Rethinkingthepsychologyoftyranny:TheBBCprisonstudyStephenReicher*andS. - PDF document

Rethinkingthepsychologyoftyranny:TheBBCprisonstudyStephenReicher*andS.AlexanderHaslamUniversityofSt.AndrewsUniversityofExeterThispaperpresentsÞndingsfromtheBritishBroadcastingCorporation(BBC)prisonstudyÐanexperimentalcasestudythatexaminedtheconsequencesofrandomlydividingmenintogroupsofprisonersandguardswithinaspeciallyconstructedinstitutionoveraperiodof8days.Unliketheprisoners,theguardsfailedtoidentifywiththeirrole.Thismadetheguardsreluctanttoimposetheirauthorityandtheywereeventuallyovercomebytheprisoners.Participantsthenestablishedanegalitariansocialsystem.Whenthisprovedunsustainable,movestoimposeatyrannicalregimemetwithweakeningresistance.Empiricalandtheoreticalanalysisaddressestheconditionsunderwhichpeopleidentifywiththegroupstowhichtheyareassignedandthesocial,organizational,andclinicalconsequencesofeitherdoingsoorfailingtodoso.OnthebasisoftheseÞndings,anewframeworkforunderstandingtyrannyisoutlined.Thissuggeststhatitispowerlessnessandthefailureofgroupsthatmakestyrannypsychologicallyacceptable.IntheintroductiontohistextonTheRootsofEvil,Staubwrites:‘thewidespreadhopeandbeliefthathumanbeingshadbecomeincreasingly‘civilized’wasshatteredbytheeventsoftheSecondWorldWar,particularlythesystematic,deliberateexterminationofsixmillionJewsbyHitler’sThirdReich’(1989,p.3).Theimpactofthisrealizationwasasmarkedinacademia,andmoreparticularlywithinacademicpsychology,asitwasinsocietyatlarge.Indeed,itisarguablethattheshadowoftheHolocaustliesoverthelasthalfcenturyofsocialpsychologyand,eitherindirectlyordirectly,informsmanyofthe *CorrespondenceshouldbeaddressedtoStephenReicher,SchoolofPsychology,UniversityofSt.Andrews,St.Andrews,Fife,KY169JU,UK(e-mail:sdr@st-andrews.ac.uk).**CorrespondenceshouldbeaddressedtoAlexHaslam,SchoolofPsychology,UniversityofExeter,Exeter,Devon,EX44QG,UK(e-mail:A.Haslam@exeter.ac.uk). TheBritish BritishJournalofSocialPsychology(2006),45,1Ð402006TheBritishPsychologicalSocietywww.bpsjournals.co.uk DOI:10.1348/014466605X48998 paper–howwecometocondonethetyrannyofothersorelseacttyrannicallyourselves.Whiletheliteratureinthesevariousareasisbothbroadandvaried,itispossibletoidentifyatleastonemajortrend.Thatis,therehasbeenashiftawayfromexplanationsthatfocusontheindividualcharacteristicsofthosewhoareprejudiced,discriminatory,orevengenocidal,towardsthosethatconcentrateonthenatureofgroupprocesseswhichcaninducethemostinoffensiveofindividualstocommitthemostoffensiveofacts(Billig,1978;Brown,1965;Milgram,1974;Sherif,1966;Tajfel,1982).Inmanycases,andcertainlywhenitcomestothepsychologyoftyranny,theoristshavetakentheargumentonestepfurtherandproposednotonlythatextremeantisocialbehaviourmustbeanalysedatthegrouplevel,butalsothatgrouppsychologynecessarilytendsinthedirectionofextremeantisocialbehaviour.Whilewestronglyendorsetheneedforagroup-levelpsychologyoftyranny(whichwedeneasanunequalsocialsysteminvolvingthearbitraryoroppressiveuseofpowerbyonegrouporitsagentsoveranother),wewilltakeissuewiththenotionthatgroupspersearetherootoftheproblem.Indeed,wewillarguethatpowerfulandeffectivegroupsprovideaneffectivepsychologicalbulwarkagainsttyrannyandthatitiswhengroupsproveineffectivethattyrannicalformsofsocialorganizationbegintobecomeattractive.Theequationofgroupsandtyrannyhasalonghistorybothwithinsocialthoughtandwithinsocialpsychology.HannahArendt(1998)describestheclassicalviewthat‘therulebymanyisnotgood’andtracesthisbacktoAristotle’scontentionthatcollectiveruleleadstohaphazardness,moralirresponsibility,andisbutadisguisedformoftyranny.Morerecently,suchideasweregivensubstancebycrowdpsychologistssuchasGustaveLeBon(1895/1947)whoarguedthat,throughsubmergenceinthecrowd,individualslosetheirindividualidentityandtheirsenseofresponsibilityandhencebecomecapableofbarbaricandatavisticacts.Thenotionofsubmergencewasdirectlytransposedintothemodernsocialpsychologicalconceptofde-individuation,whichisseentoarisefromanonymitywithinagroup(Postmes&Spears,1998;Reicher,Spears,&Postmes,1995).Asoneinuentialde-individuationresearcherhasputit,‘mythically,deindividuationistheagelesslifeforce,thecycleofnature,thebloodties,thetribe,thefemaleprinciple,theirrational,theimpulsive,theanonymouschorus,thevengefulfuries.’(Zimbardo,1969,p.249).TyrannyasroleandpowerThoughwellknownasade-individuationtheorist,ZimbardoisbetterknownforhisworkontheStanfordprisonexperiment(SPE;Haney,Banks,&Zimbardo,1973;Zimbardo,1989;Zimbardo,Maslach,&Haney,1999).Indeed,thisisoneofthemostfamoussocialpsychologicalinvestigationseverconducted,representingtheculmina-tionofaseriesof‘classic’eldstudiesintotherootsofextremebehavioursthatwereconductedintheaftermathofWorldWarTwo(Milgram,1963;Sherif,1956).Buildingonearlierstudies,itplayedacriticalpartincementingtheshiftthatwehavedescribedfromindividualtogroup-levelexplanationsofextremebehaviours(Banuazizi&Movahedi,1975).Moreover,itwasoneofthefewstudiesthatnotonlyaddressedtheissueoftyrannybut,duetothepoweroftheresearchparadigm,alsoproduceddirectevidenceoftyrannicalbehaviour.Whileitisnot,strictlyspeaking,astudyofde-individuation,ZimbardocertainlyusedhisgeneralunderstandingofthegroupasacorrosiveforcetoexplaineventsintheSPE.StephenReicherandS.AlexanderHaslam TheSPEisrememberedforshowingthat,simplyasaconsequenceofassigningcollegestudentstheroleofguardorprisoner,theformerbecameincreasinglybrutalwhilethelatterbecamepassiveandbegantoshowsignsofpsychologicaldisturbance.Suchwastheseverityofthesephenomenathatthestudy,originallyscheduledtolast2weeks,hadtobestoppedafter6days.Zimbardoandcolleaguesexplainedtheirndingsbycommentingthatguardaggression‘wasemittedsimplyasa‘natural’consequenceofbeingintheuniformofa‘guard’andassertingthepowerinherentinthatrole’(Haneyetal.,1973,p.12).Thus,immersioninagroupisseentounderminetheconstraintsthatnormallyoperateuponpeoplewhentheyactasindividuals.Inaddition,whenthosegroupshavepowerattheirdisposal,thisisbelievedtoencourageextremeantisocialbehaviour(Zimbardo,1969).Althoughthesendingsweresignicantintheirownright,theimpactoftheSPEwasasmuchethicalastheoretical(e.g.seeSmith&Mackie,2000,p.49).Indeed,theveryextremityoftheresultsledmany(includingZimbardohimself)toquestionthelegitimacyofsubjectingparticipantstosuchsituations.Theacceptabilityofconductinganysortoflarge-scaleeldinterventionsthusbecameafocusforvigorousdebate(e.g.Herrera,1997;Lindsay&Adair,1990;Sieber,Iannuzzo,&Rodriguez,1995;Smith&Richardson,1983).Paradoxically,then,atthesametimethattheSPEmarkedtheculminationofpost-wareldstudies,italsoledtotheircessation.Accordingly,sincethe1970s,socialpsychologyhasbeenincreasinglydominatedbylaboratoryexperimentsinwhichthereisminimalornointeractionbetweenparticipantsandscantattentionpaidtotheroleofpersonalandgrouphistoryortothedevelopmentofinteractionsovertime(BarTal,2004;Doosje,Spears,&Ellemers,2002;Haslam&McGarty,2001;Levine,2003;Moreland,Argote,&Krishnan,1996).Moreover,thisunwillingnesstoundertakestudiesthatcreate,manipulateandsystematicallyinvestigatetheeffectsofsocialenvironmentsonhumaninteractioncanbeseentohavecontributedtotheincreasingdominanceofexplanationsbaseduponinherentandessentiallyunavoidablegenetic,biological,orpsychologicalpropensities.Ithasalsoledtoanincreasingdisjunctionbetweentheissuesthatmotivatesocialpsychologicalstudiesandthenatureofthosestudiesthemselves.Researchreports(andcertainlymostbidsforresearchfunding)typicallystartbyalludingtolarge-scaletopicssuchasoppression,discrimination,andgenocide,butthengoontopursueanempiricalstrategythatseemsveryremotefromthesocialrealitiesofsuchphenomena(e.g.seekingtoexplainthesephenomenaintermsofindividual-levelsubconsciousprocessesfromacognitiveor,morerecently,neuroscienticperspective;Ito,Thompson,&Cacioppo,2004).AsZimbardo(quotedinBrockes,2001,p.2)hasargued,partlyasaresultofthesetrends,psychologyhasbecomeincreasinglymarginalto,andmarginalizedfrom,debatesurroundingimportantsocialissues.Intermsofthespecicissueoftyranny,theethicalconcernsthathaveplacedtheSPE‘off-limits’(withtheexceptionofapartialreplicationbyLovibond,Mithiran,&Adams,1979)haveledtoasituationinwhichtheconclusionsofthatstudyhavebecomealmostinviolateandsocialpsychologicalinquiryintotyrannyhaseffectivelygroundtoahalt.BarredfromemployingthepoweroftheSPEparadigm,itisallbutimpossibletoproducebehavioursthatarepowerfulenoughtomatchthosefoundbyZimbardoandhiscolleagues.Hence,evenifresearchersharbourdoubtsabouttheextremesituationaldeterminismandnegativeviewsofthesocialgroup,whichareusedtoexplainthesendings(andmanydo;e.g.seeEllemers,Spears,&Doosje,1999;Turner,Hogg,Oakes,Reicher,&Wetherell,1987;Haslam,2001),ithasnotbeenpossibletoproducedatathatThepsychologyoftyranny cangivesubstancetothosedoubtsandhencereopenscienticdebateaboutthepsychologicalbasesoftyranny.However,quiteapartfromtheintrinsicimportanceofthetopic,thereareatleasttwosetsofreasonswhyrevisitingtheSPEislongoverdue.First,anyassessmentoftheconclusionsdrawnfromtheSPEisinevitablylimitedbythefactthatonlyasmallproportionoftheinteractionsinthestudywererecorded(becauselmingwasintrusiveandlimited)and,ofthese,onlyaverysmallnumberareinthepublicdomain.Moreover,observationaldatawerenevercomplementedbyotherdatasourcesthatwouldallowforcontrolledmeasurementofkeybehavioursandthepsychologicalstatesseentounderliethem.Attheveryleast,thereisaneedforafullerandmoretransparentdataset,whichmightprogressempiricallygroundedandopendebateaboutthepsychologicalbasesoftyranny.However,eventhelimitedamountofdatathatisavailablefromtheSPEcastsdoubtontheanalyticconclusionsthathavebeendrawnfromit.Whereparticipantsdidbehaveinrole,itisunclearwhether,asZimbardoandhiscolleaguesclaim,thiswasduetotheir‘natural’acceptanceofrolerequirementsorduetotheleadershipprovidedbytheexperimenters(Baron,1984;Banuazizi&Movahedi,1975).Thisisbecauseduringthestudy,theguardsweregivenclearguidanceastohowtheyshouldbehave.Notably,whenZimbardobriefedhisguards,hetoldthem:Youcancreateintheprisonersfeelingsofboredom,asenseoffeartosomedegree,youcancreateanotionofarbitrarinessthattheirlifeistotallycontrolledbyus,bythesystem,you,me–andthey’llhavenoprivacy.They’llhavenofreedomofaction,theycandonothing,saynothingthatwedon’tpermit.We’regoingtotakeawaytheirindividualityinvariousways.Ingeneralwhatallthisleadstoisasenseofpowerlessness(Zimbardo,1989).TheimportanceofsuchguidanceisdemonstratedbytheresearchofLovibondetal.(1979)whoconductedastudyinwhichtheguardsweretrainedtorespecttheprisonersasindividualsandtoincludethemindecision-makingprocesses.Undertheseconditions,theensuingbehaviourofbothguardsandprisonerswasfarlessaggressiveandextreme(Lovibondetal.,1979).Yet,evenwithguidance,manyoftheparticipantsintheSPEbehavedoutofroleformuchofthetime(Baron,1984).Theavailablevideomaterialshowsthatbothprisonersandguardschallengedtheirrolesnotonlyatthestart,butthroughouttheentirestudy.Inthecaseoftheguards,Zimbardo(1989)notesthat,whilesomeexploitedtheirpower,otherssidedwiththeprisonersandyetothersweretoughbutfair.Suchdiversitysitsuneasilywiththenotionthatroleacceptanceissimplydeterminedbythesituation.Itsuggeststhattheemphasisonroleacceptanceandtyrannyisone-sidedandthatthereisaneedtofocuson(a)theconditionsunderwhichpeopledoordonotassumetheirrolesand(b)thebalancebetweentyrannyandresistanceAnalternativeanalysis:ThesocialidentityapproachItisnotonlythatsomeofthedatafromtheSPEappeartosituneasilywitharoleaccount.Increasingly,theroleaccount–andindeedthegenerallynegativeviewthatgroupmembershipleadstoalossofconstraintsonantisocialbehaviour–isatoddswithdevelopmentsingrouppsychology.Oneofthemostinuentialoftheseisthesocialidentityapproach(incorporatingsocialidentitytheory;Tajfel,1978,1982;Tajfel&Turner,1979;andself-categorizationtheory;Turner,1985;Turneretal.,1987;Turner,Oakes,Haslam,&McGarty,1994).Accordingtothisapproach,peopledonotStephenReicherandS.AlexanderHaslam automaticallyactintermsofgroupmemberships(orroles)ascribedbyothers.Rather,whetherornottheydosodependsuponwhethertheyinternalizesuchmembershipsaspartoftheself-concept(Turner,1982).Self-categorizationtheoryinparticularhasarguedthatthisactofself-denitionintermsofgroupmembership(socialidentication)formsthepsychologicalbasisofgroupbehaviourandthatthecharacterofsuchbehaviourdependsuponthenorms,values,andunderstandingsthatcharacterizetheparticularcategoryinquestion(Turner,1982,1999).Thus,whilemembersofcertaingroupsmayindeedusetheirpowertoactindiscriminatoryandoppressiveways,membersofothergroupsmayactmoreprosociallyandusetheirpowerforconstructivepurposes(Pfeffer,1981;Postmes&Spears,1998).Moreover,evenifsomegroupsaretyrannical,groupactionisalsothebasisonwhichpeoplegainthestrengthandcondencetoresist,tochallenge,andeventooverthrowtyranny(Reicher,1996;Tajfel,1978).Consistentwithsuchemphasis,thegreaterpartofearlyworkinformedbysocialidentitytheoryhasfocusedontheconditionsunderwhichpeopleacttochangeinequalitiesbetweengroups(e.g.Robinson,1996).Inbroadterms,itisassumedthatpeoplewhoarepositivelyvaluedbyvirtueoftheirgroupmembership(e.g.membersofdominantgroups)wouldidentifywithandactintermsofthegroup.Forpeoplewhoarenegativelyvaluedbyvirtueoftheirgroupmembership(e.g.membersofsubordinategroups),collectiveactioniscontingentupontwosetsoffactorsinparticular(Tajfel,1978;Tajfel&Turner,1979).Therstrelatestobeliefsaboutone’sabilitytoadvancethroughthesocialsystemdespiteone’sgroupmembership(i.e.thepermeabilitycategoryboundaries).Thesecondconcernstheperceivedofintergrouprelationsandcomprisestwofurtherelements:theperceivedfairnessofintergroupinequalities(legitimacy)andtheirperceivedstability.Whenrelationsareperceivedtobeinsecure,thisischaracterizedbythefactthatindividualsareawareofalternativestothestatusquoandhencecanenvisagespecicwaysinwhichitcouldbechanged.Permeabilityaffectswhetherpeopleactindividuallyorcollectively,sothatabeliefthatmovementacrossboundariesispossibleencouragesstrategiesofindividualmobility,butabeliefthatsuchmovementisimpossibleencouragespeopletoperceivethemselvesandactasgroupmembers(e.g.Ellemers,1993;Wright,Taylor,&Moghaddam,1990).Whetherornotpeoplethenchallengeinequalityisalsodependentuponintergrouprelationsbeingperceivedasinsecure.Thatis,peopleshouldbemostinclinedtoresistdominationwhentheyseeinequalityasbothillegitimateandunstableandcanthusenvisagecognitivealternativestoit(Turner&Brown,1978;seealsoEllemers,Wilke,&vanKnippenberg,1993;Tajfel,1978;Tajfel&Turner,1979).Itisimportanttostressthatsocialidentitytheorydoesnotconstituteacomprehensivetheoryofdominationandresistance.Mostnotably,ithaslittletosayabouttheconcomitantsofidentityprocesses(e.g.organizationalandclinicalfactors),whichmayimpactupontheabilityofgroupmemberstoacteffectively.Thesearecriticalissuesthatweaimtoinvestigatehere.Nonetheless,thesocialidentityapproachprovidesawell-articulatedandcontemporaryperspectivefromwhichtorevisittheissuesraisedbytheSPE:Whatarethepsychologicalconsequencesofintergroupinequality?Whendopeopleseektoimposesuchinequality?Andwhendotheyresistit?Beforeexplaininghowweaddressedtheseissues,itisimportanttoconsiderasecondsetofreasonsforrevisitingtheSPE.Thesehavelesstodowiththeexplanationofthendingsthemselvesthanwiththeirbroadersocialrelevance.ForZimbardoandhiscolleagues,theresultsoftheSPEwereintendedtobeardirectlyonthenatureofThepsychologyoftyranny prisonregimesintheUnitedStates.Thus,theyrefertothesettingasa‘simulatedprison’andclaimthat‘thissimulatedprisonenvironmentdevelopedintoapsychologicallycompellingprisonenvironment’(Haneyetal.,1973,p.69).However,asBanuaziziandMovahedi(1975)pointout,therearegoodreasonstodoubttheseclaims,rstly,becausetherearecriticalfeaturesoftheSPEthatareverydifferentfromarealprison(e.g.participantsknowtheyhavecommittednocrimeandcanasktoleaveatanytime)and,secondly,becauseevenwheretherearematerialsimilarities,thetwoareverydifferentphenomenologically(e.g.wherethewallsofaprisonremindtheinmatesthattheymustbekeptapartfrom‘decent’people,theyremindtheparticipantsintheSPEthattheyarehonourableparticipantsinadventurousscienticresearch).Whiletheseargumentsareinthemselvescontroversial(e.g.Thayer&Saarni,1975),ourpointisthat,eveniftheyareaccepted,theydonotrendertheSPEpracticallyirrelevant.Forthereisanotherlevelbeyondthatofphenomenalequivalencethroughwhichthestudycan(andis)claimedtohavereal-worldimplications.Thatis,asisthecasewithmostpsychologicalresearch,generalizationistheoretically,notempirically,based(Haslam&McGarty,2004;Turner,1981).Thus,Zimbardo(e.g.2001)useshisstudytoestablishthetheoreticalclaimthatpeople‘naturally’assumeroles,andthenusesthistheoreticalanalysistoexplainawiderangeofphenomenafromprisonbehaviourtoterrorism.Ofcourse,theresultantbehaviourswillnotbethesameasthoseobservedintheSPE,butwilldependupontheprecisenatureoftherolesandrolerequirementsintherelevantdomainofapplication.Itispreciselythisabilitytoapplyroletheorytoabroadrangeofdomains,despitephenomenalandbehaviouraldifferences,thathasensuredtheimpactoftheSPEwithinandbeyondpsychologyamongstthosewhohavelittleinterestinmattersrelatingtoprisons.Itisthevalidityofthistheoreticalaccountthatwewishtoaddressinthepresentpaper,notonlyfortheempiricalandtheoreticalreasonsdiscussedabove,butalsobecausewefeelthatthetraditionalanalysisoftheSPEhasprofoundandtroublingsocialimplications.Ifpeoplecannothelpbutactintermsofassignedrole,itimpliesthattheyhavelittlechoice,andhencelittleresponsibility,fortheirsocialactions.Thismakesitmoredifculttoholdtyrantstoaccountforwhattheydo.Moreover,incommunicatingthemessagethatresistanceisfutile,theanalysisdiscouragestheoppressedfromattemptingtochallengetyranny.TheBBCprisonstudyInDecember2001,weconductedamajorsocialpsychologicaleldstudy(possiblythelargestsuchstudyinthethreedecadessincetheSPE)inconjunctionwiththedocumentariesunitofBritishBroadcastingCorporation(BBC;hencethetitle‘theBBCprisonstudy’).Thestudywasindependentlydesigned,operationalized,run,andanalysedbytheauthors.Inthissense,itwasanordinarypieceofscienticresearch,whichwentthroughallthenormalscienticprocedures(includingethicalapproval),andwhosefeaturesweredesignedinrelationshiptothetheoreticalissuesthatconcernedus.ThecontributionoftheBBCwastocoordinateandmanagethelogisticaltaskof(a)creatingthestudyenvironment(inlinewithourguidelines),(b)lmingthestudy,and(c)preparingsomeoftheresultantmaterialforbroadcast.Inshort,theprojectcanbedescribedas‘originalsciencelmed’.Thismadeitauniquecollaborationthatwasmarkedlydistinctfrom‘realitytelevision’enterprisesinwhichatelevisioncompanydevisescertainscenarioswithissuesofentertainmentinmindandtheninvitesacademicstocommentonthem.StephenReicherandS.AlexanderHaslam Overaperiodof8days,thestudyexaminedthebehaviourof15menwhowereplacedinasocialhierarchyofguardsandprisonerswithinapurpose-builtenvironment.Theirbehaviourwasvideo-andaudio-recordedovertheentireperiod,andthiswascomplementedbydailypsychometricandphysiologicalmeasures.Thevideodatawereeditedintofour1-hourlongdocumentariesscreenedinMay2002(Koppel&Mirsky,2002).Theaimofthestudywasnottosimulateaprison(which,asintheSPE,wouldhavebeenimpossibleonethicalandpracticalgrounds)butrathertocreateaninstitutionthatinmanywaysresembledaprison(butalsootherhierarchicalinstitutionssuchasaschool,anofce,abarracks;seeMorgan,1979)asasitetoinvestigatethebehaviourofgroupsthatwereunequalintermsofpower,status,andresources.Whatiscritical,then,isnotthatthestudyenvironmentreplicatedarealprison(whichnosuchenvironmentevercould),butthatitcreatedinequalitiesbetweengroupsthatwererealtotheparticipants.Similarly,ouraimwasnotconductanexactreplicationoftheSPE(which,wouldalsohavebeenimpossibleforethicalreasons).Rather,itwastouseadifferentsystemofintergroupinequalityinordertorevisittheconceptualissuesraisedbytheSPE.WethereforedonotinvitecomparisonwiththeSPEintermsoftheexactdetailsofhowpeoplebehavedbutratherintermsoftheabilityofdifferentexplanatoryframeworkstomakesenseofwhathappened.Doparticipantsaccepttheirrolesuncritically?Dothoseaccordedgrouppowerexerciseitwithoutconstraint,anddothosewithoutgrouppoweraccepttheirsubordinationwithoutcomplaint?Afterall,iftheprocessofroleenactmentisindeed‘natural’,thenitshouldapplyinallcasesandanyexceptionistroublingfortheoverallclaim.Dotheconceptsusedbysocialidentityandself-categorizationtheoristsprovideamoresatisfactoryaccountofwhenpeopledo(anddonot)adoptthesocialpositionsascribedtothem?Ourstudycanthusbeseenasanexperimentalcasestudyofthebehaviourofmembersindominantorsubordinatepositionsandofthedevelopingrelationsbetweenthem.UnliketheSPE,itisnotpurelyexploratorybutratheristheoreticallyinformed(byasocialidentityperspective).Hence,andagainincontrasttotheSPE,weincludedmanipulationsoftheoreticallyrelevantvariables.Giventhepracticalimpossibilityofrunningmultiplesetsofgroups(duetothemassiveresourcesrequiredintermsofequipment,personnel,andmoney),atime-seriesapproachwasadoptedwherebyinterventionswereintroducedatpredenedpointsinthestudyandtheireffectsonthedevelopmentofintra-andintergrouprelationsthenanalysed.Tosummarize,theoverallaimsofourstudywereasfollows:(a)Toprovidecomprehensiveandsystematicdatapertainingtotheunfoldinginteractionsbetweengroupsofunequalpowerandprivilege.(b)Toanalysetheconditionsunderwhichpeople(i)denethemselvesintermsoftheirascribedgroupmembershipsandactintermsofgroupidentities,and(ii)acceptorelsechallengeintergroupinequalities.Specically,wepredictthatdominantgroupmemberswillidentifywiththeirgroupfromthestartandimposetheirpower.However,subordinategroupmemberswillonlyidentifycollectivelyandchallengeintergroupinequalitiestotheextentthatrelationsbetweengroupsareseenasimpermeableandinsecure.(c)Toexaminetherelationsbetweensocial,organizationalandclinicalfactorsingroupbehaviour.(d)Todevelopprotocolsthatprovideapracticalandethicalframeworkforexaminingsocialpsychologicalissuesinlarge-scalestudies.Thepsychologyoftyranny MethodandethicsThestudywasdesignedtocreateahierarchicalsocietyinwhichpeoplewouldliveforupto10days.ItwasconductedwithinaninstitutionalenvironmentthatwasconstructedinsideElstreeFilmStudiosinnorthLondon.Prisonerswereallocatedtolockable3-personcellsthatwerelocated,togetherwithshowers,offacentralatrium.Thiswasseparatedbyalockablesteelmeshfencefromtheguards’quarters(adormitory,bathroom,andmessroom).AplanoftheprisonispresentedinFigure1.ComprehensivedetailsoftheproceduresareavailableinHaslamandReicher(2002)orfromtheauthors.Whatfollowsisadescriptionofthekeyfeaturesofthestudy.Thestudyaimedtocreateasystemofintergroupinequalitythatwasmeaningfulbutwasnotharmfultoparticipantseitherphysicallyormentally.Toensurethatnoharmeventuated,detailsoftheexperimentalset-upandplannedmanipulationswerediscussedwithcolleaguesandsubmittedbothtotheUniversityofExeter’sethicspanelandtotheChairofTheBritishPsychologicalSociety’sEthicsCommitteepriortothestudybeingconducted.Thenoveltyoftheexperimentalmanipulationsandtheoreticalanalysisconstitutedcentralcomponentsofthescienticcasethatwaspresentedinordertojustifytheresearch.Moreover,thefollowingsafeguardswerebuiltintothe(a)Potentialparticipantswentthrough3-phaseclinical,medical,andbackgroundscreeningtoensurethattheywereneitherpsychologicallyvulnerablenorliabletoputothersatrisk(seebelow). walkway(first floor) MainatriumGuards'dorm booth Cell 1 mess bathroom g FCgBGg TMgTQg) p GPp NPp p KMp PPp Figure1.Planoftheprison.StephenReicherandS.AlexanderHaslam (b)Participantssignedacomprehensiveconsentform.Amongstotherthings,thisinformedthemthattheymaybesubjecttoaseriesoffactors–includingphysicalandpsychologicaldiscomfort,connement,constantsurveillanceandstress–whichmayinvolverisk.(c)Twoindependentclinicalpsychologistsmonitoredthestudythroughout,andhadtherighttoseeanyparticipantatanytimeortodemandthatanyparticipantberemovedfromthestudy.(d)Aparamedicwasonconstantstandbyincaseofillnessorinjury.(e)On-sitesecurityguardswereprovidedwithdetailedprotocolsclarifyingwhenandhowtointerveneincasesofdangerousbehavioursbyparticipants.(f)Anindependent5-personethicscommittee–chairedbyaBritishMemberofParliament–monitoredthestudythroughout.Thiscommitteehadtherighttodemandchangestothestudy’sset-uportoterminateitatanytime.Apartfromminorailments(blisters,etc.)thatweretreatedbytheparamedic,nointerventionswerenecessarytoaddressethicalconcerns.Afterthestudy,theethicalcommitteepublishedanindependentreport(McDermott,Opik,Smith,Taylor,&Wills,2002)andcharacterizedtheconductofthestudyas‘exemplary’.SelectionofparticipantsMaleparticipantswererecruitedthroughadvertisementsinthenationalpressandthroughleaets.Applicantswentthroughthreephasesofscreening.First,theycompletedabatteryofpsychometricteststhatmeasuredbothsocialvariables(authoritarianism,socialdominance,modernracism)andclinicalvariables(depression,anxiety,socialisolation,paranoia,aggressiveness,demotivation,self-esteem,self-harm,drugdependence).Second,theyunderwentafullweekendassessmentbyindependentclinicalpsychologists.Third,medicalandcharacterreferenceswereobtained,andpolicecheckswereconducted.Forethicalreasons(notedabove),wesoughttoincludeinthestudyonlypeoplewhowerewell-adjustedandprosocial,scoringatlowlevelsonallsocialandclinicalmeasures.Additionally,wewishedtoensurethattheindividualdispositionsofourparticipantsweresuchthat,ifthedynamicsofthestudyproducedantisocialactionsinthissample(asintheSPE),thenitcouldreasonablybesupposedthattheywouldhavesuchaneffectuponalmostanybodyinthepopulation.Thescreeningreducedaninitialpoolof332applicantsto27men(werecruitedonlymentoensurecomparabilitywiththeSPEandtoavoidethicalissuesthatwouldarisefromplacingmenandwomentogetherincells).Thenalsampleof15waschosentoensurediversityofage,socialclass,andethnicbackground.Theywererandomlydividedintotwogroupsof5guardsand10prisonersbutinsuchawayastoensurethatthetwogroupswerematchedonkeydimensions.Morespecically,the15participantswererstdividedintovegroupsof3peoplewhowereascloselymatchedaspossibleonpersonalityvariablespotentiallyimplicatedintyranny:modernracism,authoritar-ianismandsocialdominance.Fromeachgroupofthree,oneparticipantwasthenrandomlyselectedtobeaguard(andtheremainingtwotobeprisoners).Thisprocedure Thisprocedure–whichensuresequivalenceontheoreticallyrelevantvariables–wasconsideredsuperiorto‘pure’randomassignment(i.e.onewhichdoesnottakeaccountofanyindividualdifferences,asintheSPE),asthestudy’srelativelysmallsamplesizemeansthatrandomassignmenthasthepotentialtoproducelargedifferencesbetweengroups(duetothelawoflargenumbers;e.g.Haslam&McGarty,2003,pp.180–183).Thepsychologyoftyranny wasconductedblind(i.e.theidentitiesoftheparticipantswerenotknowntotheexperimenters).Note,however,thatduetoourexperimentalmanipulations(explainedbelow),theactualnumberofpeopleinthetwogroupsvariedovertime.Thus,atthestartofthestudy,therewere5guardsbutonly9prisoners–the10thbeingintroducedatalaterstage.DatasourcesTheprisonenvironmentwasdesignedinsuchawaythatparticipantscouldbebothvideo-andaudio-recordedwherevertheywere.Atalltimes,fourchannelswerevideo-recordedandallaudiochannelswererecorded.Therewasalsodailypsychometrictesting.Measuresweretakenfromabatteryofscales.Forthepurposesofthepresentanalysesthecriticalmeasureswere:socialvariables:socialidentication,awarenessofcognitivealternatives,right-wingauthoritarianism;organizationalvariables:compliancewithrules,organizationalcitizenship;andclinicalvariables:self-efcacy,depression.Inordertominimizefatigue,noteverymeasureinthefullbatterywasadministeredeveryday.However,eachwasadministeredonmultipleoccasionstoallowforananalysisofdevelopmentovertime.Finally,dailyswabsofsalivaweretakeninordertoascertaincortisollevels(asanindicationofstress;Laudatetal.,1988).Forreasonsofspace,however,wewillnotconsiderthecortisoldatahere(seeHaslam&Reicher,inpress).Fiveparticipantswereinvitedtoahoteltheeveningbeforetheyenteredtheprison.Onarrival,theyweretoldthattheywouldbeguardsinthestudy.Theywereshowntheprisontimetable–whichincludedsuchelementsascleaningchores,workduties,prisonerrollcalls,exercisetimeandarecreationalhour–andweretoldthattheirresponsibilitywastoensurethattheinstitutionranassmoothlyaspossibleandthattheprisonersperformedalltheirtasks.Theveguardswerethenaskedtodrawupaseriesofprisonrulesunderheadingsprovidedbytheexperimentersandtodrawupaseriesofpunishmentsforruleviolations.Theguardsweregivennoguidanceabouthowtheyshouldachievetheirgoals.Theonlylimitsonwhattheycoulddowereasetofethicallydetermined‘basicrights’forprisoners.Inparticular(andasintheSPE),allparticipantsweretoldthatphysicalviolencewouldnotbetolerated(forrules,punishments,andrights,seeHaslam&Reicher,2002).Beyondthis,however,itwasstressedthattheguardscouldactastheypleased.Onthemorningofthestudyitself,theguardsweretakeninablacked-outvantotheprison(sincethiswasmeanttobetheirentireexperientialworldforthedurationofthestudy,itwasimportantthattheycouldnotimaginetheoutside).Onceinside,theyweregivenafullbriengbytheexperimentersontheprisonlayoutandtheresourcesavailabletothem.Theguardshadaseriesofmeansbywhichtoenforcetheirauthority,includingkeystoalldoorsinsidetheprison(includingapunishmentisolationcell),soleaccesstoanStephenReicherandS.AlexanderHaslam upperlevel,a‘guards’station’withasurveillancesystemfromwhichtheycouldseeintotheprisoners’cells,resources(includingsnacksandcigarettes)touseasrewardsorwithdrawaspunishments–and,inaddition,theabilitytoputprisonersonabreadandwaterdiet.Theyalsohadfarbetterconditionsthantheprisoners,includingsuperiormeals,extrasuppliesofdrinksandsnacks,superiorlivingconditionsandwell-madeuniformsasopposedtotheprisoners’uniformofat-shirtprintedwitha3-digitnumber,loosetrousersandimsysandals.Theprisonersalsohadtheirhairshavedonarrival.Aftertheirbrieng,theguardschangedintotheiruniformsandpracticedtheprocedureforadmittingtheprisoners.Thenineprisonersthenarrivedoneatatime.Theyweregivennoinformationapartfromtheprisonrules,alistofprisoners’rights,whichwaspostedintheircells,andaverybriefloudspeakerannouncementfromtheexperimenters.Thisintroducedthepermeabilityintervention(seebelow)andstressedthatviolencewasnotpermissible.PlannedinterventionsAttheirinitialbrieng,theguardsweretoldthattheyhadbeenselectedonthebasisoftheirreliability,trustworthinessandinitiativeasgleanedfrompre-selectionassessmentscales.However,theywerealsotoldthatwhilethesescaleswerereasonablyreliable,theywerenotperfect.Inparticular,theexperimentersstatedthatitwaspossiblethattheyhadmisassignedoneormoreoftheprisoners.Hence,theguardsweretoldthattheyshouldobservethebehaviouroftheprisonerstoseeifanyoneshowedguard-likequalities.Iftheydid,theyweretoldthattherewasprovisionforapromotiontobemadeonDay3.Thisinformationwasalsoannouncedtotheprisonersovertheloudspeaker.Intheinitialdaysofthestudy,participantswerethusledtobelievethatmovementbetweengroupswaspossible(seealsoEllemers,vanKnippenberg,&Wilke,1990;Wrightetal.,1990).Afterthepromotionofoneprisonertoguardactuallytookplace(theselectionoftheindividualbeingmadebytheguardsonthebasisofaproceduresuggestedbytheexperimenters),thepossibilityofmovementwasremovedbyannouncingthattherewouldbenofurtherpromotions(ordemotions).LegitimacyThreedaysafterthepromotion,participantsweretobeinformedbytheexperimentersthatobservationshadrevealedthattherewereinfactnodifferencesbetweenguardsandprisonersonthekeygroup-deningqualities.However,theywouldbetoldthatitwasimpracticaltoreassignthemandhencethegroupswouldbekeptastheywere.Accordingly,whereaspreviouslythegroupdivisionhadbeenlegitimate,thiswouldnolongerbethecase(Ellemersetal.,1993).CognitivealternativesWithinadayofthelegitimacyintervention,anewprisonerwastobeintroduced.Althoughhewasasnaveastheothers,hewaschosenforthisrole(fromthepoolof10participantsrandomlyassignedtobeprisoners)becauseofhisbackgroundasanexperiencedtradeunionofcial.Onthisbasis,weexpectedthathewouldintroduceanewperspectivetotheprisonbasedonnotionsofgroup-basednegotiationandcollective-andequalrights(i.e.aperspectivethatsuggestedtheexistingregimewasbothillegitimateandchangeable).ItwasalsothoughtthathemightprovideskillsThepsychologyoftyranny necessarytoorganizecollectiveaction(e.g.seeHaslam,2001).Hence,itwasenvisagedthathisintroductionwouldenabletheprisoners(andtheparticipantsmoregenerally)toenvisagetheachievementofamoreequalsetofsocialrelations.Thendingsofthestudycanbedividedintotwophases.Intherstphase,theguardsfailedtoidentifywitheachotherasagroupandtocoherecollectively.Bycontrast,afterthepromotiononDay3,theprisonersdidincreasinglyidentifyasagroupandworkcollectivelytochallengetheguards.Thisledtoashiftofpowerandultimatelytothecollapseoftheprisoner–guardsystem.Inthesecondphase,participantsdecidedtocontinueasasingleself-governing‘commune’.However,theywereunabletodealwithinternaldissentandlostcondenceinthecommunalsystem.Bytheendofthestudy,theywereincreasinglydisposedtotolerateanewandmuchmoredraconiansystemofinequalitythatsomeparticipantsnowwishedtoimpose.Resultsthatpertaintothesetwophaseswillbepresentedinturn.Foreachphase,wecombineadescriptionbasedontheobservationaldatawithstatisticalanalysesbasedonthequantitativemeasures.Thestatisticalanalysesarebasedonthedataofindividualparticipantswithingroups(excludingdatafromtheparticipantwhowaspromotedfromprisonertoguardonDay3andfromtheprisonerwhowasintroducedonDay5).Giventheinteractionbetweenparticipants,itcouldbearguedthatthegroup,ratherthantheindividualgroupmember,shouldbetheunitofanalysishere.Forthisreason,thepresentdatawerealsoanalysedusingmethodsadvocatedbyMcGartyandSmithson(2005),whichdonotassume(orrequire)independenceofobservations.Theseanalysesconrmedthereliabilityofallthepatternsreportedbelow.However,forreasonsofspaceandinlightofthenoveltyofthesealternativemethods,weaccordwithgeneralusagebypresentingstatisticsthatarecommonlyusedevenwherethereisinteractionbetweenparticipants(Hoyle,Georgesen,&Webster,2001).Phase1:RejectinginequalitySocialidenticationFortheprisoners,thedevelopmentofsocialidenticationwasconsistentwithpredictions.Fromthestart,theywereclearlydissatisedbytheirinferiorconditions.Initiallythough,aspredictedinlightofthepermeabilityofgroupboundaries,manysoughttoimprovetheirlotbydisplayingtheindividualqualitiesnecessaryforpromotionasopposedtomountingacollectivechallengetotheguards.Asaresult,therewasnosharedidentityamongtheprisonersandnoconsensusabouthowtheyshouldbehave(seePhotograph1).However,afterthepromotion,whengroupboundarieswereimpermeableandparticipantscouldonlyaltertheirpositionbychangingthegeneralprisoner–guardrelationship,theprisonersbegantodevelopamuchstrongersenseofsharedsocialidentityandtodevelopmoreconsensualnorms–particularlyinrelationtotheirtreatmentoftheguards.Thiscontrastbetweenthepre-andpost-promotionperiodsisexempliedinexchangesbetweenparticipantsinCell2.Beforethepromotion,twooccupantsofthiscell,JEandKM,workedconscientiouslyandexplicitlysoughttoimprovetheirpositionbydisplayingbehavioursrequiredtobecomeaguard.AsJE Thesubscriptsafterparticipants’initialsdenotetheirstatusasprisonerorguard,respectively.StephenReicherandS.AlexanderHaslam it,‘I’dliketobeaguardbecausetheygetalltheluxuriesandwearenot’.However,almostimmediatelyafterthepromotion,allthreeoccupantsofthiscell(i.e.including)recognizedthattheonlywaytoimprovetheirpositionwastochangethesystem.Accordingly,theybegantodiscusshowtheycouldachievethistogether.Ifanyoneexpresseddoubtsaboutthisobjective,theywereremindedofthecollectiveinequalitiesintheprisonandoftheneedforcollectiveresistance.Thisisexempliedbythefollowinginterchangebetweenthecellmates:Extract1:Hopefullywe’llget[TQ]in.That’stheperson,he’sthetarget.:No.ImeanobviouslyIthinkit’sgoingtobealotoffunforustodothisbutIdon’tthink[TQIfeelsoIjustfeel::::Listen,listenmateI,you’vegotto,you’vegottostartforgettingaboutotherpeople’sfeelingsandwhatthey’redoingbecausethedayswhenyou’resittingherestarvinghungryandyou’vegotfuckallandyou’vegotnothingmateandyou’vegotarattylittlebedandastupidlittleblankettositunderandthey’reunderthereintheirduvets,they’vegoteverythingtheywantandthey’renotgivingtwofucksaboutyou.So–thinkonandfuckthem:Ithinktheydocareaboutus.ButguysI’mgoingtobackyoualltheway.Youshouldno’doubtme.Fortheprisoners,then,thepromotionledtoaperceptionofimpermeability.Thiswasaccompaniedbyashiftfromindividualactionandidentication(i.e.astressonwhat‘I’willdo)tocollectiveactionandidentication(i.e.discussionofwhat‘we’willdo)andfromcompliancetoconictwiththeguards.Fortheguardsthemselves,theresultswereverydifferent.Moreover,thepatterningoftheirsocialidenticationwentagainstourpredictions.AlonglinesreportedbyZimbardoandhiscolleagues(Haneyetal.,1973),wehadexpectedthat,fromtheoutset,theywouldidentifywithwhatwasahigh-statusandpositivelyvaluedgroupwithintheprison.Therewassomeevidenceofthisintherstdayofthestudy,butitwasalsoclearthatseveralguardswerewaryofassumingandexertingtheirauthority.Consequently,someguardswerealwaysambivalentaboutinternalizingtheirassignedidentity.Eventotheextentthattheyovercamethisambivalence,thegroupasawholecouldneverreachconsensusaboutitsnormsandpriorities.Inaddition,over Photograph1.Day2:TheprisonerscomplywiththeguardsÕregime.Thepsychologyoftyranny time–particularlyoncetheboundariesbetween-groupsbecameimpermeableandtheprisonershaddevelopedastrongersenseofsharedidentity–thesecleavageswereexacerbatedandthefragilityoftheguards’collectiveidentitybecamemoreapparent.However,asnotedabove,incontrasttotheSPE,oursystematiccollectionofquantitativedataallowedustocomplementbehaviouralobservationsthatspeaktothesepatternswithinferentialstatisticalanalysis(andhencetotriangulatebehaviouralandpsychometricdata).Tothisend,socialidenticationwithprisonerandguardgroupswasmeasuredeverydaybymeansoftwo3-itemscalescontainingitemsusedwidelyinpreviousresearch(e.g.Doosje,Ellemers,&Spears,1995;‘Ifeelstrongtieswiththeprisoners[guards]’/‘Iidentifywiththeprisoners[guards]/Ifeelsolidaritywiththeprisoners[guards])’.Aswithallothermeasures,responsesweremadeon7-itemscaleswithappropriatelylabelledend-points(e.g.notatall,extremelyIdenticationscoreswerecomputedbyaveragingresponsesonthethreemeasuresthatcomprisedeachscaleandsubtractingtheout-groupidenticationscorefromthein-groupidenticationscore.Notetoothat,aswithallothermeasures,datawerecollectedearlyeachmorning.Accordingly,scoresrelateprimarilytoeventsofthepreviousday.Giventhesmallnumberofparticipants,thestudyinevitablyhasalowlevelofstatisticalpower.Inlightofthis,statisticalsignicancewasconsideredinconjunctionwitheffectsize(Cohen,1977;Smithson,2000).Forallanalyses,effectswereonlyconsideredmeaningfulwhen,aswellasbeingsignicantatconventionallevels.05),effectsizewasstrongbyCohen’s(1977,p.283)criteria(i.e.Figure2presentsmeansocialidenticationscoresasafunctionofparticipantgroupandtime.Ascanbeseenfromthisgraph,thedatahereareconsistentwiththeaboveobservations.Intheveryrstdaysofthestudy,theguardsidentiedmorestronglywiththeirsocialpositionthantheprisoners.However,assoonastheyhadtoimplementthedisciplinaryregime(fromDay2onwards),theguards’identicationfellwhilethatoftheprisonersrose.Prisoneridenticationwasalsoparticularlyhighafterthepromotion(Day4).Statisticalanalysisofin-groupidenticationscoreswasconductedbymeansofa2(participantgroup:guards,prisoners)6(studyphase:Day1toDay6)ANOVAwithrepeatedmeasuresonthesecondfactor.Thisanalysissupportedobservationaldatainrevealingasignicantandlargeinteractionbetweengroupandtime,(5,55).22.Analysesofpolynomialtrendswereusedasthemostpracticalwayofdecomposingthiseffect(Norusis,1985).Consistentwithobservations,theseshowed 0.511.522.5 Day 1Day 2Day 3Day 4Day 5Day 6 Prisoners Figure2.SocialidentiÞcationasafunctionofassignedgroupandtime.StephenReicherandS.AlexanderHaslam thatsocialidenticationamongtheprisonersincreasedlinearlyovertime,.05.Ontheotherhand,identicationamongtheguardsdeclinedasthestudyprogressed,butnon-signicantly,0.77,SecurityofintergrouprelationsAfterthepromotion,thenormativeconsensusamongprisonersledtoeffectiveorganizationbasedonboththeexpectationandtheprovisionofmutualsocialsupportbetweengroupmembers(Haslam,O’Brien,Jetten,Vormedal,&Penna,2005;House,1981;Underwood,2000;seeHaslam&Reicher,inpress,forafullertreatmentofthispoint).Conversely,theguards’inabilitytoagreeongroupnormsandprioritiesmeantthattheycouldnottrustotherstorepresentthemoractappropriately.Asaconsequence,theyfelt(andwere)weak,inconsistent,andineffectiveasagroup.Thesedevelopmentsdirectlyunderminedtheperceivedlegitimacyoftheintergroupinequalityastheguardsshowednoevidenceofthequalitiesbywhichtheirselectionwassupposedlyjustied.Suchdevelopmentsalsocontributeddirectlytotheemergenceofcognitivealternatives;notonlycouldtheprisonersenvisagechangestotheexistinghierarchy,buttheyalsohadagrowingcondenceintheirabilitytoachievethem.Consider,forinstance,theoutcomeoftherstconfrontationbetweenprisonersandguards.ThethreeprisonersofCell2hadorchestratedthismainlytoseehowtheguardswouldrespondwhenchallenged.Aspartofapre-arrangedplan,JEthrewhislunchplatetothegroundanddemandedbetterfood.Astheguardstriedtointervene,KMandPPthenjoinedinwithfurtherdemandsforsmokingrightsandtreatmentforablister.Theguardsweretotallydisunitedintheirresponse,withsomewantingtotakeadisciplinarianlineandotherswantingtomakeconcessions(seePhotographs2and3).Thiscontinueduntiloneguard,TQ,decidedtoresolvethesituationbyaccedingtoPP’srequestsforacigaretteinordertoencouragetheotherprisonerstoreturntotheircell.Aftertheincident,theprisonersbackintheircellandtheguardsbackintheirmessexpressedagrowingrealizationthatthesystemwasopentochange.However,theydidsowithaverydifferentevaluativetone.Theprisonersliterallydancedwithjoy.PPexclaimed‘thatwasfuckingsweet’,towhichKMresponded(admiringly)‘youwasfuckingqualityman’.Thenallthreecellmatesexchanged‘highves’.Theguardsbickereddespondently.startedbysaying‘ThisisonlyDay4.Theycanseewhathappenedtodayandnowtheyknowtheycandowhatevertheywant’.AsBGandthenTQweaklydisagreed,TAonlyinsistedmore,culminatingintheclaimthat‘It’shappened.Imean,comeon.Youknowwhatthathasdone?Thathaslitthefuseon[PB’s]arse’.Hence,theinductionofinsecurerelationsbetweenthegroupswasnotdependentuponourinterventionsbutwasanemergentpropertyoftheintergroupdynamics.Correspondingly,plannedinterventionswerenotnecessaryfortheprisonerstostartchallengingtheguards.Thechallengestartedmountingimmediatelyafterthepromotion.Consequently,thelegitimacyinterventionwasnotimplemented.However,becausethenewprisonerhadalreadybeentoldhewouldbebroughtin,thiswentaheadonDay5(andhewaslaterwithdrawnonDay6).Giventhesituation,hedidnotsomuchsuggestcognitivealternativeswherenoneexistedassuggestadditionalalternativestothestatusquo.Mostnotably,hebegantoquestiontheprisoner–guard Inthisinstance(andwithdepressiondatabelow),analysisofpolynomialcontrastswasbasedonarandomsubsetofvephasesbecausethenumberoflevelsofthewithin-subjectsvariable(studyphase)exceededthenumberofsubjects.Thepsychologyoftyranny divisionitselfand,byquestioningthelegitimacyofcertainaspectsofthestudy(particularlytheheatintheprison),hesuggestedbothtohisfellowcellmates(DD)andtooneoftheguards(TQ)thattheparticipantsmountaunitedchallengetotheexperimenters.Asthefollowingextractshows,thisprovidednewwaysofenvisaginghowthesystemmightwork:Extract2:Ifthiswasareal-lifesituation::::Yes.:::andyouwereworkinginthiskindofheat,thenyouasanemployeecouldwellgototheemployerandsay‘Theconditionisunacceptable,I’mnotpreparedtoworkinit’.Nowlet’streatthisasareal-lifesituation.YouandI–yourgroupandthegroupI’min–bothhavethisproblemoftheheat.AndifI’vegottosleepinthis,thereisnowayIwill.AndI,youknow,won’tbearit.AndIthinkcollectivelyweshoulddosomethingaboutittothepeoplewhoarerunningtheexperiment.Nowyouknowinanormal,day-to-day,real-lifesituation,that’swhatwouldhappen.:Well,Iammostimpressedwithyournew-foundkindofangleonthis,whichpossiblyshouldn’tcomeasasurprisetome.ButIthinkthatisaveryveryvalidpointyouaremakingandI’mgoingtogoalongwithitcompletely.Asarstpracticalsteptowardsparticipantunity,DMcreatedanewnegotiatingstructure,whichbroughtguardsandprisonerstogetheronabasisthatwasfarmoreequalthanhadhithertobeenthecase.Theguardswereeagerandpleasedtoacceptthisarrangement,foreveniftheysurrenderedmuchoftheirhierarchicaladvantagebydoingso,thenewsystemconrmedtheirpositioninwhatpromisedtobeaviablesocialorder.Quantitativeconrmationofthesepatternsemergesonmeasuresofparticipants’awarenessofcognitivealternativestothestatusquo(Tajfel&Turner,1979).ThesewereadministeredonDays1,3,4,and6usinga4-itemscale(‘Icannotimaginetherelationshipbetweenguardsandprisonersbeinganydifferent(reverse-scored)’/‘Ithinkthattheguardswillalwayshavemoreprivilegesthantheprisoners(reverse-scored)’/‘Ithinkthattherelationshipbetweenguardsandprisonersislikelytochange’/‘Ithinkthatitwouldbepossiblefortheprisonerstohavemorepowerthantheguards)’.Asinglescorewascomputedbyaveragingresponsestothese4items. Photographs2and3.Day4:TheprisonersstarttochallengetheguardsÕregime.StephenReicherandS.AlexanderHaslam MeanscoresarepresentedinFigure3asafunctionofparticipantgroupandtime.Ascanbeseenfromthisgraph,atthestartofthestudy,prisonersandguardshadanequalandrelativelylowsenseofcognitivealternatives,butthisincreasedasthestudyprogressed.Thesepatternswereconrmedbystatisticalanalysisofcognitivealternativesdatathatwasconductedbymeansofa2(participantgroup:guards,prisoners)4(studyphase:Days1,3,4,6)ANOVAwithrepeatedmeasuresonthesecondfactor.Thisanalysisrevealedasignicantandlargemaineffectfortime,(3,33).34.Consistentwiththeaboveobservations,analysisofpolynomialtrendstodecomposethiseffectindicatedthatparticipants’senseofcognitivealternativesincreasedlinearlyovertime,(12)2.83,Acceptanceoftheunequalregime:ComplianceandorganizationalcitizenshipAssociatedwiththeaboveeffects,astheprisonersdevelopedasenseofsharedsocialidentitythatwasdenedinoppositiontotheguards,andastheybecameawareofthepossibilityofalternativestoexistingstatusrelations,theystartedtoworkactivelyagainsttheregimethattheguardsweretryingtouphold.Thiswasmanifestbothinminorchallengestotheguards’status(e.g.insubordinationduringrollcall)andinmoreovertchallenges–includingtheincidentalludedtoabove,inwhichanumberofprisonerscollectivelyprotestedaboutthequalityoftheirfood.Dataconsistentwiththisanalysisemergefrommeasuresof(a)participants’willingnesstocomplywithauthorityand(b)theirwillingnesstoengageinactsoforganizationalcitizenship(i.e.todomorethanwasaskedoftheminordertomaketheprisonsystemwork;Organ,1988,1997).CompliancewasmeasuredonDays1,3and5bymeansofa2-itemscale(‘Itrytodowhattheguardswant’/‘Itrytocomplywiththeguards’rules’).Responsesonthesemeasureswereaveragedtoprovideasinglescore.TheresultantmeanscoresarepresentedinFigure4asafunctionofparticipantgroupandtime.Ascanbeseenfromthisgraph,intheearlystagesofthestudy,bothguardsandprisonerswerewillingtocomplywiththerules,buttheprisonersbecamemorereluctanttocomplyafterthepromotion.Thesepatternswereconrmedbystatisticalanalysisofcompliancescoresthatwasconductedbymeansofa2(participantgroup:guards,prisoners)3(studyphase:Days1,3,5)ANOVAwithrepeatedmeasuresonthesecondfactor.Thisrevealedasignicant 23456 Day 1Day 3Day 4Day 6 Prisoners Awareness ofcognitive alternativesFigure3.Awarenessofcognitivealternativesasafunctionofassignedgroupandtime.Thepsychologyoftyranny andlargeinteractionbetweengroupandtime;(2,22)Analysesofpolynomialtrendsindicatedthatcomplianceonthepartoftheguardsdidnotvarysignicantlyovertime,,butthatoftheprisonersdeclinedinalinearfashion,4.48,.01.Therewasalsoevidenceofaquadratictrend,.07,suggestingthattheprisoners’declineincompliancewasparticularlymarkedaftergroupboundarieshadbecomeimpermeable.Organizationalcitizenship(Organ,1988,1997)wasmeasuredonDays2,4,and5bymeansofa3-itemscalecontainingitemsusedwidelyinpreviousresearch(e.g.Haslam,Powell,&Turner,2000;Tyler&Blader,2000;‘Iamwillingtodomorethanisaskedofmebytheguards’/‘IwilldowhateverIcantohelptheguards’/‘WheneverpossibleIwilltrytomaketheguards’workdifcult’;reversescored).Asinglescorewascomputedbyaveragingresponsestothese3items.TheresultantmeanscoresarepresentedinFigure5asafunctionofparticipantgroupandtime.Ascanbeseenfromthisgraph,theguardswerealwaysmorewillingthantheprisonerstoengageincitizenshipbehavioursthatwouldhelpthemruntheregime.However,whiletheguardsmaintainedtheirwillingnesstoworkfortheregimeacrossthethreetestingphases,overtimetheprisonersbecamemuchmorereluctanttosupporttheguards’regimeinthisway.Again,thiswasparticularlytrueafterthepromotiononDay3. 2 Day 2Day 4Day 5 Figure5.Organizationalcitizenshipasafunctionofassignedgroupandtime. 234 Day 1Day 3Day 5 Figure4.Compliancewithprisonrulesasafunctionofassignedgroupandtime.StephenReicherandS.AlexanderHaslam Thesepatternswereconrmedbystatisticalanalysisoforganizationalcitizenshipscoresperformedbymeansofa2(participantgroup:guards,prisoners)3(studyphase:Days2,4,5)ANOVAwithrepeatedmeasuresonthesecondfactor.Thisanalysisrevealedasignicantandlargemaineffectforgroup,(1,11)butthiswasqualiedbyasignicantandlargeinteractionbetweengroupandtime;(2,22).32.Analysesofpolynomialtrendsrevealedthattheguards’willingnesstodisplayorganizationalcitizenshipdidnotvaryacrossthethreephasesoftesting,,butthatoftheprisonersdeclinedlinearlyovertime,Collectiveself-efcacyandmentalhealthThedisorganizationoftheguardsandtheunityoftheprisonersdidnotjustproduceconditionswherethelatterchallengedtheformer,italsoledtotheprisonersbecomingprogressivelymoredominant.Throughplanningandmutualsupport,theybecameincreasinglymoreextremeandmoresuccessfulintheireffortstounderminetheguards’control.Bycontrast,theguards’attemptstoimposetheirauthoritybecameincreasinglyunsuccessful,andtheybecamemoredividedandmutuallyrecriminatory.Moreover,aswehavealreadyillustratedwithreferencetoobservationaldata,theeffectivenessoftheprisonersinpursuingtheircollectivegoalsledtostrongpositiveaffectwhiletheinabilityoftheguardstoactcollectivelyledtodespondency.Thesedynamicsarereectedinquantitativedataobtainedonmeasuresof(a)collectiveself-efcacyand(b)depression.Collectiveself-efcacy(Bandura,1995)wasmeasuredonDays2,4,and6bymeansofa5-itemscalecontainingitemsusedwidelyinpreviousresearch(e.g.Chen&Bliese,2002;‘Myprisongroupiscondentthatwecoulddealefcientlywithunexpectedevents’/‘Myprisongroupcanremaincalmwhenfacingdifcultiesbecausewecanrelyonourcopingabilities’/‘Myprisongroupcanalwaysmanagetosolvedifcultproblemsifwetryhardenough’/‘Whenmyprisongroupisconfrontedwithaproblem,wecanusuallyndseveralsolutions’/‘Myprisongroupcanusuallyhandlewhatevercomesourway’).Asinglescorewascomputedbyaveragingresponsestothese5items.Figure6presentstheresultantmeancollectiveself-efcacyscoresasafunctionofparticipantgroupandtime.Ascanbeseenfromthisgraph,thesedatamirrorthoseobtainedonmeasuresofsocialidentication,sothat,atthestartofthestudy,theguards 234 Day 2Day 4Day 6 Collective self-efficacyFigure6.Collectiveself-efÞcacyasafunctionofassignedgroupandtime.Thepsychologyoftyranny hadagreatersenseofcollectiveself-efcacythantheprisoners.However,byDay4,theprisoners’self-efcacyhadincreasedmarkedlyandwasnowgreaterthatthatoftheguards.ThispatternwasmaintainedatDay6,althoughitisinterestingtonotethatafterthetradeunionistwasremovedfromtheprison,bothgroupsshowedsomedeclineinself-efcacy(seeminglybecausetheopportunitiesforgroup-basednegotiationandorderthathepresentedwereremoved;seeEggins,Haslam,&Reynolds,2002;Haslam,Eggins,&Reynolds,2003).Thesepatternsareconsistentwithourobservationalaccount–althoughthegeneraldeclineintheguards’senseofself-efcacyislesspronouncedinthesequantitativedata.Collectiveself-efcacyscoreswerestatisticallyanalysedbymeansofa2(participantgroup:guards,prisoners)3(studyphase:Days2,4,6)ANOVAwithrepeatedmeasuresonthesecondfactor.Thisanalysisrevealedasignicantmaineffectfortime,(2,22).36,thatwasconditionedbyasignicantandlargeinteractionbetweengroupandtime,(2,22).28.Inlinewiththeaboveobservations,analysesofpolynomialtrendsrevealedthattheprisoners’senseofcollectiveself-efcacyincreasedlinearlyovertime,.05–aneffectthatwasalsoqualiedbysignicantquadraticvariation,4.25,.01.Ontheotherhand,thecollectiveself-efcacyofguardsdeclinedovertime,althoughthiseffectwasnotsignicant,0.90,Depressionwasmeasuredduringthescreeningprocessandtheneverydayduringthestudy(partlyinordertomonitortheongoingwelfareofparticipants).Forthispurpose,a7-itemscalewasadministered(‘Ingeneral,howhasyourmoodbeenoverthelastfewdays?’/‘Doyoueverfeellowordepressed?’/‘Doyoufeelhopelessaboutthefuture?’/‘Doyouhavedifcultydealingwitheverydayproblems?’/‘Areyouself-condent?’/‘Doyouthinkthatyouareaworthwhileperson?’/‘Doyouthinkaboutharmingyourself?’).Duringthescreeningphase(332),theseitemswerefoundtoformareliablescale(.80)andsotheitemswereaveragedtoformasinglemeasure.MeandepressionscoresarepresentedinFigure7.Fromthisgraph,itcanbeseenthatalthoughoveralllevelsofdepressionwerelow,theyclearlyvariedasafunctionofparticipantgroupandtime.Specically,whiletheprisonersweremoredepressedthantheguardsatthestartofthestudy,byitsend,thissituationhadreversed.Thispatternwasconrmedbya2(participantgroup:guards,prisoners)7(studyphase:pre-test,Days1to6)ANOVAwithrepeatedmeasuresonthesecondfactor.Thisrevealedasignicantandlargeinteractionbetweengroupandtime;(6,66)3.73, 2 Day 1Day 2Day 3Day 4Day 5Day 6 Figure7.Depressionasafunctionofassignedgroupandtime.StephenReicherandS.AlexanderHaslam .25.Consistentwithobservations,polynomialcontrastsshowedthattheprisoners’depressiondecreasedlinearlyovertime,4.22,.01.Atthesametimetheguards’depressionincreased,thoughnotsignicantly,CombinedimpactThecombinedoutcomeoftheseinterrelateddynamicswasthatastheprisonersbecameincreasinglyawareof,andcondentin,theircollectiveidentity,andastheguards’condenceandcollectiveself-efcacydeclined,theprisonersdevelopedandexecutedplanstotakeonanddestroytheguards’regime.Theguardsweredivided,exhaustedanddemoralized.Theywereunabletoorganizethemselvesortheprisonerseffectively.Asaresult,lateontheeveningofDay6,theprisonersinCell2brokeoutoftheircellandoccupiedtheguards’-quarters.Atthispoint,theguards’regimewasseenbyalltobeunworkableandatanend(seePhotograph4).Phase2:EmbracinginequalityAfterthecollapseoftheguard–prisonerhierarchy,allbuttwooftheparticipantsdecidedthattheywantedthestudytocontinuethroughtheinstitutionofasingleself-governingcommune.WithFCinthechair,theymetwiththeexperimentersanddrewupthetermsunderwhichthecommunewouldoperate.Initially,thenewsystemwashighlyeffective.Anumberofparticipantswhohadbeenmutuallyhostilewhentheyweredividedintoprisonersandguardsformedstrongandpositiveaffectivetiesnowtheywererecategorizedpartofacommongroup(Gaertner,Mann,Murrell,&Dovidio,1989;Turneretal.,1987).Collectively,theyperformedtheirworktasksandchoreswithmoreeffortandtoahigherstandardthanatanyotherpointinthestudy.However,thoseparticipantswhohadbeencentrallyinvolvedinchallengingtheoldregimefeltmarginalizedinthisnewsystem.Atrst,theysimplyfailedtocontributetocollectivetasks.Later,theybegantoviolatecommunalrules.Withinaday,theywereplottingtodestroythecommune.Theproblemsthiscreatedwereexacerbatedbecausethecommune’smembershadneverdevelopedproceduresfordealingwithdissidence,andhencetheyhadnomeansofrespondingtothreatstotheirsocialorder.Bythemorningofthecommune’ssecondday,itwascleartomanyparticipantsthatthenewsocialstructurewasincrisis.Thissituationwasexacerbatedwhen,bychance, Photograph4.Day6:Cell2breakoutbringingtheguardsÕregimetoanend.Thepsychologyoftyranny thebreakfastwasofverypoorquality.Thiswas(incorrectly)takenasasignthattheexperimentersdisapprovedofthecommunesystem.Itledtodespondencyamongthecommune’ssupporterswhofelt,asonelaterputit,thattheyhadtheburdenofresponsibilityforthesystemwithouttheabilitytomakeitwork.Moreover,theemergentcrisiswasexploitedbyopponentsofthecommune,fourofwhom(oneex-guardandthreeex-prisoners)formulatedaplantocreateanewandharsherguard–prisonerhierarchy.Thenatureandtoneofthisnewregimewasmadeabundantlyclearindiscussionsabouttheformthiswouldtake.AsPBputit,‘Wewanttobetheguardsandfuckingmakethemtoetheline,Imeanonthefuckingline.Nofuckingtalkingwhileyouareeating.Getonwithyourfoodandgetthefuckinghellbacktoyourcell’.Shortlyafterbreakfast,thisgroupconvenedameeting(seePhotograph5).Theirleaderberatedthecommuneanditssupportersandheintroducedtheideaofthenewhierarchy.Thesupportersofthecommunewerelargelypassiveinresponse.Theylookeddespondentandlistenedinsilenceuntilhehadnearlynished.Duringdebriengs,anumberofthemacknowledgedthat,althoughtheywouldnothaveopenlyendorsedsuchahierarchy,theywerelessopposedtoitthantheyhadbeenpreviouslyandthattheyfeltlessrepulsedbytheideaofastrongsocialorderinwhichsomeoneelseassumedresponsibilityformakingthesystemwork.Again,thesetrendsaresupportedbyanalysisofpsychometricdata.Mostnotably,thisisapparentwhenonelooksatthelevelofparticipants’right-wingauthoritarianism(Altmeyer,1981,1996).Thiswasmeasuredusing8itemsabstractedfroma30-itemscaleadministeredduringpre-testing(Reynolds,Turner,Haslam,&Ryan,2001;‘Thingswouldgobetterifpeopletalkedlessandworkedharder’/‘Itisbettertoliveinasocietyinwhichthelawsarevigorouslyenforcedthantogivepeopletoomuchfreedom’/‘Peopleshouldalwayscomplywiththedecisionsofthemajority’/‘Youhavetogiveupanideawhenimportantpeoplethinkotherwise’/‘Therearetwokindsofpeople:strongandweak’/‘Whatweneedarestrongleadersthatthepeoplecantrust’/‘Oursocialproblemswouldbesolvedif,inonewayoranother,wecouldgetridofweakanddishonestpeople’/‘Peopleshouldalwayskeeptotherules’).Inpre-testing332),thisscalewasfoundtobereliable(.71)andtocorrelateveryhighlywiththefull30-itemscale(.97).Accordingly,scoresontheaboveitemswereaveragedtoformasinglemeasure.ThiswasadministeredduringthescreeningprocessandthenonDays1,3and7. Photograph5.Day8:Newguardsmakethecaseforanauthoritarianregime.StephenReicherandS.AlexanderHaslam Intherstinstance,itisrelevanttoexaminelevelsofauthoritarianismovertimeasafunctionofthegroupstowhichparticipantswereassignedbytheexperimenters.DatapertainingtothisanalysisarepresentedinFigure8.Thesewereanalysedbymeansofa2(participantgroup:guards,prisoners)4(studyphase:pre-test,Days1,3,7)ANOVAwithrepeatedmeasuresonthesecondfactor.Consistentwithqualitativeobservations,theonlyeffecttoemergefromthiswasasignicantandlargeeffectfortime,(3,33).20.Analysisofpolynomialtrendsrevealedthattherewasalinearincreaseinallparticipants’authoritarianismasthestudyprogressed,However,asafascinatingvariantontheaboveanalysis,itisalsopossibletolookatauthoritarianismasafunctionofthegroupstowhichtheremainingparticipantsassignedthemselvesattheendofthestudy–thatis,asparticipantswhoeithersupportedthecommuneorwhoproposedsettingupanewhierarchy(withthemselvesasthenewguardswithinit).DatapertainingtothisanalysisarepresentedinFigure9.Thesewereagainanalysedbymeansofa2(participantgroup:newguards,newprisoners)4(studyphase:pre-test,Days1,3,7)ANOVAwithrepeatedmeasuresonthesecondfactor.Theonlyeffecttoemergefromthisanalysiswasasignicantandlargeinteractionbetweengroupandtime;(3,36).20.Analysisof 234 PretestDay1 Day 3Day 7 Figure8.Right-wingauthoritarianismasafunctionofassignedgroupandtime. 2 Pre-testDay 1Day 3Day 7 Figure9.Right-wingauthoritarianismasafunctionofself-selectedgroupandtime.Thepsychologyoftyranny polynomialcontrastsindicatedthatthisinteractionarosefromthefactthattheauthoritarianismofthosewhosoughttoassumetheguards’roleattheendofthestudyhaddeclinednon-signicantlyasthestudyprogressed,.However,theauthoritarianismoftheremainingparticipantshadincreasedsteadilyovertime,3.67,.01(sothatthetwogroupsweredemonstratingaverysimilarlevelofauthoritarianismatthetimethatthenewguardswereactuallyseekingtotakepower).Despitethegeneralshifttowardsauthoritarianism,andalthoughitisprobablethatthenewguardswouldhavehadtheforcetoimposetheirregimeinthefaceofweakeningresistance,inthisstudysuchforcewasprohibitedunderthepre-establishedethicalguidelines.Hence,inourjudgment,thestudywasgridlockedandhadreachedwhatwesawasanaturalpointoftermination;theexistingsystemwasnotworkingandthenewsystemcouldnotbeimposed.Accordingly,thestudywasbroughttoaconclusionatnoononDay8.However,theparticipantsremainedforafurtherdayinordertoundertakeaseriesofstructureddebriengsdesignedtoobtainandprovidefeedbackontheirexperience,toexplaintherationaleforthestudyandtoovercomeanyhostilitybetweenindividualsderivingfromeventsinthestudy.TheBBCprisonstudywasdesignedtoexaminethefactorsthatdeterminehowpeoplerespondwhenasystemofinequalityisimposeduponthembyothers.Atthestart,almostalltheparticipantsrejectedthissystem.However,bytheend,theywereclosetoinstitutinganewandmoretyrannicalsocialsystem.Inadditiontoouroriginalquestionsconcerningthewayinwhichpeoplerespondtoasystemofinequalitythathasbeenimposeduponthem–dotheyacceptitordotheyresistit?–thisraisesanewandunexpectedissue.WhataretheconditionsunderwhichpeoplecreateasystemofinequalityforthemselvesWewillarguethatourndingsconcerningreactionstoinequalitycannotbeexplainedthroughageneralor‘natural’tendencytoassumerolesandassertpower.Rather,theyrequireanunderstandingoftheconditionsunderwhichanexternallyimposedcategorizationbecomesasubjectiveself-perception(Turner,1982).Theexistingtermsofsocialidentityandself-categorizationtheoriesprovidethebasisforsuchanunderstanding,althoughitisnecessarytoelaborateuponnotionsofcontextinordertoaccountforthefailureoftheguardstocohereasagroup.Ourndingsconcerningthecreationofinequalityleadustomoreoriginaltheoreticalconclusions.Ontheonehand,wesuggestthatgroupsarethebasisforcollectiveself-realization–thatis,thecreationofasocialorderbasedonsharedvaluesandnorms(Drury&Reicher,2005).However,wheregroupsfail,wearguethatpeoplewillbemoreinclinedtoaccepttheimpositionofasocialorderbyothers,evenwherethatviolatestheirvaluesandnorms.Therefore,incontrasttothosewhoexplaintyrannyandotherextremesocialphenomenaintermsofthepsychologicaldysfunctionalityofgroups,weinterpretthemintermsofthedysfunctionalityofgroupfailure.Beforeweareinapositiontodevelopthesearguments,however,itisnecessarytodealwithfourpotentialcritiquesofthestudyanditsabilitytosayanythingmeaningfulaboutthegroupprocessesrelatingtoinequalityandtyranny.Therstisthatthebehaviourofparticipantswasdeterminedbythefactthattheyknewtheywerebeingobservedbytelevisioncamerasandthatthisrendersthestudysoarticialastohavelittleornogeneralvalue.ThesecondisthattheeffectsweobservedwereaproductofStephenReicherandS.AlexanderHaslam participants’personalitiesandthatthereforethestudyhaslittletosayaboutgroupprocesses.Thethirdisthatwefailedtocreaterealpowerdifferencesormeaningfulinequalitiesbetweengroupsandhenceourstudyhaslittletosayaboutthepsychologyofeitherpowerorinequality.Thefourthisthatthevariablesuponwhichourpredictions(andhenceourplannedinterventions)focusedwerenotresponsiblefortheeffectsobtainedand,therefore,evenifgroupandpowerprocesseswereatplay,wecannotbesurewhattheyare.FourcritiquesTheroleoftelevisionFromtheirmomentofrstcontact,volunteersknewthattheBBCwasinvolvedinthisproject,andbythetimethatthestudystarted,allparticipantswereawarethattheywouldbeconstantlyobservedbytelevisioncamerasandthatanythingtheydidmightbeshownonnationaltelevision.Althoughitwasstressedthroughoutthatthiswasaspecialistscienticprojectthatwouldnotmake‘stars’outofthoseinvolved,andalthoughthescreeningprocesswasusedtoexcludeanyonewhowasmotivatedbythedesireforpublicity,thisisahighlyunusualsituationandwasboundtoimpactonbehaviour–butwhatimpact,andwithwhatimplications?Themostdamningargumentwouldbethatparticipantsweresimplyfakingtheirbehavioursforthecameras.Webelievesuchanexplanationtobeimplausibleinlightoftheimmenseeffortthatwouldberequiredtocontinuouslymonitorandfakeone’sownbehaviourfornearly9days.Moreover,itwouldbemuchharder,ifnotimpossible,tofakethepsychometricandphysiologicaldata.However,evenifparticipantswerecapableofsuchplay-acting,onewouldstillneedtoexplainthecomplexpatternofresultsnotonlybetweengroups,butalsowithingroupsacrosstime.Onemight,forexample,suggestthatthereisacertaindiscreditinbeingatyrantandacertainglamourinbeingarebel,whichmayexplainwhyourguardsweresomildandourprisonerssorebellious.But,whythendidtheprisonersbecomemorerebelliousafterthesystemwasmadeimpermeableandinsecure?Evenmoreproblematically,whydidtheparticipantsmovetowardstyrannyatthestudy’send?Toarguethatourndingscanbeexplainedbysuggestingthatparticipantsweremerelyplay-actingorseeking‘celebritystatus’isnotonlyimplausiblebutalsounhelpful.Inparticular,itfailstoexplain(a)whypeopleactedastheydidwhentheydidand(b)whytheirbehaviourandattitudesclearlychangedinparticular(predicted)waysoverthecourseofthestudy.Bythesamelogic,surveillancecannotbethewholestory,asitremainedconstantatthesametimethatbehaviouritselfwaschanging.Havingsaidthis,wereadilyaccept(seebelow)thatthetelevisingofthestudyispartofthestory–particularlytowardsthestartofthestudy(participantsreportedthattheywereacutelyawareofthecamerasonenteringthe‘prison’,butthat,astimewentby,theyincreasinglyforgotaboutthemexceptduringquietmoments;e.g.whenseeingacameramovelateatnightintheircells).However,wesuggestthatthenatureofthisimpactaddsto,ratherthandetractsfrom,therichnessandwiderrelevanceofourndings.Thisisbecausealthoughitmaybeunusualtobeinapositionwhereanythingonedoesmightbebroadcastintomillionsofhomes,thisisanextremeexampleofsomethingthatisanincreasinglycommonfeatureofoureverydaylives–namely,surveillance.Formostofoursociallives,weareunderobservationandourbehaviourcanbeexaminedbyaudienceswhoarenotpresent(Reicher&Emler,1985).Sometimes,thesurveillanceisamatterofsurveillancecameras,scrutinyofcomputerThepsychologyoftyranny transactions,workplacereview,orunobtrusiveproling(Lyon,1994;Lyon&Zuriek,1996).Atothertimes,itissimplyamatterofthosewhoarewithusinoneplacetalkingaboutourbehaviourtootherpeopleinotherplaces(e.g.ingossip,auniversalhumanbehaviour;Emler,1994).However,whicheveristhecase,wecanrarelyignoresurveillancewhenwearealoneandalmostneverwhenwithothers.Whatthetelevisioncamerasdo,therefore,istohighlightindramaticformanaspectofhumanexperience,whichisalltoooftenoverlookedinpsychology.Indeed,itisarguablethatthetypicalpsychologyexperimentinwhichpeopleareisolatedandguaranteedabsoluteanonymityencountersgreaterproblemsofarticiality(Cronin&Reicher,2006).Ofcourse,thesizeandvarietyofaudiencestowhichoneismadeaccountablebytelevisionismuchlargerthanwecommonlyexperience.However,aswewilldiscussinmoredetailbelow,thefactofaccountability,andthefactthatourparticipantshadtoconsiderothercontextsandotheraudiencesevenwhenactingwithinthecontextofthestudy,enhancesratherthandiminishesitsrelevanceandmeaningfulness.Inthisrespect,weareremindedofThayerandSaarni’s(1975)retorttothosewhoimpugnthevalidityoftheSPEonthegroundsthatparticipantswereguidedbytheexpectationsofothers.Thisissummarizedinthetitleoftheirpaper:‘demandcharacteristicsareeverywhere(anyway)’.Likewise,tothosewhocriticizetheBBCstudyonthegroundsthatitwasbroadcast,wewouldrespondthatsurveillanceiseverywhere(anyway).TheroleofpersonalityForthosewhohavewatchedTheExperiment(Koppel&Mirsky,2002),itishardnottobestruckbytheforceofmanyofthecharacterswhocontributedtothevariousoutcomesdescribedabove.Inpart,thishastodowiththevisualnatureofthemedium,wheretheconcreteandvisibleimpactofindividualsisinevitablymoresalientthanthatofmoreintangibleand‘invisible’elementssuchasgroupprocesses(Asch,1952).Nonetheless,itcouldclearlybearguedthatalltheobservedeffectswereaproductofstrong(andweak)personalitiesandthatthemorepowerfulcharacters(inparticular,,PP,andPB)endedupasprisoners.Onecouldarguethattheirpersonalitiesdrovetheseindividualstooverwhelmtheguardsandthen,astheygrewboredofthecommune,todestroyitinturn.Therstresponsetosuchanargumentistorememberthatwematchedtheprisonersandguardsontheobviousindividualdifferencefactorsrelatingtotyranny:modernracism,authoritarianism,andsocialdominance.Nonetheless,onecouldstillarguethattherewereothercharacteristicsonwhichwedidnotmatchthegroups.Asaresult,the‘stronger’individualsmightstill,bychance,havebeenallocatedtotheprisonergroup.However,evenifthiswerethecase,aswithourdiscussionofsurveillance,isdoesnotproveparticularlyhelpfulinexplaininghoweventsinthestudyunfolded.Mostfundamentally,thisduetotheinterrelatedobservations(a)thatpeople’s‘character’onrelevantdimensionsappearedtochangeoverthecourseofthestudyand(b)thatrelevantindividualdifferencesweremuchlessapparentattheoutsetthantheywereasthestudyprogressed.Wereadilyaccept,however,thatindividualdifferencesareofthestory(seeReicher&Haslam,inpress;Turner,Reynolds,Haslam,&Venstra,inpress).However,aswiththeanalysisofleadershipmoregenerally(e.g.seeHaslam,2001;Haslam&Reicher,2005;Reicher&Hopkins,2001b;Turner&Haslam,2001),theproblemisthatitishardtoexplainchangingpatternsofbehaviourinourstudywithreferencetoabluntandconstantfactorsuchas‘personality’.Toclaimthatcertainpeoplewere‘rebellious’doesStephenReicherandS.AlexanderHaslam notexplainwhyrebellionwasmuteduntilafterthepromotionandfailstoexplainwhythoseprisonerswhorebelledagainstoneformofauthority(theguards)nonethelessremaineddeferentialtoanother(theexperimenters).InthecaseofJE,forexample,weseethatbeforethepromotion,heinvestedhisenergiesinsupportingthesystemanditwasonlyafterpromotionwasruledoutthatheputthemintounderminingthesystem.Therefore,itisonethingtocallhim‘forceful’,buttoexplainhowthatforcewasdirected,weneedtoinvokesystemicfactors.Similarly,inthecaseofJE’scellmatePPwhilehewasopposedtotheguards’authorityfromthestartofthestudy,hisabilityactuallytounderminetheirregimewascontingentoncollectivewillandsupport,whichonlymaterializedafterthepromotion(inwayspredictedbysocialidentitytheory;Tajfel&Turner,1979;seebelow).Moregenerally,thereweremanypointsinthestudywheretheprominenceandimpactofindividualsdependedupontheirrelationshiptothegroup(Reicher&Haslam,inpress).Thus,whentheywereseentoreectsharedidentityandsharedvalues,eventhemildestofmencametothefore.ThisisapparentinthecaseofFC,anenvironmentalistwhohadbeenquietthroughouttherst6daysofthestudy,butwhobecamealeadingandvocalgureincreatingandmanagingthecommuneandwithoutwhosecommitment,knowledge,andexperienceitmayneverhavecomeintobeing.Furthermore,itisalsoapparentthatwhentherewasnosharedidentityorsharedvalues,eventhemostheroicofindividualswasboundtofail.ThiswasseenearlyinthestudywhenFC,madeastrenuousefforttoimposetheguards’authorityandmaintainstrictdiscipline.However,asitbecameclearthathelackedthesupportofhisfellowguards,hegaveup,becameincreasinglypassive,andwithdrewintothebackground.Thus,themanifestdifferencesintheforcefulnessofprisonersandguardsderivednotfromtheirinherentpersonalitiesbutwereinsteadanemergentproductofthesuccessorfailureoftheirrespectivegroups.Again,thisisnottodismisstheimportanceofindividualvariables.Aswehaverepeatedlystressed,theskills,knowledge,values,andpre-existingcommitmentsofindividualsplayedacrucialpartinfacilitating,shaping(andindeedinblocking)groupformation(Postmes,Baray,Haslam,&Morton,inpress;Reicher,2004;Reicher,Drury,Hopkins,&Stott,2001;Reicher&Hopkins,1996,2001b,2004).Farfrombeingmindless,ourparticipantswerecreativeandthoughtfulasgroupmembers,continuouslystrivingtoshapegroupandintergrouprealities.However,ifindividualsplayedakeypartinshapinggroups,theconverseisequallytrue–theabilityofindividualstoaffectoutcomeswasdependentuponboththeexistenceofgroupsandtheirprototypicalitywithinthegroup(Turner,1991).Toputitslightlydifferently,individualagencywasnotdestroyedbythegroup,butratherachievedthroughit(Reicher&Haslam,inpress).Thus,whilepersonalityandotherindividualdifferencefactorsareanimportantaspectofourstudy(andwewillconsiderthemaspartofouranalysisbelow),theycannotsubstituteforthatanalysis(Asch,1952;Brown,1965;Sherif,1966;Tajfel,1978).Itfollowsthathaddifferentindividualsparticipatedinthestudy,wemaywellhaveobservedquitedifferentoutcomesfromthosereportedabove.Thisisbecausemanyindividual(and,indeed,chance)factors(suchasthepoor-qualitybreakfastonDay8)impactedonthefateofgroupsandtheirsubsequenttrajectoryinthestudy.However,asweexplainedintheIntroduction,thisiswhywearenotseekingtomakeempiricalgeneralizationsonthebasisofndingsalonebutratherareattemptingtomakegeneralizationsonthebasisofthetheoreticalanalysisthatthosendingssupport(Turner,1981).Ourclaimstogeneralitythusrelateto(a)theeffectofparticularfactorsThepsychologyoftyranny (e.g.permeability,cognitivealternatives,roles)ongroupformationand(b)theconsequencesofgroupsuccessandfailureforbothindividualsandthesystemasawhole.TherealityofinequalityandpowerAthirdpotentialcriticismofourstudyisthattheresultsand,inparticular,thefailureoftheguardstoexerttheirauthority(ortheprisonerstoacceptit),reectsthetwinfactsthat(a)thestudycontextwassimplytoopallidforourparticipantstobecomeengagedinitand(b)theguardshadnoauthoritytowieldeveniftheywantedto.If,asZimbardoandhiscolleaguesargue,tyrannyisafunctionofgroupsandpower,thenonemightassertthatourstudylackedameaningfuldegreeofeitherandthatwearethereforeinnopositiontosayanythingworthwhileabouttyranny.Letusconsidertheseissuesofengagementandpowerinturn.Inthecaseoftheformer,itisimportanttodistinguishbetweenengagingwiththesituationandengagingwithone’sgroup.Theevidencesuggeststhattheprisonersexperiencedandresentedtheirsubordinatestatusfromthestartofthestudy.Theydislikedtheirfood.Theydislikedbeinglockedup.Theydislikedtheboredom.Thesmokersinparticulardislikedbeingdeprivedofcigarettes.Indeed,afteradayinhiscell,PPadmittedthathewas‘fallingforit[theexperimentalsituation]hook,lineandsinker’.Attherstmeal,severalprisonersexpressedoutrageatthepoorqualityofthefood,especiallygiventhesuperiorfareoftheguards.Thelackofgroupidenticationamongstprisonersintheearlydayscannotthereforebeputdowntoapathyabouttheirplight.Equally,theguardsengagedwiththesituationfromthestart.Theirinitialconversationsfocusedonthepowerofthesituation,theimpactofwearingaguard’suniform,andthedangersofbecomingtyrannical.Attheirrstmeal,theyalsonotedwithdiscomfortthedisparityinfoodqualityandquantity.Theythentriedtoeasethisdiscomfortbyofferingtheirleftoverstosomeoftheprisoners.Therefore,iftheguardsfailedtoactasagroup,itwasnotbecausetheydidnotcareaboutthedisparitiesinthestudybut,onthecontrary,becausetheyweretroubledbythem.Theirdisidenticationdidnotreectthefactthatthestudywasnotcompelling,butratherthatitwasalltoocompelling(albeitinverydifferentwaysfromtheSPE).Theissuesaresimilarwhenitcomestothematterofpower.Itiscertainlytruethattheguardsfailedtoexercisepower.However,thiswasnotbecausetheyhadnopowertoexercisebutpreciselybecausetheyhadsomuch.Indiscussion,theguardsrecognizedthevariousoptionsthatwereopentothem.Theseincludedindividualpunishments,collectivepunishments,removalofprivileges,extratasksandrollcallsand,mostparticularly,thepowertopromoteaprisonerwhowouldhelpthemrunthesysteminthewaytheywanted.However,theirfearoftheguardidentity–ofbeingauthoritarianandofbeingseenasauthoritarian–madethemshuntheseoptions,topromoteaprisonerwhoembodiedtheirambivalence(Herriot,2002),andeventogiveawaysomeofthesourcesoftheirpower.Inotherwords,theparticipants’psychologicalstateexplainstheirfailuretoexercisepowermuchmorethantheirlackofpowerexplainstheirpsychologicalstate(Turner,2005).Wewilladdressthereasonsforthisstateshortly.Fornow,wesimplynotethatourparticipantsdidengagewithasituationinwhichtherewererealinequalitiesofresourcesandpower.Hence,thestudycanbeviewedasameaningfulexplorationofthepsychologyofunequalgroups.TheimpactofinterventionsandkeyvariablesItmaybethatourstudysayssomethingofinterest,itmaybethatitsayssomethinginterestingaboutgroupprocesses,itmayevenhavesomethingimportanttosayaboutStephenReicherandS.AlexanderHaslam tyrannyandresistance,butwhatofourclaimtohaveconductedan‘experimentalcasestudy’inwhichweinvestigatedtheimpactofspecictheoreticallyinformedvariables?Canourinterventionsreallybedescribedasoperationalizationsofpermeabilityandcognitivealternatives?Aretherenotahostofotherwaysinwhichtheseinterventionsandtheirconsequencescanbeinterpreted,anddoesthisfactnotcastdoubtonanyparticulartheoreticalglosswemightprovide?Thisisareasonableconcern.Inthepresentstudy,thereiscertainlyadisjunctionbetweenthesimplicityofourvariablesandthecomplexityofourinterventions.Inparticular,thepromotionandtheintroductionofDMcouldhavehadmanyeffects,suchasmakingparticipantsuncertainaboutwhatmighthappennextandmakingthemfeelhelplessanddistrustfuloftheexperimenters.However,suchproblemsarenotuniquetothepresentstudy–notleastbecausethetheoreticalstatusofindependentvariablesisapotentialconcerninexperimentalresearch(Haslam&McGarty,2004).Indeed,wheneverresearchersclaimthataconcreteinstance(aspecicoperationaliza-tionofanindependentvariable)reectsanabstractgenerality(thevariableasitoperatesintheworldatlarge),interpretationisinvolvedandalternativesarepossible(Billig,1987).Whentheoperationalizationisascomplexasours,however,thespaceforargumentisallthegreater.Inexperimentalpractice,thisissueisnormallyaddressedintwoways(Haslam&McGarty,2003).Therstisthroughtheuseofmanipulationchecks.Aswehaveseenabove(Figure3),checksofthisformprovidesomeevidenceconsistentwithourtheoreticalinterpretation(specicallyinrelationtotheexistenceofcognitivealternatives).Moreover,oneoftheadvantagesofhavingmultipledatasources(i.e.behaviouralaswellaspsychometric)isthatweareabletoseewhetherparticipantsspontaneouslyreactedtoourinterventionsinwaysthatwereexpected.Wehaveprovidedqualitativeevidencetosuggestthattheydid.Specically,afterthepromotionprisonerssawnopointinplacatingtheguardsandrealizedthattheonlywaytoimprovetheirpositionwastochallengethesystem(Extract1).Equally,wesawthatDMtransformedthewayinwhichparticipantsconceptualizedtheirsituationandtheiroptions(Extract2).Aswithanymanipulationcheck,suchdatasupportthenotionthatwedidmanipulatethevariablesofconcerntous.However,whilemanipulationchecksexcludefalsepositives,theycannotexcludeallpossiblefalsenegatives–thatis,additionalwaysinwhichthemanipulationmayhavehadanimpact.Thisiswherethesecondwayofreducinguncertaintyaboutthestatusofindependentvariablescomesintoplay.Here,itisnotenoughmerelytoarguefortheimportanceofaparticularindependentvariable(oraparticularconfound;e.g.personality,surveillance,thepassingoftime).Itisalsonecessarytoexplainhowthatvariable(orconfound)mightcomprehensivelyaccountforthepatternsofobservedresults.Thatis,itisnecessarilytohaveanintegratedtheoreticalaccountofhowtheeffectsoccurred.Asweexplainedwhenintroducingthem,inthisrespect,ourclaimsaccordwithawell-developedandextremelyinuentialtheoreticaltradition(afterTajfel&Turner,1979)andwithawell-establishedandrobustpatternofpreviousexperimentalndings(e.g.Ellemers,1993;Wrightetal.,1990).Thistheoreticalandempiricaltraditionlendscoherenceandplausibilitytoouranalysis.Havingsaidthat,wecertainlydonotviewouranalysisasinviolate,andwefullyacceptthatouraccountoftheimpactofparticularinterventionsisopentoargument.However,forcriticswhowishtochallengethisaccount,itisincumbentuponthemtodoaswehaveandexplainhowanalternativeunderstandingofourinterventionsprovidesabetterunderstandingofthestudy’sndingsintheirentirety.Thepsychologyoftyranny Withthisinmind,itistimetomovebeyondpotentialcriticismsandtobeexplicitaboutwhatourexplanationofeventsis.Thisinvolvesthreeelements.Therstrelatestotheconditionsofsocialidentication,thesecondaddressestheconsequencesofsocialidentication,andthethirdconcernsreactionstogroupfailure.AsocialidentityaccountoftyrannyTheconditionsofsocialidenticationThesimplestandclearestndingofourstudyisthatpeopledonotautomaticallyassumerolesthataregiventotheminthemannersuggestedbytheroleaccountthatistypicallyusedtoexplaineventsintheSPE(Haneyetal.,1973;Zimbardo,1989).Instead,therearearangeoffactorsthatdeterminewhetherpeoplethemselvesidentifywiththesocialpositionstowhichtheyareascribedbyothers.Someofthosefactorsoperatedinwaysspeciedbysocialidentitytheory(Tajfel&Turner,1979).Thus,theshiftfrompermeabilitytoimpermeabilityofgroupboundarieshadastrongimpactonprisoneridentication(alonglinespreviouslydemonstratedinseparateprogrammesofresearchbyEllemersandWrightandtheircolleagues;Ellemers,vanKnippenberg,DeVries,&Wilke,1988;Ellemersetal.,1990,1993;Wright,1997;Wright&Taylor,1998,Wrightetal.,1990;seealsoLalonde&Silverman,1994).Equally,theincreasinginsecurityofintergrouprelationsaffectedthewillingnessoftheprisonerstoworkasagroupinordertochallengetheguards.Inaddition,changingperceptionsofsecurity,andofthecomponentoflegitimacyinparticular,helpexplainnotonlywhytheprisonerschallengedtheguards,butalsowhytheparticipantsasawholedidnotchallengetheexperimenters,despitetheadverseconditionsthatlivingintheprisonimposeduponthem(cf.Spears&Smith,2001).Mostofthetime,whenparticipantsdiscussedthematter,theysawtheseconditionsasalegitimatepartofthestudyinwhichtheyhadagreedtoparticipate.Ononlytwooccasionsweretheconditionsseentoviolatethatcontractuallegitimation–whenthenewprisoner,DM,begantoquestiontheheatintheprison,andwhenonenight,theexperimentersfailedtoreplenishbasicsuppliesfortheguards.Onbothoccasions,thisledtoexplicitconsiderationof‘mutiny’.However,ifthestudysupportedsocialidentitytheoryindemonstratingtheroleofcontextualfactorsinmoderatingtherelationshipbetweenroleandidentity,italsoshowstheimportanceofextendingthewayinwhichweconceptualizethenatureofcontextanditsrelationtohumanaction.Thus,althoughtheroleofguardwaspositivelyvaluedintheimmediatecontextoftheprison,thoseassignedtothispositionwereconcernedwiththepossibilityofnegativeevaluationsbyfutureaudiences,andhencesomeofthemwerereluctanttoidentifywiththerole.Itisherethattheprospectivetelevisingofthestudymostclearlyhadanimpactonthebehaviourofparticipantsandservestoraiseimportantpracticalandtheoreticalpoints.Practically,itdemonstrateshowextremebehaviourscanberestrainedbyrenderingactorsvisible,andhenceaccountable,tobroaderoryet-to-beencounteredaudiences(Postmes,Spears,Lea,&Reicher,2000;Reicher&Levine,1994;Tedeschi,1981;Tetlock,1985).Theoretically,itshowsthatcontext,forhumanbeings,cannot SomecommentatorshavearguedthatthisanalysisisinconsistentwithevidencearisingfromthetortureofinmatesbyUStroopsatAbuGhraibprisonin2004.However,aswehavearguedelsewhere(Reicher&Haslam,2004),itappearsthattheseactswereperformedforthebenetofanin-groupaudiencethat(liketheexperimentersintheSPE)wasassumedbytheperpetratorstoapproveoftheaction.StephenReicherandS.AlexanderHaslam simplybeassumedtobewhereweareinthehereandnow.Peoplehaveimaginationsthatallowthemtothinkofothertimesandplaces,andtoevaluatetheirbehaviourinrelationtothese.Indeed,surveillancehasanimpactonbehaviourpreciselythroughitscapacitytomakesalientsuchalternativecontextsandalternativeaudiences(Reicher&Haslam,2003;Reicher&Levine,1994).Twopointsfollowfromthis.First,underconditionsofsurveillance,individualdifferencesrelatingtothedifferentsocialcommitmentspeoplehaveoutsideagivencontexthavethecapacitytoimpactuponbehaviourwithinthecontext.Thus,inourstudy,thefactthatTQwaspubliclyvisibleasasuccessfulentrepreneurwhoownedaliberalenterprisegoessomewaytowardsexplainingwhyhewasparticularlyambivalentaboutadoptingtheidentityofguard.However,ifindividualfactorswereimportanthere,thentheyoperatedinaparadoxicalway.First,TQ’sveryeffectivenessoutsidethestudyiswhatrenderedhimineffectivewithinit.Second,oncehumanimaginativecapacitiesareacknowledged,formsofextremesituationism(suchastheroleaccount),whichsuggestthatbehaviourisalwaysdominatedbythepresentcontext,becomeuntenable.Instead,itisimportanttotakepastandfuturecontextsintoaccountaswell(Reicher&Hopkins,2001a).TheconsequencesofsocialidenticationOneofthedistinctivefeaturesofthepresentstudywasthatitsintensivenatureallowedestablisheddisciplinaryboundariestobetranscendedsothatitwaspossibletoinvestigatetheinterrelationshipsbetweensocial,organizationalandclinicalvariables.Theresultspresentedabovepointtotheexistenceofsuchlinksandtotheirrichnessandcomplexity.Ontheorganizationalside,theachievementofacommonsocialidentitywasseentobenecessaryinorderforgroupmemberstotrustotherstoactappropriately,tosupportothersintheiractionsandtoexpectsupportfromtheminreturn(Haslametal.,2005).Inthisway,therudimentsofeffectiveorganization(e.g.taskdifferentiation,delegation,leadership,trust)wereseentobecontingentuponsharedsocialidenticationinamannerarguedbyarangeoforganizationaltheorists(e.g.Ellemers,DeGilder,&Haslam,2004;Haslam,2001;Haslam,vanKnippenberg,Platow,&Ellemers,2003;Hogg&Terry,2001;Postmes,2003;Smith,Tyler,&Huo,2003;Turner&Haslam,2001;Tyler&Blader,2000).Forthisreason,theeffectivenessoftheprisoners,whowereabletodevelopasenseofsharedsocialidentity,contrastedmarkedlywiththeineffectivenessoftheguards,whowerenot(Drury&Reicher,2005).Ontheclinicalside,theincreasingsupportamongstprisonersandtheirsuccessfulchallengestotheguardscontributedtomentalstatesthatwereincreasinglypositive.Ontheotherhand,theatomizationoftheguardsandtheirfailuretofulltheircollectivetasksledtoincreasinglynegativestates.Thisisreectedinthedepressiondata(andalsoonothermeasuresnotreportedhereincludingparanoia,anxietyandburnout;seeHaslam&Reicher,inpress).Althoughthenatureofourquantitativedatadoesnotallowforanalysisofcausalrelationsbetweenthesevariables(Haslam&McGarty,2004),wecansaywithcondencethatparticipants’mentalstatesevolvedinrelationtothesocialdynamicsbetweengroups,andthatvariablessuchassocialsupportandcollectiveself-efcacyareimplicatedinthatrelationship(e.g.asarguedbyBranscombe,Schmitt,&Harvey,1999;Cohen&Wills,1985;Contrada&Ashmore,2000;Hall&Cheston,2002;etal.,2005;Orford,1992;Schwarzer,2001;Terry,Carey,&Callan,2001).Clearlythough,theexactnatureoftherelationshipmeritsfurtherinvestigation.Thepsychologyoftyranny ReactionstogroupfailureThereweretwoinstancesofgroupfailureinthestudy.Therstconcernedthefailureoftheguardstofulltheirassignedtaskofmakingahierarchicalprisonsystemwork,andthesecondconcernedthefailureofthecommune’ssupporterstoestablishaviableegalitariansocialorder.Itisnotablethatineachcase,groupmembersbecamemorewillingtoacceptasystemthatpromisedtobeviableevenifitmeantcedingsomeofthecoreprinciplesofthegroup.OnDay5,aftertheintroductionofthenewprisoner,theguardswerewillingtocedetheirauthorityandacceptamoreequalsocialsystemandonDay8communesupporterswerepreparedtocedeequalityandtolerateamorehierarchicalsocialsystem.Thus,ratherthanpeople‘naturally’preferringanygivenformofsocialorder,itappearsthat,whengroupmembersfailtoimposeanorderbasedontheirownexistingnormsandvalues,theyarewillingtoadaptthosevalues(ortoadoptnewones)inordertocreateaviableorderratherthanhavenoorderatall.Inthisregard,thefactthattheparticipants’authoritarianismincreasedsignicantlyovertimeconstitutesoneofthemostimportantndingsofthestudyasawhole.Traditionally,authoritarianismhasbeenviewedastableindividualdifferencevariablethathasthecapacitytoexplaintheemergenceofhierarchicalandtyrannicalsocialstructures(Adornoetal.,1950;Altmeyer,1981,1996;forreviewsseeBillig,1978;Brown,1965;Duckitt,1994).However,incontrasttothisanalysis,thepresentstudyillustratesthatauthoritarianismisavariableoutcomeofsocialstructure(seealsoOakes,Haslam,&Turner,1994;Reynoldsetal.,2001).Morespecically,weseethatauthoritariansolutions–andthepersonalitiesthatwouldpromotethem–appearedmoreattractiveafterattemptstomakedemocracyworkwereseentohavefailed.Againstthisargument,itmightbesuggestedthatthesystemweobservedmovedtowardstyrannysimplybecausethemoreauthoritarianindividualswerewaitingforanopportunitytoimposetheirpreferredsystemonothers(indeed,asimilarpointcouldbemadeinrelationtotheSPE).However,alonglinesintimatedabove,twopointsspeakagainstthisconclusionandagainstrelatedargumentscouchedintermsofindividualpersonality-baseddynamics.Therstisthattyrannyaroseoutofthesocial(dis)orderoftheprison,andauthoritarianism(likeotherdimensionsofrawpersonality)wasclearlynotastraightforwarddeterminantofthat(dis)order.Thesecondisthatalthoughitmaybetruethatthosewhoproposedthenewregimewerethosewhohadinitiallybeenmoredisposedtowardsauthoritarianism,theirinuenceandleadershipwasclearlycontingentupontheirviewshavingbecomemorerepresentativeoftheviewsoftheparticipantsasawholeasthesehadevolvedoverthecourseofthestudy(formorein-depthtreatmentofthisissueseeHaslam&Reicher,2005;Reicher&Haslam,inpress;seealsoHaslam,2001;Haslam&Platow,2001;Navas,Morales,&Moya,1992;Reicher&Hopkins,1996,2001b;Turner,1991;Turner&Haslam,2001;Turneretal.,inpress).Indeed,theinteractionbetweennalgroupmembershipandstudyphasesuggeststhat,ifanything,theauthoritarianismoftheguardsinthevauntednewregimehaddeclinedovertime,whereasthatofotherparticipantswasincreasing(seeFigure9).Italsoisworthaddingthatpreviousresearchhasbeenunabletoobservedynamicsofthisformatworkortogenerateempiricalsupportforargumentssuchasthese,preciselybecauseithasbeenunabletoexaminetheimpactofevolvinggrouphistoryonbothindividualpsychologyandsocialstructure(Haslam&McGarty,2001;Levine,2003).StephenReicherandS.AlexanderHaslam Conclusion:Rethinkingtherelationshipbetweengroups,powerandtyrannyAtonelevel,ourstudyconrmsthendingsoftheSPE.Itshowsthatanunderstandingofcollectiveconictandtyrannycannotbeachievedsimplybylookingatindividualsbutrequiresananalysisofgroupprocessesandintergrouprelations.Inthissense,weagreewithZimbardo(andmanyothers;e.g.Asch,1952;Sherif,1966;Tajfel,1978)thatsuchphenomenacanonlybeexplainedthroughgroup-levelanalysis.OurdisagreementwithprioranalysisoftheSPEthusrelatestothenatureofgroupprocessesandoftheconditionsunderwhichtheyleadtosocialpathologies.Asalmosteverypsychologystudent(andanunusuallylargeproportionofthegeneralpublic)knows,themessageoftheSPEisthatthetoxiccombinationofgroupsandpowerleadstotyranny.TheimplicationsoftheBBCprisonstudyaredifferent.Incommonwithrecenttheoreticaldevelopmentsinsocialpsychology,theycontestthepremisethatgroupbehaviourisnecessarilyuncontrolled,mindlessandantisocial(Ellemersetal.,1999;Oakesetal.,1994;Postmesetal.,2000;Reicher,1982,2001;Spears,Oakes,Ellemers,&Haslam,1997;Turner,1999).Incontrast,theresultsoftheBBCprisonstudysuggestthatthewayinwhichmembersofstronggroupsbehavedependsuponthenormsandvaluesassociatedwiththeirspecicsocialidentityandmaybeeitheranti-prosocial(Jetten,Spears,&Manstead,1997).However,basedonthepresentdata,wewouldarguethatfailinggroupsalmostinevitablycreateahostofproblemsfortheirownmembersandforothers.Theseproblemshaveadeleteriousimpactonorganization,onindividuals’clinicalstate,and–mostrelevanthere–onsociety.Foritiswhenpeoplecannotcreateasocialsystemforthemselvesthattheywillmorereadilyacceptextremesolutionsproposedbyothers.Itiswhengroupslackthepowertoexercisechoicethatanauthoritarianideologythatpromisestocreateorderforthemappearsmoreseductive.Inshort,itisthebreakdownofgroupsandpowerlessnessthatcreatestheconditionsunderwhichtyrannycantriumph(forrelatedargumentsseeKanter,1979;Pfeffer,1981;Reynolds&Platow,Wewouldarguethataswellasbeingconsistentwithcontemporarythinkinginsocialpsychology(e.g.afterTajfel&Turner,1979),thisanalysisalsoarticulatesmorecloselythanZimbardo’soriginalroleaccountwiththeanalysisoftyrannyputforwardbyresearchersinotheracademicdisciplines.Mostnotably,itaccordswithinuentialanalysesproposedbymodernhistorians(e.g.Abel,1986;Gellately,2001;Hobsbawm,1995;Rees,2002).Consider,forinstance,Hobsbawm’saccountoftheconditionsthatgaverisetothefalloftheWeimarrepublicandtheemergenceofNazismin1930sGermany:Theoptimalconditionsforthetriumphoftheultra-rightwereanoldstateanditsrulingmechanismswhichcouldnolongerfunction;amassofdisenchanted,disorientedanddisorganizedcitizenswhonolongerknewwheretheirloyaltieslay;strongsocialistmovementsthreateningorappearingtothreatensocialrevolution,butnotactuallyinapositiontoachieveit....Theseweretheconditionsthatturnedmovementsoftheradicalrightintopowerful,organizedandsometimesuniformedandparamilitaryforce(1995,p.127).WewouldalsoarguethatthisanalysiscanbeusedtomakesenseofwhathappenedintheSPEwhenonelooksmorecloselyattheeventsthatunfoldedthere.Aswehavenoted,thatstudy,likeours,appearstohavestartedoffwithprisonersthreateningtobecomeascendantovertheguards.ThingschangedwhenZimbardointervenedinsuchawayastoleadtheprisonerstobelievethattheycouldnotleavethestudy.Atthispointtheybecamedisorientedastotheirposition–inZimbardo’sownwords,theyThepsychologyoftyranny experienced‘roleconfusion’(Zimbardo,1989).Theyceasedtosupporteachotheragainsttheguards.Theycollapsedasagroupandthisallowedtyrannicalguardstoprevail.Thus,whereasintheSPE,thefailureoftheprisonersallowedanexistingtyrannytobeconsolidated,intheBBCstudy,thefailureofthecommunepavedthewayfortheemergenceofanewtyranny.These,ofcourse,arelargeandcontroversialclaims.However,theyhaveimportanttheoreticalimplicationsalongwithconsiderablepracticalimplications.Theypointtonewwaysofthinkingbothaboutthenatureofgrouppsychologyandaboutthepsychologicalunderpinningsoftyranny.Becausethescopeoftheseanalysesissolarge,itwouldbebothpresumptuousandfoolishtoexpectthemtobeacceptedonthebasisofasingledataset–especiallyonereectingthecomplexitiesofthepresentstudy.YetwhatistrueofourstudyislargelytrueoftheSPE.Aswehaveargued,oneofthegreatproblemsarisingfromthatstudywasthatitmadestrongclaims(thathadanenormousimpactonpublicconsciousness)butthenledtodebatebeingclosedoffbecausefurtherresearchwasdeclaredtobeethicallyunacceptable.Hence,irrespectiveofourresultsandofouranalysis,wewouldclaimthatoneofthesignicantachievementsoftheBBCprisonstudyistoshowthat,ifsufcientcareistaken,itispossibletorunpowerfulandinuentialeldstudiesintosocialprocessesthatarealsoethical.Inaeldincreasinglydominatedbyreductionistaccountsofhumanbehaviour,suchstudiescanrestorebalancebydemonstratingtheimpactofsystematicvariationsinsocialrelationsuponhumanbehaviour.Therichness,immediacyandrelevanceoftheirndingscanalsohelpreconnectpsychologistswithpolicymakers.Forthesereasons,weofferourconclusionsnotsomuchasanalwordonthematter,butmoreinthehopeofstimulatingresearcherstorenewtheirinterestintheimportantdebatesthatourdisciplinecan,andmust,advance.AcknowledgementsWewouldliketothanktheparticipantsaswellasthestaffoftheBBCfortheircommitmenttothisresearchandforcommentsonideasinthispaper.TheassistanceandinputofNylaBranscombe,LloydCarson,DavidCorner,AndrewEagle,LibertyEaton,ScottGalloway,NickHopkins,MikeHowe,JolandaJetten,AndrewLivingstone,EvaLoth,GrantMuir,LucyO’Sullivan,MichaelSchmitt,HeatherSmith,TomPostmes,FabioSani,DenisSindic,StephanieSonnenberg,PaulWebley,StephenWilks,HuwWilliamsandBrianYoungisalsogratefullyacknowledged,togetherwithdetailedandveryhelpfulcommentsfromseveralanonymousreviewers.ReferencesAbel,T.(1986).WhyHitlercametopower.Cambridge,MA:HarvardUniversityPress.Adorno,T.W.,Frenkel-Brunswik,E.,Levinson,D.J.,&Sanford,R.N.(1950).Theauthoritarian.NewYork:Harper.Altmeyer,B.(1981).Rightwingauthoritarianism.Winnipeg:UniversityofManatobaPress.Altmeyer,B.(1996).Theauthoritarianspecter.Cambridge,MA:HarvardUniversityPress.Arendt,H.(1998).Thehumancondition.Chicago:UniversityofChicagoPress.Asch,S.E.(1952).Socialpsychology.EngelwoodCliffs,NJ:PrenticeHall.Bandura,A.(1995).Exerciseofpersonalandcollectiveefcacyinchangingsocieties.InA.BanduraSelf-efÞcacyinchangingsocieties(pp.1–45).NewYork:CambridgeUniversityPress.Banuazizi,A.,&Movahedi,S.(1975).Interpersonaldynamicsinasimulatedprison:Amethodologicalanalysis.AmericanPsychologist,152–160.StephenReicherandS.AlexanderHaslam Baron,R.S.(1984).Groupdynamics.InA.S.Kahn(Ed.),Socialpsychology(pp.260–289).Dubuque,IA:W.C.Brown.BarTal,D.(2004).Thenecessityofobservingreallifesituations:Palestinian–Israeliviolenceasalaboratoryforlearningaboutsocialbehaviour.EuropeanJournalofSocialPsychologyBillig,M.(1978).Fascists:AsocialpsychologicalviewoftheNationalFront.London:AcademicBillig,M.(1987).Arguingandthinking.Cambridge:CambridgeUniversityPress.Branscombe,N.R.,Schmitt,M.T.,&Harvey,R.D.(1999).PerceivingpervasivediscriminationamongAfricanAmericans:Implicationsforgroupidenticationandwell-being.JournalofPersonalityandSocialPsychology,135–149.Brockes,E.(2001,October16).TheExperiment.TheGuardian,G2,pp.1–3.Brown,R.(1965).Socialpsychology.NewYork:FreePress.Chen,G.,&Bliese,P.D.(2002).Theroleofdifferentlevelsofleadershipinpredictingself-andcollectiveefcacy:Evidencefordiscontinuity.JournalofAppliedPsychology,549–556.Cohen,J.(1977).Statisticalpoweranalysisforthebehavioralsciences.NewYork:Wiley.Cohen,S.,&Wills,T.A.(1985).Stress,socialsupportandthebufferinghypothesis.Psychological,310–357.Contrada,R.J.&Ashmore,R.D.(Eds.),(2000).Self,socialidentity,andphysicalhealth:Interdisciplinaryexplorations.NewYork:OxfordUniversity.Cronin,P.&Reicher,S.(2006).Astudyofthefactorsthatinuencehowseniorofcerspolicecrowdevents:OnSIDEoutsidethelaboratory.BritishJournalofSocialPsychology,175–196.Doosje,B.,Ellemers,N.,&Spears,R.(1995).Perceivedintragroupvariabilityasafunctionofgroupstatusandidentication.JournalofExperimentalSocialPsychology,410–436.Doosje,B.,Spears,R.,&Ellemers,N.(2002).Thedynamicanddeterminingnatureofgroupidentication:Responsestoanticipatedchangesinthestatushierarchy.BritishJournalofSocialPsychology,57–76.Drury,J.,&Reicher,S.(2005).Explainingenduringempowerment:Acomparativestudyofcollectiveactionandpsychologicaloutcomes.EuropeanJournalofSocialPsychologyDuckitt,J.(1994).Thesocialpsychologyofprejudice(2nded).NewYork:Preager.Eggins,R.A.,Haslam,S.A.,&Reynolds,K.J.(2002).Socialidentityandnegotiation:Subgrouprepresentationandsuperordinateconsensus.PersonalityandSocialPsychologyBulletinEllemers,N.(1993).Theinuenceofsocio-structuralvariablesonidentityenhancementEuropeanReviewofSocialPsychology,27–57.Ellemers,N.,DeGilder,D.,&Haslam,S.A.(2004).Motivatingindividualsandgroupsatwork:Asocialidentityperspectiveonleadershipandgroupperformance.AcademyofManagement,459–478.Ellemers,N.,Spears,R.,&Doosje,B.(1999).Socialidentity:Context,contentandcommitmentOxford:Blackwell.Ellemers,N.,vanKnippenberg,A.,DeVries,N.K.,&Wilke,H.(1988).Socialidenticationandpermeabilityofgroupboundaries.EuropeanJournalofSocialPsychology,497–513.Ellemers,N.,vanKnippenberg,A.,&Wilke,H.(1990).Theinuenceofpermeabilityofgroupboundariesandstabilityofgroupstatusonstrategiesofindividualmobilityandsocialchange.BritishJournalofSocialPsychology,233–246.Ellemers,N.,Wilke,H.,&vanKnippenberg,A.(1993).Effectsofthelegitimacyoflowgrouporindividualstatusonindividualandcollectiveidentityenhancementstrategies.JournalofPersonalityandSocialPsychology,766–778.Emler,N.(1994).Gossip,reputation,andsocialadaptation.InB.F.Goodman&A.Ben-Ze’evGoodgossip(pp.117–138).Lawrence:UniversityofKansasPress.Thepsychologyoftyranny Gaertner,S.L.,Mann,J.,Murrell,A.,&Dovidio,J.F.(1989).Reducingintergroupbias:Thebenetsofrecategorization.JournalofPersonalityandSocialPsychology,239–249.Gellately,R.(2001).BackingHitler.Oxford,UK:OxfordUniversityPress.Hall,S.,&Cheston,R.(2002).Mentalhealthandidentity:Theevaluationofadrop-incentre.JournalofCommunityandAppliedSocialPsychology,30–43.Haney,C.,Banks,C.,&Zimbardo,P.(1973).Astudyofprisonersandguardsinasimulatedprison.NavalResearchReview9,1–17[ReprintedinE.Aronson(Ed.),Readingsaboutthesocial(3rded.,pp.52–67).SanFrancisco,CA:W.H.Freeman].Haslam,S.A.(2001).Psychologyinorganizations:Thesocialidentityapproach.London:Sage.Haslam,S.A.,Eggins,R.A.,&Reynolds,K.J.(2003).TheASPIRemodel:ActualizingSocialandPersonalIdentityResourcestoimprovestrategicplanningandencouragesustainableproductivityinorganizations.JournalofOccupationalandOrganizationalPsychologyHaslam,S.A.,O’Brien,A.,Jetten,J.,Vormedal,K.&Penna,S.(2005).Takingthestrain:Socialidentity,socialsupportandtheexperienceofstress.BritishJournalofSocialPsychologyHaslam,S.A.,&McGarty,C.(2001).Ahundredyearsofcertitude?Socialpsychology,theexperimentalmethodandthemanagementofscienticuncertainty.BritishJournalofSocialPsychology,1–21.Haslam,S.A.,&McGarty,C.(2003).Researchmethodsandstatisticsinpsychology.London:Sage.Haslam,S.A.,&McGarty,C.(2004).Experimentaldesignandcausalityinsocialpsychologicalresearch.InC.Sanson,C.C.Morf,&A.T.Panter(Eds.),Handbookofmethodsinsocialpsychology(pp.235–264).ThousandOaks,CA:Sage.Haslam,S.A.,&Platow,M.J.(2001).Thelinkbetweenleadershipandfollowership:Howafrmingasocialidentitytranslatesvisionintoaction.PersonalityandSocialPsychologyBulletinHaslam,S.A.,Powell,C.,&Turner,J.C.(2000).Socialidentity,self-categorizationandworkmotivation:Rethinkingthecontributionofthegrouptopositiveandsustainableorganizationaloutcomes.AppliedPsychology:AnInternationalReview,319–339.Haslam,S.A.,&Reicher,S.(2002).Auser’sguidetoTheExperiment–Exploringthepsychologyofgroupsandpower:ManualtoaccompanytheBBCvideo.London:BBCWorldwide.Haslam,S.A.,&Reicher,S.D.(2005).Identityentrepreneurshipandtheconsequencesofidentityfailure:ThedynamicsofleadershipintheBBCPrisonStudy.ManuscriptsubmittedforHaslam,S.A.,&Reicher,S.D.(inpress).Stressingthegroup:Socialidentityandtheunfoldingdynamicsofstress.JournalofAppliedPsychologyHaslam,S.A.,vanKnippenberg,D.,Platow,M.&Ellemers,N.(Eds.),(2003).Socialidentityatwork:Developingtheoryfororganizationalpractice.Philadelphia,PA:PsychologyPress.Herrera,C.D.(1997).AhistoricalinterpretationofdeceptiveexperimentsinAmericanpsychology.HistoryoftheHumanSciences,23–26.Herriot,P.(2002).Selectionandself:Selectionasasocialprocess.EuropeanJournalofWorkandOrganizationalPsychology,385–402.Hobsbawm,E.(1995).Ageofextremes:Theshorttwentiethcentury1914Ð1991.London:Hogg,M.A.,&Terry,D.J.(Eds.),(2001).SocialidentityprocessesinorganizationalcontextsPhiladelphia,PA:PsychologyPress.House,J.S.(1981).Workstressandsocialsupport.Reading,MA:Addison-Wesley.Hoyle,R.H.,Georgesen,J.C.,&Webster,J.M.(2001).Analysingdatafromindividualsingroups:Thepast,thepresent,thefuture.GroupDynamics:Theory,PracticeandResearch,41–47.Ito,T.A.,Thompson,E.,&Cacioppo,J.T.(2004).Trackingthetimecourseofsocialperception:Theeffectsofracialcuesonevent-relatedbrainpotentials.PersonalityandSocialPsychology,1267–1280.StephenReicherandS.AlexanderHaslam Jetten,J.,Spears,R.,&Manstead,A.R.(1997).Strengthofidenticationandintergroupdifferentiation:Theinuenceofgroupnorms.EuropeanJournalofSocialPsychologyKanter,R.(1979).Powerfailureinmanagementcircuits.HarvardBusinessReview,July–August,Koppel,G.(Seriesproducer),&Mirsky,N.(Executiveproducer).(2002,May14,15,20,21).TheExperiment.London:BritishBroadcastingCorporation.Lalonde,R.N.,&Silverman,R.A.(1994).Behavioralpreferencesinresponsetosocialinjustice:Theeffectsofgrouppermeabilityandsocialidentitysalience.JournalofPersonalityandSocialPsychology,78–85.Laudat,M.H.,Cerdas,S.,Dournier,C.,Guiban,D.,Guilhaume,B.,&Luton,J.P.(1988).Salivarycortisolmeasurement:Apracticalapproachtoassesspituitary-adrenalfunction.JournalofClinicalEndocrinologyandMetabolism,343–348.LeBon,G.(1895,translated1947).Thecrowd:Astudyofthepopularmind.London:ErnestBenn.Levine,R.M.(2003).Times,theoriesandpracticesinsocialpsychology.TheoryandPsychology,53–72.Leyens,J.P.,Cortes,B.,Demoulin,S.,Dovidio,J.F.,Fiske,S.T.,Gaunt,R.,Paladino,M.P.,Rodriguez-Perez,A.,Rodriguez-Torres,R.,&Vaes,J.(2003).Emotionalprejudice,essentialism,andnationalism.EuropeanJournalofSocialPsychology,703–717.Lindsay,R.C.L.,&Adair,R.G.(1990).Doethicallyrecommendedresearchproceduresinuencetheperceivedethicalityofsocialpsychologicalresearch?CanadianJournalofBehavioral,282–294.Lovibond,S.H.,Mithiran,X.,&Adams,W.G.(1979).Theeffectsofthreeexperimentalprisonenvironmentsonthebehaviourofnon-convictvolunteersubjects.AustralianPsychologist,273–287.Lyon,D.(1994).Theelectroniceye.Oxford:Polity.Lyon,D.,&Zuriek,E.(1996).Computers,surveillanceandprivacy.Minneapolis:UniversityofMinnesotaPress.McDermott,M.,Opik,L.,Smith,S.,Taylor,S.,&Wills,A.(2002).TheExperiment:Reportoftheindependentethicspanel.http://www.experimentethics.org.uk.McGarty,C.,&Smithson,M.(2005).Independenceandnon-independence:Asimpletestforcomparinggroupsusingmultiplemeasuresandthebinomialtest.EuropeanJournalofSocialPsychology,35,171–180.Milgram,S.(1963).Behavioralstudyofobedience.JournalofAbnormalandSocialPsychology,371–378.Milgram,S.(1974).Obediencetoauthority:Anexperimentalview.NewYork:HarperandRow.Moreland,R.L.,Argote,L.,&Krishnan,R.(1996).Sociallysharedcognitionatwork:Transactivememoryandgroupperformance.InJ.Nye&A.Brower(Eds.),WhatÕssocialaboutsocialcognition?Researchonsociallysharedcognitioninsmallgroups(pp.57–84).NewburyPark,CA:Sage.Morgan,A.H.(1979).CommentsonS.H.Lovibond,X.MithiranandW.G.Adams“Theeffectsofthreeexperimentalprisonenvironmentsonthebehaviourofnon-convictvolunteersubjects”.AustralianPsychologist,285–286.Navas,M.,Morales,J.M.,&Moya,M.C.(1992).Categorizationandprototypicalityasdeterminantsofleadereffectiveness.InternationalJournalofPsychology,289–290.Norusis,M.J.(1985).SPSSx:Advancedstatisticsguide.NewYork:McGraw-Hill.Oakes,P.J.,Haslam,S.A.,&Turner,J.C.(1994).Stereotypingandsocialreality.Oxford:Orford,J.(1992).Communitypsychology.Chichester:Wiley.Organ,D.W.(1988).Organizationalcitizenshipbehavior:ThegoodsoldiersyndromeLexington,MA:Lexington.Organ,D.W.(1997).Organizationalcitizenshipbehavior:Itsconstructclean-uptime.Performance,85–97.Thepsychologyoftyranny Pfeffer,J.(1981).Powerinorganizations.Boston,MA:Pitman.Postmes,T.(2003).Asocialidentityapproachtocommunicationinorganizations.InS.A.Haslam,D.vanKnippenberg,M.J.Platow,&N.Ellemers(Eds.),Socialidentityatwork:Developingtheoryfororganizationalpractice.Philadelphia,PA:Taylor&Francis.Postmes,T.Baray,G.Haslam,S.A.,&Morton,T.(inpress).Personalandsocialidentityformationthroughinteraction:Individualityinsmallgroups,socialmovementsandcategories.Individualityandthegroup:Advancesinsocialidentity.London:Sage.Postmes,T.,&Spears,R.(1998).Deindividuationandanti-normativebehaviour:Ameta-analysis.PsychologicalBulletin,238–259.Postmes,T.,Spears,R.,Lea,M.,&Reicher,S.D.(2000).SIDEissuescentrestage.Amsterdam:RoyalNetherlandsAcademyofArtsandSciences.Rees,L.(2002).TheNazis:Awarningfromhistory.London:BBCWorldwide.Reicher,S.D.(1982).Thedeterminationofcollectiveaction.InH.Tajfel(Ed.),Socialidentityandintergrouprelations(pp.41–84).Cambridge:CambridgeUniversityPress.Reicher,S.D.(1996).Socialidentityandsocialchange:Rethinkingthecontextofsocialpsychology.InP.Robinson(Ed.),Socialgroupsandidentities:DevelopingthelegacyofHenriTajfel.Oxford:Butterworth-Heinemann.Reicher,S.D.(2001).Thepsychologyofcrowddynamics.InM.Hogg&S.Tindale(Eds.),Blackwellhandbookofsocialpsychology:Groupprocesses(pp.175–197).Oxford:Reicher,S.D.(2004).Thecontextofsocialidentity:Domination,resistanceandchange.PoliticalPsychology,921–946.Reicher,S.D.,Drury,J.,Hopkins,N.,&Stott,C.(2001).Amodelofcrowdprototypesandcrowdleadership.InC.Barker,M.Lavalette,&A.Johnson(Eds.),LeadershipandsocialmovementsManchester:ManchesterUniversityPress.Reicher,S.D.,&Emler,N.(1985).Delinquentbehaviourandattitudestoformalauthority.JournalofSocialPsychology,161–168.Reicher,S.D.,&Haslam,S.A.(2003).Socialpsychology,science,andsurveillance:UnderstandingTheExperimentSocialPsychologyReview,7–17.Reicher,S.D.,&Haslam,S.A.(2004).Whynoteveryoneisatorturer.RetrievedMay10,from:http://news.bbc.co.uk/1/hi/magazine/3700209.stm.Reicher,S.D.,&Haslam,S.A.(inpress).Ontheagencyofindividualsandgroups:LessonsfromtheBBCPrisonStudy.InT.Postmes&J.Jetten(Eds.)Individualityandthegroup:Advancesinsocialidentity.London:Sage.Reicher,S.D.,&Hopkins,N.(1996).Seekinginuencethroughcharacterisingself-categories:Ananalysisofanti-abortionistrhetoric.BritishJournalofSocialPsychology,297–311.Reicher,S.D.,&Hopkins,N.(2001a).Psychologyandtheendofhistory:Acritiqueandaproposalforthepsychologyofsocialcategorization.PoliticalPsychology,383–407.Reicher,S.D.,&Hopkins,N.(2001b).Selfandnation:Categorization,contestationand.London:Sage.Reicher,S.D.,&Hopkins,N.(2004).Onthescienceoftheartofleadership.InD.vanKnippenberg&M.A.Hogg(Eds.),Leadership,powerandidentity(pp.197–209).London:Sage.Reicher,S.D.,&Levine,M.(1994).Deindividuation,powerrelationsbetweengroupsandtheexpressionofsocialidentity:Theeffectsofvisibilitytotheoutgroup.BritishJournalofSocialPsychology,145–163.Reicher,S.D.,Spears,R.,&Postmes,T.(1995).AsocialidentitymodelofdeindividuationEuropeanReviewofSocialPsychology,161–198.Reynolds,K.J.,&Platow,M.J.(2003).Whypowerinorganizationsreallyshouldbeshared:Understandingpowerthroughtheperilsofpowerlessness.InS.A.Haslam,D.vanKnippenberg,M.J.Platow&N.Ellemers(Eds.),Socialidentityatwork:Developingtheoryfororganizationalpractice.Philadelphia,PA:PsychologyPress.StephenReicherandS.AlexanderHaslam Reynolds,K.J.,Turner,J.C.,Haslam,S.A.,&Ryan,M.K.(2001).Theroleofpersonalityandgroupfactorsinexplainingprejudice.JournalofExperimentalSocialPsychology,427–434.Robinson,W.P.(Ed.).(1996).Socialgroupsandidentities:DevelopingthelegacyofHenriTajfelOxford:Butterworth-Heinemann.Schwarzer,R.(2001).Stress,resources,andproactivecoping.Appliedpsychology:Aninternationalreview,400–407.Sherif,M.(1956).Experimentsingroupconict.ScientiÞcAmerican,54–58.Sherif,M.(1966).Groupconßictandco-operation:Theirsocialpsychology.London:RoutledgeandKeganPaul.Sieber,J.E.,Iannuzzo,R.,&Rodriguez,B.(1995).Deceptionmethodsinpsychology:Havetheychangedin23years?EthicsandBehavior,67–85.Smith,E.R.,&Mackie,D.M.(2000).Socialpsychology(2nded).NewYork:Worth.Smith,H.J.,Tyler,T.R.,&Huo,Y.(2003).Interpersonaltreatment,socialidentityandorganizationalbehaviour.InS.A.Haslam,D.vanKnippenberg,M.J.Platow,&N.EllemersSocialidentityatwork:Developingtheoryfororganizationalpractice.Philadelphia,PA:PsychologyPress.Smith,S.S.,&Richardson,D.(1983).Ameliorationofdeceptionandharminpsychologicalresearch.JournalofPersonalityandSocialPsychology,1075–1082.Smithson,M.(2000).StatisticswithconÞdence.London:Sage.Spears,R.,Oakes,P.J.,Ellemers,N.,&Haslam,S.A.(Eds.),(1997).Thesocialpsychologyofstereotypingandgrouplife.Oxford:Blackwell.Spears,R.,&Smith,H.J.(2001).Experimentsaspolitics.PoliticalPsychology,309–330.Staub,E.(1989).Therootsofevil:Theoriginsofgenocideandothergroupviolence.Cambridge:CambridgeUniversityPress.Tajfel,H.(1978).Interpersonalbehaviourandintergroupbehaviour.InH.Tajfel(Ed.),Differentiationbetweensocialgroups:Studiesinthesocialpsychologyofintergrouprelations(pp.27–60).London:AcademicPress.Tajfel,H.(Ed.).(1982).Socialidentityandintergrouprelations.Cambridge:CambridgeUniversityPress.Tajfel,H.,Flament,C.,Billig,M.G.,&Bundy,R.F.(1971).Socialcategorizationandintergroupbehaviour.EuropeanJournalofSocialPsychology,149–177.Tajfel,H.,&Turner,J.C.(1979).Anintegrativetheoryofintergroupconict.InW.G.Austin&S.Worchel(Eds.),Thesocialpsychologyofintergrouprelations(pp.33–47).Monterey,CA:Tedeschi,J.T.(Ed.).(1981).ImpressionmanagementtheoryandsocialpsychologicalresearchNewYork:AcademicPress.Terry,D.J.,Carey,C.J.,&Callan,V.J.(2001).Employeeadjustmenttoanorganizationalmerger:Anintergroupperspective.PersonalityandSocialPsychologyBulletin,267–290.Tetlock,P.E.(1985).Accountability:Theneglectedsocialcontextofjudgementandchoice.ResearchinOrganizationalBehavior,297–332.Thayer,S.,&Saarni,C.(1975).Demandcharacteristicsareeverywhere(anyway):AcommentontheStanfordPrisonExperiment.AmericanPsychologist,1015–1016.Turner,J.C.(1981).Someconsiderationsingeneralizingexperimentalsocialpsychology.InG.M.Stephenson&J.H.Davis(Eds.),Progressinappliedsocialpsychology(Vol.1,pp.3–34).Chichester:Wiley.Turner,J.C.(1982).Towardsacognitiveredenitionofthesocialgroup.InH.Tajfel(Ed.),identityandintergrouprelations(pp.15–40).Cambridge:CambridgeUniversityPress.Turner,J.C.(1985).Socialcategorizationandtheself-concept:Asocialcognitivetheoryofgroupbehaviour.InE.J.Lawler(Ed.),Advancesingroupprocesses(Vol.2,pp.77–122).Greenwich,CT:JAIPress.Turner,J.C.(1991).Socialinßuence.MiltonKeynes:OpenUniversityPress.Thepsychologyoftyranny Turner,J.C.(1999).Somecurrentissuesinresearchonsocialidentityandself-cate-gorizationtheories.InN.Ellemers,R.Spears,&B.Doosje(Eds.),Socialidentity:Context,commitment,(pp.6–34).Oxford:Blackwell.Turner,J.C.(2005).Explainingthenatureofpower:Athree-processtheory.EuropeanJournalofSocialPsychology,1–22.Turner,J.C.,&Brown,R.J.(1978).Socialstatus,cognitivealternativesandintergrouprelations.InH.Tajfel(Ed.),Differentiationbetweensocialgroups:Studiesinthesocialpsychologyofintergrouprelations(pp.201–234).London:AcademicPress.Turner,J.C.,&Haslam,S.A.(2001).Socialidentity,organizationsandleadership.InM.E.TurnerGroupsatwork:Advancesintheoryandresearch(pp.25–65).Hills-dale,NJ:Erlbaum.Turner,J.C.,Hogg,M.A.,Oakes,P.J.,Reicher,S.D.,&Wetherell,M.S.(1987).Rediscoveringthesocialgroup:Aself-categorizationtheory.Oxford:Blackwell.Turner,J.C.,Oakes,P.J.,Haslam,S.A.,&McGarty,C.A.(1994).Selfandcollective:Cognitionandsocialcontext.PersonalityandSocialPsychologyBulletin,454–463.Turner,J.C.,Reynolds,K.J.,Haslam,S.A.,&Veenstra,K.(inpress).Reconceptualizingpersonality:Producingindividualitythroughdeningthepersonalself.InT.Postmes&J.JettenIndividualityandthegroup:Advancesinsocialidentity.London:Sage.Tyler,T.R.,&Blader,S.L.(2000).Cooperationingroups.Proceduraljustice,socialidentity,andbehavioralengagement.Philadelphia,PA:PsychologyPress.Underwood,P.W.(2000).Socialsupport:Thepromiseandreality.InB.H.Rice(Ed.),ofstress,copingandhealth(pp.367–391).NewburyPark,CA:Sage.Wright,S.C.(1997).Ambiguity,socialinuence,andcollectiveaction:Generatingcollectiveprotestinresponsetotokenism.PersonalityandSocialPsychologyBulletin,1277–1290.Wright,S.C.,&Taylor,D.M.(1998).Respondingtotokenism:Individualactioninthefaceofcollectiveinjustice.EuropeanJournalofSocialPsychology,647–667.Wright,S.C.,Taylor,D.M.,&Moghaddam,F.M.(1990).Respondingtomembershipinadisadvantagedgroup:Fromacceptancetocollectiveprotest.JournalofPersonalityandSocialPsychology,994–1003.Zimbardo,P.(1969).Thehumanchoice:Individuation,reasonandorderversusdeindividuation,impulseandchaos.InW.J.Arnold&D.Levine(Eds.),Nebraskasymposiumonmotivation(Vol.17,pp.237–307).Lincoln,NE:UniversityofNebraskaPress.Zimbardo,P.(1989).Quietrage:TheStanfordprisonstudyvideo.Stanford,CA:StanfordUniversity.Zimbardo,P.(2001).Opposingterrorismbyunderstandingthehumancapacityforevil.onPsychology,48–50.Zimbardo,P.G.,Maslach,C.,&Haney,C.(1999).Reectionsonthestanfordprisonexperiment:Genesis,transformations,consequences.InT.Blass(Ed.),Obediencetoauthority:CurrentperspectivesontheMilgramparadigm(pp.193–237).Mahwah,NJ:Erlbaum.Received6October2004;revisedversionreceived22April2005StephenReicherandS.AlexanderHaslam