THE PSYCHOLOGY ‐ OF SELF DEFENSE: ‐ SELF AFFIRMATION THEORY David K. Sherman GeoVrey L. Cohen I.Introduction Inmajorleaguebaseball,ahittercouldhavealongandproductivecareerby maintaining a .300 average, that is, by getting a base hit 30% of the time. Agreatdealofmoneycouldbeearnedandfameaccrued.Yettheother70% of the time, this player would have failed. The vast majority of attempts to hittheballwouldresultin‘‘makinganout’’andthusposeapotentialthreat to the player’s sense of personal worth and social regard. Like major league baseball players, people in contemporary society face innumerable failures and self‐threats. These include substandard perfor- mance on the job or in class, frustrated goals or aspirations, information challengingthevalidityoflong‐heldbeliefs,illness,thedefeatofone’spoliti- calpartyinanelectionorofone’sfavoritesportsteaminaplayoV,scientific evidence suggesting that one is engaging in risky health behavior, negative feedbackatworkorinschool,rejectioninaromanticrelationship,realand perceivedsocialslights,interpersonalandintergroupconflict,themisbehav- iorofone’schild,thelossofalovedone,andsoon.Inthecourseofagiven day,thepotentialnumberofeventsthatcouldthreatenpeople’s‘‘moraland adaptiveadequacy’’—theirsenseofthemselvesasgood,virtuous,successful, and able to control important life outcomes (Steele, 1988)—seems limitless and likely to exceed the small number of events that aYrm it. A major undertaking for most people is to sustain self‐integrity when faced with the inevitable setbacks and disappointments of daily life—the 70% of the time 183 ADVANCESINEXPERIMENTAL Copyright2006,ElsevierInc. SOCIALPSYCHOLOGY,VOL.38 Allrightsreserved. DOI:10.1016/S0065-2601(06)38004-5 0065-2601/06$35.00 184 DAVID K. SHERMAN AND GEOFFREY L. COHEN ‘‘atbat’’whentheydonotgetabasehit.Howdoindividualsadapttosuch threatsanddefendself‐integrity? Muchresearchsuggeststhatpeoplehavea‘‘psychologicalimmunesystem’’ that initiates protective adaptations when an actual or impending threat is perceived(Gilbert,Pinel,Wilson,Blumberg,&Wheatley,1998).Psychologi- cal adaptations to threats include the various cognitive strategies and even distortions whereby people come to construe a situation in a manner that renders it less threatening to personal worth and well‐being. Many of these psychologicaladaptationscanbethoughtofasdefensiveinnature,insofaras they alter the meaning of the event in a way that shields people from the conclusion that their beliefs or actions were misguided. Psychologists have documented a wide array of such psychological adaptations that help peopletoprotecttheirself‐integrityinresponsetothreat. Indeed,defensiveadaptationsaresostubbornandpervasivethatGreenwald (1980)describedtheegoas‘‘totalitarian’’initsambitiontointerpretthepast and present in a way congenial to its desires and needs. People view them- selves as a potent causal agent even over events that they cannot control (Langer,1975);theyviewthemselvesasselectivelyresponsibleforproducing positive rather than negative outcomes (Greenwald, 1980; Miller & Ross, 1975; Taylor, 1983). They resist change or—if they do change—become more extreme versions of what they were before (Lord, Ross, & Lepper, 1979).Peopledismisshealthinformationsuggestingthattheyareatriskfor disease or should change their risky behavior (Jemmott, Ditto, & Croyle, 1986;Kunda,1987).Studentsmaydisidentifywith,ordownplaytheperson- al importance of, domains where they fail, thus sustaining self‐worth but precluding the opportunity for improvement (Major, Spencer, Schmader, Wolfe,&Crocker,1998;Steele,1997).Peopleareoveroptimisticintheirpre- dictions of future success and estimations of their current knowledge and competence(Dunning,GriYn,Milojkovic,&Ross,1990;Kruger&Dunning, 1999).Indeed,thesedefensiveadaptationsmayevenbenefitpsychologicaland physicalhealth(Taylor&Brown,1988).Althoughwesuspectthatpeoplecan bemorerealisticandmoreself‐criticalthanthisresearchsuggests,andthattheir optimism and positive illusions may be magnified in certain contexts rather thanothers(seeArmor&Taylor,2002),theideathatpeopleareegodefensive resonates both with psychological research and lay wisdom. An important question, then, concerns the circumstances under which people are less ego defensiveandmoreopen‐mindedintheirrelationshipwiththesocialworld. We see defensive responses as adaptations aimed at ameliorating threats to self‐integrity. The vast research on defensive biases testifies to their robustnessandtothefrequencywithwhichpeopleusethem.Althoughthese defensive responses are adaptive in the sense of protecting or enhancing an individual’ssenseofself‐integrity,theycanbemaladaptivetotheextentthey forestall learning from important, though threatening, experiences and SELF‐AFFIRMATIONTHEORY 185 information.Moreover, peoples’eVortstoprotectself‐integritymaythreat- en the integrity of their relationships with others (Cohen et al., 2005; Murray,Holmes,MacDonald,&Ellsworth,1998).Yet,thesenormaladap- tations can be ‘‘turned oV’’ through an altogether diVerent psychological adaptation to threat, an alternative adaptation that does not hinge on distorting the threatening event to render it less significant. One way that these defensive adaptations can be reduced, or even eliminated, is through the process of self‐aYrmation (Aronson, Cohen, & Nail, 1999; Sherman & Cohen,2002; Steele, 1988). Steele(1988)firstproposedthetheoryofself‐aYrmation.Itassertsthatthe overallgoaloftheself‐systemistoprotectanimageofitsself‐integrity,ofits moralandadaptiveadequacy.Whenthisimageofself‐integrityisthreatened, people respond in such a way as to restore self‐worth. As noted previously, onewaythatthisisaccomplishedisthroughdefensiveresponsesthatdirectly reducethethreat.Butanother wayisthrough theaYrmationofalternative sources of self‐integrity. Such ‘‘self‐aYrmations,’’ by fulfilling the need to protect self‐integrity in the face of threat, can enable people to deal with threateningeventsandinformationwithoutresortingtodefensivebiases. Inthispaper,weupdatethefieldonresearchconductedusingself‐aYrma- tiontheoryasaframework.Thisresearchilluminatesboththemotivational processes underlying self‐integrity maintenance and the implications of such processes for many domains of psychology. We illustrate how self‐ aYrmation aVects not only people’s cognitive responses to threatening information and events, but also their physiological adaptations and actual behavior. The research presented has implications for psychological and physical health, education, social conflict, closemindedness and resistance to change, prejudice and discrimination, and a variety of other important applied areas. We also examine how self‐aYrmations reduce threats to the self at the collective level, such as when people confront threatening in- formation about their groups. We then review factors that qualify or limit theeVectivenessofself‐aYrmations,includingsituationswhereaYrmations backfire, and lead to greater defensiveness and discrimination. We discuss the connection of self‐aYrmation theory to other motivational theories of self‐defense and review relevant theoretical and empirical advances. We conclude with adiscussion ofthe implications ofself‐aYrmation theory for interpersonal relationships and coping. A. OVERVIEW OFSELF‐AFFIRMATIONTHEORY Self‐aYrmation theory (Aronson et al., 1999; Sherman & Cohen, 2002; Steele,1988)beginswiththepremisethatpeoplearemotivated tomaintain the integrity of the self. Integrity can be defined as the sense that, on the 186 DAVID K. SHERMAN AND GEOFFREY L. COHEN whole, one is a good and appropriate person. Cultural anthropologists use the term ‘‘appropriate’’ to refer to behavior that is fitting or suitable given the cultural norms and the salient demands on people within that culture. Thus, the standards for what it means to be a good person vary across cultures,groups,andsituations(e.g.,Heine,2005).Suchstandardsofinteg- rity can include the importance of being intelligent, rational, independent, and autonomous, and exerting control over important outcomes. Such standardsofintegritycanalsoincludetheimportanceofbeingagoodgroup memberandofmaintainingcloserelationships.Threatstoself‐integritymay thustakemanyformsbuttheywillalwaysinvolverealandperceivedfailures to meet culturally or socially significant standards (Leary & Baumeister, 2000).Consequently,peoplearevigilanttoeventsandinformationthatcall their self‐integrity into question, both in their own eyes and in the eyes of others.Insuchsituations,peopletrytorestoreorreasserttheintegrityofthe self. Thus, the goal of protecting self‐integrity, and the impact of that goal onpsychologyandbehavior,becomesapparentwhenintegrityisthreatened. There are three categories of responses that people deploy to cope with suchthreats.First,theycanrespondbyaccommodatingtothethreat.That is,theycanacceptthefailureorthethreateninginformationandthenuseit asabasisforattitudinalandbehavioralchange.However,totheextentthat thethreateneddomainconcernsanimportantpartofone’sidentity,theneed to maintain self‐integrity can make it diYcult to accept the threatening informationandtochangeone’sattitudeorbehavioraccordingly.Asecond responsethusinvolvesamelioratingthethreatviadirectpsychologicaladap- tations. While some direct adaptations preserve the fundamental infor- mational value of the event while also changing one’s construal of that event (e.g., framing a failure as a learning opportunity; Dweck & Leggett, 1988),otherdirectpsychologicaladaptationsaredefensiveinnatureinthat they involve dismissing, denying, or avoiding the threat in some way. We refer to these responses as defensive biases (see Sherman & Cohen, 2002). Although a defensive bias can restore self‐integrity, the rejection of the threateninginformationcanlessentheprobabilitythatthepersonwilllearn from thepotentially important information. Self‐aYrmation theory proposes a third alternative, a diVerent kind of psychological adaptation—one that, under many circumstances, enables both the restoration of self‐integrity and adaptive behavior change. People canrespondtothreatsusingtheindirectpsychologicaladaptationofaYrming alternativeself‐resourcesunrelatedtotheprovokingthreat.Such‘‘self‐aYr- mations’’ include reflecting on important aspects of one’s life irrelevant to the threat, or engaging in an activity that makes salient important values unconnectedtothethreateningevent.Whereasdefensivepsychologicaladap- tations directly address the threatening information, indirect psychological SELF‐AFFIRMATIONTHEORY 187 adaptations, such as self‐aYrmation, allow people to focus on domains of self‐integrity unrelated to the threat. When self‐aYrmed in this manner, peoplerealizethattheirself‐worthdoesnothingeontheevaluativeimplica- tions of the immediate situation. As a result, they have less need to distort or reconstrue the provoking threat and can respond to the threatening informationin a more open and evenhanded manner. B. BASIC TENETS OF SELF‐AFFIRMATION THEORY Much research within the self‐aYrmation framework examines whether an aYrmation of self‐integrity, unrelated to a specific provoking threat, can attenuate or eliminate people’s normal response to that threat. If it does, thenonecaninferthattheresponsewasmotivatedbyadesiretoprotectself‐ integrity.Theself‐aYrmationframeworkencompassesfourtenets,whichare enumerated below: 1. People are Motivated to Protect the PerceivedIntegrity and Worth of theSelf The most basic tenet of self‐aYrmation theory (Steele, 1988) is that people are motivated to protect the perceived integrity and worth of the self. As Steeleobserved,thepurposeoftheself‐systemisto‘‘maintainaphenomenal experience of the self ... as adaptively and morally adequate, that is, competent, good, coherent, unitary, stable, capable of free choice, capable ofcontrollingimportantoutcomes...(p.262).’’Theseself‐conceptionsand imagesmakinguptheself‐systemcanbethoughtofasthediVerentdomains that are important to an individual, or the diVerent contingencies of a person’s self‐worth (Crocker & Wolfe, 2001). Figure 1 presents a schematic oftheself‐system.The selfiscomposed ofdiVerent domains,whichinclude an individual’s roles, such as being a student or a parent; values, such as being religious or having a sense of humor; social identities, such as mem- bershipingroupsororganizationsandinracial,cultural,andgendergroups; andbeliefsystems,suchaspoliticalideologies.The selfisalsocomposed of people’s goals, such as the value of being healthy or succeeding in school. Theself‐systemisactivatedwhenapersonexperiencesathreattoanimpor- tant self‐conception or image. Such threat poses a challenge to a desired self‐conception. Thus, failure feedback could threaten a person’s identity as a student, negative health information could threaten a person’s self‐ conception as a healthful individual, news about anti‐American sentiment couldthreatenaperson’spatrioticidentity,andevidenceofsocialinequality could challenge a person’s belief in a just world (Lerner, 1980). All of these 188 DAVID K. SHERMAN AND GEOFFREY L. COHEN Fig.1. Schematicrepresentationofself‐system. eventsarethreateningbecause theyhave implicationsforaperson’soverall sense of self‐integrity. 2. Motivationsto Protect Self‐Integritycan Result in DefensiveResponses Whenself‐integrityisthreatened,peoplearemotivatedtorepairit,andthis motivation can lead to defensive responses. The defensive responses may seem rational and defensible, though they are more ‘‘rationalizing’’ than ‘‘rational’’ (Aronson, 1968; Kunda, 1990; Pyszczynski & Greenberg, 1987). They serve to diminish the threat and consequently, restore the perceived integrity of the self. These defensive responses can be automatic and even unconscious in nature, and indeed, the rapidity with which people respond to threatsspeaks to theimportance ofself‐integrity maintenance. 3. The Self‐System isFlexible People often compensate for failures in one aspect of their lives by empha- sizing successes in other domains. Personality theorists, such as Allport (1961) and Murphy (1947), have advanced this notion of compensation, and self‐aYrmation theory is consistent with this claim (see also Brown & Smart, 1991). Because the goal of the self‐system focuses on maintaining the overall worth and integrity of the self, people can respond to threats in one domain by aYrming the self in another domain. This fungibility in the sourcesofself‐integrityiswhatcanenablesmokers,forexample,tomaintain a perception of worth and integrity despite the potentially threatening conclusion that they are acting in a maladaptive, harmful, and irrational SELF‐AFFIRMATIONTHEORY 189 way (Steele, 1988). AYrmations satisfy the motivation to maintain self‐ integrity—thus, they reduce the normal psychological adaptations people engageto ameliorate a specific provoking threat. 4. People canbe AYrmed by Engaging inActivities that Remind them of‘‘WhoThey Are’’ (And Doing soReduces theImplicationsfor Self‐Integrity ofThreatening Events) Those qualities that are central to how people see themselves are potential domains of self‐aYrmation. Such aYrmations can concern friends and family, making art or music, a charity, or the observance of one’s religion. In a diYcult situation, reminders of these core qualities can provide people with perspective on who they are and anchor their sense of self‐integrity in thefaceofthreat.A‘‘self‐aYrmation’’makessalientoneoftheseimportant core qualities or sources of identity. Operationally, self‐aYrmations are typically ideographic, in that people first report an important value or life domain, and then they are given the opportunity either to write an essay about it or to complete a scale or exercise that allows them to assert its importance (McQueen &Klein,2005). Whenglobalperceptionsofself‐integrityareaYrmed,otherwisethreaten- ing events or information lose their self‐threatening capacity because the individualcanviewthemwithinabroader,largerviewoftheself.Peoplecan thus focus not on the implications for self‐integrity of a given threat or stressor,butonitsinformationalvalue.Whenself‐aYrmed,individualsfeel asthoughthetaskofprovingtheirworth,bothtothemselvesandtoothers, is ‘‘settled.’’ As a consequence, they can focus on other salient demands in the situation beyond ego protection. II. Self‐AYrmationand Threats tothe Individual Self A great deal of research has used self‐aYrmation theory to address a wide rangeofpsychologicalphenomena,includingbiasedinformationprocessing, causal attributions, cognitive dissonance, prejudice and stereotyping, stress andrumination.Whatconnectsthesedisparateareasofresearchisthatthey all address situations or events where people contend with a threat to a valued self‐image. We first review research on the impact of experienced threatstoself‐identities,suchasone’spoliticalidentityorone’sidentityasa healthfulorintelligentindividual.Wethenreviewresearchontheimpactof experienced threats to collective identities, such as one’s team or racial group. 190 DAVID K. SHERMAN AND GEOFFREY L. COHEN A. MOTIVATEDINFERENCES AND BIASED ASSIMILATION Peopleofteninterpretnewinformationinawaythatreinforcestheirbeliefs and desires. One striking example occurred during the 2000 presidential controversy in Florida. Evaluations of seemingly arbitrary events (e.g., beliefs about appropriate policies for dealing with hanging chads, whether to count votes from certain counties) were consistently and predictably aligned with people’s political identities. The Florida situation presented a series of novel issues where partisanship lines had not, at first, been clearly defined. Yet, in little time, people were able to determine their criteria for establishing voter intent, and in most cases, their criteria were consistent with their partisanship, and the position that they ultimately advocated served theinterestsoftheir candidateofchoice (cf.Gerber &Green, 1999). ItisinterestingtoimaginewhatwouldhavehappenedduringtheFlorida debacleifthesituationwerereversed,andifVicePresidentAlbertGorewere leading rather than Governor George W. Bush as the vote was contested. That is, if the situation was ‘‘counterbalanced’’ like a proper psychology experiment, would Republican leaders have argued for the sanctity of the vote and the need to count each chad and would Democratic leaders have argued against them? Research on biased assimilation by Lord et al. (1979) suggeststhattheywould.Inthisclassicstudy,proponentsandopponentsof capital punishment evaluated two studies on the eYcacy of capital punish- ment as a deterrent. The two studies featured diVerent designs (a panel design comparing murder rates from states before and after the implemen- tation of capital punishment policy, and a concurrent design comparing murder rates from states that either used capital punishment or did not). Unlike the situation in Florida, the study was completely counterbalanced. Thatis,twoversionsofeachstudywerecreated,onethatsupportedandone that refuted the deterrent value of capital punishment. Thus, the study allows an examination of whether and to what extent prior beliefs (or identities) bias the interpretation of information. When the study was consistent with participants’ prior beliefs, they thought it was better conducted and more convincing regardless of the specific design (concurrent versus panel) of the study. Moreover, reading a mixed bag of evidence—one study supporting and one study contradicting their beliefs—led participants to report becoming more confident in the validity of their beliefs about the deterrent eYcacy of the death penalty. Thestudyprovidesanexperimentalanalogtothehypotheticalswitchinthe Florida election scenario. Regardless of what evidence people were pre- sented, they evaluated it in a way consistent with their prior beliefs and, in turn, this biased assimilation appeared to strengthen the very prior beliefs that give rise to the bias in evaluation. SELF‐AFFIRMATIONTHEORY 191 Lord et al. (1979) emphasized the role of cool cognitive inferential pro- cesses in producing biased assimilation. Like Bayesian theorists, people use their prior beliefs to evaluate the validity of incoming data. It is thus logically permissible for them to reject belief‐incongruent evidence and to acceptwithlittle scrutiny belief‐congruentevidence (Nisbett&Ross,1980). Incontrast,wearguethatsuchbiasesissuefromthemotivationtomaintain andprotectidentity(seealsoMunro&Ditto,1997).Thatis,partisansinthe Lordetal.(1979)studyscrutinizedtheevidenceinawaythatwouldprotect theiridentityeitherasa‘‘lawandorder’’conservativeorasa‘‘humanitarian’’ liberal(Cohen,2003;Ellsworth&Ross,1983). According to our analysis, the need to protect a valued identity or self‐ view is a major source of such biased processing and closed mindedness. Becauselong‐heldbeliefsareoftentiedtoimportantidentities,theymaybe given up only with great reluctance, and they may be embraced even when they conflict with the demands of fact, logic, or material self‐interest (e.g., Abelson, 1986; Sears & Funk, 1991). Yet, people possess other important identities and values that they can draw on when they encounter belief‐ threateninginformation.ProvidingthemwithanaYrmationofoneofthese alternative sources of self‐integrity should enable them to evaluate the threatening information in a less biased and defensive manner. We tested this logic in a study of biased assimilation in the domain of capital punish- ment (Cohen, Aronson, & Steele, 2000). In a study patterned after Lord et al. (1979), proponents and opponents of capital punishment read an article from the Journal of Law and Human Behavior titled ‘‘The Death Penalty:NewEvidenceInformsanOldDebate.’’Thearticlewasfabricated, butcontainedfacts,statistics,andargumentsthatwagedapersuasiveassault on participants’ attitudes toward capital punishment. Thus, proponents of capital punishment read an anti‐capital punishment report and opponents of capitalpunishment read a pro‐capital punishment report. Prior to reading the article on capital punishment, all participants com- pleted a writing exercise that constituted our self‐aYrmation manipulation. Participantsintheself‐aYrmationconditionwroteanessayaboutaperson- alvaluethattheyhadrated,duringpretest,aspersonallyimportant(suchas their relationships with friends or sense of humor). Specifically, they were asked to describe three to four personal experiences where the value had beenimportanttothemandhadmadethemfeelgoodaboutthemselves.The value they wrote about was, in all cases, unrelated to their political views. Participants in the no‐aYrmation condition wrote abouta neutral topic. In the no‐aYrmation control condition, our findings mirrored those of Lordetal.(1979).Participantsfoundflawsinthemethodologyofthestudies thatcontradictedtheirpoliticalbeliefs,theysuspectedbiasonthepartofthe authors of the report, and they persisted in their attitudes about capital 192 DAVID K. SHERMAN AND GEOFFREY L. COHEN punishment. By contrast, the responses of participants who aYrmed a valuedself‐identityprovedmorebalanced.Thatis,self‐aYrmedparticipants werelesscriticalofthereportedresearchandsuspectedlessbiasonthepart oftheauthorsofthereport.Participantsevenchangedtheirglobalattitudes toward capital punishment in the direction of the report they read. That is, proponents of capital punishment supported the death penalty less, and opponentsofcapitalpunishmentsupporteditmore(Cohenetal.,2000,Study 2). That both partisan groups showed the eVect attests to the power of the psychologicalmechanism.Notonlydidproponentscometoprivilege‘‘life’’ over‘‘lawandorder’’morewhenself‐aYrmedthanwhennot,butopponents ofcapitalpunishment,whenself‐aYrmed,cametosupportstate‐sanctioned executiontoagreaterextent(seealsoJacks&O’Brien,2004). A third study examined how people evaluate others who either agree or disagreewith their beliefs (Cohen et al., 2000). Pro‐choice and pro‐life parti- sans werepresentedwitha debate between two activists on opposite sidesof theabortiondispute.Participantswhodidnotreceiveaself‐aYrmationjudged theactivistwhosharedtheirconvictionsmorefavorablythantheactivistwho didnot(seealsoHastorf&Cantril,1954).Incontrast,participantswhowere given a self‐aYrmation became morebalancedintheir evaluation ofthe two activistsandassertedthattheywererelativelylessconfidentofthevalidityof theirabortionattitudes,relativetotheirnonaYrmedpeers. The motivation to protect identity may prove especially consequential in thecontextofnegotiation.Innegotiation—betweenparties,nations,indivi- duals—there exists a barrier to compromise, a barrier that often leads disputing factions to reject even mutually beneficial settlements and instead persist in mutually destructive conflict (Ross & Ward, 1995; Sherman, Nelson, & Ross, 2003). This barrier issues, in part, from a motivation to defendone’spolitical, national, or regional identity—a motivationthat can result in intransigence and stalemate (De Dreu & Carnevale, 2003). To accept compromise entails acting in a way that could exact painful costs to self and social identity. Compromise entails that one must accept policies anathema to one’s past ideological commitments, or take courses of action contrarytotheinterestsofone’sconstituenciesandgrouployalties.Accord- ingly,negotiationshouldbefacilitated when people’s partisan identitiesare made less vital to their sense of self‐integrity through the aYrmation of alternative sources ofself‐worth. In one study, pro‐choice participants entered into a negotiation with a pro‐life advocate about appropriate federal abortion policy (Cohen et al., 2005). To make their partisan identity salient prior to the negotiation, participantswerefirstaskedtoasserttheir‘‘truebeliefs’’onabortionpolicy. Additionally, prior to the negotiation, half of the participants received a values aYrmation and half did not. It was found that aYrmation increased