneasytask,butitcanbedone.Thosewhoparticipateinthe
designofsuchenvironmentsmusthaveanexperimentalattitudeandbewillingtoadjustandfinetunethesystembasedonmeasuresofitseffectiveness.Aspartofan
ongoingassessment,peoplewithinthesystemmustbeconsultedandtheirevaluationsmustbeusedtoimproveit.Mostimportant,peoplewillexpresssatisfactionand
likewhattheyaredoingifpositivereinforcementandrewardsareproperlyarranged.

ANEFFECTIVEREWARDSYSTEMREQUIRES:
anexperimentalattitude
continualfinetuningofthesystem
inputfrompeoplewithinthesystem
ongoingevaluationoftheeffectivenessofthesystem
ANEFFECTIVEREWARDSYSTEMLEADSTO:
increasedsatisfactionforstudentsandemployees
recognitionofaccomplishments
adesiretoattainhighstandards
ameanstoachievepersonalandsocialgoals
highproductivity
feelingsofcompetenceandfreedom
Page7

Theclaimthatrewardsandreinforcementaredestructivehasalonghistorythatinvolvesmanycherishedbeliefsabouthumannatureaswellasnumerous
misunderstandingsaboutascienceofhumanbehavior.Thisbookisourattempttoclarifymajorhistoricalandcontemporaryissuessurroundingthistopic,present
currentfindings,addressmisconceptionsabouttheuseofrewardwithinascienceofbehavior,anddescribehowrewardscanbeusedeffectivelywithinapplied
settings.
Inthenextsection(PartII),wedescribeandassesstheinitialresearchonrewardsandintrinsicmotivationthatwasconductedintheearly1970s.Althoughthe
findingsfromtheinitialstudieswereweakandinconclusive,theyarefrequentlycitedasevidencethatrewardshavegeneralizednegativeeffectsonhumanbehavior.
Thisviewhasbeentoutedforthepastthreedecadesinnewspaperarticles,majorpsychologytextbooks,educationandbusinessjournals,andmanyotheroutlets.Our
evaluationoftheearlystudies,however,showsthatproducinganegativeeffectofrewardrequiresanunusualcombinationofconditionsthatisnotcharacteristicofthe
actualuseofrewardsineverydaylife.Nonetheless,theideathatrewardsdisruptanindividualsintrinsicmotivationhascapturedagreatdealofattention.Numerous
theorieshavebeenputforwardtopredictandaccountfortheeffectsofrewards,andalargebodyofexperimentalresearchhasaccumulatedonthetopic.
DifferenttheoreticalperspectivesarepresentedinPartIIIofthebook.Chapter4dealsspecificallywiththeoriesthatviewrewardsasharmfulChapter5covers
theoriesthatfocusonpositiveeffectsofrewardandreinforcement.InPartIV,wereviewandevaluatemorethan30yearsofresearch(over100experiments)on
rewardsandintrinsicmotivation.Amajorfeatureofthissectionisourmetaanalysisoftheliterature.ThemetaanalysispresentedinChapter8isdesignedtoaddress
shortcomingsofpreviousreviewsonthetopic,whiledrawingonstrengths.Basedonourfindings,weshowthatnegativeeffectsofrewardarealimitedphenomenon.
Atthesametime,weidentifyconditionsunderwhichrewardscanbeusedeffectivelytoenhanceandmaintainperformanceandmotivation.InChapter9,thesefindings
arediscussedintermsoftheoreticalandpracticalimplications.
InPartVofthisbook,therewardsandintrinsicmotivationliteratureisexaminedinasociohistoricalcontext,specifically,inlightofthedebatebetweenbehaviorism
andotherpsychologies.Ourcontentionisthatthehypothesisthatrewardsundermineintrinsicmotivationwaspropagatedbecauseitprovidedanapparentempirical
attackonbehavioralconceptionsofhumannature.Thenotionthatrewardssappeoplesintrinsicmotivationcanbeunderstoodasanoffshootofothersentimentsand
concernsthatwerepopularinthe1960sand1970s.Externalrewards,structuredworkplacesandclassrooms,competition,performanceevaluations,andsoonwere
allviewedasformsofcontrolthatconstrainindividualfreedomandexpression.Thus,althoughtheresearchfindingsdonotsupporttheviewthatrewardsareinherently
harmful,giventheviewsthatdominatedtheperiodwhenthisresearchwasinstigated,itisnotsurprisingthattheliteraturewasinterpretedinthisway.
Page8

PartVIofthebookdealswithpracticalapplicationsofrewards.InChapter11,wediscusshowtoenhanceperformanceandmotivationineducationalandbusiness
settingsthroughtheeffectivearrangementofrewardsandreinforcement.Thefinalsectionofthebook,PartVII,providesanoverallsummaryofthecontroversyover
rewardsandintrinsicmotivation.
Ourpremiseinthisbookisthatthereisnoinherentnegativepropertyofrewards.Theyarenotharmfultohumannature.Ifanything,rewardsareanaturalpartof
humannature.Whatisimportantistoensurethatrewardsystemspromote,ratherthanrestrict,humanfreedom.Thisisexactlywhatwemeanbythephrase,the
effectiveuseofrewards.Asweenterthenewmillennium,thetimeisripetodebunkoldmythsandtobringrewardsbackintothepicture.Anuncriticalrelianceon
weakexperimentaldataandonaphilosophyassertingthatrewardsareharmfultohumannaturehasbeenmaintainedattheexpenseofbuildingabetterworldinwhich
tolive.

NOTES
1.Sidman,1989.Sidmanprovidesananalysisofpunishmentandotherformsofaversivecontrolandshowsthattheseformsofcontrolofhumanbehaviorproducenumerousproblems
forsociety.
2.Kohn,1993a.
3.Deci,1971.
4.ThereaderisreferredtotheJournaloftheExperimentalAnalysisofBehavior,theJournalofAppliedBehaviorAnalysis,andtextbooksonmodernbehaviortheory(e.g.,Pierceand
Epling,1999Mazur,1998).
5.AzrinandHoltz,1966.
6.Skinner,1953.
7.Kazdin,1994MartinandPear,1999.
Page9

PARTII
REWARDSANDINTRINSICMOTIVATION:ALOOKAT
THEEARLYRESEARCH
Page10

Thispageintentionallyleftblank.
Page11

Chapter2
HowRewardsGotaBadReputation
Thecontemporaryviewthatrewardsandreinforcementareharmfulisgroundedinaliteraturefromsocialpsychologyconcerningintrinsicmotivation.Aprominentview
statesthatrewardsandreinforcementdecreaseapersonsintrinsicmotivationtoengageinanactivity.Essentially,theconcernisthatalthoughmoney,highgrades,
prizes,andevenpraisemaygetpeopletoperformanactivity,themotivationtocontinuetheactivitywillbelostoncetherewardsstopcoming.Forexample,ifachild
whoenjoysreadingisrewardedwithmoneyforreading,theclaimisthatheorshewillreadlessandenjoyreadinglessoncethemoneyisdiscontinued.Inotherwords,
rewardsaresaidtohaveanegativeeffectonpeoplesintrinsicmotivation.
Althoughthissoundsstraightforward,therelationshipbetweenrewardsandintrinsicmotivationisriddledwithproblems.Thereareproblemswiththeconceptof
intrinsicmotivation,thebasicexperimentalsituationsusedtoshowthatrewardsareharmful,andtheconclusionsdrawnfromtheseexperiments.Webeginthischapter
withadefinitionofintrinsicmotivationweshowthattheconceptinvolvescomplexitiesandconfusions.Next,wedescribetheresearchhypothesisthatrewardsare
harmfultointrinsicmotivation.Thishypothesishasguidedmostoftheresearchinthisarea.Weprovideanoverviewoftwohighlycitedexperimentsontheeffectsof
rewardsandintrinsicmotivation.Althoughtheresultsofthesestudiesarenotdefinitive,theysetthestagefortheviewthatrewardsareharmful.Aconsiderationofthe
detailsofthesestudieshelpstorevealproblemswithprocedures,measures,andovergeneralizedconclusions.
Page12

WHATISINTRINSICMOTIVATION?
Intrinsicmotivationisatermusedmostlyinsocialpsychology.1Ineverydaylanguage,thetermissimplyanotherwayofsayingthatpeopleareinterestedin,and
enjoy,whattheyaredoing.Accordingtothesocialpsychologicalliterature,peoplearesaidtobeintrinsicallymotivatedwhentheydoanactivityforitsownsake,
ratherthanforanyextrinsicreward.2Thetermintrinsicmotivationisoftenusedincontrasttoextrinsicmotivation.Extrinsicallymotivatedbehaviorsarethosein
whichanexternalcontrollingfactorcanbereadilyidentified.Forexample,ifpeoplesolvepuzzles,playgames,orpaintpicturesfornoobviousexternalreason,theyare
saidtobeintrinsicallymotivated.Ontheotherhand,studentswhostudyhardtoobtainhighgrades,employeeswhoworkextrahoursforpay,andchildrenwhodo
theirhomeworktopleasetheirparentsaresaidtobeextrinsicallymotivated.

INTRINSICMotivation EXTRINSICMotivation
usedtorefertobehaviorsforwhichthereisnoapparentrewardexceptthe usedtorefertobehaviorsinwhichanexternalcontrollingvariablecanbereadily
activityitself identified
performanceintheabsenceofreward

Now,youcanprobablyseeadifficultywiththesedefinitions.Thetermintrinsicmotivationisdefinedbytheabsenceofobviousexternalfactorssuchasextrinsic
rewards.Ifthereasonsfordoingandenjoyinganactivityarearesultofpastconsequences,duetoanticipatedfuturebenefits,notobvious,orunknown,thebehavioris
characterizedasintrinsicallymotivated.3Inotherwords,whenwedonotknowwhyapersonengagesinaparticularactivity,weinferintrinsicmotivation.Thus,the
motivesforengaginginmanyactivitiesgetlumpedintothecategoryofintrinsicmotivation.Asaresult,behaviorduetodistant,hidden,orobscureexternalcausesgets
mistakenlylabeledasintrinsicallymotivated.4
Afurthercomplicationisthatthetermsinterestandmotivationarefrequentlyusedinterchangeablybothareinferredfrombehavior.Technicallyspeaking,intrinsic
motivationissupposedtobetheinnercauseofanindividualsinterestinanactivity.Substitutingtheterminterestformotivationresultsinconsiderableconfusion.
Nonetheless,severalsocialpsychologistsusetheterms,andaresearchhypothesisconcerningrewardsandintrinsicmotivationhasbeenformulated.Theconjectureis
thatexternalrewardsdestroyintrinsicmotivation.

THEDETRIMENTALEFFECTSOFREWARDHYPOTHESIS
Inthe1950sand1960s,theviewwasthatintrinsicandextrinsicmotivationwereindependent.Sometheoristsassumedthatintrinsicandextrinsicrewards
Page13

wereadditiveandcombinedtoincreaseoverallperformanceandmotivation.5Intermsofwork,theviewwasthatthehighestmotivationtoperformwouldoccurina
systeminwhichjobswereinterestingandchallengingandinwhichtheemployeewasextrinsicallyrewardedforperformance(e.g.,withpay,recognition,etc.).Thatis,
productivityandsatisfactionwouldbehighestwhenintrinsicmotivationwassupplementedwithextrinsicincentives.
Thenotionthatextrinsicrewardscouldbeharmfultooverallmotivationwasraisedinthelate1960sbythepsychologistR.DeCharms.6DeCharmsspeculatedthat
intrinsicandextrinsicmotivationmaynotbeadditiveandthatexternalrewardsmightactuallyinterferewithintrinsicmotivation.Hefurthersuggestedthatexternal
rewardswouldchangepeoplesperceptionsaboutthecausesoftheirbehavior.7Ifpeoplewererewardedforengaginginactivities,theywouldbegintoseethemselves
asdoingtheactivityfortherewardratherthanforinterestandenjoyment.Inthisway,DeCharmssuggested,externalrewardsundermineintrinsicmotivation.
Theideathatextrinsicrewardscoulddisruptanindividualsintrinsicmotivationledtothedetrimentaleffectsofrewardhypothesis,whichstatesthatexternalrewards
undermineintrinsicmotivation,eitherbysubvertingperceptionsofcompetenceandselfdeterminationorbydeflectingthesourceofmotivationfrominternaltoexternal
causes.

THEDETRIMENTALEFFECTSOFREWARDHYPOTHESIS
Extrinsicrewardsunderminepeoplesintrinsicmotivation.

Thedetrimentaleffectsofrewardhypothesishasbeenhighlyinfluentialinsocialpsychologyandeducation,leadingmanyresearcherstoinvestigatetherelationship
betweenexternalrewardsandintrinsicmotivation.Sincetheearly1970s,over100studieshavebeenconductedonthistopic.PartIVofthisbookdealswiththis
enormousliterature.Inthischapter,weexaminetwoearlystudiesonrewardsandintrinsicmotivation.Thesestudiesarethemostcitedexamplesofthenegativeeffects
ofreward.Basedontheseearlyexperiments,thehypothesisgainedwidepublicacceptance.Giventhiswideacceptance,itisinformativetocloselyexaminethe
proceduresanddetailsoftheearlyexperiments.

DECISORIGINALEXPERIMENTS

ThefirstlaboratoryinvestigationstotesttheeffectsofrewardsonintrinsicmotivationwereconductedbyEdwardDeciintheearly1970s.8Inthefirstexperiment,24
collegestudents,fulfillingacourserequirement,werepresented
Page14

withapuzzlesolvingtask(usingSoma,acommercialpuzzlecomposedofsevendifferentshapesthatcanbefittedtogethertoformaninfinitevarietyof
configurations).TheSomapuzzlewaschosenbecausethecollegestudentsindicatedhighinterestinit.9Thestudywasmadeupofthreeonehoursessionsovera
threedayperiod.Twelvestudentswereassignedtoanexperimentalgroupandtheremainder,toacontrolgroup.Duringeachsession,theparticipantswere
individuallytakentoaroomandaskedtoworkontheSomapuzzlesinordertoreproducevariousconfigurationsthatweredrawnonapieceofpaper.Fourpuzzles
werepresentedinasession,andthestudentsweregiven13minutestosolveeachone.Inthesecondsessiononly,experimentalparticipantsweretoldthattheywould
receive$1foreachpuzzlesolved.Theywerepaidpriortothethirdsession.Thecontrolgroupwasnotofferedanymoney.
Inthemiddleofeachsession,theexperimentermadeanexcusetoleavetheroomforeightminutes.Theparticipantsweretoldthattheycoulddoastheypleased.
Thepuzzles,threemagazines,andanashtraywereavailableintheroom.Duringtheeightminuteperiod,theexperimenterobservedtheparticipantsthroughoneway
glassandrecordedthetimethateachindividualspentengagedontheSomatask.Theamountoftimespentonthetaskduringthefreechoiceperiodwastakenas
onemeasureofintrinsicmotivation.Asecondmeasureofintrinsicmotivationwastheparticipantsselfreportedinterest(attitudetowardsolvingpuzzles).10
Inaccordwiththedetrimentaleffectshypothesis,Deciproposedthatreward(money)woulddecreasesubsequentintrinsicmotivationandthatparticipantsinthe
experimentalgroupwouldspendlesstimeontheSomapuzzlesinthethirdsessionthantheyhadinthefirst.Hepredictedthattherewouldbeastatisticallysignificant
differencebetweentheexperimentalandcontrolgroupsonthisfreetimemeasure.Decisfindingsindicatedthattherewardedgroupdidspendlesstimeonthetaskin
thefreechoiceperiodthanthenonrewardedgroup,butthedifferencedidnotmeetconventionallevelsofstatisticalsignificance.11Onthesecondmeasureofintrinsic
motivationtheselfreportsoftaskinterestnosignificantdifferenceswerefoundbetweenthetwogroups.Thatis,thosewhohadreceivedarewardreportedas
muchinterestintheSomataskasdidthoseinthecontrolgroup.
Thus,Decisfirststudyonthetopicofrewardsandintrinsicmotivationshowednodifferencesbetweentherewardgroupandthecontrolgrouponthetaskinterest
measure.Andalthoughtherewasasmalldifferencebetweenthetwogroupsonthefreetimemeasure,thisdifferencewasnotstatisticallysignificant.Inotherwords,
thefindingswerenotinaccordwiththedetrimentaleffectshypothesis.Nonetheless,Deciclaimed,Ifapersonisengagedinsomeactivityforreasonsofintrinsic
motivation,andifhebeginstoreceivetheexternalreward,money,forperformingtheactivity,thedegreetowhichheisintrinsicallymotivatedtoperformtheactivity
decreases.12
Inanotherstudy,Deciusedasomewhatdifferentexperimentaldesignandfurtherinvestigatedtheeffectsofexternalreward(money)oncollegestudents
Page15

DECIS(1971)INITIALSTUDYONTHEEFFECTSOFREWARDSONINTRINSICMOTIVATION

ExperimentalGroup:
12collegestudentsofferedmoneytosolvepuzzles

ControlGroup:
12collegestudentsaskedtosolvepuzzles(noofferofmoney)

Task:
SolvingSomapuzzles

Reward:
$1foreachoffourpuzzlessolved

MeasuresofIntrinsicMotivation:
Differencebetweenexperimentalandcontrolgroupson:
(1)Freetimeontasktimespentduringfreechoiceperiodwithoutreward
(2)Selfreportedtaskinterest

Results:
(1)Freetimerewardedgroupspentlessfreetimeontaskthancontrolgroup,butdifferencewasnotstatisticallysignificant
(2)Selfreportedtaskinterestnodifferencebetweengroups

intrinsicmotivation.13Thestudyinvolvedtwo,ratherthanthree,sessions.Inthefirstsession,experimentalparticipantswereofferedmoneytosolveSomapuzzles
thecontrolgroupwasnotofferedmoneybutwasaskedtosolvepuzzles.Thesecondsessioninvolvedafreechoiceperiodwithoutrewardparticipantswerefreeto
domoreSomapuzzlesortoengageinalternateactivities.AsinDecisfirststudy,intrinsicmotivationwasindexedbythedifferencebetweenrewardedandcontrol
participantsontimespentontheSomapuzzlesduringthefreechoiceperiod(freetimeintrinsicmotivation).Interestingly,thosewhowerepromisedandgiventhe
rewardpriortothefreechoiceperiodspentmorefreetimeonSomapuzzlesthanthecontrolgroup.14Thisdifferencewasstatisticallysignificant.AlthoughDeci
continuedtoclaimthatexternalrewardsreduceintrinsicmotivation,whatthisstudyactuallyshowedwasthatstudentswhowererewardedshowedanincreaseinthe
mainmeasureofintrinsicmotivation(freetime).Thus,intermsofmoneyasanextrinsic
Page16

reward,Decisearlyfindingsdidnotshowsignificantdecreasesinmeasuresofintrinsicmotivation.15
Decialsoinvestigatedtheeffectsofverbalrewards(praise)onstudentsintrinsicmotivationintwoearlystudies.16ParticipantswereaskedtosolveSomapuzzles.
Experimentalparticipantsweretoldaftereachpuzzlethattheirperformancewasverygoodormuchbetterthanaverage.Inbothstudiesusingverbalpraiseasthe
reward,Decifoundthatthepraisedgroupspentmorefreetimeonthetaskthanthosewhoreceivednopraise.ThesefindingswerestatisticallysignificantinDecisfirst
studybutnotsignificantinthesecondexperiment.
Overall,theresultsfromDecisearlystudiesshowedthatverbalrewardsappearedtoincreasemotivationtoperformanactivitytangiblerewards(money)produced
eitheranincreaseinintrinsicmotivationmeasuresornoreliableeffects.InspiteofthefactthatthefindingsfromDecisinitialexperimentsdidnotprovidesupportfor
thedetrimentaleffectsofrewardhypothesis,Decisearlyexperimentsarefrequentlycitedintextbooksasevidencefornegativeeffectsofreward.

THESTANFORDEXPERIMENT
OneofthebestknownandmostcitedstudiesonthedetrimentaleffectsofrewardonbehaviorcomesfromanexperimentalstudyconductedbyLepper,Greene,and
Nisbettinthe1970satStanfordUniversity.17Inthisstudy,nurseryschoolchildrenwereobservedinafreeplayperiodinordertodeterminetheirinitialinterestona
drawingactivity.Twoobserverssatbehindaonewaymirrorandrecordedtheamountoftimeeachchildwasengagedindrawing.Thechildrenwhospentthemost
timeonthetaskwereselectedasparticipantsfortheexperiment.Threeconditionswerethensetup.Intheexpectedrewardcondition,childrenwereoffereda
goodplayeraward,whichtheyreceivedfordrawingwithmagicmarkers.Childrenintheunexpectedrewardgroupreceivedtheawardbutwerenotpromisedit
beforehand,andnorewardparticipantsdidnotexpectorreceiveanaward.
Inasubsequentfreeplaysession,thosechildrenwhowerepromisedanaward(expectedrewardgroup)spentsignificantlylesstimedrawingthantheothertwo
groups.Furthermore,theexpectedrewardgroupspentlesstimedrawinginthepostexperimentalsessionthantheyhadintheinitialsession(preexperimentalfreeplay
session).Theunexpectedrewardandnorewardgroupsdidnotdiffersignificantlyfromoneanotherbothgroupsshowedslightincreasesintimeontaskfromthepre
experimentaltopostexperimentalsessions.
Page17

LEPPER,GREENE,ANDNISBETTS(1973)STUDYONTHEEFFECTSOFREWARDSONINTRINSICMOTIVATION

ExperimentalGroup1(expected[promised]reward):
18nurseryschoolchildrenofferedanawardtodraw

ExperimentalGroup2(unexpectedreward):
18nurseryschoolchildrenaskedtodraw(noofferofrewardbutrewardwasgivenafterchildrenhaddrawn)

ControlGroup:
15nurseryschoolchildrenaskedtodraw(noofferofreward,norewardgiven)

Task:
Drawingwithmagicmarkers

Rewards:
Goodplayeraward

MeasuresoflntrinsicMotivation:
Differencebetweengroupsonfreetimeontasktimespentduringfreechoiceperiodwithoutreward

Results:
Freetime
(1)expectedrewardgroupspentsignificantlylessfreetimeontaskthanunexpectedrewardgroupandcontrol(p<.05).
(2)nosignificantdifferencesbetweenunexpectedrewardandcontrolgroups

Basedonthefindings,Lepperandhiscolleaguesconcludedthattheirresultsprovidedempiricalevidenceofanundesirableconsequenceoftheunnecessaryuseof
extrinsicrewards.18Althoughtheymadethisconclusion,itisinterestingtonotethatthosewhohadreceivedanunexpectedrewardspentmoretimeonthetaskduring
thepostexperimentalfreeplayperiodthaneithertheexpectedrewardorthecontrolgroups.Becausetheunexpectedandexpectedrewardgroupsarebothreward
conditions,theconclusionthattheseresultsdemonstrate
Page18

Table2.1
ASummaryoftheEarlyFindingsontheEffectsofRewardsonIntrinsicMotivation
INTRINSICMOTIVATIONMEASURE
Study Typeofreward Freetime Taskinterest
Deci(1971) Expectedtangible(money) n.s. n.s.
Verbalpraise Increase* n.s.
Deci(1972b) Expectedtangible(money) Increase*
Verbalpraise n.s.
Lepper,Greene,andNisbett(1973) Expectedtangible(goodplayeraward) Decrease*
Unexpectedtangible(goodplayeraward) n.s.
Note:*=statisticallysignificantatp<.05n.s.=notsignificant.

thenegativeeffectsofrewardisnotcorrect.Thisisbecauserewardwasheldconstantintheunexpectedrewardandexpectedrewardgroupswhatdifferedwasthe
conditionofpromiseornopromise.Thatis,thepromisesmademusthaveproducedtheresults.19Asthisisthecase,theexperimentactuallyshowedthatpromisesof
reward(tellingchildrenthattheywouldreceivearewardfordoingtheactivity),andnotrewardperse,reducedthetimespentonthetask.

ASUMMARYOFTHEEARLYRESEARCHFINDINGS
Thedetrimentaleffectofrewardhypothesisbecameprominentintheearly1970s.Theclaimwasthatrewardsunderminepeoplesintrinsicmotivation.Atthepoint
whenthedetrimentaleffectshypothesisbecamepopular,however,andrewardsgotabadreputation,theexperimentaldatadidnot,infact,provideconclusive
evidencefornegativeeffectsofreward.FindingsfromtheearlystudiesaresummarizedinTable2.1.
Decis(1971)firststudyshowedthattheeffectsoftangiblerewards(money)werenotstatisticallyreliableandthatverbalrewardshadpositiveeffects.Table2.1
alsoshowsthatinasubsequentstudybyDeci(1972b),tangiblerewardswerefoundtoincreasefreetimeintrinsicmotivation,whereasverbalrewardsshowedno
significanteffect.ThestudybyLepperandhisassociates(Lepper,Greene,andNisbett,1973)showedthatwhentangiblerewardsweredeliveredunexpectedly,free
timeintrinsicmotivationwasunaffected.However,Lepperandcolleaguesdidshowasignificantdecreaseinfreetimeintrinsicmotivationwhentherewardwastangible
andofferedbeforehand(expected).InDecisfirst
Page19

study,intrinsicmotivationwasalsoindexedbyselfreportsoftaskinterest.Nosignificantfindingsweredetectedforeithertangibleorverbalrewards.
Table2.1showsthattheresultsfromtheearlystudiesproducedmixedeffects.Clearly,theresultsdidnotindicatethatrewardsaregenerallyharmful.Infact,the
earlyresearchshowedthatrewardscouldbeusedtoincreasemeasuresofintrinsicmotivation.Nevertheless,thefindingsfromtheearlyresearchareoftencitedin
newspapers,journalarticles,andintroductorypsychologytextbooksasevidencethatextrinsicrewardsundermineintrinsicmotivation.Somewritersequatereward
withreinforcementandclaimthattheearlystudiesaredemonstrationsofdetrimentaleffectsofpositivereinforcement.20

POPULARACCEPTANCEOFTHEVIEWTHATREWARDSAREHARMFUL
Followingthepublicationoftheinitialstudiesonrewardsandintrinsicmotivation,numerousstatementsbegantoappearinthepsychologicalandeducationalliterature
warningpractitionersofthedangersofimplementingincentiveprogramsbasedonrewardsandreinforcement.Inthe1970s,themessagewasthatpositive
reinforcementtechniquesandtokeneconomyprograms(programsinwhich,forappropriatebehavior,individualsaregiventokensthatareexchangeableforgoods,
services,andactivities)maybeworsethanineffective.Drawingontheresultsoftheearlystudiesonrewardsandintrinsicmotivation,theauthorsofanAmerican
Psychologistarticlearguedthatsincetokenstendtodecreasetheintrinsicvalueofanactivity,theymayactuallydomoreharmthangood.21
Suchcommentsandcautionshavebeenechoedthroughoutpsychologyandeducationeversince.Forexample,inthe1990s,theauthorsofoneprominenttextbook
citedLeppersworkandstatedthat

whenanextrinsicrewardisgiven,themotivationbecomesextrinsicandthetaskitselfisenjoyedless.Whentheextrinsicrewardsarewithdrawn,the
activitylosesitsmaterialvalue.Themoralis:Arewardadaymakesworkoutofplay.22

AtradebookentitledPunishedbyRewards,byAlfieKohn,hasreceivedconsiderableattentionforitsgeneralthesisthatinbusinessandeducation,theuseofrewards
hasadetrimentalimpact.23Theexperimentsfromsocialpsychologyonrewardsandintrinsicmotivationareusedtosubstantiatethisclaim.Kohnextendedhiscritique
ofexternalrewardstoincludeverbalrewards,particularlytheuseofpraise.24
Inarecentarticleinatopeducationaljournal,Deciandhiscolleaguescautionedteachersandsuggestedthatthereisindeedreasonforteacherstoexercisegreat
carewhenusingrewardbasedincentivesystems.25Aswehaveseen,theinitialresearchfindingswerenotconclusive.Nonetheless,theoriginalexperi
Page20

mentscontinuetobecitedasevidencethatrewardsaredetrimental.Inacurrenteditedvolumetitled,IntrinsicandExtrinsicMotivation,26severalofthewriters
pointtotheearlyresearchandclaimthatthefindingsindicatethatextrinsicincentivesundermineintrinsicmotivation.27
Thepopularizationofthenegativeviewofextrinsicrewardshasfosteredpublicattitudesagainsttheuseofrewardsandincentives.IntheUnitedStates,therehas
beenpublicpressuretoimprovechildrensreadingabilities.Businessandphilanthropicorganizationshavefundedprogramsinwhicheverythingfrommoneytopizza
hasbeenofferedtoincreasethenumberofbookschildrenread.ArticlesappearingintheNewYorkTimesandU.S.NewsandWorldReporthavecomplainedthat
readforrewardprogramsaremakingchildrenintononreadersbydestroyingtheirenjoymentofreading.28RecentarticlesinCanadiannewspapershavealso
focusedonthetopic.Aconcernisthatrewardingchildrenisakintobriberyandsuchbriberywillpreventchildrenfromeverbeingmotivatedtolearnforthesakeof
learning.29
Thereisnodoubtthattheearlyresearchonrewardsandintrinsicmotivationhashadapowerfulimpact.ManydrewconclusionsfromDecisandLepperandhis
colleaguesearlystudiesthatwentwellbeyondthescopeoftheresearch.Statementsthattheresultsconfirmedthenegativeeffectsofrewardweresoonpickedupby
themediaandpublic.Therumorthatrewardsareharmfulhasfilteredintopsychologytexts,educationaljournals,andbusinessmagazines.
Dorewardsdeservesuchabadreputation?Shouldrewardsberemovedfromeducational,business,andotherappliedsettings?Ouranswerisafirmno.Rewards
canbeusedaseffectivemeansforcultivatinginterestandincreasingmotivationandperformance.Inthefollowingchapter,wedescribewhyrewardsdonotdeservea
badimage.

NOTES
1.AnativisticapproachtomotivationcanbefoundinthefirstsocialpsychologytextbookbyMcDougall(1908)andmaybeonereasonwhyconceptssuchasintrinsicmotivation
havesomesupportinthisdiscipline.Thatis,intrinsicmotivationisthoughtbysometobeaninnatepropertyofhumannatureinthesamewaythatinstinctswereformerlythought
toaccountforvariationinhumanbehavior.Indisputeofanativisticapproachtomotivation,Kuo(1931)raisedcatswithratsandshowedthattherewasnobasicinstinctforthecats
tokilltherats.Hearguedthatmuchbehaviorwaslearned(acquiredduringtheindividualslifetime).Onebasisoflearningthatweemphasizeinthisbookislearningby
consequences.Rewardsareasourceofconsequencesforhumanbehaviorthatarearrangedbyotherpeople.Itisthesesociallyarrangedconsequencesthataresaidtodisrupt
intrinsicmotivation.
2.Deci,1975.
3.Theremaybemanyreasonswhypeopledothingsincaseswhereexternalcontrolisnotobvious.SeeKohn,1993a,pp.270276.
4.Theconsequencesofactionareoftendifficulttoobservebecausetheyareremoteorintermittent.
Page21

5. PorterandLawler,1968Vroom,1964.
6. DeCharms,1968.
7. Whydopeoplewanttoseethemselvesascausesoftheirownactions?DeCharms(1968)tellsusthatthisisanotherbasic,innateneed.Skinner(1971)suggeststhatthereasonsare
social.Thelessobviousaretheexternalcausesofbehavior,themorecreditfortheiractionspeoplewillreceivefromothers.Actionsthatareduetoclearexternalreasonsreceive
littlerespectfromothers.Ontheotherhand,ifwecannotseethesourcesofinfluencethataffectapersonsbehavior,weattributeittothatperson.Thus,theneedtoseeoneselfas
acausalagentisnotinnate.Rather,thesocialconsequencesofappearingtobeselfmotivatedmaintainthisbehavior.
8. Deci,1971,1972a,1972b.
9. Theissueoftaskinterestisproblematic.Onlytasksofhighinitialinterestaresaidtobeunderminedbyexternalrewards.Studentsareoftenfacedwithtasksthattheydonotinitially
findinteresting,andrewardscanbeusedeffectivelytobuildinterestandskillfulperformanceinsuchcases(Bandura,1986,pp.241242).Thepointatwhichrewardschangefrom
beneficialtoharmfulisnotaddressedbytheoriesofintrinsicmotivation.
10.DeciandRyan(1985)initallyclaimedthattheselfreportinterestmeasureismorecloselyrelatedtointrinsicmotivationthanisthefreetimemeasure.Inarecentpublication,Deci,
Koestner,andRyan(1999)suggestedthatacompositemeasurecalledfreechoice(freetimeandperformanceonataskduringthefreechoicesession)isthebestmeasure.
11.Socialscientistsworkunderthealpha.05levelofstatisticalsignificance.Resultsthatdonotmeetthiscriterionarefailurestorejectthenullhypothesis.Thismeansthatthe
researcherisunabletorejectthehypothesisthatthereisnodifferencebetweenexperimentalandcontrolgroupsandtherefore,shouldnotfavorthealternate,researchhypothesis.
Decisresultswerenotsignificantatthe.05level(thefindingswerereportedassignificantatp<.10,onetailed).
12.Deci,1971,p.108.
13.Deci,1972b.
14.ThereweretworewardgroupsinDecis(1972b)study.BothgroupswereofferedmoneytosolveSomapuzzles.Onegroupwaspaidpriortothefreechoiceperiodthesecond
groupwaspaidafterthefreechoiceperiod.Participantswhowererewardedpriortothefreechoicesessionspentsignificantlymorefreetimeonthetaskthancontrolparticipants
therewasnosignificantdifferencebetweencontrolsandparticipantswhowererewardedfollowingthefreetimesession.
15.Deci(1972a)conductedanotherstudyusingmonetaryrewards,inwhichparticipantswereofferedmoneysimplyforparticipatingintheexperiment.Nosignificantdifferenceswere
detectedonthefreetimemeasureforrewardversuscontrolparticipants.
16.Deci,1971,1972b.
17.Lepper,Greene,andNisbett,1973.
18.Ibid.,p.136.
19.Promisesofrewardfallunderthecategoryofverbalstimulicalledinstructions.Instructionsregulatebehaviorbecauseofapasthistoryofconsequencesintheirpresence.For
example,aninstructionfromafriendsuchas,EatatJoes,youllreallylikethefood,willaffectyouif,inthepast,theadviceofthisfriendhasbeencredible.Thatis,ifpast
recommendationsmadebyyourfriendhaveturnedoutwell,youarelikelytotakefutureadvice.Intermsofpromisesofreward,apersonwillreacttothepromiseon
Page22

thebasisofpastexperiencesofpromises.Sometypesofpromisedrewardsareusuallydescribedasbribes.Often,whenabribeispresented,peoplewillreacttoitbydoingthe
opposite.Thus,ifthepromiseofrewardistakenasabribe,apersonmaydolessofthatactivityinreactiontothesocialcontrolimpliedbythebribe.Ratherthanalossofintrinsic
motivation,thisreactancetoimpliedcontrolmayaccountfortheresultsoftheearlyexperimentsonexpectedandunexpectedreward.
20.Behavioralpsychologistsmakeanimportantdistinctionbetweenthetermsreinforcementandreward.Reinforcementisdefinedbyitseffectsonbehavior.Areinforcerisanevent
thathasbeenshowntoincreasethefrequencyofthebehaviorthatitfollows.Areward,however,isdefinedsocially.Rewardsarestimulithatareassumedtobepositiveeventsbut
havenotbeenshowntostrengthenbehavior.Incentivesystems(e.g.,classroomtokeneconomies)maybebasedoneitherrewardorreinforcementandaredesignedtoincrease
motivation.
21.LevineandFasnacht,1974,p.819.
22.Zimbardo,1992,p.454.
23.Kohn,1993a.
24.AlthoughKohn(1993a)recognizedthattheliteratureonintrinsicmotivationdoesnotsupporttheconclusionthatpraisehasdetrimentaleffects,hestillcondemnedtheuseofpraise
andotherverbalreinforcement.Hisconcernwasthatpeopleusepraisetogetotherpeopletodowhattheywant.Thisiswhatwewouldcallinsincerepraise.Sincerepraiseispraise
givenwhenapersonaccomplishessomelevelofperformancethatevokesapositivereaction.Sincerepraisehaspositiveeffectsoninterestandperformance.Thepointisnotto
condemnpraiseperse,buttoeffectivelyuseittorewardaccomplishments.
25.Deci,Koestner,andRyan,2001,p.2.
26.SansoneandHarackiewicz,2000.
27.Forexample,seeLepperandHenderlong,2000.
28.Egan,1995Hawkins,1995.
29.Smyth,2001Vanderberg,2001.
Page23

Chapter3
WhyRewardsDoNotDeserveaBadReputation
Theearlyexperimentsonrewardsandintrinsicmotivationhaveledmanypsychologists,parents,educators,andotherpractitionerstorejecttheuseofrewardsandbe
suspiciousofprogramsbasedontheuseofpositivereinforcement.Numerousconfusionsandmisunderstandingsunderliethecontinuedpopularityoftheviewthat
rewardsandreinforcementareharmful.Inthischapter,wecarefullyanalyzeaspectsoftheearlyexperimentsinordertoclarifymanyofthemisunderstandingsand
misconceptionsthatpermeatethisliterature.Basedonthisanalysis,weconcludethatthereisnoneedtorejecttheuseofrewardsineverydaylife.

PROBLEMSWITHTHEEARLYEXPERIMENTS

ReinforcementversusReward
Theearlyresearchonrewardandintrinsicmotivationisoftenusedtoarguethatrewardsandreinforcementaredestructivetohumannature.Inthiscontext,itisuseful
todistinguishbetweenrewardandreinforcement.Technically,thetermreinforcementmeanstoincreaseorstrengthenreinforcementinvolvesaprocedurefor
strengtheningbehavior.Abehaviorislikelytoberepeated(orincreaseinfrequency)ifcertainkindsofconsequencesfollow.Theconsequencesthatincreasebehavior
arecalledreinforcersandareusuallythingsthatpeople(orotherorganisms)findrewardingorsatisfying.Forexample,ateachermaysay,Correct,goodjob!whena
studentsuccessfullysolvesmathprob
Page24

lems.Moststudentsliketosucceedattheirschoolwork,sotheteacherssocialrewardisexpectedtoreinforcethebehaviorofsolvingmathproblems.
Althoughonemightsuspectthatrewardssuchascorrect,right,oryeswillstrengthenbehavior,theactualeffectsofthesewordsonbehaviorisanopen
question.Atestforreinforcementinvolvesshowingthattheteachersverbalrewardsactuallyincreaseastudentsmathperformance.Ifthisisthecase,verbalrewards
aredefinedasreinforcementforthebehaviorofaparticularstudent.Ontheotherhand,verbalrewardsmaynotreinforceacademicpeformancebyanotherstudent.
Thepointisthatreinforcementmustbetailoredtotheindividual.Thismeansthatsocialandmaterialrewardswillnotserveasreinforcementforallpeople.Thisisan
importantpoint.Teacherscannotcountontheirrewardsservingasreinforcementforallstudents.Whatmayworkasareinforcerforonepersonmaynotworkfor
another.Dependingontheperson,rewardscouldhavenoeffect,haveareinforcementeffect,orevenfunctionaspunishment(decreasingbehavior).Whenrewardsare
perceivedasbribes,theyoftenhavethislattereffect.
Reinforcementproceduresalsousuallyinvolverepeatedpresentationofthereinforcerfollowingtheoccurrenceofthedesiredbehavior.Thismeansthattheteacher
arrangescorrectivecommentsforappropriateanswerstomathproblemsoveranextendedperiod(notonaonetimebasis)soastostrengthenthisbehavior.
Eventsthatarecalledrewardsaresociallydefinedandusuallyinvolvesuchthingsasgoldstars,awardsofexcellence,patsontheback,recognition,money,
opportunitiestodopreferredactivities,andsoon.Theseeventsarepresumedtobepositiveandmaybegivenforavarietyofreasons.Rewardsmaybegivenfor
honorableservice,toestablishapositiveattitudetowardtherewardgiver,ortosetupanobligationtoreciprocate.Onlywhenrewardsareshowntostrengthen
behaviorcantheybeequatedwithreinforcement.
IntheoriginalexperimentsbyDeci(1971,1972b)andbyLepperandhiscolleagues(Lepper,Greene,andNisbett,1973),therewasnotesttoascertainwhether
therewardsthatwereusedfunctionedasreinforcementforthetasksofpuzzlesolvinganddrawing.Inaddition,therewardswerenotdeliveredrepeatedly.Instead,
repeateddeliveryofreinforcementwasreplacedbyinstructions,promisesofreward,anddescriptionsofrewardcontingencies.Inotherwords,neitheroftheseearly
studiesinvolvedpositivereinforcementandintrinsicmotivation.Infact,oncloseinspection,thestudiesdidnotreallyinvolverewardseitherbutratherpromisesand
instructionsaboutreward.

PromiseofRewardversusReward
Theearlystudiesonrewardandintrinsicmotivationinvolvedpromisesofrewardanddescriptionsofrewardcontingencies.Thatis,peopleweretoldwhattheyhadto
dotogetareward.Thisiswhattheresearcherscalledexpectedreward.Itisthisexpectedrewardconditionthatwasfoundtodecreasetime
Page25

spentonthetaskinthefreechoicephaseoftheexperimentbyLepperandhiscolleagues.1
AlyceDickinsonprovidesausefulanalysisofthedistinctionbetweenreinforcementcontingencyandthedescriptionofcontingencies.2Shemakesthepointthat
describingwhatpeoplearetodoandofferingthemarewardfordoingitmayresultintheoccurrenceofthedesiredbehavior.Butverbaldescriptionsofcontingencies
usuallydonotshowallthefeaturesofbehaviormaintainedbyreinforcement.Similarly,whathappensfollowingthepromisesandactualexposuretocontingenciesmay
bequitedifferent.Skinnermadeasimilardistinctionwhenhestated:

Themanipulationofindependentvariablesappearstobecircumventedwhen,insteadofexposinganorganismtoasetofcontingencies,thecontingencies
aresimplydescribedininstructions.Insteadofshapingaresponse,thesubjectistoldtorespondinagivenway.Ahistoryofreinforcementor
punishmentisreplacedbyapromiseorthreat....
Descriptionsofcontingenciesare,ofcourse,ofteneffective....Verbalcommunicationisnot,however,asubstituteforthearrangementand
manipulationofvariables.
Thereisnoreasonwhyadescriptionofcontingenciesofreinforcementshouldhavethesameeffectasexposuretothecontingencies.Asubjectcan
seldomaccuratelydescribethewayinwhichhehasbeenactuallyreinforced.Evenwhenhehasbeentrainedtoidentifyafewsimplecontingencies,he
cannotthendescribeanewcontingency,particularlywhenitiscomplex.Wecanscarcelyexpecthim,therefore,toreactappropriatelytodescriptionsby
theexperimenter.Moreover,theverbalcontingenciesbetweenthesubjectandexperimentermustbetakenintoaccount.Instructionsmustinsomeway
promiseorthreatenconsequencesnotgermanetotheexperimentifthesubjectistofollowthem.3

DickinsonelaboratesonSkinnerspoints,statingthatpromisesofrewardcanbeinfluencedbymanyfactors,including

thesophisticationofthesubjectsverbalrepertoire,thesubjectshistorywithrespecttowhetherthepromisedrewardswereactuallyreceived,thenature
ofthesubjectspriorexposuretotheobjectbeingofferedasreward,whethertheparticularwordingoftherequesttoperformthetaskhasbeen
correlatedwithpunishmentfornoncomplianceandeventsthatoccurduringthedelaybetweenthepromiseandrewarddelivery,suchasthewaythe
experimenterinteractswiththesubject.4

Mostimportant,peoplemayreacttodescriptionsandpromisesofrewardinadifferentwaythanthewayinwhichtheyreacttoactualrewardsoractualcontingencies
ofreinforcement.Infact,theearlyresearchshowedthatthedeliveryofrewardwithoutapromise(unexpectedreward)didnotleadparticipantstospendlesstimeon
thetask.Inspiteofthis,thedetrimentaleffecthypothesisiscastintermsofactualrewardsratherthanintermsofpromisesofrewardordescriptionsofreward
contingencies.Basedonfindingsoftheearlyresearch,a
Page26

morecorrectstatementofthenegativeeffectwouldbethattimeontaskdeclineswhenpeoplearepromisedarewardfordoingsomeactivity.5

WithdrawalofPromisedReward
Theearlyexperimentsonrewardsandintrinsicmotivationareoftencastasevidenceforthenegativeeffectsofdeliveringrewardsorreinforcement.However,itshould
bepointedoutthatthemajoreffectcamewhenthepreviouslypromisedrewardwaswithdrawn.Theparticipantsinthestudieswereofferedanddeliveredarewardfor
doingsometask.Followingthis,therewardwasremovedandtimeontaskwasmeasured.Thefindingsindicatedthatthosewhohadbeenpromisedarewardspent
lesstimeonthetaskoncetherewardwasremovedthanthoseparticipantswhohadnotbeenrewarded.Thisfindingwasinterpretedasevidencethatrewardsreduced
peoplesintrinsicmotivation.Amorelogicalinterpretation,however,isthatthechangefromastatedrewardcontingencytoanorewardsituationproduceda
temporarydeclineintimeontaskratherthanalossofintrinsicmotivation.
Toillustrate,supposealocalfastfoodrestaurantofferstogivefreemealsforeachbookastudentreadsduringamonthlongcampaign.Supposethatonechild
decidestoenrollinthecampaignwhileasecondchilddoesnot.Priortothecampaign,bothchildrenhavebeenreadingapproximatelythesamenumberofbooksatthe
samelevel.Duringthecampaign,thechildwhoisenrolledincreasesthenumberofbooksread(duetothepromisedreward).Thatis,thepromisedrewardenhances
motivationtoread.Theotherchildcontinuestoreadathisorherownpace.Thus,whiletheofferofrewardisineffect,thechildwhoisenrolledintheprogramreads
morethanthechildwhoisnotenrolled.Theofferedrewardincreasesthedesiredbehavior,whichistheobjectiveofthecampaign.
Oncethecampaignends,theconcernisthatthechildenrolledintheprogramwillreadlessbecauseofalossofintrinsicmotivationduringtherewardperiod.Will
thishappenintherestaurantexample?Onepossibilityisthatthechildenrolledintheprogramwillreadaboutasmuchasheorshedidbeforethecampaign.Thisseems
highlylikelyasalongrangeoutcome.Thatis,inthelongrun,therestaurantscampaignmayhavenolastingeffectsonthechildsreading.Therecouldbeashortterm
declineinreading.Thistemporaryreductionwouldnotbeduetoalossofintrinsicmotivationbuttothefactthatchildmayhavereadsomuchduringthemonthlong
campaignthatheorsheismomentarilytiredofreading.
Inaddition,thesignalthatrewardshavestoppedforanactivityoftenproducesitsimmediatecessation.Forexample,anoutofordersignonavendingmachinewill
preventapersonfrominsertingmoneytogetadrink.Butthisisamomentaryeffect,andthepersonwillcontinuetousevendingmachines.Inthesamesense,thechild
whoistoldthatthereadingcampaignhasendedmaystopreadingtemporarilybutwillreturntoreadinglateron.Thesetypesoftransient
Page27

effectscouldbewhattheearlyresearchersobservedinthelaboratorystudiesduringthebrieffreetimeplayperiod.Thepointisthatthesideeffectsofwithdrawalof
rewardarefleeting.Statementsthatrewardshavepervasive,longlastingnegativeeffectsonpeoplesintrinsicmotivationarenotsupportedbytheearlystudies.

EffectsofChoiceandtheFreeTimeMeasure
Inordertomeasureintrinsicmotivation,intheearlystudies,afreetimephasewasarrangedfollowingtheperiodofreward.Duringthisphase,participantshadthe
opportunitytoengageintheexperimentaltaskorinavarietyofotheractivities(e.g.,readingamagazine,smokingacigarette).Aproblemwiththisprocedureisthat
theseactivitiesintroduceanaspectofchoicethatwasnotpresentintheotherphases.
Theintroductionofchoicemeansthattherewardedparticipantswentfromaperiodofpromisedrewardforaspecificactivitytoaperiodofchoiceamongactivities.
Thedeclineintimespentontherewardedactivitycouldbeduetothischange.Considerthefactthatindividualswhowerepromisedarewardforsolvingpuzzlesor
drawingpicturesdidmoreofthisactivityintherewardphasethanthecontrolgroup,whichwasnotpromisedareward.Whenplacedinthefreetimesetting,the
rewardedparticipantscouldnowdistributetheirtimetotheexperimentaltaskortotheotheractivities.Becausetheyhadspentmoretimedoingtheexperimentaltaskin
therewardphase,thealternativeactivitiesinthechoicephasewouldtendtobecomemoreattractive.Thatis,theexperimentalparticipantswouldchoosethealternate
activitiesratherthanspendmoretimeonpuzzlesolvingordrawing.Thiswouldleadtoadeclineinabsolutetimespentontheexperimentalactivityandaccountfora
majorpartofthefindings.Inotherwords,thefindingsmayhavebeenduetotheintroductionofchoiceratherthantoalossofintrinsicmotivation.Ifthisisso,the
resultsreflectanartifactoftheproceduresratherthanafundamentalrelationshipbetweenrewardandintrinsicmotivation.

VerbalversusTangibleRewards
Thedetrimentaleffectsofrewardhypothesisconcernstheuseofexternalrewards.Externalrewardsarethosethatcomefromoutsidethepersonandareusually
arrangedbyotherpeople.InDecisfirststudy(1971)andinLepper,Greene,andNisbetts(1973)experiment,timeontaskwasfoundtodecreasefollowingthe
removalofapromisedtangiblereward(e.g.,money,goodplayeraward)givenforpuzzlesolvingordrawingpictures.ThefirststudybyDeci,however,alsoindicated
thatnotallexternalrewardsproducedadeclineinintrinsicmotivation.6Specifically,Decifoundthatwhenverbalpraisewasgivenforsolvingpuzzlesandthen
removed,theexperimentalparticipantsspentmoretimeonthetaskthantheunrewardedcontrolgroup.Attheveryleast,this
Page28

findinglimitsthegeneralityofthedetrimentaleffectsofrewardhypothesis.Thatis,onlysomeexternalrewardsproducedadeclineintimeontask,whereasothers
enhancedperformance.
Anotherpointisthatthereisnoinherentdifferencebetweenverbalandtangiblerewards.Itislikelythatverbalpraisecanbeusedinwaysthatmakepeopleless
enthusiasticaboutanactivity.Forexample,peoplewhousepraisetoingratiatethemselvestoothersmayfindthatotherpeopledolessforthem.Usingpraisetoconfer
statusdoesnotnecessarilymeanthatitwillenhanceperformance.Thatis,thefindingthatpraiseenhancesintrinsicmotivationdoesnotmeanthatitdoessoinall
situationsandcontexts.Clearly,itisthemannerinwhichrewardsarearrangedandimplementedthatdeterminestheireffectsonbehavior.

TangibleRewardsOfferedforDrawingversusSolvingPuzzles
InDecisearlystudies(1971,1972b),participantswereofferedrewardstosolvepuzzles.AsnotedinChapter2,Decifoundanegativeeffectofthistypeofrewardin
hisinitialstudy(althoughtheresultwasnotstatisticallysignificant),7butthesametypeofrewardwasfoundtoincreasefreetimeontaskinDecissubsequentstudy.8
Lepperandhiscolleagues(Lepper,Greene,andNisbett,1973),ontheotherhand,offeredarewardtoparticipantssimplyfordoingthetask(drawing).Therewasno
requirementtoachieveacertainlevelofperformance.Lepperandhisassociatesfoundastrongnegativeeffectwiththistypeofrewardarrangement.
Thesedifferingfindingssuggestthatitisnottangiblerewardortheofferofrewardthatproducedthenegativeeffect.Instead,theearlyfindingsindicatethatreward
effectsdependonwhatisrequiredforparticipantstoreceivethereward(thestatedrewardcontingency).TheLepperandcolleaguesstudy(Lepper,Greene,and
Nisbett,1973)showedthatwhentangiblerewardswereofferedbutnoperformancerequirementwasspecified,anegativeeffectoccurred.However,inDecis(1971,
1972b)studies,inwhichparticipantswereofferedatangiblerewardtomeetaperformancestandard(solvepuzzles),therewasnoevidenceofnegativeeffects.
Thesefindingsshowthattangiblerewardsarenotinherentlyharmful.Theimplicationisthattangiblerewardscanbeusedtoenhanceinterestandperformancewhen
theyareallocatedcorrectly.Forexample,ifatangiblerewardisgivenforachievingsomelevelofperformance,thereisnoreasontoexpectadeclineinperformance
andinterest.Clearly,itisnothelpful,atthepracticallevel,tocondemntheuseofallexternalrewards.Abetterstrategyistounderstandhowtouserewardseffectively
inappliedsettings.

AttitudeMeasureofIntrinsicMotivation
Accordingtotheearlyresearchers,intrinsicmotivationhasadirectlinkwithtaskinterest.Thatis,thegreateristheintrinsicmotivation,thegreaterthein
Page29

terest.Rewardsareexpectedtodecreaseintrinsicmotivation,andtherefore,expressedinterestinthetaskisalsoexpectedtodecline.Reportedinterestinpuzzle
solvingwasmeasuredinDecis(1971)originalexperimenttheresultsshowedthatinterestinthetaskwasnoterodedbythepromisedrewards.Ofcourse,thisfinding
isratherdevastatingtothedetrimentaleffectshypothesis,asitindicatesthatrewardscanbeusedwithoutreducingpeoplesinterestinatask.
Therelationshipbetweentaskinterestandthefreetimemeasureisalsoproblematic.InPartIVofthisbook,weshowthatthereislittleagreementbetweenthese
measures.Thatis,timeontaskmaydecrease,yettaskinteresttendstoremainunchangedoreventoincrease.Weonlymentionherethatthisisaseriousproblemfor
theoreticalinterpretationsofintrinsicmotivationbothmeasuresshouldcorrelateiftheyreflectthecommonenergysourceofintrinsicmotivation.

PROBLEMSWITHTHEEARLYEXPERIMENTSONREWARDSANDINTRINSICMOTIVATION
Rewardswerenotshowntobepositivereinforcers.
Rewardswereofferedorpromised(byinstructions)peopledidnotexperiencethecontingencybetweendoingthetaskandthereward.
Changefromrewardtonorewardphasemayproduceatemporarydeclineintimeontasknotalossofintrinsicmotivation.
Changefromnochoiceintherewardphasetochoiceinthefreetimephasemaylowertimeontaskasanexperimentalartifact.
Nodetrimentaleffectwasfoundwithverbalrewards.
Rewardsformerelydoinganactivityshowdetrimentaleffectsoffreetime,butrewardsforeachunitcompleteddonot.
Timeandattitudemeasuresdonotcovarywithrewardmanipulations.

SINGLESUBJECTDESIGNSASSESSINGTHEEFFECTSOFREWARD
FollowingtheworkofDeci(1971,1972b)andLepperandhiscolleagues(Lepper,Greene,andNisbett,1973),someresearchersdesignedstudiesinanattemptto
addresssomeoftheconcernsraisedinthisdiscussion.Oneoftheproblemsintheoriginalexperimentswasthatthepromisedrewardswerenotshowntoactas
reinforcement.Thatis,therewasnodemonstrationthatthe
Page30

promisedrewardsincreasedthefrequencyofpuzzlesolvingordrawingpictures(i.e.,thetargetactivities).Also,theproceduresdidnotinvolvearepeatedpresentation
oftherewardsovertime.CriticsofDecisandLepperandcolleaguesstudieshavenotedthatineverydaylife,rewardsarearrangedanddeliveredoveraconsiderable
lengthoftime.IntheexperimentsbyDeciandbyLepperandassociates,therewardandintrinsicmotivationproceduresinvolvedasingledeliveryofanofferedreward.
Inordertoaddresstheseissues,in1975,FeingoldandMahoneyconductedanexperimentusingasinglesubject,repeatedmeasuresdesign.9Fivesecondgrade
childrenwerestudiedindividually.Thechildrenweregivenaccesstotwoactivities(DottoDotorEtchaSketchgames).Theexperimenterobservedthechildrens
performanceoftheseactivitiesforeight15minutesessions.Followingthesebaselinesessions,eachchildwasindividuallyinformedthathisorherDottoDot
performancewouldsubsequentlyberewardedwithpointsexchangeableforprizes.Theresearchersdescribedtherewardproceduresasfollows:

[Participants]earnedonepointforexceedingtheirhighestbaselineperformanceandanadditionalpointforevery50subsequentdotsconnected.Thus,a
childwhohadcompleted130dotconnectionsduringhishighestbaselinesessionwasrequiredtocomplete131foronepoint,181fortwopoints,andsoon.
Thesecontingencieswerespecifiedbymarkingthe15pointcriteriaineach[participants]bookletpriortoeachsession.Pointswereredeemedona
dailybasisandwerenotaccumulatedacrossdays.Tokenreinforcementprocedureswerecontinuedforfoursessions.10

ThisrewardphasewasfollowedbyaperiodduringwhichtherewardcontingencywaswithdrawnandthechildrensrateofperformingtheDottoDotactivitywas
observedforeightsessions.Twoweeksaftertheexperiment,thechildrenreturnedandwereagainobservedfortensessions.
ResultsindicatedthattherateofDottoDotconnectionsmorethandoubledduringtherewardphaseascomparedtobaseline.Thus,areinforcementeffectwas
demonstrated.Removaloftherewardsfortheactivityresultedinadropinperformancetoalevelequaltotheinitialbaseline.Thus,therewasnodifferencebetween
thepreandpostrewardphases,indicatingnoevidenceofalossofintrinsicmotivation.Forthetwoweeksfollowingtheexperiment,childrencontinuedtoengageinthe
DottoDotactivityatbaselinelevel.Again,therewasnoevidenceofadetrimentaleffectofrewardorreinforcement.Theseresultssuggestthatthereisnodetrimental
effectofrewardwhentherewardsareshowntobereinforcersanddeliveredrepeatedlyandrepeatedassessmentsofintrinsicmotivationaremade.
Page31

FEINGOLDANDMAHONEYS(1975)STUDYONTHEEFFECTSOFREINFORCEMENTONINTRINSICMOTIVATION

Participants
Fivesecondgradechildren

Task:
Makingdottodotconnections

Reward:
Pointsexchangeableforprizes

Procedure:
BaselinephaseeightsessionsontaskwithoutrewardReinforcementphaseeightsessionsontaskwithrewardPostreinforcementphaseeight
sessionsontaskwithoutrewardFollowupphase(twoweekslater)tensessionsontaskwithoutReward

MeasuresofIntrinsicMotivation:
Differencesbetweenpostreinforcementphase,followupphase,andbaselinephaseonrateofperformance

Results:
(1)nodifferences
(2)nonegativeeffectsofreinforcement

AlthoughFeingoldandMahoneysstudyshowedthatrewardswerenotharmful,thefindingshavenotbeenreadilyacceptedoracknowledgedbythosewhoargue
againstrewardsandreinforcement.Criticshavesuggestedthattheresultsofsinglesubjectdesignsarelimitedbecausetoofewparticipantsareinvestigatedandno
rewardcontrolgroupsareomitted.11Instead,thosewhoadvocatethedetrimentaleffectshypothesisrelyonevidencefrombetweengroupdesignstudieslikethose
conductedbyDeci(1971,1972b)andbyLepper,Greene,andNisbett(1973).

SUMMARY
Theviewthatexternalrewardsnegativelyaffectpeoplesintrinsicmotivationhasitsbasisinearlylaboratoryinvestigationsintothisphenomenon.Theearly
Page32

experimentsbyDeci(1971,1972b)andbyLepper,GreeneandNisbett(1973)havebeenheraldedinpsychology,education,business,andthemediaasevidence
thatrewardsandreinforcementaredamagingtobasichumannature.Basedonthisassumption,manyresearchersandwritershavearguedagainstusingrewardsand
positivereinforcementinclassrooms,theworkplace,andotherappliedsettings.
Acloseexaminationoftheearlyexperimentsonexternalrewardsandintrinsicmotivationisrevealing.Allinall,itturnsoutthattheevidencedoesnotsupportthe
claimthatrewardsaregenerallyharmful.Inaddition,thereisnoevidencethattherewardsusedintheinitialstudiesoperatedasreinforcementfortheactivities.Feingold
andMahoney(1975)showedthatwhenrewardswereshowntofunctionasreinforcement,intrinsicmotivationwasnotaffected.IntheinitialstudiesbyDeciandby
Lepper,Greene,andNisbett,intrinsicmotivationwasmeasuredastimespentonataskfollowingtheremovalofapromisedrewardDecialsousedaselfreport
measureoftasklikingtoindexintrinsicmotivation.Theearlystudiesdemonstratedthatexternalrewardshavedifferenteffectsonthesemeasuresasafunctionofthe
typeofreward(verbalortangible),whethertherewardswerepromisedbeforehand(expectedorunexpected),andwhatparticipantswererequiredtodotoobtainthe
reward(rewardcontingency).
Basedontheearlyexperiments,thedetrimentaleffectsofrewardappeartobehighlyrestricted.Therewardproceduresofthesestudieshadnoeffectsonthe
participantsattitudeorinterestintheactivities.Theparticipantswereasinterestedintheactivitiesaftertherewardswerepromisedanddeliveredastheywerebefore.
Intermsoftimespentontheactivityinthefreechoiceperiod,astatisticallyreliabledecreasewasfoundintheLepperandcolleaguesstudy,butanunusual
combinationofconditionswasnecessarytoproducetheeffect.Specifically,anegativeeffectwasfoundwhenthedeliveryofrewardswasstatedbeforehand.Thatis,
therewardswerepromisedtoparticipantsbeforetheydidthetask(expected)unexpectedrewardsdidnotproduceanegativeeffect.Inaddition,therewardswere
materialortangibleverbalpraisedidnotproduceanegativeeffect.Inaddition,therewardswereofferedsimplyfordoingthetaskwithnoperformancerequirement
rewardsofferedtomeetaperformancestandard(solvepuzzles)didnotproduceanegativeeffect.
Anotherconditionnecessarytoproduceadetrimentaleffectwasthattimeontaskhadtobemeasuredaftertherewardswereremovedmeasuringtimeontask
duringtherewardedperioddidnotproduceanegativeeffect.Finally,anegativeeffectwasproducedonlywhentherewardwasdeliveredinasinglesessionfollowed
byaonetimeassessmentoftimeontaskaftertherewardwasremovedrepeatedpresentationoftherewardfollowedbyrepeatedassessmentsofintrinsicmotivation
didnotproduceanegativeeffect.
Page33

IMPLICATIONSOFTHEEARLYRESEARCH
Conditionsnecessarytoproduceanegativeeffectofreward:
Rewardistangibleormaterial
Rewardisofferedbeforehand
Rewardisofferedregardlessoflevelofperformance
Rewardisdeliveredonceoverasinglesession
Intrinsicmotivationismeasuredastimeontaskfollowingthewithdrawalofreward
Intrinsicmotivationisindexedbyasingleassessmentoftimeontaskfollowingtheremovalofreward

Sincethepublicationoftheinitialstudies,anenormousamountofresearchonrewardsandintrinsicmotivationhasbeenconducted.PartIVofthebookprovidesanin
depthanalysisoftheseexperimentsandshowsthattheresultsfromtheearlystudieshavestoodthetestoftime.Thatis,negativeeffectsofrewardremainhighly
circumscribedandoccuronlyundertheunusualconditionsalreadydescribed.Otherrewardprocedureseitherhavenoeffectonmeasuresofintrinsicmotivationor
resultinanincrease.
Rewardandreinforcementarecriticalelementsinthedesignofeffectivelearningandworksystems.Theliteratureonintrinsicmotivationandrewardhasbeenused
toconvinceeducatorsandbusinessmanagersthatrewardsareharmfultohumannature,andconsequently,thetrendhasbeentoshyawayfromtheuseofrewards.
However,thereisnoneedtoavoidtheuseofrewardwhenproperlyarranged,itcanenhanceperformancewithoutalossofinterestorenjoyment.Whatweare
saying,then,isthatrewardsdonotdeserveabadreputation.

NOTES
1.Lepper,Greene,andNisbett,1973.
2.Dickinson,1989.
3.Skinner,1969,pp.114115.
4.Dickinson,1989,p.11.
5.Theproblemis,therefore,toaccountforthenegativeeffectsofpromises,instruc
Page34

tions,anddescriptionsofcontingencies.Onepossibilityisthatvagueinstructionsaboutwhatisrequiredtogetarewardaretakenasattemptsatsocialcontrol.Participantsinan
experimentmaysubvertattemptsofconspicuouscontrolbyunderminingtheexperimentersauthorityspendinglesstimedoingtherequiredactivity.Reactanceagainstsocial
controlratherthanalossofintrinsicmotivationmayexplainthedecreaseintimespentonanactivity.Rewardsthataretiedtospecificperformanceaccomplishmentsmayserveto
acknowledgeapersonsattainmentsandnotevokereactance.
6. Deci,1971.
7. Ibid.
8. Deci,1972b.
9. FeingoldandMahoney,1975.
10.Ibid.,p.370
11.SeeDeci,Koestner,andRyan,1999.
Page35

PARTIII
THEORETICALDISPUTESOVERREWARDSAND
INTRINSICMOTIVATION
Page36

Thispageintentionallyleftblank.
Page37

Chapter4
TheoreticalPerspectivesonRewardsasHarmful
Theinitialstudiesintheearly1970sontheeffectsofrewardsandreinforcementonintrinsicmotivationinstigatedagreatdealofresearchonthetopic.Numerous
experimentshavebeendesignedtotestandgenerateavarietyofhypothesesabouthowrewardsalterpeoplesintrinsicinterestinactivities.Inthissectionofthebook,
weexaminevarioustheoreticalperspectivesontheeffectsofrewards.Thischapterfocusesontheoriesandperspectivesthatemphasizeharmfuleffectsofrewardsand
reinforcement.
Assertionsthatrewardsandreinforcementareharmfultopeoplesintrinsicmotivationcomemainlyfromanareaofsocialpsychologythatembracesphilosophical
notionsofhumannaturethatstressindividualism1andinnateneedsforfreedomandcompetence,whicharesaidtobesuppressedorunderminedbyexternal(social)
constraintsimposedbysociety.2Somewriterswhoadoptthispositionclaimthatallrewards,whethertangibleorverbal,inducecompliancebutunderminethe
motivationtocontinueanactivityoncetherewardsarewithdrawn.3
AnexampleofhowrewardsdestroyintrinsicmotivationispresentedinAlfieKohnsbookPunishedbyRewards.4Kohnpresentsastoryaboutanoldmanwhois
tauntedbyagroupoftenyearoldswhopasshishouseonthewayhomefromschooleachday.Kohndescribesthescene:

Oneafternoon,afterlisteningtoanotherroundofjeersabouthowstupidanduglyandbaldhewas,themancameupwithaplan.Hemetthechildrenon
hislawnthefollowingMondayandannouncedthatanyonewhocamebackthenextdayandyelledrudecommentsabouthimwouldreceiveadollar.
Amazedandexcited,theyshowedupeven
Page38

earlieronTuesday,holleringepithetsforalltheywereworth.Truetohisword,theoldmanambledoutandpaideveryone.Dothesametomorrow,he
toldthem,andyoullgettwentyfivecentsforyourtrouble.ThekidsthoughtthatwasstillprettygoodandturnedoutonWednesdaytotaunthim.At
thefirstcatcall,hewalkedoverwitharollofquartersandagainpaidoffhishecklers.Fromnowon,heannounced,Icangiveyouonlyapennyfor
doingthis.Thekidslookedateachotherindisbelief.Apenny?theyrepeatedscornfully.Forgetit!Andtheynevercamebackagain.5

Althoughthisexampleismeanttobeajoke,itisfrequentlycitedasanillustrationofhowrewardskillintrinsicinterest.Thestoryismeanttoshowthatthechildrenwere
initiallyintrinsicallymotivatedtoscorntheoldman.Oncetheybegantoreceivemoneytoinsulthim,however,theirinterestintauntinghimbecamemotivatedbythe
payratherthanbythefunofit.Whentherewardwasnolongerthere,thechildrenhadnoreasontocontinuetheirgibes.Thepointtobemadeisthattherewards
destroyedthechildrensintrinsicmotivation.
Thosewhoespousetheviewthatrewardsareharmfulsuggestthatthistypeofsituationhappensallthetimeinhomes,educationalsettings,businesses,andother
environments.Thecontentionisthatteachers,parents,andmanagersaredestroyingintrinsicinterestbygivingrewardstotheirstudents,theirchildren,andtheir
employeesforengaginginparticularactivities.
Ahistoricalillustrationofthesupposeddamageofrewardsconcernscraftspersonsandthechangefromfeudalrelationshipstoapayforserviceeconomy.6
Proponentsofthenegativeeffectsofrewardarguethatinfeudaltimes,craftspersonsmadecabinetsandotherartifactsforsheerenjoyment(intrinsicmotives).Withthe
riseofcapitalism,theviewisthatthereasonformakingcabinetschangedfromintrinsictoextrinsicsources(payforservices).Fromthisperspective,contemporary
craftspersonsmakeitemsformoney.Theynolongerengageintheircraftbecausetheyenjoyitinstead,today,theymakecabinetsforthemoney.Thatis,theirintrinsic
motivationtomakecabinetshasbeenunderminedbythepayforservicecapitalistsystem(asystempresumablybasedonprinciplesofreinforcement).
Ofcourse,onecanseeproblemswiththisargument.Allofusknowprofessionalcarpenters,artists,andmusicianswhobuild,paintpictures,orplaymusicwhenthey
arenotbeingpaidtodoso.Manyofusknowprofessionalathleteswhocontinuetoplayandenjoytheirsportaftertheyhaveretired.7Andmanyparentshavewatched
theirchildrenbecomeavidreadersfollowingparticipationinarewardforreadingprogram.
Yetitispreciselythesecontextsthatareofgreatconcerntothosewhoargueagainstrewards.Theviewisthatifwerewardpeopleforengaginginenjoyable
activities,wewillsaptheirintrinsicmotivation.Fromthisposition,oncethereisnolongerarewardfordoingso,thechildwillnotread,theartistwillnotpaint,the
athletewillnotplay,andthecarpenterwillnotbuild.Oureverydayobservationsofbehaviordonotsupportthisview,nordotheresearchfindings.Nonetheless,
severalpsychologistsandeducatorsvehementlyarguethatpayand
Page39

otherrewardshavelastingnegativeeffectsonpeoplesmotivationinthesecontexts.Infact,avarietyoftheoriesandhypotheseshavebeenformulatedtoexplainhow
andwhyrewardshavenegativeeffects.Currently,themostpopularaccountsofthepresumednegativeeffectofrewardsonintrinsicmotivationcomefromcognitive
evaluationtheoryandfromanattributionapproachtorewardsandotherexternalinfluences.
Inthischapter,wedescribeacognitiveevaluationapproachtointrinsicmotivation,outlineitsclaimsintermsoftheunderminingeffectsofrewardcontingencies,and
criticallyevaluateitsmerits.Weendthischapterwithabriefconsiderationofattributionapproachestorewardsandintrinsicmotivation.

COGNITIVEEVALUATIONTHEORY

OneexplanationfortheexistenceofdecrementaleffectsofrewardonintrinsicmotivationcomesfromDeciandRyanscognitiveevaluationtheory.8Inorderto
understandthetheory,itisinformativetofirstexaminehowtheconceptofintrinsicmotivationisunderstoodfromthisperspective.Theclaimisthatpeoplederiveinnate
satisfactionfromcertainactivities.Thisinnatefulfillmentissaidtomotivatepeopletoengage,andsustaininterest,intheseactivities.Incognitiveevaluationtheory,this
innateenergizingforceiscalledintrinsicmotivation.Deciandhiscolleaguesdiscussedtheconceptinarecentarticle:

Intrinsicmotivationenergizesandsustainsactivitiesthroughthespontaneoussatisfactioninherentineffectivevolitionalaction.Itismanifestinbehaviors
suchasplay,exploration,andchallengeseekingthatpeopledofornoexternalrewards.Itisthusaprototypicinstanceofhumanfreedomorautonomyin
thatpeopleengageinsuchactivitywithafullsenseofwillingnessandvolition.9

Accordingtocognitiveevaluationtheory,intrinsicallymotivatedactionsareonesthatevokeahighdegreeoflikingandinterestbutforwhichthereisnorewardbeyond
theactivityitself.
Cognitiveevaluationtheoryproposesthatintrinsicmotivationspringsfromtwoinnatesources(theneedforcompetenceandtheneedforselfdetermination).10On
thebasisoftheirhumanness,peoplestrivetofeelcompetentandfreefromexternalcontrol.Thus,intrinsicmotivationarisesfromactivitiesthatleadpeopletofulfilltheir
needsforcompetenceandselfdetermination.Whenpeopleengageinactivitiessuchaspuzzlesolving,theyaresaidtofeelfreeandhighlycompetentthese
perceptions,inturn,activatetheinnateenergyofintrinsicmotivation.Theenergyofintrinsicmotivation,accordingtocognitiveevaluationtheory,heightensinterestand
sustainsinvolvementinanactivity.Extrinsicallymotivatedbehaviors,ontheotherhand,refertoactionsthataredirectlylinkedtoanexternalcausesuchasanexplicit
reward,incentive,orthreat.Extrinsicallymotivatedactionsaresaidtobecharacterized
Page40

bypressureandtensionandtoresultinalossofperceivedcompetenceandpersonalfreedom.
Althoughintrinsicandextrinsicmotivationarealluringnotionsandbasictocognitiveevaluationtheory,numerousresearchersandtheoristshavepointedout
difficultieswiththeconcepts.Oneissueisthatmanybehaviorsthatappeartobeintrinsicallymotivatedareinfactmotivatedbyanticipatedfuturebenefits,previous
environmentbehaviorinteractions,andthephysicalandsocialcontexts(e.g.,culturalpatternsandpractices).11
Anotherissueconcernstheclaimthatintrinsicmotivationinvolvesaninnateenergysource.Peoplearesaidtobeintrinsicallymotivatedwhentheydrawpictures,
writeorplaymusic,solvecrosswordpuzzles,orplaygamesorsports.Butarechildrenbornwithaninterestintheseactivities?Clearly,theanswerisno.Learning
experiencesarenecessarytobuildinterestandenjoymentofactivities.Almostanyactivitycanbecomeinterestingandenjoyablewheninstructionandrewardsare
initiallyusedtopromoteengagementinit.Thetypesofactivitiesthatchildrenfindenjoyablevarywithcultureandsociallearning.Forexample,intheUnitedStates,
childrenplaybaseball,whereasCanadianchildrenplayhockeyandEuropeanchildrenplaysoccer.Thepointisthatchildrenarenotbornwithaninnatemotivational
forcethatdrivesthemtoengageinthesegames.Thesourceoftheirinterestandenjoymentliesintheircultureandsociallearninghistory,notinaninnateenergy
source.12
Afurtherdifficultyisthatexplanationsintermsofintrinsicmotivationinvolvecircularreasoning.Toillustrate,considerthefollowingsituation.Youhaveafriend,Joe,
whogoesouttohislakesidecabineachweekendandspendshistimedrawingpictures.WhenaskedwhyJoedrawspicturesontheweekends,apsychologiststates
thatJoeisintrinsicallymotivatedtodraw.ButhowdoesoneknowthatJoeisintrinsicallymotivated?TheanswerfromthepsychologististhatJoespendsmuchofhis
timedrawingand,becausetherearenoobviousexternalreasonsforhimtodothis,hemustbeintrinsicallymotivated.Inotherwords,Joesintrinsicmotivationis
inferredfromhisbehavior(timespentdrawingatthelake)itisthenusedtoexplainthesamebehavior(Joedrawsbecauseheisintrinsicallymotivated).Because
intrinsicmotivationisinferredfromtheverybehavioritissaidtocause,theconceptinvolvescircularreasoningandlacksexplanatorypower.13
Page41

DIFFICULTIESWITHTHECONCEPTOFINTRINSICMOTIVATION

Behaviorsaidtobeintrinsicallymotivationisoftendueto:
(a)anticipatedbenefits
(b)previousbehaviorenvironmentinteractions
(c)culturalcontext
Learningexperiencesarenecessarytobuildinterestandenjoymentofactivities.
Theconceptofintrinsicmotivationinvolvescircularreasoning.
Circularreasoningmeansthattheconceptlacksexplanatorypower.

Clearly,intrinsicmotivationisnotaclearcutconcept.Inspiteofconceptualdifficultieswiththeconcept,cognitiveevaluationtheoristscontinuetocontrastintrinsic
motivationwithextrinsicmotivation.Ofmajorconcernishowextrinsicrewardsacttoharmonesintrinsicinterests.

CognitiveEvaluation,Rewards,andIntrinsicMotivation
BuildingontheearlyworkofsocialpsychologistsF.HeiderandR.DeCharms,DeciandRyanarguedthattheeffectsofextrinsicrewardsonintrinsicmotivation
dependonanindividualsinterpretation(i.e.,cognitiveevaluation).14Peoplewhoreceivearewardaresaidtointerprettherewardinrelationtotheirfeelingsofself
determinationandcompetence.Whenrewardsareinterpretedascontrollersofbehavior,theyinterferewithinnateneedsforfreedomandunderminecompetenceby
shiftingperceptionsofcausalityfrominternaltoexternalsources,witharesultinglossofintrinsicmotivation.Ontheotherhand,whenrewardsareinterpretedas
indicatorsofcompetence,theseeventscanhavepositiveeffectsonintrinsicmotivation.Thatis,whenrewardsprovidepositiveinformationaboutperformance
accomplishments,theyareexpectedtosatisfytheneedforcompetenceandtherebyincreaseapersonsintrinsicmotivation.
Cognitiveevaluationtheorypresumesthatrewardshavetwoconflictingeffects:controllingandinformational.Thedegreeofloss(orenhancement)ofintrinsic
motivationdependsonapersonsinterpretationoftherewardintermsofcontrolversusinformation.Iftherewardisinterpretedasmorecontrolling(undermining
autonomy)thanpositivelyinformational(indicatingcompetence),theneteffectisalossofintrinsicmotivation.Rewardsthatconveystrongpositiveinformationabout
competencebutwithsomedegreeofperceivedcontrolcanhavepositiveneteffectsonintrinsicmotivation.
Thus,fromacognitiveevaluationperspective,rewardscanleadtoareductionorenhancementinintrinsicmotivation,dependingontheircontrollingandin
Page42

formationalvalues.Giventhatcognitiveevaluationtheorypredictsbothpositiveandnegativeeffectsofreward,itisnoteworthythatDeciandhisassociatescontinueto
predictandfindmainlynegativeeffectsofvariousrewardcontingencies.15AsweshallseeinPartIVofthebook,thebulkoftheresearchcomingfromthecognitive
evaluationtheoristshasemphasizedthenegativeeffectsofreward.Thatis,overthepast30years,mostoftheresearchonrewardsandintrinsicmotivationhasbeen
explicitlydesignedtofindnegativeeffects.Positiveinformationaleffectsofrewardsaredownplayed,andcognitiveevaluationtheoristshighlightthecontrollingaspects
ofrewards.Bydoingso,theyoftenleavereadersofthevastexperimentalliteratureonrewardsandintrinsicmotivationwiththeimpressionthatallrewardsareharmful.

PredictionsabouttheEffectsofRewardonIntrinsicMotivation
Since1971,whenDecifirstproposedcognitiveevaluationtheorytoaccountfornegativeeffectsofreward,thetheoryhasbeenreformulatedandrevisedmanytimes.
Arecentstatementofcognitiveevaluationtheoryappearedina1999articleinPsychologicalBulletin,inwhich,Deciandhiscolleaguesoutlinedtheircurrentposition
regardingcognitiveevaluationtheoryandmadepredictionsabouttheeffectsofrewardsonintrinsicmotivation.16
Cognitiveevaluationtheoryproposesthatintrinsicmotivationisalteredbyfeelingsofcompetenceandselfdetermination.Eventsthatincreasepeoplesbeliefsthat
theyareskilledinperformingataskorthattheirperformanceisbasedonpersonalpreferenceareassumedtoenhanceintrinsicmotivation.Eventsthatdecreasea
personsperceptionsofcompetenceorselfdeterminationwilldiminishintrinsicmotivation.Fromthisperspective,itisarguedthattangiblerewards(prizes,money,
awards,medals,goldstars,andsoon)elicitthestrongestperceptionsofexternalcontrol.

TangibleRewardsandIntrinsicMotivation
Accordingtocognitiveevaluationtheory,tangiblerewardsaremostharmfulwhentheyareexpected,thatis,offeredtoindividualspriortotheirengagementinactivities.
Iftherewardsareunexpected,theyarenotpredictedtoaffectintrinsicmotivation.Theexplanationforthisisthatwhentherewardsarenotofferedbeforehand,people
donotexperienceaconnectionbetweenthetaskandthereward.Therefore,therewardcannotbeinterpretedascontrollingandcannotundermineperceptionsof
autonomyandintrinsicmotivation.Unexpectedrewardsarealsosaidtohavenoimpactonintrinsicmotivationbecausethecognitiveevaluationprocessonlytakes
placewhentherewardedactivityisinprogress.Whenrewardsarepresentedunexpectedly(withoutannouncement),cognitiveevaluationoftherewardedactivityisnot
activatedandnolossofintrinsicmotivationisexpectedtooccur.
Page43

Withexpectedtangiblerewards,however,individualsareabletomakeaconnectionbetweentheofferofrewardandtheactivityforwhichtheyarebeingrewarded.
Therewardsaresaidtobeexperiencedascontrolling,perceptionsofautonomyandselfdeterminationarereduced,andtheresultisadeclineinintrinsicmotivation.
Astheresearchonrewardsandintrinsicmotivationwasamassedovertheyears,notallstudiesfoundnegativeeffectsfromexpectedtangiblerewards.Thus,
cognitiveevaluationtheorywasalteredanumberoftimestomakeitconsistentwiththeevidence.Inthemostrecentstatementofcognitiveevaluationtheory,the
precisepredictionswithregardtoexpectedtangiblerewardaresaidtodependonthemannerinwhichspecificrewardcontingenciesaffectperceptionsofboth
competenceandselfdetermination.
Cognitiveevaluationtheorystipulatesfourtypesofexpectedtangiblereward:tasknoncontingentrewards,engagementcontingentrewards,completioncontingent
rewards,andperformancecontingentrewards.Tasknoncontingentrewardsarerewardsofferedregardlessofanyinvolvementinanactivity.Engagementcontingent
rewardsareofferedsimplyfordoingataskwithoutarequirementtocompleteit.Completioncontingentrewardsareofferedfordoingandcompletingtheassigned
task,andperformancecontingentrewardsarethoseofferedforperformingwell,meetingorsurpassingaspecifiedcriterion,orachievingastandardofperformance
(forexample,betterthan80percentoftheotherparticipants).
Cognitiveevaluationtheoryissaidtoconsiderthecontrollingversusinformationalaspectsofdifferentrewardcontingenciesinordertopredictwhetherthe
contingencywillundermineintrinsicmotivation.Withtasknoncontingentrewards,thereisnocontingencybetweentherewardandthetaskthus,cognitiveevaluation
theorydoesnotpredictnegativeeffectsonintrinsicmotivation.Deciandhisassociatesarguethatengagementcontingentrewards(thoseofferedforworkingonatask
withnorequirementtocompleteit)arelikelytobeexperiencedascontrollingengagementcontingentrewardsshouldactprimarilytoreduceperceptionsofself
determinationandtherebyreduceintrinsicmotivation.Engagementcontingentrewardsaresaidtocarrylittleornoneofthecompetenceinformationthatcanoverride
theperceivedcontrolbyrewards.Thus,rewardsofferedmerelyfordoing(butnotcompleting)ataskaresaidtohaveanetnegativeeffectonintrinsicmotivation.The
majorityofresearchstudiesonrewardsandintrinsicmotivationhavebeenconductedusingthiscontingency.
Completioncontingentrewardsrequirepeopletocompleteataskinordertoobtainthereward.Thecontingencybetweencompletingthetaskandreceiving
rewardsissaidtobeexperiencedasmorecontrollingthanforrewardsgivensimplyfordoingthetask(engagementcontingent),althoughtheexactdetailsofthe
evaluationprocessarenotgiven.Ontheotherhand,completioncontingentrewardsaresaidtoconveysomelevelofcompetence,especiallyforskilledtasks.The
positiveinformationforcompetenceisclaimedtooffset
Page44

someofthecontrollingaspectsofcompletioncontingentrewards,sotheserewardsshouldhavelessofanunderminingeffectonintrinsicmotivationthanengagement
contingentrewards.
Afinaltypeofrewardcontingencyaddressedbycognitiveevaluationtheoryiscalledperformancecontingent(rewardsfordoingwellatatask,meetingastandard,
orperformingtoasetcriterion).Performancecontingentrewardsaresaidtolessenselfdeterminationbecausetherecipientinterpretstheuseoftherewardasan
attemptatbehavioralcontrol.Atthesametime,performancecontingentrewardscanincreaseonesperceptionsofcompetencebyprovidinginformationconcerning
ability.Deci,Koestner,andRyan(1999)concludedthatthedecrementinperceivedselfdeterminationwillhaveastrongereffectthantheincrementinperceived
competenceandthattheneteffectwillbeanunderminingofintrinsicmotivation.Theyfurtherpredictedthatthenegativeeffectofperformancecontingentrewards
shouldbelessthanthoseofengagementorcompletioncontingentrewards.Theywentontostatethattheeffectsofperformancecontingentrewardswillbemore
variableandwilldependonadditionalfactors(e.g.,interpersonalcontextforadministeringtherewards).17

VerbalRewardsandIntrinsicMotivation
InDecisinitialresearchonrewardsandintrinsicmotivation,verbalrewards(e.g.,praiseandpositivefeedback)werefoundtoresultinpositiveeffectsonmeasuresof
intrinsicmotivation.18Thatis,peoplewhoreceivedverbalrewardsspentmorefreetimeontheactivityandshowedincreasedinterest.19Accordingtocognitive
evaluationtheory,verbalrewardsaregenerallysaidtoconveypositiveinformationaboutcompetencethatoverridesthecontrollingcomponentofrewards.Inthiscase,
cognitiveevaluationtheoristsproposeanetpositiveeffectofverbalrewardsonintrinsicmotivation.20
Althoughcognitiveevaluationtheoristsacknowledgethebeneficialeffectsofverbalrewards,theytendtodismissthepositiveimpactandinsteademphasizepossible
negativeeffectsoftheserewards.Intheirrecentstatementofcognitiveevaluationtheory,Deciandhisassociatesconcededthepositiveeffectsofverbalrewardsbut
thenprovideawarning:

AccordingtoCET[cognitiveevaluationtheory],theinformationalaspectofverbalrewardsisgenerallyexpectedtobesalientandthusverbalrewardsare
generallypredictedtoenhanceintrinsicmotivation.However,verbalrewardscanalsohaveasignificantcontrollingcomponentthatis,peoplesometimes
engageinbehaviorstogainacknowledgmentorapprovalsoverbalrewardscanalsohavethepotentialtoundermineintrinsicmotivation.CETsuggests
that,justaswithtangiblerewards,theinterpersonalcontextwithinwhichpositivefeedbackisadministeredinfluenceshowitisinterpretedandthuswhat
effectithas.21

Thisreasoningisdifficulttofollow.Theauthorsaretellingusisthatverbalrewardsusuallyconveyhighcompetencetopeoplebutcanalsohaveasignif
Page45

icantcontrollingcomponent.Whatismeantbythisisthattherearewaystoadministerverbalrewardsthatarecoerciveandthatresultinalossofintrinsicmotivation.
Onewaytoreduceintrinsicmotivationbyverbalrewardsistohaveanauthorityfigure(theexperimenter)saysomethinglike,Excellent,youshouldkeepupthe
goodwork,asopposedtosayingthesamestatementwithoutthewordshouldorthepressuretobehave.Underconditionsofsocialpressure,peopleshowalossof
intrinsicmotivation.Thislossofintrinsicmotivation,however,cannotbeattributedtotheuseofverbalrewardrather,itisthecoerciveimplications(impliedthreat)of
thesocialpressurethatlowersmotivation.Equatingverbalrewardswithimpliedthreatsisconfusingandleadstomisconceptionsaboutthisformofreward.Teachers,
parents,andadministratorsmaydrawtheconclusionthatverbalrewardsareoftenharmfulwhen,infact,itistheuseofthreatsorothercoercivestrategiesthatisat
issue.22

RewardsandInterpersonalContext
Cognitiveevaluationtheoristsstatethattheeffectsofrewards(bothverbalandtangible)dependontheinterpersonalcontext.Theinterpersonalcontext,accordingto
cognitiveevaluationtheory,referstothesocialambianceofsettingssuchashomes,classrooms,orworkgroupsastheyinfluencepeoplesexperienceofautonomy,
competence,orrelatedness.23
Themostimportantaspectofinterpersonalcontextiswhethertheambiance(mood,feel,oratmosphere)isperceivedascontrolling.Whentheinterpersonalcontext
issetuptoreflectcontrolorpressure,cognitiveevaluationtheorystatesthatperformancecontingentrewards(andverbalrewards)willbeviewedascontrollingand
willundermineintrinsicmotivation.Rewardsthatareadministeredinanoncontrollingstylearesaidtobeexperiencedasmoreinformationalandleadtolessofalossof
intrinsicmotivation.
Inastudypublishedin1982,Ryancomparedperformancecontingentrewardsadministeredinacontrollingmannerwiththesamerewardsadministeredwithout
pressure.Astrongunderminingeffectwasreportedonlywhenpeoplewerepressuredtoperformandofferedarewardfortheirperformance.24Thisfindingsuggests
thatperformancecontingentrewardshavenodirectunderminingeffectonintrinsicmotivation.Thatis,performancecontingentrewards,perse,arenotharmful.Itis
theinterpersonalcontextthatproducespositiveornegativeeffects.
Theimplicationofthisfindingisthatperformancecontingentrewards(andverbalrewards)taketheirmeaningfromtheinterpersonalcontext.Whenpeopleare
pressured,givingthemrewardsforperformanceleadstoalossofinterestinthetargetactivity(recallthatsocialpressurecanevenalterthepositiveeffectsofverbal
rewards).Thisisnotthecasewhenrewardsaregivenwithoutpressure.Thus,theeffectsofarewardproceduredependonthestyleofadministration.Inotherwords,
cognitiveevaluationtheorypredictsalossofintrinsicmotivationonlywhenrewardsarepresentedinanauthoritariansupervisorystyle(usingpressuretogetpeopleto
perform).Theimplicationofthispredictionisthat
Page46

evenfromacognitiveevaluationperspective,rewardsystemshavenoinherentnegativeeffectsitistheaversivecontroltacticsofsupervisorsandadministratorsthat
areatissue.

RewardsandInitialTaskInterest
Deciandassociateshaveindicatedthatthefieldofresearchonrewardsandintrinsicmotivationisconcernedwithactivitiesthatpeoplefindintrinsicallymotivating(e.g.,
solvingpuzzles).25Ofcourse,notallactivitiesareinitiallyinteresting(i.e.,intrinsicallymotivating).Manyacademicskillsandperformancesholdnoinitialinterest,and
otherroutinesmaybedullandboring(e.g.,completingpagesofadditionandsubtractionproblems).However,accordingtoDeciandhiscolleagues,thecognitive
evaluationprocessisonlyoperativeforactivitiesthatarehighoninitialinterest.Theauthorsstated:

CETwasexplicitlyformulatedtoexplaintheeffectsofexternaleventssuchasrewardsonintrinsicmotivationforinterestingtasks.Ifanactivity
stimulateslittleornoinitialintrinsicmotivation,onewouldnotexpecttofindexternaleventsunderminingintrinsicmotivation,andtheprinciplesofCET
wouldnotapply.Instead,withboringtasks,thecriticaltheoreticalissueishowtofacilitateinternalizationoftheregulationofsuchtasksiftheyare
deemedimportantwithinthesocialenvironment....Thetheoreticalprinciplesrelevanttomotivationfordull,boringtasksaredifferentfromthose
relevanttomotivationforinterestingtasks.26

Thus,cognitiveevaluationtheoryappliesonlytoinitiallyinterestingtasksthatareintrinsicallymotivating.Thetheoryhasnothingtosayaboutactivitiesthatlackinitial
interest.
Althoughthisviewsoundsstraightforward,ananalysisofinitialinterestrevealsseveralproblemsforcognitiveevaluationtheory.Considerfirstthefactthatnotall
activitiesareabsolutelydullorabsolutelyinteresting.Itismoreusefultothinkofactivitiesasvaryingindegreeofinitialinterest.Iftasksvaryonacontinuumofinterest
(fromdulltoextremelyinteresting),thiscreatesaproblemofscope(i.e.,boundaryconditions)forcognitiveevaluationtheory.Atwhatpointdoesincreasinginterest
becomeindicativeofintrinsicmotivation,sothatcognitiveevaluationtheoryapplies?Inotherwords,howarewetoknowwhenrewardingpeoplefordoingataskof
someinitialinterestwillactivatethecognitiveevaluationprocess(reducingintrinsicmotivation)andwhenitwillnot?Theproblemisthatthereisnoanswertothis
questionfromcognitiveevaluationresearchers.Allthatwecanfallbackonisthatcognitiveevaluationtheoryworksbestfortasksofhighinitialinterest.
Thisfallback,however,leadstomoreproblemsandquestions.Forinstance,whatconstituteshighinitialinterestandintrinsicmotivation?Doesanactivitypossessan
absoluteamountofinterestorisinterestarelativeconcept?Consideranactivitysuchassolvingcrosswordpuzzlesandtheusualmeasureofintrinsicmotivation(or
interest)basedontimespentontheactivity.Ifthisactivityis
Page47

presentedbyitself,peoplemayspendmuchoftheirtimeonitinabsoluteterms,accordingtocognitiveevaluationtheory,thiswouldbeataskofhighinitialinterestand
intrinsicmotivation.
Butnowconsiderasituationinvolvingseveraloptionaltasksaswellascrosswordpuzzles.Whenfacedwithachoiceamongcrosswordpuzzles,computergames,
andbooksormagazines,peoplemayspendmoretimeoncomputersandbooksthanoncrosswordpuzzles.Inarelativesense,crosswordpuzzlesarelowininitial
interest.Ifthisisso,istheactivityofsolvingpuzzlesnolongerintrinsicallymotivating?Bychangingthenumberofoptionaltasks,wecanmaketherelativetimespenton
crosswordpuzzlesriseandfall.Inonecontextataskisinteresting,whereasinanotheritisnot(orlessso).Theseconceptualproblemsaboutinitialinterestmeanthatit
isimpossibletoknowwhenataskishighininitialinterestandwhentoapplycognitiveevaluationtheory.
Deciandcolleaguesconcedethatperformanceofdulltasksmayrequiretheinternalizationofbehaviorregulationbasedonrequirementssetbythesocial
environment(i.e.,sourcesofexternalcontrol).Presumably,rewardsandincentiveswouldbeanessentialaspectoftheexternalinfluencesthatgenerateinternalized
regulationofbehaviorhowever,thisisnotaddressedbyDeciandhiscolleagues.Wesuggestthatduringtheprocessofinternalization,dullorneutraltasksbecome
moreinteresting.Basedonthisprocess,whenpeopleexpressinterestincertainactivities,theirinterestmayreflectpastexperiencewithrewardsratherthanthe
fulfillmentofinnateneedsforcompetenceandselfdetermination.Inotherwords,individualsmaybeinterestedinavarietyofactivitiesforwhichtheywerenotinitially
intrinsicallymotivated.
Accordingtocognitiveevaluationtheory,onlythoseinitiallyinterestingtasksthatpossessintrinsicmotivation(bysatisfyingtheinnateneedsforcompetenceand
autonomy)aresaidtobesusceptibletotheunderminingeffectsofreward.Othertasksinwhichinitialinterestreflectsapasthistoryofrewardwouldnotbeexpectedto
showanunderminingeffectofrewardbecauseinsuchtasks,thereisnointrinsicmotivationtoundermine.Becausetheonlywaywecanknowifsomeoneisinterested
inataskisbyobservingthetimespentonthetaskorbyaskingthepersonabouthisorherattitudetowardtheactivity,cognitiveevaluationtheoryisfacedwitha
problem.
Inordertomakeuseofcognitiveevaluationtheoryanditspredictionsaboutrewards,itisnecessarytoknowifaninitiallyinterestingtaskreflectsintrinsicmotivation
orapasthistoryofrewards.Unfortunately,cognitiveevaluationtheoristsdonotprovideamethodtodistinguishthesealternativesourcesofinterest,andtherefore
cannotpredictwhichinterestingtaskswillshowtheunderminingeffectsofrewards.Instead,cognitiveevaluationtheoristsrelyonselfreportsoftaskinterestandtime
ontaskasindicationsthatpeopleareintrinsicallymotivated.Ifindividualsreporthightaskinterestandspendtimeontheactivity,theassumptionisthattheyare
intrinsicallymotivatedandthatundertheseconditions,rewardswillaffectthecognitiveevaluationprocess.
Page48

Table4.1
PredictionsAccordingtoCognitiveEvaluationTheoryabouttheEffectsofRewardonIntrinsicMotivation
HIGHINTERESTTASKS
Rewardcondition Predictedchangeinperceptionsof Predictedchangeinperceptionsof Predictedchangeinintrinsic
competence selfdetermination motivation
Verbal Increase Increase INCREASE
Tangible
Unexpected Nochange Nochange NOCHANGE
Expected
Tasknoncontingent Nochange Nochange NOCHANGE
Engagementcontingent Decrease Decrease DECREASE
Completioncontingent Smallincrease Largedecrease DECREASE
Performance Smallincrease Largedecrease DECREASE
contingent

SummaryofPredictionsMadebyCognitiveEvaluationTheory
Since1971,numerousexperimentshavebeenconductedtotestandgeneratethepredictionsofcognitiveevaluationtheory.Thespecificpredictionsofthetheoryare
summarizedinthissection.
First,cognitiveevaluationtheoryconcernstheeffectsofrewardsonactivitiesofhighinitialintrinsicinterest.Cognitiveevaluationtheorymakesnopredictionsabout
theeffectsofrewardonlowinteresttasksas,fromthisperspective,thereisnointrinsicmotivationtoalter.Intermsofhighinteresttasks,generalpredictionsofthe
theoryarepresentedinTable4.1.Rewardsaresaidtobeexperiencedascontrollingorinformational,andchangesinintrinsicmotivationcomeaboutfromchangesin
perceptionsofcompetenceandselfdetermination.Table4.1presentscognitiveevaluationtheoryspredictionsabouthowrewardswillaffectfeelingsofcompetence
andselfdeterminationaswellaspredictedchangesinintrinsicmotivation.
AninspectionofTable4.1showsthatnegativeeffectsofrewardarepredictedwhentherewardsaretangible,expected(offeredbeforehand),andengagement
contingent(offeredfordoingatask),completioncontingent(offeredforcompletingatask),andperformancecontingent(offeredfordoingwellorforsurpassinga
performancestandard).Ineachofthesesituations,thepredictionisthatparticipantswhoreceivearewardwillfeelcontrolledbytherewardandtheirfeelingsofself
determinationandautonomywilldecline.Inthecaseofcompletioncontingentandperformancecontingentrewards,however,thereceiptofrewardmaysignala
degreeofcompetence.Nonetheless,thecontrollingaspectoftheserewardsisexpectedtobestrongerthanthecompetenceinformationandintrinsicmotivationis
predictedtodecrease.
Table4.1indicatesthatverbalrewardsarepredictedtoleadtoanincreaseinintrinsicmotivation.However,verbalrewardscanreduceintrinsicinterestwhen
Page49

theyareadministeredinacoercivecontext.Underconditionsofthreatandsocialpressure,accordingtocognitiveevaluationtheory,verbalrewardswilldecrease
feelingsofcompetenceandselfdeterminationwillbereduced.
Afurtherqualificationofthetheoryregardstheeffectsofexpected,tangible,performancecontingentrewards.Again,aswithverbalrewards,performance
contingentrewardscanhaveoppositeeffectstothosepredictedinTable4.1.Performancecontingentrewardsareonlypredictedtobeharmfulwhentheyare
presentedinanauthoritariansupervisorycontext.Whensuchrewardsaregivenwithoutpressure,however,negativeeffectsarenotpredicted.
Althoughtheeffectofrewardsondifferentagegroupshasnotbeenacriticalcomponentofcognitiveevaluationtheory,Deciandhiscolleaguesrecentlymadesome
newpropositionswithregardtochildrenversusadults.Specifically,theysuggestthatalltypesofrewardsmaybemoreunderminingforchildren.AccordingtoDeciand
associates,thisisbecausechildrenfrequentlyexperienceadultsascontrollingtheirbehaviorinaddition,childrenarelessabletoseparatethecontrollingand
informationalaspectsofrewards.
Insum,cognitiveevaluationtheorypredictsbothpositiveandnegativeeffectsofrewards.However,themajorfocusofthetheoryisonharmfuleffectsofreward
and,asweshallseeinPartIVofthebook,themajorityofexperimentsconductedbycognitiveevaluationtheoristshavebeendesignedtodetectnegativeeffects.

ACRITICALASSESSMENTOFCOGNITIVEEVALUATIONTHEORY
Cognitiveevaluationtheoryhasprovidedaconceptualframeworkthathasguidedaconsiderableamountofresearchontheeffectsofrewardsonpeoplesintrinsic
motivation.However,majorambiguitiesexistinthisframework.Ouroverallassessmentofcognitiveevaluationtheoryisthatitlackssufficientdetailsonanumberof
theoreticalissues.Thislackofspecificationoftenmeansthatthereislowpredictiveutilitytothetheory.Withlowpredictivepower,thetheoryisdifficulttodisconfirm
byscientificevidenceandisoftenappliedtoexperimentalfindingsinanafterthefactmanner.
Oneproblemisthatitisdifficulttoknowhowcontrollingorinformationalarewardcontingencyisbeforeitseffectsareknown.Thatis,ifarewardprocedureresults
inlesstimeonthetaskinthefreechoiceperiod,thecontingencyissaidtobemorecontrollingthaninformational.Thus,cognitiveevaluationtheoryisoflimited
predictivepowerbecausetheprocessofcognitiveweightingofarewardcontingencyisnotwellspecified.Howmuchcontrolorinformationalperceptionsdoesagiven
rewardcontingencyactivate?Thereisnothinginthetheorytoprovidetheanswer.Further,howdoperceptionsofcontrolcombinewithperceptionsofcompetenceto
yieldtheneteffectsofarewardcontingencyonintrinsicmotivation?Areperceptionsofcontrolandcompetencecombinedadditively,averaged,orbroughttogetherin
someotherway?These
Page50

issuesarenotresolved,eveninrecentstatementsofcognitiveevaluationtheory.Instead,theinformationalandcontrollingaspectsofrewardareinferredfromtheir
effects.
AsecondissueisthatitisdifficulttoknowhowDeciandhisassociatesarrivedattheneteffectsofdifferentrewardcontingencies.Forexample,completion
contingentandperformancecontingentrewardsaresaidtohavesomecontrollingaspectsandsomepositiveinformationalaspects.Buthowmuchcontrollingversus
informationalvaluedotheserewardproceduresactivate?Couldthesetypesofrewardsactivatehighperceptionsofcompetencethatoverridethepresumedcontrolling
effectofthesecontingencies?Ifso,completioncontingentandperformancecontingentrewardsshouldhavenetpositiveeffectsonintrinsicmotivation.Overall,itisnot
clearhowcognitiveevaluationtheoryclaimsanetnegativeeffectofcompletioncontingentandperformancecontingentrewards.
Afurtherconcerninvolvingbothcompletioncontingentandperformancecontingentrewardsisthatsomepeopleinthesesituationsreceivelessthanmaximal
reward.27Forexample,apersonwhoisrequiredtocompletepuzzlesformoneymaysolveonlyhalftheproblemsandthusreceivehalftheavailablemoney.Inthis
case,lessthanmaximumrewardindicatesafailuretosolvesomeoftheproblemsandlowcompetenceatthetask.Thus,theunderminingeffectsofcompletion
contingentandperformancecontingentrewards(predictedbycognitiveevaluationtheory)maybeduetofailurefeedbackandderogationofcompetence,andnotto
thecontrollingnatureoftheserewards.Thisissueisnotaddressedincognitiveevaluationtheory.
Ourpointhereisthatfailurefeedbackshouldnotbeequatedwitharewardsystem.Inbusinesses,schools,andhomes,rewardscanbearrangedtoencourage
behaviorandgenerateinterestwithoutindicatingfailure.Whenrewardsarearrangedforsuccessiveapproximationstoskillfulperformanceorarebasedongraduated
stepsfromoneskillfulperformancetoanother,theyareprogrammedforsuccessandnotforfailure.Rewardstiedtosuccess,gradualchallenge,andmasterycan
enhancemeasuresofintrinsicmotivation.28Unfortunately,thesepositiverewardprocedureshavenotbeeninvestigatedbysocialpsychologistswhoinsteadpromote
theviewthatrewardsareinherentlydamaging.
Cognitiveevaluationtheoryclaimsthatpeopleevaluaterewards,weighingthecontrollingaspectsofrewardsagainstthecompetencyinformationthattheycontain.In
thisweighingofinformation,cognitiveevaluationresearcherstypicallypredictthatrewardswillhavepervasivenegativeeffects.However,itisclearfromthetheorythat
rewardscanhavepositiveeffectswhentherewardproceduresconveyhighcompetencewithlowperceivedcontrol.Todate,cognitiveevaluationresearcherscontinue
todenouncetheuseofrewards,eventhoughthisisnotarequirementoftheirtheory.
Aproblematicaspectofcognitiveevaluationtheoryisthenotionofintrinsicmotivation.Cognitiveevaluationtheoristsimplythatpeoplemustbeintrinsicallymotivated
toengageincertainactivities.However,taskinterestisnotabsolute
Page51

instead,itvariesonacontinuum.Inaddition,therearemanytasksthattakeagreatdealofskillandinstruction(e.g.,learningtoplaythepiano,learningtoread,etc.)
andareoflittleinitialinteresttoindividuals.Cognitiveevaluationtheorydoesnotaddresstheissueofhowpeoplebecomeinterestedinsuchtasks.Mostactivities
peopleengageinwerenotinitiallyinterestingtothem.Becausecognitiveevaluationtheoryrestrictsitsclaimstotheconceptofinnateintrinsicmotivation,ithasverylittle
tosayaboutwhatactuallygoesonineverydaysettings.
Afinalconcernisthatcognitiveevaluationtheorydoesnotconsiderdifferencesinsociallearningandacculturationthatmayeliminateorreducethedecremental
effectsofrewardsonselfdetermination.Cognitiveevaluationtheoryincorporatestheviewofphilosophicalromanticismthatselfdeterminationisaninherenthuman
motive.29Currently,however,thereisnoevidencethatselfdeterminationisinnate.Whatwedoknow,however,isthatsuchmotivationisgreatlyinfluencedbyculture.
WhereasWesternculturehighlypromotesindividualism,manyEasternculturespromotecollectivistvaluesworkingtogetherasafamily,aworkgroup,anethnic
group,andsoon.30Theregulationofbehaviorbyrewardsandbyotherpeoplemaybemuchlesstroublingforpeopleacculturatedtocollectivistvaluesthantothose
whohavebeensocializedtoseekindividualism.Additionally,manypeoplevaluebothindividualismandcollectivismthestrengthoftheseorientationsdiffersfromone
individualandonecontexttoanother.31Becauseindividualismandcollectivismarecompetingvalues,wemustlooktopeopleslearninghistories,personality
characteristics,andculturalbackgroundtounderstandwhysomeindividualsmayreactmorenegativelythanotherstoconstraintsontheirfeelingsoffreedomandself
determination.32

ATTRIBUTION,OVERJUSTIFICATION,ANDDISCOUNTING
Whereassomesocialpsychologistsadvocatecognitiveevaluationtheory,othersocialpsychologistshaveusedanattributionapproachtounderstandwhenandwhy
rewardsundermineintrinsicmotivation.33Attributiontheoryconcernsourreasoningaboutthecausesofbehavior.Fromanattributionperspective,thecausesof
behaviormaybeeitherinternalorexternal.Peopleattributetheirbehaviortointernalorexternalsources.Forexample,apersonwhorunsawayfromadogmaysay
thatheorsheranawaybecauseoffear(internalcauses)orbecausetherewasaviciousdog(externalcauses).
Fromtheperspectiveofselfperceptiontheory,DarylBemarguedthatpeopleinfertheirfeelingsandattitudes(internalcauses)fromananalysisoftheirbehaviorand
thesituation.34Whenpeoplereasonthattheirbehavioriscausedbyanexternalsource,theydonotattributethebehaviortotheirfeelingsandattitudes.Duringthe
summer,ifyoureasonthatyouaresprayingfertilizeronlawnsbecauseyouarepaidwelltodoso,youwillnotassumethatyoudoitbecauseyouenjoyit.Thatis,after
lookingatyourbehaviorandthesituation,youwill
Page52

makeanexternalattribution(Ispraylawnsforthemoney)andnotaninternalattribution(Ienjoysprayinglawnsandservingpeople).Ofcourse,apersoncouldenjoy
thejobandalsodoitformoney.Theinternalandexternalfactorscouldaddtogether,givinggreateroverallmotivationtobecomingalawnservicerepresentative.
Accordingtoattributiontheory,whentheexternalcausesofbehaviorareconspicuousorsalient,peoplediscounttheextenttowhichinternalcausesplayarole.
Moregenerally,thediscountingprinciple(orsubtractionrule)statesthatpeoplereduce(discount)apotentialcauseofbehaviortotheextentthatotherpotentialcauses
alsoexist.Theprocessofdiscountinghasbeenappliedtoananalysisofmotivationtoengageinactivitiesandoursubsequentinterestin,andenjoymentof,those
activities.
AsdiscussedinChapters2and3,MarkLepperandhisassociatespublishedoneofthemostcitedarticlesonrewardsandlossofinterestinyoungchildren.35The
discountingprinciplewasusedtopredictanoverjustificationeffectwhenrewardswereusedtomotivatechildrentoengageininterestingactivities(i.e.,drawing
pictures).Accordingtotheoverjustificationhypothesis,intrinsicinterestinanactivityisassumedtobeinfluencedbyattributionsconcerningwhetherperformanceofthe
activityresultedfrominfluencesinternalorexternaltotheperson.Whenindividualsareofferedarewardtodoanactivitytheyalreadyenjoy,thereasoningisthatthey
willhaveanewpotentialexplanationfortheirbehavior(doingtheactivityforthereward).Basedonthediscountingprinciple,thisshouldincreasethelikelihoodthat
theywillattributetheirbehaviortothereward(externalcause)ratherthantotaskenjoyment(internalcause).Thus,becausethereismorethanadequatejustification
(overjustification)forperformingthetask,intrinsicmotivationispredictedtodecline.
Afewresearchershavesuggestedthatuseofthediscountingprincipledoesnotoccuruntilachildreachesadevelopmentalstagecommonlyfoundatabouteight
yearsofage.36Inatypicaldiscountingexperiment,participantsaregivenascenarioinwhichonepersonengagesinataskforintrinsicreasonsandanotherperson
engagesinthetaskbecauseofexternalpressure.Whenparticipantsareaskedwhichpersoninthescenarioprefersthetask,youngchildrendonotreasoninaccord
withthediscountingprincipletheytendtoinferthatextrinsicreasonsindicategreatertaskliking.BecauseLepperandhisassociatesfounddecrementaleffectsof
rewardonintrinsicmotivationinveryyoungchildren,someresearchershavearguedthatoverjustificationcannotaccountforthechangesinintrinsicmotivation.37
Ontheotherhand,otherresearchershavesuggestedthat,iftestedappropriately,thediscountingprinciplecanbeusedbyyoungchildrenandappliedto
overjustificationtypesituations.Intwoexperiments,NewmanandRubledemonstratedthatchildrenwereabletouseandunderstandthediscountingprinciplewhenthe
comparisontheywereaskedtomakewasbetweenobjectsandevents(e.g.,preferenceforatoyoveranactivity)andnotbetweenpeople.38
Page53

NewmanandRublepointedoutthatthistypeofchoicesituationissimilartothesituationinoverjustificationexperimentsinwhichparticipantsmayattributetheir
behaviortoarewardoranactivity.Giventhesefindings,ithasbeenarguedthatthediscountingeffectisaplausibleexplanationfordecrementaleffectsofreward.
AccordingtoLepperandhisassociates,youngchildrendevelopsimplerulesabouttheuseofrewardsineverydaylife.39Bythisdevelopmentalaccount,young
childrenlearntherule,Ifadultsgivemerewardstodosomething,theactivitymustbeboring.Thisruleisextractedfromspecificexperiences,aswhenaparentsays,
Whenyoucleanyourroomyoucangooutandplay,orIfyouwashthecar,youcanhave$10foryourdatetonight.Givenanumberoftheseexperiences,thechild
learnsthattheofferofrewardistantamounttotellingthechild,Therewardedactivityissomethingyoudonotwanttodo.
Inthisexpandedversionofattributiontheory,themoresalientisthemeansendrelationship(instrumentality)betweenanactivityandreward,themorelikelyitisthat
therewardwillundermineintrinsicmotivation.Inchildren,salientrewardsactivatethesimplerulethattheactivityisprobablynotenjoyableadultsusethemore
abstractdiscountingprincipletoarriveatthesameconclusion.Ontheotherhand,whenrewardsconveyinformationaboutcompetence,thisinformationissaidtooffset
theinstrumentalityoftherewardandlessenthedetrimentaleffectonintrinsicmotivation.

PredictionsabouttheEffectsofRewardsonIntrinsicMotivation
Theattributionapproachtorewardsandintrinsicmotivationmakespredictionssimilartothoseofcognitiveevaluationtheory,eventhoughtheunderlyingprocessesare
different.Attributiontheoryemphasizestheinstrumentalityoftherewardswhereascognitiveevaluationtheorypointstoperceptionsofcontrol.Attributiontheorists
predictdecrementsinintrinsicmotivationwhenthereisahighdegreeofinterestinanactivityandextrinsicrewardsareofferedbeforehand(expected).Becausean
individualwouldhavetoknowabouttherewardinadvancetoattributeperformancetoitsinfluence,unexpectedrewardisnotpredictedtoalterintrinsicinterest.In
addition,Lepperandhiscolleaguesaccepttheassumptionofcognitiveevaluationtheorythatintrinsicinterestwillnotbeunderminedbyrewardsthatincreaseperceived
competencewithoutlesseningperceivedselfdetermination,suchasverbalapprovaldeliveredinanuncontrollingmanner.40
AccordingtoLepperandhisassociates,rewardsthatenhanceanindividualssenseofcompetenceandaccomplishmentwillincreasehisorherintrinsicmotivation
conversely,contingenciesthatreduceapersonssenseofcompetenceoraccomplishmentwilldecreasehisorherintrinsicmotivation.41Thus,aswithcognitive
evaluationtheory,Lepperandhisassociatespredictthatexpectedtan
Page54

giblerewardscontingentonengaginginaninterestingactivity(engagementcontingentrewards)willreduceintrinsicmotivationbecausetheydonotconveycompetence
information.
Lepperandhisassociatesdonotmakespecificpredictionsabouthowcompletioncontingentandperformancecontingentrewardswillaffectintrinsicmotivation.
Theydosuggest,however,that

enhancedperceptionsofcompetencemayalsoplayaroleindeterminingmotivation...butonlyaftersuccessfulperformanceattheactivity.Ifonelacks
confidenceinonesabilitytoperformwellenoughtoearnareward,thatrewardisnotlikelytoproveaneffectiveincentive.42

Theimplicationofthisstatementisthattheeffectsofcompletioncontingentandperformancecontingentrewarddependonwhetheranindividualissuccessfulatan
activity.Itfollowsfromthisthatifindividualsaresuccessfulandabletoobtainthefullreward,theirsenseofcompetenceshouldbeenhancedandintrinsicmotivation
shouldincrease.Ontheotherhand,whenpeopledonotperformwellenoughtoobtainthefullreward,theirintrinsicmotivationshoulddecrease.
Thus,althoughLepperandhiscolleaguesdonotpredictpositiveeffectsofcompletionorperformancecontingentrewards,theirtheoreticalapproachwouldsuggest
thattheserewardcontingenciesshouldproducepositiveeffectsonmotivationifpeoplesucceedatataskandobtainthemaximumreward.Incontrast,completion
contingentandperformancecontingentrewardsshouldproducenegativeeffectswhenindividualsarenotentirelysuccessfulandreceivelessthanthemaximumreward,
whichwouldindicatefailurefeedback.Thesepredictionswerenotmadebycognitiveevaluationtheoristswhodonotaddresstheissueofrewardsgivenforsuccessful
performance.Instead,cognitiveevaluationtheorypredictsoverallnegativeeffectsforbothcompletionandperformancecontingentrewards,regardlessofwhether
rewardedparticipantsaresuccessfulatthetask.
Likecognitiveevaluationtheory,anattributionapproachdoesnotrequirethatallrewardsdecreasepeoplesintrinsicmotivation.Butalsolikecognitiveevaluation
theorists,thosewhoadvocatetheoverjustificationhypothesisfocustheirresearchanddiscussiononthenegativeeffectsofreward.Thebulkofresearchcomingfrom
thisperspectiveisdesignedtodetectnegativeeffects.Thisfocusonnegativeeffectshasledtothemisleadingconclusionthatallrewardsareharmful.

EVALUATIONOFATTRIBUTIONTHEORYANDTHEOVERJUSTIFICATIONHYPOTHESIS
Theattributionapproachtorewardsandintrinsicmotivationisbasedonananalysisofcausalreasoning,thediscountingprinciple,andthegeneralization
Page55

thatrewardedactivitiesareboringtochildren.Atthepresenttime,thereisalackofevidencethatthiskindofcausalreasoningmediatestheeffectsofreward
contingenciesonmeasuresofintrinsicmotivation.Thatis,rewardsgivenfordoinganactivitymayresultinlessfreetimespentonthetask,butthereisalackof
evidencetoshowthat(1)peoplescausaljudgmentsarealteredbyrewardcontingencies,and(2)thatthesecausaljudgmentsproduceareductioninmeasuresof
intrinsicmotivation.
Inordertosolvethesedifficulties,attributionresearchersshouldrepeatedlymeasureparticipantscausaljudgmentsduringtheexperiment.Measurementofcausal
judgmentsduringtherewardandfreechoicephasesisnecessarytoshowthemediatingeffectsofattributions.Oncemeasured,statisticaltechniquescanbeusedto
establishthemediatingroleofcausaljudgmentsforanydeclineinintrinsicmotivation.Asacautionarynote,themeasurementofcausaljudgmentsaftertheexperiment
(duringdebriefing)isnotadequate.Thisisbecausepostexperimentalattributionsmaybearesultoftheexposuretotherewardandfreechoicephasesratherthana
mediatorofthelossofinterestandtimespent.
Afinalissueconcernsthediscountingprincipleanditsapplicationtorewardsandintrinsicmotivation.Attributiontheory,likecognitiveevaluationtheory,generally
predictsthatexternalrewardsundermineintrinsicmotivation.Theinternalfactorsofinterestandenjoymentaresaidtobediscountedbyattributingtheactivitytothe
salientexternalrewards.Theunderminingeffectsofrewardsarenot,however,aninevitableoutcomeofanattributionprocess.Thatis,thediscountingprincipledoes
notrequirethattheexternalrewardsalwayssubtractfrominternalcauses.
Whenexternalrewardsareprovidedforanactivity,interventionsthathighlighttheinternalfactorscouldresultinthediscountingoftheexternalrewards.For
example,priortoimplementingarewardcontingency,anattributionprocedurecouldbeusedtofocusparticipantsattentionontheinterestandenjoymentofthe
activity(e.g.,Youmustreallyenjoythetaskbecauseyouspendmuchmoretimeonitthanmostpeople.)Undertheseconditions,theexternalrewardcontingency
mightbediscountedandtheactivityattributedtotheinternalcausesofenjoymentandinterest.Ifthiswereso,rewardcontingencies,combinedwithattribution
procedurestoenhancethesalienceofinternalfactors,couldbeusedtoincreaseperformancewithoutalossofinterestintheactivity.

NOTES
1.Sampson,1988.
2.Rousseau,1762/1974.
3.SeeKohn,1993a.Mostsocialpsychologistsacceptthefindingsthatverbalrewardsarenotdestructivetointrinsicmotivation.Kohn,however,questionsthesefindingsandargues
thatmostpraiseandfeedbackareharmful.
4.Kohn,1993a.
5.Ibid.,pp.7172.
Page56

6. Schwartz,Schuldenfrei,andLacey(1978)arguedthatreinforcementcapturesbehaviorbutisultimatelydestructiveofhumanvalue(seealsoSchwartz,1990).
7. ThefamoushockeyplayerWayneGretzkyrecentlyretiredfromthegame,however,hecontinuestocoachit.Eventhoughtherewardsforplayinghockeyarenolongerforthcoming,
Gretskycontinuestobeinvolvedinthesport.Thisanecdoteiscontrarytotheviewthatrewardsdestroyonesmotivationforanactivity.
8. DeciandRyan,1985.
9. Deci,Koestner,andRyan,1999,p.658.
10.Ibid.,p.658.
11.Bandura(1986)suggeststhatbehaviorcanshowpersistenceandhavenothingtodowithintrinsicmotivation(p.243).Bandurapointsoutthatapersonmayappeartobe
intrinsicallymotivatedinaspecificactivitywhenthereisnothingbettertodo.Ontheotherhand,thepersonwillappeartobeunmotivatedtodothesameactivitywhenmore
attractiveoptionsareavailable.Hence,fromBandurasperspective,anytheoryofmotivationmusttakeintoaccountthewidevarietyofinteractionsthattakeplacebetween
personalandsituationalvariables.
12.Anotherissueconcernstherelativelylowfrequenciesofhumanbehaviorsaidtobeduetoanunderlyingforceofintrinsicmotivation.Forexample,accordingtocognitive
evaluationtheory,musicandartisticpaintingareactivitiesthatshouldbesustainedbyintrinsicmotivation.Basedoninnateenergyordrive,allwehavetodoisexposeourchildren
totheseactivitiesandtheyshouldbecomecaptivated,pursuinglifelongcareersinthearts.Butisthiswhatwesee?Mostchildrenwhotakepiano(oranotherinstrument)donot
becomeaccomplishedmusicians.Also,fewchildrenwhotakeupdrawingandpaintingbecomeacclaimedartists.Ifeachchildhasareservoirofinnatedrivereleasedbythese
activities,howisitthatsofewcontinuetothehighestlevels?Thepointisthatthereismorevariabilityinproficiency,interest,andenjoymentofartisticactivitiesthanwouldbe
expectedifallofushadtheinnateenergyofintrinsicmotivation.Amoreparsiomoniousexplanationisthatvariationinproficiencyandinterestisduetocultureandsociallearning.
13.Inthemorecomplexcaseoftheoriesrelatingintrinsicmotivationtoinnateneedsforcompetenceandselfdetermination,circularreasoningoccurswhenthehypotheticalmotivator
(intrinsicmotivation)andtheinnateneeds(competenceandselfdetermination)areallinferredfromthesamebehaviortheyaresaidtoexplain.Thatis,theneedsforself
determinationandcompetencyareinferredfromthefactthatJoespendsmuchofhistimedrawingatthelake,whichisthesamefactthatindicatedtousthatJoewasintrinsically
motivatedtodraw.BoththeinnateneedsandtheresultingintrinsicmotivationhavenoreferentotherthanJoesbehavior.Ourpointisthatinnateneedsandmotivatorsfailto
explainhumanbehaviorunlesstheseconceptsareanchoredtomeasuresorproceduresindependentofthebehavioritself.
14.Heider,1958DeCharms,1968DeciandRyan,1985.
15.Deci,Koestner,andRyan,1999.
16.Ibid.
17.Ibid.
18.Deci,1971,Experiment3.
19.Thedifferencebetweenverbalandtangiblerewardsandtheireffectsonintrinsicmotivationmayhavelesstodowiththetypeofrewardthanwithitsallocation.Studies
Page57

thatusetangiblerewardsusuallyinvolveasinglepresentationoftheincentive,andtherewardscannotfunctionaspositivereinforcement.Theeffectsoftangiblerewardsaremore
likelyduetotheofferofarewardthantoareinforcementprocess.Incontrast,studiesthatuseverbalrewardsusuallyinvolverepeatedpresentationsofpraiseorpositivefeedback.
Suchrewardsarelikelytofunctionaspositivereinforcementfortheactivity,andindoingso,enhanceinterestinit.SeeCarton,1996.
20.Deciandhiscolleaguespredictedthatexpectedtangiblerewardcontingencies(basedonthecognitiveevaluationofpositiveinformationandcontrollingaspectsofrewards)always
resultinanetlossofmotivation.However,itisnotobviouswhythepositiveinformationaboutcompetenceoutweighsthecontrollingcomponentwithverbalrewardsbutnotwith
expected,tangible,performancecontingentrewards.Ratherthanexplainthisinconsistency,cognitiveevaluationtheoryismadetofittheevidence,providinglittleunderstandingor
predictionaboutthecognitiveprocessthatleadsverbalrewards(asopposedtotangiblerewardsbasedonperformance)toenhanceintrinsicmotivation.
21.Deci,Koestner,andRyan,1999,p.629.
22.Thecoerciveuseofrewardsmayunderliemuchofthenegativeeffectsonintrinsicmotivation.Kohn(1993a)talkedaboutpunishmentbyreward,andSidman(1989)talkedaboutthe
coerciveuseofrewards.Whenrewardsaregivenandpeoplethenthreatentoremovethem,theproceduresareaversive.Thecoerciveuseofrewardsgeneratesescapeand
avoidanceandshouldnotbeconfusedwithpositivereinforcement.
23.Deci,Koestner,andRyan,1999,p.629.
24.Ryan,1982.
25.Researchersintheareaofrewardsandintrinsicmotivationoftenusepuzzlesolvingastheintrinsicallymotivatedactivity.Itshouldbeapparent,however,thatnotall(orperhaps
evenmost)peoplelikesolvingpuzzles.Ifinterestinpuzzlesolvingstemsfromaninnatesourceofenergy,howisitthatthereissomuchvariationinpeoplesinterest?
26.Deci,Koestner,andRyan,1999,p.635.
27.Deci,Koestner,andRyan(1999)broughtuptheissueofmaximumandlessthanmaximumrewards.Intheirmetaanalysisonthetopic,theyshowedthatforperformancecontingent
rewards,lessthanmaximumrewardsledtoalargedecreaseinintrinsicmotivation.Theyattributedthiseffecttotherewardcontingencyratherthantofailurefeedback.
28.Bandura,1986.
29.Geller,1982Hogan,1975.
30.Triandis,1995.
31.Kuhlman,Camac,andCunha,1986KuhlmanandMarshello,1975KuhlmanandWiberley,1976.
32.ThiscriticismofcognitiveevaluationtheorywasfirstnotedbyBobEisenberger(personalcommunication),1996.
33.Lepper,Greene,andNisbett,1973.
34.Bem,1972.
35.Lepper,Greene,andNisbett,1973.
36.KarniolandRoss,1977Morgan,1981.
37.DeciandRyan,1985.
Page58

38.NewmanandRuble,1982.
39.Lepper,Sagotsky,Dafoe,andGreene,1982.
40.LepperandGilovich,1981.
41.Lepper,KeavneyandDrake,1996,p.24.
42.Ibid.,p.24.
Page59

Chapter5
TheoreticalPerspectivesonRewardsasHelpful
Manyoftheactivitiesthatpeopleenjoydoingfortheirownsakeheldlittleornointerestfortheminitially.Fromalearningviewpoint,youngchildrenarenotbornwith
aninnateinterestinartisticactivitiessuchasdrawingandpainting.Aninterestinplayingpiano,singingoperas,solvingmathematicalequations,writingnovels,orplaying
baseballneedstobecultivated.Psychologistsintheareaoflearningandmotivationrejecttheclaimthatrewardsarebasicallyharmfultohumanmotivation.Instead,
theypointtothebeneficialeffectsofrewardsinestablishingperformanceandinterest.
Thecultivationofinterestinactivities,fromalearningviewpoint,dependsonenvironmentalexperiencesthatoriginallyinvolvedtheuseofextrinsicrewards.For
example,achildmayinitiallypracticescalesonapianoforanopportunitytoplaywithfriends.Therelationshipbetweenpracticingscalesandtherewardofplaying
withfriendsisarbitraryandsociallyarranged.Asproficiencyinmusicincreases,externalrewardsaregraduallywithdrawninfavoroftheusualconsequencesofmusical
performance(i.e.,thesoundsofmusic).1Inthisexample,externalrewardsareanecessarypartofteachingcomplexperformancesthateventuallyresultintheirown,
automaticrewardingconsequences.
Today,theviewthatrewardsareanecessarypartofteachingandlearningcomesfromtwobranchesofpsychology:sociallearningandbehavioralpsychology.Both
theseapproachestolearningstatethatrewardsandreinforcementplayanimportantroleintheregulationofhumanbehavior.Rewardsareviewedaspowerful
influencesonhumanperformanceandinterest,andatthesametime,rewardproceduresthatareusedincorrectlymayresultinnegativeeffectsorby
Page60

products.Thesebyproductsofreinforcementhavebeenobservedinthelaboratory,andsomeofthenegativeeffectshavebeenwelldocumentedandanalyzed.2What
isclear,however,isthatrewardsandotherreinforcementsdonothaveinherentnegativeeffectsonhumanmotivation.Fromanappliedlearningperspective,the
objectiveistospecifyhowtousetheprinciplesofreinforcementineverydaysettingswithoutgeneratingasbyproductsnegativeeffectsonperformanceandinterest.
Inthischapter,wepresentAlbertBandurassociallearningtheoryofrewardsandintrinsicinterestandpointtotherolethatrewardsplayinhumanlearningand
motivationfromthisperspective.Sociallearningtheoryemphasizestheinformationalaspectsofrewardsandhowrewardsinfluencethoughtsaboutfutureoutcomesof
action.Inthisregard,sociallearningresearcherssharemuchwithcognitivesocialpsychologists.Thedifferenceisthatsociallearningtheoristssuggestthatrewardsplay
apositiveandimportantroleinlearningandmotivation.
Incontrasttothesociallearningapproach,behaviortheoryemphasizestheroleofrewardsinthelearningofcontingentrelationsbetweenbehaviorandits
consequences.3Fromthebehavioralperspective,peopledothingsthatresultinrewardingconsequencesandthuslearnthecontingenteffectsoftheiractions(i.e.,IfI
doX,thenYhappens.).Basedoncontingencylearning,individualsaremorelikelytodosimilaractionsinthefuture.Thinkingaboutthecontingenciesaswellasthe
actionsthataretakenbothresultfromtheinteractionofthepersonsbehaviorwithitsconsequences.Thatis,boththoughtandactionbyapersonareviewedas
behaviorthathasbeenselectedbyitssuccessinmeetingtherequirementsforrewardorreinforcement.
Inthischapter,wewilldescribehowsociallearningtheoristsstresstheantecedenteffectsofrewardsasincentivesforthoughtandactionwhereasbehavioral
psychologists(behavioranalysts)pointtothelearningofcontingenciesthroughtheconsequencesofaction.Throughoutthechapterweprovidecriticalcommentaryon
thesociallearningandbehavioralperspectivesonrewardsandintrinsicmotivation.

SOCIALLEARNING,REWARDS,ANDMOTIVATION
Aspartofhiscomprehensivetheoryofhumanbehavior,Bandurapresentsasociallearninganalysis(alsocalledsocialcognitivetheory)ofrewardsandhuman
motivation.4Bandurasanalysisofincentivemotivatorsisparticularlyrelevanttotherewardandintrinsicmotivationdebateandthehypothesisthatexternalrewards
reduceintrinsicmotivation.
Incontrasttosocialpsychologistswhoseerewardsandincentivesasintrusionsonhumanmotivation,Banduracontendsthathumanbehaviorisextensivelyregulated
byitsconsequences.Bandurastates:

Ifpeopleactedwithforesightonthebasisofinformativecuesbutremainedunaffectedbytheresultsoftheiractions,theywouldbetooinsensibleto
surviveverylong.Behavior
Page61

is,infact,extensivelyregulatedbyitseffects.Actionsthatbringrewardsaregenerallyrepeated,whereasthosethatbringunrewardingorpunishing
outcomestendtobediscarded.Humanbehavior,therefore,cannotbefullyunderstoodwithoutconsideringtheregulatoryinfluenceofresponse
consequences.5

Insociallearningtheory,rewardsandotherresponseconsequencesaresaidtoinfluencehumanbehaviorthroughtheirincentivefunction.Peopleusethepatternof
feedbackfromrewardsovertimeasinformationabouthowtoproduceorincreasetheseoutcomes.Thatis,rewardsforcertainactionscreateexpectationsofthese
outcomesinthefuture.Insociallearningtheory,behaviorinasituationisincreasedbyexpectedrewardsandreducedbyanticipatedpunishment.
Bandurafurtherproposesthatpeopleshowlittlechangeinbehaviorunlesstheyareawareofwhatactionsarebeingrewardedorpunished.Awarenessofthe
contingencybetweenbehavioranditsconsequencesbuildsexpectationsofreward.Onceexpectationsofrewardhavebeenestablished,behaviorcanbemaintainedby
intermittentrewardsthatoccasionallyconfirmoutcomeexpectations.Generally,rewardsandpunishmentaresaidtohelppeopleanticipatetheoutcomesoffuture
actions,andinsodoing,theseeventsaidintheregulationofadaptivebehavior.6

ExtrinsicandIntrinsicMotivators
Incontrasttosocialpsychologicaltheoriesthatdepictrewardsasharmfultotheinnatemotivationforcompetenceandselfdetermination,sociallearningtheoryholds
thataspectsofpersonalagency(e.g.,selfmotivationandselfdirectedness)arise,inpart,fromexternalinfluences.Thatis,sociallearningtheoryrejectstheappealto
innatesourcesofmotivationandcontendsthatpersonalcompetencies(andotheraspectsofselfregulation)aredevelopedwiththeaidofexternalincentives.Bandura
indicatesthat

Manyoftheactivitiesthroughwhichcompetenciesarebuiltareinitiallytiresomeanduninteresting.Aspianoplayers,andthemuchlargerranksofformer
pianoplayers,willattest,thereislittlejoyinpracticingtherudimentsofthekeyboard,especiallywhenonespeersareatplay.Itisnotuntilsome
proficiencyisacquiredthattheactivitybecomesrewarding.Withouttheaidofpositiveincentivesduringtheearlyphasesofskillacquisition,potentialities
arelikelytoremainundeveloped.7

Banduraspointisthatexternalsocialrewardsbuildinitialinterestandaccomplishedperformanceuntilnaturalconsequencestakeover.Onceskillfulperformance
producesitsownreward(e.g.,themelodioussoundsofmusicalperformance),socialincentivescanbeslowlywithdrawn.Apersonwiththiskindofrewardpattern
pursuestheactivityseeminglyforitsownsakeandenjoysdoingit.
Page62

Banduramakesausefuldistinctionbetweenextrinsicandintrinsicmotivators,basedonaconsiderationoftherewardcontingencyandthelocusoftherewarding
effects.Arewardcontingencycanbearbitraryornatural.8Also,thelocusofrewardcanbeexternalorinternal.Whentherewardcontingencyisarbitraryandthe
rewardingeffectsareexternaltotheperson,theactivityissaidtobeextrinsicallymotivated.Workingforpay,solvingpuzzlesforapproval,practicingamusical
instrumentfortimewithfriends,ordoingmathematicsproblemstogainprivilegesareexamplesofbehaviorthatisextrinsicallymotivated.Ineachcase,thereward
contingencyisarbitraryandthesourceofrewardisexternaltotheperson.
Withextrinsicmotivators,therewardsforbehavioraresociallyarranged.Whensociallyarrangedrewardsareremoved,Banduraindicatesthatthebehaviordeclines
unlessitproducesotherrewardingeffects(i.e.,naturalconsequences).Insociallearningtheory,thedeclineinbehaviorissaidtobeduetochangesinexpectationsof
reward.Noticethatthedecreaseinbehaviorisnotassignedtoapermanentlossofinnateenergyordrive.Inotherwords,sociallyarrangedrewardsdonotdestroy
innatemotivation.Ifthebehaviorissociallyimportant,arbitraryrewardscanbereinstatedtoincreaseexpectationsofrewardforthedesiredbehavior.Oncetheperson
regainsexpectationsofreward,rewardsneedbegivenonlyoccasionallyandslowlyreplacedbynaturalconsequences.
Forsomepeople,therewardingnaturalconsequencesofanactivitymayremainweakorineffective.Thus,findingsolutionstomathematicalproblemsmayremaina
weaknaturalrewardformanystudents.Inthesecases,sociallyarrangedrewards(grades,praiseandrecognition)mustremainineffecttomaintainanappropriatelevel
ofmathematicalperformance.Manyprofessionalmusicians,artists,andathletesperformatleastpartiallyformoney,prestige,andrecognition(e.g.,awardslikethe
GrammiesandOscars),evenwhenmorenaturalisticrewardsalsoregulatetheirbehavior.
Bandurasanalysisofrewardcontingency(arbitraryornatural)andlocusofreward(externalorinternal)leadstodifferentkindsofintrinsicmotivatorsofbehavior.
Onetypeofintrinsicmotivatoroccurswhentherewardingeffectsareexternalbutnaturallyrelatedtobehavior(i.e.,externallocusandnaturalcontingency).Thus,
touchingablanketproducestactilestimulation,strikingguitarstringsresultsinmusicalsounds,andlookingthroughamicroscopemakestheunobservableworld
observable.Ineachoftheseinstances,behaviorproducesnatural,externalsensoryeffectsthatincreaseorsustaintheactions.
Anotherformofintrinsicmotivatorinvolvesbehaviorthatproducesnaturallyrewardingeffectsinternaltotheperson.Behaviorthatgeneratesneurophysiological,
neurochemical,orhormonaleffectsdemonstratesthiskindofintrinsiccontingency.Forexample,fooddeprivationincreasesphysicalactivity,whichinturnisrewarded
bythereleaseofbrainopiates(betaendorphin)andneuro
Page63

transmitters(dopamine).Asphysicalactivityincreasestoexcessivelevels,thereinforcingefficacyofeating(foodintake)isreduced,perhapsduetotheinhibitory
effectsofdopamineonpeptidesthatcontroleating(e.g.,NeuropeptideY).Thiskindofintrinsiccontingencycouldexplainmanycasesofanorexianervosa,inwhich
patientsexhibitexcessiveexercise,hyperactivity,andpacing,inconjunctionwithreducedfoodintake.9
TheextrinsicandintrinsicrewardcontingenciesidentifiedbyBanduraarenotuniquetosociallearningtheory.Sociallearningtheoryrequiresthatreward
contingenciesoperateasincentivesbyactivatingthoughtsaboutfutureoutcomes.Itisnotclearhowextrinsicandintrinsicmotivatorsaredistinguishedintermsof
expectationsofrewardandmotivation.Infact,eventhoughBandurarejectstheresponsestrengtheningeffectsofrewardsadvocatedbybehaviortheory,hisanalysis
iscompatiblewiththebehavioralprincipleofreinforcement.Fromabehavioralperspective,rewardcontingencies(bothexternalandinternal)selectbehaviorfora
givensituation,therebyincreasingformsofbehaviorthatresultinreinforcementanddecreasingformsthatarenolongerrewarded(orarepunished).Generally,social
learningtheoryandbehaviortheoryagreethatconsiderationofrewardcontingenciesintermsoftherewardssourceandcontingenciesarecentraltoananalysisof
humanperformanceandinterest.

SelfRegulationandIntrinsicMotivation
Sociallearningtheoryisdistinguishedfrombehaviortheorybyitsattentionto,andemphasison,selfevaluativemechanisms.Selfevaluativeprocessesconstituteathird
kindofintrinsicmotivationinvolvinginternal(selfgenerated)rewardsthatarearbitrarilyrelatedtoonesbehavior(i.e.,arbitrarycontingencyandinternallocus).
Banduraclarifiesthiskindofintrinsicmotivationinthefollowingpassage.

Inmostactivitiesfromwhichpeoplegainlastingenjoyment,neitherthebehavioritselfnoritsnaturalfeedbackisinherentlyrewarding....Itispeoples
affectiveselfreactionstotheirownperformancesthatconstitutetheprincipalsourceofreward.Tociteanuncommonexample,thereisnothing
inherentlygratifyingaboutplayingatubasolo.Toanaspiringtubainstrumentalist,however,aperformancethatfulfillsahopedforstandardisasourceof
considerableselfsatisfactionthatcansustainmuchtubablowing.Improvementsinintellectual,athletic,andartisticpursuitssimilarlyactivateself
reactionsthatprovideasenseoffulfillmentandcreatepersonalincentivesforaccomplishments.Peopledifferwidelyinthepursuitsinwhichtheyinvest
theirselfevaluation.Hence,whatisasourceofselfsatisfactionforonepersonmaybedevaluedorofnoconsequenceforanother.10

Insociallearningtheory,selfmotivationbyevaluationofonesownperformanceisasourceofpersonalregulation(oragency)ofbehavior.Thatis,people
Page64

aresaidtoexertinternalcognitiveinfluencesovertheirownbehavior.Thecognitionsofselfmotivationinvolveknowledgeabouthowtosequenceactions,set
challengingstandardsforperformance,andgenerateselfevaluationsofperformanceaccomplishments.Oncethismotivationalsystemisinplace,Banduraindicates,
peoplewillderivepersonalsatisfactionfromperformancesthatmeetorexceedtheiradoptedstandards.
Inabehavioralview,selfgeneratedrewardsandtheadoptionofpersonalstandardsbeginwith(andareultimatelydependenton)sociallyarrangedcontingencies.
Socialcontingenciesrelatebehaviortorewardsinarbitraryways,whichexplainswhyselfgeneratedrewardsarealsodispensedonanarbitrarybasis.Weproposethat
peoplelearnfromothers,throughsocialmodelingandreinforcementprocesses,tosequenceactionsintoskillfulperformancesandtoevaluatetheirperformances
againstpersonalstandards.Selfgeneratedstandardsandevaluativereactions,fromourperspective,aretreatedaspartofbehaviorthatisacquiredandmaintainedby
socialmodelingandreinforcement(external)contingencies.Selfregulatorybehaviorservestoguideandmotivateactions,butitdoesnotstandalone.Inthelongrun,
selfregulationdependsonintermittentrewardsfromothersandfromthenaturalconsequencesofactionsovertime.
Sociallearningtheorydisputeswhetheranybehavioroccursforitsownsake.Theapparentmotivationofbehaviorbyaninnateenergydriveisrejectedinasocial
learninganalysis.11Insociallearningtheory,performanceandinterestresideintheeffectsofextrinsiccontingencies,intrinsiccontingenciesrelatingbehaviortoitsnatural
effects,orintrinsicrewardsbasedontheselfevaluationofpersonalaccomplishments.Forexample,tothegymnast,theactofperformingabalancebeamroutineisnot
innatelysatisfying.Thiskindofskillfulperformanceisbuiltupovertimebycorrectivefeedbackfromthecoach,sensoryfeedbackfromtheactionsthemselves,andthe
selfevaluationofperformancesthatmeetorexceedpersonalperformancestandards.Banduraassertsthatitisthepersonalachievementofchallenginggoalsthat
resultsinjoyandinterest.Withoutthepersonalchallengesofperformingaroutineonthebalancebeam,itbecomesaboringactivity.

Rewards,Competence,andInterest
Banduraindicatesthatrewardsgivenforachievinghighlevelsofperformancearecontrolling,inthesensethatmoney,praiseandrecognitionareonlydispensedifa
prescribedlevelofaccomplishmentismet.However,accordingtoBandurasanalysis,thiskindofrewardsystembuildscompetence,selfefficacy,orbeliefsabout
onescapabilities.Indoingso,achievementbasedrewardshelppeopletodeveloppersonalcontroloveroutcomes,adoptchallengingpersonalstandards,andlearnto
makeapositiveselfevaluationoftheiraccomplishments.
Thecompetencefeedbackfromexternalrewardcontingenciesdoesnotinevitablyresultinmoreinterestinanactivity.Sociallearningtheoryholdsthatrewardsgiven
forprogressandgradedachievementsarelikelytoactaspositive
Page65

feedbackforjudgmentsofcompetenceand,indoingso,toincreaseinterest.Incontrast,negativefeedbackfromareductioninrewardsbasedoninadequate
performancecandecreaseperceivedselfefficacy,createobstaclestoperceptionsofpersonalcontrolorselfdetermination,andtherebylowerinterest.Perceived
competencemediatestheeffectsofrewardsoninterestandmotivationfromasociallearningperspective.12
Acommonalityexistsbetweensociallearningandcognitiveevaluationtheory.Boththeoriesclaimthatrewardscanconveyinformationaboutcompetence,which
can,inturn,enhanceintrinsicinterest.Cognitiveevaluationtheory,however,hasplayeddownthisaspectofrewardcontingencies,emphasizinginsteadthecontrolling
effectsofrewardsinunderminingselfdeterminationandintrinsicmotivation.Thisemphasisoncontrolcomesfromthenativisticrootsofthecognitiveevaluation
perspective.Becauseintrinsicmotivationissaidtobeaninnateenergysourcethatdriveshumanbehavior,externalsourcesofinfluencearecastasintrusionson
autonomy,ratherthansourcesofpersonalregulation.Banduradisagreeswiththeemphasisofcognitiveevaluationtheoryoncontrolandsuggeststhatitleadstoa
strangeparadox.Hestates:

Ontheonehand,intrinsicmotivationissaidtobethewellspringofmotivity.Ontheotherhand,itiscontendedthatincentives,constraints,deadlines,and
directives,allreadilysapintrinsicmotivation.Ifthiswereso,suchamotivatorwouldbethoroughlyunderminedbythecountlessextrinsicpressures
impingingconstantlyonpeopleintheirdailylives.Ifintrinsicmotivationissoeasilywipedoutbyexternalinfluences,itcouldhardlyfunctionasapervasive
motivatorofbehavior.13

Incontrasttocognitiveevaluationnotions,sociallearningtheorysuggeststhatextrinsicrewardscandifferentiallyaffectthevalueofanactivity.Whenothersuse
rewardstoconveythefactthatanactivityisuninteresting,peoplelearntodevalueit.Incontrast,whenrewardsareusedtoindicateproficiency,peoplelearntoenjoy
theactivityforitsownsake.Fromasociallearningperspective,rewardshavenoinherentnegativeeffects.Itisthemessageconveyedbytheuseofrewardsthatresults
indifferenteffectsoninterestandperformance.

NonCompetencyContingentandCompetencyContingentRewards
Incognitiveevaluationtheory,Deciandhiscolleaguesdistinguishedamongtask,completion,andperformancecontingentrewards,emphasizingthecontrolling
aspectsoftheserewardcontingencies.Allthesecontingenciesinvolveexternalrewardsthatareclaimedtoundermineselfdeterminationandtherebyproducenegative
effectsonintrinsicmotivation.Incontrasttocognitiveevaluationtheory,Bandurassociallearningtheorydistinguishesbetweennoncompetencycontingentrewards
(rewardsgivenwithoutregardtomastery),andcompetencycontingentrewards(rewardsformasteringactivitiesortasks).The
Page66

emphasisinsociallearningisonhowrewardcontingenciesrelatetoperceivedcompetenceorselfefficacyratherthantoperceivedcontrolandselfdetermination.
Rewardcontingenciesthatenhanceperceivedcompetenceorselfefficacyareexpectedtoincreaseinterestandperformanceofanactivity.
NonCompetencyContingentRewardsincluderewardsgivenformerelydoing,completing,orrepeatinganactivity.Thistypeofrewardcontingencyincludesmany
ofthestudiesthatDeciandassociatesclassifiedseparatelyastask,completion,andperformancecontingentrewards.14Thatis,mostoftheresearchonrewardsand
intrinsicmotivationfromasociallearningviewhasinvolvedrewardsofferedforengaginginanactivitywithoutregardtosomestandardorcriterionofperformance
(mastery).
Extrinsicrewardsgivenformerelydoinganinterestingactivityrepeatedlyaremostlikelytoreduceintrinsicinterest.Thiskindofrewardprocedure,accordingto
Bandura,impartslittleindicationofcompetencyinthattherewardsareonlylooselytiedtobehavior.Inaloosecontingency,therewardsareallocatedwithoutregard
toqualityofperformance.Banduranotedthataloosecontingencyofrewardexertsweakinfluenceoverbehavior,andasaresult,otherfactors(ormoderator
variables)canreadilyalteroroverridetheeffectsoftherewardprocedure.15

SOCIALLEARNINGANDNONCOMPETENCYCONTINGENTREWARDS

Factorsthatmoderatetheeffectsoflooseornoncompetencycontingentrewardsonintrinsicmotivationincludethelevelofinitialtaskinterestabilityonthetaskthe
size,salience,andvalueoftherewardsthetypeoftasktheageoftheparticipantsandwhenandhowintrinsicmotivationismeasured.Banduraarguesthatthe
analysisofnoncompetencycontingentrewardsis

ofnogreatsocialimportbecauserewardsarerarelyshoweredonpeopleregardlessofhowtheybehave.Noristheremuchcallforincentivesystemsfor
activitiespeoplefindhighlyinterestingandthusreadilypursueontheirownwithoutextrinsicmotivators.16

Thecriticalissue,accordingtoBandura,ishowtocultivateinterestandperceptionsofpersonalefficacythroughrewardsbasedonperformance.Insociallearning
theory,performancecontingentrewardsneedtobeseparatedintorewardsgivenfortaskmastery(e.g.,recognitionforoutstandingperformance)and
Page67

rewardsgivenforroutineactivities(e.g.,payforeachunitproduced).Afactoryworkerwhoispaidforeachpieceofleathercuttomanufactureshoes(i.e.,piecerate
payments)isunlikelytoderivemuchinterestandsatisfactionfromthework.
Rewardsgivenformastery(i.e.,achievingrelativelychallengingbehavioralstandards)aretermedcompetencycontingentrewardsandarethetypeofreward
contingencythatissaidtodevelopperceptionsofselfefficacyandtaskinterest.17Thecabinetmakerwhoisrewardedwithrecognition,praise,andmoneybasedonhis
orherwoodworkingskillsandfinelycraftedproductwillprobablydevelopbeliefsinpersonalcompetencyandenjoymentofthework.Inthisexample,external
rewardshelptoinstillpersonalregulatorsofbehavior(e.g.,perceptionsofcompetenceandinterest)ratherthanusurpintrinsicmotivation.

SOCIALLEARNINGANDCOMPETENCYCONTINGENTREWARDS

Whenrewardsaregivenformasteringknowledgeandsubskillsrelatedtoanindividualslongtermgoals,thepersonwillgainasenseofcompetence.Oneresearcher
comingfromthisperspectivefoundthatforchildrenwhowerepooratmathematics,incentivesformasteringsubskillsraisedmathematicalperformanceandincreased
perceivedcompetenceatmathematics.18Inaworksetting,anemployeewhotakesaseriesoftrainingcoursesrelatedtocomputersandaccountingmayreceivepraise,
recognition,andmonetarybonusesfordailyandweeklyaccomplishments.Thiskindofrewardsystemcouldpromotethelearningofcareerorientedbehaviorand
therebyenhanceapersonsselfassuranceandcareerinterest.
Rewardsgivenforachievingrelativelychallengingstandardsarealsoindicativeofcompetence.19Banduranotesthatrewardsgivenformeetingorexceedinga
challengingcriterionverifypeoplescompetence.Whenincentivesaregivenfortryinghard,peoplelearnwhatskillstheypossessandwhattheyareabletodo.20
Rewardstiedtoachievingperformanceobjectivesalsocausepeopletocaremoreaboutdoingwellatanactivityandincreaseintrinsicinterestmorethanpositive
performancefeedbackwithoutreward.21Generally,whenrewardsarelinkedtoovercomingchallenges,theseincentiveshelppeopledevelopasenseofselfefficacy,
interest,andinvolvementinactivities.22
Page68

RewardsandtheSocialValidationofCompetence
Insociallearningtheory,rewardscanalsoserveasasocialvalidationofonescompetence.Thisisespeciallythecasewhenpeopleareunabletojudgetheir
competencefromperformancealone.Whenthemagnitudeofrewardistiedtoqualityofperformance,greaterrewardscanindicatethatotherpeoplebelieveinthe
personsunderlyingabilities.Banduraexplainedthatthecompetencevalidationprocessissomewhatcomplex,involvingtheassessmentofmultiplebehaviors.Thus,for
example,baseballplayersmustjudgecompetenciesfromqualityofbatting,fielding,stealingbases,productionofruns,andsoon.Aplayersoveralljudgmentof
competenceatbaseballisbasedonthepersonalimportance(subjectiveweighting)ofeachaspectofthegame.
Theambiguityanduncertaintyofthisjudgmentprocess,accordingtoBandura,ensuresthatpeoplelooktoothersforindicationsofperformancequality.Inthe
searchforsocialvalidation,themagnitudeoftherewardprovidesevidenceofperformanceworthfromwhichpeoplemakeinferencesofcompetenceandself
efficacy.Researchshowsthatthemoretheextrinsicrewardforperformanceindicatescompetence,thegreatertheinterestinanactivity.23
Aninterestingaspectofsocialvalidationisthatapersonstalentsandabilities(internalfactors)aretakenasthereasonsfortherewards,ratherthantherewards
beingseenasthecauseofhighlevelperformance.24Contrarytotheclaimthatrewardsareusuallyviewedascontrolling,peopleoftenseethemselvesascontrollersof
therewards(selfdetermined),especiallywhentherewardsservetovalidatecompetence.25

TemporalLagandtheThresholdbetweenCompetenceandInterest
Banduraindicatedthatthereisprobablyadelaybetweenanincreaseinperceivedselfefficacyandthecultivationofinterestinanactivity.26Intheworkplace,an
incentivesystemthatrewardsimprovementinjobskillsandknowledge,theovercomingofjobrelatedchallenges,andhighqualityofjobperformanceislikelytobuild
perceptionsofpersonalcompetenceovertime.Asasenseofefficacydevelops,workersseekoutadditionalmasteryexperiencesthatprovideopportunitiesfor
positiveselfevaluations.Selfevaluativereactionsrelatedtocompetence(orsocialfeedbackofcompetence)eventuallyleadtointerestinanactivity.
Inadditiontoatimelagbetweenperceivedselfefficacyandintrinsicinterest,Banduraspeculatesthatpersonalcompetencemustreachamoderatelevel(or
threshold)beforeselfefficacycanupholdinterestinanactivity.Inthisscenario,anincentivesystembasedonperformanceaccomplishmentsbuildsbeliefsinpersonal
competencytothethresholdlevel.Atthispoint,peoplebecomehighlyinterestedinataskoractivity.Onestudyshowedthatchildrensmoderateto
Page69

highbeliefsinpersonalcompetencypredictedhighintrinsicinterestinanactivity,regardlessofthegoalstheyhadsetforthemselves.27Aspeoplebecomehighly
confidentintheirabilities,theymayfindlessandlesschallengeinanactivityandloseintrinsicinterest.BandurasanalysisthereforesuggeststhataninvertedU
relationshipexistsbetweenselfefficacyandintrinsicinterest.Atlowlevelsofperceivedcompetency,interestisminimal.Ascompetencyrisestomoderatelevels
throughtheuseofmasterybasedrewards,peoplefindhighinterestinanactivity.Finally,atextremelevelsofselfassurance,intrinsicinterestdropsagaintheactivity
isnolongerviewedaschallengingandselfsatisfying.
Inconclusion,sociallearningtheoryviewsexternalrewardsasanessentialcomponentinthedevelopmentofselfregulation.Rewardsgivenforthemasteryof
challengingtaskshelptobuildbeliefsinpersonalcompetency.Asapersongainsasenseofselfefficacybasedonperformanceaccomplishments,activitiesthatheld
littleinitialinterestbegintotakeonpersonalvalueorsatisfaction.Thus,rewardscontingentonqualityofperformanceandmasteryofdifficulttasksdonotdirectly
increaseinterestintheactivity.Onlywhenarewardprocedureinstills(orincreases)selfassurancewilltheactivitytakeon(orgain)ininterest.Fromasociallearning
perspective,then,aperceptionofpersonalcompetencyisthecentralmediatorofperformanceandinterestinanactivity.

BEHAVIORTHEORY,REWARDS,ANDMOTIVATION
Sociallearningtheoryemphasizescognitivemediatorsofbehavior(selfefficacy,personalstandardsofperformance,andselfevaluation)intheanalysisofrewards,
performance,andhumanmotivation.Behaviortheoryagreesthatrewardsandotherbehavioralconsequencesarecentraltoananalysisofhumanmotivationhowever,
thebehavioralapproachdoesnotpositbeliefsabouttheselfasthereasonsforhumanconductandmotivation.Fromabehavioralperspective,peoplelearnthe
contingentrelationshipsbetweenactionsandconsequenceswithoutcognitivemediators.Actionsthataresuccessfulbecomemorefrequentovertimeactionsthatare
lesssuccessfuldecreaseandeventuallyareextinguished.Fromabehavioralperspective,behaviorisnotdrivenbyinternalmotivesorthoughtsitisselectedbyits
consequences.28
Behavioralresearchershavearguedforthebasicandappliedimportanceofthelawofreinforcementandotherprinciplesofbehavior.Severalbehavioranalystshave
reviewedtheliteratureonrewardandintrinsicmotivation,offeringanalysesandevidencecompatiblewiththeprinciplesofbehavior.Thegeneraltenorofthese
behavioralexplanationsisthatrewardsarebasicallyhelpful(e.g.,increasingdesiredbehavior)butunusualconditionscanarisethatleadtoadeclineinbehavior
followingtheremovalofreward.Whentheseconditionsareidentifiedandcorrected,rewardsmaybeusedineverydaysettingswithoutdetrimentaleffects.
Page70

BehaviorAnalysis,Rewards,andIntrinsicMotivation
Oneissueofcontentionamongbehaviorists,cognitiveevaluationtheorists,andattributiontheoristsistheanalysisofthebehaviorenvironmentrelationshipsattributedto
intrinsicinterestormotivation.Behavioralresearchersmakeadistinctionbetweentheautomaticornaturalconsequencesofbehaviorandtheconsequencesofbehavior
arrangedbyotherpeople.29Thedifferencebetweennaturalandsocialconsequencesisseeminglyattheheartofthecontroversyinvolvingintrinsicversusextrinsic
motivation.Cognitiveevaluationresearchersturnintrinsic(ornatural)consequencesofactionintoaninnateenergysource(intrinsicmotivation)thatisunderminedby
extrinsic(social)consequences.Fromthisnativisticperspective,extrinsicrewardsaregiventostrengthendesiredbehavior(i.e.,reinforcement)butultimatelysubvert
peoplesnaturalinterestinactivities.Incontrast,behavioristsarguethatboththeintrinsicandextrinsicconsequencesofbehaviorobeythelawofreinforcement.Any
declineinperformanceisaresultofconditionsthatarefullyexplainedbytheprinciplesofreinforcementratherthanbyalossofinnateenergyormotivation.

DoingThingsforNaturalRewards
Anactivitymayappeartobedoneforitsownsakewhenthebehavioris,infact,maintainedbynaturalconsequences.Studiesconductedinthe1950sbyHarry
Harlowandhiscolleaguesillustratetheoperationofnaturalconsequences.30Harlowandhisassociatesreportedthatmonkeyssolvedcomplexpuzzleproblemswithout
foodrewardspuzzlesolvingincreasedanddecreasedinanorderlyfashion,basedonthenaturalconsequencesoffindingacorrectsolution.Thus,Harlowshowed
thatlearningcouldtakeplacewithonlynaturalrewards.
AnimportantaspectofHarlowsexperimentisthedistinctionbetweenintrinsicandextrinsicrewards.AccordingtoHarlow,thereweretwotypesofrewards.The
monkeysweresolvingpuzzleproblemsforintrinsic,ornatural,rewards,whichactedmuchlikeextrinsicrewardssuchasfood.Thatis,thelearningofmonkeysusing
naturalrewardswassimilartothelearningofotheranimalsusingfoodrewards.Puzzlesolvingbytheanimalsdidnotinvolveaninnateinterestbutresultedfromits
naturalconsequences.Theseautomatic,ornatural,rewardsstrengthenedbehaviorasexpectedbythelawofeffect.Thatis,naturalandextrinsicrewardsfollowthe
samelawsofreinforcement.
Ineverydaylife,thereareavarietyofconsequencesthataffectbehavior.Someoftheseconsequencesarenaturalinthesensethattheyautomaticallyoccurwhenwe
dosomething.ThesearetherewardsthatHarlowidentifiedinthestudiesofpuzzlesolvingbymonkeys.Thesearealsotherewardsthatpresumablyunderliehuman
activitiesapparentlydonefortheirownsake.Otherconsequencesaresociallyarranged,aswhenateachergradesapaperoranemployeegetspaidforwork.
Fromabehavioralperspective,behaviorisalwaysafunctionofitsconse
Page71

quences,whetherthoseconsequencesarenaturalorsocial.Itisconfusinganddeceptivetotalkaboutintrinsicmotivationandextrinsicrewards.Behaviorthatappears
tobeintrinsicallymotivatedmayinfactbeduetoitsnaturalconsequences.Considerthefollowingquestions:Howlongwouldachildplaythepianoifnosoundcame
fromit?Howlongwouldonesolvepuzzlesifasolutionneveroccurred?Howlongwouldonereadabookifthetextwerewordsalad?Thepointisthatbehaviorthat
looksintrinsicallymotivatedisactuallyregulatedbyitsnaturalconsequences.Thatis,puzzlesolving,musicalsounds,andagoodstoryarenaturalconsequencesof
theseactivitiesthatresultininterestandenjoyment.Thenaturalconsequencesresideintheenvironment,althoughotherpeopledonotarrangethem.Generally,people
dothingsforreasons,someofwhicharethenaturalconsequencesoftheirbehavior.

DoingThingsforNaturalandSocialRewards
Therearealsoconsequencesofbehaviorthatareprovidedbyothers.Rewardssuchasgoldstars,money,praise,correctivefeedback,andrecognitionare
consequencesthatcanbesociallyarrangedwhenthenaturalconsequencesofbehaviorareweak.Peoplesaygood,right,correct,orexcellent,tohelpa
personbuildskillfulperformance.Peoplealsogivenonverbalrewardssuchassmiles,winks,thumbsupsigns,hugs,congratulatoryhandshakes,patsontheback,or
applause.Atothertimespeoplegivecertificates,candy,prizes,andevenmoneyasrewardsforaccomplishments.Thesearethesocialreasonswhypeopledowhat
theydo.
Socialrewardsareoftenusedtosupplementweaknaturalconsequencesortobuildskillfulactionsthatwilleventuallyresultinnaturalrewards.Thisiswhypeople
oftenusesocialrewardsineverydaylife.Forexample,forthebeginninggolfer,thefirstnaturalconsequenceistohittheball.Thatis,thepersonmustswingtheclubin
suchawaythatitcontactstheball.Manyfirsttimegolfershavegreatdifficultydoingthisandtakealotofairshots.Thegolfprofessionalusesdemonstrationand
correctivesocialfeedbacktoshapeeffectiveswingingofthegolfclub.Eventually,thepersonconsistentlyhitstheballandthispartofthegolfgamenolongerrequires
thesociallyarrangedconsequencesoftheprofessional.
Atthispoint,thetrainerconcentratesonprovidingcorrectivefeedbackfordistanceandaccuracy.Again,oncedistanceandaccuracyhavebeenobtained,the
professionalgoesontoteachtheuseofdifferentclubs,howtomakeapproachshots,howtogetoutofbunkers,howtoputt,andsoforth.Thisprocesscontinuesuntil
thepersonhaslearnedthegame.Oncethegamehasbeenlearned,thenaturalrewardsofsinkingtheballwithafewgoodshotskeepthepersongoing.Inaddition,
learningthegameproducesnewsocialconsequences.Thegolferisnowabletoplaythegamewithothers,whosupplysocialinteraction,praiseforgoodshots,and
conversationaboutthegameattheclubhouse.Thegolferisnowabletoplayintournamentsandreaptherewardsofrecognitionandstatus.
Page72

Thepointisthatlearningtoplaythegameofgolf(oranyothercomplexactivity)involvesthearrangementofbothnaturalandsocialrewards.Withoutbothsources
ofconsequences,apersonwouldneverlearnthegameorfindanyinterestorenjoymentinplayingit.Tosaythatthegolferisintrinsicallymotivatedismisguidedmost
behaviorisacquiredandmaintainedbyacombinationofnaturalandsocialrewards.

DoingThingsWhenRewardIsIntermittentorDelayed
Althoughmanyhumanactionsappeartooccurintheabsenceofanyobviousorapparentextrinsicrewards,theymay,infact,beduetointermittentorinfrequent
consequences.Inthegolfexample,animportantpartoftheinteractionbetweenthegolferandtheprofessionalisthatthesocialrewardsgivenbythetrainerbecome
lessandlessfrequentasthegolfergainsproficiency.Evenwhenthegolferhasachievedahighlevelofaccomplishment,theprofessionalmayintermittentlyreward
aspectsofthegame.
Inaclassroomsituation,ateachermayproviderewardsdependentonthelevelofachildsproficiencyataparticularsubject.Duringtheinitialphasesoflearning,
rewardsfromtheteachermaybefrequentandconsistent.Later,asthechildgainsskillintheacademicsubject,theteacherprovidesrewardsonlyoccasionally.This
practiceinvolvesgoodeconomics:thatis,oneshouldfrequentlyrewardweakbehaviorandinfrequentlyrewardstrongbehavior.Itturnsoutthatskillfulperformanceis
bettermaintainedbyintermittentrewardthanbycontinuousreward.Whenrewardsaregiveninfrequently,itistemptingtocallthebehaviorintrinsicallymotivated.To
anoutsideobserver,thegolferandthestudentmayappeartobeintrinsicallymotivated.However,itistheintermittentrewardsthataccountforthebehavior.
Behaviorthatisregulatedbyitslongtermconsequencesisalsomistakenlyattributedtointrinsicmotivation.Insuchcases,theuseofthetermintrinsicmotivation
servesonlytoconcealamoreindepthanalysisofhumanperformance.Forsomeactivities,suchasplayingthepianoorpaintingapicture,ittakesconsiderabletime
andefforttobecomeproficientenoughtoobtainthelongrangeeffects.Theseactivitiesareusuallyarrangedinaseriesofprogressivesteps,withtheeffectsof
improvedperformancesupportingbehaviorateachlevel.Aspartofthisprocess,selfevaluativereactionsbridgethegapbetweenbehavioranditslongrangeeffects.
Formanyactivitiesinwhichpeopleengage,thenaturalconsequencesarenotyetrewarding.Aspreviouslynoted,Bandurapointedtopeoplesselfevaluativereactions
totheirbehaviorastheprincipalsourceofreward.31
Fromabehavioralperspective,selfevaluativereactionsthemselvesareacquiredandmaintainedbysocialrewardsthatresideoutsidetheindividual.Peoplelearnto
setstandardsofperformanceforthemselvesandtoevaluatetheseperformances.Thesocialconsequencesofjudgingonesperformanceandthe
Page73

actualattainmentofskillfulperformancegiveselfevaluativereactionstheirvalue.Thus,thetubaplayerhaslearnedtoidentifyimprovementsinperformanceandto
describethisimprovementaspositive.Selfevaluationbridgesthestepsbetweenlearningtoplaythetubaandthelongrangeskillfulperformanceofamusical
symphony.Inotherwords,farmoreisinvolvedinanaccountofmusicalperformancethantosayitisintrinsicallymotivatedbehavior.

BehaviorAnalysis,Rewards,andtheDisruptionofPerformance
Animplicationofthebehavioralargumentisthatresearcherswhousethetermintrinsicmotivationconfusebehaviorwiththeeffectsofthatbehavior.Theeffectsof
behaviorinvolverewardsthatarenaturalorsociallyarranged,andwhichmaybeintermittentordelayed.Labelingbehaviorasintrinsicallymotivatedmakesnoneof
thesedistinctions.Onceitisclearthatpeopleengageinactionsfornaturalandsocialrewards,theclaimthatrewardsaregenerallyharmfulisempty.
Behavioralpsychologistsacknowledgethatitispossibletodisruptorsubvertbehaviorthroughextrinsicsocialrewardsbutitisjustasplausiblethatbehaviorbeing
maintainedbysocialconsequencesisoccasionallydisruptedorsubvertedbynaturalrewards.Theeffectiveuseofrewardsisatrickybusiness,andsideeffectsfroma
rewardprocedurearealwayspossible.
Peopletakemedicineforsickness,andsomemedicinehasunwantedsideeffects.Thesesideeffectsdonotusuallystopusfromtakingmedicine,althoughwewould
liketoknowaboutthemandavoidthemifpossible.Inasimilarvein,naturalandsociallyarrangedrewardsmaysometimescreateunwantedsideeffects.Thereisno
need,however,tobanishrewardsfromourschoolsandworkplaces,especiallywhentheseconsequencesaretheprimarywaytoestablishlearningandpromote
persistenceinanactivity.Instead,wemustidentifyandcorrectthepossiblesideeffectsofrewardcontingencies,andatthesametimestipulatehowtouserewards
effectively.
Inthisregard,itisinformativetoreconsidertheresearchonrewardsandintrinsicmotivation.Recallthatbehavioristsinsistthatbehaviorisalwaysduetoits
consequencesandthatitobeystheprinciplesofreinforcement.Giventhisreality,howcanweunderstanddisruptionsinperformancethatresearchersreportinthe
experimentsonrewardandintrinsicmotivation?Theanswerliesinadetailedanalysisoftheexperimentalproceduresandhowaspectsoftheexperimental
arrangementsaffectthebehaviorofparticipants.Mostimportant,decrementaleffectsofrewardfoundinstudiesareduetofactorsotherthanadeclineininnateintrinsic
motivation.Instead,sucheffectsresultfromwellknownbehavioralprocessesassociatedwiththepresentationandsubsequentwithdrawalofreward.
Behavioristshavearguedthatwhenarewardispromisedorpresentedonly
Page74

once,asinthestudiesconductedbyDeci(1971,1972b)andLepperandhiscolleagues(Lepper,Greene,andNisbett,1973),performancemaybestronglyinfluenced
bytheindividualsinterpretationoftaskinstructionsandreinforcementhistory.Forexample,theeffectonperformanceofpromisingachildamonetaryrewardfor
workingonapuzzlemaybeinfluencedbywhetherpreviouspromisesbyauthorityfigureshavebeenmetandbypreviouswaysinwhichpuzzleshavebeenrewarded.
Todifferentiatetheeffectsofcurrentreinforcementconditionsfrompriorlearning,manybehavioranalystsfavortheuseofrepeatedrewardsessionsfollowedby
repeatedsessionsaftertherewardhasbeenremoved.TheearlystudybyFeingoldandMahoneywasconductedusingthisprocedure,andnonegativeeffectsof
rewardwerefoundonmeasuresofintrinsicmotivation.32

DifferentialEffectsofTangibleRewardsandPraise
Cartonaddressedthedifferentialeffectsoftangiblerewardsandpraisefromabehavioranalysisperspective.33Cartonnotedthatpraiseusuallyincreasesorleaves
measuresofintrinsicmotivationunaffectedincontrast,anofferofatangiblerewardforanactivityoftenreducesthesemeasures.Ratherthanclaimthatpraisehas
informationalpropertiesaboutcompetencethatenhanceintrinsicmotivation,whereastangiblerewardsarecontrollinganddecreaseintrinsicinterest,abehavioranalysis
pointstoproceduraldifferencesinstudiesthatusepraiseversustangiblerewards.Theproximityofrewardtothetargetbehavior,thenumberofreward
administrations,andthestimulioreventsassociatedwithrewardavailabilityfrequentlydifferbetweenexperimentsthatusetangiblerewardsandthosethatusepraise.
Theseproceduraldifferencescouldaccountfortheapparentlossofintrinsicmotivationwithtangiblerewardsandtheincrease(orabsenceofeffect)withpraise.
Anexaminationoftheliteratureindicatesthattangiblerewardsandpraisearedelivereddifferentlyinexperimentsonintrinsicmotivation.Onedifferenceconcernsthe
temporalproximityofrewardsandthetargetbehavior.Inatypicalexperimentusingatangiblerewardlikemoney,paymentisgiventotheexperimentalparticipants
followingtherewardphase(thecontrolgroupdoesnotreceiveanypayment).Insomeexperiments,thetangiblerewardisdeliveredfollowingthefreechoiceperiod,
andinafewstudies,therewarddeliveredonlyafteraperiodofweeks.Thebasicissueisthatthedeliveryoftangiblerewardsofteninvolvesadelaybetweendoingthe
requisitebehaviorandreceivingthereward.Ontheotherhand,praiseisusuallydeliveredthroughouttherewardphase,immediatelyfollowingthetargetbehavior.
Thus,anydifferenceinintrinsicmotivationbetweentangiblerewardsandpraisecouldbeduetoavariationindelaybetweenbehavioranditsconsequences(as
opposedtotherelativedegreeofcontroloftheserewards).
Arelatedissueconcernsthenumberofoccurrencesofrewardinstudiesofintrinsicmotivationusingpraiseandtangiblerewards.Asexplainedinthedis
Page75

cussionofoperantorsinglesubjectdesigns,repeatedrewardsgivenforbehaviorarelikelytoactasreinforcement,increasingthetargetresponse(thisisdiscussed
furtherinthenextsection).Inmanystudiesusingtangiblerewards,therewardsaregivenonlyonce,attheendoftherewardphase.Asinglerewardadministrationis
mostcharacteristicofstudiesthatoffertangiblerewardsfordoingorengaginginatask.Thebulkofthesestudiesfindnegativeeffectsofrewardsonmeasuresof
intrinsicmotivation(seeourmetaanalysisinChapters8and9).Incontrast,experimentsconcernedwithpraiseandintrinsicmotivationtypicallyadministertheverbal
rewardsmultipletimesfollowingtherequisitebehavior.Mostofthesestudiesfindpositiveeffectsofpraiseonintrinsicmotivation.Onepossibilityisthatthedifferential
effectsofpraiseandtangiblerewardonintrinsicmotivationaredue,inpart,tothenumberofrewardadministrationsratherthanthenatureoftheserewards(i.e.,
controllingversusinformational).
Anotherconcernofbehavioralresearchersregardingpraiseandtangiblerewardstudiesisthedifferenceinstimulithatsignaltheavailability(andwithdrawal)of
rewards.Ingeneral,instudiesofpraiseandintrinsicmotivation,participantsarenottoldthatthepraisehasstoppedbeforethefreechoicephase.Incontrast,
experimentersusingtangiblerewardsoftensignaltheavailabilityoftheseconsequencesduringtherewardphaseandtheunavailabilityofrewardduringthefreechoice
period.Forexample,inmanytangiblerewardstudies,theexperimenterleavestheroombeforethefreechoiceperiod,signalingthattherewards,suchasmoney,are
nolongeravailable.34Insometangiblerewardstudiesbeforethefreechoicephase,explicitstatementsaregiventoindicatethatrewardshavestopped.35The
differenceinstimulithatsignalrewardavailabilitycouldaccountforthereducedfreetimespentontaskfortangiblerewardsandthecontinued(orincreased)
engagementinthetaskwhenpraiseisused.
Considermachinesthatdispensesoftdrinkswhenpeopleinsertcoins.Gettingdrinksfromthesedispenserstypicallyreinforcesthebehaviorofinsertingacoinand
pressingabuttonforthedesireddrink.Thisbehaviorreliablyoccurswhenthemachineflashesthesign(stimulus),Insertcoinshere.Thus,signalsofreward
availabilityplayaroleinkeepingthebehaviorgoing.Butwhatwillhappenifthesoftdrinkdispenserisbrokenandthemachineoperatorplacesanoutofordersign
ontheunit?Thesignwillusuallyactasastimulusthatindicatestheunavailabilityofreward(i.e.,nosoftdrinks),andmostpeoplewillnotinsertmoneyintothemachine
orpressthebuttons.Thepointisthatastimuluschangefromrewardavailabilitytorewardunavailabilitycanshifttheprobabilityofbehaviorfromhightolow.Thiskind
ofstimuluschangeandshifttoalowprobabilityofresponseshouldoccurmoreofteninstudiesoftangiblerewardandintrinsicmotivationthaninstudiesofpraise.
Page76

PROCEDURALDIFFERENCESINSTUDIESOFINTRINSICMOTIVATIONUSINGPRAISEVERSUSTANGIBLE
REWARDS

1.Immediacyofrewardandtargetbehavior
Comparedwithtangiblerewards,praiseismoreimmediate.
2.Numberofrewardadministrations
Comparedwithasingleadministrationoftangiblereward,praiseinvolvesrepeatedpresentationsofrewardovertime.
3.Signalsofrewardavailability
Comparedwithstudiesusingpraise,experimentswithtangiblerewardoftensignaltheavailabilityofreward.

ReinforcingversusNonreinforcingRewards
Adeclineintaskperformanceduringthefreechoicephasecouldreflectparticipantsattemptsatcountercontrol,deliberatenoncompliance,andfeelingsofcontrol
generatedbytheofferofrewards.36
Thatis,participantsrespondtotheofferedrewardsascontrollingand,whengivenachoicetodomoreofthetask,actuallydoless,asareactiontoperceived
controlbytheexperimenter.Thebehavioralhypothesisofcountercontrolinducedbypromisesofrewardisseeminglyrelatedtocognitiveevaluationtheorysemphasis
onthecontrollingnatureofrewards.Amajordifference,however,isthatbehavioraltheoryseparatesrewardsintononreinforcingandreinforcingconsequences,both
ofwhicharecontrolling,butwithdifferentbehavioraleffects.Inabehavioralview,rewardsthatarenonreinforcingaremorelikelytobeperceivedascontrollingand
morelikelytogeneratedeliberatenoncompliance.Thesearetherewardsusedinthebulkoftheintrinsicmotivationstudies.Incontrasttothenegativeimpactof
nonreinforcingrewards,reinforcingrewards(i.e.,rewardsthatincreaseperformance)arelesslikelytoseemcontrollingandlesslikelytoinducecountercontrolby
participants.Skinneraddressedthisissueinthefollowingpassage.

Thefactthatpositivereinforcementdoesnotbreedcountercontrolhasnotgoneunnoticedbywouldbecontrollers,whohavesimplyshiftedtopositive
means.Hereisanexample:Agovernmentmustraisemoney.Ifitdoessothroughtaxation,itscitizensmustpayorbepunished,andtheymayescape
fromthisaversivecontrolbyputtinganotherpartyinpowerinthenextelection.Asanalternative,thegovernmentorganizesalottery,andinsteadofbeing
forcedtopaytaxes,thecitizenvoluntarilybuystickets.Theresultisthesame:thecitizensgivethegovernmentmoney,buttheyfeelfreeanddonot
protestinthesecondcase.Neverthelesstheyarebeingcontrolled.37

Skinnersanalysis,whenappliedtointrinsicmotivationexperiments,suggeststhatthenonreinforcingrewardsgivenfordoingataskactlikedemands.Participantsview
thesenonreinforcingrewardsasattemptstoforcethemtobehave
Page77

inagivenwayandtoreactagainsttheexperimenterscontrolbyshowingdeliberatenoncompliance(i.e.,reducingtimeandbehavioronthetaskinthefreechoice
phase).Behavioralresearchers,ontheotherhand,haveensuredthatreinforcingrewardsareusedintheirexperiments.Presumably,participantsregulatedbypositive
reinforcement(reinforcingrewards)perceivedtheirbehaviorasvoluntary,feltfree,anddidnotshowdeliberatenoncompliance.Thisdifferencebetweennonreinforcing
andreinforcingrewardscouldexplainwhybehavioralexperimentsonrewardandintrinsicmotivationhavenotfounddetrimentaleffectsofrewardsandwhy
nonbehavioralexperimentshave.

OFFERSOFNONREINFORCINGREWARDSCANINDUCEDELIBERATENONCOMPLIANCE(REFUSALTOENGAGEIN
TARGETACTIVITY)
Comparedwithreinforcingrewards,noncompliance(refusingtochoosethetargetactivityinthefreechoiceperiod)isgreaterwith
nonreinforcingrewards.

AssociatingRewardswithFailure
Anotherpossibleexplanationforthenegativeeffectsofrewardisthattherewardproceduresinintrinsicmotivationexperimentsareoftenaversiveratherthan
nonreinforcing.Fromthisperspective,participantsbehaviorispunishedbyrewards,inthesensethattherewardshavebeentiedtofailure,usedinacoercive
manner,orcorrelatedwiththelossofsocialreinforcement.38Forexample,rewardsaresometimestiedtotheachievementofperformancestandards,andexperimental
participantsmayfailtomeetthesestandardsandthusnotbeabletoobtainmaximumreward.Inthiscontext,failurecouldactasaconditionedpunisherandtemporarily
decreasetaskperformanceduringthefreechoiceperiod.39Also,becausefailureispairedwiththenaturalrewardsofperformingtheactivityortask,theintrinsic
rewardscoulddecreaseinvalue(atleasttemporarily).Repeatedencounterswithfailurecould,inthisway,underminemotivation.
Incontrast,whenrewardsaregivenformeetingorexceedingaperformancestandardorgroupnorm,theyaresaidtobesuccesscontingent(maximalreward).
Fromabehavioralview,rewardsthatareprogrammedforsuccessarenotexpectedtohavedetrimentaleffectsonfreechoiceperformance.Infact,evenstrong
opponentsofcontingentrewardsseemtorecognizethatsuccesscontingentrewardsdonothaveharmfuleffects.DeciandRyanacknowledged
Page78

thepotentialbenefitsofsuccessbasedrewardsintheirdiscussionofinformationandreward:

Theimportantpointisthatrewards,likefeedback,whenusedtoconveytopeopleasenseofappreciationforworkwelldone,willtendtobeexperienced
informationallyandwillmaintainorenhanceintrinsicmotivation,butwhentheyareusedtomotivatepeople,theywillsurelybeexperiencedascontrolling
andwillundermineintrinsicmotivation.40...[R]ewardsthatareappropriatelylinkedtoperformance,representingpositivefeedbackinaninformational
context,oughtnottobedetrimental.Thecosttothesystem,however,insignifyinggoodperformancethroughtheuseofperformancecontingentrewards
isthatmanypeopleendupreceivingthemessagethattheyarenotdoingverywellandthisislikelytobeamotivating.41

Whenperformancestandardsaregraded,reasonable,andattainable,rewardsareexpectedtohavepositiveeffectsonmotivation.Ontheotherhand,rewardsgiven
forgenerallyunattainablelevelsofperformanceensurefailureandreducemotivationtoperformatschooloratwork.Thegeneralpointistotierewardstosuccess
ratherthanfailure.

REWARDSASSOCIATEDWITHFAILUREREDUCEFREECHOICEINTERESTANDPERFORMANCE
Lessthanmaximumrewardisusuallyassociatedwithfailurepeoplewhoreceivelessthanmaximumrewardforthetargetbehaviorwillbeless
interestedintheactivityinafreechoiceperiod.
Achievingmaximumrewardatataskcanindicatesuccessrewardsbasedonachievingacriterionandrewardsbasedonachievingacriterion
andrewardsobtainedforsurpassinganormativestandardcanalsoindicateSuccess.

RewardsUsedasaFormofCoercion
Rewardsareoftenusedtoinducepeopletodothingsthattheydonotwanttodo.Thatis,rewardsthatareusedtomotivatepeoplecouldinvolveaformofcoercion.42
Ineverydaysettings,socialagents(e.g.,parents,managers,teachers,etc.)oftenuserewardsonlybecausetheycanberemovedforinstancesofnoncompliance.That
is,children,workers,andstudentsbehavewellinordertoavoidthelossofpromisedrewards,impendingpunishment,orthreatsofpunishmentarrangedbyothers.
Onewaytotestwhetherrewardsarebeingusedaspositivereinforcementorcoercionistoaskwhobenefitsfromtherewardprocedure.Rewardsusedtothe
advantageofthesocialagentsareusuallycoercive,whereasrewardsgiventobenefittherecipientoftenactaspositiveconsequences.43
Intermsofexperimentsonrewardandintrinsicmotivation,behavioranalystspointtoparticipantspreexperimentalhistoryandthecoerciveuseofrewards.
Page79

Theynotethatmostparticipantshaveexperiencedoffersofrewardtoengageinactivitiesthatwerelaterfollowedbylossofreward,punishment,orthethreatof
punishment.Inthisway,offersofreward(especiallyfordoingsomeactivity)cometoactasconditionedaversivestimuli.Inthecontextofthelaboratory,theseverbal
statementsinduceexperimentalparticipantstoengageinthedesignatedtask.Atthesametime,theoffersofreward,whenpairedwiththeintrinsicrewardsofthetask,
reducethevalueofthetargetactivity.Theresultisadecreaseintaskperformanceinthefreechoicephase,incomparisontoacontrolgroupthatdoesnotreceivethe
offerofreward.
Alternatively,performanceinthefreechoicephasecoulddeclineduetodeliberatenoncomplianceasareactiontothecoerciveofferofrewardbytheexperimenter.
Inthisscenario,participantsidentifysourcesofcoercionandreactagainstthem(againbecauseofapreexperimentalhistory).44Becausetheofferedrewardhasbeen
tiedtoadesignatedactivity,participantscanshowtheirnoncompliancebyrefusingtoengageintheactivityduringthefreechoiceperiod.Dickinsonexplainedthat
duringrewardadministration,therewardsmaymaintainbehaviorhowever,whentheyarenolongeravailable,behaviorthathasirritatedorinconveniencedcoercers
maybemomentarilystrong.45

REWARDSBECOMEPUNISHERSTHATSUBVERTTHENATURALCONSEQUENCESOFACTIVITIESANDLEADTO
REDUCEDINTRINSICMOTIVATION
Offersofrewardgiven"tomotivate"people(givenforthebenefitofthesocialagent)areusuallybackedupbypunishmentorthreatof
punishment.Theseincentivesbecomeconditionedpunishers,which,whenassociatedwiththenaturalrewardsofthetargetbehavior,
decreaseperformanceduringfreechoice.

OFFEREDREWARDSCANINDICATECOERCIVECONTROLBYTHEREWARDGIVERANDPEOPLEWILLREACT
WITHDELIBERATENONCOMPLIANCE
Peoplereactwithcountercontrolmeasurestothosewhousecontrol.Peoplewillactinwaysthatupsetorinconveniencethesourceof
coercivecontrol.Inthelaboratory,peoplereactagainstthecoerciveinfluenceoftheexperimentbyrefusingtoengageinthetarget
behavior.

SocialandCulturalIssues
Another,morecomplexanalysisinvolvestheculturalpracticeofwithholdingapproval,praise,andcreditforbehaviorthatisexplicitlyrewarded(e.g.,behaviordonefor
money).Behaviorissaidtobeartistic,original,creative,orintelligentwhenitsoccurrenceisunrelatedtoexplicitconsequencessuchasoffersofmoney,grades,orgold
stars.Forexample,asolutiontoaproblemisadmiredasoriginaliftherehasbeennoexplicittraining(i.e.,comingfromtheperson),
Page80

butitisconsideredlessworthyifitisduetoobvioussocialinfluence.Thesecontingenciescouldextendtoexplicitrewardandoffersofreward.Explicitrewardwould
indicatethatbehaviordidnotemanatefromthepersonandwasnotasworthyofsocialrecognition.Peoplewhoarereinforcedbyapproval,praise,andcreditcould
learntoactinwaysthatminimizetheappearanceofcontrolbyexplicitrewards.
Oneresponsetominimizetheappearanceofcontrolisrefusaltoengageintherewardedactivity.Noncompliancewiththeofferedrewardindicatesweakexternal
control,strongintrinsicmotivation,andbehaviorworthyofsocialrecognitionandapproval.Inaddition,suchnoncompliancewillseemevenmorepraiseworthyifthe
personclaimedpersonalmotivationorinterestintheactivity.Thus,thealcoholicwhodeclinesanofferofadrinkseemsmoredeservingofcreditthanthesocialdrinker.
Intermsofexperimentsonrewardsandintrinsicmotivation,theseculturalcontingenciesensurethatmanyparticipantswillshowadecreaseinthefreetimeontask(i.e.,
noncompliance)and,atthesametime,claimastronginterestintherewardedactivity.46

CONCLUDINGREMARKS
Abehavioranalysisofrewardsandintrinsicmotivationinvolvesaseriesofhypothesesconcerningthevariablesthatchangetheprobabilityofbehaviorfromhightolow.
Behavioralresearchersarguethatwellknownprocessesofstimuluscontrol,punishment,andreinforcementaccountfordifferencesinfreechoiceperformancein
studiesofintrinsicmotivation.
AccordingtoCarton,therearethreeadvantagestobehavioralaccountsofrewardandintrinsicmotivation:(1)parsimony,(2)falsifiability,and(3)externalvalidity.47
Behavioralhypothesesaresaidtobemoreparsimoniousthancognitivetheories(e.g.,cognitiveevaluationtheory)becauseonlyafewvariables,suchasreward
availabilityandfrequency,ratherthanmany,areusedtoaccountfordifferencesinfreechoiceperformanceoverstudies.Referencetoavarietyofinternalmediators
suchasselfdeterminationandperceivedcompetenceisnotrequiredfromabehavioralviewpoint.
Also,behavioralaccountsorhypothesesareclaimedtobemorereadilyfalsifiedthantheoriesthatrelyonunobservablemotivationalconstructsandinnatesourcesof
energyforexplanation.Thatis,experimentscanbearrangedtocontrolfor,ormanipulate,thereinforcementvariablesthataresaidtoaffectfreechoicebehavior.Thus,
explicitprocedurescanbeusedtoinvestigatewhetherchangesinstimulisignalingrewardavailabilityactuallyexplaindifferencesinfreechoiceperformance.48
Athirdadvantageistheclaimthatbehavioralaccountshavegreaterexternalvaliditythancognitiveevaluationtheoryintermsofimplementingrewardprogramsin
appliedsettings.Fromtheperspectiveofcognitiveevaluationtheory,itmakessensetoofferatangiblerewardtopeopleformerelyengaginginsomeactivity,
administertherewardonlyonce,andthenwithdrawallrewardsforthe
Page81

activity(producingthepurporteddetrimentaleffectonintrinsicmotivation).Ineverydaylife,parents,teachers,andadministratorswouldseldom(ifever)usesucha
rewardsystem.Mostrewardsystemsinappliedsettingsuserepeatedrewardstiedtospecificbehavior,proceduresthatareclosertotheonessuggestedby
reinforcementprinciples.
Asafinalcomment,itisnoteworthythatmoststudiesonrewardsandintrinsicmotivationhaveusedcollegestudentsandinvolvedcontingenciesofgradesand
credits.Thatis,studentswererequiredtoparticipateinanexperimentinordertofulfilltheircourseobligationsinintroductorypsychology.Thismeansthatboththe
experimentalandcontrolparticipantswereextrinsicallymotivatedtodothetargetactivitybeforerewardssuchasmoneywereofferedtotheexperimentalgroup.One
interpretationofthisisthatmuchoftheresearchonthetopicactuallyconcernsoffersofextrinsicrewardsforbehaviorthatisalreadyextrinsicallymotivatedratherthan
theeffectsofextrinsicrewardsonintrinsicallymotivatedbehavior.

NOTES
1.Therewardingaspectsofmusicalsoundsmaydependonalearninghistoryasdifferentlistenersandmusicianswillpreferdifferentmusicalsounds,includingclassical,jazz,rock,
andothergenres.
2.Forexample,negativecontrast,seeWilliams,1983.
3.Adifferencebetweensociallearningtheoryandabehavioralapproachisthatfromasociallearningperspective,thoughtsandbeliefsaboutrewardsareseenasthemajor
determinantsofhumanbehaviorandtheinteractionofhumanbehaviorwithactualrewardcontingenciesisdownplayed.
4.Bandura,1986,1997.
5.Bandura,1986,p.228.
6.RecentevidenceabouthumansensitivitytoreinforcementseemstocontradictBandurasclaimsthatawareness(i.e.,verbalizationofrewardcontingency)isnecessaryforrewards
andotherconsequencestoaffectbehavior.InanexperimentconductedbySvartdal(1992),humanparticipantsshowedsensitivitytoreinforcementwhenawareness(and,
presumably,expectancy)wasreducedbyadistractionprocedure.Usingacomplexdiscriminationtaskasadistractor,theresearcherprovidedreinforcementbasedontheforceofa
keypressingresponse.Participantsincreasedordecreasedtheforceoftheirkeypressinginaccordwiththerequirementsforreinforcement.
Sensitivitytothereinforcementprotocoloccurredeventhoughparticipantswereunawareofthelinkbetweenforceofresponseandreinforcement.Inaddition,greaterdistraction
(i.e.,complexityofthediscriminationtask)increasedthecontrolbythereinforcementcontingency.Overall,thedatafromSvartdal(1992)supportabehavioralview,indicatingthat
reinforcementhasdirecteffectsonhumanbehavior.Moreover,awarenessseemstointerferewith,ratherthanpromote,theregulationofhumanbehaviorbyreinforcement.
7.Bandura,1986,p.240.
8.Notethatthedistinctionbetweennaturalandarbitraryconsequencescomesfrombehavioralpsychology,asdoesthenotionofautomaticreinforcement.
Page82

9. SeeEplingandPierce,1991/1992.
10.Bandura,1997,p.219.
11.Bandura,1986,p.241.
12.SeeEisenberger,Rhoades,andCameron,1999,forevidenceonthemediationhypothesis.
13.Bandura,1986,p.245.
14.Deci,Koestner,andRyan,1999.
15.AlsoseeRosenfield,Folger,andAdelman,1980.
16.Bandura,1986,p.246.
17.HarackiewiczandSansone(2000)presentatheoreticalanalysisofrewardandintrinsicmotivationthatissimilartoBanduras(1986)analysisofcompetencycontingentreward.
HarackiewiczandSansonearguethatwhenrewardsareofferedtomeetaperformancestandard,theymaygeneratepositivemotivationalprocessesasindividualsapproachand
performactivitieseagertoattaincompetence(p.87).Suchrewardsaresaidtoleadtofeelingsofaccomplishmentandpride.Undertheseconditions,therewardsignifiesthevalue
ofcompetenceinthesituation.Rewardsthatsymbolizehigherlevelsofachievementareexpectedtogenerategreaterpositivemotivation.Arewardofferedandgivenforplacing
firstinagolftournamentissaidtosymbolizehigherlevelsofcompetencethanarewardgivenforfinishingthetournamentinthetopthree.Fromthisperspective,thesamereward
willhavegreatervalueinanationalgolftournamentthaninalocalone.Incontrasttothepredictionsofcognitiveevaluationtheory,HarackiewiczandSansone(2000)suggestthat
sometypesofrewardsleadtogreaterintrinsicmotivation.
18.Schunk,1983.
19.Bandura,1997.
20.Bandura,1986,p.246.
21.SeeHarackiewicz,Manderlink,andSansone,1984.
22.Bandura,1997,p.222.
23.EnzleandRoss,1978.
24.KarniolandRoss,1977.
25.Eisenberger,Rhoades,andCameron,1999.
26.Bandura,1997,p.220.
27.BanduraandSchunk,1981.
28.Skinner,1969.
29.Dickinson,1989Horcones,1983,1987VaughanandMichael,1982.
30.Harlow,1950Harlow,Harlow,andMeyer,1950.
31.Bandura,1986.
32.FeingoldandMahoney,1975.
33.Carton,1996.
34.Takingtheexperimenteroutofthesettingisnotanexplicitsignalofrewardwithdrawal,andtheinterpretationofthestimuluschangecouldvaryoverparticipantsandexperiments.
35.Cognitiveevaluationtheoryrequiresthattherebenoexpectationofrewardsduringthefreechoiceperiod.Thus,insomestudies,participantsaretoldthatrewardsarenolonger
forthcoming.
36.Dickinson,1989.
37.Skinner,1974,p.219italicsintheoriginal.
38.SeeKohn,1993a,whoarguesthatrewardsarenecessarilypunishers.
Page83

39.Dickinson,1989.
40.DeciandRyan,1985,p.300.
41.Ibid.,p.310.
42.Sidman,1989.
43.Skinner,1953,p.321.
44.ThisargumentconcerningfreedomanddignityisbasedonSkinner(1971).
45.Dickinson,1989,p.6.
46.Otherbehavioralexplanationsfortherewardandintrinsicmotivationeffectshavebeenofferedbyavarietyofresearchers.Onebehavioraleffectthatcouldbemistakenforachange
inintrinsicinterestinvolvesnegativebehavioralcontrast(Balsam&Bondy,1983Dunham,1968Williams,1983).Ifthedeclineinperformanceduringafreechoiceperiodisa
contrasteffect,theeffectsofrewardshouldbeatemporaryreactiontotherewardsremoval(Daly,1969a,1969bFlora,1990White&Cameron,2000).
Anotheralternativeisthatofferedrewardsinterferewithtaskmotivationbyelicitingcompetingresponses(Reiss&Sushinsky,1975,1976).Competingresponsesmayinclude
perceptualorcognitivedistraction,excitementrelatedtoanticipatedrewards,orfrustrationduetodelayofreward.Theseresponsescouldinterferewithtaskmotivation.Recallthat
experimentsusingrepeatedrewardsdidnotfindalossofintrinsicmotivation.Onewaytoexplainthisfindingisintermsofthereduction,throughhabituation,ofcompeting
responseselicitedbyofferedrewardsastherewardsarepresentedrepeatedly.
Alearnedhelplessnessexplanationhasalsobeenofferedforthedetrimentaleffectsofreward(Eisenberger&Cameron,1996).Learnedhelplessnesstheoryassumesthat
uncontrollablenegativeconsequencesresultingeneralizedmotivationalandemotionaldeficits(MaierandSeligman,1976OvermierandSeligman,1967Overmier,1985Seligman,
1975).Thetheoryhasbeenappliedtorewardsandnoncontingentrewardpresentationsandhasbeenfoundtoproduceperformancedeficitsinchildren(Eisenberger,Kaplan,and
Singer,1974Eisenberger,Leonard,Carlson,andPark,1979).Whenappliedtorewardsandintrinsicmotivation,thelearnedhelplessnesshypothesisstatesthatwhenrewardsare
offeredunrelatedtoanyperformancestandards,peoplelearnthatthedeliveryoftherewardhasnothingtodowiththeirperformanceandsogiveuponthetask.
47.Carton,1996.
48.SeeCartonandNowicki,1998.
Page84

Thispageintentionallyleftblank.
Page85

PARTIV
THEEMPIRICALEVIDENCEFORTHEIMPACTOF
REWARDSONINTRINSICMOTIVATION
Page86

Thispageintentionallyleftblank.
Page87

Chapter6
AnOverviewoftheExperimentsonRewardsandIntrinsicMotivation

FollowingtheworkofDeciandLepperandhisassociates,1manyresearcherssetouttotestthehypothesisthatrewardsproduceddecrementaleffectsonpeoples
intrinsicmotivation.Since1971,almost150controlledexperimentshavebeenconductedonthetopic.Inthissectionofthebook,weexaminethislargebodyof
empiricalresearch.Inthischapter,weoutlinethevariousresearchparadigmsusedtoassesstheeffectsofrewardsonintrinsicmotivationanddescribedifferent
conditionsthathavebeeninvestigatedaspotentialmoderatorsofrewardeffects.InChapter7,weevaluatemetaanalyticreviewsoftheliterature.Ametaanalysis
basedonourrecentworkispresentedinChapter8.2Ourmetaanalysisisdesignedtoupdateandimproveonpreviousreviewsoftheliterature.Theoreticaland
practicalimplicationsofourfindingsarepresentedinChapter9.
ThemajorityofstudiesonrewardsandintrinsicmotivationhaveusedabetweengroupsdesignsimilartothedesignsusedintheoriginalstudiesdescribedinChapter
2.Thebulkofthebetweengroupexperimentshavebeendesignedfromanoverjustificationorcognitiveevaluationtheoreticalperspective.Thefocusofthesestudies
hasbeentoshowwhenrewardsproducedecrementsinintrinsicmotivation.Afewresearchershaveinvestigatedthetopicwithsinglesubjectdesignslikethatof
FeingoldandMahoney(outlinedinChapter3).3Thesestudiesweredesignedtoassesswhetherrepeatedpresentationsofrewardfollowedbyrepeatedassessmentsof
intrinsicmotivationfollowingtheremovalofrewardwouldresultindecrementsinintrinsicmotivation.Whatfollowsisabriefdescriptionofthesedifferentresearch
designsandadiscussionofrewardconditionsthathavebeenfoundtoproducevaryingeffects.
Page88

BETWEENGROUPDESIGNS

Thevastmajorityofstudiesdesignedtoassesstheeffectsofrewardonintrinsicmotivationhavebeenconductedinalaboratorysettingusingabetweengroupdesign.4
Inatypicalstudy,participantsarepresentedwithatask(e.g.,solvingandassemblingpuzzles,drawingwithmagicmarkers,playingwordgames).Experimental
participantsarerewardedwithmoney,grades,candy,praise,opportunitytoengageinapreferredactivity,goodplayercertificates,andsoforthforperformingthe
activity.Rewardsareeithertangible(e.g.,money,candy,goldstars)orverbal(e.g.,praise,approval,positivefeedback).Inaddition,therewardsmaybeoffered
beforehand(expectedreward)orpresentedunexpectedlyaftertheactivity(unexpectedreward).Insomeexperiments,rewardisofferedsimplyfordoinganactivityin
otherstudies,therewardsaregivenforcompletingataskorforeachpuzzleorotherunitsolved.Inanumberofexperiments,therewardsareofferedformeetingor
exceedingaspecificstandard.
Participantsinacontrolconditionengageintheactivitywithoutreceivingareward.Therewardinterventionisusuallyconductedoveratenminutetoonehour
period.Rewardedandnonrewardedgroupsarethenobservedduringanonrewardperiod(typically,anywherefromtwominutestoonehour),inwhichparticipantsare
freetocontinueperformingthetargettaskortoengageinsomealternativeactivity.Thenonrewardperiodusuallyoccursimmediatelyaftertheexperimentalsession,
althoughsomeresearchershaveobservedparticipantsseveralweekslater.Thetimeparticipantsspendonthetargetactivityduringthisnonrewardphase,their
performanceonthetaskduringthefreechoiceperiod,orselfreportedtaskinterestareusedasmeasuresofintrinsicmotivation.Ifrewardedparticipantsspendless
freetimeontheactivity,performatalowerlevel,orexpresslesstaskinterestthannonrewardedparticipants,rewardissaidtoundermineintrinsicmotivation.

TheImpactofModeratorVariables
Thefindingsfromthebetweengroupdesignstudieshavebeenmixed(positive,negative,andnoeffectshavebeenreported).Fromthebeginning,whenDeciand
Lepperandhisassociatesconductedtheinitialstudiesonthetopic,itwasclearthatrewardshaddifferenteffectsdependingonhowtheywereadministered.5
Specifically,verbalrewardswerefoundtoincreasemeasuresofintrinsicmotivation,andtangiblerewardsproducednegativeeffectsonlywhentheywereoffered
beforehandandnotcontingentonaspecifiedlevelofperformance.
Asresearchonthetopicwasamassed,otherconditionswerefoundtoinfluencetheimpactofrewards.Anumberofreviewsappearedonthetopictheiraimwasto
assesstheadequacyofdifferenttheoreticalaccountsandtoidentifyconditionsunderwhichrewardsproducedecrementsandincrementsinintrinsic
Page89

TYPICALBETWEENGROUPONREWARDSANDINTRINSICMOTIVATION
Page90

motivation.6Nextwedescribesomeoftheconditionsthoughttobecriticalindeterminingtheimpactofrewards.

MODERATORVARIABLESFORTHEIMPACTOFREWARDSONINTRINSICMOTIVATION
Rewardtype(verbalortangible)
Rewardexpectancy(expectedorunexpected)
Rewardcontingency

Cognitiveevaluationtheory

Tasknoncontingent
Taskcontingent
Completioncontingent
Performancecontingent

Sociallearningtheory

Noncompetencycontingent(loose)
Competencycontingent(specific)

Maximumversuslessthanmaximumreward
Rewardsalience
Initialtaskinterest
Interpersonalcontext

TypeofReward
Intheintrinsicmotivationliterature,rewardsaredefinedasverbalortangible.Verbalrewards(praise,positivefeedback)havegenerallybeenfoundtoincrease
measuresofintrinsicmotivation.Inoneoftheoriginalexperiments,Deciexaminedtheeffectsofverbalrewardsandfoundthatparticipantswhowerepraisedfordoing
ataskspentmoretimeontheactivityafterthepraisewasnolongeravailablethannonrewardedparticipants.7Accordingtocognitiveevaluationtheory,allrewardsare
experiencedascontrollingbyindividuals,butverbalrewardsprovideaninformationalfunctionthatoverridesthefeelingsofcontrol.Inthemostrecentstatementof
cognitiveevaluationtheory,Deciandhiscolleaguessuggestthatverbalrewardsdonotalwaysresultinincreasesinmotivation.8Specifically,theresearcherssuggest
thatverbalrewardsaremoreeffectiveforadultsthanforchildrenandthatverbalpraisepresentedinacontrollingmannerwillreducetheinformationalfunctionofthe
rewardandleadtoadecreaseinintrinsicmotivation.
Rewardsclassifiedastangibleincludemoney,candy,goldstars,goodplayerawards,theatertickets,opportunitiestoengageinpreferredactivities,andsoon.The
resultsofstudiesusingtangiblerewardsaremixedsomestudiesfoundpositiveeffects,whereasothersfoundnegativeornoeffects.Anexaminationofthestudiesusing
tangiblerewardsindicatesthattheeffectsoftangiblerewardsdependonotherconditionssuchastherewardexpectancyorcontingency.
Page91

RewardExpectancy
Intheintrinsicmotivationstudies,rewardsareclassifiedasexpectedorunexpectedthesetermsrefertowhethertheresearchparticipantsarepromisedthereward
priortoitsdelivery.Forexample,insomeexperiments,rewardedparticipantsareofferedmoneybeforetheyengageintheexperimentalactivityandtherewardis
deliveredaftertheyhaveworkedonthetask.Inotherstudies,theparticipantsaregiventherewardaftertheyhaveengagedinthetaskbutarenotoffereditpriorto
theirinvolvement.Rewardsthatarepromisedtoparticipantsarereferredtoasexpectedrewardsunexpectedrewardsarethosedeliveredattheendofthe
experimentalsessionbutnotpromisedbeforehand.
AsdescribedinChapter2,intheearlystudybyLepperandhisassociates,onlyexpectedtangiblerewardswerefoundtoproduceanegativeeffectonthefreetime
measureofintrinsicmotivationunexpectedrewardhadnoimpact.9Bothcognitiveevaluationtheoristsandthosewhosupporttheoverjustificationhypothesissuggest
thatunexpectedrewardsdonotaffectfeelingsofcompetence,selfdetermination,orlocusofcontrolbecausethecontrollingprocessonlytakesplacewhenthe
rewardsareinprogress.Withunexpectedrewards,individualsdonotknowthattheywillreceiveareward,andthus,intrinsicmotivationwillnotbeaffected.
FollowingLepperandhisassociatesfindingwithregardtounexpectedrewards,fewstudieshavebeenconductedtoexaminetheimpactofthistypeofreward.
Instead,thefocushasbeenonexpectedtangiblereward.Fromtheperspectiveofcognitiveevaluationtheory,expectedtangiblerewardsdecreasepeoplesintrinsic
motivationbyunderminingfeelingsofcompetenceandselfdetermination.Proponentsoftheoverjustificationhypothesisclaimthatexpectedtangiblerewardsdecrease
intrinsicmotivationbydeflectingthesourceofmotivationfrominternaltoexternalcauses.Fromboththeseviewpoints,expectedtangiblerewardsarepredictedto
reduceintrinsicinterest.Moreover,mostbetweengroupdesignstudieshavefoundnegativeeffectsofexpectedtangiblereward.However,afewresearchershave
shownthatwhentherewardsaretiedtoperformancestandards,intrinsicmotivationincreases.10Theimplicationofthisfindingisthatnotallexpectedtangiblerewards
producenegativeeffectstheoutcomedependsonthecontingency.

RewardContingency
Acontingencyofrewardstatesthebasisonwhichrewardsaredelivered(arrangementofrewards).Forexample,rewardsmaybegivenforattainingaspecificscore
onatask,completingapuzzle,orsimplydoinganactivity.
Theclassificationofrewardcontingencieshasbeensomewhatconfusingintheintrinsicmotivationliterature.Thedifficultyisthatsomeresearchershaveusedthe
samelabeltodescribedifferentcontingenciesordifferentlabelstodescribethesamecontingencies.Forexample,inonestudytheresearchersde
Page92

scribeaprocedurewherebyparticipantsarepaidtoparticipateinataskasnoncontingent11thesameprocedureislabeledascontingentinotherstudies.
Usingcognitiveevaluationtheoryastheirframework,in1983,Ryanandhiscolleaguesdevelopedataxonomyofrewardcontingenciesthat,theysuggested,could
bringordertotheempiricaldataregardingexpectedtangiblerewards.12Inaslightalterationofthistaxonomy,Deci,Koestner,andRyanclassifiedexpectedtangible
rewardsintofourcategories:tasknoncontingent,engagementcontingent,completioncontingent,andperformancecontingentrewards.13Tasknoncontingentrewards
aredefinedasthosegivenwithoutspecificallyrequiringthepersontoengageintheactivity14engagementcontingentrewardsarethoseofferedtoparticipantsfor
engaginginataskwithoutarequirementtocompletethetask,doitwell,orreachsomestandard.Rewardscategorizedascompletioncontingentarethoseofferedand
givenforcompletingatask,andperformancecontingentrewardsaredefinedasthoseoffereddependentupontheparticipantslevelofperformance.15
AccordingtoDeciandhisassociates,negativeeffectsshouldbedetectedwhentheexpectedtangiblerewardsarecontingent(onlytasknoncontingentrewardsare
predictedtoproducenoeffect).AsindicatedinChapter4,engagementcontingent,completioncontingent,andperformancecontingentrewardsaresaidtobe
controllingpeoplereceivingthesetypesofrewardswillfeelalossofautonomy.Theirselfdeterminationwilldiminish,andthis,inturn,willleadtoalossofintrinsic
motivation.Completioncontingentandperformancecontingentrewardsmayprovideinformationaboutcompetence,butaccordingtocognitiveevaluationtheory,
reductionsinselfdeterminationwilloverrideanyincreasesinfeelingsofcompetence.Thus,eachoftheserewardcontingenciesisexpectedtoresultinalossofintrinsic
motivation.16
AlthoughDeciandhisassociatesmadethesepredictions,itisimportanttonotethatseveralstudiesusingcompletioncontingentandperformancecontingentrewards
didnotresultindecreasesinmeasuresofintrinsicmotivation.17EvenDecihimselfusedacompletioncontingentrewardconditioninhisearlystudiesandfoundno
significantnegativeeffectsofexpectedtangiblereward.18
Adifferentconceptualizationofrewardcontingenciesissuggestedbysociallearningtheory.19AsoutlinedinChapter5,Bandurassociallearningtheorydistinguishes
betweennoncompetencycontingentrewards(rewardsgivenwithoutregardtomastery),andcompetencycontingentrewards(rewardsgivenformasteringactivities
ortasks).Incontrasttocognitiveevaluationtheory,inwhichtheemphasisisoncontrolandperceivedselfdetermination,sociallearningtheoryemphasizeshowreward
contingenciesrelatetoperceivedcompetenceorselfefficacy.Rewardcontingenciesthatenhanceperceivedcompetenceorselfefficacyareexpectedtoincrease
interestandperformanceofanactivity.
Noncompetencycontingentrewardsarethosegivenformerelydoing,completing,orrepeatinganactivityandaremostlikelytoreduceintrinsicinterest.Fromthe
perspectiveofsociallearningtheory,inthiskindoflooserewardprocedure,therewardsdonotconveycompetencytoanindividual.Inaloose
Page93

contingency,therewardsaregivenwithoutregardtoqualityofperformance,andassuch,theyexertaweakinfluenceoverbehavior.Thus,otherfactorsareexpected
tomoderatetheeffectsofloose(ornoncompetencycontingent)rewards,includingthelevelofinitialtaskinterest,abilityonthetask,valueoftherewards,typeof
task,ageoftheparticipants,andwhenandhowintrinsicmotivationismeasured.
Competencycontingentrewards,ontheotherhandinvolverewardsgivenformastery(i.e.,achievingrelativelychallengingstandards).Thistypeofreward
contingencyissaidtodevelopperceptionsofselfefficacyandtaskinterest.Rewardsgivenforachievingchallengingstandardsarealsoindicativeofcompetence.Thus,
accordingtosociallearningtheory,whenrewardsarelinkedtoattainingchallengingperformancestandards,theseincentiveshelppeopledevelopasenseofself
efficacy,interest,andinvolvementinactivities.
Almostallthebetweengroupdesignstudiesonrewardsandintrinsicmotivationhaveusednoncompetencycontingentrewards.ManyoftheexperimentsthatDeci
andassociatesclassifiedseparatelyastaskcontingent,completioncontingent,andperformancecontingentrewardswouldbeclassifiedinthenoncompetency
contingentcategoryaccordingtosociallearningtheory.OnlyahandfulofstudieshaveusedwhatBandurahasreferredtoascompetencycontingentrewardinthese
typesofstudies,participantsarerewardedformeetingorexceedingaspecificperformancestandard.Todate,nostudiesonthetopichaveexaminedtheimpactof
rewardgivenformeetingincreasinglymoredifficultstandardsandattainingmasteryonatask.Thus,fromtheperspectiveofsociallearningtheory,researchonthe
effectsofrewardsonintrinsicmotivationhasfocusedalmostexclusivelyontheeffectsofnoncompetencycontingentrewards,withnosystematicinvestigationofthe
typesofrewardproceduresthatleadtoincreasedperformanceandmotivation.

MaximalversusLessThanMaximalReward
Afurtherpotentialmoderatorofrewardconcernswhetherparticipantsreceivethemaximumamountofrewardpossibleintheexperiment.Instudiesofmaximum
reward,participantsareofferedrewardsgradedintermsofmeetingacriterionorperformancestandardallmeetthecriterionandreceivethefullamountofreward.
Lessthanmaximumrewardoccursinstudieswherethereisatimelimit,suchthatsomeparticipantsareunabletomeetalltherequirementsinthetimeallottedandare
givenlessthanthefullreward.Forexample,Decisoriginalexperimentinvolvedlessthanmaximumreward.20Participantswereoffered$1foreachoffourpuzzles
solvedwithina13minutetimelimit.Notallparticipantswereabletosolvethepuzzleswithinthetimelimit,andtherefore,notallreceivedthefullreward.
Althoughmaximumversuslessthanmaximumrewardhasnotbeenafocusoftheoreticalpredictionsinthisliterature,thedifferencebetweenthesetwoconditions
mayleadtodifferenteffectsonintrinsicmotivation.21Onewaytounderstandtheseconditionsistoconsiderwhatlessthanmaximalrewardsig
Page94

nifiestoparticipants.Ifpeoplearetoldthattheycanobtainacertainlevelofrewardbutaregivenlessthanthatlevel,theyhavereceivedfeedbackinformationthat
indicatesfailure.Inotherwords,thistypeofsituationmayrepresentfailurefeedback,notreward.Ontheotherhand,ifpeoplereceivethefullreward,thereward
signifiessuccessatthetask.Advocatesofcognitiveevaluationtheoryandoverjustificationtheorydonotmakedifferentiatedpredictionsabouttheeffectsofthese
differentconditions.22Forsociallearningtheorists,ontheotherhand,failureatataskshouldreduceinterest,whereassuccessshouldincreasefeelingsofcompetence,
selfefficacy,andinterest.23

RewardSalience
Theoverjustificationhypothesisproposesthatwhenindividualsarerewardedforengaginginaninterestingactivity,theirperceptionsshiftfromaccountingfortheir
behaviorasselfinitiatedtoaccountingforitintermsofexternalrewards.Lepper,Greene,andNisbett(1973)suggestedthatexpectedtangiblerewardsareespecially
detrimentaltointrinsicmotivationwhentheyareparticularlyconspicuousorsalient.TheissueofsalientversusnonsalientrewardwasaddressedinastudybyRossin
themid1970s.24Whenrewardsweremadeconspicuoustoparticipants(children),thosewhowereofferedtherewardshowedadecreaseinintrinsicmotivation,
whereasrewardsthatwerenotsalientledtoanincrease.Inmostbetweengroupdesignstudies,therewardsaremadesalienttotheparticipants.

LevelofInitialTaskInterest
Amajorissueintherewardsandintrinsicmotivationexperimentsconcernswhetherthetasksusedinthestudiesareinterestingtotheparticipants.Fromtheperspective
ofcognitiveevaluationtheoryandtheoverjustificationhypothesis,theeffectofrewardsonpeoplesintrinsicmotivationisonlyrelevantiftheactivitiestheparticipants
engageinareofhighinitialinterest(seethediscussioninChapter4).Thatis,theconcernisthatifindividualsaregivenrewardsfortasksthattheyalreadyenjoydoing,
removaloftherewardswillcausethemtobelessmotivatedandlessinterestedinthetaskthantheywerebeforetherewardswereintroduced.Intermsofinitialtask
interest,Deciandhisassociateshavesuggestedthattheunderminingphenomenonisrelevantonlytointerestingtasksbecausethereisnointrinsicmotivationto
underminewithtasksoflowinitialinterest.25
Sociallearningandbehavioralinvestigators,ontheotherhand,areinterestedintheeffectsofrewardsonmotivationandperformancewhenpeoplearerewardedto
engageintaskswithlowlevelsofinitialinterest.Fromasociallearningperspective,rewardscanbeusedtocultivatetaskinterestandtobuildcompetenceandself
efficacy.26
Theissueofinitialtaskinterestisofparticularrelevancetothepracticalapplicationofreward.Inthelaboratoryexperiments,negativeeffectsofrewardhaveonly
beenfoundwhenparticipantsengageinactivitiesthatareofhigh
Page95

initialinterest.Whenthetasksarenotinitiallyinteresting,itappearsthattheintroductionofrewardseitherhasnoeffectorelseenhancesmeasuresofmotivationand
performance.Ineducationalsettings,teachersusuallyuserewardandincentivesystemstoinstillinterestinacademicsubjectsinwhichstudentshavelittleinitialinterest.
Evidencefromlaboratoryexperimentssuggeststhatrewardsarenotharmfulinthistypeofsituation.Rewardshaveonlybeenfoundtoproducenegativeeffectsfor
tasksofhighinterest.Asnotedbyseveralresearchers,teachersarenotlikelytosetuprewardsystemsforactivitiesthatstudentsalreadyenjoy.27

InterpersonalContexts
Cognitiveevaluationtheoristsstatethattheeffectsofrewards(bothverbalandtangible)dependontheinterpersonalcontext(theatmosphereinclassrooms,work
environments,etc.).28Whenpeoplefeelpressuredorcontrolledbythecontext,theviewisthatrewardswillundermineintrinsicmotivation.Incontrast,ifthosewho
administerrewardsdosoinanoncontrollingstyle,therewardsaresaidtobeexperiencedasmoreinformationalandtoleadtolessofalossofintrinsicmotivation.

OtherVariablesThatModeratetheImpactofRewards
Severalothervariableshavebeenhypothesizedtoaffecttheimpactofrewardsonmeasuresofintrinsicmotivation.Researchershaveinvestigatedtheeffectsofreward
magnitude,29whethertherewardsareattractivetoparticipants,30whethertherewardsareendogenousorexogenoustoanactivity,31whethertherearenormsabout
payment,32andwhetherparticipantsaretoldpriortothemeasurementofintrinsicmotivationthatrewardisnolongeravailable.33Presentlytherearefewstudiesthat
havesystematicallystudiedthesevariablesthus,conclusionsabouttheirimpactatthispointintimewouldbepremature.

MeasuresofIntrinsicMotivation
Researchersinvestigatingtheeffectsofrewardshaveusedavarietyofmeasurestoindexintrinsicmotivation.Themainmeasureshaveincludedtimeontaskduringa
nonrewardedfreechoiceperiodperformanceonthetaskduringthefreechoiceperiodquestionnairemeasuresoftaskinterest,enjoymentorsatisfactionand
participantswillingnesstovolunteerforfuturestudieswithoutreward.Someresearchershaveusedafreechoicemeasure,whichincludestimeontaskaswellas
performanceonthetaskduringthefreechoicesession.Oneachofthesemeasures,rewardedparticipantsarecomparedtononrewardedindividuals.Ifrewarded
participantsspendlessfreetimeontheactivity,reportlesstaskinterest,performatalowerlevelduringthefreechoiceperiod,orarelesswillingtoreturnforfuture
studiesthanthecontrolgroup,rewardissaidtoundermineintrinsicmotivation.
Page96

MEASURESOFINTRINSICMOTIVATION
Freetimeontask
Performanceontaskinfreechoiceperiod
Taskinterest,enjoyment,orsatisfaction
Willingnesstovolunteerforfuturestudieswithoutreward
Freechoicemeasureacombinedmeasurebasedonfreetimeandperformanceduringfreechoiceperiod

Severalresearchershavenotedthattheresultsacrossdifferentmeasuresofintrinsicmotivationarenotconsistent.Forexample,inmanyexperimentsanegativeeffect
ofrewardwasfoundonthefreetimemeasureofintrinsicmotivationbutnotonquestionnairemeasuresoftaskinterest.34Suchresultsindicatethatthereareproblems
indefining(operationalizing)theconstructofintrinsicmotivation.Oneimplicationofthisisthatfutureresearchneedstofocusonclarifyingtheconceptofintrinsic
motivationanddevelopingsuitablemeasures.Presently,thestudiesshowthatdifferentresultsareobtainedwithdifferentmeasuresofintrinsicmotivation.

SINGLESUBJECTDESIGNS
Theimpactofavarietyofmoderatorvariablesontheeffectsofrewardsonpeoplesintrinsicmotivationhasbeeninvestigatedwithbetweengroupdesigns.One
criticismofthisdesign,however,isthatresearchersemployingitoftenrefertotheirrewardmanipulationasareinforcementprocedure.AsdiscussedinChapter3,by
definition,areinforcerisaneventthatincreasesthefrequencyofthebehavioritfollows.However,inmostofthebetweengroupdesignstudiesonintrinsicmotivation,
theresearchersdidnotdemonstratethattheeventsusedasrewardsincreasedthefrequencyofthebehaviorstudied.Inaddition,criticshavesuggestedthatthe
measurementphasesinthebetweengroupsdesignresearcharetoobrieftodetectanytemporaltrendsandtransitionstates.35Inordertoaddresstheseissues,afew
studieshavebeenconductedusingarepeatedmeasures,singlesubjectdesign.
Inthesinglesubjectdesign,participantsserveastheirowncontrols.Measuressuchastimeontaskaretakenoveranumberofsessionsinabaselinephasebefore
presentationofthereward.Reinforcementproceduresarethenimplementedforrepeatedsessions,andfinally,therewardiswithdrawnandtimeontaskisassessedon
repeatedoccasions.Anychangeinintrinsicmotivationis
Page97

indexedbyadifferenceintimespentonthetaskbetweenthebaselineandpostreinforcementphases.

TYPICALSINGLESUBJECTDESIGNONREWARDSANDINTRINSICMOTIVATION

Theadvantageofthesinglesubjectdesignsisthattheresearchercandeterminewhethertherewardsbeingusedareactualreinforcersthatis,whetherbehavior
increasesduringthereinforcementphase.Statementscanthenbemadeabouttheeffectsofreinforcementratherthanreward.Inaddition,therepeateduseofreward
followedbyextendedtaskperformancewithoutrewardismorecharacteristicofthenaturalenvironmentthanthetypicalshorttermexperimentalprocedureusedinthe
betweengroupdesigns.
Fivestudiesemployingasinglesubjectdesignwithvarioustypesoftangiblerewardcontingenciesreportedthatparticipantsperformanceonthetaskduringthe
postrewardphaseeitherexceededorremainedatthesamelevelasperformanceintheprerewardsessions.36Thesefindingsindicatethatwhenrewardsareshownto
functionasreinforcementandmultipletrialsproceduresareused,thereisnoevidenceofadecrementaleffectofreward.

SUMMARY
Agreatdealofresearchhasbeenconductedusingabetweengroupsdesigntoinvestigatenegativeeffectsofrewardsonpeoplesintrinsicmotivation.Sincetheinitial
studiesintheearly1970s,thefindingshaveindicatedthatnotallrewardsleadtonegativeeffects.Resultsfromtheexperimentalliteraturesuggestthatnegativeeffects
dependontherewardtype,therewardexpectancy,therewardcontingency,whetherparticipantsreceivemaximumorlessthanmaxi
Page98

mumreward,levelofinitialtaskinterest,theinterpersonalcontextinwhichrewardsaredelivered,andhowintrinsicmotivationismeasured.
Findingsfromsinglesubjectdesignsfurthersuggestthatnegativeeffectsdonotoccurwhentherewardsthatareusedfunctionasreinforcementandwhenrepeated
presentationsofrewardarefollowedbyrepeatedassessmentsofmeasuresofintrinsicmotivation.Thatis,negativeeffectsarefoundtobetemporary.
Acloseinspectionofallthestudiesonrewardsandintrinsicmotivationrevealscontradictions.Notallstudiesofverbalrewardhaveproducedpositiveeffectsnotall
studieshaveshownnoeffectsofunexpectedrewards,andnotallstudiesofexpectedtangiblerewardcontingencieshaveproducednegativeeffects,aspredictedby
cognitiveevaluationtheoryandtheoverjustificationhypothesis.Inaddition,Deciandhiscolleagueshavenotacceptedthefindingsofthesinglesubjectdesignsand
insteadhavesuggestedthatthenegativeeffectsofrewardareenduringratherthantemporary.37
Giventhediversepredictionsandfindingsconcerningrewardsandintrinsicmotivation,anumberofreviewershaveattemptedtodelineatethespecificconditions
underwhichrewardsareeitherharmfulorbeneficial.38Inaddition,afewresearchershaveusedmetaanalytictechniquestoassesstheliterature.Inthenextchapter,
weprovideacriticalexaminationofmetaanalysesonthistopic.

NOTES
1.Deci,1971,1972a,1972bLepper,Greene,andNisbett,1973.
2.Cameron,Banko,andPierce,2001.
3.FeingoldandMahoney,1975.
4.Researchersusingabetweengroupsdesigntypicallyemployoneoftwomethods:abeforeafterdesignoranafteronlydesign.Thebeforeafterdesigninvolvesathreesession
paradigm,inwhichabaselinemeasureofintrinsicmotivationonaparticulartaskistaken.Thisentailsmeasuringtimeontaskintheabsenceofextrinsicreward,usuallyfroma
sessionofshortduration(e.g.,tenminutes).Participantsarethenassignedtoeitherarewardornoreward(control)condition,andaninterventionwithextrinsicrewardsiscarried
out.Followingthis,therewardiswithdrawnandtimeontaskisagainmeasured.Theprocedureisidenticalforbothgroupsexceptthatcontrolparticipantsdonotexperiencethe
rewardinterventioninthesecondsession.Meandifferencesintimeontaskbetweenpreandpostinterventionarecalculatedforeachgroup,andthescoresfortheexperimentaland
controlparticipantsarethenstatisticallycompared.Anydifferencebetweenthetwogroupsisconsideredevidenceoftheeffectsofwithdrawalofreward.
Mostresearchershaveusedanafteronlydesign,inwhichnobaselinemeasureofintrinsicmotivationistaken.Experimentalparticipantsarepresentedwithataskandrewardedfor
performingtheactivitynorewardisofferedorgiventocontrolparticipants.Allgroupsarethenobservedduringanonrewardperiod.Ifexperimentalparticipantsspendlesstimeon
thetask(duringthepostrewardobservation)thanthecontrols,reinforcement/rewardissaidtoundermineintrinsicmotivation.
5.Deci,1971,1972a,1972bLepper,Greene,andNisbett,1973.
6.Bates,1979Bernstein,1990Dickinson,1989Flora,1990Morgan,1984.
Page99

7. Deci,1971,Experiment3.
8. Deci,Koestner,andRyan,1999.
9. Lepper,Greene,andNisbett,1973.
10.Forexample,seeEisenberger,Rhoades,andCameron,1999.
11.HamnerandFoster,1975.
12.Ryan,Mims,andKoestner,1983.
13.Deci,Koestner,andRyan,1999.
14.Ibid.,p.636.
15.Ibid.
16.Ibid.
17.Forexample,seeHarackiewicz,Manderlink,andSansone,1984.
18.Deci,1971,1972b.
19.Bandura,1986.
20.Deci,1971.
21.Deci,Koestner,andRyan(1999)analyzedsomestudiesthatusedlessthanmaximumversusmaximumrewards.Theirfindingsindicatedthatexpectedtangiblerewardsproduced
largenegativeeffectsonmeasuresofintrinsicmotivationonlywhenparticipantsreceivedthelessthanmaximumreward.
22.Lepper,Keavney,andDrake(1996)suggestedthattheremaybedifferenteffectsonmotivationdependingonwhetherpeoplearesuccessfulatobtainingthereward(seethe
discussionofthisissueinChapter4).Theydidnot,however,makeanyspecificpredictionsinstead,theirtheoreticalpositionwasinaccordwithcognitiveevaluationtheory,which
predictsnegativeeffectsforallthecontingentrewardconditionsstudiedinthisliterature(taskcontingent,completioncontingent,andperformancecontingent).
23.Dickinson(1989)alsopointsoutthatnegativeeffectsarenotfoundwhenrewardsaretiedtosuccess.
24.Ross,1975.
25.Deci,Koestner,andRyan,1999,p.633.
26.Bandura,1986.
27.Forexample,seeDickinson,1989.
28.Deci,Koestner,andRyan,1999.
29.SeeNewmanandLayton,1984.
30.Williams,1980.
31.Kruglanski,Riter,Armitai,Margolin,Shabatai,andZaksh,1975.
32.Staw,Calder,Hess,andSandelands,1980.
33.CartonandNowicki,1998.
34.Forexample,seeWicker,Brown,Wiehe,andShim,1990.
35.FeingoldandMahoney,1975Mawhinney,1990.
36.DavidsonandBucher,1978FeingoldandMahoney,1975Mawhinney,Dickinson,andTaylor,1989Skaggs,Dickinson,andOConnor,1992Vasta,Andrews,McLaughlin,Stirpe,
andComfort,1978.
37.Deci,Koestner,andRyan,1999.
38.Bates,1979Bernstein,1990Carton,1996Dickinson,1989Flora,1990Morgan,1984Notz,1975.
Page100

Thispageintentionallyleftblank.
Page101

Chapter7
ACritiqueofMetaanalysesoftheEffectsofRewardsonIntrinsicMotivation
Overthepastthreedecades,alargebodyofexperimentalresearchhasaccumulatedontheeffectsofrewardsonpeoplesintrinsicmotivation.Althoughmany
researcherscontinuetoarguethatrewardsproducepervasivenegativeeffects,anexaminationoftheliteratureindicatesamixedsetoffindings.Asdescribedin
Chapter6,manydifferentvariablesimpacttheeffectofrewards.Inanattempttomakesenseofthediverseresults,someresearchershaveusedmetaanalytic
techniquestoevaluatetheliterature.Metaanalysisisastatisticaltechniqueforcombiningtheresultsfromalargenumberofstudiesonthesametopic.Itinvolvesthe
statisticalanalysisofalargecollectionofresultsfromindividualstudiestointegratethefindings.Inthischapter,wedescribethelogicandproceduresofmetaanalysis
anddiscussandevaluatefindingsfromdifferentmetaanalysesonthetopicofrewardsandintrinsicmotivation.

THETECHNIQUEANDLOGICOFMETAANALYSIS
Inmanyareasofthesocialsciences,dozensofstudieshavebeenconductedtoinvestigateasingleissue.Ratherthanconductadditionalstudies,severalresearchers
havepointedoutthatwhatisneededisawaytoorganizeandmakesenseofthevastamountsofdatathathavebeenamassedonatopic.1Inrecentyears,researchers
haveusedmetaanalysisasawaytoorganizeandunderstandanareainwhichalargeamountofdatahavebeencollected.
Normally,inanindividualstudy,aresearcherconductsanexperimentandcomparestheresultswithwhatwouldbeexpectedbychance.Thedifferencebetweenthe
resultandchanceisknownasthedeviation.Ifthereisaneffect,
Page102

thedeviationislargeenoughtorejectchance.Thesizeofthedeviation(calledtheeffectsize)dependsonthesizeofthesample.Asmalleffectsizeinalargesample
maybeinformative,butthesameeffectsizeinasmallsamplemaynot.Metaanalysisassessestheeffectsizesovermanystudiesonthesametopicthatdifferinsample
size.Itinvolvesacollectionofstatisticaltechniquesdesignedtoquantitativelysummarizeabodyofdatafromconceptuallyrelatedstudies.
ThetermmetaanalysiswascoinedbyGeneGlassinthe1970sGlassusedthetechniquetoorganizeandanalyzealargenumberofstudiesconductedin
psychotherapyandeducation.2Sincethe1980s,metaanalysishasrapidlyincreasedinimportanceinthebehavioralandsocialsciences.Overthepast30years,
researchsummariesbasedonmetaanalyseshavebecomevaluedsourcesofinformationforbothpolicymakersandresearchers.
Quantitativeanalysessimilartometaanalysishavebeenconductedonsinglesubjectdesigns3however,metaanalysisistypicallyusedwithbetweengroupdesigns
inwhichatreatmentgroup(e.g.,rewardedgroup)iscomparedtoacontrolgroup(nonrewardedgroup)onacommondependentmeasure(intrinsicmotivation).The
goalsofametaanalysisaretoestablishtherelationshipbetweenindependentanddependentvariables(inthiscase,therelationshipbetweenrewardsandintrinsic
motivation)andtodeterminewhatfactorsmoderateoralterthemagnitudeoftherelationship(e.g.,typeofreward,rewardcontingency,etc.).Conductingameta
analysisentailsspecifyingtheresearchquestions,delineatingthecriteriaforincludingandexcludingstudies,collectingallexperimentsthatmeetthecriteria,codingthe
studies,andcalculatingeffectsizesforeachstudy.

STEPSINVOLVEDINAMETAANALYSIS
Specifytheresearchquestions
Outlinecriteriaforincludingandexcludingstudies
Collectallexperimentsthatmeettheinclusioncriteria
Codestudiesforvariousrewardcharacteristics
Calculateaneffectsizeforexperimentalversuscontrolgroupsforeachstudy
Conductametaanalysistoassesswhetherthereisastatisticallysignificantdifferencebetweenexperimentalandcontrolconditions
Conductmoderatoranalysestoclarifyunderwhatconditionstheexperimentalandcontrolgroupsdifferoneffectsize

Inametaanalysis,theeffectsizefromeachstudybecomestheunitofanalysis,ratherthantheindividualparticipantswithinastudy.Iftheeffectsizesfromallthe
studiespresentarandompattern,theywillhoveraroundzero,
Page103

indicatingnoevidenceforaneffect.Ontheotherhand,theeffectsizesmayclusterinapositiveornegativedirection,indicatingthepresenceofaneffect.Apositive
patternindicatesthatrewardsproduceincreasesinintrinsicmotivationmeasuresanegativepatternsupportstheclaimthatrewardsundermineintrinsicmotivation.

METAANALYSESOFTHEEFFECTSOFREWARDSONINTRINSICMOTIVATION

RummelandFeinberg(1988):EffectsofControllingRewards
ThefirstmetaanalysisontheeffectsofrewardsonintrinsicmotivationwasconductedbyRummelandFeinbergin1988.4Thepurposewastoassesstheclaimof
cognitiveevaluationtheorythatcontrollingrewardsleadtodecreasedintrinsicmotivation.Fortyfivestudieswereincluded,inwhichparticipantswhoreceived
rewardsthatweredefinedbyRummelandFeinbergasconveyingcontrollinginformationwerecomparedtogroupsreceivingothertypesofrewardornoreward.
Intrinsicmotivationwasindexedasbothfreetimemeasuresandselfreportsoftaskinterestandsatisfaction.Thesemeasureswereincludedtogetherinthemeta
analysis.Theresearchersconcludedthattheirmetaanalyticresultsprovidedsupportforcognitiveevaluationtheory.Specifically,theyarguedthatcontrollingextrinsic
rewardshaveadetrimentaleffectonintrinsicmotivation.

AnAssessmentofRummelandFeinbergsMetaanalysis
RummelandFeinbergsanalysiswaslimitedtowhattheydefinedastheeffectsofcontrollingrewards.Theinvestigatorsdidnotincludemoderatoranalyses,suchasthe
effectsofdifferentrewardtypes,expectancies,contingencies,andsoon.Verbalrewardsandtangiblerewardswerecombined,anddifferentrewardexpectanciesand
contingencieswereaggregated.Inaddition,theresearcherscombinedfreetimeandselfreportmeasures.(AsindicatedinChapter6,thesemeasuresdonotproduce
consistentfindings,andthusshouldbeanalyzedseparately.)Bycollapsingacrossdifferentrewardconditionsanddifferentmeasuresofintrinsicmotivation,Rummeland
Feinbergsanalysisproducedanoveralleffectofreward,butnoanalyseswereconductedtoshowhowrewardimpactsintrinsicmotivationmeasuresunderdifferent
conditions.
AfurtherdifficultyisthatmanyoftheeffectsizesthatRummelandFeinbergusedwerenotbasedonacomparisonofarewardgrouptoacontrolgroupona
measureofintrinsicmotivation.Instead,manyoftheeffectsizesusedintheanalysisinvolvedcomparisonsamongtypesofrewards.Theproblemwiththistechniqueis
thatitdoesnotallowonetomakeaconclusionwithregardtotheeffectsofrewardsversusnorewards.
AmajordifficultywithRummelandFeinbergsanalysisisthatrewardsweredefinedascontrollingafterthefact.Whenarewardwasfoundtoproducea
Page104

negativeeffect,itwasclassifiedascontrollingandthestudywasselectedfortheanalysis.Thatis,therewardsweredefinedascontrollingorinformationalaftertheir
effectsonperformancehadbeenmeasured.ThisisnotonlyanissueinRummelandFeinbergsanalysisitexemplifiesamajoroveralldifficultywithcognitiveevaluation
theory.Theproblemisthatcontrollingrewardsarenotdefinedintheliteratureinstead,theyareinferredfromthebehaviortheysupposedlycause.
Cognitiveevaluationtheoryproposesthatchangesinfeelingsofcompetenceandselfdeterminationarethecausesofchangesinintrinsicmotivation,butthese
constructsarenottypicallymeasured.Changesinperceptionsofcompetenceandselfdeterminationareassumedtobeoperatingbecausebehaviorchanges.Inother
words,theexistenceofcompetence,selfdetermination,andintrinsicmotivationisinferredfromtheverybehaviortheysupposedlycause.Incognitiveevaluationtheory
andRummelandFeinbergsanalysis,rewardsaredefinedascontrollingifmeasuresofintrinsicmotivationdecrease,andrewardsaredefinedasinformationalwhenthe
dependentvariableindexesanincreaseinmotivation.
Thus,thefindingsandconclusionsofthefirstmetaanalysisonthetopicaredifficulttointerpretbecause(a)nomoderatoranalyseswereconducted,(b)different
measuresofintrinsicmotivationwerecombined,(c)manyeffectsizesincludedintheanalysiswerenotbasedonacomparisonofarewardgrouptoacontrolgroup,
and(d)rewardsweredefinedascontrollingaftertheeffectsonbehaviorwereobserved.

Wiersma(1992):TheMeasurementofIntrinsicMotivation
In1992,Wiersmaaddressedtheissuethattheeffectsofrewarddependonhowintrinsicmotivationismeasured.5Wiersmasmetaanalysisincluded20intrinsic
motivationstudiesfromworkandorganizationalpsychologyjournals.Onlystudiesthatusedtangiblerewardswereincluded.Studiesweredividedintothosewithfree
timemeasuresofintrinsicmotivationandthosethatmeasuredperformanceonataskduringtheexperimentalphaseofthestudy(therewardedperiod).
Thefindingsindicatedthatrewardsenhancedperformancemeasuresbutunderminedfreetimeintrinsicmotivation.Wiersmaconcludedthatthefindingthatrewards
reducedtaskbehaviorasmeasuredinafreetimeperiodsupportedcognitiveevaluationtheorybutthatthetheorywasnotsupportedwhentaskperformancewas
measuredwhiletheextrinsicrewardwasineffect.

AnAssessmentofWiersmasMetaanalysis
Deciandhiscolleaguessuggestedthatperformancemeasurestakenwhentherewardcontingencyisineffectarenotgoodindicatorsofintrinsicmotivationbecause
suchameasurewillreflectbothintrinsicandextrinsicmotives.6In
Page105

addition,Wiersmasanalysisislessthanconclusivebecauseonlyasmallsampleoftheavailablestudieswasincluded,somestudiesdidnotincludenorewardcontrol
groups,andnomoderatoranalyseswereconducted.

TangandHall(1995):TheOverjustificationHypothesis
Inametaanalysisof50studiesonrewardsandintrinsicmotivation,TangandHalltestedseveralpropositionsoftheoverjustificationhypothesis.7Specifically,they
assessedthehypothesesthatexpected,tangiblerewardswouldproducenegativeeffectsfortasksofhighinitialinterestwhentherewardsweretaskcontingent(what
Deciandhisassociateslaterreferredtoasengagementcontingentandcompletioncontingent)andwhentherewardswereperformancecontingent.Theyfurther
evaluatedthehypothesesthatnegativeeffectswouldnotoccurforverbalrewards,unexpectedrewards,tasknoncontingentrewards,andrewardsgivenfortasksof
lowinitialinterest.
Foreachoftheiranalyses,TangandHallassessedtheeffectsofrewardonsixdifferentmeasuresofintrinsicmotivation(freetimeontask,taskenjoymentratings,
quantityofperformanceduringtherewardphase,qualityofperformanceduringtherewardphase,ratingbyothers,andtimewaitedtoinitiatethetaskinafreechoice
session).
Generally,theresultssupportedthehypotheses.Expectedtangiblerewardswerefoundtodecreasemeasuresofintrinsicmotivationonhighinteresttaskswhenthe
rewardsweretaskcontingentandperformancecontingent.Tasknoncontingentrewardsdidnotaffectmeasuresofintrinsicmotivation.Contrarytothepredictionsof
theoverjustificationhypothesis,unexpectedrewardsledtoincreasesonsomeofthemeasuresofintrinsicmotivationandverbalrewardsledtodecreases.8

AnAssessmentofTangandHallsMetaanalysis
AlthoughTangandHallsanalysiswasmorecomprehensivethantheprevioustwoanalysesdescribedhere,severalshortcomingsmaketheirresultsdifficulttointerpret.
First,severalmeasuresofintrinsicmotivationwereincludedthathavebeencriticizedaspoormeasuresoftheconstruct.Forexample,TangandHallincludedmeasures
suchastaskenjoymentratingsandquantityandqualityofperformanceduringtherewardphase.Deciandhiscolleagueshavesuggestedthatsuchmeasuresarenot
puremeasuresofintrinsicmotivationbecausetheyreflectamixofintrinsicandextrinsicmotivation.9
Inaddition,somestudiesinTangandHallsmetaanalysisdidnotincludenorewardcontrolgroups,severalstudiesthatwerepublishedpriortotheirmetaanalysis
werenotincluded,afewstudiesweremiscategorized,andsomebasicstatisticalassumptionswereviolated.10Overall,however,TangandHallsresultssupportedthe
hypothesisthatexpectedtangiblerewardsdecreasemeasuresofintrinsicmotivationonhighinteresttasks.
Page106

CameronandPierce(1994)andEisenbergerandCameron(1996):TheFirstHierarchicalMetaanalysis
In1994,ourresearchteampublishedthefirsthierarchicalmetaanalysisoftheliterature(weaddedadditionalanalysesin1996).11Thesamplewasmadeupof96
studies.Inourhierarchicalmetaanalysis,allstudieswerefirstincludedinananalysisoftheoveralleffectsofrewards.Wethensearchedformoderatorvariables.The
studieswerebrokenoutbyonekeymoderator,thenanother,andsoon.Themoderatorsthatweexaminedwerebasedontheoreticalconsiderationsandsufficient
replicationinthestudies(rewardtype,expectancy,andcontingency).Ouranalysisincludedbetweengroupdesignsinwhichrewardedparticipantswerecomparedto
nonrewardedcontrolgroupsonmeasuresofintrinsicmotivation.Theeffectsofrewardswereanalyzedseparatelyforfreetimemeasuresofintrinsicmotivationand
selfreportedattitudestowardthetask.
Ourresultsindicatedthatoverallwhenalltypesofrewardwereaggregated,rewardsdidnotnegativelyaffectintrinsicmotivationoneitherofthemeasures.When
rewardsweresubdividedintorewardtype(verbal,tangible),expectancy(expected,unexpected),andcontingency,thefindingsdemonstratedthatpeoplewhoreceived
averbalrewardspentmoretimeonataskoncetherewardwaswithdrawntheyalsoreportedmoreinterestandenjoymentthannonrewardedpersons.
Tangiblerewardproducednodecrementsinintrinsicmotivationwhenitwasreceivedunexpectedly.Expectedtangiblerewardsproduceddifferingeffectsdepending
onthemannerinwhichtheywereadministered.Individualswhoreceivedanexpectedtangiblerewardforsolvingorcompletingatask,orachievingaspecificlevelof
performancedidnotspendlesstimeonthetaskthancontrolsoncetherewardwaswithdrawn.Theydid,however,reportmoreinterest,satisfaction,andenjoymentof
thetaskwhentherewardwasgivenforacertainlevelofperformance.
Detrimentaleffectsofrewardappearedonlywhentangiblerewardswereofferedtopeoplesimplyforengaginginatask,independentofsuccessfulperformance.
Undertheseconditions,oncetherewardwasremoved,individualsspentlesstimeonthetaskthancontrolstheydidnot,however,reportalessfavorableattitude
towardthetask.
Inadditiontoouranalysisofthegroupdesignstudies,wealsoprovidedaseparateanalysisofthesinglesubjectdesigns.Nounderminingeffectsofrewardwere
detected.Basedonourresults,wearguedthatthereisnoinherentnegativepropertyofrewards.Weconcludedthatnegativeeffectsofrewardareminimal,temporary,
andeasilypreventableinappliedsettings.

AnAssessmenttheMetaanalysesofCameronandPierceandofEisenbergerandCameron

Ourresearchandrecommendationscreatedconsiderabledebateandfuror.12Ourcriticssuggestedthatourfindingswereinvalidduetointentionalbias,
Page107

deliberatemisrepresentation,andineptanalysis.ThepopularwriterAlfieKohncommentedthat

acloserlookattheir[CameronandPierces]reviewandattheempiricalliteratureasawholerevealsthatthereismorethanadequatejustification
foravoidingtheuseofincentivestocontrolpeoplesbehavior,particularlyinaschoolsetting.13

Lepperandhiscolleaguesstatedthattheproceduresusedbyuswereakintoturningsilkpursesintosowsears.Theysuggestedthatouranalysiswascomparableto
puttingabeautifuldessertintoanindustrialblenderandliquefyingtheentireconcoction.14RyanandDecidescribedouranalysisasanattempttodefendtheir
[CameronandPierces]behavioralturf[rather]thanameaningfulconsiderationoftherelevantdataandissues.15
FurtherdebateensuedinanissueoftheAmericanPsychologistfollowingEisenbergerandCameronspublicationofthemetaanalysis.16Althoughourfindings
supportedmanyofthepredictionsmadebycognitiveevaluationtheoryandbytheoverjustificationhypothesis(seepredictionsinChapter4),proponentsofthese
theories(e.g.,Deci,Lepperandcolleagues)wouldnotaccepttheresults.17
Ourfindingswereclearlycontentiousandofgreatconcerntothosewhoarguethatnegativeeffectsofrewardaresubstantial,generalized,andprevalentacrossmany
conditions.18Inrecentarticlesonthetopic,Deciandhiscolleaguessuggestedthatourworkwasseriouslyflawedandthatrewardsdosubstantiallyunderminepeoples
intrinsicinterest.19Theysuggestedthatourfailuretodetectmorepervasivenegativeeffectsinourmetaanalysiswasduetoanumberofmethodologicalinadequacies.
Specifically,theycriticizedusforthefollowing:(a)collapsingdataacrosstaskswithhighandlowinitialinterestinouroverallanalysisandomittingamoderatoranalysis
ofinitialtaskinterest,(b)includingastudythatusedaninappropriatecontrolgroup,20(c)omittingstudiesordataasoutliersratherthanattemptingtoisolate
moderatorsfromthem,(d)omittingstudiesthatwerepublishedduringtheperiodcoveredbyourmetaanalysis,(e)omittingunpublisheddoctoraldissertations,and(f)
missclassifyingstudiesintorewardcontingencies,asdefinedbycognitiveevaluationtheory.

Deci,Koestner,andRyan(1999):ANewHierarchicalMetaanalysis
Spurredbyourresearch,ina1999issueofPsychologicalBulletin,Deci,Koestner,andRyanpresentedanewhierarchicalmetaanalysisthatclaimedtosupportthe
viewthatrewardshavepervasivenegativeeffectsonintrinsicmotivation.21Theyoutlinedanumberofconcernsthattheyhadwithpreviousmetaanalysestheirmeta
analysiswasdesignedtoaddresstheseconcerns,testcognitiveevaluationtheory,andprovideamorecomprehensivereviewoftheliterature.
Page108

Theirfindingssupportedcognitiveevaluationtheory,andgenerally,rewardswerefoundtohaveasubstantialnegativeeffectonpeoplesintrinsicmotivation.The
reviewersconcludedthatalthoughrewardscancontrolpeoplesbehaviorindeed,thatispresumablywhytheyaresowidelyadvocatedtheprimarynegative
effectofrewardsisthattheytendtoforestallselfregulation.22
Deciandhiscolleaguesidentified128experimentsonrewardsandintrinsicmotivation,including20unpublishedstudiesfromPh.D.dissertations.Torectifyconcerns
thattheyhadwithourpreviouswork,Deciandassociatesexcludedstudiesdeemedtohaveinappropriatecontrolgroups,includedstudiesthatweremissedinprevious
metaanalyses,andalsoincludedunpublisheddoctoraldissertations.Theirprimarymetaanalysisfocusedonrewardeffectsonhighinteresttasks.Studiesorconditions
withinstudieswereincludedonlyifthetasksusedweremeasuredordefinedtobeinterestingstudiesorconditionswithinstudieswereexcludedifthetasksusedwere
measuredordefinedasuninteresting.Thus,theirprimarymetaanalysisbeganwiththeoveralleffectsofrewardsonintrinsicmotivationforhighinteresttasksonly.23
Theeffectsofrewardwereanalyzedseparatelyformeasuresofselfreportedtaskinterestandfreechoiceintrinsicmotivation.Thefreechoicemeasureincluded
timespentonataskafterrewardswereremoved.Whenatimemeasurewasnotreportedinastudy,thereviewersusedmeasuresoftaskpersistenceduringthefree
choiceperiod(e.g.,numberoftrialsinitiatedinagame,numberofballsplayedonapinballmachine,numberofsuccessesonatask).Hence,Decisanalysisofthe
freechoicemeasurewasbroaderthanourpreviousanalyses,whichusedonlystudiesthatassessedtimemeasures.
Onhighinteresttasks,Decisfindingsindicatedasignificantoverallnegativeeffectofrewardonthefreechoicemeasureandanonsignificantnegativeeffectof
rewardontheselfreportmeasure.Intermsofmoderatoranalyses,verbalrewardswerefoundtoincreasefreechoiceintrinsicmotivationforcollegestudents(a
nonsignificanteffectwasfoundforchildren)andtoenhancetaskinterestforbothchildrenandcollegestudents.Tangiblerewardsweredividedintounexpectedand
expectedrewardstherewerenoreliableeffectswhentherewardswereunexpected.Basedoncognitiveevaluationtheory,expectedtangiblerewardswerefurther
subdividedintotasknoncontingent,engagementcontingent,completioncontingent,andperformancecontingentrewards.Theresultsshowednoreliablenegative
effectsfortasknoncontingentrewards,whereasengagementcontingentrewardsproducedsignificantnegativeeffectsonbothfreechoiceintrinsicmotivationandself
reportedtaskinterest.Completioncontingentandperformancecontingentrewardsalsoresultedinsignificantnegativeeffectsonthefreechoiceintrinsicmotivation
measure.
Additionally,Deciandhiscolleaguesprovidedabreakdownofperformancecontingentrewardsintostudiesofmaximumandnotmaximumreward.Alarge
negativeeffectwasfoundforthesixstudiesidentifiedasinvolvingnotmaximumreward.(Inthistypeofstudy,someparticipantsfailedtoattainthecriterionandwere
givenlessthanmaximumreward.)
Page109

Asasupplementaryanalysis,Deci,Koestner,andRyananalyzedstudieswithchildreninwhichthefreechoiceassessmentofhighinterestactivitieswasconducted
immediatelyfollowingtheremovalofreward,withinaweek,andafteraweek.Theresultsshowednegativeeffectsofrewardateachtimeofassessmentthereviewers
arguedthattheseresultsindicatethattheunderminingeffectisnotatransitoryphenomenon.
Allinall,Deci,Koestner,andRyansmetaanalysisoftheeffectsofrewardsonintrinsicmotivationproducednumerousnegativeeffectsofexpectedtangible
rewardsunderthevariousrewardcontingencies.Theyclaimedthattheirfindingssupportedeachofthepredictionsmadebycognitiveevaluationtheory.

AnAssessmentofDeci,Koestner,andRyansMetaanalysis
AlthoughDeciandhiscolleaguesanalysisincludedmorestudiesthananyothermetaanalysespriortotheirpublication,thereareseveralshortcomingsintheirwork.
Oneissueconcernstheoveralleffectsofrewardsacrossalltypesoftasks.Deciandhisassociatesdidnotconductaprimaryanalysisoftheoveralleffectsofreward.
Instead,theyarguedthatthemoretheoreticallyrelevantquestionconcernstheeffectsofrewardsonhighinteresttasks.Intermsofinitialtaskinterest,Decisresearch
teamnotedthat

thefieldofinquiryhasalwaysbeendefinedintermsofintrinsicmotivationforinterestingtasksandtheunderminingphenomenonhasalwaysbeen
specifiedasapplyingonlytointerestingtasksinsofaraswithboringtasksthereislittleornointrinsicmotivationtoundermine.24

Giventhatcognitiveevaluationtheoryhaslittletosayabouttheeffectsofrewardsonlowinteresttasks,Deci,Koestner,andRyansprimarymetaanalysisfocusedon
rewardeffectsonhighinteresttasks.
Wecontendthatananalysisoftheoveralleffectofrewardiscentraltoanunderstandingofthiscomplexareaofresearch.Onapracticallevel,manyeducators,
parents,andadministratorshavetakenthepositionthatrewardsandincentivesystemsareharmful.Theviewisthatrewardsnegativelyaffectstudentsintrinsicinterest
acrossalltypesofactivities(e.g.,reading,math,science,computergames,etc.)nodistinctionismadebetweenlowandhighinitiallevelsoftaskinterest.
Writerswhocautionagainsttheuseofrewardsandreinforcementfrequentlyuseexamplestoillustratetheirpoint.Moreoftenthannot,activitiessuchasreading,
lawnmowing,andmathematicsarecitedasactivitiesinwhichpeoplewillloseinterestiftheyaregivenrewardsforperformingtheactivity.Mostoftheseactivitiesare
notonesthatindividualsbegindoingwithhighlevelsofinitialinterest.Importantly,policymakerswhoadopttheviewthatrewardsareharmfulrarelydistinguish
betweenhighandlowinteresttasks.Becauseofthis,ananalysisoftheoveralleffectsofrewardiswarranted.Itisourcontentionthatamorecompleteanalysisofthe
effectsofrewardsonintrinsicmotivation
Page110

shouldbeginatthelevelofallrewardsoveralltypesoftasks.Followingthis,abreakdownofrewardeffectsonhighandlowinteresttaskswouldbeappropriate.
Anotherconcernisthatforsomestudiesincludedintheiranalysisofhighinteresttasks,Deciandhiscolleaguesomittedconditionsthatwererelevanttotheir
analyses.25Inaddition,theyalsomissedafewexperimentsthatmettheirinclusioncriteriaandwerepublishedduringtheperiodcoveredbytheirmetaanalysis.As
well,severalstudiesusinghighinteresttasksthatmeasuredselfreportedtaskinterestwereeitherexcludedorinadvertentlyomittedfromtheanalyses.Manyofthese
studiesfoundpositiveeffectsontheselfreportmeasureoftaskinterest.Eachoftheseissuesisthoroughlydocumentedinasetofappendicesinourupdatedreviewof
thisliterature.26
ThemajorareaofdisagreementbetweenDeciandhiscolleaguesworkandourpreviousmetaanalysesconcernstheclassificationofexpected,tangiblerewardinto
variousrewardcontingencies.Asindicated,Deciandhiscolleaguesusedcognitiveevaluationtheorytoguidetheirclassificationofstudies.Theyestablishedthe
categoriesoftasknoncontingent,engagementcontingent,completioncontingentandperformancecontingentreward.
Althoughthiscategorizationsystemmaybeinformativeforcognitiveevaluationtheory,theproblemisthatthecategoriesaretoobroad.Studiesthatusedvery
differentprocedureswerepooledintooverallcategoriesofengagementcontingent,completioncontingentandperformancecontingentreward.Forexample,
experimentsinwhichparticipantswereofferedarewardeitherfordoingwell,foreachproblem/unitsolved,forobtainingacertainscore,orforexceedinganormwere
allcategorizedasperformancecontingentbyacognitiveevaluationtheoryframework.
InaresponsetotheworkofDeciandhisassociates,Eisenberger,Pierce,andCameronexaminedsomeofthesediverserewardproceduresandfoundvery
differenteffectsonmeasuresofintrinsicmotivation.27Theirfindingsindicatedthatwhentangiblerewardswereofferedformeetingorexceedingaspecificcriterionor
surpassingthescoresofothers,measuresofintrinsicmotivationincreased.Presently,itisdifficulttoreachadefinitiveconclusionabouttheeffectsofdifferentreward
contingencies.Studiesusingdifferentproceduresandproducingdifferentresultsneedtobeanalyzedseparatelyandnotlumpedintooverallcategories.
OnewaytoresolvethisissueistoreturntotheMethodssectionoftheoriginalstudies,writedowntheprecisestatementoftherewardcontingencyusedbythe
researchers,andcodethestudiesaccordingtotheproceduresactuallyemployedintheexperiment.Bycategorizingstudiesintermsoftheactualcontingenciesused
ratherthanbythosepresumedtobeaneffectbyanytheoreticalperspective,practitionerscandeterminewhethertherewardproceduresusedinthelaboratory
experimentsarecomparabletothoseusedinappliedsettings.Furthermore,aproceduralcategorizationallowsforatestofcognitiveevaluationtheoryand
Page111

alsoprovidesatestofalternativeaccountsoftheeffectsofrewardsonintrinsicmotivation.
Theliteratureonrewardsandintrinsicmotivationisfraughtwithcompetingtheories(e.g.,cognitiveevaluationtheory,theoverjustificationhypothesis,socialcognitive
theory,generalinteresttheory,thecompetingresponsehypothesis,behavioraltheory).Theproblemwithorganizingstudiesaccordingtoaparticulartheoreticalstance
isthateachtheorycouldbeusedtoorganizetheliteratureand,usingcategoriesappropriatetothetheory,couldthusgainsupport.Usingatheoreticalapproachto
guidetheclassificationoftherewardproceduresdoesnotprovideadefinitiveanswerabouttheeffectsofrewardcontingenciesonmeasuresofintrinsicmotivation.
Instead,weproposethataproceduraldescriptionofrewardcontingencies,notonlyallowsustoassesswherewestandintermsoftheeffectsoftheactualreward
contingencies,butalsoprovidesuswithatestofalternativeaccountsoftheeffectsofrewardsonintrinsicmotivation.
Researchersusingasinglesubjectdesignfoundthatwhenrewardsweredeliveredrepeatedlyandrepeatedassessmentsofperformanceonthetaskweretaken
withoutreward,detrimentaleffectswerenotevident.Deciandhiscolleaguesdidnotprovideananalysisofthesestudies.Instead,theyarguedthatresultsfromsuch
studiesarenotgeneralizablebecausetoofewparticipantsarestudiedinanyoneexperiment.Theyfurtherarguedthatinthesinglesubjectdesigns,thereisnogroup
thatperformstheactivitywithoutreinforcementthus,onecannotknowifthereisanunderminingeffectrelativetoacontrolgroup.Hence,theconclusionsreachedby
Deciandassociatesconcernedresultsfromthebetweengroupdesignstudiesonly.
Finally,weshouldpointoutthatalthoughDeciandhiscolleaguesclaimedthattheirfindingssupportcognitiveevaluationtheory,theydidnotprovideanytestofthe
mediators(feelingsofcompetenceandselfdetermination)thatarethoughttobecriticaltoproducingchangesinpeoplesintrinsicmotivation.Instead,Deciandhis
associatesusedevidenceofdecreasesinmeasuresofintrinsicmotivationtoinferthecontrollingnatureofrewards.However,Eisenberger,Pierce,andCameron
showedthatrewardsofferedfordoing,completing,ormeetingaperformancecriterionoftenincreasedpeoplesperceivedfreedomandindependence(autonomy).28
Thisevidencesuggeststhatrewardsmaynotbeviewedbypeopleascontrollingorrestrictivetotheirsenseoffreedom.Atpresent,cognitiveevaluationtheoristshave
notprovidedanyevidencetoindicatewhypeopleshowalossofintrinsicmotivationonexpectedtangiblerewardcontingencies.
Page112

PROBLEMSWITHDECI,KOESTNER,ANDRYANS(1999)METAANALYSISOFREWARDSANDINTRINSIC
MOTIVATION
Noprimaryanalysisoftheoveralleffectsofrewards
Noprimaryanalysisoftheeffectsofrewardsforbothhighandlowinteresttasks
Omittedconditionswithinstudiesthatwererelevanttotheanalyses
Excludedexperimentsthatmettheinclusioncriteriaandwerepublishedduringtheperiodcoveredbythemetaanalysis
Omittedstudiesthatmettheinclusioncriteriaandshowedpositiveeffectsofreward
Collapseddataacrossdifferentrewardproceduresthatcouldhavedifferenteffectsonmeasuresofintrinsicmotivation

Insum,Deci,Koestner,andRyansmetaanalysisontheeffectsofrewardsonintrinsicmotivationwasbroaderthanpreviousmetaanalyses,butthereareproblems
thatmaketheirconclusionslessthansatisfactory.Specifically,theresearchers(a)didnotincludeananalysisoftheoveralleffectsofrewardsacrossalltypesoftasks,29
(b)omittedconditionswithinstudiesthatwererelevanttotheiranalyses,(c)missedexperimentsthatmettheirinclusioncriteriaandwerepublishedduringtheperiod
coveredbytheirmetaanalysis,and(d)omittedseveralstudiesthatmettheirinclusioncriteriaandshowedpositiveeffectsofrewardontaskinterestmeasures.In
addition,byorganizingstudiesaccordingtothecognitiveevaluationframeworkratherthantheactualproceduresusedinthestudies,Deciandhiscolleaguescollapsed
dataacrossdistinctrewardprocedures.Thisstrategyresultedinafindingofpervasivenegativeeffectsofrewardwhen,infact,manyofthestudiesshowedthat
rewardsincreasedmeasuresofintrinsicmotivation.Thepointisthatstudiesusingdifferentproceduresthatproducedifferentresultsneedtobeanalyzedseparatelyand
shouldnotcombinedintooverallcategories.

SUMMARY
Overthepast30years,alargebodyofresearchhasbeendevotedtoassessingtheeffectsofrewardsonpeoplesintrinsicmotivation.Morethan100experiments
havebeenconductedusingabetweengroupsdesigninwhichrewardedparticipantsarecomparedtononrewardedparticipantsonvariousmeasuresofintrinsic
motivation.Thefindingshavebeenmixed.Metaanalyticreviewsof
Page113

theliteraturesuggestthatnegativeeffectsoccurwhentherewardsaretangibleandexpectedbutthatdifferenteffectsoccurunderdifferentcontingencies.
Incontrasttothetypicaluseofasinglepairingoftaskperformancewithrewardfollowedbyasinglesessionofunrewardedtaskperformance,fivestudieshave
comparedrepeatedtaskperformancebeforeandafterrepeatedrewardpresentations.Theresultsofthesesinglesubjectdesignsshownonegativeeffectsofreward.
Thesefindingssuggestthatwhenrepeatedtrialsproceduresareused,rewardsdonotresultindecrementaleffects.
Presently,itisdifficulttomakemanydefinitivestatementsaboutthevariablesthataffecttheinfluenceofrewardsonintrinsicmotivationmeasures.In1994,we
publishedametaanalysisofthisliteratureandconcludedthatnegativeeffectsofrewardwerecircumscribedandeasilyavoidable(CameronandPierce,1994).In
contrast,amorerecentmetaanalysisbyDeci,Koestner,andRyan(1999)showspervasivenegativeeffectsofexpectedtangiblerewardcontingencies.
Ourexaminationofmetaanalysesonthetopicshowsthatdifferentconclusionsreachedbydifferentreviewersarearesultofconceptualandmethodological
shortcomings.Inthenextchapter,wereanalyzethedataonrewardsandintrinsicmotivation.Thepurposeofouranalysisistoaddressshortcomingsinpreviousmeta
analyticreviewswhiledrawingontheirstrengths.

NOTES
1. SeeHunterandSchmidt,1990.
2. Glass,1976.
3. Forexample,seeKollins,Newland,andCritchfield,1997.
4. RummelandFeinberg,1988.
5. Wiersma,1992.
6. Deci,Koestner,andRyan,1999,p.635.
7. TangandHall,1995.
8. SeeTangandHall,1995,pp.380381,Tables7and8.
9. Deci,Koestner,andRyan,1999,p.632.
10.SeethecommentsinCameronandPierce,1996.AlsoseethecommentsintheNotessectionofAppendix8A.
11.CameronandPierce,1994EisenbergerandCameron,1996.
12.CameronandPierce,1996Lepper,Keavney,andDrake,1996Kohn,1996RyanandDeci,1996.
13.Kohn,1996,p.3.
14.Lepper,Keavney,andDrake,1996,pp.2526.
15.RyanandDeci,1996,p.34.
16.EisenbergerandCameron,1998HennesseyandAmabile,1998Lepper,1998SansoneandHarackiewicz,1998.
17.HennesseyandAmabile,1998Lepper,1998SansoneandHarackiewicz,1998.
18.SeethecommentsbyCameronandPierce(1996)andEisenbergerandCameron(1998),whorespondedtocriticismoftheirresearch.
Page114

19.Deci,Koestner,andRyan1999,2001.
20.Specifically,Deci,Koestner,andRyan(1999)wereconcernedbecausewehadincludedastudybyBoalandCummings(1981).Thisstudywasslightlydifferentfromotherincluded
studies.Insteadoftakingplaceinalaboratory,thestudywasconductedinaworkplace.Theparticipantswereemployees,andallwerepaidonthejob.Duringthereward
intervention,theexperimentalparticipantswerepaidextracontrolparticipantswerenot.Decietal.s(1999)concernwasthatbecauseallparticipantswerereceivingpay,therewas
nononrewardedcontrolgroup.
21.Deci,Koestner,andRyan,1999,p.633.
22.Ibid.,p.659.
23.AnadditionalanalysisoftheeffectsofrewardsonlowinteresttaskswasconductedbyDeci,Koestner,andRyan(1999)noreliableeffectsweredetected.Aproblemwiththis
analysis,however,isthatseveralstudiesthatusedlowinteresttaskswereexcluded(e.g.,FreedmanandPhillips,1985OverskeidandSvartdal,1996).
24.Deci,Koestner,andRyan,1999,p.650.
25.Forexample,inanexperimentbyWilson(1978),onegroupwasoffered$.50toengageinthetargetactivity,asecondgroupwasoffered$2.50forthesametask,andacontrolgroup
performedthetaskwithouttheofferofreward.InDeci,Koestner,andRyans(1999)analyses,onlyoneoftherewardedgroupswasincluded.Forotherstudiesthatusedmorethan
onelevelofrewardmagnitude(e.g.,Earn,1982McLoyd,1979NewmanandLayton,1984),Decietal.includedallrewardconditions.Theomissionofcertainconditionswithin
studiesdoesnotappeartobesystematic(forexample,rewardmagnitudewasnotexaminedbyDecietal.asamoderator),yetthereareanumberofdifferenttypesofcaseswhere
thisoccurred.
26.Cameron,Banko,andPierce,2001.
27.Eisenberger,Pierce,andCameron,1999.
28.Ibid.
29.Deci,Koestner,andRyan(1999)didconductasupplementalanalysisoftheoveralleffectsofrewardsacrosshighandlowinteresttasks.Theproblem,however,isthatmany
studiesthatusedlowinteresttasks(whichwereexcludedfromtheirprimaryanalysisofhighinteresttasks)werenotincludedintheiroverallanalysis(e.g.,FreedmanandPhillips,
1985PhillipsandFreedman,1985).
Page115

Chapter8
AMetaanalysisoftheEffectsofRewardsonIntrinsicMotivation

Inthischapter,wepresentanupdatedmetaanalysisoftheexperimentalstudiesontheeffectsofrewardsonpeoplesintrinsicmotivation.1Thepurposeofourcurrent
metaanalyticreviewisto(a)resolvedifferencesbetweenpreviousmetaanalyticfindings,(b)provideametaanalysisofrewardsandintrinsicmotivationthatallowsfor
testsofcompetingtheoreticalexplanations,and(c)provideeducators,managers,andotherpractitionerswithaclearunderstandingofwhattheycanextrapolatefrom
thelaboratoryinvestigationsofrewardsandintrinsicmotivationtoappliedsettings.
Ourreanalysisisinformedbyaconsiderationofdecisionsandproceduresusedinthepreviousmetaanalyses.Inaddition,theconcernsraisedinChapter7about
priormetaanalyses(includingours)areaddressed.Ouranalysisisnotguidedbyanyparticulartheoryinstead,wecategorizestudiesinordertotestdifferent
theoreticalpredictions.Basedonourfindings,inthenextchapterweevaluatetheadequacyofcognitiveevaluationtheory,theoverjustificationhypothesis,social
learningtheory,andbehavioralexplanationsfortheeffectsofrewardsonintrinsicmotivation.
Webeginwithadescriptionoftheproceduresusedinthecurrentmetaanalysisandthenpresentourresults.Next,weprovideareviewofstudiesassessinglong
termeffectsofrewardaswellasanexaminationofstudiesthathaveassessedmediationalprocessesofreward.Technicalaspectsofourmetaanalysisandadetailed
statisticaldescriptionofourfindingsarepresentedinAppendix8A,attheendofthechapter.
Page116

THEPRESENTMETAANALYSIS
Thepurposeofthepresentmetaanalysiswastomakeacausalstatementabouttheeffectsofextrinsicrewardsonintrinsicmotivation.Thisanalysiswasdesignedto
addressanumberofconcerns.Ofmajorimportanceiswhetherthebulkofevidencesuggeststhatextrinsicrewardsproducedecrementsinintrinsicmotivation.Ifso,
whatisthesizeofanyrelationshipsbeinguncovered?Also,dodifferentpatternsemergewithdifferenttypesoftasks(highorlowlevelsofinitialtaskinterest),different
rewardtypes(e.g.,tangibleorverbalrewards),rewardexpectancies(expected,unexpected),rewardcontingencies(e.g.,rewardsdeliveredforengaginginatask,
completingorsolvingatask,ormeetingaspecifiedlevelofperformance),andsoon?Inordertoaddresstheseissues,ouranalysisfocusedonresultsfrombetween
groupdesignstudies.Ourgoalwastoassesshowrewardaffectsmeasuresoffreechoiceintrinsicmotivation2andselfreportedtaskinterest.

ResearchQuestions
Thereareseveralstepstoconductingametaanalysisthatwereusedinthecurrentresearch.Thefirststepwastospecifytheresearchquestions.Thefollowing
questionswereaddressedinthepresentmetaanalysis:
1.Overall,whatistheeffectofrewardonintrinsicmotivation?Ouranalysisbeginswithanassessmentoftheoveralleffectsofreward.Studieswereincludedin
whichparticipantswhoreceivedarewardwerecomparedtoanonrewardedcontrolgroup.Thepurposeofthisanalysiswastoassesstheeffectsofrewards
acrossalltypesoftasksandoverdifferentrewardtypes.
2.Underwhatconditionsdorewardsleadtodecreasedorincreasedintrinsicmotivation?AsdiscussedinChapters6and7,severalmoderatorvariablesinteract
withrewardtoproduceincrementsordecrementsinintrinsicmotivation.Specifically,ourmetaanalysisisdesignedtoinvestigatethefollowing:
a.theeffectofrewardonintrinsicmotivationwhenthetasksusedareofeitherhighorlowinitialinterest.
b.theeffectofrewardtypeonintrinsicmotivation(i.e.,whetherrewardsareverbalortangible)
c.theeffectofrewardexpectancyonintrinsicmotivation(i.e.,whetherrewardsareexpectedpromisedanddeliveredtoparticipantsorunexpected
deliveredtoparticipantsbutnotpromised)
d.theeffectofrewardcontingencyonintrinsicmotivation(i.e.,whetherrewardsaredeliveredforparticipatinginanexperimentalsessionregardlessof
whatparticipantsdo,forengaginginatask,forcompletingorsolvingatask,etc.)
e.theeffectofdeliveringmaximumorlessthanmaximumreward.
Page117

Allanalysesperformedonthesefeatureswereconductedwithbetweengroupdesignstudiesinwhicharewardedgroupwascomparedtoanonrewardedcontrol
group.Separateanalyseswereconductedonthefreechoiceandtaskinterestmeasures.Theseanalysesweredesignedtoleadtoagreaterunderstandingofthe
specificconditionsunderwhichrewardaffectsintrinsicmotivation.

SampleofStudies
Studiesincludedinthepresentanalysisincorporatedthedatabasesfrompriormetaanalyses.3Inaddition,inasearchofthePsycINFOdatabase(oftheAmerican
PsychologicalAssociation),welocatedafewnewstudiesandafewstudiesthatwereinadvertentlymissedinpreviousmetaanalyses.Thecriteriaforincludingastudy
inasamplewere:arewardedgroupwascomparedtoanonrewardedgroup,therewardsweredistinguishedasverbal(praise,positivefeedback)ortangible(e.g.,
money,candy,goodplayerawards),andintrinsicmotivationwasindexedasfreechoicebehavior(timespentonthetaskfollowingtheremovalofrewardor
performanceonthetaskduringthefreechoiceperiod)orbyselfreportedmeasuresoftaskinterest(taskliking,enjoyment,satisfactionortaskpreference).
Inadditiontoincludingpublishedwork,andinkeepingwithDeci,KoestnerandRyans(1999)analysisofthisliterature,weincludedunpublisheddoctoral
dissertations.4Theresultingsampleconsistedof145independentstudies(21oftheexperimentswerefromunpublisheddoctoraldissertations).Atotalof115studies
includedafreechoicemeasureofintrinsicmotivationand100studiesincludedaselfreportmeasureoftaskinterest.Alistofthestudiesincludedinourmetaanalysis
canbefoundintheReferencesectionofthisbook.

ClassificationandCodingofStudies
Allstudieswerefirstcodedforinitiallevelsoftaskinterest.5Ifameasureofinitialtaskinterestwasreportedinthearticle,thestudywasclassifiedasalowinteresttask
iftheaverageonthatmeasurewasbelowthemidpointofthescalefortheactivityandahighinteresttaskiftheaveragewasatorabovethemidpoint.Studieswithout
preinterestmeasureswereclassifiedashighorlowinterestdependingonhowtheresearcherdefinedthetaskorwhetherthetaskhadbeendescribedasinterestingor
uninterestinginpriorexperiments.
Studieswerealsoclassifiedaccordingtorewardtype(tangible,verbal),expectancy(expected,unexpected),andcontingency.Tangiblerewardsweremoney,
ticketstoatheater,candy,toys,andsoonverbalrewardsinvolvedgivingparticipantspraiseorpositivefeedbackontheexperimentaltask.Intermsofreward
expectancy,expectedrewardswerethoseofferedtoparticipantspriortoengagingintheactivitytherewardsweredeliveredaftertheyhadworkedonthetask.
Unexpectedrewardswerethosedeliveredattheendoftheexperimentalsessionbutnotofferedbeforehand.
Page118

Table8.1
DescriptionofExpectedTangibleRewardContingencies
RewardContingency Description
Tasknoncontingent rewardisofferedforagreeingtoparticipate,forcomingtothestudy,orforwaitingfortheexperimenter
offerofrewardisunrelatedtoengaginginthetask
Rewardsofferedfordoingwell rewardisofferedfordoingwellonthetaskorfordoingagoodjob
nospecificationisgivenastowhatitmeanstodoagoodjobortodowell
Rewardsofferedfordoingatask rewardisofferedtoparticipantstoengageintheexperimentalactivity
noinstructionsaregivenabouthowwellparticipantsmustperformorwhethertheymustcompletethetask
Rewardsofferedforfinishingorcompletingatask rewardisofferedtofinishanactivity,completeatask,orgettoacertainpointonthetask
rewardisnotrelatedtoqualityofperformance
Rewardsofferedforeachunitsolved rewardisofferedforeachunit,puzzle,problem,etc.,thatissolved
Rewardsofferedforexceedinganorm rewardisofferedtomeetorexceedtheperformanceofothersonthetask(relativestandard)

Forrewardcontingency,ratherthanclassifystudiesaccordingtoacognitiveevaluationtheoryframework(aswasdonebyDeciandhiscolleagues)orinaccordwith
anyothertheoreticalframework,wecategorizedrewardcontingencyintermsoftheactualproceduresusedinthestudies.Thisprocedureallowedustotestavariety
oftheoreticalpredictions.
Toclassifystudiesbyrewardcontingency,wewentbacktotheoriginalstudies,readthepreciseproceduresusedforrewarddelivery,andrecordedwhatwassaid
totheparticipantsandhowtherewardwasdelivered.Wethenorganizedthestudiesintosevenmaincategoriesofrewardcontingency:rewardsdeliveredregardlessof
taskinvolvement(tasknoncontingent)fordoingataskfordoingwellforfinishingorcompletingataskforeachproblem,puzzle,orunitsolvedforachievingor
surpassingaspecificscoreandformeetingorexceedingtheperformanceoftheothers.AdefinitionofeachoftherewardcontingenciesispresentedinTable8.1.6
Finally,weidentifiedstudiesthatinvolvedeithermaximumorlessthanmaximumreward.Suchstudiesinvolvedofferingparticipantsarewardfordoing
Page119

well,finishingatask,eachproblem/unitsolved,surpassingascore,orexceedinganorm.AsdiscussedinChapter6,studieswerelabeledmaximumrewardif
participantsintherewardconditionmettheperformancerequirementsandreceivedthefullreward.Lessthanmaximumrewardoccurredwhentherewasatimelimit
suchthatsomeparticipantswereunabletomeetalltherequirementsinthetimeallottedandweregivenlessthanthefullreward.

CalculationandAnalysisofEffectSizes
Afterallstudieswerecoded,wecalculatedeffectsizesforeachcomparisonofarewardedgrouptoanonrewardedgrouponthefreechoiceandselfreportmeasures
ofintrinsicmotivation.Inthepresentmetaanalysis,positiveeffectsizesindicatethattherewardproducedincreasesinintrinsicmotivation,negativeeffectsizessupport
theclaimthatrewardsundermineintrinsicmotivation,andzeroeffectsindicatenoevidenceforaneffectofreward.Alistofstudiesusedineachoftheanalyses,the
classificationandcodingofeachstudy,andeffectsizesarepresentedinAppendix8B,attheendofthischapter.Afteralltheeffectsizeswerecalculated,statistical
analysesoftheeffectsizeswereperformedusingacomputerprogramcalledMETA.7

ProceduresUsedinOurHierarchicalAnalysis
Weconductedahierarchicalanalysisthatbeganatthelevelofallrewardsacrossalltypesoftasks.Wethenexaminedtheeffectsofdifferentmoderatorvariables.Our
firstbreakdownwasintermsofhighandlowinitialtaskinterest.Itisimportanttorecognizethatthisisnottheonlywayinwhichahierarchicalmetaanalysisonthis
topiccouldproceed.AswepointedoutinChapter7,Deci,Koestner,andRyansfocusonhighinteresttasksisbasedontheirtheoreticalpositioncognitive
evaluationtheory.8However,thefieldisnotunifiedtheoretically.Thus,onecanimaginethatresearchersfromothertheoreticalpositionswouldbeginadifferent
moderatoranalysis.Forexample,aresearcherwithanoverjustificationviewpointmightselectforrewardsalienceandsubdividestudiesintothosewithhighandlow
rewardsalience.Asocialcognitiveinvestigatormightfocussolelyontheeffectsofrewardsonlowinteresttasksbecauserewardscanbeusedtocultivatetaskinterest
andtobuildcompetenceandselfefficacy.Abehavioralresearcher,ontheotherhand,mightwishtobreakstudiesintothoseinwhichtherewardsareshownto
functionasreinforcementandthoseinwhichtheyarenot.Thepointisthatiftherewereenoughstudiesineachofthesecategories,thechoiceofwhichmoderatorto
emphasizewouldvarydependingononestheoreticalstance.
BecauseDeciandhiscolleaguesfocusedtheiranalysisontheeffectsofrewardsonhighinteresttasks,wechosehighandlowinitialtaskinterestasourfirst
breakdown.SubdividingthestudiesbyhighandlowinitialtaskinterestallowedustodirectlycompareourfindingswiththoseofDeciandhisassoci
Page120

ates.Indoingso,wefavorcognitiveevaluationtheory.Ontheotherhand,failuretofindpervasivenegativeeffectsevenwithhighinteresttasksfavorstheconclusion
thatrewardcontingenciesdonotdestroyinterest.Inotherwords,thestrongestwaytotestDeci,Koestner,andRyansclaimswastousetheirrequirementthattasks
usedinthestudiesmustbebrokenoutbyhighorlowinitialinterest.
Fewstudiesusedtasksoflowinitialinterest,andthus,nofurtherbreakdownofthesestudieswascarriedout.Ontasksofhighinitialinterest,however,studieswere
subdividedbyrewardtype(verbal,tangible)tangiblerewardswerefurtherbrokendownbyrewardexpectancy(expected,unexpected),andtheeffectsofexpected
tangiblerewardswereassessedbytherewardcontingency.Atthefinallevelofanalysis,studieswereanalyzedthatusedmaximumversuslessthanmaximumreward.

RESULTS

OverallEffectsofReward
Ourfirstsetofanalysesconcernedtheoveralleffectsofrewardonfreechoiceintrinsicmotivationandselfreportedtaskinterest.Freechoiceintrinsicmotivationwas
measuredastimespentonthetaskaftertherewardwaswithdrawnorperformanceonthetaskduringthefreechoiceperiod.Taskinterestwasassessedby
questionnairemeasuresofparticipantstaskinterest,enjoyment,orsatisfaction.Theseanalysesincludedstudieswithbothhighandlowinteresttasksaswellasstudies
thatusedverbalortangiblerewards.Thatis,allstudieswereincludedintheseanalyses.
Onthefreechoicemeasureofintrinsicmotivation,thestatisticaltestshowedthattherewasnosignificanteffectofreward.Intermsofselfreportedtaskinterest,the
metaanalysisrevealedasmallbutsignificantpositivefinding.Thatis,overall,individualsreportedgreatertaskinterestwhentheyreceivedareward.Thesefindings
suggestthattheevidencedoesnotsupporttheclaimthatrewardshavegeneraldetrimentaleffects.Theresultsindicatethatwhenrewardsarecollapsedacrossdifferent
taskandrewardtypes,overall,rewardsdonotsignificantlyundermineintrinsicmotivationasmeasuredbyfreechoiceandselfreportmeasures.9

EffectsofRewardonLowandHighInterestTasks
Althoughtheanalysisshowsthatrewardsdonothavegeneralizednegativeeffects,severalresearchershavesuggestedthatlevelofinitialtaskinterestisofprime
importance.Theclaimfromcognitiveevaluationtheoryisthatifpeopleenjoy,andarehighlyinterestedin,anactivity,theintroductionofrewardwilldestroytheir
intrinsicinterest.Cognitiveevaluationtheoryhaslittletosayabouttheeffectsofrewardsonlowinteresttasks.Ontheotherhand,fromtheper
Page121

spectiveofsociallearningtheory,rewardscanbeusedtoenhanceinterestandperformanceontasksinwhichpeoplearenotinitiallyinterested.Thus,ournextanalysis
concernstheeffectsofrewardsonlowinterestversushighinteresttasks.
Fortheanalysisoftheeffectsofrewardonlowinteresttasks,12studiesusedafreechoicemeasureofintrinsicmotivationand11studiesmeasuredtaskinterest.
Themetaanalysisrevealedastatisticallysignificantpositiveeffectonthefreechoicemeasuretherewasnoreliableeffectonselfreportedtaskinterest.10These
findingsindicatethatwhenataskisnotinitiallyinteresting,rewardsenhancetimeandperformanceonthetaskbutdonotenhanceverbalexpressionsoftaskinterest.
Thebulkofstudiesusedtasksthatwereofhighinteresttoparticipants.Atotalof114studiesusedhighinteresttasksandmeasuredtheeffectsofrewardonfree
choiceintrinsicmotivation,whereas98studiesusedselfreportmeasures.Forhighinteresttasks,onfreechoice,themetaanalysisshowedasmallbutsignificant
negativeeffect.Themeaneffectsize(average)forselfreportedtaskinterestwassignificantitwassmall,butinapositivedirection.Thus,forhighinteresttasks,
rewardswerefoundtounderminefreechoiceintrinsicmotivationbutrewardedparticipantsreportedgreatertaskinterest.

BreakdownofRewardEffectsonHighInterestTasks
Giventhelargenumberofstudiesthatusedhighinteresttasks,itwaspossibletoexamineaspectsoftherewardproceduresthatcouldleadtoincrementsor
decrementsinintrinsicmotivation.Specifically,theconcernwaswhen,andunderwhatconditions,rewardproceduresproducepositiveornegativeeffectsonfree
choiceintrinsicmotivationandselfreportedtaskinterest.

EffectsofRewardType
Onepossibilityisthatdifferenttypesofrewardsmayhavedifferenteffects.Toassessthispossibility,studiesthatusedeitherverbalortangiblerewardswereevaluated.
Verbalrewardsincludedpraiseorpositivefeedbacktangiblerewardsweremoney,toys,goldstars,andsoon.
Onewaytovisuallycomparetheresultsofmanystudiesinvolvestheuseoffunneldiagrams.Funnelgraphsareusedtoploteffectsize(xaxis)againstsamplesize(y
axis)foreachstudy.Theadvantageofafunneldisplayisthatitcapitalizesonawellknownstatisticalprinciple:thelargerthesample,theclosertheeffectsizewillcome
torepresentthetrueunderlyingeffectsize.Smallersamplesaremorepronetosamplingerrorandarelikelytodeviateconsiderablyfromthetruemeaneffectsize.For
thesereasons,thedistributionisexpectedtotaketheshapeofaninvertedfunnelthetipofthefunnelwillhoneinaroundthetrueeffectsize.
Figure8.1presentsthepatternofeffectsizesonhighinteresttasksforverbalandtangiblerewardonfreechoiceintrinsicmotivation.11Eachdatapointrep
Page122

Figure8.1
FunnelDistributionofEffectSizesonHighInterestTasksforVerbalandTangibleRewardsonFreeChoiceIntrinsic
Motivation

resentstheeffectsizefromanindividualstudy.Studiesfallingtotheleftofzeroareconsistentwiththeviewthatrewarddecreasesintrinsicmotivation.Thoseonthe
rightofzeroindicateincreasesinthemeasureofintrinsicmotivation.
Consideringtheresultsofverbalandtangiblerewardseparately,visualinspectionofthegraphsuggestssometendencyfortangiblerewards(opencircles)toreduce
freechoiceintrinsicmotivation.Verbalrewards(closedtriangles)appeartohaveproducedapositiveeffect.Inotherwords,verbalrewards(praiseandpositive
feedback)increasedfreechoiceperformancewhiletangiblerewardsreducedfreechoiceintrinsicmotivation.Metaanalysesshowedthattheseeffectswere
statisticallysignificant.
Onthetaskinterestmeasure,aninspectionofthefunnelgraphpresentedinFigure8.2indicatesthatpositiveeffectsemergefromverbalrewardstudies,whereas
tangiblerewardshoneinaroundzero.Themetaanalyticresultsshowedapositiveeffectforbothverbalandtangiblerewardshowever,themeaneffectfortangible
rewardswasverysmall.Thesedifferencessuggestthattheeffectsofrewarddependonthetypeofreward.
Page123

Figure8.2
FunnelDistributionofEffectSizesonHighInterestTasksforVerbalandTangibleRewardsonSelfReportedTask
Interest

EffectsofRewardExpectancy
Inexperimentsinvolvingtheuseoftangiblerewards,theproceduresforadministeringanddeliveringsuchrewardsvaried.12Oneprocedurewaswhethertherewards
wereexpected(promisedbeforeparticipantsengagedintheexperimentalactivity)orunexpected(deliveredbutnotpromisedbeforehand).Thus,thenextbreakdown
inouranalysisinvolvedassessingtheeffectsofdifferentrewardexpectancies.
Tangiblerewardsweresubdividedintounexpectedandexpectedcategories.Nosignificanteffectsweredetectedforunexpectedtangiblerewardsoneithermeasure
ofintrinsicmotivation.Thatis,ifparticipantsweregivenatangiblerewardbutnotoffereditpriortoengagingintheexperimentalactivity,intrinsicmotivationwasnot
altered.Expectedtangiblerewardsproducedasignificantnegativeeffectonthefreechoicemeasureandareliablepositiveeffectontheselfreportmeasure.

EffectsofRewardContingency
Atthenextlevelofanalysis,expectedtangiblerewardsweresubdividedintovariousrewardcontingencies.Thestudieswereclassifiedintosevencategories
Page124

Table8.2
HighInterestTasks:TheEffectsofTangibleExpectedRewardContingencies
FREECHOICEINTRINSICMOTIVATION
ExpectedRewardCondition NumberofStudies MeanEffectSize ChangeinIntrinsicMotivation
Tasknoncontingent 7 0.10(n.s.) Nochange
Fordoingwell 11 0.31* Decrease
Fordoingtask 50 0.30* Decrease
Forfinishingtask 6 0.24(n.s.) Nochange
Foreachunitsolved 20 0.16* Decrease
Forsurpassingascore 11 0.02(n.s.) Nochange
Forexceedingothers 11 0.18 Increase
SELFREPORTEDTASKINTEREST
ExpectedRewardCondition NumberofStudies MeanEffectSize ChangeinIntrinsicMotivation
Tasknoncontingent 6 0.17(n.s.) Nochange
Fordoingwell 6 0.04(n.s.) Nochange
Fordoingtask 38 0.13* Decrease
Forfinishingtask 6 0.32* Increase
Foreachunitsolved 20 0.15* Increase
Forsurpassingascore 11 0.24* Increase
Forexceedingothers 14 0.14* Increase
Note:*=significantatp<.05,n.s.=notsignificant.Positiveeffectsizesindicatehigherintrinsicmotivationforrewardedgroupscomparedtononrewardedgroups.
Negativeeffectsizesindicatelowerintrinsicmotivationforrewardedgroups

ofrewardcontingency:rewardsdeliveredregardlessoftaskinvolvement(tasknoncontingent)fordoingataskfordoingwellforfinishingorcompletingataskfor
eachproblem,puzzle,orunitsolvedforachievingorsurpassingaspecificscoreandformeetingorexceedingothers.
TheresultsofouranalysesarepresentedinTable8.2.,whichpresentsthenumberofstudiesincludedineachanalysis,themeaneffectsizeforeachreward
contingency,andanindicationofwhetherornottheeffectsizewasstatisticallysignificant.Aneffectofaround.20isconsideredsmall,.50ismoderate,andgreater
than.80islarge.13
Onfreechoiceintrinsicmotivation,noreliableeffectsweredetectedwhen
Page125

therewardsweretasknoncontingent,offeredforfinishingorcompletingatask,orofferedforattainingorsurpassingascore.Table8.2showssignificantnegative
effectsonfreechoicewhentherewardswereofferedfordoingatask,fordoingwellonatask,orforeachunitsolved.Asignificantpositiveeffectwasfoundwhenthe
rewardswereofferedformeetingorexceedingtheperformancelevelofothers.
Fortaskinterest,theanalysispresentedinTable8.2showsnoreliableeffectfortasknoncontingentreward,asmallbutsignificantnegativeeffectforrewards
offeredfordoing,andsignificantpositiveeffectsforeachoftheotherrewardcontingencies.
Thesefindingsindicatethatwhentherewardcontingencyisdefinedintermsoftheactualproceduresusedintheexperiments,negative,neutral,andpositiveeffects
areobtained.

EffectsofMaximumversusLessThanMaximumReward
Atthefinallevelofanalysis,weexaminedtheeffectsofmaximumversuslessthanmaximumreward.Maximumrewardreferstoaprocedurewherebyallparticipantsin
therewardconditionmettheperformancerequirementsofthestudyandreceivedthefullamountofreward.Lessthanmaximumrewardoccurredwhensome
participantswerenotabletofulfilltheperformancerequirementandweregivenlessthanthefullreward.
Therewasonlyonerewardcontingencyrewardsofferedperunitsolvedthatallowedforacomparisonbetweenmaximumandlessthanmaximumreward.14For
otherrewardcontingencies,toofewstudiesinvolvedlessthanmaximumreward,andsoametaanalysiswasnotfeasible.Whenrewardswereofferedforeachunit
solved,thefindingsonfreechoiceintrinsicmotivationshowedanonsignificanteffectinstudiesofmaximumrewardsandasignificantnegativeeffectinstudiesofless
thanmaximumreward.Theseresultssuggestthatthenegativeeffectofrewardperunitisassociatedwithparticipantsreceivinglessthanmaximumrewards.

SUMMARYOFOURMETAANALYTICFINDINGS
AsummaryofourmetaanalyticfindingsispresentedinFigure8.3.Theanalysesarearrangedhierarchically,withthemoregeneralproceduralcategoriesclosertothe
topofthediagram.Firstwepresenttheoveralleffectsofrewardstudiesarethenbrokendownintosubsetsofdifferentmoderatorvariables.Apositiveeffect(+)
indicatesthatrewardsenhancedthemeasureofintrinsicmotivationrelativetoacontrolcondition,anegativeeffect()indicatesadecreasefortherewardedgroup,
andazeroeffect(0)indicatesnoreliableeffect.
Onthefreechoiceintrinsicmotivationmeasure,intermsoftheoveralleffectsofreward,ourmetaanalysisindicatesnoevidenceforadetrimentaleffectofreward.
Thisfindingisimportantbecausemanyresearchersandwritersespousetheviewthatrewards,ingeneral,reducemotivationandperformance.Inaddi
Page126

Figure8.3
ASummaryoftheMetaanalysisComparingFreeChoiceIntrinsicMotivationandSelfReportedTaskInterest


Note:(0)indicatesnoreliableeffect()indicatesastatisticallysignificantnegativeeffectofreward(+)indicatesastatisticallysignificantpositive
effectofreward.

tion,manystudentsofpsychologyandeducationaretaughtthatrewardsareharmfulandthatrewardproceduresshouldbeavoidedinappliedsettings.Ourfindings
indicatethatthereisnoinherentnegativepropertyofrewardinstead,ourresearchdemonstratesthatrewardshaveeitherpositive,neutral,ornegativeeffects,
dependingonhowtheyareadministered.
InFigure8.3,theeffectsofallrewardsarefirstbrokenintohighandlowinteresttasks.Thefindingsconcerningfreechoiceintrinsicmotivationareinteresting.
Whenthetasksusedinthestudiesareoflowinitialinterest,rewardsincreasefreechoiceintrinsicmotivation.Thisfindingindicatesthatrewardscanbeusedto
enhancetimeandperformanceontasksthatinitiallyholdlittleenjoyment.Thisresultsuggeststhatrewardproceduresareonewaytocultivateinterestinanactivity.In
education,amajorgoalistoinstillmotivationandenjoymentofacademicactivities.Manyacademicactivitiesarenotofhighinitialinteresttostudents.Animplicationof
ourfindingisthatrewardscanbeusedtoincreasemotivationandperformanceonlowinterestacademicactivities.
Onhighinteresttasks,rewardsproduceadecreaseinfreechoiceintrinsicmotivation.Atfirstglance,thisseemstoprovidesupportforpopularassertionsabout
detrimentaleffectsofreward.Butasweproceedfromthegeneralclas
Page127

sificationofallrewardsunderhighinteresttaskstomorespecificclassifications,weseethatnegativeeffectsareconditionalandspecialized.
Verbalrewardsarefoundtoincreasefreechoice.Thisfindingreplicatestheresultsofpreviousmetaanalyses.Mostsocialinteractioninbusiness,education,and
clinicalsettingsinvolvestheuseofverbalpraiseandpositivefeedbackfrommanagers,teachers,andtherapists.Whenpraiseandotherformsofpositivefeedbackare
givenandlaterremoved,ourfindingindicatesthatpeoplecontinuetoengageintheactivityandperformatahighlevel.
Theeffectsoftangiblerewardonfreechoiceintrinsicmotivationdifferbyrewardexpectancy.Expectedrewardsarethosepromisedtoparticipantsbeforethe
experimentalsession.Unexpectedrewardsaredeliveredduringorfollowingthesessionbutnotpromisedbeforehand.Tangiblerewardsalonehavebeensodivided
becauseinmoststudiesthepresentationofverbalrewardwasunexpected.Atthislevelofanalysis,thedetrimentaleffectsofrewardonthefreechoicemeasureare
restrictedtotheinfluenceofexpectedtangiblereward.Whenrewardsaredeliveredunexpectedly(withoutadescriptionoftherewardcontingency),thereisno
evidenceofareliableeffect.Thissuggeststhatitisnottangiblerewards,perse,thatunderminemotivationinstead,theparticulareffectdependsoninstructionandthe
statementofcontingency.
Atthenextlevelofanalysis,inFigure8.3,expectedtangiblerewardsarecategorizedaccordingtothedescriptionoftherewardcontingency.Whentheofferof
rewardisunrelatedtotaskbehavior(tasknoncontingent),thereisnoevidenceforaneffectofrewardonthefreechoicemeasure.Ontheotherhand,whenpeopleare
offeredatangiblerewardfordoingataskorfordoingitwell,theychoosetodotheactivitylessinafreechoiceperiod.Generally,whenthedescriptionofthereward
contingencyimpliesthatrewardsarelooselytiedtoperformance,theevidencesuggeststhatpeopleshowasmallreductioninperformanceandinterest.
Figure8.3showsthatrewardsofferedforfinishingorcompletingataskhaveastatisticallynonsignificanteffectonfreechoice.Becausetherewerefewstudiesinthis
category,drawingafirmconclusionabouttheeffectispremature.Astrongerconclusioncanbedrawnfortheanalysisofrewardsofferedforeachunitsolved.When
participantsareofferedarewardforeachproblem,puzzle,orunitsolved,ourfindingsindicateanegativeeffectonfreechoice.Atthefinallevelofanalysis,involving
lessthanmaximumrewardandmaximumreward,theresultsshowthatthenegativeeffectonfreechoiceoccurswhenparticipantsobtainlessthanthefullreward.In
studiesoflessthanmaximumreward,participantsaregivenatimelimittosolveproblems.Thus,thenegativeeffectmaybearesultoftimepressureratherthanreward.
Anotherwaytounderstandthisresultistoconsiderwhatlessthanmaximalrewardsignifiestoparticipants.Ifpeoplearetoldthattheycanobtainacertainlevelof
rewardbutaregivenlessthanthatlevel,theyhavereceivedfeedbackinformationthatindicatesfailure.Inotherwords,thistypeofsituationmayrepresentfailure
feedbackratherthan
Page128

reward.Whenparticipantsarenotundertimepressureandareabletoobtainthemaximalreward,thereisnoreliableeffectonthefreechoicemeasure.
Whenrewardsareofferedformeetingorsurpassingascore,wefoundnoreliableeffectonfreechoice(seeFigure8.3).However,whenrewardsaregivenfor
exceedingtheperformancelevelofothers,theresultsshowasignificantpositiveeffectonfreechoice.
Thedataconcerningselfreportedtaskinterestprovidelittleevidencefordecrementaleffectsofanytypeofreward.Onhighinteresttasks,bothverbalandtangible
rewardshadanoverallpositiveeffect.Fortangiblerewards,asweproceededtosubsequentlevelsofanalysis,wedeterminedthatunexpectedrewardsdidnotalter
taskinterestandexpectedrewards(a)hadnoeffectwhentheywerenoncontingentorofferedfordoingwellonatask,(b)producedanegativeeffectwhentheywere
offeredfordoingthetask,and(c)ledtoincreasedexpressedtaskinterestwhentheywereofferedforfinishingatask,foreachunitsolved,forsurpassingascore,or
forexceedingothers.
Thus,ontaskinterest,theonlynegativeeffectoccurredwhentherewardsweretangible,expected,andgivensimplyforengaginginanactivity,withoutregardtoany
performancestandards.Nonegativeeffectwasdetectedwhentherewardswereofferedfordoingwell.Weshouldpointout,however,thattherewereonlysixstudies
inthedoingwellcategory(seeTable8.2).Thus,itispossiblethatthetrueeffectforthisrewardcontingencymayalsobenegative,butatthispointintime,thereare
toofewstudiestoyieldareliableestimate.
Overall,ourmetaanalysisshowedthatrewardscanbeusedtoproducebothnegativeandpositiveeffectsonmeasuresofintrinsicmotivation.Rewardscanbeused
toincreasemotivationandperformanceontasksthatareoflowinitialinterest.Onhighinteresttasks,positiveeffectsareobtainedwhenparticipantsareverbally
praisedfortheirworkandwhentangiblerewardsareofferedandexplicitlytiedtoperformancestandardsandtosuccess.Negativeeffectsareproducedwhenrewards
signifyfailureorareonlylooselytiedtobehavior.

DURABILITYOFREWARDEFFECTSONINTRINSICMOTIVATION

Deciandhiscolleaguesclaimedthatnegativeeffectsofrewardsarenottemporary.15Inasupplementalanalysis,Deciandhisresearchteamexaminedstudiesof
childreninwhichthefreechoiceassessmentwasconductedbothwithinoneweekfollowingtheremovalofrewardandafteroneweek.Theiranalysisshowednegative
effectsonfreechoiceforeachtimeofassessment.Theresearchersconcludedthattheirresultsindicatequiteclearlythatthephenomenonofextrinsicrewards
underminingintrinsicmotivationisnotmerelytransitory.16
AnexaminationofthestudiesincludedinDeciandhiscolleaguessupplementaryanalysisindicatesthatalmostallthestudiesinvolvedrewardsofferedfordoingthe
taskorfordoingwell.Whenthefreechoiceassessmentwas
Page129

conductedwithinaweekfollowingtheremovalofreward,10of12studiesinvolvedrewardsofferedfordoingwellorfordoingthetask.Of14studieswith
assessmentsconductedmorethanaweeklater,13wereconcernedwithrewardsofferedfordoingwell.Giventhis,ourinterpretationofDeciandhiscolleagues
findingsisthatitistangiblerewardsthatareofferedfordoingataskorfordoingwellonataskthatcontinuetoproduceanegativeeffectonfreechoiceintrinsic
motivation(evenwhentheassessmentofthefreechoicemeasureistakenaweekormorethanaweeklater).Thereisnoevidencetosuggestthatotherreward
contingencies(rewardsofferedforsuccess,formeetingorsurpassingaperformancestandard,orforexceedingothers)producenegativeeffectsovertime.
Anunresolvedissue,however,iswhetherthereisachangeinfreechoiceintrinsicmotivationovertime.InmostofthestudiesincludedinDeciandhisresearch
teamsanalysisofthedurabilityofrewardeffects,thefreechoiceassessmentwasonlymeasuredatonetime.Thatis,thefreechoiceassessmentwasdone
immediatelyfollowingtheexperimentalintervention,aweeklater,ormorethanaweeklater,butonlyatoneofthesetimes.Weexaminedsevenstudiesofrewards
offeredfordoingthetaskthatassessedwhethernegativeeffectsweremaintainedovertime.17Thesestudiesincludedtwomeasuresoffreechoiceintrinsicmotivation,
oneaftertheremovalofrewardandasecondmeasuredoneafewweekslater.Onlytwoofthesevenstudiesshowedasignificantnegativeeffectonthesecond
measure.18Theseresultssuggestthatrewardsofferedfordoingataskhavetransitoryeffectswhenmultiplemeasuresoffreechoicemotivationareused,butthesmall
numberofstudiesmakesitdifficulttodrawastrongconclusion.
Thisconclusionisstrengthened,however,byexaminingtheresultsfromstudiesusingrepeatedpresentationsofrewardfollowedbyrepeatedassessmentsofintrinsic
motivationmeasuresfollowingrewardremoval.AsdiscussedinChapter6,someresearcherstestedtheeffectsofrewardsonintrinsicmotivationbyexperimental
designsinwhichthesameindividualwasexposedtoabaselineperiod,arewardintervention,andareturntobaseline.19Participantsweremeasuredrepeatedlyduring
eachphaseoftheexperiment,andrewardswereshowntoincreasemeasuresofperformance,indicatingthattherewardsfunctionedasreinforcement.Theresultsof
theseexperimentsshowedthatparticipantsspentasmuch(ormore)timeonthetargetactivityinthepostrewardphaseastheydidintheinitial,baselineperiod.These
findingsindicatethatnegativeeffectsofrewarddonotpersistwhentaskperformanceisrewardedonrepeatedoccasions.
Takentogether,theresultsfromgroupdesignstudiesthatprovidedmorethanoneassessmentoffreechoiceintrinsicmotivationandtheresultsfromsinglesubject
designsinwhichparticipantswererewardedoveraperiodoftimeandrepeatedassessmentsofintrinsicmotivationweretakensuggestthatthenegativeeffectsof
rewardsare,infact,atransitoryphenomenon.
Page130

ASSESSMENTOFMEDIATIONALPROCESSES
InChapters4and5,wedescribedanumberoftheoriesthatmakepredictionsandattempttoaccountfortheeffectsofrewardsonintrinsicmotivation.Mosttestsof
theoreticalclaims(e.g.,cognitiveevaluationtheory,theoverjustificationhypothesis)considertheinfluenceofrewardonbehavioralmeasuresofintrinsicmotivation(free
choice)andattitudestowardthetask(selfreportedtaskinterest)withoutattemptingtodirectlyassesspresumedunderlyingpsychologicalprocesses.However,afew
studieshaveconsideredwhetherrewardproducedchangesofintrinsicmotivationaremediatedbychangesinperceivedselfdeterminationorachangeintheperceived
locusofcausality.Inthissection,weexaminestudiesthathavedirectlyassessedproposedmediationalprocesses.
Cognitiveevaluationtheoryholdsthatverbalpraisepresentedinanonthreateninginformationalmannershouldenhanceperceptionsofcompetencewithoutaffecting
perceivedselfdetermination.Sixstudieshaveevaluatedtheeffectsofverbalpraiseonperceivedcompetence.Perceivedcompetencewasindexedbyparticipants
reportsof(a)howcompetenttheyfeltfollowingtheirperformance,20(b)whetherfeelingsofcompetenceinfluencedhowhardtheyworkedonthetask,21and(c)how
wellparticipantsthoughttheyperformedrelativetoothers.22Inallsixexperiments,verbalrewardwasfoundtoincreasefeelingsofcompetencerelativetocontrol
participantswhodidnotreceivereward.However,thethreestudiesthatusedcausalmodelingtoassesstherelationshipsamongpraise,competence,andintrinsic
motivationproducedmixedresults.Onestudyreportedmediatingeffectsofcompetence23whereastwostudiesdidnotfindcompetencetohaveareliablemediating
effect.24Althoughadditionalreplicationsareneededbecauseofthemixedresults,thefindingssuggestthatverbalpraisemayincreasefeelingsofcompetencebutgive
littlesupporttotheviewthatintrinsicinterestismediatedbyenhancedperceptionsofcompetence.
Forcontingent,tangible,expectedrewards,cognitiveevaluationtheoryinvokesalossofperceivedselfdeterminationastheunderlyingmechanismthatleadsto
decreasesinintrinsicmotivation.Ina1999article,Eisenberger,Pierce,andCameronidentifiedfivestudiesthatmeasuredtheeffectsofexpectedtangiblerewardson
perceivedselfdetermination.25Allfivestudiesshowedpositiveeffectsofrewardonselfdetermination.26Contrarytothepredictionsofcognitiveevaluationtheory,the
resultsindicatedthatrewardsincreasedperceivedautonomy.Ametaanalysisoftheresultsproducedastatisticallysignificantpositiveeffect(d+=0.37).
Theoverjustificationhypothesisalsopredictsnegativeeffectsofexpected,tangible,contingentrewardsonmeasuresofintrinsicmotivation.Accordingto
overjustificationtheory,tangiblerewardsleadpeopletoviewthereward,ratherthantheirowninterest,asthelocusofcausalityforanactivity.Intwostudiesusing
expectedtangiblerewards,participantswereaskedtoindicatewhethertheyattributedtheirperformancetointrinsicfactors(e.g.,abilityandpersonaleffort)orto
extrinsicfactorssuchasthereward.Contrarytotheoverjustification
Page131

hypothesis,oneoftheexperimentsreportedanincreasedattributionofperformancetointrinsicfactors27theotherexperimentreportednoreliableeffect.28
Overall,andconsistentwithcognitiveevaluationtheory,allsixstudiesofverbalrewardreportedincreasedperceptionsofcompetence.However,twoofthree
studiesemployingcausalmodelingfailedtofindamediatingeffectofcompetence.Itisimportanttoconductadditionalcausalmodelingstudieswithverbalrewardto
providemorecompleteevidenceonthemediatinginfluenceofcompetence.Ofthefivestudiesforwhichcognitiveevaluationtheorypredictsadecrementaleffecton
selfdetermination,allreportedanincrementaleffect.Theseresultsdonotsupporttheassumptionofcognitiveevaluationtheorythattangible,expected,contingent
rewardsreducepeoplesselfdetermination.Intermsofevidenceaboutmediationalprocessesproposedbytheoverjustificationhypothesis,atpresent,theevidenceis
toolimitedtodrawdefinitiveconclusions.

CONCLUDINGREMARKS
Overall,themetaanalysispresentedinthischaptershowsthatrewardscanbeusedtoproduceeithernegative,neutral,orpositiveeffectsonmeasuresofintrinsic
motivation.Rewardscanbeusedtoincreasemotivationandperformanceontasksthatareoflowinitialinterest.Onhighinteresttasks,positiveeffectsareobtained
whenparticipantsareverballypraisedfortheirworkandwhentangiblerewardsareofferedandexplicitlytiedtoperformancestandardsandtosuccess.
Althoughseveralwritershavearguedthatthenegativeeffectsofrewardarepervasiveandlonglasting,theresultsofourmetaanalysisindicatethatanydetrimental
effectsofrewardarelimited.Specifically,negativeeffectsareproducedonhighinteresttaskswhentangiblerewardssignifyfailureorarelooselytiedtobehavior.This
effectdoesnotappeartobelonglastinginfact,whenrewardsaredeliveredrepeatedlyandintrinsicmotivationisassessedonmorethanoneoccasion,theevidence
suggeststhatthenegativeeffectsarefleetingandtransitory.
Intermsofmediationalprocesses,weevaluatedstudiesthatexaminedtheeffectsofmediatorsproposedbycognitiveevaluationtheoryandtheoverjustification
hypothesistounderliechangesinintrinsicmotivation.Atpresent,thebestevidenceisthatrewardsarenotviewedbypeopleascontrollingorrestrictivetotheirsenseof
freedom.

NOTES
1.ThisanalysisisbasedontheworkofCameron,Banko,andPierce,2001.
2.InaccordwiththemetaanalysisbyDeci,Koestner,andRyan(1999),ourfreechoicemeasureincludesfreetimeontaskwhentherewardswereremovedand,whentimemeasures
werenotavailable,performanceduringthefreechoiceperiod.AsdidDecietal.,wecombinedperformanceandtimemeasurestomakeupthefreechoice
Page132

intrinsicmotivationindex(wefoundnosignificantdifferencesintheanalyseswhenonlytimemeasureswereanalyzed).
3. ThedatabasesofCameronandPierce(1994)andDeci,Koestner,andRyan(1999)wereincludedinthepresentmetaanalysis.
4. ThesamedissertationsincludedbyDeci,Koestner,andRyan(1999)wereincludedinthepresentmetaanalysis,andoneadditionaldissertation(Adorney,1983)wasadded.
5. Tocodeforinitiallevelsoftaskinterest,weusedthesameproceduresasDeci,Koestner,andRyan,1999.
6. Althoughalltypesofrewardswereclassifiedintermsofcontingency,itwasattheleveloftangibleexpectedrewardthatitbecamenecessarytoanalyzestudiesintermsofthe
variousrewardcontingencies.
7. TheMETAprogramwasdevelopedbySchwarzer(1991)themetaanalyticmethodusedinthisanalysisisbasedontheweightedintegrationmethoddescribedinHedgesandOlkin
(1985).
8. Deci,Koestner,andRyan,1999.
9. ThisanalysiswasnotconductedbyDeci,Koestner,andRyan(1999).Theresults,however,areinaccordwithCameronandPierce(1994).
10.Thereweretoofewstudiesthatusedlowinteresttaskstoconductmetaanalysesoftheseparateeffectsofrewardtype,rewardexpectancy,andrewardcontingency.Howeverwe
didexaminewhethertherewereanydifferencesasaresultofthesemoderators.Onthefreechoicemeasure,onlyonestudyincludedaconditionthatusedaverbalreward(theeffect
waspositive).Fortangiblereward,onlyonestudyincludedanunexpectedrewardcondition(theeffectwaspositive).All12studieswithlowinteresttasksincludedanexpected
tangiblerewardconditioncomparedwithanonrewardcontrol,themeaneffectwassignificantlypositive.Ninestudiesinvolvedofferingthetangiblerewardfordoingthetaskon
thefreechoicemeasure,theeffectremainedsignificantlypositive.Forselfreportedtaskinterest,noreliableeffectswerefoundunderanyoftheconditions.
11.Becausethesamplesizesintheoriginalstudiesrangedfromasfewas10participantstoasmanyas500,inFigures9.1and9.2,samplesizeisrepresentedincommonlogs.Log1
indicatesasamplesizeof10,2indicates100,and3indicates1,000.
12.Inmoststudiesofverbalreward,therewardswereunexpectedandthemeaneffectwaspositiveapositiveeffectwasalsofoundinthefivestudiesthatusedexpectedrewards.In
addition,verbalrewardsweregenerallydeliveredsimplyfordoingataskandwerenotcontingentonanyspecificlevelofperformance(again,theeffectswerepositive).
13.Cohen,1988.
14.Othercontingenciesthatusedmaximumorlessthanmaximumrewardinvolvedtangiblerewardsofferedfordoingwell,formeetingorsurpassingaspecificstandard,andfor
exceedingothers.Ineachofthesecategories,themajorityofstudiesusedmaximumreward.Thus,ananalysisofmaximumversuslessthanmaximumrewardcouldnotbeconducted
inthesecategories.
15.Deci,Koestner,andRyan1999,2001.
16.Ibid.,p.650.
17.Chung,1995LovelandandOlley,1979Morgan,1983,Experiments1and2OgilvieandPrior,1982Ross,1975,Experiment1ShiffmanKauffman,1990.
18.Morgan,1983,Experiments1and2.
19.DavidsonandBucher,1978FeingoldandMahoney,1975Mawhinney,Dick
Page133

inson,andTaylor,1989Skaggs,Dickinson,andOConnor,1992Vasta,Andrews,McLanghlin,Stirpe,andComfort,1978.
20.Vallerand,1983VallerandandReid,1984.
21.Shanab,Peterson,Dargahi,andDeroian,1981.
22.Sansone,1989Sansone,Sachau,andWeir,1989.
23.VallerandandReid,1984.
24.Sansone,1989Sansone,Sachau,andWeir,1989.
25.Eisenberger,Pierce,andCameron,1999.
26.Eisenberger,Rhoades,andCameron,1999,Experiment1FreedmanandPhillips,1985OverskeidandSvartdal,1996,Experiments1and2ShiffmanKauffman,1990.
27.WimperisandFarr,1979.
28.Arnold,1985.
Page134

Appendix8A:TechnicalAspectsofOurMetaAnalysisandStatisticalFindings
Inthissectionweprovideadetaileddescriptionofthemetaanalyticproceduresweusedtoassesstheeffectsofrewardsonintrinsicmotivation.Inaddition,we
presentstatisticalanalysesofourresults.Manyoftheissuesdiscussedinthissectionarehighlytechnicalbutnonethelessfundamentaltoanunderstandingofameta
analysisofthisliterature.

OURMETAANALYTICPROCEDURES
Metaanalysisinvolvesseveralsteps,thefirstofwhichistospecifytheresearchquestions.Ourmetaanalysiswasdesignedtoaddressthefollowingquestions:
Dorewardsgenerallyleadtoadecreaseinintrinsicmotivation?
Underwhatconditionsdorewardsleadtoeitherdecreasedorincreasedintrinsicmotivation?
Otherstepsinthemetaanalysisinvolvedspecifyingthecriteriaforincludingandexcludingstudies,collectingallexperimentsthatmetthecriteria,readingandcodingthe
studies,calculatingeffectsizesforeachstudy,andperformingstatisticalanalysesofeffectsizes.

SelectionofStudies
Ourmetaanalysisincludedstudiesthatwereusedinpreviousmetaanalysesonthetopic.1Aswell,afewotherrelevantstudieswerelocatedthroughasearchof
PsycINFO.Inaddition,weincludedunpublisheddoctoraldissertations.ThemetaanalysisstartedwithDecis1971experiment2andincludesresearchpublished
before1999.
Theanalysisentailedassessingtheeffectsofrewardfromstudiesthatusedbetweengroupdesigns.Criteriaforincludingstudiesinthesamplewerethat:(a)thestudy
includedanexperimentalmanipulationofarewardconditionandanonrewardedcontrolgroup,(b)anycharacteristicofrewardedparticipantswaseitherheldconstant
orvariedandwasrepresentedidenticallyforbothrewardedandcontrolgroups,3(c)studieswerewritteninEnglish,and(d)studiesassessedtheeffectsofreward
versusnorewardonthetwomainmeasuresofintrinsicmotivation(freechoiceandselfreportedtaskinterest).
Intrinsicmotivationhasbeenmeasuredinavarietyofways.Researchershaveexaminedtheeffectsofrewardonmeasuressuchasquantityandqualityof
performanceduringarewardedphase,participantswillingnesstovolunteerfor
Page135

futureresearch,timespentonanexperimentaltaskinafreechoiceperiodafterrewardisremoved,performanceduringthefreechoiceperiod,andquestionnaire
measuresoftaskinterest.InChapter7,wedescribedhowresearchershavefounddifferenteffectsofrewardsdependingonhowintrinsicmotivationisindexed.4Deci
andhiscolleaguesarguedthatameasuretakenduringarewardphaseisnotapuremeasureofintrinsicmotivationbecauseitreflectsbothextrinsicandintrinsic
motives.5Instead,theyclaimedthatthebestmeasuresofintrinsicmotivationarefreechoiceintrinsicmotivation(timespentonthetaskfollowingtheremovalofreward
orperformanceonthetaskduringthefreechoiceperiod)andselfreportsoftaskinterest(taskliking,enjoyment,satisfaction,ortaskpreference).Thus,forthe
presentmetaanalysis,studieswereselectedthatusedeitherthefreechoiceorselfreportmeasuresofintrinsicmotivation.6
Onecriticismofthemetaanalytictechniqueisthatresearchersoftenlumptogetherdifferentmeasures.7Thishasbeenreferredtoastheapplesandoranges
problem,inthatitisarguedthatlogicalconclusionscannotbedrawnfromcomparisonsofstudiesusingdifferentmeasuresofthesameconstruct(intrinsicmotivation).8
Toavoidthisproblem,weconductedseparateanalysesfortheeffectsofrewardonfreechoiceintrinsicmotivationandselfreportedtaskinterest.Theoverallsample
consistedof145studies,ofwhich115studiesassessedtheeffectsofrewardonfreechoicemeasuresofintrinsicmotivationand100includedquestionnairemeasures
oftaskinterest.

CodingofStudies
Onceallrelevantstudieshadbeencollected,eachstudywasreadandcoded.Thefollowingaspectsofthestudieswerecoded:(a)tasktype(highorlowinterest),(b)
rewardtype(verbalortangible),(c)rewardexpectancy(expectedorunexpected),(d)rewardcontingency,and(e)whetherparticipantsreceivedmaximumorless
thanmaximumreward.
Theclassificationofstudiesbyrewardcontingencyhasbeensomewhatcontroversial.Inthemostrecentmetaanalysisofthisliterature,Deciandassociates(Deci,
Koestner,andRyan,1999)usedcognitiveevaluationtheorytoguidetheirclassificationofrewardcontingencies.However,theircategoriesweretoobroad.The
problemisthatstudiesusingdifferentproceduresandproducingdifferentfindingswerepooledtogether,withtheresultthattheeffectswereobscuredwhentheresults
wereaggregated.Oursolutiontothisproblemwastoreturntotheoriginalstudiesandcodethecontingencyintermsoftheactualproceduresused.Usingaprocedural
definitionofrewardcontingencies,weidentifiedsevendifferenttypesofcontingencies(tasknoncontingentrewardsgivenregardlessoftaskinvolvementand
rewardsgivenfordoingthetask,doingwell,finishingorcompletingatask,eachproblemorunitsolved,achievingorsurpassingaspecificscore,andmeetingor
exceedingtheperformanceofothers).AdefinitionofeachofthesecontingenciesisprovidedinChapter8inTable8.1.
Insomestudies,therewasnotenoughinformationtocodespecificcharac
Page136

teristicsofreward.9Inaddition,acoupleofstudiesusedacontingencythatdidnotfitintoanyofthesevencategoriesdescribedhere.10Thesestudieswereincludedin
overallanalysesofrewardbutwereomittedfromanalysesdesignedtoassesstheeffectsofspecificrewardconditions.Descriptivecharacteristicsandeffectsizesofthe
reviewedstudiesaresummarizedinAppendix8B.

IntercoderReliability
Toensurereliabilityofcoding,anindependentcoderwasgiventhedefinitionsforeachcontingency(Table8.1)andasampleof32studiestocode(eachofthestudies
involvedexpectedtangiblerewards).11Reliability,calculatedaspercentageagreement,was97percent(31of32studies).Onestudyincludedaconditioninwhich
participantswereofferedarewardtotakepictures.12Theissuewaswhetherthiscontingencyinvolvedrewardsimplyfordoingthetaskorforfinishingthetask.The
secondauthorwasbroughtintocodethestudyhepointedoutthatparticipantsintherewardconditionwerenotrequiredtocompleteorfinishthetaskinorderto
obtaintherewardandthattheauthorofthestudystatedthattherewarddidnotimplythatthesubjecthaddonewellonthetask,onlythats/hehadengagedinit.13
Hence,therewardcontingencywasclassifiedasrewardofferedfordoingthetask.

ComputationandAnalysisofEffectSizes
Thepresentmetaanalysiswasconductedbyemployingahierarchicalbreakdownoftheeffectsofrewardsonintrinsicmotivation.Theproceduresdescribedhereare
basedontheweightedintegrationmethod.14
Onceallrelevantstudieshadbeenidentified,thestatisticalresultofeachstudywastransformedintoaneffectsizebyconvertingthefindingsfromeachstudyintoa
standarddeviationunit.Intherewardsandintrinsicmotivationliterature,theeffectsizefromeachstudyindicatestheextenttowhichtheexperimentalgroup(rewarded
group)andthecontrolgroup(nonrewardedgroup)differedinthemeanscoresonmeasuresofintrinsicmotivation(e.g.,freechoicetaskinterest).Initssimplestform,
theeffectsize(g)isthedifferencebetweenthemeansoftherewardedgroupandthemeansofthenonrewardedcontrolgroupdividedbythepooledstandard
deviationofthisdifference.
Foreachstudy,wecalculatedeffectsizes(g)foreachcomparisonofarewardedgrouptoanonrewardedgrouponfreechoiceandselfreportmeasuresofintrinsic
motivation.Positiveeffectsizesindicatethatrewardsproducedanincreaseinmeasuresofintrinsicmotivationrelativetoacontrolgroup,negativeeffectsizesdenotea
decrease,andaneffectsizeof0.00indicatesnodifference.
Oneprobleminmetaanalysisariseswhenstudiesfailtoprovideenoughinformationtocalculateeffectsizes.Whenmeansandstandarddeviationsarenotavailable,
effectsizescanbecalculatedfromttests,Fstatistics,andplevel
Page137

values.15However,insomecases,theremaystillbeinsufficientinformationtoobtainaneffectsize.Themetaanalystcanwritetotheresearchersandtrytoobtainthe
missingdata.Whenthedatacannotbeprocured,thestudycanbeexcludedfromtheanalysesorassignedaneffectsizeof0(indicatingnodifferencebetweenthe
experimentalandcontrolgroups).
Ithasbeenarguedthatincludingzeroeffectsizesisaconservativestrategyifasignificanteffectisdetectedinspiteoftheinclusionofzeros,thecontentionisthatthe
resultswouldnotalterifmissingdatawereavailable.16Ontheotherhand,ifonesbiasistowardnoeffect(thatis,wearesatisfiedifthetreatmentisnotharmful),
includingzerosfavorsthisconclusion.Onestrategyfordealingwiththisissueistoconducttheanalyseswithzerosincludedandexcluded.17
Whentherewasnotenoughinformationtocalculateaneffectsizefromastudy,weattemptedtocontacttheresearchers.Fromalistof22researchers,wewere
abletolocateemailaddressesfor9.Emailmessagesweresentrequestingthemissingdata.Although8peoplereplied,6couldnotlocatethedataandsoonly2
provideduswithdata.18Whenwecouldnotobtainmissingdata,weimputedaneffectsizeof0.Eachanalysiswasconductedwithzerosbothincludedandexcluded.
Wereporttheanalyseswiththezerosincluded.However,whenmeaneffectsizeswerealteredtoanyextentbytheinclusionofzeros,wereporttheanalysiswithand
withoutzeros.
Aftereffectssizes(g)werecalculatedforeachrelevantstudy,anoverallmeaneffectsize(d+)wasobtained.First,gswereconvertedtodsbycorrectingthemfor
bias(gisanoverestimationofthepopulationeffectsize,particularlyforsmallsamples).19Theoverallmeaneffectsizeisobtainedbyweightingeacheffectsizebythe
reciprocalofitsvarianceandaveragingtheweightedds.Thisproceduregivesmoreweighttoeffectsizesthataremorereliablyestimated.Thecalculationofmean
effectsizesprovidesasignificancetest(whetherthevaluedifferssignificantlyfrom0)anda95percentconfidenceinterval(whentheconfidenceintervalcontains0,the
resultssuggestthatthereisnoevidenceofastatisticallyreliableeffect).

ProceduresinaHierarchicalMetaanalysis
Inthepresenthierarchicalmetaanalysis,allstudieswereincludedinanoverallanalysis.Wethensearchedformoderatorvariables.Thestudieswerebrokenoutby
onekeymoderator,thenanother,andsoon.Themoderatorsthatwechosetoexaminewerebasedontheoreticalconsiderations(e.g.,levelsofinitialtaskinterest,
rewardtype,rewardexpectancy,andrewardcontingency).
Somemetaanalystsrecommendusinghomogeneityteststoascertainwhetheramoderatoranalysisisnecessary.20Ahomogeneitystatisticisusedtodetermine
whetherasetofeffectsizesinasamplesharesacommoneffectsize(i.e.,isconsistentacrossstudies).Essentially,theprocedureistouseachisquarestatistic,Q,with
k1degreesoffreedom,wherekisthenumberofeffectsizes.
Page138

Thenullhypothesisisthattheeffectsizesarehomogeneous(i.e.,effectsizesinagivenanalysisareviewedasvaluessampledfromasinglepopulationvariationineffect
sizesamongstudiesismerelyduetosamplingvariation).WhenQisstatisticallysignificant,theimplicationisthatmoderatoranalysesshouldbeconducted.Theoriginal
setofstudiesisthenbrokenintosubsetsuntilthechisquarestatisticswithinthesubgroupsarenonsignificant.
Ateachlevelofouranalysis,ahomogeneitystatistic(Q)wascalculatedtodeterminewhetherthesetofeffectsizescouldbeconsideredhomogeneous.WhenQwas
significant,weproceededwithfurthermoderatoranalyses.Inafewcases,homogeneitycouldnotbeobtainedevenafterathoroughexaminationofpotential
moderators.Inthesecases,weconductedtheanalysisbyremovingoutliers,aswasdoneinpreviousmetaanalyses.21Outlierswereexaminedinanattempttoexplain
theirextremevalues.Ateachlevelofouranalysis,wereportmeaneffectsizesand95percentconfidencelevelintervals.However,weshouldpointoutthatmaking
conclusionsbasedonheterogeneoussamplesmaybemisleading.Inahierarchicalbreakdown,interpretationsshouldfocusonthehomogeneouseffectsatthebottom
leveloftheanalysis.22
Forseveralstudiesinouranalyses,morethanoneeffectsizewascalculated.Forexample,ifasinglestudyassessedfreechoiceandusedtwotypesofexpected
tangiblerewards(e.g.,rewardsofferedfordoingthetaskandrewardsofferedforsurpassingacertainscore)plusacontrolgroup,twoeffectsizeswerecalculated.In
ordertosatisfytheindependenceassumptionofmetaanalyticstatistics,onlyoneeffectsizeperstudywasenteredintoeachanalysis.23Whentwoormoreeffectsizes
fromonestudywereappropriateforaparticularanalysis,theseeffectsizeswereaveraged.24Toillustrate,fortheestimateoftheoveralleffectofrewardonthefree
timemeasureofintrinsicmotivation,somestudiesassessedtheeffectsofseveraltypesofrewards.Ifasinglestudy,forexample,containedtwoormorerewardgroups
(e.g.,expectedrewardandunexpectedreward)andacontrolcondition,thetwoeffectsizeswereaveragedsothatthestudycontributedonlyoneeffectsizetothe
overallanalysisofreward.
Forananalysisoftheeffectsofexpectedrewardonintrinsicmotivation,onlytheoneappropriateeffectsizefromthestudywasused.Thisstrategyretainedasmuch
dataaspossiblewithoutviolatingtheassumptionofindependence.25Thus,subcategories(e.g.,rewardsofferedfordoingthetask,fordoingwell,etc.)maycontain
moreeffectsizesthanthesuperordinatecategory(expectedtangiblereward).Forexample,forallrewardsonthefreechoicemeasure(overbothhighandlow
interesttasks),therewere126effectsizes,butonly115oftheseareindependent(severalarewithinthesamestudy).
Afteralleffectsizeswerecalculated,thepresentanalyseswererunontheMETAcomputerprogram.26Theprogramconvertsgs(effectsizes)todsmeanweighted
effectsizes(d+s)areobtainedstatisticaltestsareprovided,95percentconfidencelevelintervalsareconstructedaroundthemeans,andahomogeneitystatistic,Q,is
computed.
Page139

STATISTICALRESULTSOFOURMETAANALYSIS
InTable8A.1,wepresenttheresultsforourmetaanalysisuptothelevelofrewardcontingency.Table8A.1presentsmeanweightedeffectsizes(d+s)and95
percentconfidencelevelintervalsforeachanalysis.Meaneffectsareconsideredstatisticallysignificantwhentheconfidenceintervaldoesnotincludezero.Inthe
presentmetaanalysis,positiveeffectsizesindicatethatrewardproducesincreasesinintrinsicmotivation,negativeeffectsizessupporttheclaimthatrewardsundermine
intrinsicmotivation,andzeroeffectsindicatenoevidenceforaneffectofreward.Aneffectsizeofaround.20isconsideredsmall,.50ismoderate,andgreaterthan
.80islarge.27

AllRewards
First,theoveralleffectsofrewardwereanalyzedacrossallconditionsandacrosshighandlowinteresttasks.Onthefreechoicemeasure,Table8A.1indicatesthat
therewasnoreliableeffect(d+=.08,CI=.12,.02).Onthemeasureofselfreportedtaskinterest,asmall,significantpositiveeffectwasdetected(d+=.12,CI
=.07,.16).28Onboththefreechoiceandselfreportmeasures,however,thesetsofstudiesweresignificantlyheterogeneous,suggestingthenecessityofamoderator
analysis.Thus,atthenextlevelofanalysis,wedividedstudiesintothosewithlowversushighinteresttasks.

TheEffectsofRewardsonLowInterestTasks
Whenrewardeffectswereanalyzedfortaskswithlowinitialinterest,(seeTable8A.1),wefoundastatisticallysignificantpositiveeffectonthefreechoicemeasure
(d+=.28,CI=.07,.47)therewasnoreliableeffectonselfreportedtaskinterest(d+=.12,CI=.06,.30).Thesefindingsindicatethatwhenataskisnotinitially
interesting,rewardsenhancefreechoiceintrinsicmotivationbutverbalexpressionsoftaskinterestdonot.29
Althoughthestudiesinthisanalysiswereconsideredhomogeneous(thatis,Qwasnotsignificant),weexaminedwhethertherewereanydifferencesamongdifferent
typesofrewards,expectancies,andcontingencies.Onthefreechoicemeasure,onlyonestudyincludedaconditionthatusedaverbalreward(theeffectwaspositive).
Fortangiblereward,onestudyincludedanunexpectedrewardcondition(theeffectwaspositive).Allthe12studieswithlowinteresttasksincludedanexpected
tangiblerewardconditioncomparedwithanonrewardcontrol,themeaneffectwassignificantlypositive(d+=0.26,CI=.06,.45).Atotalof9studiesinvolved
offeringtherewardfordoingthetaskonthefreechoicemeasure,theeffectremainedsignificant(d+=.26,CI=.03,.48).Forselfreportedtaskinterest,noreliable
effectswerefoundunderanyoftheconditions.
Page140

Table8A.1
HierarchicalAnalysisoftheEffectsofRewardsonMeasuresofIntrinsicMotivation
FREECHOICEINTRINSICMOTIVATION
Analysisoftheeffectsofreward K N d+ 95%CI
Allreward 115 8176 0.08 0.12,0.02
Lowinitialtaskinterest 12 429 0.28* 0.07,0.47
Highinitialtaskinterest 114 7888 0.09* 0.14,0.04
Verbalreward 25 1374 0.31* 0.20,0.41
Tangiblereward 102 6942 0.17* 0.22,0.12
Unexpectedreward 9 375 0.02 0.18,0.22
Expectedreward 101 6703 0.18* 0.23,0.13
SELFREPORTEDTASKINTEREST
Analysisoftheeffectsofreward K N d+ 95%CI
Allreward 100 8028 0.12* 0.07,0.16
Lowinitialtaskinterest 11 503 0.12 0.06,0.30
Highinitialtaskinterest 98 7547 0.12* 0.07,0.17
Verbalreward 24 1584 0.32* 0.22,0.43
Tangiblereward 83 6354 0.08* 0.03,0.13
Unexpectedreward 5 299 0.03 0.20,0.26
Expectedreward 81 6138 0.08* 0.03,0.13
Note:Analysesmarkedwithaindicatethatthesampleofeffectsizeswassignificantlyheterogeneous.K=numberofstudiesN=totalsamplesized+=mean
weightedeffectsizeeffectsizesmarkedwithanasteriskwerestatisticallysignificantatp<.0595%CI=95%confidenceinterval.

TheEffectsofRewardsonHighInterestTasks
Forhighinteresttasksthemeaneffectsizeonfreechoice(Table8A.1)showedasmallbutsignificantnegativeeffect(d+=0.09,CI=.14,.04)thesetofeffect
sizes,howeverwasheterogeneous.30Themeaneffectsizeforselfreportedtaskinterestwaspositiveandsmallbutsignificant(d+=.12,CI.07,.17)thesampleof
effectsizeswasalsoheterogeneous.

VerbalRewards
Table8A.1showsthatverbalrewardswerefoundtosignificantlyenhancebothfreechoiceintrinsicmotivation(d+=.31,CI=.20,.41)andselfreportedtask
interest(d+=.32,CI=.22,.43).31Thatis,whenparticipantsweregivenpositivefeedbackorverballypraised,theyspentmoretimeonataskandper
Page141

formedatahigherleveloncethepraisewasnolongerforthcomingthandidanonrewardedcontrolgroup.Theyalsoexpressedgreaterinterestinthetask.
Onthefreechoicemeasure,thesetofeffectsizeswashomogeneous,suggestingthatnofurtherbreakdownswerenecessary.Inmoststudiesofverbalreward,the
rewardswereunexpectedandthemeaneffectwaspositiveapositiveeffectwasalsofoundinthefivestudiesthatusedexpectedrewards.Inaddition,verbalrewards
weregenerallydeliveredsimplyfordoingatask,notcontingentonanyspecificlevelofperformance(again,theeffectswerepositive).Whentheeffectsofverbal
rewardonfreechoicewereexaminedwithchildrenversusadults(mainly,collegestudents),childrenshowedasmallerpositiveeffect(k=10,N=320,d+=.22,CI
=.04,.39)thanadults(K=15,N=844,d+=.36,CI=.22,.49).32
Onthetaskinterestmeasure,thesetofeffectsizesforverbalrewardwassignificantlyheterogeneous.Weconductedmoderatoranalysesofchildrenversusadults
andexpectedrewardversusunexpected.Meaneffectsizesforeachoftheseanalysesremainedsignificantlypositivebuthomogeneitywasstillnotobtained.Inalmost
allthestudies,therewardsweregivenfordoingthetaskhence,thisrewardcontingencycouldnotbeamoderator.
Toobtainhomogeneity,threestudieswereremovedfromtheanalysisthesewerethesameoutliersremovedbyDeci,Koestner,andRyan(1999).Inspectionofthe
outliersindicatedthattwoofthestudiesproducedlargepositiveeffects33thesestudiesdidnotdifferinobviouswaysfromotherstudiesinthesampleexceptfortheir
tendencytogenerateextremevaluesofeffectsize.Thethirdoutlierproducedanegativeeffect(.46).34Inthisstudy,controlparticipantswerecomparedto
participantswhowerepraisedfortheirperformanceonthetaskaswellastoanothergroupwhosememberswerealsopraisedbutalsotoldthattheyshouldbedoing
well.ThesecondverbalrewardconditionproducedthenegativeeffectandwasdifferentfromverbalrewardusedinotherstudiesDeciandhiscolleagueshavetermed
thisacontrollingreward.35Whentheoutlierswereremovedfromtheanalysisofverbalrewardsonthetaskinterestmeasure,thesetofstudieswashomogeneous
andthemeaneffectremainedsignificantlypositive(K=21,N=1,194,d+=.32,CI=.21,.44).36

TangibleRewards
Whentheeffectsoftangiblerewardsonhighinteresttaskswereanalyzed,(seeTable8A.1),wefoundasmallsignificantnegativeeffectonthefreechoicemeasure
(d+=.17,CI=.22,.12)andasmall,positive,reliableeffectonselfreportedtaskinterest(d+=0.08,CI=.03,.13).Boththesesamplesofeffectssizeswere
significantlyheterogeneousandrequiredafurthermoderatoranalysis.
TheEffectsofTangibleRewardExpectancies.Tangiblerewardsweresubdividedintounexpected(rewardsdeliveredwithoutastatementofthecontingency)and
expected(rewardsdeliveredafterastatementofcontingency).Nosignificanteffectsweredetectedforunexpectedtangiblerewards(seeTable8A.1)
Page142

Figure8A.1
TheEffectsofExpectedTangibleRewardContingenciesonFreeChoiceIntrinsicMotivationunderHighLevelsofInitial
TaskInterest


Note:K=numberofstudiesN=totalsamplesized+=meanweightedeffectsizestatisticallyreliableeffectsizesaremarkedwithanasterisk(*
=p<.05**=p<.01).Positiveeffectsizesindicatehigherintrinsicmotivationforrewardedversuscontrolgroupnegativeeffectsizesindicatelower
intrinsicmotivationforrewardedgroups.Numbersinparenthesesrepresent95%confidenceintervals.Alleffectsizesarebasedonhomogeneous
samples.

andthesampleswerehomogenous.Expectedtangiblerewardsproducedanegativeeffectonthefreechoicemeasure(d+=.18,CI=.23,.13)andapositive
effectontheselfreportmeasure(d+=.08,CI=.03,.13),butboththesesamplesweresignificantlyheterogeneous.
TheEffectsofExpectedTangibleRewardContingencies.Forthenextlevelofanalysis,expectedtangiblerewardsweresubdividedintovariousreward
contingencies.
ResultsofouranalysisonthefreechoicemeasurearepresentedinFigure8A.1.Noreliableeffectsweredetectedwhentherewardsweretasknoncontingent,
offeredforfinishingorcompletingatask,orofferedforattainingorsurpassingascore.Figure8A.1showssignificantnegativeeffectswhentherewardswereoffered
fordoingatask,fordoingwellonatask,andforeachunitsolved.Asignificantpositiveeffectwasfoundwhentherewardswereofferedformeetingorexceedingthe
performancelevelofothers.37
Whenrewardswereofferedfordoingatask,theeffectwassignificantly
Page143

Figure8A.2
TheEffectsofExpectedTangibleRewardContingenciesonSelfReportsofTaskInterestunderHighLevelsofInitial
TaskInterest


Note:K=numberofstudiesN=totalsamplesized+=meanweightedeffectsizestatisticallyreliableeffectsizesaremarkedwithanasterisk(*)
=p<.05**=p<.01).Positiveeffectsizesindicatehigherintrinsicmotivationforrewardedversuscontrolgroupnegativeeffectsizesindicatelower
intrinsicmotivationforrewardedgroups.Numbersinparenthesesrepresent95%confidenceintervals.Alleffectsizesarebasedonhomogeneous
samples.

negative(K=57,N=2,910,d+=.35,CI=.43,.27),butnothomogeneous.Althoughwesearchedformoderators(salientversusnonsalientreward,children
versusadults,andtimeofrewarddelivery),analysesofthesevariablesdidnotresultinhomogeneoussamples.Asaresult,outlierswereidentifiedandomitted.The
meaneffectsizewithoutliersremovedispresentedinFigure8A.1.38
Ourfindingsforfreechoiceindicatethatwhenrewardcontingencyisdefinedintermsofexperimentalprocedures,negative,neutral,andpositiveeffectsare
obtained.OurresultsforthetaskinterestdataarepresentedinFigure8A.2.Theanalysisshowsnoreliableeffectfortasknoncontingentrewardsasmall,reliable,
negativeeffectforrewardsofferedfordoing,andsignificant,positiveeffectsforeachoftheothercontingencies.39Inoneanalysis,rewardsofferedforeachunit
completed,whenallstudieswereincluded,theeffectwaspositive(K=22,N=1,161,d+=.19,CI=.08,.31),butsignificantlyheterogeneous.Twostudieshada
largepositiveeffectsizewhenthesestudieswereomitted,homogeneitywasattained.40
MaximumversusLessThanMaximumReward.Therewasonlyonereward
Page144

contingency(rewardsofferedperunitsolved)thatallowedforacomparisonbetweenmaximumandlessthanmaximumreward.Forotherrewardcontingencies,most
studiesinvolvedmaximumrewardacomparisonwithlessthanmaximumrewardwouldthusbeunreliable.Whenrewardswereofferedforeachunitsolved,the
findingsshowedanonreliablenegativeeffectforstudiesofmaximumrewards(K=6,N=345,d+=.03,CI=.25,.18)andasignificantnegativeeffectforstudies
oflessthanmaximumreward(K=14,N=749,d+=.22,CI=.37,.07).Thesetwosetsofeffectsizeswerehomogeneous.Theseresultssuggestthatthe
negativeeffectofpayperunitisassociatedwithparticipantsreceivinglessthanmaximumrewards.
Noanalyseswereconductedondifferencesbetweenmaximumandlessthanmaximumrewardsontheselfreportmeasure.Mostofthecontingencieshadtoofew
studiesthatusedlessthanmaximumrewardforstudiesinvolvingtheofferofrewardforeachproblemsolved,thereweretoofewexperimentsusingmaximumreward
(seeAppendix8B).

NOTES
1. StudiesidentifiedbyCameronandPierce(1994)andbyDeci,Koestner,andRyan(1999)wereincludedinthepresentanalysis.
2. Deci,1971.
3. TwostudiesincludedinCameronandPierces(1994)researchwereomittedinthepresentanalyses.Inonestudy(BoggianoandHertel,1983),thedependentmeasurewasassessed
beforeallparticipantsworkedonthetaskintheotherstudy(BoalandCummings,1981),allparticipants(includingthecontrolgroup)receivedmonetarypayments.Thesestudies
werealsonotincludedinDeci,Koestner,andRyans(1999)analyses.
4. Forexample,seeWiersma,1992.
5. Deci,Koestner,andRyan,1999.
6. InaccordwiththemetaanalysisbyDeci,Koestner,andRyan(1999),ourfreechoicemeasureincludesfreetimeontaskwhentherewardswereremovedand,whentimemeasures
werenotavailable,performanceduringthefreechoiceperiod.AsdidDecietal.,wecombinedperformanceandtimemeasurestomakeupthefreechoiceintrinsicmotivationindex
(wefoundnosignificantdifferencesintheanalyseswhenonlytimemeasureswereanalyzed).
7. RummelandFeinberg(1988)combinedfreetimeandquestionnairemeasuresintheirmetaanalysisoftheeffectsofrewardsonintrinsicmotivation.
8. SeeGlass,McGaw,andSmith,1981.
9. Forexample,seeChung,1995,Horn,1987.
10.Forexample,Smith(1975)offeredrewardstoparticipantsforshowingsignsoflearning.Thiswastheonlystudythatusedthistypeofrewardcontingency.
11.WewouldliketothankKatherineBankoforcodingstudiesanddoingreliabilitychecks.
12.Goldstein,1977.
13.Ibid.,p.30.
14.Themetaanalyticproceduresdescribedandusedinthecurrentanalysisarebased
Page145

onHedgesandOlkin(1985)thesewerethebasicproceduresusedbyCameronandPierce(1994)andbyDeci,Koestner,andRyan(1999).
15.HedgesandBecker(1986)describehowtocalculateeffectsizesfromttests,Fstatistics,andplevelvalues.CameronandPierce(1994,p.404)alsoprovideformulasforthese
calculations.
16.Foradiscussionofthisissue,seeLightandPillemer,1984.
17.InCameronandPierces(1994)metaanalysis,studiesinwhichtherewasinsufficientinformationtocalculateeffectsizeswerenotincludedDeci,Koestner,andRyan(1999)
includedsuchstudiesintheirmetaanalysisandimputedeffectsizesof0.00.
18.Wicker,Brown,Wiehe,andShim,1990DollingerandThelen,1978.
19.SeeHedges,1981.
20.HedgesandOlkin,1985.
21.SeeCameronandPierce,1994Deci,Koestner,andRyan,1999.
22.SeeHunterandSchmidt,1990.
23.HedgesandOlkin,1985.
24.Averageeffectsizeswereobtainedbyweightingeachgindexbythenumberofparticipantsonwhichitwasbased.SeeCooper,1989.
25.ThisprocedurewasusedbyCameronandPierce(1994)andbyDeci,Koestner,andRyan(1999).HedgesandOlkin(1985)discussedtheindependenceassumption.Itisimportantto
pointoutthataseriousstatisticalviolationoccurswhenmorethanoneeffectsizefromanindividualexperimentisenteredintoasinglemetaanalysis.Typically,insuchcases,a
controlgroupiscomparedwithmorethanoneexperimentaltreatmentwithinastudy,severaleffectsizesarecalculated,andeachisenteredintoasinglemetaanalytictest.Themajor
problemisthattheeffectsizesarenotindependent(errorsamongobservationsarecorrelated).Anotherproblemisthataparticularstudywillcontributemoreweighttotheoverall
metaanalyticoutcomethanastudyyieldingonlyoneeffectsize.Othermetaanalysesofrewardandintrinsicmotivationhaveviolatedtheindependenceassumptionbyentering
severalormanyeffectsizesfromonestudy(sometimesoverteneffectsizes)intoasinglemetaanalytictest(e.g.,TangandHall,1995RummelandFeinberg,1988).Theimplication
isthatconclusionsbasedonthesemetaanalysescouldbeincorrect.
26.TheMETAprogramwasdevelopedbySchwarzer(1991)themetaanalyticmethodusedinthisanalysisisbasedontheweightedintegrationmethoddescribedinHedgesandOlkin
(1985).
27.Cohen,1988.
28.TheseresultsareinaccordwithCameronandPierce(1994)andEisenbergerandCameron(1996).
29.InDeci,Koestner,andRyans(1999)supplementalanalysisoflowinteresttasks(p.651),fewerstudieswereincludedandnoreliableeffectswerefoundoneitherthefreechoiceor
theselfreportmeasuresofintrinsicmotivation.
30.Deci,Koestner,andRyan(1999)alsoreportedasignificantnegativeeffectonthefreechoicemeasurefortheeffectsofrewardsonhighinteresttasksbutanonreliableeffectonthe
taskinterestmeasure.Ontheselfreportmeasure,Decietal.(1999)omittedormissedseveralselfreporteffectsizes.
31.TheseresultswerealsoobtainedbyDeci,Koestner,andRyan(1999),whoreportedsimilarsmalltomoderatepositiveeffectsofverbalrewards.
32.Deci,Koestner,andRyan(1999)alsoreportedalargereffectofverbalrewards
Page146

onfreechoiceforadultsbutanonsignificanteffectforchildren.Oureffectforchildrenwasstatisticallysignificanthowever,weincludedmorestudiesthanDecietal.
33.Butler,1987Vallerand,1983.
34.KastandConnor,1988.
35.Deci,Koestner,andRyan,1999.
36.Forhighinteresttasks,inthedatasetfortheeffectsofverbalrewardonselfreportedtaskinterest,thereweresixstudiesthatdidnotprovideenoughinformationtoobtainaneffect
sizeestimate(thesestudieswereassignedaneffectsizeof0.00).Whenthesestudieswereremovedthemeaneffectsizefortaskinterestshowedaslightincrease(K=15,N=981,
d+=0.40,CI=0.27,0.53).
37.Inthestudiesinwhichparticipantswereofferedatangiblerewardforsurpassingascoreorforexceedingtheperformanceofothers,thereweretwopossibletypesofcontrol
comparisons.Insomestudies,thecontrolgroupwastoldtheperformanceobjectivesandgivenperformancefeedback(completecontrol)inothers,thecontrolgroupwasnottolda
performanceobjectiveandnofeedbackwasgiven(partialcontrol).Eisenberger,Pierce,andCameron(1999)examineddifferencesbetweenthesetwotypesofcomparisons(reward
versuspartialcontrolandrewardversuscompletecontrol).Onesmalldifferencewasdetectedonthefreechoicemeasure.Whenrewardswereofferedtoexceedothers,reward
versusapartialcontrolconditionresultedinanonsignificantpositiveeffectthemeaneffectforrewardversusacompletecontrolwassignificantlypositive(noothercomparisons
resultedindifferences).Becausethisdifferencewassmallandbothmeaneffectswereinthesamedirection,inthepresentanalysesweincludedstudieswitheithertypeofcontrol
condition.Ifastudycontainedbothtypesofcontrols(e.g.,Harackiewicz,Manderlink,andSansone,1984),oneeffectsizewascalculatedtocomparetherewardconditiontoboth
controls.
38.TwooftheoutliersproducedpositiveeffectstheonlydifferencesbetweenthesetwostudiesandthemajorityofthestudieswerethatthestudybyTripathiandAgarwal(1988)was
conductedinIndiaandthestudybyBrennanandGlover(1980)wasdesignedtoassesstheeffectsofrewardswhentherewardswereshowntofunctionasreinforcement.Other
outliers(Chung,1995Danner&Lonkey,1981Fabes,Eisenberg,Fultz,andMiller,1988Morgan,1981,Experiment1Okano,1981,Experiment2)hadlargenegativeeffectsbut
revealednocommonfactorthatcouldexplaintheirextremevalues.
39.Intheanalysisofrewardsofferedfordoingthetaskonthetaskinterestmeasure,14studiesweregiveneffectsizesof0.00whenthesestudieswereremovedfromtheanalysis,the
negativeeffectincreasedfrom0.13to0.22(K=24,N=1201,d+=0.22,CI=0.33,0.10).
40.Kruglanski,Riter,Amitai,Margolin,Shabatai,andZaksh,1975,Experiment1WimperisandFarr,1979.
Page147

Appendix8B:StudiesIncludedinourMetaAnalysis
StudiesIncludedintheAnalysisoftheEffectsofRewardsonIntrinsicMotivationforTaskswithLowInitialInterest
Study RewardType RewardExpectancy RewardContingency NENC FreeChoiceEffectSize(g) SelfReportEffectSize(g)
Calder&Staw(1975) T E forfinishingtask 10 10 0.61
Chung(1995) T E fordoingtask 5 5 1.93
T E insufficientinfo 5 5 1.22
Crino&White(1982) V U perunitsloved 10 5 0.05
V U yolked/perunit 10 5 0.32
Daniel&Esser(1980) T E fordoingquickly 16 16 0.28 0.08
Eisenstein(1985) T U forfinishingtask 6 6 0.62
T E forfinishing 16 6 0.22
Freedman&Phillips(1985) T E perunitsolved 24 25 0.24
T E forfinishingtask 26 25 0.53
Griffith(1984)D T E fordoingtask 44 44 0.25
Hamner&Foster(1975) T E fordoingtask 16 15 0.28
T E perunitsloved 19 15 0.52
Hittetal.(1992) T E fordoingtask 30 15 0.57 0.16
Loveland&Olley(1979) T E fordoingtask 6 6 1.20
McLoyd(1979) T E forfinishingtask 18 9 0.61 0.00
Mynattetal.(1978) T E fordoingtask 5 5 1.35
Newman&Layton(1984) T E fordoingtask 20 10 0.41
Overskeid&Svartdal(1996)Exp1 E T fordoingtask 10 10 0.29 0.39
Overskeid&Svartdal(1996)Exp2 T E fordoingtask 64 32 0.15
Phillips&Freedman(1985) T E forfinishingtask 12 12 0.63
T E perunitsolved 12 12 0.10
Smith(1980)D T E fordoingtask 21 27 0.04
V U fordoingtask 22 26 0.17
Wilson(1978)D T E fordoingtask 46 23 0.03 0.12
Page148

StudiesIncludedintheAnalysisoftheEffectsofVerbalRewardsonIntrinsicMotivationforTaskswithHighInitialInterest
Study RewardExpectancy RewardContingency NENC FreeChoiceEffectSize(g) SelfReportEffectSize(g)
Anderson&Rodin(1989) U fordoingtask 10 10 0.20 0.40
Andersonetal.(1976) U fordoingtask 18 19 0.40
Blancketal.(1984)Exp.1 U fordoingtask 70 69 0.56 0.69
Blancketal.(1984)Exp.2 U fordoingtask 12 12 0.73 0.00
Boggiano&Barrett(1985) U fordoingtask 18 18 0.35
Boggianoetal.(I988) U fordoingtask 66 34 0.42
Butler(1987) E fordoingtask 50 50 1.59*
Cohen(1974)D U fordoingtask 52 52 0.07 0.42
Crino&White(1982) U perunitsolved 10 5 0.05
U yolked/perunit 10 5 0.79
Danner&Lonkey(1981) U fordoingtask 30 30 0.10 0.08
Deci(1971)Exp.3 U fordoingtask 12 12 0.82 0.00
Deci(1972a) U fordoingtask 48 48 0.29
Decietal.(1975) noinfo noinfo 32 32 0.02
Dollinger&Thelen(1978) E fordoingwell 12 12 0.07 0.00
Effron(1976)D U fordoingtask 15 13 0.89
Goldstein(1977)D U fordoingtask 32 32 0.77 0.12
Harackiewicz(1979) U fordoingtask 31 31 0.59
Horn(1987)Exp.2 noinfo noinfo 28 28 0.37
Kast&Connor(1988) U fordoingtask 180 60 0.46*
Koestneretal.(1987) U Fordoingtask 35 18 0.51 0.00
Orlick&Mosher(1978) U Fordoingtask 11 12 0.34
Pallaketal.(1982) U Fordoingtask 16 12 0.47
E fordoingtask 14 12 0.32
Pittmanetal.(1980) U fordoingtask 24 12 0.80
Pretty&Seligman(1984)Exp1 U fordoingtask 30 30 0.35 0.46
Ryanetal.(1983) E fordoingtask 32 16 0.53 0.00
Sansone(1986) U fordoingtask 44 11 0.68
Sansone(1989) E fordoingtask 82 41 0.46
Sansoneetal.(1989) U fordoingtask 40 40 0.12
Shanabetal.(1981) U fordoingtask 20 20 0.64 0.43
Smith,W.E.(1975)D U forlearning 20 20 0.04 0.00
Smith,A.T.(1980)D U fordoingtask 21 27 0.24
Tripathi&Agarwal(1985) E fordoingtask 20 20 1.61 0.48
Vallerand&Reid(1984) E fordoingtask 28 28 0.53
Vallerand(1983) E fordoingtask 40 10 1.98*
Zinseretal.(1982) U fordoingtask 64 32 0.08
Page149

StudiesIncludedintheAnalysisofTangibleRewardsonIntrinsicMotivationforTaskswithHighInitialInterest
UNEXPECTEDTANGIBLEREWARDSALLREWARDCONTINGENCIES
Study RewardContingency NE NC FreeChoiceEffectSize(g) SelfReportEffectSize(g)
Eisenstein(1985) forfinishing 10 10 0.46
Greene&Lepper(1974) fordoingwell 26 15 0.14
Harackiewiczetal.(1984)Exp.2 exceedingothers 15 15 0.44a 0.15
Kruglanskietal.(1972) forwinning 36 33 0.65
Lepperetal.(1973) fordoingtask 18 15 0.12
Orlick&Mosher(1978) fordoingwell 12 12 1.28
Pallaketal.(1982) fordoingtask 15 12 0.43
Pretty&Seligman(1984)Exp.1 fordoingtask 30 30 0.06 0.42
Pretty&Seligman(1984)Exp.2 fordoingtask 30 30 0.06 0.38
Smith(1975)D forlearning 20 20 0.08 0.00

EXPECTEDTANGIBLEREWARDSLISTEDBYREWARDCONTINGENCY
Tasknoncontingent
Study NE NC FreeChoiceEffectSize(g) SelfReportEffectSize(g)
Dafoe(1985)D 25 28 0.20 0.73
Deci(1972a) 24 16 0.08
Earn(1982) 40 20 0.28 0.18
Kruglanskietal.(1971) 16 16 0.69
Okano(1981)Exp.2 11 11 0.47 0.27
Pittmanal.(1982)Exp.1 10 10 0.26 0.00
Rossetal.(1976) 12 12 0.44
Swann&Pittman(1977)Exp.1 20 20 0.21
Wimperis&Far(1979) 16 16 0.56

Rewardsofferedfordoingtask
Study NE NC FreeChoiceEffectSize(g) SelfReportEffectSize(g)
Amabileetal.(1986)Exp.1 56 57 0.00 0.00
Amabileetal.(1986)Exp.3 30 30 0.00
Andersonetal.(1976) 36 19 0.53
Amold(1976) 17 36 0.00
Arnold(1985) 13 16 0.04
Boggiano&Ruble(1979) 20 20 0.61
Boggianoetal.(1982) 81 84 0.28
Page150

Boggianoetal.(1985) 26 13 0.79
Brennan&Glover(1980) 20 19 1.00*
Brewer(1980)D 24 24 0.13 0.12
Chung(1995) 5 5 1.61*
Danner&Lonkey(1981) 30 30 1.33* 1.23
DeLoachetal.(1983) 26 26 0.00
Dimitroff(1984)D 108 36 0.27 0.00
Effron(1976)D 12 13 0.19
Fabesetal.(1986) 24 24 0.06 0.14
Fabesetal.(1988) 14 14 1.34* 0.76
Fabesetal.(1989) 15 14 0.73
Feehan&Enzle(1991)Exp1 24 12 0.97
Goldstein(1977)D 16 16 0.99 0.87
Greene&Lepper(1974) 15 15 0.70
Griffith(1984)D 44 44 0.23
Hamner&Foster(1975) 15 15 0.14
Harackiewicz(1979) 31 31 0.38
Hittetal.(1992) 30 15 0.82 0.47
Hyman(1985)D 32 32 0.42
Kamiol&Ross(1977) 17 20 0.08
Lepperetal.(1973) 18 15 0.72
Lepperetal.(1982) 32 32 0.13
Loveland&Olley(1979) 6 6 1.20
Morgan(1981)Exp.1 27 27 0.98 0.31
Morgan(1981)Exp.2 20 20 0.77 0.04
Morgan(1983)Exp.1 40 40 1.94*
40 20 0.54
Morgan(1983)Exp.2 20 20 0.66 0.00
Mynattetal.(1978) 5 5 0.19
Newman&Layton(1984) 20 10 0.37
Ogilvie&Prior(1982) 26 26 0.08
Okano(1981)Exp.1 15 15 0.99 0.45
Okano(1981)Exp.2 10 11 1.31* 0.00
Patrick(1985)D 33 31 0.00 0.00
Perryetal.(1977) 32 32 0.43 0.21
Picek(1976)D 10 10 0.00 0.65
Pittmanetal.(1982)Exp.1 10 10 0.17 0.00
Page151

Pittmanetal.(1982)Exp.2 27 27 0.05
Pretty&Seligman(1984)Exp.1 30 30 0.75 0,05
Pretty&Seligman(1984)Exp.2 30 30 0.13 0.16
Reiss&Sushinski(1975) 16 16 0.83
Ross(1975)Exp.1 40 20 0.01 0.45
Ross(1975)Exp.2 52 14 0.66 0.00
Rossetal.(1976) 12 12 0.64
Ryanetal.(1983) 16 16 0.35 0.00
Sarafino(1984) 85 15 0.41 0.00
ShiffmanKauffman(1990)D 20 20 0.06 0.04
Smith(1980)D 21 27 0.82
Swann&Pittman(1977)Exp.1 20 20 0.78
Swann&Pittman(1977)Exp.2 26 13 1.01
Thompsonetal.(I993) 34 33 0.003 0.14
Tripathi&Agarwal(1988) 20 10 0.34* 0.72
Tripathi(1991) 20 5 0.00 0.00
Weiner&Mander(1978) 30 30 0.34 0.00
Williams(1980) 24 24 0.18 0.00
Wilson(1978)D 46 23 0.06 0.01
Yuen(1984)D 60 60 0.40 0.12

Rewardsofferedfordoingwellordoingagoodjobonthetask
Study RewardDelivery NE NC FreeChoiceEffectSize(g) SelfReportEffectSize(g)
Brewer(1980)D M 48 24 0.08 0.12
Dafoe(1985)D M 26 28 0.00 0.59
Dollinger&Thelen(1978) L 36 12 0.55 0.00
Enzleetal.(1991) M 40 10 0.53
Fabes(1987)Exp.1 M 18 19 0.87
Goldstein(1977)D M 16 32 0.08 0.48
Greene&Lepper(1974) M 15 15 0.57
Hyman(1985)D M 16 16 0.11
Orlick&Mosher(1978) M 14 12 0.53
Pallaketal.(1982) M 15 12 0.17
Ryanetal.(1983) M 32 32 0.46 0.00
Taub&Dollinger(1975) NI 124 124 0.00
Page152

RewardsandIntrinsicMotivation
Study RewardDelivery NE NC FreeChoiceEffectSize(g) SelfReportEffectSize(g)
Calder&Staw(1975) M 10 10 0.46
Eisenstein(1985) M 18 10 0.53
Fabes(1987)1 M 19 19 0.82
Fabes(1987)2 M 14 14 0.45
Freedman&Phillips(1985) M 26 22 0.94
Griffithetal.(1984) M 64 32 0.00
McLoyd(1979) M 18 9 1.04 0.00
Phillips&Freedman(1985) M 12 12 0.74
Stawetal.(1980) M 47 46 0.19
Tripathi&Agarwal(1985) M 20 20 0.41 0.54

Rewardsofferedforeachproblem,puzzle,unit,etc.solved
Study RewardDelivery NE NC FreeChoiceEffectSize(g) SelfReportEffectSize(g)
Arkes(1979) M 32 32 0.16 0.03
Arnold(1985) L 13 16 0.05
Bartelme(1983)D M 35 34 0.04a 0.03
Boggianoetal.(1985) M 26 13 0.10
Brockner&Vasta(1981) L 26 26 0.37 0.58
Carton&Nowicki(1998)1 L 44 22 0.36a
Carton&Nowieki(1998)2 L 40 20 0.20a 0.71
Cohen(1974)D L 52 52 0.18 0.13
Deci(1972a) L 64 32 0.33
Deci(1971)Exp.1 L 12 12 0.54 0.00
Effron(1976)D L 43 28 0.04
Feehan&Enzle(1991)Exp.2 M 30 15 0.31a
Freedman&Phillips(1985) L 23 22 1.12
Goldstein(1980)D L 14 14 0.32 0.68
Hamner&Foster(1975) L 18 15 0.21
Kruglanski(1975)Exp.I M 24 24 1.15*
Lee(1982)D M 40 40 0.36a 0.35
Liberty(1986)Exp1D L 23 23 0.86a 0.34
Liberty(1986)Exp.2D L 44 42 0.22a 0.04
McGraw&McCullers(1979) NI 20 20 0.04
Phillips&Freedman(1985) L 12 12 0.77
Page153

Porac&Meindl(1982) L 20 20 0.78
Shapira(1976) L 30 30 0.41
Sorenson&Maehr(1976) L 20 20 0.54
Vasta&Stirpe(1979) L 4 5 0.16
Weiner&Mander(1978) L 30 30 0.54 0.00
Weiner(I980) M 24 24 0.35 0.00
Wickeretal.(1990) L 29 29 0.46 0.18
Wimperis&Farr(1979) NI 16 16 1.36*

Rewardsofferedformeetingaspecificstandardorsurpassingascore
Study RewardDelivery NE NC FreeChoiceEffectSize(g) SelfReportEffectSize(g)
Adorney(1983)D L 35 36 0.39 0.48
Bartelme(1983)D M 35 34 0.19a 0.03
Boggiano&Ruble(1979) M 20 20 0.17
Dafoe(1985)D M 28 28 0.15 0.59
Eisenberger(1999) M 214 316 0.08 0.31
Eisenbergeretal.(1999)Exp.1 M 110 113 0.10 0.34
Harackiewiczetal.(1987) M 13 25 0.28
Hyman(1985)D M 16 16 0.04
Kruglanskietal.(1975)Exp.2 M 40 40 0.38
Patrick(1985)D M 30 31 0.00 0.00
Pittmanetal.(1977) L 60 20 0.50a 0.20
Smith&Pittman(1978) L 66 33 0.56a 0.00
Tripathi(1991) M 20 5 0.00 0.00

Rewardsofferedformeetingorexceedingothers
Study RewardDelivery NE NC FreeChoiceEffectSize(g) SelfReportEffectSize(g)
Dafoe(1985)D M 25 28 0.00 0.59
Eisenbergeretal.(1999)Exp.1 M 106 106 0.38 0.22
Harackiewicz(1979) M 31 31 0.87
Harackiewicz&Manderlink(1984) M 47 47 0.33
Harackiewiczetal.(1984)Exp.1 M 32 64 0.27 0.12
Harackiewiczetal.(1984)Exp.2 M 15 15 0.43a 0.18
Harackiewiczetal.(1984)Exp.3 M 26 52 0.34a 0.40
Harackiewiczetal.(1987) M 11 29 0.12
Kamiol&Ross(1977) L/M 20 20 0.15
Luyten&Lens(1981) L 10 10 0.90 0.08
Page154

Rosenfieldetal.(1980) L/M 30 27 0.30 0.22


Salancik(1975) M 38 39 0.34 0.01
ShiffmanKauffman(1990)D M 20 20 0.35 0.00
Tripathi&Agarwal(1988) M 20 10 0.87 1.01
Weinberg&Jackson(1979) L 40 40 0.00

StudiesorConditionsWithinStudiesIncludedintheOverallAnalysesofRewardandTangibleRewardThatCouldNotBeClassifiedintoReward
Contingencies
Study RewardContingency NE NC FreeChoiceEffectSize(g) SelfReportEffectSize(g)
Chung(1995) insufficientinfo 5 5 1.02
Daniel&Esser(1980) fordoingquickly 16 16 0.75 0.71
Hom(1987)Exp.I noinformation 26 26 0.12 0.00
Hom(1987)Exp.2 noinformation 28 28 0.37a
Smith(1975)D forshowinglearning 20 20 0.56 0.00
Note:StudiesfollowedbyaDrefertounpublisheddoctoraldissertations.Forrewardtype,T=tangiblerewardV=verbalreward.Forrewardexpectancy,E=
expected(offeredbeforehand)U=unexpected(deliveredbutnotoffered).Rewardcontingencyreferstowhattheparticipantswererewardedfor.Forrewarddelivery,
M=maximumrewardL=lessthanmaximumrewardNI=notenoughinformation.NE=samplesizeforexperimentalgroup(rewardedgroup)NC=samplesizeof
controlgroupg=effectsize.Effectsizeslistedunderfreechoiceintrinsicmotivationarebasedonthedifferencebetweenrewardedandcontrolgroupsontheamountof
timespentonataskoncetherewardwasremovedainthiscolumnsignifieseffectsizesbasedonperformancemeasuresonthetaskduringthefreechoiceperiod(e.g.,
numberofballsplayedinapinballgame,numberoftrialsinitiatedinaLabyrinthgame,numberofwordsfoundinawordsearchgame).Selfreporteffectsizesarebased
onquestionnairemeasuresoftaskliking,satisfaction,enjoymentortaskpreference.Effectsizesmarkedwithanasterisk(*)wereoutliersinthedataset.
Page155

Chapter9
DiscussionandImplicationsofOurMetaanalyticFindings
Ourmetaanalysisoftheempiricalliteratureontheeffectsofrewardsonintrinsicmotivation(presentedinChapter8),showsthatthereisnoinherentnegativeproperty
ofreward.Ourfindingsdonotsupporttheclaimthatrewardsproducesignificantandsubstantialdecreasesinpeoplesintrinsicinterest.1InChapters2and3,an
examinationoftheearlystudiesonthetopicshowedthatnegativeeffectsofrewardwereextremelylimited.After30yearsofresearchandmorethan100experimental
investigationsonthetopic,thefindingscontinuetoshowthatnegativeeffectsofrewardoccurunderahighlycircumscribedsetofconditions.Importantly,ourmeta
analysisindicatesthat,whenproperlyarranged,rewardscanbeusedtoenhanceperformanceandmotivation.Inthischapter,wediscussourmetaanalyticresultsand
elucidatethetheoreticalandpracticalimplicationsofourfindings.

WHATOURMETAANALYTICRESULTSREVEAL

Intermsoftheoveralleffectsofreward,inaccordwithourearlierreviews,2ourmetaanalysisindicatesnoevidenceforageneraldetrimentaleffectofrewardon
measuresofintrinsicmotivation.Thisfindingisimportant.Overtheyears,manystatementshaveappearedinnewspaperarticles,magazines,andacademictexts
condemningtheuseofallrewards.Asaresult,manypractitionershavebeenconcernedthattheincentivesystemsthattheyhavedesignedareactuallymoreharmful
thanbeneficial.Inaddition,manystudentsofpsychologyandeducationaretaughtthat,overall,rewardsareharmfulandshouldbeavoidedinappliedsettings.Our
metaanalyticresultssuggestthattheargu
Page156

Table9.1
ASummaryofOurMetaAnalyticFindingsontheEffectsofRewardsonMeasuresofIntrinsicMotivation
ChangeinIntrinsicMotivation
RewardCondition FreeChoiceBehavior TaskInterest
Allreward
Lowinitialtaskinterest Increase Nochange
Highinitialtaskinterest
Verbalreward Increase Increase
Tangiblereward
Unexpectedreward Nochange Nochange
Expectedreward(offered)
Tasknoncontingent Nochange Nochange
Rewardofferedfordoingtask Decrease Decrease
Rewardofferedfordoingwell Decrease Nochange
Rewardofferedforfinishingtask Nochange Increase
Rewardofferedforeachunitsolved Decrease Increase
Maximumreward Nochange
Lessthanmaximumreward Decrease
Rewardofferedforsurpassingascore Nochange Increase
Rewardofferedforexceedingothers Increase Increase

mentthatrewards,ingeneral,destroypeoplesintrinsicmotivationisnotsupportedbytheempiricaldata.
Ourfindingofnooveralleffectofreward,however,mustbetreatedwithcaution.Inourmetaanalysis,theoverallrewardcategorylackedhomogeneity,indicating
theappropriatenessofamoderatoranalysis.Inotherwords,theoverallrewardcategoryistooinclusiverewardshavedifferenteffectsunderdifferentmoderating
conditions.
InTable9.1,wepresentasummaryofourfindings.Thetableshowswhichrewardconditionsproduceincreasesordecreasesinthetwomainmeasuresof
Page157

intrinsicmotivationfreechoicebehaviorandselfreportedtaskinterest.Onlyresultsfromcategoriesthatwereconsideredhomogeneousarepresented.Other
categoriesweretoobroad,suggestingthatafurthersubdivisionofrewardswasnecessary.
InTable9.1,theeffectsofallrewardsarefirstbrokenintohighandlowinteresttasks.Whenthetasksusedinthestudiesareoflowinitialinterest,rewardsincrease
freechoicebutdonotaffectselfreportedtaskinterest.Thisfindingindicatesthatrewardscanbeusedtoenhancetimeandperformanceontasksthatinitiallyholdlittle
appeal.AsnotedinChapter5,Bandurapointedoutthatfewtasksareinitiallyinterestingtopeople.Butwhenindividualsaregivenappropriatelearningexperiences,
manyactivitiescanbecomeinterestingandimportant.Manyofthetasksusedineducationalsettingsarenotofhighinitialinteresttostudents.Ourresultssuggestthat
rewardproceduresareonewaytocultivateinterestinanactivity.
Onhighinteresttasks,theeffectsofrewarddependontherewardtype,theexpectancy,andcontingency.Table9.1showsthatverbalrewardssignificantlyenhance
bothfreechoiceintrinsicmotivationandselfreportedtaskinterest.Thesefindingssuggestthatwhenpraiseandotherformsofpositivefeedbackaregivenandlater
removed,peoplecontinuetoengageintheactivityandalsoexpresshighlevelsoftaskinterest.
Theeffectsoftangiblerewardsarefoundtodifferbyrewardexpectancy.Whenrewardsaredeliveredunexpectedly(withoutadescriptionofthereward
contingency),thereisnoevidenceofsignificanteffectsoneithermeasureofintrinsicmotivation.Thefindingsfromourmetaanalysisindicatethatonlyexpectedtangible
rewardsproduceadecrementaleffectonintrinsicmotivationandtheeffectofexpectedtangiblerewardsdependsonthecontingency.
InTable9.1,expectedtangiblerewardsarecategorizedaccordingtothedescriptionoftherewardcontingency.Theresultsindicatedecreasesinfreechoice
intrinsicmotivationwhenrewardsaretangibleandofferedsimplyfordoingataskorfordoingwell.Anegativeeffectisalsofoundwhenparticipantsareoffereda
rewardforeachunitorproblemsolved.However,thenegativeeffectofrewardofferedforeachunitsolveddoesnotholdifparticipantsaregiventhemaximum
reward.Thatis,ifpeopleareofferedarewardforeachindividualtaskthattheysolve,theirmotivationisnotalterediftheyaresuccessfulatthetask.Anegativeeffect
occursifindividualsareofferedarewardforeachproblemsolvedbuttheyarealsogivenatimelimitandareunabletoobtainthemaximumreward,atypeofsituation
thatrepresentstheeffectsoffailurefeedbackratherthantheeffectsofreward.Onthetaskinterestmeasure,onlyparticipantswhoareofferedarewardfordoinga
taskshowadecreaseinexpressedtaskinterest.Noothernegativeeffectsontaskinterestwerefound.
Finally,theresultsinTable9.1showthatwhenrewardsareofferedformeetingorsurpassingascore,thereisnosignificanteffectonfreechoicebutasignificant
positiveeffectontaskinterest.Whenrewardsaregivenforexceeding
Page158

theperformancelevelofothers,theresultsshowasignificantincreaseonbothfreechoiceintrinsicmotivationandselfreportedtaskinterest.
Ourmetaanalyticresultssuggestthatthefindingsonrewardsandintrinsicmotivationhavestoodthetestoftime.Ourreview(Chapters2and3)oftheinitialstudies
byDeci(1971,1972b)andbyLepperandhiscolleagues(Lepper,Greene,andNisbett,1973)showedthatobtaininganegativeeffectofrewardrequiredaparticular
combinationofconditions.Anevaluationofover100experimentsonthetopicshowsthatthenegativeeffectcontinuestobehighlycircumscribed.Toproducethe
phenomenon,thefollowingmustoccur:
Theactivitymustbeofhighinitialinterest(usingrewardswithlowinterestactivitiesdoesnotproduceanegativeeffect).
Thedeliveryofrewardsmustbestatedbeforehand(unexpectedrewardsdonotproduceanegativeeffect).
Therewardsmustbematerialortangible(verbalrewardsdonotproduceanegativeeffect).
Therewardcontingencymustbelooseorvague(rewardsgivenforsuccessorformeetingaperformancestandarddonotproduceanegativeeffect).
Therewardmustbedeliveredonlyonceoverasinglerewardsession(therepeateddeliveryofrewardsdoesnotproduceanegativeeffect).
Intrinsicmotivationmustbeindexedbyfreechoicebehaviororselfreportedtaskinterestfollowingthewithdrawalofreward(measuringintrinsicmotivation
duringtherewardedperioddoesnotproduceanegativeeffect).
Intrinsicmotivationmustbeassessedonlyoncefollowingtheremovalorrewards(repeatedassessmentsofintrinsicmotivationmeasuresfollowingtheremovalof
rewardshownonegativeeffects).
Clearly,onecanseethatittakesanunusualcombinationofconditionstoproduceanegativeeffectofrewardinthelaboratory.Ofparticularinterestisthefactthatthe
bulkofthebetweengroupdesignexperimentsonthetopicwereprimarilydesignedtodetectnegativeeffects.Inspiteofthis,manyofthestudiesshowedthatrewards
canbeusedeffectivelytoenhanceormaintainmotivationandinterest.
Overall,inaccordwithourpreviousreviews,3ourupdatedmetaanalysisshowsthatrewardscanbeusedtoreduce,enhance,orhavenoeffectonmeasuresof
intrinsicmotivation.4Rewardsincreasemotivationandperformanceontasksthatareoflowinitialinterest.Onhighinteresttasks,positiveeffectsareobtainedwhen
participantsareverballypraisedfortheirworkandwhentangiblerewardsareofferedandexplicitlytiedtoperformancestandardsandsuccess.Producinganegative
effectofrewardrequiresaparticularcombinationofcircumstancesnegativeeffectsareobtainedprimarilyonhighinteresttaskswhentangiblerewardssignifyfailureor
arelooselytiedtobehavior.
Page159

FINDINGSFROM30YEARSOFRESEARCHONREWARDSANDINTRINSICMOTIVATION
ConditionsnecessarytoproduceanegativeeffectofReward:
Taskisofhighinitialinterest
Rewardistangibleormaterial
Rewardisofferedbeforehand(expected)
Rewardisofferedanddeliveredwithoutregardtosuccessortoaspecificlevelofperformance
Rewardisdeliveredonceoverasinglesession
Intrinsicmotivationisindexedasfreechoicebehaviororselfreportedtaskinterestfollowingthewithdrawalofreward
Intrinsicmotivationismeasuredwithasingleassessmentfollowingtheremovalofreward

ACOMPARISONOFOURMETAANALYTICFINDINGSTOTHOSEOFDECI,KOESTNER,AND
RYAN(1999)
Deciandhiscolleaguesclaimedthatrewardshavesignificantandsubstantialnegativeeffectsonpeoplesintrinsicmotivation.Ouranalysis,ontheotherhand,showsthat
negativeeffectsarelimited.Giventhesediscrepancies,inthissectionwecompareouranalysisandfindingswiththemetaanalysisconductedbyDeciandhisresearch
team.
First,itisimportanttopointoutthatthereareseveralareasofagreementbetweenourcurrentanalysis,Deci,Koestner,andRyans(1999)metaanalysis,andour
previousreviews.Ineachofthesemetaanalyses,verbalrewardsareshowntoincreasemeasuresofintrinsicmotivation.Thefindingsalsoshowthatunexpected
tangiblerewardsdonotaffectmeasuresofintrinsicmotivation.Inaddition,whenrewardsaretangible,offeredbeforehand(expected),andnotrelatedtothetaskat
hand(tasknoncontingent),intrinsicmotivationisunaffected.Clearly,notallrewardsinevitablyresultinalossofintrinsicmotivation.
Deciandhisassociatesclaimthattangiblerewardsaregenerallyharmfulisbasedontheiranalysisofexpectedtangiblerewardcontingencies.5Ourpatternof
findingsforexpectedtangiblerewardcontingenciesdiffersfromtheresultsofDeciandcolleaguesmetaanalysis.InFigures9.1and9.2,wecompareourpresent
analysisofexpectedtangiblerewardcontingencieswiththemetaanalysisofDeci,Koestner,andRyan.Figure9.1showstheeffectsofrewards
Page160

Figure9.1
AComparisonofFindingsUsingaCognitiveEvaluationFrameworkversusaProceduralAnalysisoftheEffectsof
ExpectedTangibleRewardContingenciesonFreeChoiceIntrinsicMotivationforHighInterestTasks


Note:Inthecognitiveevaluationclassification,thecompletioncontingentandperformancecontingentrewardgroupingscontainedstudiesthat
involvedrewardofferedforeachunitsolved.
Page161

onfreechoicebehaviorwhenstudiesareclassifiedaccordingtocognitiveevaluationtheoryversusaproceduralclassificationofthecontingencies.Figure9.2presents
acomparisonofthefindingsforselfreportedtaskinterest.
AnexaminationofFigure9.1indicatespervasivenegativeeffectswhenrewardcontingenciesareorganizedbycognitiveevaluationtheory.Incontrast,aprocedural
classificationshowscircumscribednegativeeffects.Forexample,onfreechoiceintrinsicmotivation,Deciandhisassociatesfoundanegativeeffectforperformance
contingentrewards.Thecategoryofperformancecontingentincludedsomestudiesofrewardsofferedforeachunitsolved,aswellasrewardsofferedfordoingwell,
surpassingascore,andexceedingothers.Figure9.1showsthatthesediverserewardproceduresproducedifferenteffectsonfreechoicehence,itisunwiseto
collapsethemintoasinglecategoryofperformancecontingentreward.Bycombiningthesedistinctrewardprocedures,Decisresearchteamobtainedanoverall
negativeeffectforperformancecontingentreward.Incontrast,whencontingenciesaredefinedbytheproceduresusedinthestudies,Figure9.1showsthatdifferent
proceduresproducedifferenteffectsonfreechoice.
Similarly,onthetaskinterestmeasure,Figure9.2showsthatDeciandhisassociatescollapseddataoverrewardcategories,withsimilarproblemsresulting.In
addition,Decisresearchteamomittedseveralpositiveeffectsthat,whenincluded,resultedinpositivefindingsfortaskinterest.
Insum,themajordifferencebetweenDeci,Koestner,andRyansmetaanalysisandourcurrentresearchconcernstheeffectsofexpectedtangiblerewards.Decis
researchteamusedrewardcontingenciesthatwererelevanttocognitiveevaluationtheorybutcollapseddataoverdistinctrewardprocedures.Thisstrategyresultedin
pervasivenegativeeffectsofexpectedtangiblerewardcontingencies.Whenthecategoriesareorganizedaccordingtotheactualproceduresusedinthestudies,
negativeeffectsarelimitedtoaspecificsetofcircumstances.

MAGNITUDEANDIMPACTOFREWARDEFFECTS
Ineachofthemetaanalysestodate,anegativeeffecthasbeendetectedwhenataskisofhighinitialinterest,whentherewardsaretangibleandofferedbeforehand,
andwhentherewardsaredeliveredwithoutregardtosuccessonthetaskortoanyspecifiedlevelofperformance.Specifically,whentangiblerewardsareoffered
simplyfordoingatask(orfordoingwellonit),somerewardedparticipantsspendlesstimeatthetaskinafreechoiceperiodandreportlessinterestthana
nonrewardedgroup.Giventhatthiseffectwasfoundtobestatisticallysignificantineachofthemetaanalysestodate,itmaybeinformativetoconsiderhowseriousthe
negativeeffectis.
Inallthestudiesinvolvingrewardofferedfordoingatask(orfordoingwellonit),timespentonthetaskduringthefreechoiceperiodwasthemeasureoffree
choiceintrinsicmotivation.Usingthefreetimemeasure,onecouldaskhow
Page162

Figure9.2
AComparisonofFindingsUsingaCognitiveEvaluationFrameworkversusaProceduralAnalysisoftheEffectsof
ExpectedTangibleRewardContingenciesonSelfReportedTaskInterestforHighInterestTasks


Note:Inthecognitiveevaluationclassification,thecompletioncontingentandperformancecontingentrewardgroupingscontainedstudiesthat
involvedrewardofferedforeachunitsolved.
Page163

muchlesstimestudentswouldspendonhighinteresttasks(e.g.,art,music,reading,drama)ifateacherimplementedarewardsystemfordoingthetask(orfordoing
itwell)andthenremovedit.Resultsfromourmetaanalysisindicatethattheaverageeffectsizeforacomparisonbetweenstudentswhoreceivethisrewardprocedure
andnonrewardedindividualsontimeontaskfollowingthewithdrawalofrewardisabout.30.
Intheoriginalexperiments,freetimeontaskwastypicallymeasuredoveraneightminuteperiod.Inordertoconverttheeffectsizeof.30torealtime,oneneeds
toknowthepooledstandarddeviationofrewardedandnonrewardedgroups.BecausemanyresearchersreportonlytorFstatistics,whichcannotbeconvertedto
theoverallpooledstandarddeviation,wewereunabletoprovideanestimateofthisparameter.Instead,weusedawelldesignedstudyconductedbyPrettyand
Seligmanin1984thatprovidesapooledstandarddeviation.6PrettyandSeligmanconductedtwoexperimentswithlargesamplesandreadilyavailablestatistical
information.Bothexperimentscomparedaconditionoftangiblerewardsofferedfordoingahighinteresttask(Somapuzzles)withanonrewardedcontrolgroupon
eightminutesoffreetime.Thepooledstandarddeviationwas2.6minutes.7
Using2.6minutesastheestimateoferror,wewereabletoconvertthenegativeeffectsizefromthemetaanalysisintorealtime.Aneffectsizeof.30wouldmean
thatinaneightminuteperiod,theaverageindividualwhoisofferedatangiblerewardfordoingthetask(ordoingwell)willspendabout47secondslesstimeonthe
taskwhentherewardiswithdrawnthantheaveragenonrewardedindividual.
Giventhisresult,whatwouldhappenifateacherimplementedthisincentiveprocedureinareadingprogram(forchildrenwhoalreadyenjoyreading)andthen
removedit?Accordingtothisestimate,studentswhowereofferedgoldstarsforreadingwouldspendaboutfourminuteslesstimereadinginafortyminutefreechoice
periodthanstudentsnotgiventheincentive.Ifweassumethatstudentswithoutrewardspendabout30minutesreadinginthe40minutefreechoiceperiod,then
rewardedstudentswouldspendabout26minutesreading.8Afourtofiveminutereductioninfreetimereadingcouldbebehaviorallyimportantifcumulatedover
manysuccessiveopportunitiestoread,buttherearenostudiesthathaveaddressedthisissue.Infact,thefewstudiesthathavemaderepeatedmeasuresofintrinsic
motivation(followingtheremovalofreward)havefoundthat,overtime,participantsintrinsicmotivationisnotreduced.
Acautionarynoteisinorder.Ourexampleofreadingandrewarddependsontheuseofastandarddeviationfromasingle,thoughwelldesignedstudy.Italso
dependsontheabilitytoextrapolatefromaneightminuteexperimentalperiodtolongerones.Itispossiblethatthenegativeeffects,suchastheyare,areevidentonly
forashorttimeatthebeginningofthefreechoiceperiod.Thatis,itmaywellbethecasethatifanhouroffreechoiceweregiven,resultsmightlookverydifferent.The
pointisthatthisisahypotheticalexample.Fur
Page164

therevidenceisrequiredtogeneralizethefindingstoexperimentswithlongerfreechoiceperiodsortoeverydaysettingswherechoiceisdistributedoverlongperiods
oftime.
Giventhestateoftheliterature,weconcludethatthenegativeeffectoftangiblerewardsofferedfordoingahighinteresttask(orfordoingitwell)isstatistically
significant,butthesizeoftheeffectdoesnotsuggestastrongimpact.Ofcourse,ourconclusionwithregardtothemagnitudeofthenegativeeffectsofreward
contingenciesappliesequallytopositiveeffects.Thatis,whilethepositiveeffectsarestatisticallysignificant,they,too,aresmall.

THEORETICALIMPLICATIONS
Howdotheresultsofourmetaanalysessquarewiththedifferenttheoriesthathavebeenproposedtoaccountforeffectsofrewardsonintrinsicmotivation?In
Chapter4wedescribedthebasictenetsofcognitiveevaluationtheoryandtheoverjustificationhypothesis(attributiontheory)bothofthesetheoriesviewrewards
primarilyasharmful.InChapter5,weexaminedsociallearningtheoryandbehavioralaccountsofrewardsandintrinsicmotivationfromthesetwoperspectives,
rewardscanbeusedinhelpfulways.Inthissection,weexamineourmetaanalyticresultsinlightofthepredictionsmadebythesedifferenttheoreticalpositions.

CognitiveEvaluationTheory
Accordingtocognitiveevaluationtheory,underlyingintrinsicmotivationaretheinnateneedsforcompetenceandselfdetermination.9Intrinsicmotivationissaidtobe
alteredbychangesinfeelingsofcompetenceandselfdetermination.Theeffectofrewardsonpeoplesintrinsicmotivationdependsonhowtheseeventsimpactan
individualsperceptionsofcompetenceandselfdetermination.Eventsthatincreasepeoplesbeliefsthattheyareskilledinperformingataskandthattheirperformance
isbasedonpersonalchoicewillenhanceintrinsicmotivation.Ontheotherhand,wheneventsdecreaseselfdeterminationandperceivedcompetence,intrinsic
motivationwillbeundermined.Cognitiveevaluationtheoristsassertthatrewardshavetwoaspects:aninformationalfunctionandacontrollingaspect.Theinformational
aspectconveysselfdeterminedcompetencetoanindividual,andthus,intrinsicmotivationisenhanced.Ontheotherhand,rewardsarealsoexperiencedascontrolling,
andhence,selfdeterminationandintrinsicmotivationwillbereduced.
Cognitiveevaluationtheoryaccountsfortheincrementaleffectsofverbalpraiseonintrinsicmotivationmeasures,asfoundinthemetaanalysestodate,byclaiming
aninformationalroleforverbalrewards.ForDeciandhiscolleagues,verbalpraisedeliveredinanonjudgementalwaycontainspositiveinformationalfeedbackthat
servestoincreaseapersonsperceptionsofcom
Page165

petenceandselfdetermination,whichinturnleadstoincreasedintrinsicmotivation.10AsnotedinChapter8,afewstudiesusingverbalpraiseshowedthatfeelingsof
competenceincreased.Themediatingrolesofcompetenceandselfdeterminationarelessclearcut,however.
Cognitiveevaluationtheoristsarguethat,incontrasttoverbalrewards,tangiblerewardselicitastrongperceptionofexternalcontrol.However,becausethe
cognitiveevaluationprocessissaidtotakeplacewhilearewardedactivityisoccurring,unexpectedtangiblerewardsarenotpredictedtoalterpeoplesintrinsic
motivation.Onhighinteresttasks,whentangiblerewardsareofferedtopeoplefordoingatask,completingatask,ormeetingaperformancestandard,cognitive
evaluationtheoristsclaimthattherewardswillbeexperiencedascontrolling,andhence,anindividualssenseofselfdeterminationwillbeundermined.Althoughin
someinstances,contingentrewardsmayconveycompetence,thepredictionisthatthelossofselfdeterminationwilloverridefeelingsofcompetenceandthenetresult
willbeadecreaseinintrinsicmotivationforengagementcontingent,completioncontingent,andperformancecontingentrewards.
Therearetwoproblemswiththisprediction.First,whenexpectedtangiblerewardsareclassifiedaccordingtotheproceduresused,nonegativeeffectsaredetected
onintrinsicmotivationmeasureswhentherewardsarelinkedtosuccess,surpassingascore,orexceedingothers.Aseconddifficultyconcernstheeffectsofrewards
onmeasuresofperceivedselfdetermination.AsdiscussedinChapter8,Eisenberger,Pierce,andCameronevaluatedstudieswithmeasuresofselfdeterminationand
showedthatrewardsofferedfordoing,completing,ormeetingaperformancecriterionoftenincreasedpeoplesperceivedsenseoffreedomandautonomy.11These
findingsareindirectcontrasttothepredictionsmadebycognitiveevaluationtheory.
Ourpatternoffindingscontradictsastrictapplicationofcognitiveevaluationtheory.Cognitiveevaluationtheoryemphasizesthecontrollingaspectofexpected,
contingent,tangiblerewardsinreducingpersonalautonomyorselfdetermination.Thelossofperceivedautonomyleadstoalossinintrinsicmotivation.Ourfindingthat
rewardsgivenforsuccessorrewardsspecificallytiedtolevelofperformance(surpassingascoreorexceedingothers)donotunderminemeasuresofintrinsic
motivationisincompatiblewiththeclaimsofcognitiveevaluationtheory.
Cognitiveevaluationtheorycouldhandlethepatternofresultsifrewardsofferedfordoingataskorfordoingitwellwereshowntobecontrollingwhilerewardstied
toperformancelevelwereshowntoenhanceperceptionsofcompetency.Thisisonepossiblewaytorelatecognitiveevaluationtheorytothecurrentresults.Ofcourse,
theimplicationisthatcognitiveevaluationtheorywouldrequiremodificationinordertohandlepositiveeffectsofrewardstiedtolevelofperformanceandthefactthat
rewardcontingenciescanincreaseperceptionsofselfdetermination.
Page166

TheOverjustificationHypothesis
Thepredictionsofoverjustificationtheoryregardingtheeffectsofrewardsonintrinsicmotivationareessentiallythesameasthoseofcognitiveevaluationtheory.12
Overjustificationtheoryemphasizestheshiftinattributionfrominternaltoexternalsourcesthatrewardsaresaidtoproduce.Fromthisperspective,peoples
perceptionsaboutthecausesoftheirbehaviorinfluencefuturemotivation.Unexpectedrewardisnotpredictedtoalterintrinsicmotivationbecauseanindividualwould
havetoknowabouttherewardinadvanceinordertoattributehisorherperformancetoit.Proponentsofoverjustificationtheoryaccepttheassumptionofcognitive
evaluationtheorythatintrinsicinterestwillnotbeunderminedbyrewardsthatincreaseperceivedcompetencewithoutlesseningperceivedselfdetermination,suchas
verbalapprovaldeliveredinanuncontrollingmanner.13
Decrementsinintrinsicmotivationarepredictedtooccurwhenthereisahighdegreeofinterestinanactivityandwhentheextrinsicrewardsaresalientandoffered
beforehand(expected).Undertheseconditions,peoplesperceptionsshiftfromaccountingfortheirbehaviorasselfinitiatedtoaccountingforitintermsofexternal
rewards.Inaccordwithcognitiveevaluationtheory,expectedtangiblerewardscontingentondoingatask,doingwellonatask,completingatask,andmeetinga
performancestandardareallexpectedtoreducepeoplesintrinsicmotivation.Thus,thepredictionsoftheoverjustificationhypothesisarehighlysimilartothoseof
cognitiveevaluationtheoryandappeartogreatlylimitthedifferencesbetweenthetwoaccounts.
Thepositiveeffectsofverbalpraisefoundinourmetaanalysisandthenegativeeffectsoftangiblerewardsofferedfordoingataskorfordoingitwellarecorrectly
predictedbytheoverjustificationhypothesis.However,theoverjustificationhypothesisalsopredictsnegativeeffectsforotherexpectedtangiblerewardcontingencies,
whichwasnotsupportedbyourmetaanalyticresults.Inaddition,atpresent,theevidencedoesnotindicatethatrewardscausepeopletoattributetheirperformance
toexternalratherthaninternalcauses.

SocialLearningTheory
Basedonaproceduralclassificationofrewardcontingencies,ourmetaanalyticfindingsareinaccordwithasociallearning(socialcognitive)perspective.14The
emphasisinsociallearningisonhowrewardcontingenciesrelatetoperceivedcompetenceorselfefficacy.Rewardcontingenciesthatenhanceperceivedcompetence
orselfefficacyareexpectedtoincreaseinterestandperformanceofanactivity.Socialcognitivetheorypredictsthatrewardstiedtolevelofperformanceenhanceself
efficacytotheextentthatapersonisabletoattaintheperformancestandard(i.e.,succeed).Greaterselfefficacyleadstohigherinterestinataskandmoretimespent
ontheactivity.
Sociallearningtheorydistinguishesbetweennoncompetencycontingentand
Page167

competencycontingentrewards.Noncompetencycontingentrewardsincluderewardsgivenwithoutregardtomasteryofperformance(e.g.,rewardsofferedfor
doinganactivity,doingitwell,completingit,orrepeatingit).ThistypeofrewardcontingencyincludesmanyofthestudiesthatDecisresearchteamclassifiedasusing
taskcontingent,completioncontingent,andperformancecontingentrewards.15Fromasocialcognitiveperspective,thebulkofexperimentsonrewardsandintrinsic
motivationhasinvolvedrewardsofferedforengaginginanactivitywithoutregardtoastandardorcriterionofperformance.Accordingtosocialcognitivetheory,non
competencycontingentrewardsimpartlittleindicationofcompetency,inthattherewardsareonlylooselytiedtobehavior.
Rewardsgivenformastery(i.e.,achievingrelativelychallengingbehavioralstandards)aretermedcompetencycontingentrewards,anditisthistypeofreward
contingencythatissaidtodevelopperceptionsofselfefficacyandtaskinterest.Inouranalyses,rewardsgivenforsurpassingascoreorforexceedingotherscouldbe
consideredasubsetofcompetencycontingentrewards,andpositiveeffectsmaybearesultofincreasedfeelingsofcompetenceandselfefficacy.Thus,thepredictions
ofsocialcognitivetheoryarecompatiblewithourmetaanalyticresults.Atpresent,however,therearenostudiesthathaveexaminedthemediatingroleofselfefficacy.

BehavioralAccounts
AsdiscussedinChapter5,behavioralresearchershavenotprovidedatheoryofrewardandintrinsicmotivation.Rather,theyhavesuggestedanumberofhypotheses
thatcouldaccountforthefindings,basedonaconsiderationofreinforcementprinciples.Intermsofthedifferencebetweenhighandlowinteresttasks,behavioral
researchershavetakenthesamepositionasBanduraandquestionwhyanyonewoulduseextrinsicrewardsforbehaviorthatisalreadyathighstrength(highinterest).
Thefactthatrewardsincreasefreechoicebehavioronlowinteresttasksismostimportantfromabehavioranalyticperspective.Thatis,ourfindingthatrewarding
performanceontasksoflowinitialinterestincreasesmeasuresofintrinsicmotivationsuggeststhatrewardsmaybeusedwithoutdetrimentaleffectsinconditionsunder
whichtheyaremostimportant(i.e.,whenpeoplehavelittleinitialinterestinthetask).Undertheseconditions,theextrinsicrewardsdonotconflictwiththenatural
consequencesoftheactivity.
Intermsofhighinteresttasks,ausefuldistinctionismadebyCartoninhisanalysisoftangibleandverbalrewards.16Cartonnotedthatinstudiesofverbalreward,
therewardsweregivenimmediately,frequently,andwithoutanyindicationofwithdrawal.Theseproceduresaretypicalofacontingencyofreinforcementandwouldbe
expectedtoincreaseperformanceandinterest.Ourfindingsthatverbalrewardsincreasefreechoiceandtaskinterestareinaccordwiththishypothesis.
Page168

Incontrast,Cartonsuggestedthatinstudiesofrewardsandintrinsicmotivation,tangiblerewardsareusuallydeliveredonlyonce,withsomedelaybetweenthe
performanceandthereward.Inaddition,insuchstudies,participantsarepresentedwithsignalsthatindicatethattherewardsarenotavailableforthefreechoice
session.Generally,then,tangiblerewardsareexpectedtohaveanegativeeffectbecausetherewardsareunlikelytobefunctioningasreinforcement.Ourmetaanalytic
resultsdoindicateageneralnegativeeffectfortangiblerewardthatcouldsupportCartonshypothesis.Thisaccountisgivenmoresupportbythesinglesubjectstudies
onthetopic,inwhichtangiblerewardsweredeliveredrepeatedlyandimmediately.Nodetrimentaleffectofrewardwasdetectedinthesestudies.Instead,when
rewardswerewithdrawnandrepeatedassessmentsofperformanceweretaken,therewasnoindicationofachangeinintrinsicmotivation.
Resultsfromourmetaanalysisindicatethattangiblerewardsproducenegativeeffectswhentheyareofferedfordoingataskorfordoingitwell.Onewayto
understandthisnegativeeffectfromabehavioralviewpointistonotethattherewardsusedinmostofthesestudieswerenotshowntofunctionasreinforcement.When
nonreinforcingrewardsareofferedforsimplydoingatask(withnoregardtoqualityorlevelofperformance),onepossibilityisthattheyactlikedemands.Participants
mayviewthesenonreinforcingrewardsasanattempttoforcethemtobehaveinagivenwayandmayreactagainsttheexperimenterscontrolbyshowingdeliberate
noncompliance.Deliberatenoncompliancemayalsobeareactiontothecoerciveofferofrewardbytheexperimenter.Thecoerciveaspectofthesituationwillbe
moreprominentwhenrewardsaregivenwithoutregardtolevelofperformance.Undertheseconditions,participantsmayreactbyreducingtheirtimeandperformance
onthetaskinthefreechoicephase.
Inourmetaanalysis,whentangiblerewardswereofferedforeachproblemorunitsolved,anegativeeffectoccurredwhenparticipantsreceivedlessthanmaximum
reward.Inthiscontext,lessthanmaximumrewardindicatesfailure.Failureisaconditionedpunisherthattemporarilydecreasestaskperformanceduringthefree
choiceperiod.Also,becausefailureispairedwiththenaturalrewardsofperformingtheactivity,thevalueofthenaturalconsequencescoulddecrease.Repeated
encounterswithfailurecould,inthisway,underminemotivation.
Whenrewardsareofferedforsuccess(formeetingorsurpassingaperformancestandardorgroupnorm),therewardprocedurecanbecalledsuccesscontingent.
Rewardsprogrammedforsuccesswouldnotbeexpectedtohaveadetrimentaleffectonfreechoiceintrinsicmotivation.Thisisbecausetherewardsfunctionas
reinforcement.Ourmetaanalyticfindingssupportthisinterpretationinthreeways:(1)whenparticipantsweresuccessfulatsolvingproblemsandreceivedmaximal
reward,therewasnodecreaseinintrinsicmotivation,(2)rewardsofferedforsurpassingaspecifiedscoredidnotdecrease
Page169

intrinsicmotivation,and(3)rewardsofferedforexceedingothersledtoincreasesinmeasuresofintrinsicmotivation.

SUMMARYOFTHEORETICALIMPLICATIONS
Intermsoftheoreticalconsiderations,theresultsfromourmetaanalysisdonotsupportcognitiveevaluationtheoryortheoverjustificationhypothesis.Thesetwo
theoriesfocusprimarilyonnegativeeffectsofrewardandgivelittleattentiontorewardcontingenciesthatareshowntoenhanceperformance,interest,ormotivation.
Ourfindingsarewellexplainedbytheoriesthatpredictthattheeffectsofrewardonintrinsicmotivationdependonaclearspecificationoftherewardcontingency.
Socialcognitivetheorypredictsthatrewardstiedtolevelofperformanceenhanceselfefficacytotheextentthatthepersonisabletoattaintheperformancestandard
(i.e.,succeed).Greaterselfefficacyleadstohigherinterestinataskandtomoretimespentontheactivity.Inouranalyses,thepositiveeffectsofrewardsgivenfor
surpassingascoreorexceedingothersareinaccordwiththisaccount.Theresultsalsosupportbehavioralexplanationsinthatrewardsthatarenonreinforcing,or
coercive,decreaseintrinsicmotivation,whilerewardsthatarereinforcing,orbasedonsuccess,maintainorenhanceinterestandperformance.

PRACTICALIMPLICATIONS
Ourmetaanalyticfindingsindicatethatextrinsicrewardsdonothavepervasivenegativeeffectsonpeoplesintrinsicmotivation.Ontasksoflowinitialinterest,extrinsic
rewardscanbeusedtoincreasemotivationandperformance.Onhighinteresttasks,verbalpraiseandtangiblerewardslinkedtosuccessortoobtainingorexceeding
aspecificperformancestandardcanenhancepeoplesinterestwithoutdisruptingperformanceoftheactivityinafreechoicesetting.Theserewardcontingenciescan
beviewedasasubsetofthemanypossiblearrangementsoftheuseofrewardineverydaylife.Rewardscanbearrangedtoprogressivelyshapeperformance,17
cultivateinitialinterestinanactivity,18andmaintainorenhanceeffortandpersistenceatatask.19
Anegativeeffectoccurswhenataskisofhighinitialinterestandtherewardsaretangible,offeredbeforehand,anddeliveredwithoutregardtosuccessonthetask
oranyspecifiedlevelofperformance.Underthiscombinationofconditions,experimentalfindingsindicatethatsomerewardedparticipantsspendlesstimeonthetask
(inafreechoiceperiodwithoutreward)andreportlesstaskenjoymentthananonrewardedgroup.Thiseffectisstatisticallysignificantinallthemetaanalysestodate
onthistopic,buttheeffectissmall.Inaddition,thenegativeeffectappearstobeatemporaryphenomenonnolongtermdecrementaleffectsofrewardaredetected
whenrewardsarepresentedrepeatedlyandintrinsicmotivationmeasuresareassessedovertime.
Page170

Howrelevanttoeverydaylifeisthedecrementaleffectofrewardfoundintheexperimentalstudies?Onewouldbehardpressedtofindanyreallifecircumstances
thatdirectlyparallelthelaboratoryconditionsnecessarytoproduceanegativeeffect.Itisdifficulttoimagineanemployerormanagerwhowouldshowerincentiveson
employeesforatemporaryperiodregardlessofhowtheybehave,withdrawthesystem,andthenexpectthemtoworkhardafterhours.Ineducationalenvironments,
studentsarerarelyrewardedsimplyfordoingactivitiesthattheyalreadyenjoy.Rewardsaremostoftenusedtoshapesuccessfulperformanceandrecognizestudent
accomplishment.Inaddition,therewardsareusuallypresentedoveraperiodoftime,andasproficiencyinataskincreases,theyaregraduallyfadedout.Incontrast,in
thetypicalbetweengroupsrewardandintrinsicmotivationexperiment,theprocedureinvolvesasinglerewarddeliveryfollowedbyasingleassessmentofintrinsic
motivationwithoutreward.Thepointisthattheproceduresusedintheexperimentalstudiesthatobtainednegativeeffectsofrewardonintrinsicmotivationarenot
characteristicoftheuseofrewardsinappliedsettings.
Thecomparabilityoftherewardcontingenciesusedinthelaboratoryexperimentstothoseusedineverydaysettingsisfrequentlyignoredindiscussionsthat
emphasizethedetrimentaleffectsofrewardonintrinsictaskinterest.Instead,thosewhoarguethatnegativeeffectsarepervasivetendtocondemntheuseofrewardsin
practicalsettings.Thus,itisimportanttoconsideranycircumstancesthatcouldremotelyresemblethelaboratoryconditionsshowntoproduceanegativeeffectof
reward.
Inworksettings,therearesomecircumstancesinwhichindividualsarerewardedirrespectiveofsuccessorperformancelevel.Forexample,becauseof
compensationandpromotionsystemsthatareinsensitivetoperformance(e.g.,wagessetbyjobclassification),someemployeescanvarytheirperformance
substantiallyyetwithlittleeffectontangiblereward.Insuchasituation,employeesmayhaveconsiderablelatitudeinhowwellorpoorlytheyperformtheirjobs,without
anychangeinpayorfringebenefits.Ifanemployeehashighintrinsicinterestinthejobandreceivesarewardindependentofperformance,interestinthejobmay
decrease,therebyproducingpoorerperformancethaniftherewardwerecontingentonperformance.
Ineducationalsettings,ifastudenthashighinterestinaparticularsubjectmatter,sayahistorycourse,andreceivesthesamegraderegardlessofperformance,
interestinhistorycoulddeteriorate.Thestudentmaybelesslikelytospendtimereadinghistorybothduringthecourseandfollowingitsconclusion.
Inordertoavoidpoorperformanceandreducedtaskinterestinbusinessandeducationalsettings,employers,teachers,andadministratorsneedtoconsiderthe
basisonwhichtheyallocaterewards,recognition,andadvancement.Ourmetaanalysissuggeststhatwhenpeoplearepraisedorgivenpositivefeedbackfortheir
work,taskinterestandperformanceincrease.Whentangiblerewards
Page171

(e.g.,money)aremadecontingentonsuccessoronmeetingaperformancestandard,motivationisenhanced.
Thefindingsfromourmetaanalysisandtheresultsofsinglesubject,repeatedmeasuresexperimentsareinaccordwitharetrospectivesurveyontheeffectsof
extrinsicrewardofferedtochildrenforreading.FloraandFloraexaminedtheeffectsofparentalpayforreadingaswellasparticipationintheBookItreading
programsponsoredbythePizzaHutchain.20TheBookItprograminvolvedover22millionchildreninAustralia,Canada,andtheUnitedStates.Thechildrenset
readinggoalsandwererewardedwithcouponsredeemableforpizzasiftheymettheirobjectives.FloraandFlorasfindingsindicatethatneitheroffersofmoneynor
pizzasnegativelyaffectedreadingorintrinsicmotivationforreadingineverydaylife.Theseresultsindicatethatthefindingsfromourmetaanalysishaveexternalvalidity.
Thatis,inbothlaboratorysituationsandeverydaysettings,rewardsofferedcontingentonmeetingaspecificlevelofperformancedonotnegativelyaffectpeoples
intrinsicmotivation.
Insomeofthestudiesofrewardslinkedtoperformancestandards,individualsweretoldinadvancethatrewarddependedonsurpassingtheperformanceofsome
percentageofotherswhohadpreviouslyperformedthetask.Inbusinessandeducation,thistypeofrewardcontingencyisquitecommon.Rewardsarefrequently
givenforsurpassingthepastorcurrentperformanceofothers.Forexample,thecriteriathatmanyteachersusetogradestudentstypicallydependsonhowwellother
studentshavedoneinthepast.
Inbusiness,bonusesorpromotionsareoftenpaidtoindividualsorgroupsforoutperformingothers.Inmanycases,thecomparisonofperformancewithothersis
implicitorganizationstypicallyattempttosetcommissionsonsaleshighenoughtomotivateincreasedperformancebutnotsohighastoallowincometoexceedthat
paidtohigherlevelemployees.Becausetheexperimentalstudiesfoundthatrewardofferedforexceedingothersincreasedparticipantsexpressedinterestandtime
spentonanactivity,sucharewardcontingencymightbeusedbyemployerswithoutthedetrimentaleffectsthataccompanyrewardsofferedsimplyfordoingatask.
Unfortunately,atthispointintime,welackinformationconcerningthedurabilityoftheeffectofthistypeofrewardonintrinsictaskinterest.
Inmanypracticalsettings,therewardcontingenciesarenotclearlyrelatedtothoseexaminedintheintrinsicintereststudiesconductedinthelaboratory.An
employeemayhavetomeetaratherlowperformancestandard,inordertoavoidbeingfired.Systematicresearchisneededconcerningtheeffectofstringencyofthe
standardonintrinsictaskinterest.Inotherjoborschoolsettings,themagnitudeofrewardisoftenacontinuousfunctionorstepfunctionofperformancethegreaterthe
performance,thegreateristhereward.Thisdiffersfromthetypicallaboratoryuseofanallornoneperformancestandardtostudyreward.Slidingscalesofreward,
whicharefrequentlyusedineverydaylife,may
Page172

havestrongbeneficialeffectsonintrinsicinterestbecausetheyallowindividualstoexperienceincreasinglevelsofrewardwhilefeelingincreasinglycompetent.

CONCLUSION
Inthischapter,wehaveshownthatrewardsarenotinherentlyeitherbadorgoodforpeople.Rewardscanhavenegativeeffects,butsucheffectsarecircumscribed,
limited,andeasilyprevented.Acarefularrangementofrewardsineducationalsettingsandtheworkenvironmentcanenhanceemployeesinterestandperformance.
Thisoccurswhenrewardsarecloselytiedtotheattainmentofperformancestandards.
Ourmetaanalyticfindingscanbeplacedinthebroadercontextofrewardsgivenformasteryorcompetency.21Undertheseconditions,rewardsfosterperformance
accomplishments.Whentasksarechallengingandgradedinstepstowardthemasteryofskills,rewardsgivenateachstepofaccomplishmentcaninstillinterestand
highpersonalstandardsforperformance.22Inaddition,whenrewardsaremadecontingentoneffort,peopleworkharderatthetasktowhichtheyareassignedand
showincreasedperformanceonotheractivities.Rewardinghigheffortbuildsageneralizedlearnedindustriousness.23Theimplicationsarethatrewardsbasedon
challenge,mastery,andhigheffortbuildskills,interest,performancestandards,andpersistence.
Educators,businessmangers,andadministratorsareinterested,notonlyinhighlevelsofinterest,performance,andmotivation,butalsoinpromotingcreativeand
novelachievementsbytheirstudentsandemployees.Creativeperformanceinvolvesthegenerationofnovelbehaviorthatmeetsastandardofqualityorutility.Reward
forworkinghardatbeingcreativemayincreasecreativity.Eisenbergerhasshownthatwhenrewardsaregivenforcreativethinkingandperformance,peopleshow
generalizedcreativityonothertasks.Whenrewardsaregivenformenialactivitiesinvolvinglittleeffort,however,creativitywillbelowonthetaskandgeneralizetolow
creativityinothersituations.24Thepointisthatyougetwhatyoureinforce.Creativitycanbeinstilledbyrewardproceduresthatspecifycreativeperformance.Onthe
otherhand,peoplecanlearntobehaveinagenerallyuncreativemanneriflowcreativityandeffortarereinforced.
Takentogether,thefindingsfromourmetaanalysisandotherexperimentalliteraturesuggestthatrewardsincreaseperformanceandinterestwhenrewardsare:
madecontingentonsuccessoraregivenformeetingclearstandardsofperformance
madecontingentonchallengingactivities
givenformasteringeachcomponentofacomplexskill
deliveredforhigheffortandcreativity
Page173

Ineachofthesesituations,itisimportanttoarrangetherewardsothatitcloselyfollowsthedesignatedbehavior.Rewardsthataredelayedallowforother
performancestooccurduringtheinterval,andconsequently,thoseinvolvedmaynotlearntheactualbasisofreward.Inaddition,researchindicatesthatrewardplans
aremosteffectiveinorganizationsinwhichtheemployeesareinvolvedinthedesignandimplementationoftheplan.25Insum,whenstudentsandemployeesparticipate
inperformance/rewardplansandrewardsareimmediateandbasedonpersonalcompetence,effort,andaccomplishment,individualswillenjoytheirworkanddoit
well.
Giventhesefindings,itissurprisingthatsomeeducators,psychologists,andbusinessspokespersonscontinuetoarguethatrewardshavegeneralizednegativeeffects
onhumanbehavior.Theseadvocateswouldhaveusremoveallformsofincentivesystemsinbusiness,education,andothermajorinstitutionsofoursociety.Ifthis
couldnotbeaccomplished,theywouldadvocatethatrewardsshouldbelooselytiedtobehavior,sincethiswouldhavetheleastnegativeeffectonthesupposedinnate
humandriveforcompetenceandselfdetermination.Inanswertothisposition,wenotetheunmitigatedfailureofcollectivefarmsintheformerSovietUnion,asituation
inwhichrewardsforproductivitywereonlylooselytiedtoactualperformance.Ourresearchsuggeststhatwhenrewardsarelinkedtospecificstandardsof
performance,peoplewillbemorecontentedandproductiveemployees.Why,then,wouldmanagersimplementlooseincentivesystemsthatfailtorewardsuccess,
effort,accomplishment,andqualityofperformance?Theanswerfromourresearchisthatthiswouldbeabackwardstep.Instead,anevaluationoftheexperimental
findingsindicatesthatinterestandperformancewillincreasewhenbusinessesandschoolssucceedintyingrewardstohighlevelpersonalaccomplishments.

NOTES
1. Deci,Koestner,andRyan(2001)madethisclaimandsuggestedthattheuseofrewardsshouldbeofparticularconcerntothosewhoworkineducationalsettings.
2. CameronandPierce,1994EisenbergerandCameron,1996.
3. Ibid.
4. Asindicated,thisanalysisisbasedonCameron,Banko,andPierce(2001).
5. Deci,Koestner,andRyan,1999.
6. PrettyandSeligman,1984.
7. Deci(1971)alsousedtheSomapuzzle,assessedthefreetimemeasureoveraneightminuteperiod,andthepooledstandarddeviationwas2.4minutes.
8. BasedonDeci,Koestner,andRyans(1999)analysisofengagementcontingentreward,rewardedchildrenwouldspendabout25minutesreadingina40minuteperiod.
9. ThemostrecentstatementsofthetenetsofcognitiveevaluationtheorycanbefoundinDeci,Koestner,andRyan(1999)andinRyanandDeci(2000).
10.Deci,Koestner,andRyan,1999.
11.Eisenberger,Pierce,andCameron,1999.
12.Lepper,Greene,andNisbett,1973.
Page174

13.LepperandGilovich,1981LepperandHenderlong,2000.
14.Bandura,1986,1997.
15.Deci,Koestner,andRyan,1999.
16.Carton,1996.
17.Schunk,1983,1984.
18.Bandura,1986.
19.Eisenberger,1992.
20.FloraandFlora,1999.
21.AlsoseeBandura,1986.
22.SeeSchunk,1983.
23.Eisenberger,1992.
24.EisenbergerandArmeli,1997EisenbergerandCameron,1996.
25.McAdamsandHawk,1992.
Page175

PARTV
REWARDSANDINTRINSICMOTIVATION:A
SOCIOHISTORICALPERSPECTIVE
Page176

Thispageintentionallyleftblank.
Page177

Chapter10
ASociohistoricalAnalysisoftheLiteratureonRewardsandIntrinsicMotivation
Throughouthumanhistory,manywrongheadedideashavebeenpopularizedascherishedbeliefs.Onceprevalent,theseideasarepropagatedandmaintainedfora
longtime,eveninthefaceofdisconfirmingevidence.Therewasatimewheneveryonebelievedthattheearthwasflat,thatthesunrevolvedaroundtheearth,andthat
humanflightwasimpossible.Todaythesebeliefsseemsilly.Butanenormousamountofeffortandevidencewasrequiredandagreatdealofresistancewas
encounteredbeforesuchideaswerereplaced.Theviewthatrewards(oranyexternalcontrol)disruptordestroyintrinsicmotivationisalsoamisguidednotion.
AsshowninChapters2and3,theearlystudiesdidnotproduceevidenceofageneralizednegativeeffectofrewards.InPartIVofthisbook,weshowedthatwhen
allstudiesareconsidered,thenegativeeffectremainshighlycircumscribedandminimal.Nonetheless,theviewthatrewardsreducepeoplesintrinsicmotivation
continuestobeheldinhighregard.Whyhavemillionsofdollarsandthousandsofhumanhoursbeenspentonthistopic?Whydoresearcherscontinuetoadvocate
againsttheuseofrewards,evenwhentheevidencedoesnotsupportthisposition?Andwhydoestheviewthatrewardsareharmfulremainpopular?Onewayto
understandwhythisviewbecamepopularistoexaminetheliteratureinthecontextinwhichitwasinstigated.
Ifrewardsarenotharmful,whatsociohistoricaltrendscontributedtothenegativeeffectviewandmadeitacceptable?Canwetracecertainpatternsofthoughtthat
culminatedintheearlyresearchonrewardsandintrinsicmotivation?Theseareimportantquestionstheanswershelpustoseethatthebadreputationofrewardsarose
inahistoricalcontext.Inthischapter,weanalyze
Page178

therewardsandintrinsicmotivationcontroversyfromasociohistoricalperspective.Ouranalysisindicatesthattheliteratureonrewardsandintrinsicmotivationisbest
understoodaspartofahistoricaldebatebetweenbehaviorismandotherviewsinpsychology.
Thechapterbeginswithadescriptionoftheriseofbehavioralviewsinpsychology,thedisputeovermentalismandintrospection,argumentsaboutfreedomand
control,andthereemergenceofmentalismascognitivepsychology.Eachoftheseelementscontributedtoanoppositiontowardbehaviorismandtheviewthattheuse
ofreinforcementandrewardsisharmful.

THERISEOFBEHAVIORISMANDTHEDECLINEOFMENTALISM
Themostcommonlyaccepteddateoftheoriginofpsychologyasascienceis1879,whenWilhelmWundtfoundedthefirstpsychologicallaboratory.Inthebeginning,
psychologywasthestudyofthemind,orpsyche(i.e.,mentalevents,structures,andprocesses).Theaimofapsychologyofmindwastodescribeandexplain
consciousness.Inordertounderstandconsciousness,psychologistsusedintrospection(lookinginsideoneself)andselfobservationasbasicmethods.Thesemethods
involvedanalyzingthecontentofconsciousnessintocomponentpartssuchassensations,images,feelings,andthoughts.Animportantaspectofapsychologyofmindis
thatthehumanpsycheisviewedasthecenterofactionandcreativethought.Asconsciousbeings,peoplearesaidtocreaterepresentationsoftheworldandtoactin
accordwiththeirperceptions,memories,andplans.Thisearlyviewisreflectedtodayinpsychologiesthatviewpeopleaswillful,intentional,andselfdetermined.From
thisperspective,externalinfluences(e.g.,reward)interferewithpeoplesbasiccapacityforselfdirectionand,indoingso,disruptanindividualsintrinsicmotivation.
Duringthe1890sinNorthAmerica,psychologygraduallymovedawayfromapsychologyofthemindandindividualagencytowardabehavioristview.In1892,the
AmericanPsychologicalAssociation(APA)wasfoundedandprofessionalismrequiredthatpsychologistsdefinetheirroleinsociety.Basedonapragmaticviewpoint,
reform,efficiency,andprogresswerethemajorvaluesofAmericansociety.Theperiodhasbeencalledtheageofthenewstheneweducation,thenewethics,the
newwoman,andthenewpsychology.1Psychologistsbegantalkingaboutascienceofbehaviorthatcouldbepractical,objectiveandverifiablethroughnaturalscience
methods(directobservationandexperimentation).
Thenewbehavioralpsychologystoodincontrasttoapsychologyoftheindividualandthemind.Behavioristsarguedthatstimulationandfeedbackfromthe
environment(bothphysicalandsocial)areimportantdeterminantsofhumanbehavior.Inthebehavioristview,creativethinking(oftenattributedtothemind)arisesfrom
theinteractionbetweenhumanbehaviorandthesocialandphysicalenvironments,ratherthanfromanautonomouscreativemind.Fromabehavioral
Page179

perspective,externalrewardscanbearrangedtoenhancecreativity,feelingsofselfdetermination,andpersonalcompetence.
Somepsychologistssawthenewpsychologyofbehaviorasquantitativeandexperimentalandclaimedwidereachingpracticalapplicationsineducation,medicine,
thefinearts,politicaleconomy,andindeed,inthewholeconductoflife.2JohnDewey,aphilosopherofeducationandpresidentoftheAPAin1899,saweducational
reformasacentralconcernforthenewpsychology.OneimplicationofDeweyspragmaticphilosophywasthatpsychologywouldhavetounderstandthecontrolof
humanbehaviorbystimulationandfeedbackfromtheenvironment(e.g.,praisebyteachers).3Thismovementtowardanobjectiveandpracticalscienceofbehavior
eventuallypredominatedoverthestudyofthemindandintrospection.Severalprominentresearcherscontributedtotheriseofbehaviorismandtotherejectionof
mentalisticpsychology.

IvanPetrovichPavlov(18491936)
OnepersonwhoseideaswerehighlyinfluentialinbehavioralpsychologywasIvanPavlov,aRussianphysiologistwhowaswellknownatthetimeforhisworkon
digestion.Atthetime,manypeoplethoughtthatallanimals,withtheexceptionofhumans,werecomplexbiologicalmachines.Theideawasthataspecificstimulus
wouldevokeaparticularresponseinmuchthesamewaythatturningakeystartsanengine.Inotherwords,animalswouldreacttotheenvironmentinasimplecause
andeffectmanner.Humans,ontheotherhand,wereseenasdifferentfromotheranimalsinthattheiractionswerepurposive.Humansweresaidtoanticipatefuture
events.
Inthecourseofhisexperimentsondigestivesecretionsindogs,Pavlovnoticedthatwhentheexperimenterappeared,thedogswouldsalivate,seeminglyin
anticipationoffood.Theappearanceoftheexperimenterhadnotoriginallyelicitedsalivation.Theeffectwasobservedonlyaftertheexperimenterspresencehadbeen
pairedwiththepresentationoffood.Pavlovrecognizedthatsucharesultchallengedtheconventionalwisdomthatonlyhumansdisplayedforethought.
Pavlovhadmadeanimportantobservationintermsofthestudyofbehavior.Hereasonedthatanticipatoryreflexeswerelearnedorconditioned.Further,Pavlov
concludedthattheseconditionedreflexeswereanessentialpartofthebehavioroforganisms.Althoughsomebehaviorsweredescribedasinnatereflexes,otheractions
werebasedonconditioningthatoccurredduringtheanimalslife.
Takingadvantageofthisobservation,Pavlovsetouttoinvestigatethephenomenonexperimentally.Thequestionwashowtostudyconditionedreflexes
systematically.Pavlovsanswertothisquestionrepresentsamajoradvanceintheexperimentalscienceofbehavior.Ifdogsreliablysalivatedatthesightofalabcoat,
Pavlovreasoned,anyarbitrarystimulusthatprecededfoodmightalsobeconditionedtoevokesalivation.Pavlovreplacedtheexperimenterslabcoatwithastimulus
thathecouldsystematicallymanipulateandreliablycontrol.In
Page180

someexperiments,ametronome(adeviceusedtokeepthebeatwhileplayingthepiano)waspresentedtothedogjustbeforeitwasfed.Thisprocedureresultedin
thedogeventuallysalivatingtothesoundofthemetronome.
Overall,Pavlovsresearchshowsthatpartofhumanandanimalbehavioriscausedbyexternalevents.Intermsofbehaviorism,Pavlovshowedthatexternalstimuli
cancontrolspecificresponses.Moreimportant,heshowedthatthecontrolbythesestimulicanbetransferredtoother(arbitrary)partsoftheenvironment(e.g.,a
whitelabcoatorametronome).Oneimplicationofthisresearchisthatmuchhumanbehaviorthatappearstobepurposive,intentional,andconsciousis,infact,dueto
conditioning.Pavlovsworkrepresentedamajorbreakfromtheviewthathumanbehaviorarosefrommentalprocessesandeventsinthemind.

JohnBroadusWatson(18781958)
PavlovsresearchbecameprominentinNorthAmerica,andtheconditionedreflexwasincorporatedintoamoregeneraltheoryofbehaviorbythefamousbehaviorist
JohnB.Watson.BuildingontheworkofPavlov,Watsonarguedthattherewasnoneedtomakeupunobservablementalassociationstoaccountforhumanand
animalbehavior.Heproposedthatinstead,psychologyshouldbeasciencebasedonobservablebehavior.
InWatsonsbehaviorism,thoughtsandfeelingshadnoplaceinascientificaccountandresearcherswereencouragedtodirecttheirattentiontoobservable
movementsandmeasuredneuralactivity.Althoughthiswasanextremeposition,Watsonsucceededindirectingtheattentionofpsychologiststothestudyofbehavior
anditsenvironmentaldeterminants.
In1913,WatsonpublishedhismostinfluentialworkinPsychologicalReview,inanarticletitled,PsychologyastheBehavioristViewsIt.4Thearticleoutlined
Watsonsviewsonbehaviorismandarguedthatobjectivitywastheonlywaytobuildascienceofpsychology.Inparticular,Watsonrejectedpeoplesreportsoftheir
thoughtsandfeelingsasscientificdata.Selfreportswereseenassubjectiveandunreliable.Inferringmentalstatesfromselfreportswasnot,accordingtoWatson,a
scientificenterprise.Finally,Watsonnotedthatthepsychologyofmindhadlittlepracticalvalueforbehaviorcontrolorpublicaffairs.
Watsonsbehaviorismdidnotsitwellwiththosepsychologistswhoclaimedtheexistenceofamentalworldwithintheskin.Forhisopponents,behaviorismseemed
todehumanizepeople,makingthemintopuppetsreactingtopullsontheirstrings.BasedonthischaracterizationofWatsonsposition,manypsychologistsand
educatorsrejectedtheevidenceofconditioningandclaimedthathumanswerenotsubjecttoconditioningprinciples.Thecontrolofhumanbehavior,evenbypositive
externalmeans,becameviewedasanencroachmentonhumannature.Fromahumanistic(andapsychologyofmind)perspective,peoplewereconsciousbeings
strivingforfreedom(fromcontrol),competence,anddignity.Controlbyexternalmeanswasnecessarilybad.Theseviewswerea
Page181

forerunnertocurrentclaimsthatexternalrewardsdamagepeoplesintrinsicmotivation.

EdwardLeeThorndike(18741949)
Watsonsbehaviorismemphasizedobservablebehaviorandtheconditionedreflex.Hisanalysisfocusedontheeventsthatprecedeactionandisusuallycalleda
stimulusresponse(SR)approach.AnotherAmericanpsychologist,EdwardThorndike,wasmoreconcernedwithhowsuccessandfailureaffectthebehaviorof
organisms.Hisresearchemphasizedtheeventsorconsequencesthatfollowbehavior.ThorndikewasthefirstscientisttosystematicallystudywhatB.F.Skinnerlater
calledoperantbehavior.Thorndikecalledthesebehavioralchangestrialanderrorlearning.5
Thorndikewasalwaysintriguedwithanimalbehavior.AtColumbiaUniversity,Thorndikebeganhisfamousexperimentsontrialanderrorlearningincats.Cats
wereplacedinwhatThorndikecalledapuzzlebox,andfoodwasplacedoutsidethebox.Acatthatstruggledtogetoutoftheboxwouldaccidentallystepona
treadle,pullastring,andsoon.Theseresponsesresultedinthecatopeningthepuzzleboxdoorandobtainingfood.Thorndikefoundthatmostcatstooklessandless
timetosolvetheproblemaftertheywererepeatedlyreturnedtothebox(i.e.,repeatedtrials).Fromtheseobservations,Thorndikemadethefirstformulationofthelaw
ofeffect:

Thecatthatisclawingallovertheboxinherimpulsivestrugglewillprobablyclawthestringorlooporbuttonsoastoopenthedoor.Andgraduallyallthe
othernonsuccessfulimpulseswillbestampedoutandtheparticularimpulseleadingtothesuccessfulactwillbestampedinbytheresultingpleasure,until
aftermanytrials,thecatwill,whenputinthebox,immediatelyclawthebuttonorloopinadefiniteway.6

Today,Thorndikeslawofeffectisrestatedastheprincipleofreinforcement,whichstatesthatbehaviormaybefollowedbyconsequencesthatincreaseordecrease
thatbehaviorinthesamesituation.
Intermsofbehaviorism,theresearchbyThorndikedemonstratesthatbehaviorisregulatedbyitsfeedback,orconsequences.Thecatpullsastringandgetsoutof
theboxapersonlooksinadrawerandfindsalostobject.Inbothcases,thebehaviorislikelytoberepeatedundersimilarconditions.Thelawofeffectisnot
somethingapersonchoosestoobey.Justasthelawofgravitydeterminesthatwewillfalltothegroundifwejump,thelawofeffectrequiresthathumanbehaviorwill
increase(ordecrease)duetoitspastconsequences.
Thosewhocontendthatrewardsdecreaseintrinsicmotivationoftenargueagainstthegeneralityofthelawofeffect.Fromthisposition,thelawofeffectisnotabasic
principleofhumannature.Humansareviewedasselfdetermining.Applyingexternalconsequencestoactivitiesissaidtounderminethisbasicaspectofhumannature
andsubvertinterestintheactivities.7Thelargerdebate
Page182

concernstheuniquenessofhumanscomparedtootheranimals.Behavioristsviewbothanimalandhumanbehaviorasdeterminedbythelawsofconditioning,whereas
thosewhoargueagainstthelawofeffectdonotbelievethatthesameprinciplesoperateforbothanimalsandhumans.

BurrhusFredrickSkinner(19041990)
Althoughtheideasofmanyscientistsandphilosophershavehadanimpactonbehaviorism,B.F.Skinnerislargelyresponsibleforthedevelopmentofbehavioranalysis
andthephilosophyofmodernbehaviorism.Skinneremphasizedoperantbehavior,whichisbehaviorthatoperatesupontheenvironmenttoproduceeffectsor
consequences.8Someoftheseeffectsincreasebehaviorandarecalledreinforcement.Theclaimbycriticsthatrewardandreinforcementhavenegativeeffectson
intrinsicmotivationis,inalargepart,achallengetotheviewsofSkinnerandhisscienceofbehavior.
InaccordwithWatson,Skinnertalkedaboutascienceofbehaviorratherthanofphysiologyormentallife.IncontrasttoWatson,however,Skinnerdidnotreject
mentaleventsfromscientificanalysis.Skinnerarguedthatremembering,thinking,feeling,andsoonareprivatebehaviorsoftheorganismthatrequireexplanation.He
proposedthatcontingenciesofreinforcementcontrolbothpublicandprivatebehavior(e.g.,thinking)ofanimalsandhumans.Acontingencyofreinforcementinvolves
events(SD,ordiscriminativestimulus)thatsettheoccasionforbehavior,theoperant(R),andtherequirementsforreinforcement(Sr).
Skinnersmodelofoperantconditioningbreakswithallothermodesofpsychology.Behaviorisselectedbythecontingenciesofreinforcement,andnotdirectedby
mental,cognitive,orphysiologicalprocesses.Behaviorhasarangeofvariation,andtheprocessofreinforcementselectstheformofresponsethatisappropriatetothe
situation.

SKINNERSBEHAVIORALMODEL
Studythebehavioroforganisms,includinghumans
Behaviorisselectedbyitsconsequencesitisnotdirectedbymentalprocesses
Behavior(includingthinkingandfeeling)isduetothebiologyofthespeciesandselectionbyreinforcementduringthelifetimeoftheindividual


ReinforcementisaprocessatthebehaviorallevelthatisanalogoustoCharlesDarwinsconceptofnaturalselectionatthebiologicallevel.9DarwinandSkin
Page183

nerwerebothvehementlyattackedbycritics.Darwinshowedthathumanswerepartofaprocessofevolutionandnaturalselectionandnotuniqueintermsofasoulor
spiritinstead,hestatedthathumansweresimplyoneofmanytypesoforganismsontheplanet.Skinnershowedthatbehaviorwasnotduetoacreativemindor
cognitiverepresentationsbuttothebiologyofthespecies(Homosapiens)andselectionofbehaviorbyreinforcementduringthelifetimeoftheindividual.BothDarwin
andSkinnerchallengedtheWesternculturalidealthathumansareuniquebeingspursuingtheircreativeandpurposivegoalsandobjectives.
Skinneradvocatedtheuseofpositivereinforcementbutnotedthatthemostprevalentformofbehaviorcontrolinoursocietyispunishment.10Thatis,peopleoften
usepunishmenttogetwhattheywantormakepeoplebehaveastheywant.Aliteratureoffreedomhasidentifiedobvioussourcesofpunitivecontrolandadvocated
methodsofescapefromcoerciveenvironments.Becausepunishmentandcontrolareoftenassociated,mostpeopletakethecontrolofbehavior(evenbypositive
rewards)tobecoercive.Forthisreason,thegeneralideaofbehavioralcontrolisaconcerntomanypeople.
Skinnerquestionedwhetheritispossibletoeliminatebehaviorcontrolandwhethercontrolisthesameascoercion.Experimentalevidenceconvincedhimthatthe
controlofbehaviorisafactofnaturethatmaybeinvestigatedanddescribed.Thatis,thecontrolofhumanconductbythesocialandphysicalenvironmentisabuiltin
aspectoftheworld.Peoplecannotchoosetoacceptorrejectbehaviorcontrol,becauseitisafactoflife.Weallacceptthecontrolofphysicalobjects,chemical
reactions,andphysiologicalprocesses.Skinnerarguedthatthecontrolofbehaviorwasjustasnaturalhumansadjusttheirbehavioronthebasisofstimulationand
feedbackfromthephysicalandsocialenvironment.Ifsuchstimulationandfeedbackaresociallyarranged(eitherinadvertentlyorbyplan)forparticularformsof
conduct,thenitfollowsthathumanbehavioriscontrolledbytheactionsofothers.
Page184

Iscontrolthesameascoercion?Skinnersansweristhattheprocessofpositivereinforcementisaformofcontrolbutthecontrolisnotpunitive.Positivereinforcement
isatypeofcontrolthatisseldomidentifiedandlessfrequentlyusedinoursociety.Apositivereinforcementsystemtakestimetoincreaseparticularformsofconduct,
butonceithasbeenestablished,peopleoftendescribethemselvesashappy,contented,andselffulfilledinsuchasystem.AccordingtoSkinner,peopledonotshow
thenegativebyproductsofpunitivecontrol(generalizedescapeandavoidance)whenpositivereinforcementisarrangedfortheirbehavior.Aculturebasedonpositive
reinforcementnotonlyteachesdesiredactions,italsoinstillsacommitmenttothesurvivaloftheculture,itsvaluesandpractices.11

CHALLENGESTOBEHAVIORISM

ThepostWorldWarIIyearshavebeendescribedasaperiodofcomfortableeclecticisminpsychology.12Manyvarietiesofbehaviorismwereofferedinthe
collegesanduniversitiesofNorthAmerica.13Atthesametime,otherschoolsofthoughtbegantoemerge(i.e.,humanism,developmentalpsychology,andcognitive
psychology).Threeimportantdevelopmentsoccurredthatsetthestagefortheresearchonrewardsandintrinsicmotivation.First,thereweredebatesbetween
humanistsandbehavioristsconcerningtheroleoftheenvironmentandthecontrolofhumanbehavior.Atthesametime,thecomputerrevolutionledtotheresurgence
ofmentalrepresentationsandtheriseofcognitivepsychology.Finally,socialpsychologistschallengedlearningtheorysclaimsabouttherelationshipbetweenrewards
andattitude.Thischallengepavedthewayforthesubsequentattackonbehaviorismbysocialpsychologists,whoclaimedthatrewardsdecreasedintrinsicmotivation.

HumanismandBehaviorism
Thehumanisticmovementinpsychologyaroseduringthe1950sand1960sasareactiontobothpsychoanalysisandbehaviorism.Humaniststypicallyhavestrong
negativereactionstobehaviorism,andespeciallytoSkinnersideas.TheybelievethatSkinneropposedtheconceptsoffreewill,purposefulaction,selfactualization,
andinnermeaningsofself.Infact,however,Skinnerprovidedanaccountoffreedom,purpose,andtheselfinbehavioralterms.14Themajorpointofdisagreementwas
thathumanistsbelievedintheautonomoushumanastheguidingforceinhumanbehavior.Skinnerrejectedtheindividualasacausalagent.Hearguedthathuman
behaviorchangedtheenvironmentandthatthesechangesinturncausedsubsequenthumanconduct.Onlybyacknowledgingtheroleoftheenvironment,Skinner
stated,arehumansabletoshapebehaviorthatiscalledcreative,meaningful,andselffulfilling.
Thedisputebetweenhumanistsandbehavioristsishighlightedbytheideo
Page185

logicaldebatesbetweenB.F.SkinnerandCarlRogers.15CarlRogersfoundedaformofhumanisticclinicalpsychologycalledclientcenteredtherapy.Headvocated
fortheuseofencountergroups(atherapygroupthatfocusesonselfawarenessandencouragesopenexpressionoffeelings)anddevelopedatheoryofperson
centeredaction.16Rogersbelievedthathumanswerebasicallygoodbynature,seekinggrowth,fulfillment,andmeaningfulrelationshipswithothers.
Humanistsstatethatinordertoachieveselfactualizationandpositivesocialrelationships,peoplemustbecomeawareof,andexpress,theirinnerexperiences
(perceptions,thoughts,feelings,sensations,etc.).AccordingtoRogers,manypeoplefailinthisquestbecauseoftheirneedforpositiveregard(acceptanceandlove)
fromothers.Usually,significantothersgiveloveandacceptanceonlywhenthepersonlivesuptotheirexpectations.Thatis,positiveregardisgivenonaconditional
basis,and,accordingtoRogers,thisinterfereswithapersonsawarenessoftruefeelingsandbeliefs.Ifpeoplearetoachieveselfactualization,arguesRogers,positive
regardmustbegivenunconditionally.Thatis,peoplemustbecompletelyacceptedforwhattheyareiftheyaretodevelopinpositiveways.
Rogersfoundedclientcenteredtherapytoenhanceselfawarenessandreducethepresumeddetrimentalinfluenceofconditionalpositiveregardbyothers.The
therapyrestedonthreemajortenets:(1)empathicunderstanding,(2)unconditionalpositiveregard,and(3)genuineness.Aclientcenteredtherapistmustsense,
throughempathy,thesubjectivemeaningsofaclientsexperiencesandreflectthisunderstanding.Inthisway,peoplegainmoreawarenessabouttheirexperiences.At
thesametime,thetherapistshowsunconditionalcaringandacceptingoftheclientsfeelingsandbeliefs,allowingthepersontobecomemoreselfacceptingofthefull
rangeofsensations,thoughts,andemotions.Finally,thetherapiststrivestomaintaingenuineness,orhonestyofinteraction,withtheclient.Honestyofinteractionleads
togreatertrustandmoreeaseandcomfortinexpressingonestruefeelings.
Skinnerrejectedtheideathatpeoplecouldlookwithinthemselvesandmakefreechoicestochangetheirselfconcept.Hisanalysisofthetherapeuticsituation
suggestedthatthetherapistandclientwereinteractinginawaythatshapedtheclientsbehavior.Ifatherapistsucceededinbeingcompletelynonjudgmentaland
accepting,Skinnersuggestedthatthisonlyallowedtheclientsbehaviortoreflectitspasthistoryofreinforcementandcurrentinteractionwithothersoutsidethetherapy
session.Forexample,atherapistmayusesubtlewaystoencouragetheclienttobemoreassertiveintalkingabouthisorhertruefeelings.Assertivetalkingmayhave
aneffectonsignificantothers,whothenreinforcethisbehaviorbybeingmorecompliantwiththeclientsrequests.Thesechangesinthesocialenvironment,stated
Skinner,andnottheneedforselfactualization,accountforanymodificationofbehavior.
Intermsofeducationandmotivation,RogersandSkinneralsodifferedinopinions.RogerssviewoflearningwasinaccordwiththeromanticismofJeanJacques
Rousseautheviewwasthatlearningisbestaccomplishedinunstruc
Page186

turedsettings,withlittleinterferencebyteachersandotheradults.17Apersonisfreetopursuepersonalinterestsandtochoosehisorherown,selfguidedpathto
knowledge.Thus,Rogersemphasizedfreedom,independence,andselfactualizationasthebasisofeducation.Socialinfluencebyteachers,parentsandother
educatorswasviewedasdetrimentaltomotivationandlearningaccordingRogershumanistpointofview.18
Skinneragreedthatthegoalsofeducationweretodeveloppositivefeelingsaboutlearningandpromotegreaterfeelingsofindependence.ForSkinner,however,a
personfeelsfreeandenjoyslearningwhentheeducationalenvironmentisarranged(structured)toproducepositivereinforcementforacademicandsocialbehavior.
Thetaskforeducatorsistoreducepunitivecontrolthatleadsstudentstotuneoutordropoutandtoimplementeffectivesystemsofpositivereinforcement.19Skinner
disagreedwithRogerssclaimthatunstructuredsettingsandlowinvolvementbyteachersimproveslearning.Ifstudentslearnundertheseconditions,claimedSkinner,it
isbecauseothersourcesofcontrol,outsidetheclassroom,haveshapedtheirbehavior.Ifeducatorsdonotteachbypositivereinforcement,studentsmaylearnwaysof
actingfromothersourcesthatarenotdesirablefromthepointofviewofparents,teachers,andothersegmentsofsociety.20
Generally,behavioristsandhumanistsagreeonthegoalsofeducationbutdisagreeonthemeanstotheseends.Humanists,onthebasisoftheirphilosophyofhuman
nature,advocateforunstructuredclassroomsandlessinstructionbyteachers.Behaviorists,onthebasisofexperimentalevidence,proposetoreducepunitivecontrol
andincreasetheregulationofacademicandsocialbehaviorbypositivereinforcement.

ACOMPARISONOFHUMANISMANDBEHAVIORISM
Humanism Behaviorism
freewillandpersonascasualagent controlofhumanbehavior
purposefulactiondirectedatgoals operantbehaviorduetopastconsequences
innermeanings contingenciesofreinforcement
selfactualization interplayofbiologyandenvironment

Theclaimthatrewardslessenapersonsintrinsicmotivationisalogicalextensionofthehumanistthesisforeducation.Fromthehumanistperspective,unstructured
classroomsandminimalteacherinvolvementlessenexternalcontrol.Rewardsaresourcesofcontrol,andhumanistsarguethatanyformofcontrolhasnegative
implicationsforselfactualization.21Behavioristsopposethisclaimaboutrewards,justastheyhaveopposedthegeneralviewthatselfactualizationisthecenterof
learningandeducation.Thebehavioralviewisthatstructuredclassroomsandplannedinterventionbyteachersarenecessaryforsuccessfullearningbecause,fromthe
behavioralperspective,controlisinevitable.Ratherthanallowstudentbehaviortobecontrolledbyunplannedcontin
Page187

gencies,educatorscanmotivatestudentsbydesigningapositivereinforcementsystem.

CognitivePsychologyandBehaviorism
Theriseofcognitivepsychologyisanotherhistoricalforcecontributingtothedebateaboutrewardsandintrinsicmotivation.Thisrisehasbeenheraldedasarevolution,
butitismorelikelythatcognitivepsychologyrepresentsareturntopurposivemodelsofhumanbehavior.22EdwinTolmansbook,PurposiveBehaviorinAnimals
andMen,writtenin1932,focusedonanticipatorythinking(cognitivemapsandexpectations)inhumansandanimals.23Forwardthinkingaboutgoals,actions,and
outcomeswassaidtodirectbehavior.ModerncognitivepsychologistshaveadoptedTolmansviewsbuthavesubstitutedcognitiverepresentationsforforward
thinking.
Thereemergenceofcognitivepsychologyisalsoassociatedwiththecomputerrevolutionandworkonartificialintelligence.24Cognitivetheoriespostulatethought
andmemoryprocessesasdeterminantsofhumanbehavior,muchassoftwareprogramsofthecomputerdetermineoutput.Thus,thecomputermetaphorgaveamajor
impetustotheriseofcognitivepsychology.
Cognitiveresearchersclaimthathumanbehaviorisduetounderlyingcognitiveprocessesthataremuchlikeinformationprocessingbycomputers.Thecausesof
behaviorliewithinapersonandthatpersonscognitiverepresentationsoftheworld.Cognitivepsychologistsemphasizethepersonasasourceofcausationandself
determination.25Externalrewardsareharmfulinthisviewbecausetheseeventschangepeoplescognitiverepresentations.Morespecifically,rewardsshiftattributions
ofcontrolfromthepersontotheenvironment,thusunderminingperceptionsofselfdetermination.Thischangeinattributionresultsinlessinterestintheactivitiesfor
whichthepersonwasrewarded.
ThenotionofcognitiverepresentationsofactionsandoutcomesisincontrasttothemodernbehavioralviewthatbeganwithB.F.Skinner.26Inabehavioralaccount,
remembering,thinking,andfeelingarenotcognitiveprocessesthatdirectbehaviorrather,theyareprivatebehaviorsofanindividual.Theseprivateeventsdonot
explainbehaviorbutratherconstitutemorebehaviortobeexplained.Theexplanationofbothprivatebehaviorsandovertactionsrestsonananalysisofthebiologyof
thespeciesandthecontingenciesofreinforcementencounteredbypeopleovertheirlifetimes.27

SocialPsychology,Rewards,andAttitudes
Anotherdevelopmentinthedebateaboutrewardsandintrinsicmotivationcamefromstudiesinsocialpsychologyconductedduringthe1950sand1960s.Social
psychologistshaddevelopedcognitiveaccountsofhumanbehaviorpriortotheemergenceoftheinformationprocessingview.Specifically,socialpsychologistswere
interestedinpersonalandinterpersonalattitudesandhowatti
Page188

tudesinfluencedhumanbehavior.Oneimportantquestionconcernedthefactorsthatproducedattitudechange.LeonFestingerproposedthetheoryofcognitive
dissonanceasanaccountofhowpeoplechangetheirattitudesbasedoninformation.28Wheninformationaboutattitudesandbehaviorisinconsistent,people
experienceastateoftensioncalledcognitivedissonance.Inordertoreducedissonance,peoplechangetheirattitudestobringtheminlinewiththeirbehavior.Thatis,
peoplemaketheirattitudesconsistentwiththeirbehavior.
Animportantaspectofattitudechangeanddissonanceistheeffectofrewardsorincentives.Festingerarguedthatthelargertheexternalrewardapersonreceives
fordoingsomethinginwhichheorshedoesnotbelieve,thelesswillbetheattitudechange.Rewardsandincentivesaregoodjustificationsforactionsthatare
inconsistentwithattitudes.Thegreaterthejustificationfortheaction,thelesswillbethedissonanceandthesmallerwillbetheattitudechange.
Intermsofdissonancetheory,considerthefollowingsituation.Twopeopleareengagedinadull,boringtask,towardwhichbothhaveanegativeattitude.One
personispaid$20tosaythatthetaskisfunandenjoyabletheotherpersonispaid$1tosaythesamething.Thepredictionfromdissonancetheoryisthattheperson
whoispaidthesmalleramountofmoneywillenduplikingthetaskmore(haveamorepositiveattitude)thanthehigherpaidindividual.Thepersonwhowaspaid$20
hasgoodjustificationforsayingthatheorshelikesthetask.Thelargerewardjustifiesthestatementandreducesthedissonancecausedbytheinconsistencybetween
theattitude(thetaskisboring)andthebehavior(sayingthatthetaskisenjoyable).Becausedissonanceislow,attitudechangeisminimal.Thatis,thepersoncontinues
tobelievethatthetaskisdullandboring.Ontheotherhand,thepersonwhowaspaid$1haslessjustificationforthecounterattitudinalstatementandthusencounters
highdissonance.Inordertoreducethedissonance,thepersonchangeshisorherattitudetowardthetask(beginstolikeit).FestingerandCarlsmithcarriedoutan
experimentthatconfirmedthattherewasmoreattitudechange(likingofaboringtask)withlessmonetaryreward.29
Generally,dissonanceresearchappearedtoshowthatexternalincentivesweredetrimentaltoattitudechange.Fromthisperspective,inordertogetchildrentolike
reading,writing,andarithmetic,onewouldusetheleastamountofinducementtogetthemtodotheseactivities.30Becausechildrencannotjustifyreading(orother
academicactivities)intermsofexternalrewards,theywillhavehighdissonance,whichinturnwillleadtoanattitudechange.Inthiscase,thechildrenwillchangetheir
attitudestowardtheschoolsubject(cometolikeit).Thatis,thesmallerthereward,themorepositivewillbetheattitudetowardanactivity.
Inthe1960s,dissonancetheoryspredictionsaboutrewardandattitudechangewereviewedasadirectcontradictiontotheoriesoflearningandbehaviorism.The
claimwasthatlearningtheorieswouldpredictthatthepersonwhowaspaid$20shouldbemoreenthusiasticaboutthetaskthanthepersonpaidonlyonedollar.
Becausetheoriesaboutrewardsandreinforcementdidnotac
Page189

countforthecreationofmoreattitudechangewithlessreward,behavioralprinciplesofreinforcementwerebroughtintoquestion.
DarylBemcarriedoutresearchbasedonabehavioralanalysisofreward,attitude,anddissonanceinthelate1960s.31Usingstandardbehaviorprinciplesand
Skinnersanalysisofverbalbehavior,Bemshowedthatpeoplejudgedtheirattitudesbasedontheirbehavior.Justasanoutsideobserverwouldusetheamountof
rewardtojudgethetruebeliefsofanindividual,sothepersonhimorherselfusestheamountofrewardtoinferhisorherownattitudes.Forexample,apromotional
messagebyanadvertiserorpoliticianislesscredibleandpersuasiveiftheviewerorlistenerfindsoutthatthespeakerisbeingpaidtosayit.FromBemsperspective,
thesameprocessoccursinselfpersuasion.Thus,thepersonwhoispaid$20tosaythataboringtaskisfunandenjoyableusesthelargeamountofmoneytoinferlow
credibilityofthestatementandisnotpersuadedbythemessage(thetaskcontinuestobeseenasboring).ThereisnodissonancereductioninBemsselfpersuasion
analysis.Noticealsothatrewardsdonotserveasreinforcementforbehaviorinthisscenariobutratherserveasstimuliexertingcontroloververbalbehavior(i.e.,
attitudestatements).Thus,theprincipleofreinforcementisnotinconsistentwithBemsreanalysisofdissonanceeffects.
AlthoughBemsreanalysisofdissonance,reward,andattitudechangeservesasastrongdefenseofbehaviorismandbehaviorprinciples,socialpsychologistswere
notconvinced.Ascognitivepsychologybecamemoreprominent,socialpsychologistsbegantotalkaboutcausalreasoninginhumansandattributionstointernalor
externalcauses.32Externalinfluences,likemoney,shiftedtheattributionforcausationfromthepersontotheenvironment.Whenpeopleseethemselvesasdoingthings
forexternalreasons,socialpsychologistsclaimedthattheywillconcludethatinternalcausesareweak.Thus,apersonwhoispaidtodoanactivitywillreasonthatthe
causeoftheactivitywasthemoneyandnothisorherinterest(internalcause)init.Attributiontheoryupheldthepositionthatrewardsareharmful,shiftingcausal
attributionstoexternalsourcesandundermininginternalmotivesforaction.Atthispoint,behaviorismandsocialpsychologybecamelockedintoabattleoverrewards
andintrinsicmotivation.

THEINTRINSICMOTIVATIONLITERATUREINCONTEXT
Aswehaveseen,threedevelopmentsthatbeganinthe1950ssetthestageforresearchonhowrewardsdestroyintrinsicmotivation.Thesedevelopmentsinvolved(a)
humanismsgeneraldistasteforSkinnersviewoffreedom,control,andtechnology,(b)thereemergenceofcognitivepsychologyandtheviewthatpeoplewerean
agentofcausation,and(c)socialpsychologysconcernwithcognitivedissonance,reward,andattitudechange.
Bythe1960sand1970s,bothcognitivepsychologyandbehaviorismhadbecomeestablishedschoolswithinpsychology.Behavioristsemphasizedtheim
Page190

portanceofbehaviorenvironmentrelationshipscognitivistsattendedtointernalprocessesandmentaleventswithintheindividual.Thisdifferenceinfocusledadvocates
fromthetwocampstodebateandcriticizeeachotherspositions.Inaddition,Skinnersstrictdeterminismandemphasisoncontrolrousedagreatdealofcontroversy.
ItrancontrarytothebasicbeliefinfreewillthatisheldbymostNorthAmericansandledseveralpsychologiststoputforthapositionofhumansaswillfulandself
determining.Itwasduringthisperiodofrivalryandunrestthattheresearchontheeffectsofrewardandreinforcementonintrinsicmotivationwasinstigated.
Cognitivesciencegainednumeroussupportersinthe1960sand1970s.BecauseSkinnersbehaviorismrepresentedthemajorbehavioralinfluenceinpsychology,he
andhisfollowerscameunderincreasingattack.33Manyofthecriticismswerebasedonmisconceptionsaboutbehaviorism.Onewouldoftenhearbehaviorism
describedasamechanistic,stimulusresponseaccountofhumansthatdidnotconsidergeneticdeterminantsofbehavior.Behavioristswereaccusedofrepresenting
peopleasnonthinking,nonfeelingautomatons(asanaside,itisinterestingtopointoutthatitiscognitivepsychologistswhoviewhumansascomputers,not
behaviorists).Oneauthorstated:

Ofallthecontemporarypsychologists,B.F.Skinnerisperhapsthemosthonoredandthemostmaligned,themostwidelyrecognizedandthemost
misrepresented,themostcitedandthemostmisunderstood.Somestillsaythatheisastimulusresponsepsychologist(heisnot)somestillsaythat
stimulusresponsechainsplayacentralroleinhistreatmentofverbalbehavior(theydonot)somestillsaythathedisavowsevolutionarydeterminantsof
behavior(hedoesnot).Theseandothermisconceptionsarecommon.34

HowardRachlinsuggestedthatSkinnersviewshavebeenlargelymisunderstoodbecausepeoplethinkthatmentalisticvocabularyisforbiddentothebehaviorist.35He
pointedoutthatthecriticaldifferencebetweenabehavioristandamentalististhepointatwhichexplanationstops.Thebehavioristexplainsbehavior(andmental
states)astheresultofpastexternalbehaviorenvironmentinteractions,whereasthecognitivistinfersmentalprocesses(perceptions,beliefs,expectancies)asthecausal
focusofcurrentbehavior.
Aswellasthenumerousmisunderstandingsthataresprinkledthroughoutthepsychologicalliterature,seriousargumentswerealsodirectedatbehaviorism.Several
writerscouldnotacceptSkinnersrejectionofmentalistictheories,statisticalanalyses,hypothesistesting,andinferredprocessesandarguedthatinferredprocesses
andsimilarphenomenahaveplayedamajorroleinothersciences.36B.Schwartz,R.Schuldenfrei,andH.Laceymaintainedthatbehaviorismhasrestricted
explanatorypower.37Theysawnoevidencethatbehaviorprinciplesoperateoutsidelaboratorysettingsandinsteadarguedforateleologicalexplanationofhuman
behaviorbasedonpurposesandintentions.Theat
Page191

tacksandcriticismsofbehaviorismhave,ofcourse,notgoneundefended,norinmanycaseshavetheybeenunprovoked.38
PerhapstheissuethatcreatedthemostfurorconcernedSkinnerspositionwithregardtofreedom,control,andtechnology.Manysimplycouldnotacceptaviewof
humansascontrollednoraccepttheconsequencesofdeterminism.39Toacceptarigorousdeterminismandapplyittoonesownbehaviorisextremelydifficult,
requiringtheoverthrowofalifetimeshabitsofthought.40Instead,somepsychologistsstressedthenotionofpersonalcausationandarguedthathumansaregoal
directed,selfdetermining,freeagents.41WithregardtoSkinnersproposalforatechnologyofbehavior,oneauthorcontendedthatfollowingWorldWarII,people
becamesuspiciousandskepticalabouttheprospectsforsolvingtechnologicallygeneratedproblemswithfurthertechnologyofanysort.42
Amidthedebatesandthebattlesagainstbehaviorism,ThomasKuhnpublishedhisimportantbook,TheStructureofScientificRevolutions.43Putsimply,Kuhns
thesiswasthatanomaliesorproblemsperiodicallyariseinasciencethatcannotberesolvedundertheprevailingparadigm.Arevolutionoccurswhenoneparadigmis
discardedinfavorofanother.Inevitably,suchtimesaremarkedbydebateandresistancetoparadigmchange.Kuhnsworkreceivedwiderecognition.Inparticular,in
the1960sand1970s,psychologistsbegantoclaimthatascientificrevolutionwastakingplaceinpsychologyrightundertheirnoses.44Theviewwasthatbehaviorism
wouldbereplacedbycognitivescience.Overtheyears,severalwritershavearguedthatacognitiverevolutiondid,infact,takeplaceinthe1960s.45Forexample,in
1977,MacKenziedeclaredbehaviorismdead.46
Leaheysuggestedthatthereneverhasbeenacognitiverevolutionandinstead,thattheideawasignitedinthe1960sand1970sbyKuhnspublicationitself.47
Kuhnsworkservedasarallyingflagforcognitivists.48Fromthisperspective,manyofthecriticismsaimedatbehaviorismcanbeconstruedasresultingfromazeal
forrevolutionspurredbyareadingofKuhn.
Psychologyinthe1960sand1970scanbecharacterizedasundergoingaperiodofturmoil.Manypsychologistsbecameconvincedthatcognitiveprocessesand
mentaleventsshouldbethefocusoftheirdiscipline.Giventhisview,theyattemptedtodiscreditbehavioralviews.Kuhnsworkwasoftencitedtosupportthe
cognitivepositionandpointafingeratbehaviorists.Interestingly,atthesametimethatpsychologistswereattackingbehavioralviewsandvoicingtheirsuspicionsabout
behavioraltechnology,theuseofbehavioraltechniqueswasexpandinginappliedsettings(e.g.,classroomtokeneconomiesandincentivesystems).
Giventhepervasiveinfluenceofcognitivepsychologyandthegrowingapplicationofbehaviormodificationtechniques,theresearchexaminingtheeffectsof
reinforcementonintrinsicmotivationwasatimelyandrelevantconsideration.Inaddition,becausemostofthecriticismsaimedatbehaviorismwerebasedonlogical
argumentratherthanexperimentalfindings,thetimewasripetoproduce
Page192

researchevidencedemonstratingnegativeeffectsofreinforcement.Researchfindingsthatshowedthedetrimentaleffectsofrewardsattackedtheveryheartof
behaviorismreinforcementandhadwidespreadimplicationsforbehavioraltechnology.ItiswithinthiscontextthattheearlystudiesbyEdwardDeciandbyM.R.
Lepper,49D.Greene,andR.E.Nisbettwereconductedandinterpreted.50
Asindicated,theresearchonintrinsicmotivationcameoutofsocialpsychology.Cognitivedissonancetheory,whichwasprominentinsocialpsychologyinthe
1960s,seemedtoshowthatrewardsandincentivesweredetrimentaltoattitudechange.Bythelate1960s,attributiontheoryandselfperceptionaccountsofattitudes
weregainingpopularity.51AlthoughBems(1972)selfperceptionaccountupheldbehaviorprinciples,Kellys(1967)attributiontheoryimpliedthatrewardswould
shiftattributionsofcausationfrominternaltoexternalsourcesandtherebyundermineinternalreasonsormotivesforaction.Thus,socialpsychologistsarrivedatthe
topicofrewardsandintrinsicmotivationviaaninterestinattitudesandattitudechangesubsequently,rewardswereseenasharmfulbasedonpeoplescausalreasoning
andaviewthatpeoplewerecausalagentsofaction.

SUMMARYANDCONCLUSIONS
Theargumentputforthinthischapteristhattheliteratureconcernedwiththenegativeeffectsofrewardandreinforcementonintrinsicmotivationishistorically
groundedinthedebatebetweenbehaviorismandotherpsychologies.Aspartoftheconfrontation,socialpsychologistsofthe1970sbegananempiricalattackon
reinforcementtheoryandpractice.Thiswasatimewhenbehaviorismcameunderfire,especiallyintermsofitsapplicationsandtechnology.Cognitivepsychologywas
becomingwellestablished,andmanypsychologistsarguedthatinternalcognitiveprocessesandnotreinforcementcontingencies,weredirectcausesofbehavior.
FueledbyKuhns(1962)publicationonscientificrevolutionsandafearandsuspicionofbehavioraltechnology,manycriticssetouttodemonstratetheinadequaciesof
behavioralviews.Comingoutofconceptsinsocialpsychologyandageneralclimateofdissatisfactionaboutbehaviorism,numerousstudiesontheeffectsofreward
andreinforcementweregenerated.
Decisinitialstudiesestablishedtheresearchparadigmunderwhichthenegativeeffectsofrewardandreinforcementonintrinsicmotivationwere,andcontinuetobe,
investigated.52AlthoughDecisfindingswereweak,hearguedthatreinforcementdidindeedproducedecrementsinintrinsicmotivation.Giventheprevailingcognitive
orientationtowardpsychology,itisnotsurprisingthatDecisclaimswerequicklyseizedonandfurtherinvestigationswereconductedtoconfirmhisconclusions.In
addition,anumberofpsychologistsbegantoissuewarningsthatDecisdemonstrationsofthenegativeeffectsofrewarddirectly
Page193

THESOCIOHISTORICALCONTEXTOFTHEREWARDSANDINTRINSICMOTIVATIONCONTROVERSY

contradictedreinforcementtheoryandhadrelevancetotheuseofincentiveprogramsinschoolsandotherinstitutionalsettings.
TheexperimentbyLepper,Greene,andNisbett53isperhapsthemostwidelycitedexampleofthedetrimentaleffectsofreinforcement.Whatisstrikingaboutthis
studyisthatonlychildrenwhowerepromisedarewardshowedadecreaseinintrinsicmotivationrelativetoacontrolgroup(seeChapters2and3fordetails).
Childrenwhoreceived,butwerenotpromised,arewarddemonstratedanincreaseinintrinsicinterest.Giventhesefindings,alogicalconclusionisthatthepromise,
ratherthantherewardperse,mayhaveproducedthedifferences.Unfortunately,however,theresultshaveseldom,ifever,beeninterpretedinthisway.
Instead,manypsychologistshaveassertedthatthestudybyLepper,Greene,andNisbett(1973)isyetanotherdemonstrationoftheunderminingeffectsof
reinforcement.Intheirenthusiasmtoattackthegeneraltenetsofthebehavioralposition,manyhavemistakenlyequatedpromisedrewardswithrewardsandrewards
withreinforcers.
Page194

AsshowninChapters2and3,theearlyresearchonnegativeeffectsofrewardsandreinforcementwasweakandinconclusive.Nonetheless,thefindingshave
frequentlybeenusedtodiscreditbehavioralviewsandreinforcementtheory.Oursociohistoricanalysisofthisliteraturesuggeststhatovergeneralizationsand
misinterpretationsofthestudieswerearesultofthefervorofthetimes(whichsoughttooverthrowbehaviorism)ratherthanacarefulanalysisofthedata.Our
examinationofthestudies,however,showsthattheliteratureonrewardsandintrinsicmotivationhasnotresultedinadiscreditingofbehavioraltheoryor,specifically,
theprincipleofreinforcement.Instead,whatresultedwasaliteraturethatpointstoconditionsunderwhichanegativeeffectofpromisedrewardwilloccur,and,
inadvertently,ithasrevealedseveralconditionsthatresultinpositiveeffectsofrewards.Thus,itisourcontentionthatifthestudiesonrewardsandintrinsicmotivation
hadbeenconductedinanothertimeorplace,withdifferentdominantpsychologicalviews,thefindingsmighthavebeeninterpretedinaverydifferentlight.

NOTES
1. Leahey,1987,p.262.
2. Ibid.
3. Dewey,1900/1978.
4. Watson,1913.
5. Thorndike,1898.AlthoughThorndiketalkedabouttrialanderrorlearning,thisisnotthewayinwhichmodernbehavioristsunderstandthepuzzleboxexperiments.Insteadoftrial
anderrorlearning,theexperimentsarebetterdescribedastrialandsuccesslearning.Thecatsbehaviorofpullingonastring(orpressingatreadle)resultedinopeningthedoor
tothepuzzlebox(success).Theserepeatedsuccessesaccountfortheincreasingefficiencyofthecatsinthissituation.
6. Thorndike,1911/1965,p.40.
7. DeciandRyan,1985.
8. Skinner,1938.Operantconditioningprinciplesarecriticizedasdehumanizing.CriticsarguethatSkinnercharacterizedpeopleaspassiveresponderstoeventsfromtheenvironment.
Infact,however,operantconditioningrequiresanactiveperson,whoemitsbehaviorinvarioussituations.Whenpeopleactontheworld,theychangeit.Thesechanges,inturn,
feedbackontheactionsthatproducethem,eitherincreasingordecreasingthelikelihoodofthisbehavior.
9. Skinner,1987.Inthisessayonselectionbyconsequences,SkinneralignshimselfwithDarwininadvocatingthreelevelsofselection:(1)naturalselectionandtheevolutionofthe
species,(2)behavioralselectionattheleveloftheindividualduringhisorherlifetime,and(3)culturalselection(thechangeinculturalpracticesovergenerations).
10.Skinner,1971.
11.Skinner,1953.
12.Leahey,1997.
13.SeeZuriff,1979,1985.
14.Skinner,1953.
15.RogersandSkinner,1956.ThisarticleconcernstheargumentbetweenRogers
Page195

andSkinneratasymposiumheldbytheAmericanPsychologicalAssociation.BothRogersandSkinnerdiscussedvariousissuesregardingthetopicofhowvaluesemergeinour
society.
16.Rogers,1964.
17.Rousseau,1762/1974.
18.Rogers,1969.
19.Sidman,1989.
20.Skinner,1968.
21.Inthedecadesofthe1950s,1960sand1970s,thehumanistmovementledtotheestablishmentofavarietyofalternativeschools.Theaimofmanyoftheseinstitutionswasto
reduceteacherinterventionandtocreateanatmospherewherestudentswouldcometodiscovertheirowninnerpotential.Proponentsofthesegoalsheldtheviewthatstudents
shouldnotberewardedorreinforcedforengaginginactivitiesinstead,theyassumedthatthechildrenwouldcometoengageinschoolrelatedtasksbecauseoftheself
satisfactioninherentinsuchactivities.ThisviewwasreflectedinthewritingsofA.S.Neil(1959),whostated:

Thedangerinrewardingachildisnotasextremeasthatofpunishinghim,buttheunderminingofthechildsmoralethroughthegivingofrewardsismoresubtle.
Rewardsaresuperfluousandnegative.Toofferaprizefordoingadeedistantamounttodeclaringthatthedeedisnotworthdoingforitsownsake....Areward
should,forthemostpartbesubjective:selfsatisfactionfortheworkaccomplished.(pp.162163)

Neilandotherhumanistswerehighlyinfluentialinthe1960sand1970sandmanyteachersadoptedtheviewthatchildrenwouldflourishbestinanunstructured
environment.Unfortunately,inmanycases,theconsequencesofsuchasystemwerenotgood.Manystudentsdidnotfindtheacademictasksselfsatisfyingthe
resultwasthatsuchstudentsdidnotlearnthebasicskillsofreading,writing,andmathematics.Someparentsandeducatorsbeganabacklash,andseveral
booksappearedthatwerecritiquesofthehumanistmovementanditseffectsoneducation(e.g.,seeEngelmann,1992Flesch,1965,1981Kline,1973).
22.Leahey,1992.
23.Tolman,1932.
24.Simon,1956Newell,Shaw,andSimon,1958.
25.SeeNeisser,1967Miller,1962.
26.Skinner,1978.
27.Skinner,1969.
28.Festinger,1957.
29.FestingerandCarlsmith,1959.
30.Freedman,1965.
31.Bem,1965.
32.Heider,1958DeCharms,1968.
33.AlthoughmostoftheattacksonbehaviorismwereaimedattheviewsofSkinner,severalresearcherswhoinvestigatedtheeffectsofrewardsonintrinsicmotivationincluded
Banduraasabehaviorist(e.g.,seeRyanandDeci,1996).
34.Catania,1984,p.473.
35.Rachlin,1984.
36.Forexample,seeWessels,1981.
37.Schwartz,Schuldenfrei,andLacey,1978.
Page196

38.Skinner,1974,1978.
39.Forexample,seeKrutch,1953.
40.Leahey,1987,p.461.
41.DeCharms,1968Deci,1975.
42.Smith,1992,p.221.
43.Kuhn,1962.
44.Palermo,1971.
45.Forexample,seeBaars,1986.
46.SeeMacKenzie,1977.However,alsoseeZuriff,1979,whopointedoutthatMacKenziesconceptionofbehaviorismwasHullian,notradicalbehaviorist.
47.Leahey,1992.
48.Peterson,1981.
49.Deci,1971,1972a,1972bLepper,Greene,andNisbett,1973.
50.Althoughtheearlyfindingswereweakandcircumscribed,theyhavebeenfrequentlyinterpretedasdemonstrationsofthenegativeimpactofreinforcement.Theyhavealsobeen
usedtorejectthescienceofbehavior,itsprinciples,anditsprograms.
Bythe1970s,behavioralpsychologistshaddevelopedanumberofbehavioralprogramsineducationalsettings,hospitals,andtheworkplace.Suchprogramsusedreinforcement
principlesandweredesignedtoteacharangeoflifeskillsandtoincreasesociallyappropriatebehavior.Anextensiveliteraturedocumentedthebeneficialoutcomesofsuch
programs(AyllonandAzrin,1968Kazdin,1975a,1975bKazdinandBootzin,1972).Acentralissueofconcerninthebehavioralliteraturewaswhethergainsmadeintreatment
settingswouldbemaintainedinnontreatmentenvironments(Carlson,Hersen,andEisler,1972Kazdin,1975b).Inotherwords,onceapersonhadacquiredparticularskillsand
behaviorsintheprogramsetting,theconcernwaswhethertheseskillsandbehaviorswouldbemaintainedineverydaysettings.
Manynonbehavioralpsychologistsandeducatorsseizedupontheresearchfindingsonrewardandlossofintrinsicinterestasdirectlyrelevanttothisissue.Thelaboratoryresults
indicatingthatexpectedrewardsdecreasedintrinsicmotivationwereusedtoarguethatbehavioralprogramswouldfailinthelongrunbecausetheywouldunderminepeoples
intrinsicmotivation.Ina1974articleinAmericanPsychologist,LevineandFasnachtcriticizedbehavioralprogramsasharmfultointrinsicmotivationandusedthiscriticismtoattack
behaviorism.Theauthorsstatedthatthetimehascomeforustoavoidanarrowoperant[behavioral]perspective.Operantprocedureshavetheirplaceandtheirdangers(p.819).
Oneproblemwiththisconclusionisthatitwasbasedonacircumscribedlaboratoryfindingthathasnotbeenshowntoholdinappliedsettings.Thatis,thereisnoevidencethat
behavioralprogramsundermineintrinsicmotivation.Theissueofbehavioralmaintenanceisfarmorecomplexthanasimpleunderminingofintrinsicmotivation.Maintaining
behaviorinvolvesanunderstandingofthereinforcementcontingenciesoftheprogramandthereinforcementcontingenciesoperatingineverydaysettings(seeMartinandPear,
1999StokesandBaer,1977).
Theoriginalstudiesonrewardsandintrinsicmotivationgeneratedfurtherresearch.Somereviewersoftheliteraturenotedthattheresultsweremixedandconcludedthatfew
generalizationscouldbemadeaboutthenegativeeffectsofreward,letalonethoseofreinforcement.Otherwriters,however,threwcautiontothewindandinterpretedtheresultsas
clearevidencethatreinforcementtheory(andbehaviorism,ingeneral)wasflawed(DeCharmsandMuir,1978Kohn,1993a,1993bLepperandGilovich,1981LepperandGreene,
1978Schwartz,1990).Overtheyears,theviewthatrewardsand
Page197

reinforcementareharmfulhasbeenextolledinpsychologytexts(e.g.,Zimbardo,1992),aswellasineducation(e.g.,Tegano,Moran,andSawyers,1991),business(e.g.,Kohn,
1993b),andeventhenaturalsciences(e.g.,Sutherland,1993).
51.Kelley,1967Bem,1965.
52.Deci,1971,1972a,1972b.
53.Lepper,Greene,andNisbett,1973.
Page198

Thispageintentionallyleftblank.
Page199

PARTVI
PRACTICALAPPLICATIONSOFREWARDS
Page200

Thispageintentionallyleftblank.
Page201

Chapter11
TheEffectiveUseofRewardsinEverydayLife
Theuseofrewardsineverydaylifeisatrickybusiness.Oneproblemisthatrewardsareoftengiventogetpeopletodowhatwewant.Undersuchconditions,rewards
arenotpositivemotivatorsofbehaviorinstead,theyserveasnegativeincentives.Forexample,teachersmayofferstudentsgoldstarsforbeinggoodandemployees
maybepromisedrecognitionandadvancementiftheyareproductive.Insuchinstances,teachersandmanagersofferandgiverewardsbasedontheirjudgmentsand
evaluationsofstudentandemployeeperformance.Becausethestandardsofperformanceareoftenvagueinthesecircumstances,rewardscanbemanipulatedtothe
advantageofthoseincharge.Sucharewardsystem,fromthepointofviewofstudentsandemployees,canquicklybecomeunfairandcoercive.Whatisgoodtoday
maynotbegoodtomorrow.Thus,thoseinsupervisorypositionscanwithholdtherewardsthatarecurrentlybeinggivenforacceptableperformanceandrequirebetter
andbetterlevelsofaccomplishment.Rewardsystemsarrangedinthiswayareprogrammedtobackfirepeoplewilleventuallyshowwillfulnoncompliance,escape,
avoidance,and,insomecases,rebellion.
Thesenegativesideeffectsofrewardshavenotgoneunnoticed.Theliteratureonintrinsicmotivationpointedtothesocalledunderminingeffectofrewardsand
claimedthatallextrinsicrewardshadnegativeeffectsonpeoplesinterests.Ouranalysisoftheresearchonrewardsandintrinsicmotivationmakesitclearthatcertain
rewardproceduresdoreducemotivationatthesametime,themetaanalysisshowsthattherearealsowaystouserewardstoenhanceperformanceandinterest.
Theeffectiveuseofrewardsineverydayliferequiresthatwemakeadis
Page202

tinctionbetweenrewardsasincentivesandrewardsasbehavioralconsequences.Whenrewardsareusedasincentives,theofferofrewardcomesbeforethebehavior.
Muchoftheliteratureonrewardsandintrinsicmotivationconcernstheuseofincentives(expectedrewards)andisinstructiveabouthowtoarrangeoffersofrewardto
motivateperformance.Rewardsalsocanbeusedasbehavioralconsequences.Inthiscase,rewardsoftenfunctionasreinforcementforbehavior,strengtheningthe
action.Thestudiesofrewardandintrinsicmotivationthatusedrepeatedrewardsshowthatreinforcementcanbeusedwithoutalossofintrinsicmotivation.Inthis
chapter,weprovideguidelinesforusingincentivesandreinforcementineverydaylifetomotivateperformanceandinterest.

EFFECTIVEUSEOFINCENTIVESINEVERYDAYLIFE
Onewaytouserewardsineverydaylifeistoofferpayment,prizes,orrecognitionforsomeaction.Theofferofrewarddescribeswhatwillhappenifthepersondoes
someaction,performsataspecifiedlevel,completesassignedtasks,andsoon.Offers,then,areverbaldescriptionsofcontingencesratherthantheactual
contingenciesthemselves.1Theseverbaldescriptionscomebeforethetargetbehaviorandareusuallystatedtomotivatethepersontobehaveinagivenway.
Becauseoffersofrewardprecedeactionandareattemptsbypeopletomotivateandguidebehavior,werefertotheseoffersasincentives.
Whenpeopleuseincentivesratherthanactualcontingenciestomotivateandguidebehavior,theeffectsoftendependonthepersonshistoryofbehavioral
consequences.Theeffectofsuchapersonalhistoryisusuallycalledanexpectancyand,whenofferedanincentive,peopledifferintheirexpectations.Apersonwho
hascompliedwithincentivesinthepastwillbemotivated(behaveasinstructed)ornotbycurrentincentivesdependingonthepreviousconsequencesoffollowing
incentives.
Forexample,astudentwhoisofferedrewards(e.g.,couponsforpizzas)forreadingforonehouradaymaybenefitinmanyways(e.g.,heorshemayenjoyreading
books,readingamenutoorderfood,readingdirectionstofindalocation,andsoon).Basedonthisbehavioralhistory,thestudentexpectspositiveoutcomesandis
motivatedbycurrentincentivesofferedbyparentsorteachers.Anotherstudentmaybegivenanincentivetositquietlyinhisorherseat(e.g.,agoodstudent
certificate)anddiscoverthatincentivefollowingbehaviorremovesorreducesotherrewards(e.g.,seeinginterestingthingsthroughthewindow,discussingeventswith
classmates,gettingtheteachersattention).Thisstudentmayexpectnegativeoutcomesforincentivefollowingandbelessinclinedtobehaveinaccordwithsubsequent
incentivesfromthoseinauthority.Thegeneralpointisthatforaspecificperson,theeffectsofanincentivewilldependonfeaturesassociatedwithpriorconsequences
(whetherpositiveoraversive).Thesefeaturesmayhavetodowiththepersonwhoofferstheincentive,thecontextorsettinginwhichtheofferoccurs,andtheverbal
formoftheincentiveitself(howitisstated).
Page203

Althoughtheeffectsofoffersofrewarddependonpersonalhistory,theresearchonrewardsandintrinsicmotivation,aswellasstudiesofsociallearning,suggest
guidelinesfortheeffectiveuseofincentives.Oneguidelineistouseincentives(offersofreward)forthebenefitofthelearner(oremployee)andnotforthebenefitof
socialagentsorpersonsinauthority.Rewardsthatareusedtogetpeopletodowhatwewantarefrequentlyofferedandgivenforthebenefitofthoseinauthority
(parents,teachers,managers,etc.).Offersofrewardsforstudentstositquietlyinclassareusuallyarrangedforthebenefitoftheteacher,notthestudents.Similarly,
incentivestoincreaseproductioninafactoryareoftengivenforthebenefitofmanagersratherthanworkers.Mostofthesemotivationalsystemsareultimatelybased
oncoercionratherthanpositivereinforcement.Insuchcases,theofferofrewardisoftenbackedupbypunishmentorthreatsofpunishment(lossofemployment,being
senttotheprincipal,andsoon).
Incontrast,incentivesgiventohelpapersonlearnormasterataskoperateforthebenefitofthestudentorworker.Thatis,beyondtherewardsofferedto
motivatebehavior(theincentiveitself),peoplebenefitinotherwayswhenrewardsaretiedtomasteryandskillfulperformance(e.g.,increasedperceivedcompetence
orselfefficacy,contactwithalternativesourcesofreward,etc.).Incentiveprogramsdesignedforthebenefitoflearners(andworkers)arelikelytoinvolvethepositive
motivationofbehaviorandgeneratehighperformanceandenjoymentoftheactivities.

IncentivesforPerformingUninterestingActivities
Atthepresenttime,theresearchonrewardsandintrinsicmotivationsuggeststhatincentivescanbeusedtoenhanceperformanceandinterestofactivitiesthatpeople
donotliketodo.Onewayofknowingwhethertheactivityisenjoyableistoassessitsrelativefrequencyofoccurrence.Thatis,whatproportionoftimedoesaperson
spendonthisactivityinagivensetting?Inaschoolsituation,studentsmayspendrelativelylittletimeonacademicactivitiesandmuchmoretimeonofftaskbehavior.
Ourmetaanalysisindicatesthattheuseofincentivesforactivitiesoflowinitialinterest(relativelylowfrequencybehavior)mayenhancethemotivationtoengagein
theseactivities.Animportantpointisthattheeffectsonperformanceandinterest(inthelaboratory)aremoderateandtemporary.Inappliedsettings,incentivescould
beusedtogetthebehaviorgoing,butthesestimuliwillhavetobesupplementedbybehavioralcontingenciestokeepthebehaviorathighstrength(seethesectionon
reinforcementlaterinthischapter).

TangibleIncentivesforPerformingInterestingActivities
Themostrobustfindingintherewardsandintrinsicmotivationresearchconcernsthenegativeeffectsoftangibleincentivesformerelydoingactivitiesthatpeoplelike
todo.Thatis,offersofrewardwithoutclearperformancestan
Page204

dardswereconsistentlyfoundtoreduceperformanceandinterest.Astudentwhoenjoysartisticactivitiesmaydolessofthemifateacheroffersgoldstarsfordoingart
andsculpturetotheclass(ontheotherhand,studentswhoarenotinterestedinartisticactivitiesmaybenefitfromtheincentive).Theexactbasisofthiseffectisnotyet
resolveditcouldbeduetoalossofintrinsicmotivation,ashiftinattributiontoexternalsources,aweakeningofcontrolbynaturalconsequences,ordeliberate
noncompliancetoadirectivegivenbythoseinauthority(backedupbypunishmentorthethreatofpunishment).Whatevertheexplanation,peopleseemtoturnoff
whenconfrontedwiththiskindofincentive.Thebestadviceisnottouseoffersoftangiblerewardsfordoingactivitiesthatpeoplealreadyenjoydoing.
Ifoffersoftangiblerewardsaretobeused,itisimportanttotiematerialincentivestotheattainmentofperformancestandards.Ourmetaanalysis2indicatedthat
tangiblerewardsthatareofferedforexceedingacriterionorperformancestandardcanenhanceperformanceandinterest,evenwhenpeoplealreadyenjoytheactivity.
Inaddition,researchinsociallearningindicatesthattangiblerewardsthatareofferedformastery,effort,andmeetingchallengeshavepositiveeffectsonperformance,
perceivedcompetenceandinterest.Asageneralrule,materialincentivesshouldbelinkedtospecific,reasonable,andattainableperformancestandards.3Anemployee
whoisofferedrewardsbasedonagraduallyincreasingmasteryofcomputersandtherelevantsoftwareprogramsislikelytogaininperformance,competence,and
interest.Themostimportantguidelinefortheuseofmasterybasedincentivesistoensurethatpeoplehavetherequisiteskillstomeettheperformancestandards
requiredforrewardateachstep(incentivestiedtosuccess).
Theuseofincentivesforexceedingaperformancestandardhasbeenasuccessfulstrategyinbusinessandindustry.4Inonecase(HoneywellInc.),management
implementedaprogramofrecognitioncalledTheWinningEdgeforsuperiorperformanceaboveandbeyondonesjob.Employeeswererecommendedtoa
reviewcommitteebytheirworkmatesfortheaward($100andhavingtheirnameslistedinthecompanysnewsletter).Theprogramincreasedemployeecooperation
andthegeneralmoraleoftheworkenvironmentandwasconsideredsuccessfulandcosteffectivebythecompanyandworkers.
Inanotherindustry(BlanchardTrainingandDevelopment,Inc.),themanagementestablishedtheEagleAwardtorecognizetheprovisionoflegendaryserviceto
customersbystayinglatetoshipmaterials,helpingtolocatealostorder,resolvingabillingproblem,andsoon.Again,employeeswerenominatedbytheirworkmates,
whoprovidedabriefdescriptionoftheexceptionalperformance.Areviewcommitteethatensuredthatthedescribedperformanceexceededthejobdescription
consideredthenominations.WinnersweresurprisedwithavisitfromtheEagleCommittee,andthewinnerreceivedanEagleAwardtrophy.Inaddition,apictureof
thewinnerholdingtheEagleAwardwasdisplayedinthelobbybulletinboardwithabriefdescriptionoftheexceptionalperformance.Thewinnergottokeepthe
trophyonhisorherdeskforaweek
Page205

orsountilitwasneededforanotherrecipient.Theprogramwasviewedassuccessful,andlegendaryservicebecamepartofthecompanysculture.
AdivisionofAmericanExpressofNewYork(TravelRelatedServices)implementedanexceptionalperformanceprogram.Theprogram,calledGreat
Performers,recognizedandrewardedexceptionalperformancesofemployeesbyplacinglifesizedpostersofwinnersandtheiraccomplishmentsthroughoutthe
company.WinnerstookhometheirpostersandwerealsoeligibletobecomeGrandAwardwinners(nolimitonnumberofwinners)ofanexpensepaidtrip,$4,000in
travelerschecks,andaplatinumGPlogopinandframedcertificate.TheGreatPerformersprogramincreasedTravelRelatedServicesincome500percentover11
years,andthecompanysreturnonequityhasincreasedby28percentsincetheprogrambegan.
Theresearchonrewardsandintrinsicmotivationhasexaminedalimitedsetoftangiblerewardcontingenciesfocusedmainlyonactivitiesthatpeopleliketodo.Even
so,thereisaclearmessageaboutthesuperfluoususeofmaterialrewards.Thebasicprincipleisthattangiblerewardscanundermineperformanceandinterestwhen
offered,withoutregardtoperformancestandards,foractivitiesthatpeopleenjoydoing.Thenegativeeffectsofmaterialincentivesarelessenedorreversedbytying
rewardstospecificperformancestandards.

USINGINCENTIVESEFECTIVELY
Useincentives(offersofreward)forthebenefitofthelearnerorworker,notforthebenefitofthesocialagent(teacher,manager,etc.).
Useincentivestoenhanceperformanceofactivitiesthatpeopledonotliketodo(relativelylowfrequencyofoccurrence).
Reducetheuseofmaterialincentivesforactivitiesthatpeoplealreadyliketodo(relativelyhighfrequencyofoccurrence).
Reducetheuseoftangibleincentivesofferedtopeopleformerelydoingactivitiesthattheyalreadyliketodo.
Tiematerialincentivestospecific,reasonable,andattainablestandardsofperformance(rewardsforSuccess).

Encouragement:UseofPraiseasanIncentive
Asanalternativetotangibleincentives,theresearchonrewardsandintrinsicmotivationindicatesthatincentivesbasedonpraiseusuallyenhancemeasures
Page206

ofintrinsicmotivation,evenforactivitiesthatpeoplealreadyliketodo.Inappliedresearch,5praiseisdefinedasaresponseofteachers(ormanagers)thatgoesbeyond
thecorrectivefeedbackofright,yes,OK,andsoon.Praiseinvolvesanevaluativeresponsesuchasverygoodandsignsofpositiveaffectsuchasexcitement
orwarmth.Areactiontoastudentwhosucceedsatamathematicsassignmentsuchas,Yougotthemallright,greatwork!isanexampleofpraise.
Thedistinctionbetweenpraiseandcorrectivefeedbackisusefulbecausehumanlearningrequirescorrectivefeedbackbutdoesnotrequirepraise.Thatis,people
requirefeedbackontheirperformancetogainproficiencyandskill6theydonotneedpraisefromothersinordertolearnandperformathighlevels.Thus,theroleof
praiseinworkorschoolsettingsisopentoquestion,eventhoughthelaboratorystudiessupportitsuse.Oneissueisthatlaboratoryresearchonrewardsandintrinsic
motivationhasnotmadeacleardistinctionbetweenpraiseandfeedback.Withoutsuchadistinction,theeffectsofpraiseonintrinsicmotivationpartiallymaybedueto
correctivefeedback,thusbiasingthemetaanalyticfindingsinapositivedirection.Animplicationisthatthelaboratoryresultsonpraiseandintrinsicmotivationshould
beviewedwithcaution.
Althoughpraisecanfunctionasreinforcement(seethenextsection)forspecificbehavior,appliededucationalresearchindicatesthatitisseldomusedinthiswayin
classroomsettings.7Studiessuggestthat,inorderforpraisetoserveaseffectivereinforcementintheclassroom,teacherswouldhavetomoveabout,praisingstudents
performanceaboutonceevery15seconds,or200timesin50minutesaratethatisincompatiblewithtraditionalteachingpractices(e.g.,sittingatthefrontofthe
room).8
Teachersoftenusedeliberateorplannedpraiseasencouragement(incentives)tomotivatehigherlevelsofperformancefromstudentswithlimitedability.Deliberate
praisebyteachersistypicallybasedonsignsofeffortratherthanaccomplishments.Becauseofthediscrepancybetweeneffortandaccomplishment,deliberatepraise
cansometimeslackcredibilitywithstudents,leadingtoembarrassmentordiscouragement.Incontrast,spontaneouspraisebasedonsurpriseoradmirationofstudents
accomplishmentsusuallyhashighcredibilityandenhancesstudentsacademicmotivation.
Unfortunately,theappliededucationalfindingssuggestthatplannedpraiseislesseffectiveformotivationthanunplannedpraise,makingthedeliberateuseofpraise
formotivationlessviableinclassroomsettings(andperhapsinbusinessandindustry).Also,evenspontaneouspraisecanleadtonegativesideeffects,aswhenthe
emphasisofpraiseisonsurprisemorethanadmiration(Boy,Ineverthoughtyoucoulddothat!).Thegeneralguidelineisthatpraisecan(atleastmomentarily)
enhanceperformanceandinterestbutthatthisformofencouragementcanbackfire,asincaseswherepraiselackscredibilityorisoverused.
Page207

EFFECTIVEUSEOFREINFORCEMENTINEVERYDAYLIFE
Theuseofrewardsasincentivesforbehaviorhasitsplace.Incentivescanbeusedtoguideandmotivatebehaviorintheshortrun,especiallywhenactivitieslackinitial
interest.Inthelongrun,however,incentivesmustbebackedbybehavioralconsequencesorelseperformanceandinterestinanactivitywillreturntopreviouslevels
(beforetheofferedreward).Theeffectiveuseofrewardsasbehavioralconsequencesinvolvestheprincipleofreinforcement.Inthissection,weoutlinethebasicidea
ofreinforcementandsuggestguidelinesforitssuccessfuluseineverydaylife.
Reinforcementoccurswhenapersonactsinawaythatproduceseffectsorconsequences.Someofthesebehavioralconsequencesfeedbackonthepersons
behavior,strengtheningorincreasingtheactionthatproducedthem.Wemayspeakofreinforcementonlyifaspecificbehaviorisstrengthenedbyaparticular
behavioralconsequence.Inaworksetting,anemployeefilesrecordsandobservesacountofhowmanyrecordshavebeenfiledatagivenmoment.Onepossibilityis
thatobservingthecountservesasreinforcementforfilingrecordsbytheemployee.Thatis,therecordfilingbehaviorisstrengthenedandmaintainedbyitseffects
(observingthecount).
Manyofthebehavioralconsequencesthataffectbehaviorarecalledrewardsthus,rewardscanbeusedasconsequencesofbehaviortoincreaseitsfrequencyof
occurrence.Noticethattheuseofrewardsasreinforcementisdifferentfromtheuseofrewardsasincentivesforaction.Rewardsasincentivescomebeforeaction,
whereasrewardsasreinforcementcomeafterthebehaviorhasoccurred.Ourreviewoftheliteratureonrewardsandintrinsicmotivationidentifiedahandfulofstudies
thatinvolvedtheuseofrepeatedrewardsfollowingtheoccurrenceofatargetbehavior.Intheseexperiments,thetargetbehaviorincreasedduringtherewardphaseso
thattherewardsfunctionedasreinforcement.Undertheseconditions,rewardseitherenhancedordidnotaffectmeasuresofintrinsicmotivation.Asageneralguideline,
rewardsthatoperateasreinforcementforanactivitycanbeusedtoincreaseatargetbehaviorwithoutalossofintrinsicmotivation.
Althoughrewardscanbeusedaseffectivebehavioralconsequences,mostpeoplearenotfamiliarwiththemannerinwhichtoimplementasuccessfulreinforcement
systeminthehome,classroom,orworksetting.Wearenotabletooutlinealltheinsandoutsofreinforcementprinciplesinthisbook9weare,however,abletogive
someguidelinesandreferinterestedreaderstotheliteratureonreinforcementinappliedsettings.10

AspectsofRewardandReinforcement

Positivereinforcementinvolvespresentingarewardorbehavioralconsequencewhenthepersongivesthetargetresponse.11Reinforcementisalsoa
Page208

nameforaprocesstheincreaseinanactionthatresultsfromreinforcingconsequences.Reinforcementdescribesarelationshipbetweenbehaviorandits
consequences,whichhasthreecomponents:(1)thebehaviormustproduceconsequences,(2)thebehaviormustincreasemorewiththeconsequencespresentthan
whentheyareabsent,and(3)theincreaseinbehaviormustoccurbecauseoftheconsequencesandnotforsomeotherreason.12
Theeffectsofbehavioralconsequencescanbeclassifiedasunconditionedandconditionedreinforcement.Withunconditionedreinforcement,thereinforcingeffects
ofbehavioralconsequencesdonotdependonotherreinforcementrelations.Alargeamountofhumanbehaviorisstrengthenedbyfood,water,sexualcontact,and
otherbiologicallysignificantformsofreinforcement.Sensorystimulationsuchasflashinglightscanalsofunctionasunconditionedreinforcementeventhoughthereisno
obviousbiologicalimportance.Otherbehavioralconsequencesstrengthenbehaviorbecauseofarelationshiptoothersourcesofreinforcement.Inaworksetting,a
patonthebackbythebosscanfunctionasconditionedreinforcementfordetectingaflawedcomponentbyaworkerthereinforcementeffectivenessofthebosss
responsepresumablydependsonitsassociationwithotherkindsofreinforcement(includingmoney,whichisexchangeablefornumerousgoodsandservices).
Rewardssuchasattention,approval,andrecognitionactasconditionedreinforcementwhenthesesocialconsequenceshavebeenassociatedwithothersourcesof
reinforcement.
Manybehavioralconsequencesthatincreasehumanbehaviorarenotcalledrewardsineverydaylanguage.Forexample,fallingfromhighplacesorbeingtossed,
twisted,andshakenhardlyseemlikerewardsyettheseconsequencesofamusementparkrideskeepuscomingback.Thisexampleshowsthatthereinforcingeffects
ofbehavioralconsequencesoftendependonthesituationorcontext.Theeffectivenessofabehavioralconsequencealsocandependonthebehaviorthatproducesit.
Laughtercanfunctionasreinforcementfortellingjokesbutnotfortellingpuns.Groaningcanbeusedasreinforcementforpunsbutnotfortellingjokes.Generally,the
reinforcingeffectsofbehavioralconsequencesarerelativeratherthanabsolute.
Therelativenatureofreinforcementisobservedwhentheopportunityforoneactivityisusedasreinforcementforanotheractivity.Aparentwhomakesplayingwith
friendscontingentonpracticingthepianoislikelytoseeanincreaseinpianopractice.Thebasicideaisthatahigherfrequencybehavior(playingwithfriends)will
reinforcealowerfrequencybehavior.13Formostchildren,playingwithfriendshasahigherprobabilityofoccurrencethanplayingthepianoplayingwithfriends
thereforewillreinforcepianopractice.
Ontheotherhand,takethecaseofMary,anaccomplishedpianist.ForMary,playingthepianohasahigherprobabilitythanbeingwithherfriends.Inthisinstance,
beingwithfriendswillnotreinforceplayingthepiano.Infact,forMary,pianoplayingwillreinforcebeingwithherfriends.Inaclassroom,paintinganddrawing(art)are
usuallyhigherprobabilityactivitiesthansolvingmathematicalproblems.Thus,teacherscanusepaintinganddrawingtoreinforce
Page209

mathematicalproblemsolving.Incontrast,foraparticularchild,Stephanie,whoexcelsatmathematics,paintinganddrawingmaybeatalowerfrequencythansolving
mathematicalproblems.InStepaniescase,paintinganddrawingwillnotreinforcefindingmathematicalsolutions.Practicingmathematics,however,willreinforce
Stephaniesartisticbehaviororanotherlowerprobabilityactivity,perhapsplayingsports.Noticethattherelativityofreinforcementmeansthatareinforcement
contingencyhastobetailoredtotheperson,situation,andbehavior.
Theuseofreinforcementisnotastraightforwardextensionofgivingpeoplethingsthatwecallrewards.Rewardsareusuallythingsthataresatisfyingandthat
peoplelike.Evenso,thesequalitiesdonotguaranteethatanyrewardorbehavioralconsequencewillreinforceaparticularbehavior.Althoughtheuseof
reinforcementinvolvesmanysubtleties,thereareseveralguidelinesthatcanmakereinforcementmoreeffectiveinappliedsettings.

RulesforEffectiveReinforcement

Rule1:SpecifytheTargetBehavior
Recallthatreinforcementinvolvesanincreaseinaspecifictargetbehavior.Behavioriswhatapersondoes.Atargetbehaviorshouldbeobservable,countable,and
important(anactionthat,ifincreasedinfrequency,willmakeapracticaldifferenceintheeverydaylifeofthestudent,worker,orclient).Mostparents,teachers,and
managersusegeneralizedandvaguetermstodescribetheperformanceofchildren,studentsandemployees.Astudent(orworker)shoulddevelopanappreciation
ofthewrittenmaterial,orshouldshowanunderstandingoftheassignedtask.
Suchgeneralizedtermsmustbemadeconcretebygivingexamplesofbehaviorsthatareobservedwhenwesaythatastudent(orworker)appreciatesor
understands.Inaclassroom,astudentsbehaviorisbestexpressedbyactionverbsthataredirectlyobservable.Forexample,aprecisebehavioraldescriptioncould
involvethefollowingstatements:
Thestudentswill:
writesentencesintheirnotebooksusingtheactivevoice
countoutloudtheequivalentinquarters
pointtothepicturethatisthesameastheworddog
Ausefulguidetoselectingobservableactionverbsisavailabletoteachers.14Table11.1showstheclassificationofactionverbsasobservable,ambiguous,ornot
observablebasedonagreementamongindependentclassroomobservers.Theclassificationhelpspinpointbehaviorinschoolsettingsandisinformativeforanyone
concernedwithbehaviorchange.
Inbusinessandindustry,weusuallytalkaboutperformance.AubryDaniels
Page210

Table11.1
ExampleofActionVerbsThatDifferinObservability
CLASSIFICATIONOFACTIONVERBS
DirectlyObservable Ambiguous NotDirectlyObservable
tomark touse todistinguish
tounderline toarrange torecognize
towrite todemonstrate tothinkcritically
tofillin tochoose toappreciate
toremove tosee todiscover
topointto toutilize toknow
toputon toconstruct tounderstand
tocrossout tocomplete toperceive
tocircle toidentify tobecurious
toreadorally toperform tolearn
Source:BasedonverbsfromDenoandJenkins(1967)reprintedinAlbertoandTroutman(1999,p.69).

suggestedthatwethinkofperformanceasasequenceofbehaviorsdirectedatsomeoutcomeorgoal.15Forexample,theperformanceofafilingclerkcouldinvolvea
sequenceofbehaviorssuchasopeningthemail,sortingthecorrespondence,andfilingitintoappropriatefolders.Thegoalorobjectivemightbetohavedaily
correspondencefiledsothatorderdeskclerkscaneasilyfindlettersandmemostoservicecustomers.Inanotherworksetting,performancemayconsistofthe
behaviorsofpickingupasheetofplastic,placingitonamachine,pressingabutton,removingthemoldedpart,andstackingitneatlyonapile.16Theobjectivecould
betoproduceagivennumberofmoldsinacertaintimeandwithincertainspecificationsoftolerance(i.e.,quality).
Inordertocreateachangeinperformance,amanagerneedstoanalyzethecomponentbehaviors.Forexample,allbehavioralcomponentsforproducingmolds
mightbepreciselycarriedouteachtimeexceptforthefinalbehaviorofstackingneatly.Thelackofprecisioninthisbehavioralcomponentmaycausebreakage,thus
reducingthenumberandqualityofmoldsmadeinagiventime.Inthisexample,behaviorchangewouldbetargetedatstackingneatlybehaviorratherthansomeother
componentofthebehavioralsequence.Asanaside,onereasonthatanemployeemightnotshowtherequisitebehavioristhatthetermneatlyisnotclearlydefined.
Overall,thepointisthatachangeinperformancerequiresachangeinoneormorecomponentbehaviors.
Intermsoftargetingbehavior,therewardsandintrinsicmotivationliteratureisproblematic.Mostofthestudiesinvolvevaguedefinitionsofbehaviororperformance,
anditispossiblethatanylossofinterestormotivationisduetotheimprecisespecificationofthetargetbehavior.Thestudiesusingrepeatedrewardfoundnolossof
intrinsicmotivation,butthetargetbehaviorwasalsomoreclearlydescribedandcounted.Furtherresearchisnecessarytoexamine
Page211

whetherthedefinitionofthetargetbehaviorplaysaroleinreducingmeasuresofintrinsicmotivation.

Rule2:ArrangeaFavorableSituation
AnimportantaspectoftraininganimalsinthelaboratoryistheSkinnerbox,oroperantchamber.Thechamberisdesignedtoaccommodateanimals(suchasratsor
pigeons).Thespaceinthechamberissmallsothattherangeofbehaviorsisrestricted,andthechamberisdesignedtosignalthetargetresponse(e.g.,pressingalever)
toprovideimmediatereinforcement(foodpellets)whenthebehavioroccurs.Ingeneralterms,theoperantchamberstructuresthesituationsothatthedesiredbehavior
willoccurandincompatiblebehaviorisreduced.Thus,leverpressingislikelyandexploringofthechamberisminimized.17
Thenecessityofprovidingafavorablesituationforteaching,training,andperformancehasnotgoneunnoticed.Inahome,thekitchenisafavorablesituationfor
foodpreparationandcooking.Schoolshaveclassroomsthataresetuptosupportacademicbehaviorandreduceincompatiblebehaviorsuchastalkingtoclassmates
andwanderingaroundtheroom.Aworksettingsuchasafactoryisdesignedtoenableanorderlysequenceofactionsthatresultinspecificmanufacturedunitsor
products.Althoughthereisanaveappreciationofthefavorablesituationforbehavior,manyoftheproblemsofwork,school,orhomecouldrelatetothetrainingor
performancesituation.
Forexample,aclassroomisnotthesamesituationfromoneschooltoanotherorevenwithinaparticularschool.Therearetraditionalclassroomswithdesks
arrangedinrowsandcolumnsandasupervisorydeskatthefrontoftheroom.Usually,thereareblackboardstowriteonandbulletinboardsforpostingitems.Asmall
partoftheclassroommaycontainbookshelvesandbooksandperhapsareadingarea.Incontrast,therearelessstructuredclassroomswithmovablechairsandtables,
materialsforprojectsanddiscoverythroughouttheroom,andateacherwhointeractswiththechildreninalowsupervisorymanor(e.g.,seatedonthefloorwith
studentsgatheredaround).Ratherthandebatethemeritsofdifferenttypesofclassroomsintermsofideology,abetterstrategyistoaskwhatbehaviorsaretobe
taught.Theidealsettingforteachingscientificbehaviorswouldlikelybedifferentfromasituationthatisbestarrangedtoteachreadingandwritingskills.Thepointis
thataflexibleclassroomthatallowsateachertoarrangeandrearrangethesettingintermsofthetargetbehaviorwillbeonethatpromotesthedesiredacademicor
socialbehaviorswhilereducingactionsthatareincompatiblewithlearning.
Inbusinessandindustry,thearrangementofthefactory,warehouse,orofficemaynotstructurethetargetperformance.Thatis,aspectsofthesituationmayinterfere
withthesequenceofbehaviors(oraparticularbehavior),therebyreducingperformanceandproductivity.Employersandmanagersareusedtotalkingaboutthedesign
ofmaterialsandmachinery,butthedesignofthework
Page212

settingasameanstopromotetargetbehaviorsanddecreaseincompatibleactivitiesisnotwellrecognized.

Rule3:SelectEffectiveRewardsandBehavioralConsequences
Onceyouhaveidentifiedthetargetbehaviorandestablishedafavorablesituation,thenextstepistoselecteffectiverewardsandbehavioralconsequences.Behavioral
consequencescanincreaseordecreasethetargetbehavior,butinthisbookwewilldealonlywithpositivereinforcement.Positivereinforcementinvolves(1)the
presentationofrewardsandotherbehavioralconsequencesfollowingthetargetbehaviorand(2)anincreaseinfrequencyofthetargetbehavior.Therefore,beforeone
canusepositivereinforcementitisnecessarytoselectoneormoreitems,events,oractivities(oranyotherthing)thatservestoreinforcethetargetbehavior.
Mostconsequencesofbehaviorinvolveconditionedreinforcement(ratherthanprimaryreinforcement).Thatis,thethings,activities,andbehaviorofothers(social
consequences)thataffectourbehavioraremostlybasedonanassociationwithothersourcesofreinforcement.Therewardsandotherbehavioralconsequencesthat
enterintopositivereinforcementareuniquetoeachindividual.Fortunately,somereinforcingconsequencesareeffectiveformanypeoplehowever,thelargerthegroup
ororganization,thelessistheprobabilitythatanyparticularrewardorconsequencewillserveasreinforcementforeveryone.How,then,canteachersandmanagers
findtheeffectiverewardsandconsequencesfortheindividualsandgroupswithwhichtheywork?
Inanattempttoidentifyeffectiverewardsandbehavioralconsequences,AubreyDanielssuggestedthatbehaviormanagers18(1)thinkintermsofwhatthepersonor
groupwants,desires,orvalues(2)askpeoplewhattheylikeand(3)observewhatpeopledoandwhattheygetoutofit.19Intermsofwants,desires,andvalues,itis
likelythatpeoplewhoshareacommonsocialbackgroundwillupholdsimilarvaluesandenjoysimilarrewardsandbehavioralconsequences.Evenso,theruleisthatif
somethingdoesnotincreasebehavior,therewardorconsequenceisnotreinforcing.Danielsexplainsitinthisway:

Manyplantmanagershaveofferedtotakeatopperformerandspousetodinneratafancyrestaurantwhenthatwasthelastthingthatparticularcouple
wanted.Theymaywellhavepreferredbeerandpizzaordinneralone!Ofcourse,theoppositemistakeisalsomadewhenweassumethatbecauseof
certainbackgroundfactorstheywouldpreferbeerandpizzawheninfacttheywanttogotoafancyrestaurant.20

Thepointisthatananalysisofculturalvaluescanbehelpfulforspottingpotentialreinforcingconsequences,butitisnotaguaranteethatthespecifiedrewardswillbe
reinforcingforaparticularindividualorgroup.
Anotherwaytoidentifypotentialreinforcingrewardsandbehavioralconsequencesissimplytoaskpeoplewhattheylike.Explaintotheworkersoremployeesthat
youarehopingtoimplementapositiverewardsystemandneed
Page213

theirhelp.Talktothemaboutthekindsofleisuretimeactivitiestheyliketodo,suchasstayingatacottage,attendingthemoviesortheater,playingsports,fishing,
goingtoaconcert,andsoon.Inaddition,spendtimetalkingtopeopleabouttheirfamily,friends,andothersocialrelationships.Thegeneralpointistogettoknow
peoplesintereststhroughcasualconversation.
Acasualdiscussionoflikesandinterestscanbesupplementedbyareinforcementsurvey.Thesurveyisanotherwayofgettingatpotentialreinforcingconsequences.
Again,explaintothepersonwhatyouwanttodoandwhyyouwanttodoit,beforeconductingasurvey.Theideaistoinvolvepeopleinthedesignoftheirown
positiverewardsystem.Asafirststep,managersinbusinesscouldsimplyhaveemployeeslistthethingstheylikeandwant.Theproblemsarethatpeopleoftendonot
knowwhattheywantorlike,willnottellyou,oraskforthingsthatyoucannotdeliver.Thus,itissometimesusefultouseastructuredquestionnaireforareinforcement
survey.
AsshowninTable11.2,thereinforcementquestionnairelistsactivitiesandthingsthatpeopleoftensaytheylike.Eachitemismeasuredonafivepointscaleof
liking,fromnotatalltoverymuch.Thesurveycancovereverydaysocial,tangible,andactivityrewards,aswellasrewardsthatareparticulartotheworksetting.
Therearemanyintricaciesaboutcollectingandusingreinforcementsurveysthatarebeyondthescopeofthisbookbutareaddressedintheperformancemanagement
literature.21Overall,ifusedwithsensitivityandcare,areinforcementsurveycanbehelpfulinidentifyingrewardsthatcanserveasreinforcingconsequences.
Potentialreinforcingconsequencescansometimesbespottedbydirectobservationratherthantalksandsurveys.Payattentiontowhatpeopledoandtalkabout.
Forexample,workersinaplantmaytalkalotaboutupcomingbreaksorrestperiods.Theyalsomaytakelongerbreaksthanisacceptabletomanagement.One
assumptionisthathavingabreakwillfunctionasreinforcementforcompletingaproductionquotaontime(e.g.,100unitsperhour).Aplantsupervisormighttestthis
contingencyandfindanincreaseinproductivity,indicatingthatworkbreaksarereinforcingconsequences.
Althoughareinforcementcontingencybetweenunitsperhourandopportunitiesforbreakscouldincreaseproduction,bossesmaycomplainaboutemployeessitting
aroundduringworktime(employeessittingaroundisnotusuallyreinforcingforbosses).Thepointisthatidentifyingreinforcingconsequences(bytalk,survey,and
observation)doesnotresolveallproblemsofimplementingareinforcementsystem.Othersectorsofacompanyororganizationinitiallymayopposethereinforcement
system,22evenifitiseffectiveinchangingemployees(orstudents)performance.
Onewaytoensurethatyouhaveidentifiedreinforcingconsequencesistohedgeyourbetsbyusingareinforcementmenu.Afterspecifyingpotentialreinforcing
consequencesbytalk,survey,andobservation,makeupamenuofrewardsandbehavioralconsequencesthatarelikelytobereinforcing.Aswiththesurvey,the
reinforcementmenulistssocial,tangible,andactivityrewards.
Page214

Table11.2
ReinforcementSurveyExample
Pleaseindicateyourdegreeofliking(amountofpleasureyouobtain)foreachofthefollowingthings,events,oractivities.Circleanumberbetween0and4.
Item None ABit Some Much VeryMuch
1.ListeningtoMusic(specify)
Rock 0 1 2 3 4
R&B 0 1 2 3 4
Jazz 0 1 2 3 4
Reggae 0 1 2 3 4
Classical 0 1 2 3 4
Country 0 1 2 3 4
Opera 0 1 2 3 4
2.ReadingAbout(specify)
Romance 0 1 2 3 4
History 0 1 2 3 4
Mystery 0 1 2 3 4
ScienceFiction 0 1 2 3 4
SelfHelp 0 1 2 3 4
Sports 0 1 2 3 4
Adventure 0 1 2 3 4
Eachoftheseactivitiescouldbefurtherspecified(whatkindofconcert,sports,etc.)
3.Attendingconcerts 0 1 2 3 4
4.Watchingsports 0 1 2 3 4
5.Playingsports 0 1 2 3 4
6.Watchingmovies 0 1 2 3 4
7.Goingtoparties 0 1 2 3 4
8.Singing 0 1 2 3 4
9.Dancing 0 1 2 3 4
10.Painting 0 1 2 3 4
11.Cooking 0 1 2 3 4
12.Writing 0 1 2 3 4
13.PlayingCards 0 1 2 3 4
14.Playingmusicalinstrument 0 1 2 3 4
15.Shopping 0 1 2 3 4
16.Talkingabout(topic) 0 1 2 3 4
17.Beingrecognizedatameetingfor
Solvingaproblem 0 1 2 3 4
Exceedingjobrequirements 0 1 2 3 4
Helpingcustomers 0 1 2 3 4
Page215

18.Receivingfromyoursupervisororboss
Remarksonreport 0 1 2 3 4
Phonecallonworkdonewell 0 1 2 3 4
Afavoritebook 0 1 2 3 4
Privatetalkaboutyourproject 0 1 2 3 4
Ticketstoevents(specify) 0 1 2 3 4
AfavoritemusicalCD 0 1 2 3 4
Opportunitytogiveyouropiniononaproject 0 1 2 3 4
Dinneratagourmetrestaurant 0 1 2 3 4
Acouponforabook 0 1 2 3 4
Nameonbulletinboard 0 1 2 3 4
Atouchontheshoulderforworkwelldone 0 1 2 3 4
AcouponforamusicalCD 0 1 2 3 4
19.Otheractivities,things,orevents...

Theideaisthatthepersongetstochooseitemsfromthemenuwhenevertheperformanceexceedsapredeterminedstandard.Thereinforcementmenuensuresthata
varietyofbehavioralconsequenceswillbereinforcingatanymomentareinforcementmenualsoworksagainstsatiationbynotusingthesamerewardsovertime.

Rule4:SettheReinforcementContingency
Oncepotentialreinforcingconsequenceshavebeenidentified,itisusefultoconductatest.Doestherewardorbehavioralconsequenceactuallyincreasethetarget
behavior?Inordertoconductatest,thebehaviormanager(teacher,supervisor,boss,etc.)muststatethecontingencybetweenbehavioranditsconsequences.For
example,IfJohncompletestheassignedhomeworkproblems,thenhegetstochoosearewardfromapreestablishedmenu.Eachdayforaweek,thehomeworkis
given,andifitiscompleted,therewardsfollow.IfJohnshomeworkiscompletedmoreoftenwiththereinforcementcontingencyineffectthanwithoutit(baseline
period),theevidencesuggestspositivereinforcement.Atthispoint,thereinforcementsystemcanbeimplementedonafulltimebasis.
Theperformancecriterionbetweenthetargetbehavioranditsconsequencescanbeincreasedslowlyovertime.InthecaseofJohnandhishomeworkassignments,
thenumberofhomeworkproblemseachdaycouldbeincreasedinsmallstepssothatJohnisalwayssuccessful.Thus,theteachermayassignone
Page216

takehomeproblemeachdayinthefirstweek,twointhesecond,threeinthethird,andsoon.ThisprogressivecontingencycouldbeusedtomoveJohns
performancefromalowleveltoafinalperformancethatisacceptabletoJohn,hisparents,andtheteacher(e.g.,Johnwilldofivehomeworkproblemseachdayforat
leastfouroutoffiveweekdays).

Rule5:WaitfortheTargetBehaviortoOccurandThenReinforceIt
Thoseinvolvedinteachingormanagementareusedtotellingpeoplewhattodoratherthanwaitingforthemtodoit.Theproblemisthatreinforcementworksbest
whenthetargetbehavioroccurswithoutprompting,cajoling,ortelling.Forexample,inalaboratorysituation,aratistrainedtopressaleverbywaitingforaclose
approximationandfollowingtheresponsewithfoodreinforcement.Thereisnowaytotelltheanimalwhattodo,eitherverballyorbydemonstration.Followingthe
animalexample,thebestthingtodoinappliedsettingsistodiscussthereinforcementcontingencybeforeitisimplemented,ensurethatthepersoncanmeetthe
requirementsforreinforcement,waitforthetargetbehaviortooccur,andthenreinforceit.
InthecaseofJohnandhishomework,theteachershouldgivethehomeworkassignmentforthedayand,ifJohncompletesit,lethimpickarewardfromthemenu.
Noticethatwiththisprocedure,Johnisnottoldtodohishomeworkbytheteacherorremindedbyhisparentsoftherewardshecanearn.Rather,thereinforcement
contingencyisset,andifthetargetbehavioroccurs,theconsequencesfollow.Ifthetargetbehaviordoesnotoccurorhasalowprobabilityofoccurrence,thenthe
strategyistouseshaping(nottelling),reinforcingcloserandcloserapproximationstothefinalperformance(thebehavioralgoalorobjective).
Theuseofshaping,ratherthantelling,intheregulationofhumanbehaviorwasdiscussedbyKarenPryorinherbookDontShoottheDog!23Pryorindicated:

Ininformalsituationsinreallife,however,youareprobablybetteroffshapingwithoutinstructionsorverbaldiscussion.Supposeyouhaveamessy
roommatewholeavesdirtyclothesallovertheplace....Youwouldofcoursedrawupashapingplan,theinitialandintermediatestepsbywhichyou
wouldreachthedesiredgoal.Togetdirtyclothesintothehampereverytime,forexample,youmightstartwithonesock,andtargetthebehaviorby
holdingouttheopenhamperjustasthesockisabouttogoonthefloor.Thereinforcercanbeverbal,tactile,orwhateveryouthinktheroommatewould
belikelytorespondtooraccept....Thereare,however,twotrapsinthisuseofshaping.Thefirstisthatitiseasiertonoticemistakesthantonotice
improvement,so,verbalcreaturesthatweare,itismucheasierforustoremonstratewhenthecriteriaarenotmetthantoreinforcewhentheyare.And
thatcanundotheprogress.Thesecondtrapisthatifyouarecalculatingtoshapesomeonesbehavior,itisverytemptingtotalkaboutit.Andtalking
aboutitcanruinit.Ifyousay,Iamgoingtoreinforceyouforputtingyour
Page217

laundryinthehamper,fornotsmokingmarijuana,forspendingless,orwhateveryouarebribingorpromising,notactuallyreinforcingonlearningof
yourplans,thepersonmayrebel,instantly,andescalatethemisbehavior.Togetresultsyouhavetodoshaping,nottalkaboutit.24

Pryorsdistinctionbetweentalkingaboutrewardsanddeliveringeffectiveconsequencesiswhatthisbookisallabout.Therewardsandintrinsicmotivationliterature
makesitclearthattalkingaboutrewardscanbackfirebyreducingthetargetbehaviorandsubvertinginterestinanactivity.Incontrast,reinforcementbuildsand
maintainsperformancewhileatthesametimegeneratingmotivationandinterest.

Rule6:MovefromContinuoustoIntermittentReinforcement
Togetbehaviorgoing,itisnecessarythatreinforcementbealmostcontinuous.Continuousreinforcementhastwopropertiesthatarecriticaltoestablishing
performance:(1)reinforcementisimmediateand(2)reinforcementiscertain.Thecertaintyofreinforcementconcernshowlikelyitisthatreinforcementfollowsthe
targetbehaviortheimmediacyofreinforcementinvolveshowquicklythereinforcingconsequencesaredelivered.
Themorecloselythetargetbehaviorisfollowedbythereinforcingconsequence,themorelikelyitistobeeffective.Delaysbetweenthetargetbehaviorand
reinforcementallowtimeforotherbehaviortooccur,resultinginaccidentalreinforcementoftheunwantedbehavior.Inhisrecentbookonappliedbehavioranalysis,
PaulChancediscussedimmediatereinforcementinthefollowingexample:

Supposeyouwouldliketoincreaseyour10yearolddaughterstendencytoread.Youknowfrompastexperiencethatgivinghertheopportunitytoread
aloudtoyouisreinforcing.Afterdinner,whileyouarewashingdishes,younoticethatinsteadofwatchingTV,yourdaughterhaspickedupabookand
begunreading.Youdecidethattoreinforcethisbehavior,youwillaskhertoreadtoyouassoonasyoufinishthedishes.Unfortunately,5minuteslater
whenyouareready,sheisnolongerreadingsheisnowwatchingTV.Oddasitmayseem,ifyouaskhertostopwatchingtelevisionandreadtoyou,
youmightactuallyreinforceTVviewing.25

Toavoidthedelayproblem,Chancerecommendsthatbehaviormanagersprovidereinforcementwhilethebehaviorisinprogress,therebyensuringimmediate
reinforcement.
Theotherpropertyofcontinuousreinforcementiscertainty.Thegreateristhelikelihoodofreinforcementforthetargetbehavior,themorerapidlythebehaviorwill
increase.Withcontinuousreinforcement,everytimethetargetbehavioroccurs,itisfollowedbythereinforcingconsequence.Also,thereinforcingconsequencealmost
neverfollowsanyotherbehavior.Thus,theprobabilityofreinforcementforthetargetbehaviorisessentiallyone(certain),and
Page218

itiszeroforanyotherbehavior.Chancenotedthatinhumanbehavior,unfortunately,reinforcementisoftennomorecertaintofollowdesiredbehaviorthanatother
times:

Parentsandteachersoftenignorechildrenwhentheybehavewell,forexamplebutletthosechildrenwhineorcryorbedestructive,andsuddenlythey
havetheadultsundividedattention.Ifattentionisreinforcing,theseparentsandteachersaremakingreinforcementlikelywhenundesirablebehavior
occursandunlikelywhendesirablebehavioroccurs.Strangeasitmayseem,some[professionals]actuallyadvocatemakingreinforcersmorelikelywhen
childrenmisbehaveoractcrazythanatothertimes.26

Theoverallpointisthatalmostcontinuousreinforcementisnecessarytoestablishatargetbehaviorathighstrength.Thismeansthatreinforcementmustbeimmediate
andcertainforeachinstanceofthetargetbehaviorandmustalmostneverfollowundesiredbehavior.
Onceaperformancehasbeenwellestablished,theruleistochangetheschedulefromcontinuoustointermittentreinforcement.Amajoradvantageofintermittent
reinforcementisthatbehaviorismaintainedinstrengthbetterwithintermittentthanwithcontinuousreinforcement.Forexample,takethebehaviorofgamblingonthe
slotsinLosVegas.Gamblingbehaviorinvolvesplacingmoneyinaslotmachineandpressingabuttonorpullingahandletoactivateaspinordraw.Thereinforcing
consequencesarethepayoffsfollowingthegamble.Anoviceisonlylikelytorepeatedlygambleiftheearlyhistoryofreinforcementismostoftensuccessful(almost
continuousreinforcement).Oncetheperformancehasbeenwellestablished,however,itistheintermittentpayoffsthatkeepthegamblercomingback.Thatis,slot
machinesaredesignedtooccasionallyreinforcegamblingusuallytheschedulescanbequiteextendedsothatmanygamblesoccurbeforethepersonhitsapayoff.
Thechangesintheschedulesofreinforcementfromaninitialhighrateofsuccesstoaleanrateofintermittentpayoffshelptoexplainthemaintenanceofgambling
behavior.
Inclassroomsorworkplaces,intermittentreinforcementaccountsforthecontinuedperformanceofstudentsoremployeeswithlimitedsupervision.Daniels
elaboratedonthispoint:

Salespeopleworkingawayfromtheofficeforextendedperiodsoftime,longdistancetruckdrivers,servicetechnicians,securityguards,nightshift
employees,auditors,entrepreneurs,andconsultantsareafewofthejobsthatmustbeestablishedonintermittentreinforcementforbestperformance.An
additionaladvantageofintermittentschedulesisthattheyallowamanagertobeawayfromtheofficeforalongtimeandnothavedecreasesin
performanceduetolackofreinforcement.27

Noticethatpeoplewhoarecalledselfmotivated,innerdriven,orselfstartersarethosewithahistoryofintermittentreinforcementforperformance.Ultimately,self
motivatedperformancedependsonatleastoccasionalreinforcementwhen
Page219

allsourcesofreinforcementarewithdrawn,evenselfmotivatedpeopledeclineinperformance.Asarule,teachersormanagersshouldensurethatoccasional
reinforcementisarrangedforperformanceevenforstudentsoremployeeswhoseemtokeepgoingwithoutreinforcement.
Intermittentschedulesaretypicallyclassifiedaseitherratioorinterval.Ratioschedulesofreinforcementarebasedontheamountofworkaccomplishedorthe
numberofresponsesrequiredforreinforcement.Intervalschedulesarebasedonhowmuchtimemustpassbeforethedesignatedbehaviorisreinforced.Thereason
whytheseschedulesareimportantisthateachkindproducesadistinctivepatternofperformance.
Behaviormaybereinforcedonlyafteranintervaloftimehaspassed.Afixedintervalschedulestipulatesthatthefirstresponsefollowingaspecifiedintervalis
reinforced.Lookingforabusisbehaviorthatisreinforcedafterafixedtimesetbythebusschedule.Ifyoujustmissedabus,theprobabilityoflookingforthenextone
isquitelow.Astimepasses,therateofresponseincreaseswiththehighestrateoccurringjustbeforethebusarrives.Thus,therateofresponseisinitiallyzerobut
graduallyrisestoapeakatthemomentofreinforcement.Thisresponsepatterniscalledscallopingandischaracteristicoffixedintervalreinforcement.
Ascallopingpatternofbehaviorwouldbeexpectedifateacherinspectedthestudentsmathhomeworkattheendofeachweek.OnMondaythroughWednesday,
studentswoulddofewproblems,butastheendoftheweekapproachedandinspectionbecameimminent,moreandmoreofthehomeworkwouldbe
accomplishedtobereadytosubmitforinspection(reinforcement)onFridaymorning.
Stipulatingavariableintervalscheduleofreinforcementcaneliminatescalloping.Inthiscase,thefirstresponseafteravariableamountoftimeisreinforced.A
teachercouldcheckhomeworkaftervariableamountsoftimehaveelapsedthenextday,afterthreedays,attheendoftheweek,andsoon.Withvariableinterval
reinforcement,behaviorcanbereinforcedatanytime,evenifitwasreinforcedjustashortwhileago.Dependingontheworkloadoftheteacher,shecouldinspect
homeworktwodaysinarowandthenwaitaweektodothenextinspection.Thevariableintervalscheduleproducessteadyandregularresponding.Studentswilldo
homeworkatastable,moderateratethroughouttheweekifthescheduleofteacherinspectionisvariableinterval.
Incontrasttointervalschedulesthatreinforcebehaviorafterthepassageoftime,ratioschedulesreinforceonthebasisofthefrequencyornumberofresponses.
Withratioschedules,thefasteroneaccomplishestheperformancerequirement,themorereinforcementoneobtainstheharderyouwork,themoreyouget!Thus,a
workerwhoispaidforeachfinishedOringwillproducemoreunitsthanaworkerwhoispaidbythehour.Withafixedratioschedule,reinforcementoccursaftera
fixednumberofresponses.Inafactory,piecerateofpaymentareexamplesoffixedratioschedules.Thus,aworkermayreceive$1
Page220

forsewing100ofelasticwristband.Whentheratioofresponsestoreinforcementishigh,fixedratioschedulesproducelongpostreinforcementpausesfollowedbya
burstofactivityknownasbreakandrunperformance.Withoutknowledgeofthisfixedratioeffect,managersoftencomplainaboutslackingoffbytheworkers,
especiallyifinspectionsoccurduringabreakperiod.Danielssuggestedthatmanagersviewthepostreinforcementpauseasatimetocelebrateratherthanawasteof
timeawayfromwork.Organizationsthattaketimetocelebrateaccomplishmentsorvictoriesaretypicallyhighlyproductivebecausepeopleworkhardertogetthe
celebrationandworkharderfollowingthecelebrationthantheywouldhavedonewithoutit.28
Ratioschedulesofreinforcementcanalsobeprogrammedonavariablebasis.Onavariableratioschedule,reinforcementisdeliveredafteravariableamountof
workorbehavior.Peopleneverknowwhenreinforcementwilloccuritcomesasasurprise.Thebestknownexampleofvariableratioreinforcementistheslot
machine.Thenumberoftimesyouhavetoputyourmoneyinandpullthehandlebeforeapayoffisvaried(asistheamountofwinnings).Onthiskindofschedule,there
islittlewaitingbetweenbetsbecausewinningonagivenpulldoesnotmeanthatyoucannotwinonthenextpull.Thesooneryoubetandpullthehandle,thesooner
youwilldiscoverifyouareawinner.
Variableratioscheduleseliminatethelongpausesfoundonfixedratioreinforcementbecausethenextresponsecanalwaysbereinforced.Variableratio
reinforcementthereforeproducesverysteadylevelsofperformance.Also,researchshowsthatvariableratioschedulesproducethehighestresponseratesofanyof
thefourbasicschedules.Theaverageratio(responsestoreinforcement)canbeslowlyincreasedsothathighlevelsofperformancearemaintainedbyveryinfrequent
reinforcement.Inthelaboratory,apigeonpeckingakeyforfoodmayemitmanythousandsofresponsesbeforereceivingreinforcement.Overall,thepatternof
behaviorproducedbyvariableratioschedulesishighandsteadyreferringtoahighandsteadyresponserate.
Ineverydaylife,variableratioschedulesmayexplainbehavioroftenattributedtoindividualpersistence.Ascientistmaytestthousandsofchemicalcompounds
beforefindingonethatworksandadoortodoorsalesmanmayknockonhundredsofdoorsbeforemakingasale.Also,variableratioschedulesprobablyare
involvedinbehaviorthatissaidtobeobsessive.Wecallpeoplezealots,workaholics,orneuroticwhentheypursueonethingwithoutlettingup.Thesepeoplework
incessantlybecauseexperiencehastaughtthemthatjustaroundthecorneritwillpayoff.29
Asfarasthepracticaluseofvariableratioreinforcementisconcerned,oneofthecommonmistakesistoraisetheratioofresponsestoreinforcementtoorapidly.
Manypeoplewhogiveupinschoolorworkmaybethosewhowereplacedinsettingsinwhichreinforcementwastooinfrequentbeforetheyachievedahighand
steadylevelofperformance.Inbusiness,salestraineesare
Page221

oftensentintothefieldbeforetheybecomeproficient,withtheresultthattheamountofcallstosalesisextremelylowleadingtohighdropoutrates.Givingmore
reinforcementduringtrainingandsupplementaryreinforcementwhentraineesmoveintothefieldcansolvetheproblem.
Generally,behaviorisduetothescheduleofreinforcement.Whenpeoplearelearningnewbehavior,continuousreinforcementwillquicklyestablishahighlevelof
performance.Oncebehaviorisoccurringatsomelevelofproficiency,itisbesttochangetointermittentreinforcementinordertomaintaintheperformance.Ofthefour
basicintermittentschedules,variableratioschedulesproduceveryhighandsteadyratesofbehavioroneproblemisthatpeoplewillgiveupiftheratioofresponsesto
reinforcementistoolarge.Providingsupplementaryreinforcementduringtrainingandinthefieldcanalleviatetheproblemofratiostrainonvariableratioschedulesof
reinforcement.

Rule7:MonitortheResultsofYourRewardProgram
Ateacherormanagercannotassumethattherewardprocedurewillhavetheintendedeffects.Justbecauseyouintendtoincreasethefrequencyofatargetbehavior,
thisdoesnotmeanthatyouwillgetthisresult.Theonlywayinwhichbehaviormanagerscanbesurethattheyareusingpositivereinforcementistolookattheeffects
oftherewardsonbehavior.Whatisneededisasystematicwayofmonitoringthetargetbehavior.Thatis,thebehaviormanagershouldbeabletomeasurebehavior
anddepictchangesinperformanceonagraph.Ifthedesiredbehaviorincreasesfollowingthepresentationofareward,youhavesucceededinreinforcingitifthereis
noincrease,youarenotprovidingreinforcement,nomatterwhatyourintention.
Oneofthemostusefulandleasttimeconsumingwaysofmonitoringperformanceissimplytocounteachinstanceofthetargetbehavior.Ateachercanuseagolf
counterwornonthewristtorecordeachoccurrence.30Performancelevelsarefoundbytallyingthecountoveraspecifiedperiodoftime.Forexample,inthe
classroom,ateachercouldcounteachtimeachildcompletesaproblem,answersaquestion,orwritesinanotebook.Thecountcanthenbetalliedoveraday.
Inbusiness,managersareofteninterestedintherateatwhichsomethingisdone.Therateissimplythenumberofoccurrencesinaspecifiedamountoftime.
Managersusuallyareconcernedwiththenumberofpartsassembledperhour,thelistingsperweek,orthecallscompletedeachday.Convertingcountstoratesallows
ustocomparedatawhenweareunabletostandardizetheperiodofobservationoropportunitiestorespond.Forexample,supposeTomisawayfromhisdesksix
timesina30minuteperiodofobservationandeighttimesinanotherobservationalintervalof40minutes.Inthissituation,therateinbothcasesis0.2perminute.
Othermeasuresofbehaviorareusefultoensurethatreinforcementiseffective.Whenbehaviorisnoteasilyseparatedintodiscreteresponses,durationmay
Page222

beausefulmeasure.Thedurationofbehaviorisameasureofhowlongapersondoesit.Thus,Sallyspent30minutestalkingtocustomersandJohnspentanaverage
of10minuteseachnightonhomework.Inanothercontext,thelatencyofbehaviormaybeausefulmeasure.Thelatencyisthelengthoftimebetweenaninstruction
andtheoccurrenceofbehavior.Forexample,theteachertoldSylviatositinherseatandittookher20secondstositdown.Thelatencyisusefulwhenabehavior
managerisconcernedabouthowlongittakesthestudentorworkertobeginsomethingorgetstarted.31
Itisnotalwayspossibletoobtaincount,duration,orlatencymeasuresofbehavior.Insuchcases,measurementbyjudgmentissometimesuseful.However,
judgmentsbysupervisorsareoftenarbitraryandbiased.Thatis,theevaluationofperformancevariesfromonedaytothenext.Onewaytoovercomethisproblemis
toestablishspecificperformancecriteriathatcanbeconsistentlyobservedbyseveralpeople.Systemsofjudginginsportslikegymnastics,figureskating,anddiving
illustratetheuseofjudgingperformancetodeterminewinners.
Inbusiness,subjectivityinjudgmentcanbeovercomebyconstructingabehaviorallyanchoredratingscale.32Thescaleisusuallyconstructedbyemployeesandtheir
bossandstipulatesthedegreeofcompletion(ornot)ofspecifiedbehaviorsortasks.Forexample,anemployeesperformancecouldbejudgedintermsofcarrying
outrequiredtests,identifyingproblems,andschedulingrepairs.Ascoreof1couldbeassignedtoaperformanceinwhichnoneofthesetasksareaccomplishedanda
scoreof10couldbeassignedwhenallofthesetasksarecompleted.Scoresbetween1and10areassignedonthebasisofthepercentageoftestscompleted,
problemsidentified,orrepairsscheduled.33Onceabehavioralratingscalehasbeenconstructed,contingentreinforcementcanbearrangedforthedesiredbehavioras
wellasforanassessmentofbehaviorchange.
Afinalissueinmonitoringtheeffectivenessofareinforcementprogramconcernssocialvalidation.34Inadditiontothedirectassessmentofbehaviorchange,itis
importanttoobtainanassessmentoffeelingsandsentimentsofthoseinvolvedin(orassociatedwith)theprogram.Iftheprocedures(objectivesandresults)arenot
acceptabletotheparentsorstudents,ateachersreinforcementprogrammaybackfirethestudentsmayavoidorescapeitandtheparentsmaycomplainaboutit.
Inabusinesscontext,employeesandbossesshouldbeconsultedbeforeandafterimplementingareinforcementsystem.Forexample,asupervisormayhave
identifiedtakingabreakasreinforcementforanemployeesjobrelatedbehaviorbutupperlevelmanagementmightcomplainabouttherewardprocedure(You
cantusebreaksasareward.)andsabotagetheprogram.Inanotherexample,areinforcementprogrammayincreasethetargetbehaviorbutthebehaviorchangesdo
notcorrelatewithdesiredresultsofthecorporation.Thatis,inasalesorganization,thenumberoflistingsandcallsmayincreasedueto
Page223

reinforcementbutweeklysalesmaystaystableordecline.Again,theinterventionmaynotbeacceptabletocorporatebosseseventhoughthetargetbehaviorhas
increased.Thegeneralpointisthatconsumersatisfactionisanimportantelementinareinforcementprogrambehaviormanagersshouldensurethatgoals,procedures,
andoutcomesarefavorabletothoseinvolvedin,oraffectedby,therewardsystem.

Rule8:ExperimentBeReadytoChangeandChangeAgain
Weindicatedatthebeginningofthischapterthatusingrewardsineverydaylifeisatrickybusiness.Thedictumtobereadytochangeandchangeagain,
acknowledgesthateventhebestlaidplansmaynotworkoratleast,notasyouexpected.Whenfacedwithevidencethatareinforcementproceduredoesnothave
thedesiredeffect,itistemptingtosaythattheprinciplesofreinforcementarewrongandabehavioralviewisincorrect.
Analternativetothedismissalofreinforcementprinciplesistotakeapragmaticviewoftheproblem.Perhapstheintendedreinforcingconsequencesdonotactually
reinforcethetargetbehavior.Inthiscase,oneanswermaybetofurtheridentifyitems,events,oractivitiesthatfunctionasreinforcementfortheparticularstudentor
employee.Orperhapsthereinforcingconsequencesaredelayedandotherbehaviorisstrengthenedratherthanthedesiredperformance.Switchingtoreinforcing
consequencesthatcanbedeliveredmoreimmediatelyorpresentingthereinforcementwhenthetargetbehaviorisinprogressmightovercomethisproblem.
Areinforcementproceduremayalsohavenegativesideeffects.Forexample,anincreaseinthefrequencyofthetargetbehaviormayinadvertentlyleadtoa
decreaseinsomeotherdesiredbehavior.Thedeclineinanotherbehaviormightoccurbecausetherelativerateofbehavior(orrelativetimespentonanactivity)isa
functionoftherelativerateofreinforcement.35Thatis,whenconsideringtwobehaviors(e.g.,playingmusicandreading)asreinforcementisaddedtoone(reading),the
relativerateofreinforcementfortheother(playingmusic)willdecline.Withadropintherelativerateofreinforcement,theotherdesiredbehavior(playingmusic)will
inadvertentlydecreaseanunintendedsideeffectofthereinforcementprogram.
Thegeneralpointisthatitisnotpossibletoforeseeeveryproblemthatcanarisefromareinforcementprogram.Butwhenfacedwithproblems,thereisnoreasonto
giveup.Carefulmonitoringofthetargetbehavior,possiblesideeffects,andsocialacceptancecanhelptopinpointproblemsandsuggestnewinterventionstrategies.
Basedonareanalysisofthereinforcementplan,newprocedurescanbeimplementedandtestedforeffectivenessintermsofbehaviorchange.Overall,itisbestto
maintainanexperimentalattituderegardingtheuseofrewardsineverydaylife.B.F.Skinnerencouragedustoregardnopracticeasimmutable.Changeandbeready
tochangeagain.Acceptnoeternalverity.Experiment.36
Page224

USINGREWARDSASREINFORCEMENT
Definethetargetbehaviorsothatitcanbeobservedandcounted.
Establishafavorablesituationwheretargetbehaviorislikelytooccur.
Selectappropriaterewardsofbehavioralconsequences(developareinforcementmenu).
Settherewardcontingency(e.g.,if...then...).
Waitforthetargetbehaviortooccurandthenrewardit.
Ensurethattherewardisinitiallyimmediateandreliablefollowingthetargetbehaviorthenprogramformaintenance.
Monitortheresultsofyourrewardprogram.
Bereadytochangeandchangeagainexperiment.

NOTES
1.PierceandEpling(1999,pp.225231)discussedthedistinctionbetweenrulegovernedbehaviorandcontingencyshapedbehavior.Anofferofrewardoranincentive,inthisview,is
aninstanceofrulegovernedbehavior.Analysisofrulessuchasinstructionsandadviceshowsthatcompliancewiththeruleisduetoahistoryofreinforcementforfollowingrules.
Peoplewithoutsuchahistoryarenotcompliantwithrulesandpeoplewhohavereceivedaversiveconsequencesforrulefollowingmayavoidbehavingastherulespecifies
(noncompliance).
2.Ourmetaanalysissupportstheuseoftangiblerewardstiedtoperformancestandards(evenwithhighinteresttasks).Incontrast,themetaanalysisbyDeci,Koestner,andRyan
(1999)indicatedthattheseperformancebasedcontingencieshavenegativeeffectsonmeasuresofintrinsicmotivation.Wehavesuggested,inthisbookandelsewhere(Cameron,
Banko,andPierce,2001),thatthemetaanalysisbyDeciandcolleaguescollapsedthedataforrewardprocedureswithdistincteffects,leadingthemtotheerroneousconclusionthat
allrewardcontingenciesnegativelyaffectintrinsicmotivation.
3.SeeBandura(1986).TheserulesfortheeffectiveuseofmasterybasedincentivesareextractedfromthetheoreticalwritingsofBanduraandtheresearchliteratureonsociallearning
concernedwithrewardandperformancestandards.ThereaderisreferredtoChaper5inthisbookforanoverviewofsociallearningtheory,rewards,andintrinsicmotivation.
4.TheexamplesofstrategiesusedbyHoneywellInc.,BlanchardTrainingandDevelopmentInc.,andAmericanExpressaretakenfromNelson(1996).
5.SeeBrophy(1981),whichmakesausefuldistinctionbetweenpraiseandcorrectivefeedbackthatisnotusuallymadeintheintrinsicmotivationandrewardsliterature.
Page225

6. Feedbackisveryimportantforbehaviorchangebutfeedbackalonewillnotsustainatargetbehavior(seeDaniels,1994,pp.100105).Behavioranalysisoffeedbackinindustry
showsthatfeedbackactsasanantecedentstimulus,tellingpeoplewhatbehaviortochange.Theeffectoffeedbackonpeoplesperformancedependsontheconsequencesthey
experience,haveexperienced,orexpecttoexperience(Daniels,1994,p.100).Acombinationoffeedbackandpositivereinforcementisthebestwaytosustainbehaviorchange.
7. Brophy(1981)indicatesthatfewteachersusepraiseasrewardforconductoraccomplishments.Theappliedresearchindicatesthatteacherpraiseiscuedbyavarietyofpersonal
characteristicsandbehaviorsofstudentsandnotasystematicattempttoshapebehaviorthroughreinforcement.
8. SeeChance(1992),inwhichtheauthorarguesthattoteachwithoutusingextrinsicrewardsisanalogoustoaskingourstudentstolearntodrawwiththeireyesclosed(p.200).
Thearticlepointstoproblemswiththeuseofrewardsinclassroomsbutsuggeststheseproblemscanandmustbeovercome.
9. TheinterestedreaderisreferredtoCatania(1998),Mazur(1998),orPierceandEpling(1999)forageneraloverviewoftheprinciplesofbehavior,whichinvolvereinforcement,
discriminationandgeneralization,conditionedreinforcement,schedulesofreinforcement,andseveralotherareasofbehavioranalysis.Aneasierintroductiontoteachingand
trainingbyreinforcementprinciplesispresentedinPryor(1999).ResearchonreinforcementprinciplesoftenappearsintheJournaloftheExperimentalAnalysisofBehavior.
10.ReadersinterestedintheuseofrewardsasreinforcementinappliedsettingscanbenefitfromthediscussioninChance(1998).Theuseofreinforcementinclassroomsettingscanbe
foundinAlbertoandTroutman(1999)andSchlossandSmith(1994).Inbusinessandindustry,theuseofrewardsasreinforcementisfoundintheperformancemanagement
literature(e.g.,Daniels,1994).ResearchonreinforcementinappliedsettingsoftenappearsintheJournalofAppliedBehaviorAnalysis.
11.Anotherkindofreinforcementprocessiscallednegativereinforcement.Thetermnegativereferstotheoperationofremovingortakingawaysomethingwhenaresponseoccurs.
Forexample,onarainyday,apersoninflatesanumbrellaandthusremovestherain.Iftheuseofanumbrellaonarainydayincreases,theprocessiscalledreinforcement.Thus,
negativereinforcementinvolvesanincreaseinaresponsebytheremovalofanongoingaversivestimulation.
12.SeeCatania(1998)concerningthedifficultyofattributinganincreaseinbehaviortoitsconsequences.Cataniastatesthattheinfantcoosandtheparentcomes,andnowthe
cooingincreases.Howdowedecidewhethertheinfantisnowcooingbecausecooinghasbeenreinforcedorbecausetheparentisnowpresentmakingcooingmorelikely?(p.70).
Ifweuserewardsasbehavioralconsequencesinappliedsettings,wemustbesurethattheincreasesinperformanceareduetotherewardsandnottoother,extraneousevents.
13.ThePremackprinciple(Premack,1959,1971)statesthatifresponseAismoreprobablethanresponseB,anopportunitytoengageinAcanbeusedtoreinforceB.Inthelaboratory
(withrats),ifeatinghasahigherprobabilityofoccurrencethanwheelrunning,anopportunitytoeatwillreinforcewheelrunning.
14.TheclassificationofverbsisfromDenoandJenkins(1967)andwasrepublishedinAlbertoandTroutman(1999,p.69).Anotherhelpfulsourcefordescriptionsoftargetbehavioris
Morris(1976),whoprovidesthefollowingtestquestions:(1)canyoucount
Page226

thenumberoftimesthebehavioroccurs(e.g.,fourtimesaday)?(2)Wouldastrangerknowexactlywhattolookforwhenyoutellhimorherthetargetbehavioryouaregoingto
change(canyouactuallyseethebehaviorwhenitoccurs)?(3)Areyoustillabletobreakdownthetargetbehaviorintosmallercomponentsthataremorespecificandobservable
thantheoriginaltargetbehavior(theanswershouldbeno)?
15.ThebehavioranalysisofperformanceinbusinessandindustryisfoundinDaniels(1989),whichpresentsapproachancalledperformancemanagement,ortheABCmodelof
behaviorchange.Thebasicideaistospecifytheantecedents(A),thetargetbehavior(B),andtheconsequences(C).Inthissectionofthebook,wehaveemphasizedbehaviorand
itsconsequencesratherthantheeventsorstimulithatprecedebehavior.ThefullABCsystemisnecessaryforeffectivebehaviormanagement.
16.Daniels,1989,p.13.
17.AdiscussionoftheoperantchamberasafavorablesituationforconditioningbyreinforcementappearsinPierceandEpling(1999,pp.102103).
18.Behaviormanagerisusedheretorefertoanyonewhoisinvolvedinbehaviorchange.Therulesforselectingreinforcingconsequencesnotonlyrelatetobusinessandindustry
butalsocanbeadaptedtohome,schoolsandavarietyoforganizations(e.g.,hospitals,legalsystems,andsoon).
19.Daniels,1989,pp.5663.
20.Daniels,1989,p.56.
21.SeeDaniels(1989,pp.5759)foramorecompletediscussionofthereinforcementsurvey.
22.Theanswertothiskindofoppositionistodiscussthebehavioralinterventionwiththeupperlevelmanagementbeforethereinforcementcontingencyisimplemented.Formost
bosses,productivityandprofitarereinforcingconsequencesandrestsorbreaksusuallyreducetheseoutcomes.Ifacontingencybetweenunitsperhourandtakingbreaks
increasesproductivityandprofit,thebossesshouldeventuallyadoptandsupportthereinforcementsystem.
23.Pryor,1999.
24.Pryor,1999,pp.6566.Inthesectiononshapingwithoutwords,Pryoralsopointsoutthatonceyouhaveachievedsuccessinshaping,youhadbetternotbragaboutit.Such
braggingispatronizingandusuallyresultsinmakingalifelongenemyofthepersontargetedforbehaviorchange.
25.Chance,1998,p.115.Chanceprovidesawellwrittenandsuccinctoverviewoftherulesofreinforcement.
26.Ibid.
27.Daniels,1989,p.98.Danielsnotesthatintermittentreinforcementavoidstheproblemofsatiationwithtangiblereinforcingconsequences.Satiationoccurswhenreinforcementhas
beenrepeatedlypresentedaswhenfoodispresentedfollowingeachleverpressandtherateventuallystopspressingthelever(theratissatiated).Intermittentreinforcementis
alsousefulformanagers(andteachers)becauseasinglemanagercanmaintainthebehaviorofseveralemployees(orstudents)byusingoccasionalreinforcement.
28.Daniels,1989,p.105.Danielsnotedthatwriters,directors,andotherswhotakeonlongtermprojectsmayshowpostreinforcementpausesthatlastweeks,months,orevenyears.
29.Onebyproductofvariableratioreinforcementisexcitement.Onereasonwhyathleticsandsportsareexcitingisthevariableratiopayoff,intermsofgoals,runs,or
Page227

baskets,forwatchingthegame.Thelackofexcitementinschoolandworkmayreflectthelackofvariableratioschedulesinthesesituations.
30.SeeLindsley(1968)ontheuseofwristcountersforrecordingratesofbehavior.
31.Othermeasuresofbehaviormayincludeitsformortopography,force,andlocus.AdiscussionofthesemeasuresisfoundinAlbertoandTroutman(1999,pp.97100).
32.SeeDaniels(1994,pp.9495)foradiscussionandexampleofthiskindofscale.
33.Daniels(1994,p.94)indicatedthatalthoughrankingisoneofthemostfrequentlyusedmeasures,

Istronglyadviseagainstit.Rankingshouldnotbeusedbecauseitsetsoneemployeeagainstanother....Byusingratingswecompareperformanceagainst
establishedcriteria.Inthisway,itispossibleforeveryonewhomeetstherequiredcriteriatoberatedasatopperformer.

34.Wolf(1978)introducedtheimportanceofsocialvalidationofappliedbehavioralinterventions.Theuseofquestionnaires,interviewsandsurveysareimportantwaysofassessing
theacceptabilityofreinforcementprogramstovarioussectorsandconsumers.
35.SeeHerrnstein(1961)onthematchinglaw,whichstatesthatfortwobehaviors,B1andB2,maintainedontwoschedulesofreinforcement,R1andR2,respectively,theproportional
rateofbehavior[B1/(B1+B2)]equalsormatchestheproportionalrateofreinforcement[R1/(R1+R2)].Basedonthisequation,ifreinforcementisaddedtoR2,thentherelativetime
spentonB1mustdecline.Thisrequirementofthematchinglawwouldbedetectedasasideeffectofreinforcementinappliedsettings.
36.Skinner(1979,p.346).Althoughthisquoterefersmoretobasicresearch,maintaininganexperimentalattitudeintermsofapplicationofreinforcementprinciplesisalsoimportant.In
fact,giventhecomplexitiesofappliedsettings,thewillingnesstochangerepeatedlymaybeessentialtosolvingpracticalproblemsinvolvinghumanbehavior.Inotherareasof
technology,finetuningasystemisanacceptablestrategy.Forexample,althoughthebasicprinciplesofphysicsallowsforthepropulsionofarocketintospace,theguidance
systemmustobtaindatathatallowsforcorrectionsintrajectory.Similarly,thebasicprinciplesofreinforcementcanbeusedforbehavioraltechnology,buttherealsomustbefine
tuningofthereinforcementsystem.
Page228

Thispageintentionallyleftblank.
Page229

PARTVII
CONCLUSION
Page230

Thispageintentionallyleftblank.
Page231

Chapter12
ResolvingtheControversyoverRewardsandIntrinsicMotivation
Therewardsandintrinsicmotivationcontroversyconcernsthedisputeovertheuseofrewardsandreinforcementtomotivatepeoplesperformanceandinterest.Many
teachersusegoldstars,recognition,bonuses,accesstopreferredactivities,orothertypesofrewardstoencouragehighlevelsofperformancebytheirstudents.In
business,managersfrequentlyofferincentivestoemployeesforexceptionalperformance.Overthepast30years,anumberofpsychologistshavequestionedthe
wisdomofthesepractices.Theconcernisthatrewardsunderminepeoplesintrinsicmotivationandperformance.Ifstudentsandemployeesarerewardedfordoingan
interestingtask,theclaimisthattheywillcometolikethetasklessandengageinitlessoncetherewardsarenolongerforthcoming.
Thecontentionthatrewardsundermineintrinsicmotivationrestsonabodyofexperimentalresearchfromsocialpsychology.Afewyearsago,ourresearchteam
conductedametaanalysisofthisliteraturetodeterminewhen,andunderwhatconditions,rewardsproduceeitherincreasesordecreasesinmeasuresofintrinsic
motivation.1Weconcludedthatnegativeeffectsofrewardoccurunderacircumscribedsetofconditionsandthat,whenappropriatelyarranged,rewardscanbeused
toenhancemotivationandperformance.
Ourfindingsandrecommendationswerehighlycontentioustothosewhoarguethatrewardsareinherentlyharmful.Spurredbyourresearch,certaininvestigators
whoviewrewardsasnegativeconductedareanalysisoftheliterature.2Theseinvestigatorssuggestedthatourpreviousmetaanalysiswasseriouslyflawedandthat
rewardsdo,infact,haveasubstantialunderminingeffect.Inthisbook,wehaveshownthatthereisnoinherentnegativeproperty
Page232

ofreward.Onthebasisofanupdatedmetaanalysisonthistopic(seeChapter8),3acarefulexaminationofpreviousresearch,andahistoricaloverviewofthe
literature(seeChapter10),wefoundnoreasontoaccepttheclaimthatrewardshavepervasivenegativeeffectsonpeoplesintrinsicmotivation.
Intermsofthemetaanalysispresentedinthisbook,theresearchfindingsonthetopicindicatethatrewardscanbeusedtoincreasemotivationandperformanceon
tasksthatareoflowinitialinterest.Onhighinteresttasks,positiveeffectsareobtainedwhenparticipantsareverballypraisedfortheirworkandtangiblerewardsare
offeredandexplicitlytiedtoperformancestandardsandsuccess.Theonlynegativeeffectoccurswhentangiblerewardssignifyfailureorareonlylooselytiedto
behavior.Overall,ourresearchsuggeststhatnegativeeffectsofrewardcanbeeasilypreventedandthatteachersandmanagerscanuserewardseffectivelytomotivate
performanceandinterest.
Althoughthenegativeeffectsofrewardarehighlycircumscribed,overthepast30years,manyresearchersandeducatorshavearguedagainsttheuseofrewards
andreinforcementineverydaylife.Tounderstandthisreluctancetousepositiveincentivesystems,wepointtothesociohistoricalcontextoftherewardsandintrinsic
motivationcontroversy.Studiesonnegativeeffectsofrewardwereinstigatedatatimewhenmanypeopleweresuspiciousofscienceandtechnology,especiallyasthey
appliedtohumanbehavior.
Aprominentviewinthe1960sand1970sconcernedhumanbeingsaswillfulandselfdeterminedamajorgoalwastoseekselfactualization.Anyexternal
influences(includingtheuseofrewards)wereseenasharmfultoanindividualspsychologicaldevelopment.Itiswithinthiscontextthatexperimentsonrewardsand
intrinsicmotivationwereconducted,andalthoughthefindingswereweak,theresearchwasinterpretedasevidencefortheharmfuleffectsofrewardsand
reinforcement.Thisviewledtoinertiainmanagementandteaching.Ratherthandesignprogramsthatusedpositivereinforcementtoinstillhabitsandvalues,individuals
involvedinperformancemanagementwerebeingtoldthatdoingnothingwasbetterthanusinganincentivesystem.
Whatwehaveshowninthisbookisthatrewardsandreinforcementcanbeusedtopromotedesiredbehaviorand,atthesametime,maintainpeoplesinterest.In
otherwords,thereisnoneedtobantheuseofrewardsinclassroomsandindustry.Instead,moreeffortneedstobedevotedtotheeffectivemanagementofrewardsin
appliedsettings.
ProminentvaluesofcontemporaryWesterncultureincludehumanhappiness,theindividualpursuitofselfdiscoveryandcreativepotential,andpersistenceinthe
faceofadversity.Thesevaluesarerelatedtothemoregeneralvaluesofindividualismconcerningfreedom,dignity,andindividualresponsibility.Inthepursuitofthese
values,oursocietyhasoftenresortedtopunishmentwhenbehaviorsconflictwithcherishedbeliefs.Inthisbook,wehaveshownthatthereisnoneedtousepunishment
andcoerciontoattainourvaluedgoals.Humanfreedomandhappinesscanbebetterattainedthroughtheeffectiveuseofrewardsandpositivereinforcementin
everydaylife.
Page233

NOTES
1.CameronandPierce,1994EisenbergerandCameron,1996.
2.Deci,Koestner,andRyan,1999.
3.Cameron,Banko,andPierce,2001.
Page234

Thispageintentionallyleftblank.
Page235

References
Studiesmarkedwithanasterisk(*)werethoseincludedinthemetaanalysispresentedinthisbook.
*Adorney,K.M.(1983).Facilitatingandunderminingintrinsicmotivation:Atestofattribution,cognitiveevaluationandcompetencetheories.Unpublished
doctoraldissertation,ColumbiaUniversity.
Alberto,P.A.,&Troutman,A.C.(1999).Appliedbehavioranalysisforteachers.UpperSaddleRiver,NJ:PrenticeHall.
*Amabile,T.M.,Hennesey,B.A.,&Grossman,B.S.(1986).Socialinfluencesoncreativity:Theeffectsofcontractedforreward.JournalofPersonalityandSocial
Psychology,50,1423.
*Anderson,R.,Manoogian,S.T.,&Reznick,J.S.(1976).Theunderminingandenhancingofintrinsicmotivationinpreschoolchildren.JournalofPersonalityand
SocialPsychology,34,915922.
*Anderson,S.,&Rodin,J.(1989).Isbadnewsalwaysbad?Cueandfeedbackeffectsonintrinsicmotivation.JournalofAppliedSocialPsychology,19,449467.
*Arkes,H.R.(1979).Competenceandtheoverjustificationeffect.MotivationandEmotion,3,143150.
*Arnold,H.J.(1976).Effectsofperformancefeedbackandextrinsicrewarduponhighintrinsicmotivation.OrganizationalBehaviorandHumanPerformance,17,
275288.
*Arnold,H.J.(1985).Taskperformance,perceivedcompetence,andattributedcausesofperformanceasdeterminantsofintrinsicmotivation.Academyof
ManagementJournal,28,876888.
Ayllon,T.,&Azrin,N.H.(1968).Thetokeneconomy.UpperSaddleRiver,NJ:PrenticeHall.
Page236

Azrin,N.H.,&Holz,W.C.(1966).Punishment.InW.K.Honig(Ed.),Operantbehavior:Areasofresearchandapplication(pp.380447).NewYork:
AppletonCenturyCrofts.

Baars,B.J.(1986).Thecognitiverevolutioninpsychology.NewYork:GuilfordPress.
Balsam,P.D.,&Bondy,A.S.(1983).Thenegativesideeffectsofreward.JournalofAppliedBehaviorAnalysis,16,283296.
Bandura,A.(1986).Socialfoundationsofthoughtandaction:Asocialcognitivetheory.UpperSaddleRiver,NJ:PrenticeHall.
Bandura,A.(1997).Selfefficacy:Theexerciseofcontrol.NewYork:Freeman.
Bandura,A.,&Schunk,D.H.(1981).Cultivatingcompetence,selfefficacyandintrinsicinterestthroughproximalselfmotivation.JournalofPersonalityandSocial
Psychology,41,586598.
*Bartelme,L.A.(1983).Theeffectsofchoiceandrewardsonintrinsicmotivationandperformance.Unpublisheddoctoraldissertation,UniversityofIowa.
Bates,J.A.(1979).Extrinsicrewardandintrinsicmotivation:Areviewwithimplicationsfortheclassroom.ReviewofEducationalResearch,49,557576.
Bem,D.J.(1965).Anexperimentalanalysisofselfpersuasion.JournalofExperimentalSocialPsychology,1,199218.
Bem,D.J.(1972).Selfperceptiontheory.InL.Berkowitz(Ed.),Advancesinexperimentalsocialpsychology(Vol.6)(pp.262).NewYork:AcadmicPress.
Bernstein,D.J.(1990).Ofcarrotsandsticks:AreviewofDeciandRyansIntrinsicmotivationandselfdeterminationinhumanbehavior.Journalofthe
ExperimentalAnalysisofBehavior,54,323332.
*Blanck,P.D.,Reis,H.T.,&Jackson,L.(1984).Theeffectsofverbalreinforcementofintrinsicmotivationforsexlinkedtasks.SexRoles,10,369386.
Boal,K.B.,&Cummings,L.L.(1981).Cognitiveevaluationtheory:Anexperimentaltestofprocessesandoutcomes.OrganizationalBehaviorandHuman
Performance,28,289310.
*Boggiano,A.K.,&Barrett,M.(1985).Performanceandmotivationaldeficitsofhelplessness:Theroleofmotivationalorientations.JournalofPersonalityand
SocialPsychology,49,17531761.
*Boggiano,A.K.,Harackiewicz,J.M.,Besette,J.M.,&Main,D.S.(1985).Increasingchildrensinterestthroughperformancecontingentreward.SocialCognition,
3,400411.
Boggiano,A.K.&Hertel,P.T.(1983).Bonusesandbribes:Moodeffectsinmemory.SocialCognition,2,4961.
*Boggiano,A.K.,Main,D.S.,&Katz,P.A.(1988).Childrenspreferenceforchallenge:Theroleofperceivedcompetenceandcontrol.JournalofPersonalityand
SocialPsychology,54,131141.
*Boggiano,A.K.,&Ruble,D.N.(1979).Competenceandtheoverjustificationeffect:Adevelopmentalstudy.JournalofPersonalityandSocialPsychology,37,
14621468.
*Boggiano,A.K.,Ruble,D.N.,&Pittman,T.S.(1982).Themasteryhypothesisandtheoverjustificationeffect.SocialCognition,1,3849.
*Brennan,T.P.,&Glover,J.A.(1980).Anexaminationoftheeffectofextrinsicreinforcersonintrinsicallymotivatedbehavior:experimentalandtheoretical.Social
BehaviorandPersonality,8,2732.
Page237

*Brewer,J.(1980).Theunderminingandenhancingeffectsofintrinsicmotivation:Adevelopmentalstudy.Unpublisheddoctoraldissertation,OhioState
University.
*Brockner,J.,&Vasta,R.(1981).Docausalattributionsmediatetheeffectsofextrinsicrewardsonintrinsicinterest?JournalofResearchinPersonality,15,201
209.
Brophy,J.(1981).Onpraisingeffectively.TheElementarySchoolJournal,81,270278.
*Butler,R.(1987).Taskinvolvingandegoinvolvingpropertiesofevaluation:Effectsofdifferentfeedbackconditionsonmotivationalperceptions,interest,and
performance.JournalofEducationalPsychology,79,474482.

*Calder,B.J.,&Staw,B.M.(1975).Selfperceptionofintrinsicandextrinsicmotivation.JournalofPersonalityandSocialPsychology,31,599605.
Cameron,J.,Banko,K.M.,&Pierce,W.D.(2001).Pervasivenegativeeffectsofrewardsonintrinsicmotivation:Themythcontinues.TheBehaviorAnalyst,24,1
44.
Cameron,J.,&Pierce,W.D.(1994).Reinforcement,rewardandintrinsicmotivation:Ametaanalysis.ReviewofEducationalResearch,64,363423.
Cameron,J.,&Pierce,W.D.(1996).Thedebateaboutrewardsandintrinsicmotivation:Protestsandaccusationsdonotaltertheresults.ReviewofEducational
Research,66,3951.
Cameron,J.,&Pierce,W.D.(1997).Rewards,interestandperformance:Anevaluationofexperimentalfindings.AmericanCompensationAssociation[ACA]
Journal,6,615.
Carlson,C.G.,Hersen,M.,&Eisler,R.M.(1972).Tokeneconomyprogramsinthetreatmentofhospitalizedadultpsychiatricpatients:Currentstatusandrecent
trends.JournalofNervousandMentalDisease,155,192204.
Carton,J.S.(1996).Thedifferentialeffectsoftangiblerewardsandpraiseonintrinsicmotivation:Acomparisonofcognitiveevaluationtheoryandoperanttheory.The
BehaviorAnalyst,19,237255.
*Carton,J.S.,&Nowicki,S.(1998).Shouldbehaviortherapistsstopusingreinforcement?Areexaminationoftheunderminingeffectofreinforcementonintrinsic
motivation.BehaviorTherapy,29,6586.
Catania,A.C.(1984).TheoperantbehaviorismofB.F.Skinner.BehavioralandBrainSciences,7,473.
Catania,A.C.(1998).Learning.UpperSaddleRiver,NJ:PrenticeHall.
Chance,P.(1992,November),Therewardsoflearning.PhiDeltaKappa,200207.
Chance,P.(1998).Firstcourseinappliedbehavioranalysis.PacificGrove,CA:Brooks/ColePublishing.
*Chung,K.T.(1995).Theeffectsofextrinsicreinforcementonintrinsicmotivationamongstmildlymentallyhandicappedchildren.CurriculumForum,4,98114.
*Cohen,D.S.(1974).Theeffectsoftaskchoice,monetary,andverbalrewardonintrinsicmotivation:AcloserlookatDeciscognitiveevaluationtheory.
Unpublisheddoctoraldissertation,OhioStateUniversity.
Cohen,J.(1988).Statisticalpoweranalysisforthebehavioralsciences(2nded.).Mahwah,NJ:Erlbaum.
Cooper,H.M.(1989).Integratingresearch:Aguideforliteraturereviews.BeverlyHills,CA:Sage.
*Crino,M.D.,&White,M.C.(1982).Feedbackeffectsinintrinsic/extrinsicrewardparadigms.JournalofManagement,8,95108.

*Dafoe,J.L.(1985).Useofrewardsinteachingaskill:Effectsoncompetence,selfefficacyandintrinsicmotivation.Unpublisheddoctoraldissertation,
StanfordUniversity.
Page238

Daly,H.B.(1969a).Isinstrumentalrespondingnecessaryfornonrewardfollowingrewardtobefrustrating?JournalofExperimentalPsychology,80,186187.
Daly,H.B.(1969b).Learningofahurdlejumpresponsetoescapecuespairedwithreducedrewardorfrustrativenonreward.JournalofExperimentalPsychology,
79,146157.
*Daniel,T.L.,&Esser,J.K.(1980).Intrinsicmotivationasinfluencedbyrewards,taskinterestandtaskstructure.JournalofAppliedPsychology,65,566573.
Daniels,A.C.(1989).Performancemanagement:Improvingqualityproductivitythroughpositivereinforcement.Tucker,GA:PerformanceManagement
Publications.
Daniels,A.C.(1994).Bringingoutthebestinpeople:Howtoapplytheastonishingpowerofpositivereinforcement.NewYork:McGrawHill.
*Danner,F.W.,&Lonkey,E.(1981).Acognitivedevelopmentalapproachtotheeffectsofrewardsonintrinsicmotivation.ChildDevelopment,52,10431052.
Davidson,P.,&Bucher,B.(1978).Intrinsicinterestandextrinsicreward:Theeffectsofacontinuingtokenprogramoncontinuingnonconstrainedpreference.
BehaviorTherapy,9,222234.
DeCharms,R.(1968).Personalcausation.NewYork:AcademicPress.
DeCharms,R.,&Muir,M.S.(1978).Motivation:Socialapproaches.AnnualReviewofPsychology,29,91113.
*Deci,E.L.(1971).Effectsofexternallymediatedrewardsonintrinsicmotivation.JournalofPersonalityandSocialPsychology,18,105115.
*Deci,E.L.(1972a).Theeffectsofcontingentandnoncontingentrewardsandcontrolsonintrinsicmotivation.OrganizationalBehaviorandHumanPerformance,
8,217229.
*Deci,E.L.(1972b).Intrinsicmotivation,extrinsicreinforcement,andinequity.JournalofPersonalityandSocialPsychology,22,113120.
Deci,E.L.(1975).Intrinsicmotivation.NewYork:PlenumPress.
*Deci,E.L.,Cascio,W.F.,&Krusell,J.(1975).CognitiveevaluationtheoryandsomecommentsontheCalderandStawcritique.JournalofPersonalityand
SocialPsychology,31,8185.
Deci,E.L.,Koestner,R.,&Ryan,R.M.(1999).Ametaanalyticreviewofexperimentsexaminingtheeffectsofextrinsicrewardsonintrinsicmotivation.
PsychologicalBulletin,125,627668.
Deci,E.L.,Koestner,R.&Ryan,R.M.(2001).Extrinsicrewardsandintrinsicmotivationineducation:Reconsideredonceagain.ReviewofEducationalResearch,
71,127.
Deci,E.L&Ryan,R.M.(1985).Intrinsicmotivationandselfdeterminationinhumanbehavior.NewYork:PlenumPress.
*DeLoach,L.L.,Griffith,K.,&LaBarba,R.C.(1983).Therelationshipofgroupcontextandintelligencetotheoverjustificationeffect.BulletinofthePsychonomic
Society,21,291293.
Deno,S.,&Jenkins,J.(1967).Evaluatingpreplanningcurriculumobjectives.Philadelphia:ResearchforBetterSchools.
Dewey,J.(1900).Psychologyandsocialpractice.PsychologicalReview,7,105124.ReprintedinE.R.Hilgard(Ed.)(1978),Americanpsychologyinhistorical
perspective:AddressesofthepresidentsoftheAmericanPsychologicalAssociation18921977(pp.6579).Washington,DC:APA.
Dickinson,A.M.(1989).Thedetrimentaleffectsofextrinsicreinforcementonintrinsicmotivation.TheBehaviorAnalyst,12,115.
Page239

*Dimitroff,G.(1984).Depressionofintrinsicallymotivatedperformancebyrewards:Theroleoffrustrationmediatedcontrasteffects.Unpublisheddoctoral
dissertation,UniversityofToronto.
*Dollinger,S.J.,&Thelen,M.H.(1978).Overjustificationandchildrensintrinsicmotivation:Comparativeeffectsoffourrewards.JournalofPersonalityandSocial
Psychology,36,12591269.
Dunham,P.J.(1968).Contrastedconditionsofreinforcement.PsychologicalBulletin,69,295315.

*Earn,B.M.(1982).Intrinsicmotivationasafunctionofextrinsicfinancialrewardsandsubjectslocusofcontrol.JournalofPersonality,50,360373.
*Effron,B.(1976).Effectsofselfmediatedcompetencyfeedbackandexternalincentivesonintrinsicmotivationandqualityoftaskperformance.
Unpublisheddoctoraldissertation,UniversityofPittsburgh.
Egan,T.(1995,November12).Takethisbribe,please,forvaluestobereceived.NewYorkTimes,WeekinReview,5.
Eisenberger,R.(1992).Learnedindustriousness.PsychologicalReview,99,248267.
*Eisenberger,R.(1999).Effectsofgenderandcooperativeversusindividualperformancecontingentrewardonintrinsicmotivation.Unpublishedmanuscript.
Eisenberger,R.,&Armeli,S.(1997).Cansalientrewardincreasecreativeperformancewithoutreducingintrinsiccreativeinterest?JournalofPersonalityandSocial
Psychology,72,652653.
Eisenberger,R.,&Cameron,J.(1996).Thedetrimentaleffectsofreward:Mythorreality?AmericanPsychologist,51,11531166.
Eisenberger,R.,&Cameron,J.(1998).Rewards,intrinsicinterestandcreativity:Newfindings.AmericanPsychologist.,53,676679.
Eisenberger,R.,Kaplan,R.M.,&Singer,R.D.(1974).Decrementalandnondecrementaleffectsofnoncontingentsocialapproval.JournalofPersonalityandSocial
Psychology,30,716722.
Eisenberger,R.,Leonard,J.M.,Carlson,J.,&Park,D.C.(1979).Transfereffectsofcontingentandnoncontingentpositivereinforcement:Mechanismsandgenerality.
AmericanJournalofPsychology,92,525535.
Eisenberger,R.,Pierce,W.D.,&Cameron,J.(1999).Effectsofrewardonintrinsicmotivation:Negative,neutral,andpositive.PsychologicalBulletin,125,677
691.
*Eisenberger,R.,Rhoades,L.,&Cameron,J.(1999).Doespayforperformanceincreaseordecreaseselfdeterminationandintrinsicmotivation?Journalof
PersonalityandSocialPsychology,77,10261040.
*Eisenstein,N.(1985).Effectsofcontractual,endogenous,orunexpectedrewardsonhighandlowinterestpresechoolers.ThePsychologicalRecord,35,2939.
Engelmann,S.(1992).Waragainsttheschoolsacademicabuse.Portland,OR:HalcyonHouse.
*Enzle,M.E.,Roogeveen,J.P.,&Look,S.C.(1991).Selfversusotherrewardadministrationandintrinsicmotivation.JournalofExperimentalSocial
Psychology,27,468479.
Enzle,M.E.,&Ross,J.M.(1978).Increasinganddecreasingintrinsicinterestwithcontingentrewards:Atestofcognitiveevaluationtheory.JournalofExperimental
SocialPsychology,14,588597.
Page240

Epling,W.F.&Pierce,W.D.(1992).Solvingtheanorexiapuzzle:Ascientificapproach.Toronto:Hogrefe&Huber.(Originalworkpublished1991.)

*Fabes,R.A.(1987).Effectsofrewardcontextsonyoungchildrenstaskinterest.JournalofPsychology,121,519.
*Fabes,R.A.,Eisenberg,N.,Fultz,J.,&Miller,P.(1988).Reward,affectandyoungchildrensmotivationalorientation.MotivationandEmotion,12,155169.
*Fabes,R.A.,Fultz,J.,Eisenberg,N.,MayPlumlee,T.,&Christopher,F.S.(1989).Effectsofrewardsonchildrensprosocialmotivation:Asocializationstudy.
DevelopmentalPsychology,25,509515.
*Fabes,R.A.,McCullers,J.C.,&Horn,H.(1986).Childrenstaskinterestandperformance:Immediatevs.subsequenteffectsofrewards.PersonalityandSocial
PsychologyBulletin,12,1730.
*Feehan,G.G.,&Enzle,M.E.(1991).Subjectivecontroloverrewards:Effectsofperceivedchoiceofrewardscheduleonintrinsicmotivationandbehavior
maintenance.PerceptualandMotorSkills,72,9951006.
Feingold,B.D.,&Mahoney,M.J.(1975).Reinforcementeffectsonintrinsicinterest:Underminingtheoverjustificationhypothesis.BehaviorTherapy,6,357377.
Festinger,L.(1957).Atheoryofcognitivedissonance.Evanston,Il:Row,Peterson.
Festinger,L.,&Carlsmith,J.M.(1959).Cognitiveconsequencesofforcedcompliance.JournalofAbnormalandSocialPsychology,58,203210.
Flesch,R.(1965).WhyJohnnycantread.NewYork:Harper&Row.(Originalworkpublished1955)
Flesch,R.(1981).WhyJohnnystillcantread:Anewlookatthescandalofourschools.NewYork:Harper&Row.
Flora,S.R.(1990).Underminingintrinsicinterestfromthestandpointofabehaviorist.ThePsychologicalRecord,40,323346.
Flora,S.R.,&Flora,D.B.(1999).Effectsofextrinsicreinforcementforreadingduringchildhoodonreportedreadinghabitsofcollegestudents.ThePsychological
Record,49,314.
Freedman,J.L.(1965).Longtermbehavioraleffectsofcognitivedissonance.JournalofExperimentalSocialPsychology,1,145155.
*Freedman,S.M.,&Phillips,J.S.(1985).Theeffectsofsituationalperformanceconstraintsonintrinsicmotivationandsatisfaction:Theroleofperceivedcompetence
andselfdetermination.OrganizationalBehaviorandHumanDecisionProcesses,35,397416.

Geller,L.(1982).Thefailureofselfactualizationtherapy:AcritiqueofCarlRogersandAbrahamMaslow.JournalofHumanisticPsychology,22,5673.
Glass,G.V.(1976).Primary,secondaryandmetaanalysisofresearch.EducationalResearcher,5,38.
Glass,G.V.,McGaw,B.,&Smith,M.L.(1981).Metaanalysisinsocialresearch.BeverlyHills,CA:Sage.
*Goldstein,G.S.(1980).Theeffectsofcompetitionandexternalrewardsonintrinsicmotivation.Unpublisheddoctoraldissertation,UniversityofNew
Hampshire.
*Goldstein,L.W.(1977).Intrinsicmotivation:Theroleofrewardandfeedbackonqualityofperformanceandsubsequentinterestinphotography.
Unpublisheddoctoraldissertation,CornellUniversity.
*Greene,D.,&Lepper,M.R.(1974).Effectsofextrinsicrewardsonchildrenssubsequentintrinsicinterest.ChildDevelopment,45,11411145.
Page241

*Griffith,K.M.(1984).Theeffectsofgroupversusindividualcontext,initialinterest,andrewardonintrinsicmotivation.Unpublisheddoctoraldissertation,
UniversityofSouthFlorida.
*Griffith,K.M.,DeLoach,L.L.,&LaBarba,R.C.(1984).Theeffectsofrewarderfamiliarityanddifferentialrewardpreferenceinintrinsicmotivation.Bulletinofthe
PsychonomicSociety,22,313316.

*Hamner,W.C.,&Foster,L.W.(1975).Areintrinsicandextrinsicrewardsadditive?AtestofDeciscognitiveevaluationtheoryoftaskmotivation.Organizational
BehaviorandHumanPerformance,14,398415.
*Harackiewicz,J.M.(1979).Theeffectsofrewardcontingencyandperformancefeedbackonintrinsicmotivation.JournalofPersonalityandSocialPsychology,
37,13521363.
*Harackiewicz,J.M.,Abrahams,S.,&Wageman,R.(1987).Performanceevaluationandintrinsicmotivation:Theeffectsofevaluativefocus,rewards,and
achievementorientation.JournalofPersonalityandSocialPsychology,53,10151023.
*Harackiewicz,J.M.,&Manderlink,G.(1984).Aprocessanalysisoftheeffectsofperformancecontingentrewardsonintrinsicmotivation.JournalofExperimental
SocialPsychology,20,531551.
*Harackiewicz,J.M.,Manderlink,G.,&Sansone,C.(1984).Rewardingpinballwizardry:Effectsofevaluationandcuevalueonintrinsicinterest.Journalof
PersonalityandSocialPsychology,47,287300.
Harackiewicz,J.M.,&Sansone,C.(2000).Rewardingcompetence:Theimportanceofgoalsinthestudyofintrinsicmotivation.InC.Sansone&J.M.Harackiewicz
(Eds.),Intrinsicandextrinsicmotivation:Thesearchforoptimalmotivationandperformance(pp.79103).SanDiego,CA:AcademicPress.
Harlow,H.F.(1950).Learningandsatiationofresponseinintrinsicallymotivatedcomplexpuzzleperformancebymonkeys.JournalofComparativePhysiological
Psychology,43,289294.
Harlow,H.F.,Harlow,M.K.,&Meyer,D.R.(1950).Learningmotivatedbyamanipulationdrive.JournalofExperimentalPsychology,40,228234.
Hawkins,D.(1995,October30).Johnnycanreadforcashandfreebies:Isbriberythebestwaytogetakidtolearn?U.S.NewsandWorldReport,7273.
Hedges,L.V.(1981).DistributiontheoryforGlasssestimatorofeffectsizeandrelatedestimators.JournalofEducationalStatistics,6,107128.
Hedges,L.V.&Becker,B.J.(1986).Statisticalmethodsinthemetaanalysisofresearchongenderdifferences.InJ.Hyde&M.C.Linn(Eds.),Thepsychologyof
gender:Advancesthroughmetaanalysis(pp.1450).Baltimore,MD:JohnHopkinsUniversityPress.
Hedges,L.V.&Olkin,I.(1985).Statisticalmethodsformetaanalysis.Orlando,FL:AcademicPress.
Heider,F.(1958).Thepsychologyofinterpersonalrelations.NewYork:JohnWiley&Sons.
Hennessey,B.A.,&Amabile,T.M.(1998).Reward,intrinsicmotivation,andcreativity.AmericanPsychologist,53,674675.
Herrnstein,R.J.(1961).Relativeandabsolutestrengthofresponsesasafunctionofthefrequencyofreinforcement.JournaloftheExperimentalAnalysisof
Behavior,4,267272.
*Hitt,D.D.,Marriott,R.G.,&Esser,J.K.(1992).Effectsofdelayedrewardsandtask
Page242

interestonintinsicmotivation.BasicandAppliedSocialPsychology,13,405414.
Hogan,R.(1975).Theoreticalegocentrismandtheproblemofcompliance.AmericanPsychologist,30,533540.
Horcones.(1983).NaturalreinforcementinaWaldenTwocommunity.RevistaMexicanadeAnalisisdellaConducta,9,141143.
Horcones.(1987).Theconceptofconsequencesintheanalysisofbehavior.TheBehaviorAnalyst,10,291294.
*Horn,H.L.(1987).Amethodologicalnote:Timeofparticipationeffectsonintrinsicmotivation.PersonalityandSocialPsychologyBulletin,13,210215.
Hunter,J.E.,&Schmidt,F.L.(1990).Methodsofmetaanalysis:correctingerrorandbiasinresearchfindings.NewburyPark,CA:SagePublications.
*Hyman,C.(1985).Rewardcontingency,standards,andintrinsicmotivation.Unpublisheddoctoraldissertation,CityUniversityofNewYork.

*Karniol,R.,&Ross,M.(1977).Theeffectofperformancerelevantandperformanceirrelevantrewardsonchildrensintrinsicmotivation.ChildDevelopment,48,
482487.
*Kast,A.,&Connor,K.(1988).Sexandagedifferencesinresponsetoinformationalandcontrollingfeedback.PersonalityandSocialPsychologyBulletin,14,
514523.
Kazdin,A.E.(1975a).Behaviormodificationinappliedsettings.Homewood,IL:DorseyPress.
Kazdin,A.E.(1975b).Recentadvancesintokeneconomyresearch.InM.Hersen,R.M.Eisler,&P.M.Miller(eds.),Progressinbehaviormodification(Vol.1)
(pp.233274)NewYork:AcademicPress.
Kazdin,A.E.(1994).Behaviormodificationinappliedsettings.PacificGrove,CA:Brooks/Cole.
Kazdin,A.E.,&Bootzin,R.R.(1972).Thetokeneconomy:Anevaluativereview.JournalofAppliedBehaviorAnalysis,5,343372.
Kelley,H.H.(1967).Attributiontheoryinsocialpsychology.InD.Levine(Ed.),Nebraskasymposiumonmotivation(Vol.15).Lincoln:UniversityofNebraska
Press.
Kline,M.(1973).WhyJohnnycantadd:Thefailureofthenewmath.NewYork:VantageBooks.
*Koestner,R.,Zuckerman,M.,&Koestner,J.(1987).Praise,involvement,andintrinsicmotivation.JournalofPersonalityandSocialPsychology,53,383390.
Kohn,A.(1993a).Punishedbyrewards.Boston:HoughtonMifflin.
Kohn,A.(1993b,SeptemberOctober).Whyincentiveplanscannotwork.HarvardBusinessReview,5463.
Kohn,A.(1996).Byallavailablemeans:CameronandPiercesdefenseofextrinsicmotivators.ReviewofEducationalResearch,66,14.
Kollins,S.H.,Newland,M.C.,&Critchfield,T.S.(1997).Humansensitivitytoreinforcementinoperantchoice:Howmuchdoconsequencesmatter?Psychonomic
BulletinandReview,4,208220.Erratum.PsychonomicBulletinandReview,4,431.
*Kruglanski,A.W.,Alon,S.,&Lewis,T.(1972).Retrospectivemisattributionandtaskenjoyment.JournalofExperimentalSocialPsychology,8,493501.
*Kruglanski,A.W.,Friedman,I.,&Zeevi,G.(1971).Theeffectsofextrinsicincentive
Page243

onsomequalitativeaspectsoftaskperformance.JournalofPersonality,39,606617.
*Kruglanski,A.W.,Riter,A.,Amitai,A.,Margolin,B.S.,Shabatai,L.,&Zaksh,D.(1975).Canmoneyenhanceintrinsicmotivation?Atestofthecontent
consequencehypothesis.JournalofPersonalityandSocialPsychology,31,744750.
Krutch,J.W.(1953).Theidolsofthelaboratory.InF.Connolly(ed.),Manandhismeasure(pp.12421251).NewYork:Harcourt,Brace&World.
Kuhlman,D.M.,Camac,C.R.,&Cunha,D.A.(1986).Individualdifferencesinsocialorientation.InH.Wilke,D.Messick,&C.Rutte(Eds.),Experimentalsocial
dilemmas(pp.151176).Frankfurt:VerlagPeterLang.
Kuhlman,D.M.,&Marshello,A.F.J.(1975).Individualdifferencesingamemotivationasmoderatorsofpreprogrammedstrategyeffectsinprisonersdilemma.
JournalofPersonalityandSocialPsychology,32,922931.
Kuhlman,D.M.,&Wimberley,D.(1976).Expectationsofchoicebehaviorheldbycooperators,competitorsandindividualistsacrossfourclassesofexperimental
games.JournalofPersonalityandSocialPsychology,34,6981.
Kuhn,T.S.(1962).Thestructureofscientificrevolutions.Chicago:UniversityofChicagoPress.
Kuo,Z.Y.(1931).Thegenesisofthecatsresponsestotherat.JournalofComparativePsychology,11,135.

Leahey,T.H.(1992).ThemythicalrevolutionsofAmericanpsychology.AmericanPsychologist,47,308318.
Leahey,T.H.(1987).Ahistoryofpsychology:MaincurrentsinpsychologicalthoughtUpperSaddleRiver,NJ:PrenticeHall.
*Lee,J.(1982).Effectsofabsoluteandnormativefeedbackandrewardonperformanceandaffect.Unpublisheddoctoraldissertation,UniversityofIowa.
Lepper,M.R.(1998).Awholemuchlessthanthesumofitsparts.AmericanPsychologist,53,675676.
Lepper,M.R.,&Gilovich,T.(1981).Themultiplefunctionsofreward:Asocialdevelopmentalperspective.InS.S.Brehm,S.Kassin,&F.X.Gibbions(Eds.),
Developmentalsocialpsychology(pp.531).NewYork:OxfordUniversityPress.
Lepper,M.R.,&Greene,D.(Eds.).(1978).Thehiddencostsofrewards:Newperspectivesofthepsychologyofhumanmotivation.Hillsdale,NJ:Erlbaum.
*Lepper,M.R.,Greene,D.,&Nisbett,R.E.(1973).Underminingchildrensintrinsicinterestwithextrinsicreward:Atestoftheoverjustificationhypothesis.Journal
ofPersonalityandSocialPsychology,28,129137.
Lepper,M.R.,&Henderlong,J.(2000).Turningplayintoworkandworkintoplay:25yearsofresearchonintrinsicversusextrinsicmotivation.InC.
Sansone,&J.M.Harackiewicz(Eds.),Intrinsicandextrinsicmotivation:Thesearchforoptimalmotivationandperformance(pp.257307).SanDiego,
CA:AcademicPress.
Lepper,M.R.,Keavney,M.,&Drake,M.(1996).Intrinsicmotivationandextrinsicrewards:AcommentaryonCameronandPiercesmetaanalysis.Reviewof
EducationalResearch,66,532.
*Lepper,M.R.,Sagotsky,G.,Dafoe,J.L.,&Greene,D.(1982).Consequencesofsuperfluoussocialconstraints:Effectsonyoungchildrenssocialinferencesand
subsequentintrinsicinterest.JournalofPersonalityandSocialPsychology,42,
Page244

5165.
Levine,F.M.,&Fasnacht,G.(1974).Tokenrewardsmayleadtotokenlearning.AmericanPsychologist,29,817820.
*Liberty,H.J.(1986).Intrinsicmotivation,extraversion,impulsivity,andrewardinacomputergamesetting.Unpublisheddoctoraldissertation,CityUniversity
ofNewYork.
Light,R.J.,&Pillemer,D.B.(1984).Summingup:Thescienceofreviewingresearch.Cambridge:HarvardUniversityPress.
Lindsley,O.R.(1968).Technicalnote:Areliablewristcounterforrecordingbehaviorrates.JournalofAppliedBehaviorAnalysis,1,7778.
*Loveland,K.K.,&Olley,J.G.(1979).Theeffectofexternalrewardoninterestandqualityoftaskperformanceinchildrenofhighandlowintrinsicmotivation.Child
Development,50,12071210.
*Luyten,H.,&Lens,W.(1981).Theeffectofearlierexperienceandrewardcontingenciesonintrinsicmotivation.MotivationandEmotion,5,2536.

MacKenzie,B.D.(1977).Behaviorismandthelimitsofthescientificmethod.AtlanticHighlands,NJ:HumanitiesPress.
Maier,S.F.,&Seligman,M.E.P.(1976).Learnedhelplessness:Theoryandevidence.JournalofExperimentalPsychology:General,105,346.
Martin,G.,&Pear,J.(1999).Behaviormodification:Whatisitandhowtodoit(6thed.).UpperSaddleRiver,NJ:PrenticeHall.
Mawhinney,T.C.(1990).Decreasingintrinsicmotivationwithextrinsicrewards:Easiersaidthandone.JournalofOrganizationalBehaviorManagement,11,
175191.
Mawhinney,T.C.,Dickinson,A.M.&Taylor,L.A.(1989).Theuseofconcurrentschedulestoevaluatetheeffectsofextrinsicrewardsonintrinsicmotivation.
JournalofOrganizationalBehaviorManagement,10,109129.
Mazur,J.E.(1998).Learningandbehavior.UpperSaddleRiver,NJ:PrenticeHall.
McAdams,J.L.,&Hawk,E.J.(1992).Capitalizingonhumanassets:Thebenchmarkstudy.Scottsdale,AZ:AmericanCompensationAssociationandMaritz
Inc.
McDougall,W.(1908).Anintroductiontosocialpsychology.London:Methuen.
*McGraw,K.O.,&McCullers,J.C.(1979).Evidenceofadetrimentaleffectofextrinsicincentivesonbreakingamentalset.JournalofExperimentalSocial
Psychology,15,285294.
*McLoyd,V.C.(1979).Theeffectsofextrinsicrewardsofdifferentialvalueonhighandlowintrinsicinterest.ChildDevelopment,50,10101019.
Miller,G.A.(1962).Psychology:Thescienceofmentallife.NewYork:Harper&Row.
*Morgan,M.(1981).Theoverjustificationeffect:Adevelopmentaltestofselfperceptioninterpretations.JournalofPersonalityandSocialPsychology,40,809
821.
*Morgan,M.(1983).Decrementsinintrinsicinterestamongrewardedandobserversubjects.ChildDevelopment,54,636644.
Morgan,M.(1984).Rewardinduceddecrementsandincrementsinintrinsicmotivation.ReviewofEducationalResearch,54,530.
Morris,R.(1976).Behaviormodificationwithchildren.Cambridge,MA:WinthropPublications.
*Mynatt,C.,Oakley,D.,Piccione,A.,Margolis,R.,&Arkkelin,J.(1978).Anexaminationofoverjustificationunderconditionsofextendedobservationandmultiple
reinforcement:Overjustificationorboredom?CognitiveTherapyandResearch,2,171177.
Page245

Neil,A.S.(1959).Summerhill:Aradicalapproachtochildrearing.NewYork:Hart.
Neisser,U.(1967).Cognitivepsychology.NewYork:AppletonCenturyCrofts.
Nelson,R.(1996).Theuseofinformalrewardsinrecognizingperformance.SmartBusinessSupersite(http://www.smarth.3.com).Availableonhttp://www.p
management.com/reward/reward.htm
Newell,A.,Shaw,J.C.,&Simon,H.A.(1958).Elementsofatheoryofproblemsolving.PsychologicalReview,65,151166.
*Newman,J.,&Layton,B.D.(1984).Overjustification:Aselfperceptionperspective.PersonalityandSocialPsychologyBulletin,10,419425.
Newman,L.S.,&Ruble,D.N.(1992).Doyoungchildrenusethediscountingprinciple?JournalofExperimentalSocialPsychology,28,572593.
Notz,W.W.(1975).Workmotivationandthenegativeeffectsofextrinsicrewards.AmericanPsychologist,30,884891.

*Ogilvie,L.,&Prior,M.(1982).Theoverjustificationeffectinretardedchildren:durabilityandgeneralizability.AustraliaandNewZealandJournalof
DevelopmentalDisabilities,8,213218.
*Okano,K.(1981).Theeffectsofextrinsicrewardonintrinsicmotivation.JournalofChildDevelopment,17,1123.
*Orlick,T.D.,&Mosher,R.(1978).Extrinsicawardsandparticipantmotivationinasportrelatedtask.InternationalJournalofSportPsychology,9,2739.
Overmeier,J.B.(1985).Towardareanalysisofthecausalstructureofthelearnedhelplessnesssyndrome.InF.R.Brush&J.B.Overmeier(Eds.),Affect,
conditioning,andcognition:Essaysonthedeterminantsofbehavior(pp.211227).Mahwah,NJ:Erlbaum.
Overmier,J.B.,&Seligman,M.E.P.(1967).Effectsofinescapableshockuponsubsequentescapeandavoidanceresponding.JournalofComparativeand
PhysiologicalPsychology,63,2833.
*Overskeid,G.,&Svartdal,F.(1996).Effectofrewardonsubjectiveautonomyandinterestwheninitialinterestislow.PsychologicalRecord,46,319331.

Palermo,D.S.(1971).Isascientificrevolutiontakingplaceinpsychology?ScienceStudies,1,135155.
*Pallak,S.R.,Costomotis,S.,Sroka,S.,&Pittman,T.S.(1982).Schoolexperience,rewardcharacteristicsandintrinsicmotivation.ChildDevelopment,53,1382
1391.
*Patrick,C.(1985).Theeffectofleveloftaskdifficultyonchildrensintrinsicmotivation.Unpublisheddoctoraldissertation,PurdueUniversity.
*Perry,D.G.,Bussey,K.,&Redman,J.(1977).Rewardinduceddecreasedplayeffects:Reattributionofmotivation,competingresponses,oravoidingfrustration?
ChildDevelopment,48,13691374.
Peterson,G.L.(1981).Historicalselfunderstandinginthesocialsciences:TheuseofThomasKuhninpsychology.JournalfortheTheoryofSocialBehavior,11,
130.
*Phillips,J.S.,&Freedman,S.M.(1985).Contingentpayandintrinsictaskinterest:Moderatingeffectsofworkvalues.JournalofAppliedPsychology,70,306
313.
*Picek,J.S.(1976).Effectsofrewarduncertaintyandabilityinformationonattributionsofintrinsicmotivation.Unpublisheddoctoraldissertation,Indiana
University.
Pierce,W.D.,&Epling,W.F.(1999).Behavioranalysisandlearning(2nded.).UpperSaddleRiver,NJ:PrenticeHall.
Page246

*Pittman,T.S.,Cooper,E.E.,&Smith,T.W.(1977).Attributionofcausalityandtheoverjustificationeffect.PersonalityandSocialPsychologyBulletin,3,280
283.
*Pittman,T.S.,Davey,M.E.,Alafat,K.A.,Wetherill,K.V.,&Kramer,N.A.(1980).Informationalversuscontrollingverbalrewards.PersonalityandSocial
PsychologyBulletin,6,228233.
*Pittman,T.S.,Emery,J.,Boggiano,A.K.(1982).Intrinsicandextrinsicmotivationalorientations:rewardinducedchangesinpreferenceforcomplexity.Journalof
PersonalityandSocialPsychology,42,789797.
*Porac,J.F.,&Meindl,J.(1982).Underminingoverjustification:Inducingintrinsicandextrinsictaskrepresentations.OrganizationalBehaviorandHuman
Performance,29,208226.
Porter,L.W.,&Lawler,E.E.(1968).Managerialattitudesandperformance.Homewood,IL:IrwinDorsey.
Premack,D.(1959).Towardempiricalbehaviorlaws:1.Positivereinforcement.PsychologicalReview,66,219233.
Premack,D.(1971).Catchingupwithcommonsenseortwosidesofageneralization:Reinforcementandpunishment.InR.Glaser(Ed.),Thenatureof
reinforcement(pp.121150).NewYork:AcademicPress.
*Pretty,G.H.,&Seligman,C.(1984).Affectandtheoverjustificationeffect.JournalofPersonalityandSocialPsychology,46,12411253.
Pryor,K.(1999).Dontshootthedog!Thenewarttoteachingandtraining(rev.ed.).NewYork:BantamBooks.

Rachlin,H.(1984).TheexplanatorypowerofSkinnersradicalbehaviorism.InS.Modgil&C.Modgil(Eds.),B.F.Skinner:Consensusandcontroversy(pp.155
164).Philadelphia,PA:FalmerPress.
*Reiss,S.,&Sushinsky,L.W.(1975).Overjustification,competingresponsesandtheacquisitionofintrinsicinterest.JournalofPersonalityandSocialPsychology,
31,11161125.
Reiss,S.,&Sushinsky,L.W.(1976).Thecompetingresponsehypothesisofdecreasedplayeffects:AreplytoLepperandGreene.JournalofPersonalityand
SocialPsychology,33,233245.
Rogers,C.R.(1964).Towardascienceoftheperson.InT.W.Wann(Ed.),Behaviorismandphenomenology(pp.109140).Chicago:UniversityofChicago
Press.
Rogers,C.R.(1969).Freedomtolearn.Columbus,OH:Merrill.
Rogers,C.R.,&Skinner,B.F.(1956).Someissuesconcerningthecontrolofhumanbehavior:Asymposium.Science,124,10571066.
*Rosenfield,D.,Folger,R.,&Adelman,H.F.(1980).Whenrewardsreflectcompetence:Aqualificationoftheoverjustificationeffect.JournalofPersonalityand
SocialPsychology,39,368376.
*Ross,M.(1975).Salienceofrewardandintrinsicmotivation.JournalofPersonalityandSocialPsychology,32,245254.
*Ross,M.,Karniol,R.,&Rothstein,M.(1976).Rewardcontingencyandintrinsicmotivationinchildren:Atestofthedelayofgratificationhypothesis.Journalof
PersonalityandSocialPsychology,33,442447.
Rousseau,J.J.(1974).Emile(B.Foxley,Trans.).London:Dent.(Originalworkpublished1762)
Rummel,A.,&Feinberg,R.(1988).Cognitiveevaluationtheory:Ametaanalyticreviewoftheliterature.SocialBehaviorandPersonality,16,147164.
Page247

Ryan,R.M.(1982).Controlandinformationintheintrapersonalsphere:Anextensionofcognitiveevaluationtheory.JournalofPersonalityandSocialPsychology,
43,450461.
Ryan,R.M.,&Deci,E.L.(1996).Whenparadigmsclash:CommentsonCameronandPiercesclaimthatrewardsdonotundermineintrinsicmotivation.Reviewof
EducationalResearch,66,3338.
Ryan,R.M.&Deci,E.L.(2000)Whenrewardscompetewithnature:Theunderminingofintrinsicmotivationandselfregulation.InC.Sansone&J.M.Harackiewicz
(Eds.),Intrinsicandextrinsicmotivation:Thesearchforoptimalmotivationandperformance.(pp.1354).SanDiego,CA:AcademicPress.
*Ryan,R.M.,Mims,B.,&Koestner,R.(1983).Relationofrewardcontingencyandinterpersonalcontexttointrinsicmotivation:Areviewandtestusingcognitive
evaluationtheory.JournalofPersonalityandSocialPsychology,45,736750.

*Salincik,G.R.(1975).Interactioneffectsofperformanceandmoneyonselfperceptionofintrinsicmotivation.OrganizationalBehaviorandHumanPerformance,
13,339351.
Sampson,E.E.(1988).Thedebateonindividualism:Indigenouspsychologiesandtheirroleinpersonalandsocietalfunctioning.AmericanPsychologist,43,1522.
*Sansone,C.(1986).Aquestionofcompetence:Theeffectsofcompetenceandtaskfeedbackonintrinsicinterest.JournalofPersonalityandSocialPsychology,
51,918931.
*Sansone,C.(1989).Competencefeedback,taskfeedbackandintrinsicinterest:Anexaminationofprocessandcontext.JournalofExperimentalSocial
Psychology,25,343361.
Sansone,C.,&Harackiewicz,J.M.(1998).Realityiscomplicated.AmericanPsychologist,53,673674.
Sansone,C.,&Harackiewicz,J.M.(Eds.).(2000).Intrinsicandextrinsicmotivation:Thesearchforoptimalmotivationandperformance.SanDiego,CA:
AcademicPress.
*Sansone,C.,Sachau,D.A.,&Weir,C.(1989).Effectsofinstructiononintrinsicinterest:Theimportanceofcontext.JournalofPersonalityandSocial
Psychology,57,819829.
*Sarafino,E.P.(1984).Intrinsicmotivationanddelayofgratificationinpreschoolers:Thevariablesofrewardsalienceandlengthofexpecteddelay.BritishJournalof
DevelopmentalPsychology,2,149156.
Schloss,P.J.,&Smith,M.A.(1994).Appliedbehavioranalysisintheclassroom.NeedhamHeights,MA:Allyn&Bacon.
Schunk,D.H.(1983).Rewardcontingenciesandthedevelopmentofchildrensskillsandselfefficacy.JournalofEducationalPsychology,75,511518.
Schunk,D.H.(1984).Enhancingselfefficacyandachievementthroughrewardsandgoals:Motivationalandinformationaleffects.JournalofEducationalResearch,
78,2934.
Schwartz,B.(1990).Thecreationanddestructionofvalue.AmericanPsychologist,45,715.
Schwartz,B.,Schuldenfrei,R.,&Lacey,H.(1978).Operantpsychologyasfactorypsychology.Behaviorism,2,229254.
Schwarzer,R.(1991)META:Programsforsecondarydataanalysis.MSDOSVersion5.0[Computerprogram].Dubuque,IA:Wm.C.Brown.
Seligman,M.E.P.(1975).Helplessness.SanFrancisco:Freeman.
Page248

*Shanab,M.E.,Peterson,D.,Dargahi,S.,&Deroian,P.(1981).Theeffectsofpositiveandnegativeverbalfeedbackontheintrinsicmotivationofmaleandfemale
subjects.TheJournalofSocialPsychology,115,195205.
*Shapira,Z.(1976).Expectancydeterminantsofintrinsicallymotivatedbehavior.JournalofPersonalityandSocialPsychology,34,12351244.
*ShiffmanKauffman,S.E.(1990).Theeffectsofrewardcontingencyandtypeoflearningexperienceonintrinsicmotivation.Unpublisheddoctoraldissertation,
CityUniversityofNewYork.
Sidman,M.(1989).Coercionanditsfallout.Boston,MA:AuthorsCooperative.
Simon,H.A.(1956).Rationalchoiceandthestructureoftheenvironment.PsychologicalReview,63,129138.
Skaggs,K.J.,Dickinson,A.M.,&OConnor,K.A.(1992).Theuseofconcurrentschedulestoevaluatetheeffectsofextrinsicrewardsonintrinsicmotivation:A
replication.JournalofOrganizationalBehaviorManagement,12,4583.
Skinner,B.F.(1938).Thebehavioroforganisms:Anexperimentalanalysis.UpperSaddleRiver,NJ:PrenticeHall.
Skinner,B.F.(1953).Scienceandhumanbehavior.NewYork:Macmillan.
Skinner,B.F.(1968).Thetechnologyofteaching.NewYork:AppletonCenturyCrofts.
Skinner,B.F.(1969).Contingenciesofreinforcement:Atheoreticalanalysis.UpperSaddleRiverNJ:PrenticeHall.
Skinner,B.F.(1971).Beyondfreedomanddignity.NewYork:Knopf.
Skinner,B.F.(1974).Aboutbehaviorism.NewYork:Knopf.
Skinner,B.F.(1978).WhyIamnotacognitivepsychologist.InReflectionsonbehaviorismandsociety.UpperSaddleRiver,NJ:PrenticeHall.
Skinner,B.F.(1979).Theshapingofabehaviorist.NewYork:Knopf.
Skinner,B.F.(1987).Selectionbyconsequences.InUponfurtherreflection.UpperSaddleRiver,NJ:PrenticeHall.
*Smith,A.T.(1980).Effectsofsymbolicrewardandpositivefeedbackonhighandlowlevelsofintrinsicmotivationinpreschoolers.Unpublisheddoctoral
dissertation,UniversityofMissouriColumbia.
Smith,L.D.(1992).Onpredictionandcontrol:B.F.Skinnerandthetechnologicalidealofscience.AmericanPsychologist,47,216223.
*xSmith,T.W.,&Pittman,T.S.(1978).Reward,distraction,andtheoverjustificationeffect.JournalofPersonalityandSocialPsychology,36,565573.
*Smith,W.E.(1975).Theeffectofanticipatedvs.unanticipatedsocialrewardonsubsequentintrinsicmotivation.Unpublisheddoctoraldissertation,Cornell
University.
Smyth,J.(2001,June20).Researchvindicatesbriberyasschooltool.Academiccontroversy:studysaysrewardswhetanenduringappetiteforlearning.National
Post,p.1.
*Sorensen,R.L.,&Maehr,M.L.(1976).Towardtheexperimentalanalysisofcontinuingmotivation.TheJournalofEducationalResearch,69,319322.
*Staw,B.M.,Calder,B.J.,Hess,R.K.,&Sandelands,L.E.(1980).Intrinsicmotivationandnormsaboutpayment.JournalofPersonality,48,114.
Stokes,T.F.,&Baer,D.M.(1977).Animplicittechnologyofgeneralization.JournalofAppliedBehaviorAnalysis,10,349367.
Sutherland,S.(1993).Impoverishedminds.Nature,364,767.
Svartdal,F.(1992).Sensitivitytononverbaloperantcontingencies:Dolimitedprocessing
Page249

resourcesaffectoperantconditioninginhumans?LearningandMotivation,23,383405.
*Swann,W.B.,Jr.,&Pittman,T.S.(1977).Moderatinginfluenceofverbalcuesonintrinsicmotivation.ChildDevelopment,48,11281132.

Tang,S.H.,&Hall,V.C.(1995).Theoverjustificationeffect:Ametaanalysis.AppliedCognitivePsychology,9,365404.
*Taub,S.I.,&Dollinger,S.J.(1975).Rewardandpurposeasincentivesforchildrendifferinginlocusofcontrolexpectancies.JournalofPersonality,43,179195.
Tegano,D.W.,Moran,D.J.,III&Sawyers,J.K.(1991).Creativityinearlychildhoodclassrooms.Washington,DC:NationalEducationAssociation.
*Thompson,E.P.,Chaiken,S.,&Hazlewood,D.(1993).Needforcognitionanddesireforcontrolasmoderatorsofextrinsicrewardeffects:ApersonXsituation
approachtothestudyofintrinsicmotivation.JournalofPersonalityandSocialPsychology,64,987999.
Thorndike,E.L.(1898).Animalintelligence.PsychologicalReviewMonographSupplements(SerialNo.8).
Thorndike,E.L.(1965).AnimalIntelligence:ExperimentalStudies.NewYork:Macmillan.(Originalworkpublished1911)
Tolman,E.C.(1932).Purposivebehaviorinanimalsandmen.NewYork:Naiburg.
Triandis,H.C.(1995).Individualismandcollectivism.Boulder,CO:WestviewPress.
*Tripathi,K.N.(1991).Effectofcontingencyandtimingofrewardonintrinsicmotivation.TheJournalofGeneralPsychology,118,97105.
*Tripathi,K.N.,&Agarwal,A.(1985).Effectsofverbalandtangiblerewardsonintrinsicmotivationinmalesandfemales.PsychologicalStudies,30,7784.
*Tripathi,K.N.,&Agarwal,A.(1988).Effectofrewardcontingencyonintrinsicmotivation.TheJournalofGeneralPsychology,115(3),241246.

*Vallerand,R.J.(1983).Theeffectofdifferentialamountsofpostiveverbalfeedbackontheintrinsicmotivationofmalehockeyplayers.JournalofSport
Psychology,5,100107.
*Vallerand,R.J.,&Reid,G.(1984).Onthecausaleffectsofperceivedcompetenceonintrinsicmotivation:Atestofcognitiveevaluationtheory.JournalofSport
Psychology,6,94102.
Vanderberg,G.(2001,June21).Bribery:itmaybethekeytobettergrades.TheEdmontonJournal,p.1.
Vasta,R.(1981).Ontokenrewardsandrealdangers.BehaviorModification,5,129140.
Vasta,R.,Andrews,D.E.,McLaughlin,A.M.,Stirpe,L.A.,&Comfort,C.(1978).Reinforcementeffectsonintrinsicinterest:Aclassroomanalog.JournalofSchool
Psychology,16,161168.
*Vasta,R.,&Stirpe,L.A.(1979).Reinforcementeffectsonthreemeasuresofchildrensinterestinmath.BehaviorModification,3,223244.
Vaughan,M.E.,&Michael,J.(1982).Automaticreinforcement:Animportantbutignoredconcept.Behaviorism,10,217227.
Vroom,V.(1964).Workandmotivation.NewYork:Wiley.

Watson,J.B.(1913).Psychologyasthebehavioristviewsit.PsychologicalReview,20,158177.
*Weinberg,R.S.,&Jackson,A.(1979)Competitionandextrinsicrewards:Effectonintrinsicmotivationandattribution.ResearchQuarterly,50,494502.
Page250

*Weiner,M.J.(1980).Theeffectofincentiveandcontroloveroutcomesuponintrinsicmotivationandperformance.TheJournalofSocialPsychology,112,247
254.
*Weiner,M.J.,&Mander,A.M.(1978).Theeffectsofrewardandperceptionofcompetencyuponintrinsicmotivation.MotivationandEmotion,2,6773.
Wessels,M.G.(1981).AcritiqueofSkinnersviewsontheexplanatoryinadequacyofcognitivetheories.Behaviorism,9,153170.
White,K.G.,&Cameron,J.(2000).Resistancetochange,contrast,andintrinsicmotivation.BehavioralandBrainSciences,23,115116.
*Wicker,F.W.,Brown,G.,Wiehe,J.A.,&Shim,W.Y.(1990).Moods,goals,andmeasuresofintrinsicmotivation.TheJournalofPsychology,124,7586.
Wiersma,U.J.(1992).Theeffectsofextrinsicrewardsinintrinsicmotivation:Ametaanalysis.JournalofOccupationalandOrganizationalPsychology,65,101
114.
Williams,B.A.(1983).Anotherlookatcontrastinmultipleschedules.JournaloftheExperimentalAnalysisofBehavior,39,345384.
*Williams,B.W.(1980).Reinforcement,behaviorconstraintandtheoverjustificationeffect.JournalofPersonalityandSocialPsychology,39,599614.
*Wilson,R.L.(1978).Theeffectofrewardonintrinsicmotivation:Anintegrationofdissonanceandintrinsicmotivationstudies.Unpublisheddoctoral
dissertation,NorthCarolinaStateUniversity.
*Wimperis,B.R.,&Farr,J.L.(1979).Theeffectsoftaskcontentandrewardcontingencyupontaskperformanceandsatisfaction.JournalofAppliedSocial
Psychology,9(3),229249.
Wolf,M.M.(1978).Socialvalidity:Thecaseforsubjectivemeasurementorhowappliedbehavioranalysisisfindingitsheart.JournalofAppliedBehaviorAnalysis,
11,203214.

*Yuen,W.C.(1984).Selfschema,taskinformation,extrinsicrewardandintrinsicmotivation.Unpublisheddoctoraldissertation,SimonFraserUniversity,
Canada.

Zimbardo,P.G.(1992).Psychologyandlife(13thed.).NewYork:HarperCollinsPublishersInc.
*Zinser,O.,Young,J.G.,&King,P.E.(1982).Theinfluenceofverbalrewardonintrinsicmotivationinchildren.TheJournalofGeneralPsychology,106,8591.
Zuriff,G.E.(1979).ThedemiseofbehaviorismExaggeratedrumor?AreviewofMacKenziesBehaviorismandthelimitsofthescientificmethod.Journalof
ExperimentalAnalysisofBehavior,32,129136.
Zuriff,G.E.(1985).Behaviorism:Aconceptualreconstruction.NewYork:ColumbiaUniversityPress.
Page251

Index
actionverbs,209
actionsandconsequences,69
ageofthenews,178
AmericanExpress,205
AmericanPsychologicalAssociation(APA),178,179
AmericanPsychologist,107
attitudes,andhumanbehavior,187188
attributiontheorists,53
attributiontheory.Seeoverjustificationhypothesis

Bandura,Albert,6061,92,157,167
andextrinsic/intrinsicmotivators,6163
andtemporallag,6869.
Seealsosociallearning
behavior:
consequencesof,212213
asfunctionofconsequences,7071
measurementof,222
andreinforcement,217221
targeting,210211
behaviorcontrol,180181,183
aspunishment,183
versuscoercion,183184
behavioralanalysis,7374
ofrewards,8081
behavioralpsychology,59,178179
andintrinsicmotivation,167169
behaviorism,178179,181,189190
challengesto,184189,190191
anddehumanization,180
oneducationalgoals,186
Bem,Daryl,51,189,192
BlanchardTrainingandDevelopment,Inc.,204
bonuses/promotionsinbusinessasrewards,171

Cameron,J.,106107,110,111,130,165
Carlsmith,J.M.,188
Carton,J.S.,167168
Chance,Paul,217218
classrooms,typesof,211
cognitivedissonancetheory,188189,192
cognitiveevaluationtheory(CET),3944,4647,5354,92,95,104,108,110111,112,115,131,164165
criticalassessmentof,4951
predictionsof,4849
andverbalpraise,130
cognitivepsychology,184,189192
andbehaviorism,187
cognitiverepresentations,187
Page252

collectivism,51
competence,socialvalidationof,68.
SeealsoBandura,Albert,andtemporallag
computerrevolution:
andmentalrepresentations,184
andreemergenceofcognitivepsychology,187
contingencies.Seerewardcontingencies
creativethinking,178179.
Seealsoperformance,creative

Daniels,AubryC.,209210,212,220
Darwin,Charles,182183
DeCharms,R.,13,41
Deci,Edward,4,41,42,44,65,92,94,98,107,110111,120,158,159,161,192193
effectsofrewardonchildrenversusadults,49
experimentsonintrinsicmotivation,1316,1819,27,28,30,32,47
determinism,190,191
Dewey,John,179
Dickinson,Alyce,25
discountingprinciple,55
andchildren5253
dissonancetheory.Seecognitivedissonancetheory
DontShoottheDog!(Pryor),216217

EagleAward,204
Eisenberger,R.,106107,110,111,130,165
extrinsicmotivation,12,13,19,20,6163

Feinberg,R.,103104
Feingold,B.D.,3031,32,87
Festinger,Leon,188
Flora,D.B.,171
Flora,S.R.,171
fooddeprivation,6263
funnelgraphs,121122

gamblingbehavior,218
Glass,Gene,102
GreatPerformersprogram,205
Greene,D.,1618,27,32,94,192,193

Hall,V.C.,105
Harlow,Henry,70
Heider,F.,41
Honeywell,Inc.,204
humanism,184187
oneducationalgoals,186

incentivesineverydaylife,202203
effectiveuseof,205
andperforminguninterestingactivities,203
tangibleincentives,203205
independentvariables,25
individualism,51
intrinsicmotivation,11,1920,23,3133,5051,6163,72,73,9192,231232
attitudemeasureof,2829
andcircularreasoning,40
andcognitiveevaluationtheory,3942
decrementsin,166
definition,12
effectsofchoiceandthefreetimemethod,27
effectsofreward,42,48,5354,124125
andfreechoice,125127,129,142,143,161
literature,189192
measuresof,9596
andoccurrencesofrewards,7475
andpraise,75
problemswithearlyexperiments,2329
andselfregulation,6364
andtheStanfordexperiment,1618
summaryofearlyresearchfindings,1819
andtangiblerewards,4244,106,123124,141143,164,167168
andverbalrewards,4445,140141,167168.
Seealsobehavioralpsychology,andintrinsicmotivation
Deci,Edward,experimentsonintrinsicmotivation
metaanalysesoftheeffectsofrewardsonintrinsicmotivation
metaanalysesoftheeffectsofrewardsonintrinsicmotivation,assessmentsof
metaanalysesoftheeffectsofrewardsonintrinsicmotivation,technicalaspects
rewardsandintrinsicmotivation,experimentson

Kelly,H.H.,192
Koestner,R.,44,107,117,120,159,161
Kohn,Alfie,4,19,3738,107
Kuhn,Thomas,191,192
Page253

Lacey,H.,190
Leahey,T.H.,191
Lepper,MarkR.,1618,20,24,25,27,29,30,32,94,107,158,192,193
andtheoverjustificationeffect,52,53
andtangiblerewards,5354

Mahoney,3031,32,87
mentalism,declineof,178179
METAcomputerprogram,119,138
metaanalysesoftheeffectsofrewardsonintrinsicmotivation,112113,115,116,125128,131,155158,172173,232
assessmentofmediationalprocesses,130131
calculationofeffectsizes,119,136137
classificationofstudies,117119
comparisontoDeci,Koestner,andRyanstudies,159,161
anddurabilityofrewardeffects,128129
effectsonhighinteresttasks,120122,140
effectsonlowinteresttasks,120121,139
effectsofmaximumversuslessthanmaximumrewards,125
effectsonrewardcontingency,123125
effectsonrewardexpectancy,123
magnitudeofrewardeffects,161,163164
overalleffectsofrewards,120,139
practicalimplications,169172
proceduresutilized,119120,134
researchquestions,116117
samplestudies,117
tangiblerewards,141144
theoreticalimplications,164,169
verbalrewards,140141
metaanalysesoftheeffectsofrewardsonintrinsicmotivation,assessmentsof:
assessmentofCameron,Pierce,andEisenbergsmetaanalysis,106107
assessmentofDeci,Koestner,andRyan,109112
assessmentofRummelandFeinbergsmetaanalysis,103104
assessmentofTangandHallsmetaanalysis,105
assessmentofWiersmasmetaanalysis,104105
metaanalysesoftheeffectsofrewardsonintrinsicmotivation,technicalaspects,134
codingofstudies,135136
computationofeffectsizes,136137
intercoderreliability,136
procedures,134
proceduresinhierarchicalmetaanalysis,137138
selectionofstudies,134135,147154
tangiblerewards,141144,168
verbalrewards,140141
metaanalysis,137138
andeffectsize,102
firsthierarchical,106
logicof,101103
newhierarchical,107109
stepsinvolved,102
techniqueof,101103.
Seealsometaanalysesoftheeffectsofrewardsonintrinsicmotivation

NewYorkTimes,20
Newman,L.S.,5253
Nisbett,R.E.,1618,27,32,94,192,193

operantbehavior,181
operantchamber.SeeSkinnerbox
overjustificationhypothesis,52,5455,94,105,130131
andverbalpraise,166.
SeealsoLepper,MarkR.,andtheoverjustificationeffect

Pavlov,IvanPetrovich,179180
performance,201,209210,211
breakandrun,220
creative,172173
selfmotivated,218219
Pierce,W.D.,106107,110,111,130,165
PizzaHut,171
praise,asanincentive,205206
Pretty,G.H.,163
Pryor,Karen,216217
PsychologicalBulletin,42
PsychologicalReview,180
psychologists,178
psychology,191.
Seealsocognitivepsychology
PsychologyastheBehavioristViewsIt(Watson),180
PunishedbyRewards(Kohn),19,3738
punishment,4,6,183
punishmentsystems,34
PurposiveBehaviorinAnimalsandMen(Tolman),187
puzzlebox,18,181
Page254

puzzles/puzzlesolving,71,74,88,93.
Seealsopuzzlebox

Rachlin,Howard,190
readforrewardprograms,20
reading,andrewards,163164
reinforcement,2324,77,9697,182184,192,207209
conditioned,212
contingency,215216
detrimentaleffectsof,193194
monitoring,221222
negativeeffectsof,223
andpraise,206
principleof,181
andratioschedules,219220
relativenatureof,208
survey,213
tailoringtotheindividual,24
usingrewardsas,223
variableintervalscheduleof,219
variableratioscheduleof,220221.
Seealsorulesforeffectivereinforcement
reinforcerevents,96
reinforcingrewards,andbehavioralconsequences,212213,215
rewardcontingencies,25,61,62,111,118,135136,170
andclassificationsofstudies,135136,166
taxonomyof,92
rewardhypothesis,detrimentaleffectsof,1213
rewardsystems,6
rewards:
andchildren,20,3031,94,109
competencycontingent,6567,9293,166167
completioncontingent,4344,50,54,6465,108,110
effectiveuseof,7,23
engagementcontingent,100
extrinsic,19,41,116,167,169
asformofcoercion,7879
andinitialtaskinterest,4647,9495
intermittentordelayed,7273
andinterpersonalcontext,4546,95
natural,7071
noncompetencycontingent,6567,9293,166167
performancecontingent,44,50,54,108,110
practicalapplicationsof,7
problemswithexperimentson,29
promiseof,2426
reinforcingversusnonreinforcing,7677
andsinglesubjectassessmentdesigns,2931
social,7172
andsocialvalidationofcompetence,68
successcontingent,168169
tangible,117
taskcontingent,110
verbalversustangible,2728,116
andwithdrawalofpromisedrewards,2627.
Seealsointrinsicmotivation
reinforcement
rewards,effectiveuseof
rewards,andharmfuleffects
rewardsasharmful,theoreticalperspectives
rewardsashelpful,theoreticalperspectives
rewardsandintrinsicmotivation,experimentson
sociallearning
socialpsychology
rewards,effectiveuseof,201202
inbusinessandindustry,204205
andencouragement,205206
ineverydaylife,202203,207
forperforminginterestingactivities,203205
forperforminguninterestingactivities,203
andreinforcement,207209.
Seealsorulesforeffectivereinforcement
rewards,andharmfuleffects,47,3133,108,165,172173,186,192194,231232
associatingrewardswithfailure,7778
conditionsfor,158,159
popularacceptanceof,1920.
Seealsorewardsasharmful,theoreticalperspectives
rewardsasharmful,theoreticalperspectives,3739,186
attributiontheory,5153,5455
cognitiveevaluationtheory,3940,44,4647,4851
rewardsashelpful,theoreticalperspectives,5960,8081,179,186187
andbehaviortheory,69
andbehavioralanalysis,7374
anddifferentialeffectsoftangiblerewards,7475
rewardsandintrinsicmotivation,experimentson,87,9798
betweengroupdesigns,88,89,93
impactofmoderatorvariables,88,90
interpersonalcontexts,95
andlevelofinitialtaskinterest,9495,120121
maximalversuslessthanmaximalrewards,9394,143144
andmeasuresofintrinsicmotivation,9596
moderatingvariables,95
andrewardcontingency,9193
andrewardexpectancy,91
andsrewardsalience,94
singlesubjectdesigns,9697
andtypesofreward,90
Page255

Rogers,Carl,185186
Ross,M.,94
Rousseau,185186
Ruble,D.N.,5253
rulesforeffectivereinforcement:
Rule1(SpecifytheTargetBehavior),209211
Rule2(ArrangeaFavorableSituation),211212
Rule3(SelectEffectiveRewards),212213,215
Rule4(SettheReinforcementContingency),215216
Rule5(WaitfortheTargetBehaviortoOccur),216217
Rule6(MovefromContinuoustoIntermittentReinforcement),217221
Rule7(MonitortheResultofYourRewardProgram),221223
Rule8(Experiment),223
Rummel,A.,103104
Ryan,R.M.,41,44,45,107,117,120,159,161
andtaxonomyofrewardcontingencies,92

Schuldenfrei,R.,190
Schwartz,B.,190
selfdetermination,165,166
selfreportedtaskinterest,128
Seligman,C.,163
Skinner,B.F.,25,76,181,182186,187,189,190,191,223
behavioralmodel,182
rejectionofindividualascausalagent,184
Skinnerbox,211
slidingscalesofrewards,171172
sociallearning,5961,6366,69,92,94,166167
socialpsychology,11,187189,192
sociohistoricalanalysisoftheliteratureofrewardsandintrinsicmotivation,177178,189194.
Seealsoindividuallylistedpsychologistsandpsychologicalmovements
TM
Soma puzzle,14,163
StructureofScientificRevolutions,The(Kuhn),191

Tang,S.H.,105
Thorndike,EdwardLee,181182
Tolman,Edwin,187
TravelRelatedServices,205
trialanderrorlearning,181

U.S.NewsandWorldReport,20

Watson,JohnBroadus,180181,182
Westernculture,viewson,183,232
Wiersma,U.J.,103105
Wundt,Wilhelm,178
Page256

Thispageintentionallyleftblank.
Page257

AbouttheAuthors
JUDYCAMERONisanAssociateProfessorofeducationalpsychologyattheUniversityofAlberta.
W.DAVIDPIERCEisaProfessorintheDepartmentofSociologyattheUniversityofAlberta,DirectoroftheCentreforExperimentalSociology,andretired
AdjunctProfessorintheDepartmentofNeuroscience(Medicine).

Вам также может понравиться