ThesadtruthaboutdepressiverealismLorraineG.Allan,ShepardSiegel,andSamu - PDF document

ThesadtruthaboutdepressiverealismLorraineG.Allan,ShepardSiegel,andSamuelHannahMcMasterUniversity,Hamilton,Ontario,CanadaInoneformofacontingencyjudgementtaskindividualsmustjudgetherelationshipbetweenanactionandanoutcome.Therearereportsthatdepressedindividualsaremoreaccuratethanarenon-depressedindividualsinthistask.Inparticular,nondepressedindividualsareinßuencedbymanipula- CorrespondenceshouldbeaddressedtoLorraineG.AllanorShepardSiegel,DepartmentofPsychology,Neuroscience,andBehaviour,McMasterUniversity,HamiltonON,L8S4K1,Canada.E-mail:allan@mcmaster.caorsiegel@mcmaster.caThepreparationofthispaperwassupportedbyresearchgrantsfromtheNaturalSciencesandEngineeringResearchCouncilto AlloyandAbramson(1979)ThereferentreportbyAlloyandAbramson(1979)describesaseriesoffourexperiments.ThedesignsoftheseexperimentsaresummarizedinTable2.Oneachof40trials,participantscouldeitherpressabutton(R)ornot(R),andthenagreenlightwasilluminated(O)ornot(O).Attheendofthe40trials,theparticipantrated,ona100-pointscale,thedegreeofcontrolthattheirresponsesexertedovertheilluminationofthelight,where0indicatednocontrol,and100indi-catedcompletecontrol.Onepurposeoftheseexperimentswastoevaluatethepredictionsmadebythelearnedhelplessnesstheoryofdepression(Seligman,1975).Accordingtothistheory,depressedpeoplehavegeneralizedexpectanciesofindependencebetweentheirresponsesandout-comesÑthedepressiveischaracterizedasonewhobelievesthatheorsheisineffectiveandpowerlesstocontroloutcomesintheworld.Adeductionfromlearnedhelplessnesstheoryisthatdepressedindividualsshouldunderestimatethedegreeofcontingencybetweentheirresponsesandenvironmentaloutcomes(Abramson&Alloy,1980;Alloy&Seligman,1979),andExperiment1inAlloyandAbramson(1979)evaluatedthispre-diction.Thesizeofthecontingency(.50,and.75)andthesignofthecontingency(posi-PandnegativeP)werevaried.ThepairsofconditionalprobabilitiesÑthatis,P(OR)andR),seeEquation1ÑforthethreepositivePvalueswere.75and.50,.75and.25,and.75and.00,andthepairsforthethreenegativevalueswere.50and.75,.25and.75,and.00and.75.Therewasnoeffectofsignonratings,andtheratings,collapsedoversign,areplottedasafunctionofPinFigure1a.RatingsofcontrolincreasedwithP,andtherewasnodifferencebetweendepressedandnondepressedparticipants.TheabsenceofadifferencebetweentheratingsofthetwomoodgroupsisinconsistentwiththepredictionsoflearnedhelplessnesstheoryÑdepressivesshouldhaveunderestimatedthedegreeofcontingencybetweenresponseandoutcome,butdidnot. Figure1.ThedataarereplottedfromtheTablesinAlloyandAbramson(1979).Figure1ashowsjudgedcontrolinExperiment1asafunctionofPforeachmoodgroup(NDdepressed).Figure1bshowsjudgedcontrolinExperiment2asafunctionofP(O)foreachmoodgroup(ND Table2.AsummaryofthedesignofthefourexperimentsreportedbyAlloyandAbramson(1979)ExperimentPP(O)Ovalence.25.675.50.500.75.37520.250.7530.5win0.5lose.5.5win.5.5lose Itshouldbenotedthatalthoughtherewerenegativecontingencies,theratingscalerangedfrom0to100.THEQUARTERLYJOURNALOFEXPERIMENTALPSYCHOLOGY,2007,60(3)ALLAN,SIEGEL,HANNAH AccordingtoAlloyandAbramson(1979),learnedhelplessnesstheoryalsoregardsnonde-pressivesashavingageneralizedexpectationofcontrol,whichshouldinterferewiththedetectionofnoncontingencies.ÒJustasdepressivesÕgeneral-izedexpectationofresponseÐoutcomeindepen-denceinterfereswiththeirabilitytoperceivethatoutcomesarenowdependentonresponses,nonde-pressivesÕgeneralizedexpectationthatoutcomesaredependentonresponsesshouldinterferewiththeirabilitytoperceivethatoutcomesareindepen-dentofresponsesÓ(p.457).Thus,nondepressivesshouldshowanillusionofcontrol,anddepressivesshouldnotwheneventsarenoncontingent.InExperiment2inAlloyandAbramson(1979),therelationshipbetweenrespondingandtheoutcomewasnoncontingent(i.e.,0)andP(O)wasvaried(.25and.75).TheratingsareplottedasafunctionofP(O)inFigure1b.ItisclearthatP(O)interactedwithmoodstate.Whentheoutcomewasinfrequent,P(O)ratingswereclosetozeroforbothmoodgroups.NondepressiveratingsincreasedwhenP(O).75,whereasdepressiveratingswereunaffectedbyP(O).Thus,nondepressivesshowedanillusionofcontrolÑjudgedcontrolincreasedwithP(O)ÑwhereasdepressivesdisplayeddepressiverealismÑjudgedcontrolwasunaffectedbyP(O).AlloyandAbramsonarguedthattheresultsoftheirExperiment2,whileshowingadifferencebetweendepressivesandnondepressives,wereinconsistentwithlearnedhelplessnesstheory,whichpredictsthatthetwomoodgroupsshoulddifferwhenP(O).25aswellaswhenP(O).75.Thatis,thetheorypredictsamoodeffect,whereasthedatarevealedaMoodDensityinteraction.Intheirremainingtwoexperiments,AlloyandAbramson(1979)evaluatedtheeffectsofoutcomevalence,ratherthanprobability,oncontrolratingsindepressiveandnondepressiveparticipants.Thatis,anoutcomewasmadeeitherdesirableorunde-sirable(ratherthanfrequentorinfrequent).InExperiment3,bothPandP(O)wereconstant,0andP(O).5,andthevalenceoflightonsetwasmanipulated.FortheÒwinÓcondition,theparticipantgained$0.25oneachtrialonwhichthelightturnedon.FortheÒloseÓcon-dition,theparticipantbegantheblockoftrialswith$5.00andlost$0.25oneachtrialonwhichthelightdidnotturnon.Therewasnocontin-gencybetweenrespondingandpayoff;thatis,thesepayoffsoccurredregardlessofwhetherornottheparticipantresponded.AswithP(O),outcomevalenceinßuencednondepressiveratingsbutnotdepressiveratings.Inthelosecondition,ratingswerelowforbothmoodgroups.Nondepressiveratingsincreasedinthewincon-dition,whereasdepressiveratingsdidnot.Thus,depressiverealismwasdemonstratedwithanÒoutcomevalenceeffectÓ,aswellaswithanoutcome-densityeffect.Experiment4alsovariedthevalenceoftheoutcome.IncontrastwithExperiment3(inwhich0),however,Pwassetateither.5orÐ.5.When.5,theconditionalprobabilitieswere.75and.25.WhenÐ.5,theconditionalprobabilitieswere.25and.75.Forthewincondition,theparticipantgained$0.10oneachtrialonwhichthelightturnedon.Forthelosecondition,theparticipantbegantheblockoftrialswith$4.00andlost$0.30oneachtrialonwhichthelightdidnotturnon.AsinExperiment3,outcomevalenceinßuencednonde-pressiveratingsbutnotdepressiveratings.Inthewincondition,thetwomoodgroupsdidnotdiffer,andratingswereataboutthemiddleofthescale(i.e.,50).Nondepressiveratingsdecreasedinthelosecondition,whereasdepressiveratingsdidnot.Inadditiontoaskingtheirparticipantstoratethedegreeofcontrolofrespondingontheoutcome,inalltheirexperimentsAlloyandAbramson(1979)requiredtheirparticipants(a)toestimatetheoverallpercentageoflightonsetÑthatis,P(O),(b)toestimatethe AlloyandAbramson(1979)donotethatnotallresearchersagreewiththeirinterpretationoflearnedhelplesstheory(seep.457).THEQUARTERLYJOURNALOFEXPERIMENTALPSYCHOLOGY,2007,60(3)DEPRESSIVEREALISM university,thusextendingthegeneralityoftheAlloyandAbramsonresults.InPressonandBenassi,0andP(O).75.ConsistentwiththeÞndingsofAlloyandAbramsonandofzquez,depressivejudgementsofcontrolwerelowerthannondepressivejudgements.NonreplicationsIncontrastwiththeabovereports,othershavereportedinabilitiestoreplicatethedifferenceincontroljudgementsbetweenthetwomoodstates0.Bryson,Doan,andPasquali(1984)reportedthatneithermoodgroupshowedanillusionofcontrolÑratingswereconstantwithincreasesinP(O)ÑwhereasKapcüiandCramer(1999)reportedthatbothmoodgroupsshowedanillusionofcontrolÑratingsincreasedwithincreasesinP(O).DobsonandPusch(1995)usedclinicalpopulationsratherthancollegestudents.Therewerethreepatientgroups:clinicallydepressed(currentlyreceivingtreatment),previouslydepressed(nolongerreceiv-ingtreatment),andneverdepressed.IntheDobsonandPuschstudy,0andP(O).75.Thethreegroupsdidnotdifferintheirratings.AlthoughDobsonandPuschdidnotvaryP(O),theyusedthelevelofP(O)atwhichAlloyandAbramson(1979)foundadifferencebetweenmoodgroups.SummaryWhentherelationshipbetweenrespondingandtheappearanceoftheoutcomeisnoncontingent0),therehavebeenbothreplicationsandnonreplicationsoftheÞndingsreportedbyAlloyandAbramson(1979).Insomestudiesnon-depressivesshowanillusionofcontrol,anddepressivesdonot,butinotherstudiesthetwomoodgroupsdonotdiffer.Thereappearstobeconsistencyamongallthestudiesthatthetwomoodgroupsdonotdifferwhentherelationshipiscontingent(i.e.,0).AlloyandAbramsoncommentedonthisabsenceofamooddifferenceintheirExperiment1:Althoughmanipulationsofreinforcementfrequencyaffectedjudgementsofcontroladverselyinthenoncontingentcase,suchmanipulationswerenotsufÞcienttoproduceerrorsinthecontingentcase.Theseresultsmayimplythatnoncontin-gencyispsychologicallymoredifÞcultrelationshiptoperceiveorunderstandthancontingency.(p.474)Aswasnotedearlier,PandP(O)werecon-foundedintheirExperiment1andalsoinLennoxetal.(1990).Intheotherstudieswhere0,P(O)wasconstant.InVazquez(1987),.375forbothcontingencyvalues,andinüiandCramer(1999),P(O).5forbothcon-tingencyvalues.GiventhatAlloyandAbramsonfoundaninteractionbetweenmoodandP(O)0,itissurprisingthatinvestigatorshavenotmanipulatedP(O)when0andmoreoverthattheyhaveevaluatedonlylowvaluesofP(O)Ñthatis,.5.Onthebasisoftheavailable0data,onewouldnotexpectadifferencebetweenmoodgroupswithlowvaluesofP(O).IntheirExperiments3and4,AlloyandAbramson(1979)reportedthatmanipulationsofoutcomevalence(inadditiontooutcomeprob-ability)canbeusedtodemonstratedepressiverealism.Subsequentresearch,inspiredbytheseÞndings,haveadditionallyevaluatedtheeffectsofoutcomevalencemanipulationsonmoodeffectsincontingencyjudgements.MoodeffectsandjudgementsofInducedmoodTheAlloyandAbramson(1979)experimentsdonotprovideinformationaboutthecausaldirectionofthecorrelationbetweendepressionandaccuracyinjudgingcontingencies.Itcouldbethatthedepressivemoodstatecausespeopletoassesscon-tingenciesaccurately.Ifso,peopleshouldjudge AlloyandAbramson(1979)usedÒreinforcementÓinterchangeablywithÒoutcomeÓ.Alloy,Abramson,andcollaboratorspublishedaseriesofpapersinthe1980s.Inourreviewwehaveincludedonlypapersdirectlyconcernedwithmoodeffectsonjudgementsofcontingency.Wehavenotincludedpaperswherethemainfocuswasonlearnedhelp-lessnesstheory(e.g.,Abramson,Alloy,&Rosoff,1981;Alloy&Abramson,1982).THEQUARTERLYJOURNALOFEXPERIMENTALPSYCHOLOGY,2007,60(3)DEPRESSIVEREALISM higherthannondepressiveratings.Thus,thepre-senceofanobserverreversestheeffectsreportedbyAlloyandAbramson(1979)Ñdepressedcollegestudentsshowanillusionofcontrolrelativetonondepressedcollegestudents.zquez(1987)wasalsointerestedinselfversusothersinhisExperiments3and4.Heusedsentences(negativeorpositive)asoutcomesandvariedthereferentinthesesentences(selforother).Examplesofthefoursentencetypesare:ÒMyproblemsareunsolvableÓ(negativeself-refer-ent),ÒMyproblemsare,ingeneral,notunsolvableÓ(positiveself-referent),ÒProblemsofhumanbeings[aggressivity,selÞshness,etc.]willneverbesolvedÓ(negativeother-referent),andÒProblemsofhumanbeings[aggressivity,selÞsh-ness,etc.]willbesolvedatlastÓ(positiveother-referent).InExperiment3.25(.50and.25),andinExperiment40(.75and.75).Acrossthetwoexperiments,controljudgementswereacomplexinteractionofP,valence,andreferent.Overall,nondepressivesweremoreinßu-encedbyvalencethenweredepressives.However,0,depressivesshowedhigherjudge-mentsofcontrolthandonondepressiveswhentheoutcomeswerenegativeself-referentsen-tences.Also,outcomevalencedidnotaffectthejudgementsofeithermoodgroupwhenthesen-tenceswereother-referent.ActiveversuspassiveAlmostallreportedcontingencystudiesconcernedwithdepressiverealismhaveusedthediscrete-trial,activetaskdescribedearlier.Inabookchapter,Alloy,Abramson,andKossman(1985)brießydescribethreestudiesthatusedthepassive(Pavlovian)contingencytask.Inthepassivetask,acueiseitherpresentedorisnotpresented,andthentheoutcomeeitheroccursordoesnotoccur.Attheendofablockoftrials,theparticipantisaskedtoratethestrengthoftherelationshipbetweenthecueandtheoutcome(incontrastwiththeactivetask,inwhichjudgementsareaboutresponseÐoutcomerelationships).InExperiment1inAlloyetal.,thecuewasaredlight(presentedornotpresented),andtheoutcomewasagreenlight(presentedornotpre-sented).Attheendofa40-trialsequence,thepar-ticipantwasaskedÒtojudgethedegreeofcontingencyorpredictability(ona0to100scale)thatexistedbetweenredlightonsetandgreenlightonsetÓ(p.236).ThereweretwovaluesofP(.25and.75).Thepairsofconditionalprobabil-itiesforthetwoPvalueswere.75and.50,and.75and00.AsinExperiment1ofAlloyandAbramson(1979),ratingsincreasedwithPandthetwomoodgroupsdidnotdiffer.Experiment2wasmodelledonExperiment3inAlloyandAbramson(seeTable2):0,P(O).5,andthewinÐloseconditionswereusedtomanipulateoutcomevalence.Alloyetal.concludedthatÒtherewerenodifferencesbetweendepressedandnondepressedsubjectsÕjudgementsofpredictability,andbothgroupsjudgedrelativelyaccuratelythattheredlightprovidedlittlepredictionofthegreenlightÓ(p.237).InExperiment3,andP(O).75.Forhalftheparticipants,thetaskwasactive(asinAlloy&Abramson,1979)andforhalftheparticipantsthetaskwaspassive.Alloyetal.concludedthatÒwhereasnondepressedsubjectsasagroupweremorelikelythandepressedsubjectsasagrouptoexhibitanillusionofpersonalcontrolwhentheexperimentaloutcomewasnon-contingentbutfrequent,nondepressiveswerenomorelikelythandepressivestoexhibitanillusionofpredictionaboutthissameoutcomeÓ(p.240).Moodstatedoesnotappeartodifferentiallyaffectpredictivejudgements.Insummary,incontrastwiththeÞndingsofAlloyandAbramson(1979)withtheactiveformofthecontingencytask,Alloyetal.(1985)reportednoevidenceofdepressiverealismwiththepassiveformofthetask.However,theseAlloyetal.datawerepresentedbrießyinachapterandhavenotsubsequentlybeenpresentedinamorecomprehensiveforminajournalarticle(L.B.Alloy,personalcommunication,April18,2005). Thethreestudiesreportedinthischapterhavenotbeenpublishedinajournalarticle.Thedescriptioninthechapterisbrief,andneitherstandarderrorsnorstatisticalanalysesarereported.THEQUARTERLYJOURNALOFEXPERIMENTALPSYCHOLOGY,2007,60(3)DEPRESSIVEREALISM provideopportunitiesforanewlookatdepressiverealism.ITIhypothesisWenotedearlierthatMsetÞetal.(2005;MsetÞetal.,2007)concludedthatITIdurationinßu-encedcontroljudgements.SpeciÞcally,withashortITI,P(O)didnotaffectratings,andthiswasthecaseforbothdepressivesandnondepres-sivesÑanoutcome-densityeffectwasnotseenineithermoodgroup.WithalongITI,P(O)affectednondepressiveratingsbutnotdepressiveratingsÑanoutcome-densityeffectwasseeninnondepres-siveratingsbutnotindepressiveratings(i.e.,depressiverealismwasobserved).Toaccountfortheirresults,MsetÞetal.(2005;MsetÞetal.,2007)suggestedthattheITIcouldbeconceptual-izedasequivalenttocelldofthe22matrixdepictedinFigure1Ñthatis,O.Thus,theITIhypothesispredictsaninteractionbetweenITIdurationandP(O)Ñtheoutcome-densityeffectshouldincreasewithITIduration.MsetÞetal.(2005)explainedhowtheirITIintegrationhypothesisiscompatiblewithbothcomputationalmodels(e.g.,Cheng,1997)andassociativemodels(e.g.,Dickinson,Shanks,&Evenden,1984)ofcontingencyjudgements.ComputationalmodelsspecifythatjudgementsaredeterminedbyPÑandinsomeversions,PnormalizedbyP(OR).IntegrationofITIintocelldwouldeffectivelydecreaseP(OandtherebyincreaseP.SinceinßationofcelldwouldbeanincreasingfunctionofITIduration,effectivePwouldalsoincreasewithITI.Moreover,thehigherthevalueofP(O),thehigherthespuriousvalueofP.Withinthefra-meworkofanassociativemodel,ITIisconceptu-alizedasthecontext.IntegrationofITIintocelldwouldweakenthecontextÕsassociationwiththeoutcome,therebyallowingtheresponsetogainmoreassociativestrength.ToaccountfortheabsenceofanITIeffectintheirdepressedparticipants,MsetÞetal.(2005;MsetÞetal.,2007)suggestedthatdepressedmoodisaccompaniedbyreducedcontextualpro-cessingÑspeciÞcallydepressivesdonotintegratetheITIintocelld.Theyconcludedthatdepressiverealismeffectsoccurbecausedepressedpeopledonotusealltheavailableevidencetoarriveattheirjudgementsratherthanbecausetheyarebeingrealistic.MsetÞetal.(2005)didacknowledgethattheirITIaccountoftheoutcome-densityeffectisincomplete.Therearemanystudiesthatshowanoutcome-densityeffectinthepassivetaskwhentheITIisshort(e.g.,Allan&Jenkins,1983,Exp.3;Allanetal.,2005;Vallee-Tourangeau,Murphy,Drew,&Baker,1998).MsetÞetal.con-cludedthattheITIhypothesisis,sofar,restrictedtotheactivecontingenttask.ResponsecriterionhypothesisOnephenomenonthattriggeredinterestindepressiverealisminthecontingencyjudgementliteraturewastheÞndingthatdepressedindivid-ualsseemedimmunetotheoutcome-densityeffectÑtheirjudgementsofthecontingencybetweenrespondingandoutcomewerenotunrea-listicallyinßatedastheoutcomedensityincreased.Elsewhere(Allanetal.,2005)wehavearguedthatitisusefultoconsideranindividualÕsjudgementofthemagnitudeofacontingentrelationshipascon-sistingoftwocomponents.Itisasifthepartici-pantaskshimselforherselftwoquestionsoneachtrial:(a)ÒWhatdoIperceivethelikelihoodoftheoutcometobeonthistrial?Ó,and(b)ÒGiventhatlikelihood,howshouldIrespond?ÓThesetwocomponentsofthedecisionprocess,ofcourse,correspondtothetwovariablesofclassicsignaldetectiontheory:perception(dandresponsecriterion(C),respectively(seeWickens,2002).Variablesthatchangedecision-makingbehaviour(e.g.,outcomedensity)maybeduetotheireffectoneitherofthesetwovariables.Asindicatedearlier,depressiverealismisafragilephenomenon.Thereisampleevidenceinthepsychophysicalliteraturethatresponsecri-terioneffects(incontrastwithperceptualeffects)arenotoriouslyvariable,acrossindividualsandexperiments(e.g.,Allan,1968,2002).ThefragilityofdepressiverealismisconsistentwithitslocationTHEQUARTERLYJOURNALOFEXPERIMENTALPSYCHOLOGY,2007,60(3)DEPRESSIVEREALISM strongtestofMatuteÕshypothesissincemostofstudiesconcernedwithdepressiverealismandcon-tingencyjudgementshavenotreportedP(R).However,thefewstudiesthatdidreportP(R)(e.g.,Benassi&Mahler,1985;Dobson&Pusch,1995;Mseftietal.,2007)donotprovidesupportforMatuteÕs(1996)predictioninthattherewasnotadifferenceinP(R)betweenthemoodgroups.WhileMatute(1996)expectedmooddiffer-encesintherelationshipbetweenP(R)andratings,theresponsecriterionhypothesisexpectsmooddifferencesintherelationshipbetweenpre-dictionresponses(whetherexplicitlyrequiredbytheexperimenterorimplicitlymadebythepartici-pant)andratings.Unfortunately,sincepredictionresponseshavenotbeenexplicitlyrequiredintheactivetask,theresponsecriterionanalysiscannotbeappliedretrospectivelytoexistingdata.TheactivetaskcouldreadilybemodiÞedtoincorporatetrialpredictionresponses.Oneachtrial,aftertheparticipantdidordidnotrespond,heorshewouldthenpredictwhethertheoutcomewouldorwouldnotoccur.Accordingtotheresponsecri-terionhypothesis,theparticipantwouldbebiasedtowardspredictingthattheoutcomewouldoccurasP(O)increased.Ifdepressiverealismisacri-terioneffect,thenthebiasshouldbegreaterfornondepressivesthanfordepressives.ItisunclearinthewritingsofAlloy,Abramson,andtheircolleagueswhethertheyplacedthelocusofthedifferenceinthejudgementsofdepressivesandnondepressivesintheperceptionofthecon-tingency,orintheresponsecriterion.Ontheonehand,theytalkaboutanÒillusionofcontrolÓ,andillusionsareusuallyconsideredtobeperceptual.Ontheotherhand,theytalkaboutÒbiasesÓandsuggestthatbothdepressedandnondepressedaresusceptibletobiases,butinoppositedirections,Òwithnondepressivesdistort-ingenvironmentalinformationoptimisticallyanddepressivesdistortingitpessimisticallyÓ(Alloy&Abramson,1988,p.255).ResponsebiasesrefertothecriterionappliedindecidingwhetherornottoanswerÒyesÓorÒnoÓ,giventheperceptualinformationthatcurrentlyisavailable.Wesuggestthatunderstandingdepressiverealismrequiressomeprecisionindistinguishingthesetwoprocesses.WhendepressedpeopledonotsuccumbtotheoutcomedensityeffectitisnotbecausetheyareparticularlyÒrealisticÓ,butratherbecausetheyareparticularlyreluctanttosayÒyesÓ.Thenatureoftheoutcomedensitymanipu-lationinthecontingencyjudgementtaskissuchthatthisbiasofdepressivesisexperienced,bytheexperimenter,asanillusionofaccuracy.FortheITIhypothesis,mooddifferencesareattributedtodifferentialprocessingoftheITIÑacontextprocessingdifferencebetweendepressedandnondepressedpeopleremovesanyobjectivenotionofÔrealismÕthatwasoriginallyemployedtoexplainthedepressiverealismeffectÓ(Mseftietal.,2007).Fortheresponsecriterionhypothesis,mooddifferencesareattributedtodifferentialpropensitiesforpredictingoutcomes.Thus,thetwoaccountsareconceptuallysimilarÑdepressivesandnondepressivesdonotdifferintheirabilitytodetectthedegreeofcontrolthattheirrespondinghasoveroutcomes.SinceitwasÞrstdescribedin1979,thephenom-enonofdepressiverealismhasbeensubjectedtoextensiveempiricalandtheoreticalscrutiny.TherearereportsofbothsuccessesandfailurestoreplicateAlloyandAbramsonÕs(1979)report.Theliteraturedoesnotreadilyfallintoacoherentpicture.Wehaveconcludedthat,ingeneral,theoutcome-densityeffectisbetterunderstoodasachangeinCthanachangeind.Thatis,increas-ingtheoutcomedensitydoesnotaffectthepartici-pantÕsperceptionoftherelationshipbetweenbehaviour(orcue)andoutcome.Rather,foranyparticularcontingency,increasingtheprobabilityofanoutcomeincreasesthetendencyforthepar-ticipanttosayÒyes,theoutcomewilloccurÓ(andtorespondinwaysappropriatetotheimminentarrivaloftheoutcome).THEQUARTERLYJOURNALOFEXPERIMENTALPSYCHOLOGY,2007,60(3)DEPRESSIVEREALISM learningandmotivation(Vol.13,pp.219Ð276).NewYork:AcademicPress.Beck,A.T.(1967).Depression:Clinical,experimental,andtheoreticalaspects(2nded.).NewYork:Harper-Row.Benassi,V.A.,&Mahler,H.I.M.(1985).Contingencyjudgmentsbydepressedcollegestudents:Sadderbutnotalwayswiser.JournalofPersonalityandSocial,1323Ð1329.Bryson,S.E.,Doan,B.D.,&Pasquali,P.(1984).Sadderbutwiser:Afailuretodemonstratethatmoodinßuencesjudgementsofcontrol.JournalofBehaviouralScience,107Ð119.Cheng,P.W.(1997).Fromcovariationtocausation:Acausalpowertheory.PsychologicalReview,367ÐDickinson,A.,Shanks,D.,&Evenden,J.(1984).Judgementofact-outcomecontingency:Theroleofselectiveattribution.QuarterlyJournalofExperimentalPsychology36A,29Ð50.Dobson,K.,&Franche,R.-L.(1989).Aconceptualandempiricalreviewofthedepressiverealismhypothesis.CanadianJournalofBehaviouralScience,419Ð433.Dobson,K.S.,&Pusch,D.(1995).Atestofthedepress-iverealismhypothesisinclinicallydepressedsubjects.CognitiveTherapyandResearch,179Ð194.Herskovic,J.E.,Kietzman,M.,&Sutton,S.(1986).Visualßickerindepression:Responsecriteria,conÞ-denceratings,andresponsetimes.,187Ð197.üi,E.G.,&Cramer,D.(1999).Judgementofcontrolrevisited:ArethedepressedrealisticorCounsellingPsychologyQuarterly95Ð105.Lennox,S.S.,Bedell,J.R.,Abramson,L.Y.,Raps,C.,&Foley,F.W.(1990).Judgmentofcontingency:Areplicationwithhospitalizeddepressed,schizo-phrenicandnormalsamples.JournalofSocialBehaviorandPersonality,189Ð204.Martin,D.J.,Abramson,L.Y.,&Alloy,L.B.(1984).Illusionofcontrolforselfandothersindepressedandnondepressedcollegestudents.JournalofPersonalityandSocialPsychology,125Ð136.Matute,H.(1996).Illusionofcontrol:Detectingresponse-outcomeindependenceinanalyticbutnotinnaturalisticconditions.PsychologicalScience,289Ð293.Miller,E.,&Lewis,P.(1977).Recognitionmemoryinelderlypatientswithdepressionanddementia:Asignaldetectionanalysis.JournalofAbnormal,84Ð86.MsetÞ,R.M.,Murphy,R.A.,&Simpson,J.(2007).Depressiverealismandtheeffectofintertrialintervalonjudgementsofzero,positiveandnegativeTheQuarterlyJournalofExperimental,461Ð481.MsetÞ,R.M.,Murphy,R.A.,Simpson,J.,&Kornbrot,D.E.(2005).Depressiverealismandoutcomedensitybiasincontingencyjudgements:Theeffectofthecontextandinter-trialinterval.JournalofExperimentalPsychology:General10Ð22.Myers,D.G.,&Spencer,S.J.(2004).Socialpsychology(2nded.).Toronto,Canada:McGraw-Hill.Nolen-Hoeksema,S.(2001).Abnormalpsychologyed.).NewYork:McGraw-Hill.Pizzagalli,D.A.,Jahn,A.L.,&OÕShea,J.P.(2005).Towardanobjectivecharacterizationofananhedo-nicphenotype:Asignal-detectionapproach.BiologicalPsychiatry,57,319Ð327.Potts,A.J.,Bennett,P.J.,Kennedy,S.H.,&Vaccarino,F.J.(1997).Depressivesymptomsandalterationsinsucrosetasteperception:Cognitivebiasoratruechangeinsensitivity?JournalofPsychology,57Ð60.Presson,P.K.,&Benassi,V.A.(2003).Aredepressivesymptomspositivelyornegativelyassociatedwiththeillusionofcontrol?SocialBehavior&Personality483Ð495.Schwartz,B.(1981a).Doeshelplessnesscausedepression,ordoonlydepressedpeoplebecomehelpless?CommentonAlloyandAbramson.JournalofExperimentalPsychology:General110,429Ð435.Schwartz,B.(1981b).Helplessness,illusions,anddepression:Finalcomment.JournalofExperimentalPsychology:General,448Ð449.Seligman,M.E.P.(1975).Helplessness:Ondepression,development,anddeath.SanFrancisco:Freeman.Skinner,E.A.(1985).Action,controljudgments,andthestructureofcontrolexperience.,39Ð58.Vallee-Tourangeau,F.,Murphy,R.A.,Drew,S.,&Baker,A.G.(1998).Judgingtheimportanceofcon-stantandvariablecandidatecauses:AtestofthepowerPCtheory.QuarterlyJournalofExperimentalPsychology51A,65Ð84.zquez,C.(1987).Judgmentofcontingency:Cognitivebiasesindepressedandnondepressedsubjects.ofPersonalityandSocialPsychology,419Ð431.Wickens,T.D.(2002).Elementarysignaldetection.NewYork:OxfordUniversityPress.Zuckerman,M.,&Lubin,B.(1965).ManualfortheMultipleAffectAdjectiveCheckList.SanDiego,CA:EducationalandIndustrialTestingService.THEQUARTERLYJOURNALOFEXPERIMENTALPSYCHOLOGY,2007,60(3)DEPRESSIVEREALISM