Anarchy Is What States Make of It: the Social Construction of Power Politics Author(S): Alexander Wendt Reviewed Work(S): Source: International Organization, Vol

Anarchy is what States Make of it: The Social Construction of Power Politics Author(s): Alexander Wendt Reviewed work(s): Source: International Organization, Vol. 46, No. 2 (Spring, 1992), pp. 391-425 Published by: The MIT Press Stable URL: http://www.jstor.org/stable/2706858 . Accessed: 20/12/2011 11:58

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

The MIT Press is collaborating with JSTOR to digitize, preserve and extend access to International Organization.

http://www.jstor.org Anarchyis whatstates make ofit: the social constructionof power politics AlexanderWendt

The debate betweenrealists and liberalshas reemergedas an axisof contention in internationalrelations theory.1 Revolving in the past around competing theoriesof humannature, the debate is moreconcerned today with the extent to whichstate action is influencedby "structure" (anarchy and the distribution of power) versus "process" (interactionand learning)and institutions.Does the absence of centralizedpolitical authority force states to play competitive powerpolitics? Can internationalregimes overcome this logic, and underwhat conditions?What in anarchyis givenand immutable,and whatis amenable to change? The debate between "neorealists"and "neoliberals" has been based on a shared commitmentto "rationalism."2Like all social theories,rational choice directsus to ask some questions and not others,treating the identitiesand interestsof agentsas exogenouslygiven and focusingon how the behaviorof

This articlewas negotiatedwith many individuals. If myrecords are complete(and apologies if theyare not), thanksare due particularlyto JohnAldrich, Mike Barnett,Lea Brilmayer,David Campbell,Jim Caporaso, Simon Dalby, David Dessler, Bud Duvall, Jean Elshtain,Karyn Ertel, Lloyd Etheridge,Ernst Haas, MartinHollis, Naeem Inayatullah,Stewart Johnson, Frank Klink, Steve Krasner,Friedrich Kratochwil, David Lumsdaine,M. J. Peterson,Spike Peterson,Thomas Risse-Kappen,John Ruggie, Bruce Russett,Jim Scott, Rogers Smith,David Sylvan,Jan Thomson, MarkWarren, and JuttaWeldes. The articlealso benefitedfrom presentations and seminarsat the American University,the Universityof Chicago, the Universityof Massachusettsat Amherst, SyracuseUniversity, the Universityof Washingtonat Seattle, the Universityof Californiaat Los Angeles,and Yale University. 1. See, forexample, Joseph Grieco, "Anarchy and the Limitsof Cooperation:A RealistCritique ofthe Newest Liberal Institutionalism,"Intemational Organization 42 (Summer1988), pp. 485-507; JosephNye, "Neorealism and Neoliberalism,"World Politics 40 (January1988), pp. 235-51; Robert Keohane, "NeoliberalInstitutionalism: A Perspectiveon WorldPolitics," in hiscollection of essays entitledIntemational Institutions and StatePower (Boulder, Colo.: WestviewPress, 1989), pp. 1-20; JohnMearsheimer, "Back to the Future:Instability in Europe Afterthe Cold War," Intemational Security13 (Summer1990), pp. 5-56, along withsubsequent published correspondence regarding Mearsheimer'sarticle; and EmersonNiou and Peter Ordeshook,"Realism Versus Neoliberalism: A Formulation,"American Joumal of Political Science 35 (May 1991), pp. 481-511. 2. See Robert Keohane, "InternationalInstitutions: Two Approaches," IntemationalStudies Quarterly32 (December 1988), pp. 379-96.

IntemationalOrganization 46, 2, Spring1992 ? 1992 bythe WorldPeace Foundationand the MassachusettsInstitute of Technology 392 InternationalOrganization agents generates outcomes. As such, rationalismoffers a fundamentally behavioralconception of both process and institutions:they change behavior but not identitiesand interests.3In additionto thisway of framingresearch problems,neorealists and neoliberals share generallysimilar assumptions about agents: states are the dominantactors in the system,and theydefine securityin "self-interested"terms. Neorealists and neoliberalsmay disagree about the extentto whichstates are motivatedby relative versus absolute gains, butboth groups take the self-interestedstate as the startingpoint for theory. This startingpoint makes substantive sense forneorealists, since theybelieve anarchies are necessarily"self-help" systems, systems in whichboth central authorityand collectivesecurity are absent.The self-helpcorollary to anarchy does enormouswork in neorealism,generating the inherentlycompetitive dynamicsof the securitydilemma and collectiveaction problem.Self-help is not seen as an "institution"and as such occupies a privilegedexplanatory role vis-a-visprocess, setting the termsfor, and unaffectedby, interaction.Since statesfailing to conformto the logicof self-helpwill be drivenfrom the system, onlysimple learning or behavioraladaptation is possible;the complexlearning involved in redefinitionsof identityand interestis not.' Questions about identity-and interest-formationare thereforenot importantto studentsof internationalrelations. A rationalistproblematique, which reduces process to dynamicsof behavioral interactionamong exogenouslyconstituted actors, definesthe scope of systemictheory. By adopting such reasoning,liberals concede to neorealists the causal powersof anarchicstructure, but theygain the rhetoricallypowerful argument thatprocess can generatecooperative behavior, even in an exogenouslygiven, self-help system. Some liberals may believe that anarchy does, in fact, constitutestates with self-interestedidentities exogenous to practice. Such "weak" liberalsconcede the causal powers of anarchyboth rhetoricallyand substantivelyand accept rationalism'slimited, behavioral conception of the causal powersof institutions.They are realistsbefore liberals (we mightcall them "weak realists"), since only if internationalinstitutions can change powersand interestsdo theygo beyondthe "limits"of realism.5

3. Behavioral and rationalistmodels of man and institutionsshare a common intellectual heritagein the materialistindividualism of Hobbes, Locke, and Bentham. On the relationship between the two models, see JonathanTurner, A Theoryof Social Interaction(Stanford, Calif.: StanfordUniversity Press, 1988), pp. 24-31; and George Homans, "Rational Choice Theoryand Behavioral Psychology,"in Craig Calhoun et al., eds., Structuresof Power and Constraint (Cambridge:Cambridge University Press, 1991), pp. 77-89. 4. On neorealistconceptions of learning,see Philip Tetlock, "Learning in U.S. and Soviet ForeignPolicy," in George Breslauerand PhilipTetlock, eds., Leamingin U.S. and SovietForeign Policy(Boulder, Colo.: WestviewPress, 1991), pp. 24-27. On the.difference between behavioral and cognitivelearning, see ibid., pp. 20-61; Joseph Nye, "Nuclear Learning and U.S.-Soviet SecurityRegimes," Intemational Organization 41 (Summer 1987), pp. 371-402; and Ernst Haas, WhenKnowledge Is Power(Berkeley: University of CaliforniaPress, 1990), pp. 17-49. 5. See Stephen Krasner, "Regimes and the Limits of Realism: Regimes as Autonomous Variables,"in StephenKrasner, ed., IntemationalRegimes (Ithaca, N.Y.: CornellUniversity Press, 1983), pp. 355-68. Anarchy 393

Yet some liberals want more. When Joseph Nye speaks of "complex learning,"or Robert Jervisof "changingconceptions of self and interest,"or RobertKeohane of "sociological"conceptions of interest,each is assertingan importantrole for transformationsof identityand interestin the liberal researchprogram and, byextension, a potentiallymuch stronger conception of processand institutionsin worldpolitics.6 "Strong" liberals should be troubled bythe dichotomousprivileging of structureover process, since transformations of identityand interestthrough process are transformationsof structure. Rationalismhas littleto offersuch an argument,7which is in part why,in an importantarticle, Friedrich Kratochwiland John Ruggie argued that its individualistontology contradicted the intersubjectivistepistemology necessary forregime theory to realize itsfull promise.8 Regimes cannot change identities and interestsif the latterare takenas given.Because of thisrationalist legacy, despiteincreasingly numerous and richstudies of complexlearning in foreign policy,neoliberals lack a systematictheory of how suchchanges occur and thus must privilegerealist insightsabout structurewhile advancing their own insightsabout process. The ironyis thatsocial theorieswhich seek to explainidentities and interests do exist.Keohane has called them"reflectivist";9 because I wantto emphasize theirfocus on the social constructionof subjectivityand minimizetheir image problem,following Nicholas Onuf I will call them "constructivist.""Despite importantdifferences, cognitivists, poststructuralists, standpoint and postmodern feminists,rule theorists,and structurationistsshare a concernwith the basic "sociological" issue bracketed by rationalists-namely,the issue of identity-and interest-formation.Constructivism's potential contribution to a strong liberalism has been obscured, however,by recent epistemological debates betweenmodernists and postmodernists,in whichScience disciplines Dissent for not defininga conventionalresearch program,and Dissent celebratesits liberation from Science.1" Real issues animatethis debate, which

6. See Nye, "Nuclear Learningand U.S.-Soviet SecurityRegimes"; Robert Jervis,"Realism, Game Theory,and Cooperation,"World Politics 40 (April 1988),pp. 340-44; and RobertKeohane, "InternationalLiberalism Reconsidered," in John Dunn, ed., The Economic Limitsto Modem Politics(Cambridge: Cambridge University Press, 1990), p. 183. 7. Rationalistshave givensome attentionto the problemof preference-formation,although in so doingthey have gone beyondwhat I understandas the characteristicparameters of rationalism. See, forexample, Jon Elster, "Sour Grapes: Utilitarianismand the Genesis ofWants," in Amartya Sen and Bernard Williams,eds., Utilitarianismand Beyond (Cambridge: CambridgeUniversity Press,1982), pp. 219-38; and Michael Cohen and RobertAxelrod, "Coping withComplexity: The AdaptiveValue of ChangingUtility,"American Economic Review 74 (March 1984), pp. 30-42. 8. FriedrichKratochwil and JohnRuggie, "International Organization: A State of theArt on an Artof the State,"Intemational Organization 40 (Autumn1986), pp. 753-75. 9. Keohane, "InternationalInstitutions." 10. See Nicholas Onuf, Worldof Our Making(Columbia: Universityof South Carolina Press, 1989). 11. On Science,see Keohane, "InternationalInstitutions"; and RobertKeohane, "International Relations Theory:Contributions of a FeministStandpoint," Millennium 18 (Summer 1989), pp. 245-53. On Dissent, see R. B. J. Walker,"History and Structurein the Theoryof International Relations,"Millennium 18 (Summer 1989), pp. 163-83; and RichardAshley and R. B. J. Walker, 394 InternationalOrganization

also divides constructivists.With respect to the substance of international relations,however, both modern and postmodernconstructivists are interested in how knowledgeablepractices constitute subjects, which is not farfrom the strongliberal interestin how institutionstransform interests. They share a cognitive,intersubjective conception of process in whichidentities and interests are endogenousto interaction,rather than a rationalist-behavioralone in whichthey are exogenous. My objectivein thisarticle is to build a bridgebetween these two traditions (and, by extension,between the realist-liberaland rationalist-reflectivist debates) by developinga constructivistargument, drawn from structurationist and symbolicinteractionist sociology, on behalf of the liberal claim that internationalinstitutions can transformstate identitiesand interests.12In contrastto the "economic" theorizingthat dominatesmainstream systemic international relations scholarship, this involves a "sociological social psychological"form of systemictheory in whichidentities and interestsare the dependentvariable.13 Whether a "communitarianliberalism" is stillliberalism does not interestme here. What does is thatconstructivism might contribute significantlyto the strongliberal interestin identity-and interest-formation and therebyperhaps itself be enrichedwith liberal insights about learningand cognitionwhich it has neglected. My strategyfor building this bridge will be to argue againstthe neorealist claim that self-helpis given by anarchic structureexogenously to process. Constructivistshave not done a good job of takingthe causal powersof anarchy seriously.This is unfortunate,since in the realist view anarchyjustifies disinterestin the institutionaltransformation of identitiesand interestsand thus buildingsystemic theories in exclusivelyrationalist terms; its putative causal powers must be challenged if process and institutionsare not to be subordinatedto structure.I argue that self-helpand power politics do not follow either logicallyor causally fromanarchy and that if today we find ourselvesin a self-helpworld, this is due to process,not structure.There is no

"Reading Dissidence/Writingthe Discipline: Crisis and the Question of Sovereigntyin Interna- tionalStudies," Intemational Studies Quarterly 34 (September1990), pp. 367-416. For an excellent criticalassessment of these debates, see Yosef Lapid, "The Third Debate: On the Prospectsof InternationalTheory in a Post-PositivistEra," IntemationalStudies Quarterly 33 (September1989), pp. 235-54. 12. The fact that I draw on these approaches aligns me withmodernist constructivists, even thoughI also drawfreely on thesubstantive work of postmodernists, especially Richard Ashley and Rob Walker. For a defenseof thispractice and a discussionof its epistemologicalbasis, see my earlier article,"The Agent-StructureProblem in InternationalRelations Theory,"Intemational Organization41 (Summer 1987), pp. 335-70; and Ian Shapiro and Alexander Wendt, "The DifferenceThat Realism Makes: Social Science and the Politics of Consent," forthcomingin Politicsand Society.Among modernist constructivists, my argument is particularlyindebted to the published work of Emanuel Adler, FriedrichKratochwil, and John Ruggie, as well as to an unpublishedpaper by Naeem Inayatullahand David Levine entitled"Politics and Economics in ContemporaryInternational Relations Theory," Syracuse University, Syracuse, N.Y., 1990. 13. See ViktorGecas, "Rekindlingthe SociologicalImagination in Social Psychology,"Joumal forthe Theory of Social Behavior19 (March 1989),pp. 97-115. Anarchy 395

"logic" of anarchyapart fromthe practicesthat create and instantiateone structureof identitiesand interestsrather than another; structurehas no existenceor causal powersapart from process. Self-help and powerpolitics are institutions,not essentialfeatures of anarchy.Anarchy is whatstates make of it. In the subsequentsections of thisarticle, I criticallyexamine the claimsand assumptionsof neorealism,develop a positiveargument about how self-help and power politicsare sociallyconstructed under anarchy,and then explore threeways in whichidentities and interestsare transformedunder anarchy: by the institutionof sovereignty,by an evolutionof cooperation,and by intentionalefforts to transformegoistic identities into collective identities.

Anarchyand powerpolitics

Classical realists such as Thomas Hobbes, Reinhold Niebuhr, and Hans Morgenthauattributed egoism and powerpolitics primarily to humannature, whereas structuralrealists or neorealistsemphasize anarchy.The difference stems in part from differentinterpretations of anarchy's causal powers. KennethWaltz's workis importantfor both. In Man, the State,and War,he definesanarchy as a conditionof possibilityfor or "permissive"cause of war, arguingthat "wars occurbecause thereis nothingto preventthem."14 It is the humannature or domesticpolitics of predatorstates, however, that provide the initial impetus or "efficient"cause of conflictwhich forces other states to respond in kind.15Waltz is not entirelyconsistent about this,since he slips withoutjustification from the permissivecausal claim that in anarchywar is always possible to the active causal claim that "war may at any moment occur."'16But despite Waltz's concluding call for third-imagetheory, the efficientcauses that initializeanarchic systems are fromthe firstand second images.This is reversedin Waltz's Theoryof International Politics, in whichfirst- and second-imagetheories are spurned as "reductionist,"and the logic of anarchyseems by itselfto constituteself-help and power politicsas necessary featuresof world politics.17 This is unfortunate,since whatever one maythink of first- and second-image theories, they have the virtue of implyingthat practices determine the characterof anarchy.In the permissiveview, only if human or domesticfactors cause A to attack B will B have to defend itself.Anarchies may contain dynamicsthat lead to competitivepower politics, but theyalso maynot, and we can argueabout when particular structures of identity and interestwill emerge.

14. Kenneth Waltz, Man, the State,and War (New York: Columbia UniversityPress, 1959), p. 232. 15. Ibid.,pp. 169-70. 16. Ibid., p. 232. This pointis made by Hidemi Suganamiin "BringingOrder to the Causes of War Debates," Millennium19 (Spring1990), p. 34, fn.11. 17. KennethWaltz, Theory of Intemational Politics (Boston: Addison-Wesley,1979). 396 InternationalOrganization

In neorealism,however, the role ofpractice in shapingthe character of anarchy is substantiallyreduced, and so thereis less aboutwhich to argue:self-help and competitivepower politics are simplygiven exogenously by the structureof the statesystem. I will not here contestthe neorealistdescription of the contemporarystate systemas a competitive,self-help world;18 I will onlydispute its explanation.I develop my argumentin three stages. First, I disentanglethe concepts of self-helpand anarchyby showingthat self-interestedconceptions of security are not a constitutiveproperty of anarchy.Second, I show how self-helpand competitivepower politicsmay be produced causallyby processesof interaction betweenstates in whichanarchy plays onlya permissiverole. In both of these stages of myargument, I self-consciouslybracket the first-and second- image determinantsof state identity,not because theyare unimportant(they are indeed important),but because likeWaltz's objective,mine is to clarifythe "logic" of anarchy.Third, I reintroducefirst- and second-imagedeterminants to assess theireffects on identity-formationin different kinds of anarchies.

Anarchy,self-help, and intersubjectiveknowledge Waltz definespolitical structure on threedimensions: ordering principles (in thiscase, anarchy),principles of differentiation(which here drop out), and the distributionof capabilities.19By itself,this definition predicts little about state behavior.It does not predictwhether two stateswill be friendsor foes,will recognizeeach other'ssovereignty, will have dynasticties, will be revisionistor status quo powers, and so on. These factors,which are fundamentally intersubjective,affect states' securityinterests and thusthe characterof their interactionunder anarchy. In an importantrevision of Waltz's theory,Stephen Walt impliesas muchwhen he arguesthat the "balance of threats,"rather than the balance of power, determinesstate action, threatsbeing socially constructed.20Put more generally,without assumptions about the structureof identitiesand interestsin the system,Waltz's definitionof structurecannot predictthe contentor dynamicsof anarchy.Self-help is one such intersubjective structureand, as such, does the decisiveexplanatory work in the theory. The questionis whetherself-help is a logicalor contingentfeature of anarchy. In thissection, I develop the conceptof a "structureof identityand interest" and showthat no particularone followslogically from anarchy. A fundamentalprinciple of constructivistsocial theoryis that people act towardobjects, including other actors,on the basis of the meaningsthat the

18. The neorealist descriptionis not unproblematic.For a powerfulcritique, see David Lumsdaine,Ideals and Interests:The ForeignAid Regime,1949-1989 (Princeton,N.J.: Princeton UniversityPress, forthcoming). 19. Waltz,Theory of Intemational Politics, pp. 79-101. 20. StephenWalt, The OriginsofAlliances (Ithaca, N.Y.: CornellUniversity Press, 1987). Anarchy 397 objects have for them.21States act differentlytoward enemies than theydo towardfriends because enemies are threateningand friendsare not. Anarchy and the distributionof power are insufficientto tell us whichis which.U.S. militarypower has a differentsignificance for Canada than forCuba, despite their similar"structural" positions, just as Britishmissiles have a different significancefor the United States than do Soviet missiles.The distributionof powermay always affect states' calculations, but how it does so dependson the intersubjectiveunderstandings and expectations,on the "distributionof knowledge,"that constitutetheir conceptions of self and other.22If society "forgets"what a universityis, the powers and practices of professorand studentcease to exist;if the United States and Soviet Union decide thatthey are no longerenemies, "the cold war is over." It is collectivemeanings that constitutethe structureswhich organize our actions. Actorsacquire identities-relativelystable, role-specific understandings and expectationsabout self-by participatingin such collectivemeanings.23 Identi- ties are inherentlyrelational: "Identity, with its appropriateattachments of psychologicalreality, is alwaysidentity within a specific,socially constructed

21. See, forexample, Herbert Blumer, "The MethodologicalPosition of Symbolic Interactionism," in hisSymbolic Interactionism: Perspective and Method(Englewood Cliffs, N.J.: Prentice-Hall, 1969), p. 2. Throughoutthis article, I assumethat a theoreticallyproductive analogy can be made between individualsand states.There are at least two justificationsfor this anthropomorphism.Rhetori- cally,the analogyis an acceptedpractice in mainstreaminternational relations discourse, and since this articleis an immanentrather than externalcritique, it should followthe practice.Substan- tively,states are collectivitiesof individualsthat throughtheir practices constitute each otheras "persons" havinginterests, fears, and so on. A fulltheory of stateidentity- and interest-formation would neverthelessneed to drawinsights from the social psychologyof groupsand organizational theory,and forthat reason myanthropomorphism is merely suggestive. 22. The phrase "distributionof knowledge"is BarryBarnes's, as discussed in his work The Natureof Power (Cambridge: Polity Press, 1988); see also PeterBerger and Thomas Luckmann,The Social Constructionof Reality(New York: Anchor Books, 1966). The concernof recentinterna- tionalrelations scholarship on "epistemiccommunities" with the cause-and-effectunderstandings of the world held by scientists,experts, and policymakersis an importantaspect of the role of knowledgein worldpolitics; see Peter Haas, "Do Regimes Matter?Epistemic Communities and MediterraneanPollution Control," Intemational Organization 43 (Summer1989), pp. 377-404; and ErnstHaas, WhenKnowledge Is Power.My constructivistapproach would merelyadd to this an equal emphasison howsuch knowledgealso constitutesthe structuresand subjectsof social life. 23. For an excellentshort statement of how collectivemeanings constitute identities, see Peter Berger,"Identity as a Problemin the Sociologyof Knowledge,"European Joumal of Sociology, vol. 7, no. 1, 1966,pp. 32-40. See also David Morganand Michael Schwalbe,"Mind and Selfin Society: LinkingSocial Structureand Social Cognition,"Social PsychologyQuarterly 53 (June 1990), pp. 148-64.In mydiscussion, I drawon thefollowing interactionist texts: George HerbertMead, Mind, Self and Society(Chicago: Universityof Chicago Press, 1934); Bergerand Luckmann,The Social Constructionof Reality; Sheldon Stryker, Symbolic Interactionism:A Social StructuralVersion (Menlo Park, Calif.: Benjamin/Cummings,1980); R. S. Perinbanayagam,Signifying Acts: Structureand Meaningin EverydayLife (Carbondale: SouthernIllinois University Press, 1985); JohnHewitt, Self and Society:A SymbolicInteractionist Social Psychology(Boston: Allyn& Bacon, 1988); and Turner, A Theoryof Social Interaction.Despite some differences,much the same points are made by structurationistssuch as Bhaskar and Giddens. See Roy Bhaskar, The Possibilityof Naturalism (AtlanticHighlands, N.J.: HumanitiesPress, 1979); and AnthonyGiddens, CentralProblems in Social Theory(Berkeley: University of CaliforniaPress, 1979). 398 InternationalOrganization world," Peter Berger argues.24Each person has many identitieslinked to institutionalroles, such as brother,son, teacher,and citizen.Similarly, a state may have multiple identitiesas "sovereign," "leader of the free world," "imperial power," and so on.25The commitmentto and the salience of particularidentities vary, but each identityis an inherentlysocial definitionof the actor grounded in the theories which actors collectivelyhold about themselvesand one anotherand whichconstitute the structureof the social world. Identitiesare the basis of interests.Actors do not have a "portfolio"of intereststhat theycarry around independentof social context;instead, they definetheir interests in the process of definingsituations.26 As Nelson Foote puts it: "Motivation... refer[s]to the degree to which a human being,as a participantin the ongoingsocial processin whichhe necessarilyfinds himself, definesa problematicsituation as callingfor the performanceof a particular act, with more or less anticipatedconsummations and consequences, and therebyhis organismreleases the energyappropriate to performingit."27 Sometimessituations are unprecedentedin our experience,and in these cases we have to constructtheir meaning, and thusour interests,by analogy or invent them de novo. More often they have routine qualities in which we assign meanings on the basis of institutionallydefined roles. When we say that professorshave an "interest"in teaching,research, or goingon leave, we are sayingthat to functionin the role identityof "professor,"they have to define certainsituations as callingfor certain actions. This does not mean thatthey will necessarilydo so (expectationsand competence do not equal perfor- mance), but if theydo not,they will not get tenure.The absence or failureof roles makes definingsituations and interestsmore difficult,and identity

24. Berger,"Identity as a Problemin the Sociologyof Knowledge,"p. 111. 25. While not normallycast in such terms, foreignpolicy scholarship on national role conceptions could be adapted to such identitylanguage. See Kal Holsti, "National Role Conceptionsin the Studyof ForeignPolicy," Intemational Studies Quarterly 14 (September1970), pp. 233-309; and Stephen Walker,ed., Role Theoryand ForeignPolicy Analysis (Durham, N.C.: Duke UniversityPress, 1987). For an importanteffort to do so, see Stephen Walker,"Symbolic Interactionism and International Politics: Role Theory's Contribution to International Organization,"in C. Shih and Martha Cottam,eds., ContendingDramas: A CognitiveApproach to Post-WarIntemational Organizational Processes (New York: Praeger,forthcoming). 26. On the "portfolio"conception of interests,see BarryHindess, Political Choice and Social Structure(Aldershot, U.K.: Edward Elgar, 1989), pp. 2-3. The "definitionof the situation"is a centralconcept in interactionisttheory. 27. Nelson Foote, "Identificationas the Basis for a Theory of Motivation,"American SociologicalReview 16 (February1951), p. 15. Such stronglysociological conceptions of interest have been criticized,with some justice, for being "oversocialized"; see Dennis Wrong,"The OversocializedConception of Man in Modern Sociology,"American Sociological Review 26 (April 1961), pp. 183-93. For useful correctives,which focus on the activation of presocial but nondetermininghuman needs withinsocial contexts,see Turner,A Theoryof Social Interaction,pp. 23-69; and ViktorGecas, "The Self-Conceptas a Basis for a Theoryof Motivation,"in Judith Howard and Peter Callero, eds., The Self-SocietyDynamic (Cambridge: CambridgeUniversity Press,1991), pp. 171-87. Anarchy399

confusionmay result. This seems to be happeningtoday in theUnited States and the formerSoviet Union: without the cold war'smutual attributions of threatand hostilityto definetheir identities, these states seem unsure of what their"interests" should be. An institutionis a relativelystable set or "structure"of identitiesand interests.Such structuresare oftencodified in formalrules and norms,but thesehave motivationalforce only in virtueof actors'socialization to and participationin collectiveknowledge. Institutions are fundamentallycognitive entitiesthat do notexist apart from actors' ideas about how the world works.28 This does notmean that institutions are not real or objective,that they are "nothingbut" beliefs. As collectiveknowledge, they are experiencedas having an existence"over and abovethe individuals who happen to embodythem at themoment."29 In this way, institutions come to confront individuals as moreor lesscoercive social facts, but they are still a functionof what actors collectively "know."Identities and suchcollective cognitions do notexist apart from each other;they are "mutuallyconstitutive."30 On this view, institutionalization is a processof internalizing new identities and interests,not something occurring outsidethem and affectingonly behavior; socialization is a cognitiveprocess, notjust a behavioralone. Conceivedin thisway, institutions may be coopera- tiveor conflictual,a pointsometimes lost in scholarshipon international regimes,which tends to equate institutionswith cooperation. There are importantdifferences between conflictual and cooperativeinstitutions to be sure,but all relativelystable self-other relations-even those of "enemies"- aredefined intersubjectively. Self-helpis an institution,one ofvarious structures of identity and interest thatmay exist under anarchy. Processes of identity-formationunder anarchy are concernedfirst and foremostwith preservation or "security"of theself. Conceptsof security therefore differ in theextent to whichand themanner in whichthe self is identifiedcognitively with the other,31 and, I wantto suggest,it

28. In neo-Durkheimianparlance, institutions are "social representations."See Serge Moscov- ici, "The Phenomenonof Social Representations,"in Rob Farr and Serge Moscovici,eds., Social Representations(Cambridge: Cambridge University Press, 1984), pp. 3-69. See also Barnes, The Natureof Power.Note that thisis a considerablymore socialized cognitivismthan thatfound in muchof therecent scholarship on the role of "ideas" inworld politics, which tends to treatideas as commoditiesthat are held by individualsand intervenebetween the distributionof power and outcomes.For a formof cognitivism closer to myown, see Emanuel Adler,"Cognitive Evolution: A Dynamic Approach for the Studyof InternationalRelations and Their Progress,"in Emanuel Adler and BeverlyCrawford, eds., Progress in Postwar International Relations (New York: Columbia UniversityPress, 1991), pp. 43-88. 29. Bergerand Luckmann,The Social Constructionof Reality, p. 58. 30. See Giddens,Central Problems in Social Theory;and AlexanderWendt and RaymondDuvall, "Institutionsand InternationalOrder," in Ernst-OttoCzempiel and JamesRosenau, eds., Global Changesand TheoreticalChallenges (Lexington, Mass.: LexingtonBooks, 1989),pp. 51-74. 31. Proponentsof choice theorymight put this in termsof "interdependentutilities." For a usefuloverview of relevantchoice-theoretic discourse, most of whichhas focusedon the specific case of altruism,see Harold Hochman and Shmuel Nitzan,"Concepts of ExtendedPreference," 400 InternationalOrganization is upon thiscognitive variation that the meaning of anarchyand the distribution of power depends. Let me illustratewith a standardcontinuum of security systems." At one end is the "competitive"security system, in which states identify negativelywith each other'ssecurity so that ego's gain is seen as alter's loss. Negative identificationunder anarchyconstitutes systems of "realist" power politics: risk-averseactors that inferintentions from capabilities and worry about relativegains and losses. At the limit-in the Hobbesian war of all againstall-collective actionis nearlyimpossible in such a systembecause each actormust constantly fear being stabbed in the back. In the middle is the "individualistic"security system, in which states are indifferentto the relationshipbetween theirown and others' security.This constitutes"neoliberal" systems:states are still self-regardingabout their securitybut are concernedprimarily with absolute gains ratherthan relative gains. One's position in the distributionof power is less important,and collectiveaction is more possible (thoughstill subject to free ridingbecause statescontinue to be "egoists"). Competitiveand individualisticsystems are both "self-help" forms of anarchyin the sense that states do not positivelyidentify the securityof self withthat of othersbut insteadtreat security as the individualresponsibility of each. Giventhe lack of a positivecognitive identification on the basis ofwhich to build securityregimes, power politicswithin such systemswill necessarily consistof effortsto manipulateothers to satisfyself-regarding interests. This contrastswith the "cooperative" security system, in which states identify positivelywith one anotherso that the securityof each is perceived as the responsibilityof all. This is not self-helpin any interestingsense, since the "self" in terms of which interestsare defined is the community;national interestsare internationalinterests.33 In practice,of course, the extentto which

Joumalof Economic Behavior and Organization6 (June1985), pp. 161-76. The literatureon choice theoryusually does not link behaviorto issues of identity.For an exception,see AmartyaSen, "Goals, Commitment,and Identity,"Joumal of Law, Economics,and Organization1 (Fall 1985), pp. 341-55; and Robert Higgs, "Identityand Cooperation: A Commenton Sen's Alternative Program,"Joumal of Law, Economics,and Organization3 (Spring1987), pp. 140-42. 32. Securitysystems might also varyin the extentto whichthere is a functionaldifferentiation or a hierarchicalrelationship between patron and client,with the patronplaying a hegemonicrole withinits sphereof influencein definingthe securityinterests of its clients.I do not examinethis dimensionhere; forpreliminary discussion, see AlexanderWendt, "The States Systemand Global Militarization,"Ph.D. diss., Universityof Minnesota,Minneapolis, 1989; and AlexanderWendt and Michael Barnett,"The InternationalSystem and Third World Militarization,"unpublished manuscript,1991. 33. This amountsto an "internationalizationof the state." For a discussionof thissubject, see RaymondDuvall and AlexanderWendt, "The InternationalCapital Regime and the Internation- alization of the State," unpublishedmanuscript, 1987. See also R. B. J. Walker, "Sovereignty, Identity,Community: Reflections on the Horizonsof ContemporaryPolitical Practice," in R. B. J. Walkerand Saul Mendlovitz,eds., ContendingSovereignties (Boulder, Colo.: LynneRienner, 1990), pp. 159-85. Anarchy 401 states'identification with the community varies, from the limitedform found in "concerts"to the full-blownform seen in "collectivesecurity" arrangements.34 Dependingon howwell developedthe collectiveself is, itwill produce security practices that are in varyingdegrees altruisticor prosocial. This makes collectiveaction less dependent on the presence of active threatsand less prone to free riding.35Moreover, it restructuresefforts to advance one's objectives,or "power politics,"in termsof shared normsrather than relative power.36 On this view, the tendencyin internationalrelations scholarship to view power and institutionsas two opposing explanationsof foreignpolicy is thereforemisleading, since anarchyand the distributionof power onlyhave meaningfor state action in virtueof the understandingsand expectationsthat constituteinstitutional identities and interests.Self-help is one suchinstitution, constitutingone kind of anarchybut not the only kind. Waltz's three-part definitionof structuretherefore seems underspecified.In order to go from structureto action,we need to add a fourth:the intersubjectivelyconstituted structureof identitiesand interestsin the system. This has an importantimplication for the way in whichwe conceiveof states in the state of nature before theirfirst encounter with each other. Because states do not have conceptionsof self and other,and thus securityinterests, apart fromor prior to interaction,we assume too much about the state of nature if we concur with Waltz that, in virtue of anarchy,"international political systems,like economic markets,are formed by the coaction of self-regardingunits."37 We also assume too muchif we argue that,in virtueof

34. On the spectrumof cooperativesecurity arrangements, see Charles Kupchan and Clifford Kupchan, "Concerts, Collective Security,and the Future of Europe," InternationalSecurity 16 (Summer 1991), pp. 114-61; and Richard Smoke, "A Theory of Mutual Security,"in Richard Smoke and Andrei Kortunov,eds., Mutual Security(New York: St. Martin's Press, 1991), pp. 59-111. These maybe usefullyset alongsideChristopher Jencks' "Varieties of Altruism,"in Jane Mansbridge,ed., BeyondSelf-Interest (Chicago: Universityof ChicagoPress, 1990), pp. 53-67. 35. On the role of collectiveidentity in reducingcollective action problems, see Bruce Fireman and WilliamGamson, "Utilitarian Logic in the Resource MobilizationPerspective," in MayerZald and JohnMcCarthy, eds., TheDynamics of Social Movements(Cambridge, Mass.: Winthrop,1979), pp. 8-44; RobynDawes et al., "Cooperationfor the Benefit of Us-Not Me, or My Conscience,"in Mansbridge,Beyond Self-Interest, pp. 97-110; and Craig Calhoun, "The Problem of Identityin CollectiveAction," in Joan Huber, ed., Macro-MicroLinkages in Sociology(Beverly Hills, Calif.: Sage, 1991), pp. 51-75. 36. See Thomas Risse-Kappen,"Are DemocraticAlliances Special?" unpublishedmanuscript, Yale University,New Haven, Conn., 1991. This line of argumentcould be expanded usefullyin feministterms. For a usefuloverview of the relationalnature of feministconceptions of self,see Paula England and Barbara Stanek Kilbourne,"Feminist Critiques of the SeparativeModel of Self: Implicationsfor Rational Choice Theory,"Rationality and Society2 (April 1990), pp. 156-71. On feministconceptualizations of power, see Ann Tickner,"Hans Morgenthau'sPrinciples of Political Realism: A FeministReformulation," Millennium 17 (Winter 1988), pp. 429-40; and Thomas Wartenberg,"The Conceptof Powerin FeministTheory," Praxis International 8 (October 1988), pp. 301-16. 37. Waltz,Theory of International Politics, p. 91. 402 InternationalOrganization

anarchy,states in the stateof naturenecessarily face a "stag hunt"or "security dilemma."38These claims presuppose a historyof interactionin whichactors have acquired "selfish"identities and interests;before interaction (and stillin abstractionfrom first- and second-imagefactors) they would have no experience upon which to base such definitionsof self and other. To assume otherwiseis to attributeto statesin the state of naturequalities that they can onlypossess in society.39Self-help is an institution,not a constitutivefeature of anarchy. What,then, is a constitutivefeature of thestate of nature before interaction? Two things are left if we strip away those propertiesof the self which presuppose interactionwith others. The firstis the material substrateof agency,including its intrinsiccapabilities. For humanbeings, this is the body; for states, it is an organizationalapparatus of governance.In effect,I am suggestingfor rhetorical purposes that the rawmaterial out ofwhich members of the statesystem are constitutedis createdby domesticsociety before states enterthe constitutiveprocess of internationalsociety,40 although this process impliesneither stable territorialitynor sovereignty,which are internationally negotiatedterms of individuality(as discussedfurther below). The second is a desire to preservethis material substrate,to survive.This does not entail "self-regardingness,"however, since actors do not have a self prior to interactionwith an other;how theyview the meaningand requirementsof this survivaltherefore depends on the processes by which conceptionsof self evolve. This mayall seem veryarcane, but thereis an importantissue at stake: are theforeign policy identities and interestsof statesexogenous or endogenousto the statesystem? The formeris the answerof an individualisticor undersocial- ized systemictheory for which rationalismis appropriate;the latter is the answerof a fullysocialized systemictheory. Waltz seems to offerthe latterand

38. See Waltz,Man, theState, and War; and Robert Jervis,"Cooperation Under the Security Dilemma," WorldPolitics 30 (January1978), pp. 167-214. 39. My argumenthere parallels Rousseau's critiqueof Hobbes. For an excellentcritique of realistappropriations of Rousseau, see Michael Williams,"Rousseau, Realism,and Realpolitik," Millennium18 (Summer 1989), pp. 188-204. Williamsargues thatfar frombeing a fundamental startingpoint in the state of nature,for Rousseau the stag huntrepresented a stage in man's fall. On p. 190,Williams cites Rousseau's descriptionof man priorto leavingthe stateof nature:"Man onlyknows himself; he does not see his own well-beingto be identifiedwith or contraryto thatof anyoneelse; he neitherhates anythingnor loves anything;but limitedto no more than physical instinct,he is no one, he is an animal." For anothercritique of Hobbes on the stateof naturethat parallels myconstructivist reading of anarchy,see Charles Landesman,"Reflections on Hobbes: Anarchyand Human Nature,"in Peter Caws, ed., The Causes of Quarrel(Boston: Beacon, 1989), pp. 139-48. 40. Empirically,this suggestionis problematic,since the process of decolonizationand the subsequentsupport of many Third World states by international society point to waysin which even the raw material of "empirical statehood" is constitutedby the societyof states. See Robert Jacksonand Carl Rosberg,"Why Africa's Weak States Persist:The Empiricaland the Juridicalin Statehood,"World Politics 35 (October 1982), pp. 1-24. Anarchy 403 proposes two mechanisms,competition and socialization,by whichstructure conditionsstate action.4' The contentof his argumentabout thisconditioning, however,presupposes a self-helpsystem that is not itselfa constitutivefeature of anarchy.As JamesMorrow points out, Waltz's two mechanismscondition behavior,not identityand interest.42This explainshow Waltz can be accused of both "individualism"and "structuralism."4'He is the formerwith respect to systemicconstitutions of identityand interest,the latter with respect to systemicdeterminations of behavior.

Anarchy and the social constructionof power politics If self-helpis not a constitutivefeature of anarchy,it mustemerge causally fromprocesses in whichanarchy plays only a permissiverole.' This reflectsa second principleof constructivism:that the meaningsin termsof whichaction is organizedarise out of interaction.45This being said, however,the situation facingstates as theyencounter one anotherfor the firsttime may be such that only self-regardingconceptions of identitycan survive;if so, even if these conceptions are socially constructed,neorealists may be rightin holding identitiesand interestsconstant and thusin privilegingone particularmeaning of anarchicstructure over process. In thiscase, rationalistswould be rightto argue for a weak, behavioral conceptionof the differencethat institutions make, and realistswould be rightto argue that any internationalinstitutions which are created will be inherentlyunstable, since withoutthe power to

41. Waltz,Theory of International Politics, pp. 74-77. 42. See JamesMorrow, "Social Choice and SystemStructure in World Politics,"World Politics 41 (October 1988), p. 89. Waltz's behavioraltreatment of socializationmay be usefullycontrasted withthe more cognitive approach taken by Ikenberry and theKupchans in thefollowing articles: G. John Ikenberryand Charles Kupchan, "Socialization and Hegemonic Power," International Organization44 (Summer 1989), pp. 283-316; and Kupchan and Kupchan,"Concerts, Collective Security,and the Future of Europe." Their approach is close to my own, but they define socializationas an elite strategyto induce value change in others,rather than as a ubiquitous featureof interactionin termsof which all identitiesand interestsget produced and reproduced. 43. Regardingindividualism, see Richard Ashley,"The Povertyof Neorealism,"International Organization38 (Spring1984), pp. 225-86; Wendt,"The Agent-StructureProblem in International Relations Theory";and David Dessler, "What's at Stake in the Agent-StructureDebate?" Inter- nationalOrganization 43 (Summer1989), pp. 441-74. Regardingstructuralism, see R. B. J.Walker, "Realism, Change, and InternationalPolitical Theory," International Studies Quarterly 31 (March 1987), pp. 65-86; and MartinHollis and StevenSmith, Explaining and UnderstandingInternational Relations(Oxford: Clarendon Press, 1989). The behavioralismevident in neorealisttheory also explainshow neorealistscan reconciletheir structuralism with the individualismof rationalchoice theory.On the behavioral-structuralcharacter of the latter,see Spiro Latsis, "SituationalDeter- minismin Economics,"British Journalfor the Philosophy of Science 23 (August1972), pp. 207-45. 44. The importanceof the distinctionbetween constitutiveand causal explanationsis not sufficientlyappreciated in constructivistdiscourse. See Wendt,"The Agent-StructureProblem in InternationalRelations Theory," pp. 362-65;Wendt, "The StatesSystem and Global Militarization," pp. 110-13; and Wendt, "Bridgingthe Theory/Meta-TheoryGap in InternationalRelations," Reviewof International Studies 17 (October 1991), p. 390. 45. See Blumer,"The MethodologicalPosition of SymbolicInteractionism," pp. 2-4. 404 InternationalOrganization

transformidentities and intereststhey will be "continuingobjects of choice" by exogenouslyconstituted actors constrainedonly by the transactioncosts of behavioralchange.46 Even in a permissivecausal role, in otherwords, anarchy may decisivelyrestrict interaction and thereforerestrict viable forms of systemictheory. I address these causal issues firstby showinghow self- regardingideas about securitymight develop and then by examiningthe conditionsunder which a keyefficient cause-predation-may disposestates in thisdirection rather than others. Conceptionsof selfand interesttend to "mirror"the practicesof significant others over time. This principle of identity-formationis captured by the symbolicinteractionist notion of the "looking-glassself," which asserts that the selfis a reflectionof an actor'ssocialization. Consider two actors-ego and alter-encounteringeach otherfor the first time.47Each wantsto surviveand has certainmaterial capabilities, but neither actorhas biologicalor domesticimperatives for power, glory, or conquest(still bracketed),and thereis no historyof securityor insecuritybetween the two. What should theydo? Realists would probablyargue thateach should act on thebasis ofworst-case assumptions about the other'sintentions, justifying such an attitudeas prudentin view of the possibility of death from making a mistake. Such a possibilityalways exists, even in civilsociety; however, society would be impossible if people made decisions purely on the basis of worst-case possibilities.Instead, most decisionsare and should be made on the basis of probabilities,and these are producedby interaction, by what actors do. In the beginningis ego's gesture,which may consist,for example, of an advance,a retreat,a brandishingof arms,a layingdown of arms,or an attack.48 For ego, thisgesture represents the basis on whichit is preparedto respondto alter.This basis is unknownto alter,however, and so it mustmake an inference or "attribution"about ego's intentionsand, in particular,given that this is anarchy,about whetherego is a threat.49The contentof this inferencewill largelydepend on two considerations.The firstis the gesture's and ego's

46. See RobertGrafstein, "Rational Choice: Theoryand Institutions,"in KristenMonroe, ed., TheEconomicApproach to Politics (New York: Harper Collins,1991), pp. 263-64. A good example of the promise and limitsof transactioncost approaches to institutionalanalysis is offeredby RobertKeohane in hisAfterHegemony (Princeton, N.J.: Princeton University Press, 1984). 47. This situationis not entirelymetaphorical in worldpolitics, since throughouthistory states have "discovered" each other,generating an instantanarchy as it were. A systematicempirical studyof firstcontacts would be interesting. 48. Mead's analysisof gesturesremains definitive. See Mead's Mind,Self and Society.See also the discussionof the role of signalingin the "mechanicsof interaction"in Turner'sA Theoryof Social Interaction,pp. 74-79 and 92-115. 49. On the role of attributionprocesses in the interactionistaccount of identity-formation,see Sheldon Strykerand Avi Gottlieb,"Attribution Theory and SymbolicInteractionism," in John Harvey et al., eds., New Directionsin AttributionResearch, vol. 3 (Hillsdale, N.J.: Lawrence Erlbaum, 1981), pp. 425-58; and Kathleen Crittenden,"Sociological Aspects of Attribution," Annual Reviewof Sociology,vol. 9, 1983, pp. 425-46. On attributionalprocesses in international relations,see Shawn Rosenbergand GaryWolfsfeld, "International Conflict and the Problemof Attribution,"Journal of ConflictResolution 21 (March 1977),pp. 75-103. Anarchy 405

physicalqualities, which are in part contrivedby ego and whichinclude the directionof movement,noise, numbers,and immediateconsequences of the gesture.50The second considerationconcerns what alterwould intendby such qualitieswere it to make such a gestureitself. Alter may make an attributional "error"in its inferenceabout ego's intent,but thereis also no reason forit to assume a priori-before the gesture-that ego is threatening,since it is only througha processof signalingand interpretingthat the costs and probabilities ofbeing wrong can be determined.5"Social threatsare constructed,not natural. Consideran example.Would we assume,a priori,that we were about to be attackedif we are ever contactedby membersof an alien civilization?I think not. We would be highlyalert, of course,but whetherwe placed our military forceson alertor launchedan attackwould depend on howwe interpretedthe importof their firstgesture for our security-if only to avoid making an immediateenemy out ofwhat may be a dangerousadversary. The possibilityof error,in otherwords, does not forceus to act on the assumptionthat the aliens are threatening:action depends on theprobabilities we assign,and theseare in key part a functionof what the aliens do; priorto theirgesture, we have no systemicbasis forassigning probabilities. If theirfirst gesture is to appearwith a thousand spaceships and destroyNew York, we will definethe situationas threateningand respond accordingly.But if theyappear withone spaceship, sayingwhat seems to be "we come in peace," we will feel "reassured"and will probablyrespond with a gestureintended to reassurethem, even ifthis gesture is not necessarilyinterpreted by them as such.52 This process of signaling,interpreting, and respondingcompletes a "social act" and beginsthe process of creatingintersubjective meanings. It advances thesame way.The firstsocial act createsexpectations on bothsides about each other's futurebehavior: potentiallymistaken and certainlytentative, but expectationsnonetheless. Based on thistentative knowledge, ego makes a new gesture,again signifyingthe basis on whichit will respondto alter,and again alterresponds, adding to the pool of knowledgeeach has about the other,and so on over time. The mechanismhere is reinforcement;interaction rewards actorsfor holding certain ideas about each otherand discouragesthem from holdingothers. If repeated long enough,these "reciprocaltypifications" will createrelatively stable concepts of selfand otherregarding the issue at stakein the interaction.53

50. On the "stagecraft"involved in "presentationsof self,"see ErvingGoffman, The Presenta- tionof Self in EverydayLife (New York: Doubleday,1959). On the role of appearance in definitions of the situation,see GregoryStone, "Appearance and the Self," in Arnold Rose, ed., Human Behaviorand Social Processes(Boston: HoughtonMifflin, 1962), pp. 86-118. 51. This discussionof the role of possibilitiesand probabilitiesin threatperception owes much to StewartJohnson's comments on an earlierdraft of myarticle. 52. On the role of "reassurance"in threatsituations, see RichardNed Lebow and JaniceGross Stein,"Beyond Deterrence,"JournalofSocialIssues, vol. 43, no. 4, 1987,pp. 5-72. 53. On "reciprocaltypifications," see Bergerand Luckmann,The Social Constructionof Reality, pp. 54-58. 406 InternationalOrganization

INSTITUTIONS PROCESS

State A with identities .(1) Stimulus requiring and interests \action

(2) StatesA's definitionof | l the situationl

Intersubjective understandingsand (3) StateA's action expectationspossessed by andconstitutive of A and B

\ ~~~(4)State B's interpretation \ ~~~ofA's action and it B's owndefinition of { \/ the situation

StateB withidentities \l andinterests (5) StateB's action

FIGURE 1. Thecodetermination of institutionsand process

It is throughreciprocal interaction,in other words, that we create and instantiatethe relativelyenduring social structuresin termsof which we define our identitiesand interests.Jeff Coulter sums up theontological dependence of structureon process thisway: "The parametersof social organizationthem- selves are reproducedonly in and throughthe orientationsand practicesof membersengaged in social interactionsover time.... Social configurationsare not 'objective'like mountainsor forests,but neitherare they'subjective' like dreams or flightsof speculative fancy.They are, as most social scientists concede at the theoreticallevel, intersubjective constructions."54 The simple overall model of identity-and interest-formationproposed in Figure 1 applies to competitiveinstitutions no less than to cooperativeones. Self-helpsecurity systems evolve from cycles of interactionin whicheach party acts in ways that the other feels are threateningto the self, creating expectationsthat the other is not to be trusted. Competitiveor egoistic identitiesare caused by such insecurity;if the otheris threatening,the self is forcedto "mirror"such behaviorin its conceptionof the selfs relationshipto

54. JeffCoulter, "Remarks on the Conceptualizationof Social Structure,"Philosophy of the Social Sciences12 (March 1982), pp. 42-43. Anarchy 407

thatother.55 Being treatedas an object forthe gratificationof othersprecludes the positiveidentification with othersnecessary for collectivesecurity; con- versely,being treatedby othersin waysthat are empathicwith respect to the securityof the selfpermits such identification.56 Competitivesystems of interactionare prone to security"dilemmas," in whichthe effortsof actorsto enhance theirsecurity unilaterally threatens the securityof the others,perpetuating distrust and alienation. The formsof identityand interestthat constitutesuch dilemmas,however, are themselves ongoingeffects of, not exogenous to, the interaction;identities are producedin and through"situated activity."57We do not beginour relationshipwith the aliens in a securitydilemma; securitydilemmas are not givenby anarchyor nature. Of course, once institutionalizedsuch a dilemma may be hard to change (I returnto thisbelow), but the pointremains: identities and interests are constitutedby collectivemeanings that are alwaysin process.As Sheldon Strykeremphasizes, "The social processis one ofconstructing and reconstruct- ing self and social relationships."58If states find themselvesin a self-help system,this is because theirpractices made it thatway. Changing the practices willchange the intersubjectiveknowledge that constitutes the system.

Predator states and anarchy as permissive cause The mirrortheory of identity-formationis a crude account of how the process of creatingidentities and interestsmight work, but it does not tell us whya systemof states-such as, arguably,our own-would have ended up with self-regardingand not collective identities.In this section, I examine an efficientcause, predation,which, in conjunctionwith anarchy as a permissive cause, maygenerate a self-helpsystem. In so doing,however, I show the key role thatthe structureof identitiesand interestsplays in mediatinganarchy's explanatoryrole. The predatorargument is straightforwardand compelling.For whatever reasons-biology, domestic politics,or systemicvictimization-some states

55. The followingarticles by Noel Kaplowitz have made an importantcontribution to such thinkingin internationalrelations: "Psychopolitical Dimensions of InternationalRelations: The ReciprocalEffects of ConflictStrategies," International Studies Quarterly 28 (December 1984), pp. 373-406; and "National Self-Images,Perception of Enemies,and ConflictStrategies: Psychopoliti- cal Dimensionsof InternationalRelations," Political Psychology 11 (March 1990), pp. 39-82. 56. These argumentsare common in theoriesof narcissismand altruism.See Heinz Kohut, Self-Psychologyand theHumanities (New York: Norton,1985); and MartinHoffmann, "Empathy, Its Limitations,and Its Role in a ComprehensiveMoral Theory,"in WilliamKurtines and Jacob Gewirtz,eds., Morality,Moral Behavior,and Moral Development(New York: Wiley, 1984), pp. 283-302. 57. See C. Norman Alexander and Mary Glenn Wiley, "Situated Activityand Identity Formation," in Morris Rosenberg and Ralph Turner, eds., Social Psychology:Sociological Perspectives(New York: Basic Books, 1981), pp. 269-89. 58. Sheldon Stryker,"The Vitalizationof SymbolicInteractionism," Social PsychologyQuarterly 50 (March 1987), p. 93. 408 InternationalOrganization maybecome predisposedtoward aggression. The aggressivebehavior of these predatorsor "bad apples" forcesother states to engage in competitivepower politics,to meet firewith fire, since failureto do so may degrade or destroy them.One predatorwill best a hundredpacifists because anarchyprovides no guarantees.This argumentis powerfulin partbecause itis so weak: ratherthan makingthe strongassumption that all states are inherentlypower-seeking (a purely reductionisttheory of power politics), it assumes that just one is power-seekingand thatthe othershave to followsuit because anarchypermits theone to exploitthem. In makingthis argument,it is importantto reiteratethat the possibilityof predationdoes notin itselfforce states to anticipateit a prioriwith competitive power politicsof theirown. The possibilityof predationdoes not mean that "war mayat any momentoccur"; it mayin factbe extremelyunlikely. Once a predatoremerges, however, it maycondition identity- and interest-formation in the followingmanner. In an anarchyof two,if ego is predatory,alter must either define its security in self-helpterms or pay the price. This follows directlyfrom the above argument,in whichconceptions of self mirrortreatment by the other.In an anarchyof many,however, the effectof predationalso depends on the level of collectiveidentity already attained in the system.If predationoccurs right after thefirst encounter in the stateof nature, it will force others with whom it comes in contactto defendthemselves, first individually and thencollectively if they come to perceivea commonthreat. The emergenceof such a defensivealliance willbe seriouslyinhibited if the structureof identitiesand interestshas already evolvedinto a Hobbesian worldof maximuminsecurity, since potentialallies will stronglydistrust each other and face intensecollective action problems; suchinsecure allies are also morelikely to fallout amongstthemselves once the predator is removed. If collective securityidentity is high, however,the emergenceof a predatormay do muchless damage. If the predatorattacks any memberof the collective,the latterwill come to the victim'sdefense on the principleof "all forone, one forall," even if the predatoris not presentlya threatto othermembers of the collective.If the predatoris not strongenough to withstandthe collective, it will be defeatedand collectivesecurity will obtain. But if it is strongenough, the logic of the two-actorcase (now predatorand collective)will activate, and balance-of-powerpolitics will reestablish itself. The timingof the emergence of predation relative to the historyof identity-formationin the communityis thereforecrucial to anarchy'sexplana- toryrole as a permissivecause. Predationwill alwayslead victimsto defend themselves,but whether defense will be collectiveor notdepends on thehistory of interactionwithin the potential collective as muchas on the ambitionsof the predator.Will the disappearance of the Soviet threatrenew old insecurities amongthe membersof the NorthAtlantic Treaty Organization? Perhaps, but notif they have reasonsindependent of thatthreat for identifying their security withone another.Identities and interestsare relationship-specific,not intrinsic Anarchy 409 attributesof a "portfolio";states may be competitivein some relationshipsand solidaryin others."Mature" anarchiesare less likelythan "immature" ones to be reduced by predationto a Hobbesian condition,and maturity,which is a proxyfor structures of identityand interest,is a functionof process.59 The source of predationalso matters.If it stemsfrom unit-level causes that are immuneto systemicimpacts (causes such as human nature or domestic politics taken in isolation), then it functionsin a manner analogous to a "genetictrait" in the constructedworld of the statesystem. Even ifsuccessful, thistrait does notselect for other predators in an evolutionarysense so muchas it teachesother states to respondin kind,but since traitscannot be unlearned, the otherstates will continue competitive behavior until the predatoris either destroyedor transformedfrom within. However, in the more likelyevent that predationstems at least in part fromprior systemic interaction-perhaps as a resultof beingvictimized in the past (one thinkshere of Nazi Germanyor the Soviet Union)-then it is more a response to a learned identityand, as such, mightbe transformedby future social interactionin the formof appeasement, reassurancesthat securityneeds will be met, systemiceffects on domestic politics,and so on. In thiscase, in otherwords, there is morehope thatprocess can transforma bad apple intoa good one. The role of predationin generatinga self-helpsystem, then, is consistent witha systematicfocus on process. Even if the source of predationis entirely exogenousto the system,it is whatstates do thatdetermines the quality of their interactionsunder anarchy. In thisrespect, it is notsurprising that it is classical realistsrather than structural realists who emphasizethis sort of argument.The former'semphasis on unit-levelcauses of powerpolitics leads more easilyto a permissiveview of anarchy'sexplanatory role (and thereforeto a processual viewof internationalrelations) than does the latter'semphasis on anarchyas a "structuralcause";60 neorealistsdo not need predationbecause the systemis givenas self-help. This raises anew the question of exactlyhow much and what kind of role human nature and domesticpolitics play in world politics.The greaterand more destructivethis role, the more significantpredation will be, and the less amenableanarchy will be to formationof collectiveidentities. Classical realists, of course, assumed that human naturewas possessed by an inherentlust for poweror glory.My argumentsuggests that assumptions such as thiswere made for a reason: an unchangingHobbesian man providesthe powerfulefficient cause necessaryfor a relentlesspessimism about worldpolitics that anarchic structurealone, or even structureplus intermittentpredation, cannot supply. One can be skepticalof such an essentialistassumption, as I am, but it does

59. On the "maturity"of anarchies,see BarryBuzan, People, States,and Fear (Chapel Hill: Universityof NorthCarolina Press,1983). 60. A similarintuition may lie behindAshley's effort to reappropriateclassical realist discourse for criticalinternational relations theory. See Richard Ashley,"Political Realism and Human Interests,"International Studies Quarterly 38 (June1981), pp. 204-36. 410 InternationalOrganization producedeterminate results at the expenseof systemictheory. A concernwith systemicprocess over structuresuggests that perhaps it is time to revisitthe debate overthe relativeimportance of first-,second-, and third-imagetheories of stateidentity-formation.6" Assumingfor now that systemictheories of identity-formationin world politics are worthpursuing, let me conclude by suggestingthat the realist- rationalistalliance "reifies"self-help in the sense of treatingit as something separatefrom the practices by which it is producedand sustained.Peter Berger and Thomas Luckmanndefine reification as follows:"[It] is the apprehension of the productsof human activityas if theywere somethingelse than human products-such as factsof nature,results of cosmiclaws, or manifestationsof divine will. Reificationimplies that man is capable of forgettinghis own authorshipof the humanworld, and further,that the dialecticbetween man, the producer,and his productsis lost to consciousness.The reifiedworld is ... experiencedby man as a strangefacticity, an opusalienum over which he has no controlrather than as the opusproprium of his own productiveactivity."62 By denyingor bracketingstates' collective authorshipof their identitiesand interests,in otherwords, the realist-rationalistalliance denies or bracketsthe factthat competitivepower politicshelp create the very"problem of order" theyare supposed to solve-that realismis a self-fulfillingprophecy. Far from beingexogenously given, the intersubjective knowledge that constitutes compet- itiveidentities and interestsis constructedevery day by processes of "social will formation."63It is whatstates have made of themselves.

Institutionaltransformations of power politics

Let us assumethat processes of identity-and interest-formationhave createda worldin whichstates do not recognizerights to territoryor existence-a war of all againstall. In thisworld, anarchy has a "realist"meaning for state action: be insecureand concernedwith relative power. Anarchy has thismeaning only in virtueof collective,insecurity-producing practices, but if those practicesare

61. Waltz has himselfhelped open up such a debate withhis recognitionthat systemic factors conditionbut do not determinestate actions. See Kenneth Waltz, "Reflectionson Theoryof InternationalPolitics: A Response to My Critics,"in RobertKeohane, ed., Neorealismand Its Critics (New York: Columbia UniversityPress, 1986), pp. 322-45. The growingliterature on the observationthat "democracies do notfight each other"is relevantto thisquestion, as are twoother studies that break importantground toward a "reductionist"theory of state identity:William Bloom's PersonalIdentity, National Identityand IntemationalRelations (Cambridge: Cambridge UniversityPress, 1990) and Lumsdaine'sIdeals and Interests. 62. See Berger and Luckmann,The Social Constructionof Reality,p. 89. See also Douglas Maynardand Thomas Wilson,"On the Reificationof Social Structure,"in ScottMcNall and Gary Howe, eds., CurrentPerspectives in Social Theory,vol. 1 (Greenwich,Conn.: JAI Press, 1980), pp. 287-322. 63. See Richard Ashley, "Social Will and InternationalAnarchy," in Hayward Alker and Richard Ashley,eds., AfterRealism, work in progress,Massachusetts Institute of Technology, Cambridge,and Arizona State University,Tempe, 1992. Anarchy 411 relativelystable, they do constitutea systemthat may resistchange. The fact thatworlds of powerpolitics are sociallyconstructed, in otherwords, does not guaranteethey are malleable,for at least tworeasons. The firstreason is thatonce constituted,any social systemconfronts each of its membersas an objectivesocial fact that reinforcescertain behaviors and discouragesothers. Self-help systems, for example, tend to rewardcompetition and punish altruism.The possibilityof change depends on whether the exigenciesof such competitionleave room for actions that deviate fromthe prescribedscript. If theydo not, the systemwill be reproducedand deviant actorswill not.64 The second reason is that systemicchange mayalso be inhibitedby actors' interestsin maintainingrelatively stable role identities.Such interestsare rootednot only in the desireto minimizeuncertainty and anxiety,manifested in effortsto confirmexisting beliefs about the social world,but also in the desire to avoid the expectedcosts of breakingcommitments made to others-notably domesticconstituencies and foreignallies in the case of states-as partof past practices.The level of resistancethat these commitmentsinduce will depend on the "salience" of particularrole identitiesto the actor.65The United States, for example, is more likely to resist threats to its identityas "leader of anticommunistcrusades" than to its identityas "promoterof humanrights." But foralmost any role identity,practices and informationthat challenge it are likelyto create cognitivedissonance and even perceptionsof threat,and these maycause resistanceto transformationsof the selfand thusto social change.66 For bothsystemic and "psychological"reasons, then, intersubjective understandingsand expectationsmay have a self-perpetuatingquality, constituting path-dependenciesthat new ideas about self and othermust transcend. This does not change the fact that throughpractice agents are continuously producingand reproducingidentities and interests,continuously "choosing nowthe preferences [they] will have later."67But it does mean thatchoices may not be experiencedwith meaningfuldegrees of freedom.This could be a constructivistjustification for the realistposition that onlysimple learningis

64. See Ralph Turner,"Role-Taking: Process Versus Conformity,"in Rose, Human Behavior and Social Processes,pp. 20-40; and JudithHoward, "From ChangingSelves Toward Changing Society,"in Howard and Callero, The Self-SocietyDynamic, pp. 209-37. 65. On the relationshipbetween commitmentand identity,see Foote, "Identificationas the Basis for a Theoryof Motivation";Howard Becker, "Notes on the Concept of Commitment," AmericanJoumal of Sociology66 (July1960), pp. 32-40; and Stryker,Symbolic Interactionism. On role salience,see Stryker,ibid. 66. On threatsto identityand thetypes of resistancethat they may create, see GlynisBreakwell, Copingwith Threatened Identities (London: Methuen,1986); and TerrellNorthrup, "The Dynamic of Identityin Personaland Social Conflict,"in Louis Kreisberget al., eds.,Intractable Conflicts and TheirTransfornation (Syracuse, N.Y.: SyracuseUniversity Press, 1989), pp. 55-82. For a broad overviewof resistanceto change,see TimurKuran, "The Tenacious Past: Theoriesof Personaland CollectiveConservatism," Journal of Economic Behavior and Organization10 (September1988), pp. 143-71. 67. James March, "Bounded Rationality,Ambiguity, and the Engineeringof Choice," Bell Journalof Economics 9 (Autumn1978), p. 600. 412 InternationalOrganization possiblein self-helpsystems. The realistmight concede that such systems are sociallyconstructed and stillargue that after the corresponding identities and interestshave become institutionalized, they are almostimpossible to transform. In theremainder of this article, I examine three institutional transformations ofidentity and security interest through which states might escape a Hobbesian worldof their own making. In so doing,I seekto clarifywhat it meansto say that"institutions transform identities and interests," emphasizing that the key tosuch transformations is relatively stable practice.

Sovereignty,recognition, and security In a Hobbesianstate of nature,states are individuatedby the domestic processesthat constitute them as statesand by their material capacity to deter threatsfrom other states. In thisworld, even if freemomentarily from the predationsof others,state security does nothave any basis in socialrecogni- tion-in intersubjectiveunderstandings or norms that a statehas a rightto its existence,territory, and subjects.Security is a matterof nationalpower, nothingmore. The principleof sovereigntytransforms this situation by providing a social basisfor the individuality and securityof states.Sovereignty is an institution, and so it existsonly in virtueof certainintersubjective understandings and expectations;there is no sovereigntywithout an other.These understandings andexpectations not only constitute a particular kind of state-the "sovereign" state-butalso constitutea particular form of community, since identities are relational.The essenceof thiscommunity is a mutualrecognition of one another'sright to exerciseexclusive political authority within territorial limits. Thesereciprocal "permissions"68 constitute a spatially rather than functionally differentiatedworld-a worldin whichfields of practiceconstitute and are organizedaround "domestic" and "international"spaces rather than around the performanceof particularactivities.69 The locationof the boundaries betweenthese spaces is ofcourse sometimes contested, war being one practice throughwhich states negotiate the terms of theirindividuality. But thisdoes notchange the fact that it is only in virtue of mutual recognition that states have

68. Haskell Fain, NormativePolitics and the Communityof Nations (Philadelphia: Temple UniversityPress, 1987). 69. This is the intersubjectivebasis for the principleof functionalnondifferentiation among states,which "drops out" of Waltz's definitionof structurebecause the latter has no explicit intersubjectivebasis. In internationalrelations scholarship, the social productionof territorial space has been emphasizedprimarily by poststructuralists. See, forexample, Richard Ashley, "The Geopolitics of Geopolitical Space: Toward a Critical Social Theory of InternationalPolitics," Alternatives12 (October 1987), pp. 403-34; and Simon Dalby, Creatingthe Second Cold War (London: Pinter,1990). But the idea of space as both productand constituentof practiceis also prominentin structurationistdiscourse. See Giddens,Central Problems in Social Theory;and Derek Gregoryand JohnUrry, eds., Social Relationsand SpatialStructures (London: Macmillan,1985). Anarchy 413

"territorialproperty rights."70 This recognitionfunctions as a formof "social closure" thatdisempowers nonstate actors and empowersand helps stabilize interactionamong states.71 Sovereigntynorms are now so takenfor granted, so natural,that it is easy to overlookthe extentto whichthey are bothpresupposed by and an ongoingarti- factof practice.When statestax "their" "citizens" and not others,when they "protect"their markets against foreign "imports," when theykill thousands of Iraqis in one kindof war and thenrefuse to "intervene"to killeven one person in anotherkind, a "civil"war, and whenthey fight a globalwar againsta regime that soughtto destroythe institutionof sovereigntyand then give Germany back to the Germans,they are actingagainst the backgroundof, and thereby reproducing,shared norms about what it means to be a sovereignstate. If statesstopped acting on thosenorms, their identity as "sovereigns"(if not necessarilyas "states") would disappear. The sovereignstate is an ongoing accomplishmentof practice,not a once-and-for-allcreation of norms that somehow exist apart frompractice.72 Thus, sayingthat "the institutionof sovereigntytransforms identities" is shorthandfor saying that "regularprac- tices produce mutuallyconstituting sovereign identities (agents) and their associatedinstitutional norms (structures)." Practice is the core ofconstructiv- istresolutions of the agent-structureproblem. This ongoingprocess may not be politicallyproblematic in particularhistorical contexts and, indeed, once a communityof mutual recognitionis constituted,its members-even the disadvantagedones73 mayhave a vestedinterest in reproducingit. In fact,this is part of what having an identitymeans. But this identityand institution remain dependenton what actors do: removingthose practiceswill remove theirintersubjective conditions of existence.

70. See JohnRuggie, "Continuity and Transformationin theWorld Polity: Toward a Neorealist Synthesis,"World Politics 35 (January1983), pp. 261-85. In Mind,Self, and Society,p. 161, Mead offersthe following argument: "If we say'this is myproperty, I shall control it,' that affirmation calls out a certainset of responseswhich must be the same in anycommunity in whichproperty exists. It involvesan organizedattitude with reference to propertywhich is commonto all membersof the community.One musthave a definiteattitude of controlof his own propertyand respectfor the propertyof others. Those attitudes(as organizedsets of responses)must be thereon thepart of all, so thatwhen one says such a thinghe calls out in himselfthe responseof the others.That which makessociety possible is such commonresponses." 71. For a definitionand discussionof "social closure," see RaymondMurphy, Social Closure (Oxford:Clarendon Press, 1988). 72. See Richard Ashley,"Untying the SovereignState: A Double Reading of the Anarchy Problematique,"Millennium 17 (Summer1988), pp. 227-62. Those withmore modernist sensibili- ties will findan equally practice-centricview of institutionsin Blumer'sobservation on p. 19 of "The Methodological Position of SymbolicInteractionism": "A gratuitousacceptance of the conceptsof norms,values, social rules and the like should not blindthe social scientistto the fact thatany one of themis subtendedby a processof social interaction-a processthat is necessarynot onlyfor their change but equallywell fortheir retention in a fixedform. It is the social processin grouplife that creates and upholdsthe rules,not the rulesthat create and upholdgroup life." 73. See, for example,Mohammed Ayoob, "The Third World in the Systemof States: Acute Schizophreniaor GrowingPains?" InternationalStudies Quarterly 33 (March 1989),pp. 67-80. 414 InternationalOrganization

This may tell us somethingabout how institutionsof sovereignstates are reproduced throughsocial interaction,but it does not tell us why such a structureof identityand interestwould arise in the firstplace. Two conditions would seem necessaryfor this to happen: (1) the densityand regularityof interactionsmust be sufficientlyhigh and (2) actorsmust be dissatisfiedwith preexistingforms of identityand interaction.Given these conditions, a normof mutual recognitionis relativelyundemanding in termsof social trust,having the form of an assurance game in which a player will acknowledge the sovereigntyof the othersas long as theywill in turnacknowledge that player's own sovereignty.Articulating international legal principlessuch as those embodiedin the Peace ofAugsburg (1555) and the Peace ofWestphalia (1648) mayalso help by establishingexplicit criteria for determining violations of the nascent social consensus.74But whethersuch a consensus holds depends on whatstates do. If theytreat each otheras ifthey were sovereign,then over time theywill institutionalizethat mode of subjectivity;if theydo not, then that mode willnot become the norm. Practicesof sovereigntywill transform understandings of securityand power politicsin at least threeways. First, states will come to definetheir (and our) securityin termsof preservingtheir "property rights" over particularterrito- ries.We now see thisas natural,but the preservationof territorialfrontiers is not,in fact,equivalent to the survivalof the state or its people. Indeed, some states would probably be more secure if they would relinquish certain territories-the"Soviet Union" of some minorityrepublics, "Yugoslavia" of Croatia and Slovenia, Israel of the West Bank, and so on. The fact that sovereigntypractices have historicallybeen orientedtoward producing distinct territorialspaces, in otherwords, affects states' conceptualization of what they must"secure" to functionin thatidentity, a processthat may help accountfor the "hardening"of territorialboundaries over the centuries.75 Second, to the extentthat states successfully internalize sovereignty norms, theywill be more respectfultoward the territorialrights of others.76This restraintis not primarilybecause of the costs of violatingsovereignty norms, althoughwhen violatorsdo get punished (as in the Gulf War) it reminds everyoneof whatthese costscan be, but because partof whatit means to be a

74. See WilliamCoplin, "International Law and AssumptionsAbout the State System,"World Politics17 (July1965), pp. 615-34. 75. See AnthonySmith, "States and Homelands: The Social and Geopolitical Implicationsof NationalTerritory," Millennium 10 (Autumn1981), pp. 187-202. 76. This assumesthat there are no other,competing, principles that organize political space and identityin the internationalsystem and coexistwith traditional notions of sovereignty;in fact,of course, there are. On "spheres of influence" and "informalempires," see Jan Triska, ed., DominantPowers and SubordinateStates (Durham, N.C.: Duke UniversityPress, 1986); and Ronald Robinson,"The ExcentricIdea of Imperialism,With or WithoutEmpire," in WolfgangMommsen and JurgenOsterhammel, eds., Imperialismand After:Continuities and Discontinuities(London: Allen & Unwin, 1986), pp. 267-89. On Arab conceptionsof sovereignty,see Michael Barnett, "Sovereignty,Institutions, and Identity:From Pan-Arabismto the Arab State System,"unpublishedmanuscript, University of Wisconsin,Madison, 1991. Anarchy 415

''sovereign''state is that one does not violate the territorialrights of others without"just cause." A clear exampleof such an institutionaleffect, convinc- inglyargued by David Strang,is the markedlydifferent treatment that weak states receivewithin and outside communitiesof mutualrecognition.77 What keeps the United Statesfrom conquering the Bahamas,or Nigeriafrom seizing Togo, or Australiafrom occupying Vanuatu? Clearly,power is not the issue, and in thesecases even the costof sanctionswould probablybe negligible.One mightargue thatgreat powers simplyhave no "interest"in these conquests, and thismight be so, butthis lack of interestcan onlybe understoodin termsof theirrecognition of weak states'sovereignty. I have no interestin exploitingmy friends,not because ofthe relative costs and benefitsof such action but because theyare my friends.The absence of recognition,in turn,helps explain the Westernstates' practicesof territorialconquest, enslavement, and genocide against Native American and Africanpeoples. It is in thatworld that only powermatters, not theworld of today. Finally,to the extentthat their ongoing socialization teaches statesthat their sovereigntydepends on recognitionby other states, they can affordto relymore on the institutionalfabric of internationalsociety and less on individual national means-especially militarypower-to protect their security.The intersubjectiveunderstandings embodied in the institutionof sovereignty,in otherwords, may redefine the meaningof others'power for the securityof the self.In policyterms, this means that states can be less worriedabout short-term survivaland relativepower and can thus shifttheir resources accordingly. Ironically,it is the greatpowers, the stateswith the greatestnational means, thatmay have the hardesttime learning this lesson; smallpowers do not have the luxuryof relyingon national means and may thereforelearn fasterthat collectiverecognition is a cornerstoneof security. None of this is to say that power becomes irrelevantin a communityof sovereignstates. Sometimes states are threatenedby others that do not recognize their existence or particular territorialclaims, that resent the externalitiesfrom their economic policies, and so on. But most of the time, thesethreats are playedout within the terms of the sovereignty game. The fates of Napoleon and Hitlershow what happens when they are not.

Cooperation among egoists and transformationsof identity We began this sectionwith a Hobbesian state of nature. Cooperation for joint gain is extremelydifficult in this context,since trustis lacking,time horizonsare short,and relativepower concerns are high.Life is "nasty,brutish, and short." Sovereigntytransforms this systeminto a Lockean world of (mostly)mutually recognized property rights and (mostly)egoistic rather than

77. David Strang,"Anomaly and Commonplacein European Expansion: Realist and Institu- tionalAccounts," Intemational Organization 45 (Spring1991), pp. 143-62. 416 InternationalOrganization

competitiveconceptions of security,reducing the fearthat what states already have willbe seized at anymoment by potential collaborators, thereby enabling themto contemplatemore directforms of cooperation.A necessarycondition forsuch cooperationis thatoutcomes be positivelyinterdependent in the sense thatpotential gains existwhich cannot be realized by unilateralaction. States such as Brazil and Botswanamay recognize each other'ssovereignty, but they need furtherincentives to engage in joint action. One importantsource of incentivesis the growing"dynamic density" of interactionamong states in a world with new communicationstechnology, nuclear weapons, externalities fromindustrial development, and so on.78Unfortunately, growing dynamic densitydoes not ensurethat states will in factrealize joint gains; interdepen- dence also entailsvulnerability and the riskof being "the sucker,"which if exploitedwill become a sourceof conflictrather than cooperation. This is the rationalefor the familiar assumption that egoistic states will often findthemselves facing prisoners' dilemma, a game in which the dominant strategy,if played only once, is to defect. As Michael Taylor and Robert Axelrod have shown,however, given iteration and a sufficientshadow of the future,egoists using a tit-for-tatstrategy can escape this result and build cooperativeinstitutions.79 The storythey tell about thisprocess on the surface seems quite similar to George Herbert Mead's constructivistanalysis of interaction,part of whichis also told in termsof "games."8"Cooperation is a gestureindicating ego's willingnessto cooperate; if alter defects,ego does likewise,signaling its unwillingnessto be exploited;over time and through reciprocalplay, each learns to formrelatively stable expectationsabout the other'sbehavior, and throughthese, habits of cooperation(or defection)form. Despite similarconcerns with communication, learning, and habit-formation, however,there is an importantdifference between the game-theoreticand constructivistanalysis of interactionthat bears on how we conceptualizethe causal powersof institutions. In the traditionalgame-theoretic analysis of cooperation,even an iterated one, the structureof the game-of identitiesand interests-is exogenousto interactionand, as such, does not change.8'A "black box" is put around identity-and interest-formation,and analysisfocuses instead on the relation-

78. On "dynamicdensity," see Ruggie,"Continuity and Transformationin the World Polity"; and Waltz, "Reflectionson Theoryof IntemationalPolitics." The role of interdependencein conditioningthe speed and depthof social learningis muchgreater than the attentionto whichI have paid it. On the consequences of interdependenceunder anarchy,see Helen Milner,"The Assumptionof Anarchyin InternationalRelations Theory:A Critique,"Review of Intemational Studies17 (January1991), pp. 67-85. 79. See Michael Taylor,Anarchy and Cooperation(New York: Wiley, 1976); and Robert Axelrod,The Evolution of Cooperation(New York: Basic Books, 1984). 80. Mead, Mind,Self and Society. 81. Strictlyspeaking, this is not true,since in iteratedgames the additionof futurebenefits to currentones changesthe payoff structure of the game at Ti, in thiscase fromprisoners' dilemma to an assurancegame. This transformationof interesttakes place entirelywithin the actor,however, and as such is nota functionof interactionwith the other. Anarchy 417

ship betweenexpectations and behavior.The normsthat evolve from interaction are treatedas rules and behavioralregularities which are externalto the actorsand whichresist change because of the transactioncosts of creatingnew ones. The game-theoreticanalysis of cooperation among egoists is at base behavioral. A constructivistanalysis of cooperation,in contrast,would concentrateon how the expectationsproduced by behavior affect identities and interests.The process of creatinginstitutions is one of internalizingnew understandingsof self and other,of acquiringnew role identities,not just of creatingexternal constraintson the behaviorof exogenouslyconstituted actors.82 Even if not intended as such, in other words, the process by which egoists learn to cooperate is at the same time a process of reconstructingtheir interests in termsof shared commitmentsto social norms.Over time,this will tend to transforma positiveinterdependence of outcomesinto a positiveinterdependence of utilitiesor collectiveinterest organized around the normsin question. These normswill resist change because theyare tied to actors'commitments to their identitiesand interests,not merelybecause of transactioncosts. A constructivistanalysis of "the cooperationproblem," in otherwords, is at base cognitiverather than behavioral,since it treatsthe intersubjectiveknowledge that defines the structureof identitiesand interests,of the "game," as endogenousto and instantiatedby interaction itself. The debate over the futureof collectivesecurity in WesternEurope may illustratethe significanceof this difference.A weak liberal or rationalist analysiswould assume that the European states' "portfolio"of interestshas not fundamentallychanged and thatthe emergenceof new factors,such as the collapse of the Soviet threat and the rise of Germany,would alter their cost-benefitratios for pursuing current arrangements, thereby causing existing institutionsto break down. The European statesformed collaborative institu- tionsfor good, exogenouslyconstituted egoistic reasons, and the same reasons maylead themto rejectthose institutions; the game of European powerpolitics has not changed. A strongliberal or constructivistanalysis of this problem would suggest that four decades of cooperation may have transformeda positiveinterdependence of outcomesinto a collective"European identity"in termsof which states increasingly define their "self "-interests.83 Even ifegoistic reasons were its startingpoint, the process of cooperatingtends to redefine those reasons by reconstitutingidentities and interestsin terms of new intersubjectiveunderstandings and commitments.Changes in the distribution ofpower during the late twentiethcentury are undoubtedlya challengeto these new understandings,but it is not as if West European states have some

82. In fairnessto Axelrod,he does pointout thatinternalization of norms is a real possibilitythat mayincrease the resilienceof institutions.My pointis thatthis important idea cannotbe derived froman approachto theorythat takes identities and interestsas exogenouslygiven. 83. On "European identity,"see BarryBuzan et al., eds., The EuropeanSecurity Order Recast (London: Pinter,1990), pp. 45-63. 418 InternationalOrganization

inherent,exogenously given interestin abandoningcollective security if the priceis right.Their identities and securityinterests are continuouslyin process, and if collectiveidentities become "embedded," theywill be as resistantto change as egoistic ones.84Through participationin new forms of social knowledge,in otherwords, the European statesof 1990might no longerbe the statesof 1950.

Critical strategictheory and collective security The transformationof identityand interest throughan "evolution of cooperation"faces two importantconstraints. The firstis that the process is incrementaland slow. Actors' objectivesin such a process are typicallyto realize joint gains withinwhat theytake to be a relativelystable context,and theyare thereforeunlikely to engage in substantialreflection about how to changethe parametersof thatcontext (including the structureof identitiesand interests)and unlikelyto pursue policies specificallydesigned to bringabout such changes. Learningto cooperate may change those parameters,but this occurs as an unintendedconsequence of policies pursued for other reasons ratherthan as a resultof intentionalefforts to transcendexisting institutions. A second,more fundamental, constraint is thatthe evolutionof cooperation storypresupposes that actors do not identifynegatively with one another. Actors must be concerned primarilywith absolute gains; to the extentthat antipathyand distrustlead themto definetheir security in relativisticterms, it will be hard to accept the vulnerabilitiesthat attend cooperation.85This is importantbecause it is preciselythe "centralbalance" in the statesystem that seems to be so oftenafflicted with such competitivethinking, and realistscan thereforeargue that the possibilityof cooperation withinone "pole" (for example,the West) is parasiticon thedominance of competition between poles (the East-West conflict).Relations between the poles may be amenable to some positivereciprocity in areas such as armscontrol, but the atmosphereof distrustleaves littleroom forsuch cooperationand its transformativeconse- quences.86The conditionsof negativeidentification that make an "evolutionof cooperation"most needed workprecisely against such a logic. This seeminglyintractable situation may nevertheless be amenableto quite a differentlogic of transformation,one drivenmore by self-consciousefforts to change structuresof identityand interestthan by unintendedconsequences. Such voluntarismmay seem to contradictthe spiritof constructivism,since

84. On "embeddedness,"see JohnRuggie, "International Regimes, Transactions, and Change: Embedded Liberalism in a Postwar Economic Order," in Krasner,Intemational Regimes, pp. 195-232. 85. See Grieco,"Anarchy and the Limitsof Cooperation." 86. On the difficultiesof creatingcooperative security regimes given competitive interests, see Robert Jervis,"Security Regimes," in Krasner,Intemational Regimes, pp. 173-94; and Charles Lipson,"International Cooperation in Economic and SecurityAffairs," World Politics 37 (October 1984), pp. 1-23. Anarchy 419 would-berevolutionaries are presumablythemselves effects of socializationto structuresof identityand interest.How can theythink about changingthat to whichthey owe theiridentity? The possibilitylies in the distinctionbetween the social determinationof the self and the personal determinationof choice, betweenwhat Mead called the "me" and the "."87 The "me" is that part of subjectivitywhich is definedin termsof others;the characterand behavioral expectationsof a person'srole identityas "professor,"or of the United States as "leader of the alliance," forexample, are sociallyconstituted. Roles are not played in mechanicalfashion according to precise scripts,however, but are "taken" and adapted in idiosyncraticways by each actor.88Even in the most constrainedsituations, role performanceinvolves a choiceby the actor.The "I" is the partof subjectivityin whichthis appropriation and reactionto roles and itscorresponding existential freedom lie. The factthat roles are "taken" means that,in principle,actors always have a capacityfor "character planning"-for engagingin criticalself-reflection and choicesdesigned to bringabout changesin theirlives.89 But whenor underwhat conditionscan thiscreative capacity be exercised?Clearly, much of the timeit cannot:if actors were constantly reinventing their identities, social orderwould be impossible,and the relativestability of identitiesand interestsin the real worldis indicativeof our propensityfor habitual rather than creativeaction. The exceptional,conscious choosingto transformor transcendroles has at least two preconditions.First, there mustbe a reason to thinkof oneselfin novel terms.This would most likelystem fromthe presence of new social situationsthat cannot be managed in termsof preexistingself-conceptions. Second, the expectedcosts of intentionalrole change-the sanctionsimposed byothers with whom one interactedin previousroles-cannot be greaterthan itsrewards. When these conditionsare present,actors can engage in self-reflectionand practice specificallydesigned to transformtheir identities and interestsand thus to "change the games" in which they are embedded. Such "critical" strategictheory and practice has not received the attentionit meritsfrom students of world politics (another legacy of exogenouslygiven interests perhaps), particularlygiven that one of the most importantphenomena in contemporaryworld politics, Mikhail Gorbachev's policy of "New Thinking,"is

87. See Mead, Mind, Self and Society.For useful discussions of this distinctionand its implicationsfor notions of creativityin social systems,see George Cronk, The Philosophical Anthropologyof GeorgeHerbert Mead (New York: Peter Lang, 1987), pp. 36-40; and Howard, "From ChangingSelves TowardChanging Society." 88. Turner,"Role-Taking." 89. On "characterplanning," see JonElster, Sour Grapes:Studies in theSubversion of Rationality (Cambridge:Cambridge University Press, 1983), p. 117. For otherapproaches to the problemof self-initiatedchange, see HarryFrankfurt, "Freedom of the Will and the Concept of a Person," Joumalof Philosophy 68 (January1971), pp. 5-20; AmartyaSen, "Rational Fools: A Critiqueof the BehavioralFoundations of EconomicTheory," Philosophy and PublicAffairs 6 (Summer1977), pp. 317-44; and Thomas Schelling,"The IntimateContest for Self-Command," The PublicInterest 60 (Summer1980), pp. 94-118. 420 InternationalOrganization arguablyprecisely that.90 Let me thereforeuse thispolicy as an exampleof how states mighttransform a competitivesecurity system into a cooperativeone, dividingthe transformativeprocess into four stages. The firststage in intentionaltransformation is the breakdownof consensus aboutidentity commitments. In the Sovietcase, identitycommitments centered on the Leninisttheory of imperialism,with its belief that relationsbetween capitalistand socialist states are inherentlyconflictual, and on the alliance patternsthat this belief engendered.In the 1980s, the consensuswithin the Soviet Union over the Leninisttheory broke down for a varietyof reasons, principalamong which seem to have been the state's inabilityto meet the economic-technological-militarychallenge fromthe West, the government's declineof politicallegitimacy at home,and the reassurancefrom the West that it did not intendto invade the Soviet Union, a reassurancethat reduced the externalcosts of role change.9"These factorspaved the way for a radical leadershiptransition and for a subsequent "unfreezingof conflictschemas" concerningrelations with the West.92 The breakdown of consensus makes possible a second stage of critical examinationof old ideas about self and other and, by extension,of the structuresof interactionby which the ideas have been sustained.In periodsof relativelystable role identities,ideas and structuresmay become reifiedand thustreated as thingsthat exist independently of social action.If so, the second stage is one of denaturalization,of identifyingthe practicesthat reproduce seeminglyinevitable ideas about self and other;to that extent,it is a formof "critical"rather than "problem-solving" theory.93 The resultof such a critique should be an identificationof new "possible selves" and aspirations.94New

90. For usefuloverviews of New Thinking,see MikhailGorbachev, Perestroika: New Thinkingfor Our Countryand the World(New York: Harper & Row, 1987); Vendulka Kubalkova and Albert Cruickshank,Thinking New About Soviet "New Thinking"(Berkeley: Instituteof International Studies,1989); and Allen Lynch,Gorbachev's Intemational Outlook: Intellectual Origins and Political Consequences(New York: Institutefor East-West SecurityStudies, 1989). It is not clear to what extentNew Thinkingis a consciouspolicy as opposed to an ad hoc policy.The intensetheoretical and policydebate withinthe Soviet Union over New Thinkingand the frequentlystated idea of takingaway the Western"excuse" forfearing the SovietUnion both suggestthe former,but I will remainagnostic here and simplyassume thatit can be fruitfullyinterpreted "as if' it had the form thatI describe. 91. For useful overviewsof these factors,see Jack Snyder,"The Gorbachev Revolution:A Waningof Soviet Expansionism?"World Politics 12 (Winter1987-88), pp. 93-121; and Stephen Meyer, "The Sources and Prospects of Gorbachev's New Political Thinkingon Security," IntemationalSecurity 13 (Fall 1988), pp. 124-63. 92. See Daniel Bar-Tal et al., "ConflictTermination: An EpistemologicalAnalysis of Interna- tional Cases," PoliticalPsychology 10 (June 1989), pp. 233-55. For an unrelatedbut interesting illustrationof how changingcognitions in turn make possible organizationalchange, see Jean Bartunek,"Changing Interpretive Schemes and OrganizationalRestructuring: The Example of a ReligiousOrder,"Administrative Science Quarterly 29 (September1984), pp. 355-72. 93. See RobertCox, "Social Forces, States and World Orders:Beyond International Relations Theory,"in Keohane, Neorealismand Its Critics,pp. 204-55. See also Brian Fay, CriticalSocial Science(Ithaca, N.Y.: CornellUniversity Press, 1987). 94. Hazel Markus and Paula Nurius,"Possible Selves,"American Psychologist 41 (September 1986), pp. 954-69. Anarchy 421

Thinkingembodies such criticaltheorizing. Gorbachev wants to freethe Soviet Union fromthe coercivesocial logic of the cold war and engage the West in far-reachingcooperation. Toward thisend, he has rejectedthe Leninistbelief in the inherentconflict of interestbetween socialist and capitaliststates and, perhapsmore important, has recognizedthe crucialrole thatSoviet aggressive practicesplayed in sustainingthat conflict. Such rethinkingpaves the way for a thirdstage of new practice.In most cases, it is not enoughto rethinkone's ownideas about selfand other,since old identitieshave been sustainedby systemsof interactionwith other actors, the practicesof which remain a social factfor the transformativeagent. In orderto changethe self,then, it is oftennecessary to changethe identitiesand interests of the othersthat help sustainthose systemsof interaction.The vehicle for inducingsuch change is one's own practiceand, in particular,the practiceof "altercasting"-a techniqueof interactorcontrol in whichego uses tacticsof self-presentationand stage management in an attempt to frame alter's definitionsof social situationsin ways that create the role whichego desires alterto play.95In effect,in altercastingego triesto inducealter to take on a new identity(and therebyenlist alter in ego's effortto changeitself) by treating alter as ifit already had that identity.The logic of this followsdirectly from the mirrortheory of identity-formation,in which alter's identityis a reflectionof ego's practices; change those practices and ego begins to change alter's conceptionof itself. What these practicesshould consistof depends on the logic by whichthe preexistingidentities were sustained. Competitivesecurity systems are sustainedby practices that create insecurityand distrust.In thiscase, transforma- tivepractices should attemptto teach otherstates that one's own statecan be trustedand shouldnot be viewedas a threatto theirsecurity. The fastestway to do this is to make unilateral initiativesand self-bindingcommitments of sufficientsignificance that another state is faced with "an offerit cannot refuse."96Gorbachev has triedto do thisby withdrawing from Afghanistan and Eastern Europe, implementingasymmetric cuts in nuclear and conventional forces,calling for "defensive defense," and so on. In addition,he has skillfully cast the West in the role of being morallyrequired to give aid and comfortto the Soviet Union, has emphasized the bonds of common fate between the Soviet Union and the West, and has indicated that furtherprogress in

95. See Goffman,The Presentationof Self in EverydayLife; Eugene Weinstein and Paul Deutschberger,"Some Dimensionsof Altercasting,"Sociometry 26 (December 1963), pp. 454-66; and Walter Earle, "InternationalRelations and the Psychologyof Control:Alternative Control Strategiesand Their Consequences,"Political Psychology 7 (June1986), pp. 369-75. 96. See Volker Boge and Peter Wilke,"Peace Movementsand UnilateralDisarmament: Old Conceptsin a New Light,"Arms Control 7 (September1986), pp. 156-70; Zeev Maoz and Daniel Felsenthal,"Self-Binding Commitments, the Inducementof Trust,Social Choice, and the Theory of InternationalCooperation," Intemational Studies Quarterly 31 (June1987), pp. 177-200; and V. Sakamoto,"Unilateral Initiative as an AlternativeStrategy," World Futures, vol. 24, nos. 1-4, 1987, pp. 107-34. 422 InternationalOrganization

East-Westrelations is contingentupon the West assuming the identity being projectedonto it. These actions are all dimensionsof altercasting, the intention ofwhich is totake away the Western "excuse" for distrusting the Soviet Union, which,in Gorbachev'sview, has helpedsustain competitive identities in the past. Yet bythemselves such practices cannot transform a competitive security system,since if they are notreciprocated by alter, they will expose ego to a "sucker"payoff and quicklywither on thevine. In orderfor critical strategic practiceto transformcompetitive identities, it mustbe "rewarded"by alter, whichwill encourage more such practice by ego, and so on.97Over time, this will institutionalizea positive rather than a negativeidentification between the securityof self and otherand will thereby provide a firmintersubjective basis forwhat were initially tentative commitments to new identities and interests.98 Notwithstandingtoday's rhetoric about the end of the cold war, skeptics may stilldoubt whether Gorbachev (or somefuture leader) will succeed in building an intersubjectivebasis for a newSoviet (or Russian)role identity. There are importantdomestic, bureaucratic, and cognitive-ideologicalsources of resis- tancein bothEast and Westto sucha change,not the least of whichis the shakinessof the democraticforces' domestic position. But if myargument aboutthe role of intersubjective knowledge in creatingcompetitive structures of identityand interestis right,then at leastNew Thinkingshows a greater appreciation-consciousor not-forthe deep structureof power politics than we are accustomedto in international relations practice.

Conclusion

All theoriesof internationalrelations are based on social theoriesof the relationshipbetween agency, process, and socialstructure. Social theories do notdetermine the content of our international theorizing, but they do structure the questionswe ask aboutworld politics and our approachesto answering thosequestions. The substantiveissue at stakein debates about social theory is whatkind of foundation offers the most fruitful set of questions and research strategiesfor explaining the revolutionary changes that seem to be occurringin the late twentiethcentury international system. Put simply,what should systemictheories of internationalrelations look like? How should they conceptualizethe relationship between structure and process? Should they be

97. On rewards, see Thomas Milburn and Daniel Christie,"Rewarding in International Politics,"Political Psychology 10 (December 1989), pp. 625-45. 98. The importanceof reciprocityin completingthe processof structuraltransformation makes the logic in thisstage similarto that in the "evolutionof cooperation."The differenceis one of prerequisitesand objective:in the former,ego's tentativeredefinition of selfenables it to tryand changealter by acting "as if" bothwere alreadyplaying a new game; in the latter,ego acts onlyon the basis of given interestsand prior experience,with transformationemerging only as an unintendedconsequence. Anarchy 423 based exclusivelyon "microeconomic" analogies in which identities and interests are exogenouslygiven by structureand process is reduced to interactionswithin those parameters? Or should they also be based on "sociological" and "social psychological"analogies in which identitiesand interestsand thereforethe meaningof structureare endogenousto process? Should a behavioral-individualismor a cognitive-constructivismbe the basis for systemictheories of world politics? This articlenotwithstanding, this question is ultimatelyan empiricalone in two respects.First, its answerdepends in part on how importantinteraction amongstates is forthe constitutionof theiridentities and interests.On the one hand, it maybe that domesticor geneticfactors, which I have systematically bracketed,are in factmuch more importantdeterminants of states' identities and intereststhan are systemicfactors. To the extentthat this is true, the individualismof a rationalistapproach and theinherent privileging of structure over process in this approach become more substantivelyappropriate for systemictheory (if not forfirst- and second-imagetheory), since identitiesand interestsare infact largely exogenous to interactionamong states. On theother hand,if the bracketed factors are relativelyunimportant or ifthe importance of the internationalsystem varies historically(perhaps withthe level of dynamic densityand interdependencein the system),then such a frameworkwould not be appropriateas an exclusivefoundation for general systemic theory. Second,the answerto the questionabout what systemic theories should look like also depends on how easilystate identitiesand interestscan change as a result of systemicinteraction. Even if interactionis initiallyimportant in constructingidentities and interests,once institutionalizedits logic maymake transformationextremely difficult. If the meaningof structurefor state action changes so slowlythat it becomes a de factoparameter within which process takes place, then it may again be substantivelyappropriate to adopt the rationalistassumption that identitiesand interestsare given(although again, thismay vary historically). We cannot address these empirical issues, however,unless we have a frameworkfor doing systemic research that makes state identity and interestan issue forboth theoreticaland empiricalinquiry. Let me emphasizethat this is notto saywe shouldnever treat identities and interestsas given.The framingof problems and research strategiesshould be question-drivenrather than method-driven,and ifwe are not interestedin identity-and interest-formation, we may findthe assumptionsof a rationalistdiscourse perfectly reasonable. Nothingin this article, in other words, should be taken as an attack on rationalismper se. By the same token, however,we should not let this legitimateanalytical stance become a de factoontological stance with respect to thecontent of third-image theory, at least notuntil after we have determined thatsystemic interaction does not play an importantrole in processesof state identity-and interest-formation.We should not choose our philosophical anthropologiesand social theoriesprematurely. By arguingthat we cannot 424 InternationalOrganization derive a self-helpstructure of identityand interestfrom the principleof anarchyalone-by arguingthat anarchy is what statesmake of it-this article has challengedone importantjustification for ignoring processes of identity- and interest-formationin world politics.As such, it helps set the stage for inquiryinto the empiricalissues raised above and thus for a debate about whethercommunitarian or individualistassumptions are a betterfoundation forsystemic theory. I have triedto indicateby crude examplewhat such a researchagenda might look like. Its objectiveshould be to assess the causal relationshipbetween practiceand interaction(as independentvariable) and the cognitivestructures at the level of individualstates and of systemsof states which constitute identitiesand interests(as dependent variable)-that is, the relationship betweenwhat actorsdo and whatthey are. We mayhave some a priorinotion thatstate actorsand systemicstructures are "mutuallyconstitutive," but this tells us little in the absence of an understandingof how the mechanicsof dyadic,triadic, and n-actorinteraction shape and are in turnshaped by "stocks of knowledge"that collectivelyconstitute identities and interestsand, more broadly,constitute the structuresof international life. Particularly important in thisrespect is the role of practicein shapingattitudes toward the "givenness" of these structures.How and whydo actorsreify social structures,and under whatconditions do theydenaturalize such reifications? The state-centrismof thisagenda maystrike some, particularly postmodernists,as "depressinglyfamiliar."99 The significanceof statesrelative to multina- tionalcorporations, new social movements, transnationals, and intergovernmen- tal organizationsis clearlydeclining, and "postmodern"forms of world politics meritmore research attention than they have received.But I also believe,with realists,that in the medium run sovereignstates will remain the dominant politicalactors in the internationalsystem. Any transition to new structuresof global politicalauthority and identity-to"postinternational" politics-will be mediatedby and path-dependenton the particularinstitutional resolution of the tensionbetween unity and diversity,or particularismand universality,that is the sovereignstate.100 In such a worldthere should continue to be a place for theoriesof anarchicinterstate politics, alongside other forms of international theory;to thatextent, I am a statistand a realist.I have arguedin thisarticle, however,that statismneed not be bound by realistideas about what "state" mustmean. State identitiesand interestscan be collectivelytransformed within an anarchic context by many factors-individual, domestic, systemic,or transnational-and as such are an importantdependent variable. Such a

99. Yale Fergusonand Richard Mansbach, "Between Celebrationand Despair: Constructive Suggestionsfor Future International Theory," Intermational Studies Quarterly 35 (December 1991), p.375. 100. For excellentdiscussions of thistension, see Walker,"Sovereignty, Identity, Community"; and R. B. J. Walker,"Security, Sovereignty, and the Challengeof World Politics,"Altematives 15 (Winter1990), pp. 3-27. On institutionalpath dependencies,see StephenKrasner, "Sovereignty: An InstitutionalPerspective," Comparative Political Studies 21 (April 1988), pp. 66-94. Anarchy 425 reconstructionof state-centricinternational theory is necessaryif we are to theorize adequately about the emergingforms of transnationalpolitical identitythat sovereign states will help bringinto being. To thatextent, I hope thatstatism, like the state,can be historicallyprogressive. I have arguedthat the proponents of strongliberalism and theconstructivists can and shouldjoin forcesin contributingto a process-orientedinternational theory.Each group has characteristicweaknesses that are complementedby the other's strengths.In part because of the decision to adopt a choice- theoreticapproach to theoryconstruction, neoliberals have been unable to translate their work on institution-buildingand complex learning into a systemictheory that escapes the explanatorypriority of realism'sconcern with structure.Their weakness, in other words, is a lingeringunwillingness to transcend,at the level of systemictheory, the individualistassumption that identitiesand interestsare exogenouslygiven. Constructivists bring to thislack of resolutiona systematiccommunitarian ontology in which intersubjective knowledge constitutesidentities and interests.For their part, however, constructivistshave oftendevoted too mucheffort to questionsof ontologyand constitutionand not enougheffort to thecausal and empiricalquestions of how identitiesand interestsare producedby practicein anarchicconditions. As a result,they have not takenon board neoliberalinsights into learning and social cognition. An attemptto use a structurationist-symbolicinteractionist discourse to bridge the two research traditions,neither of which subscribesto such a discourse,will probablyplease no one. But in part this is because the two "sides" have become hungup on differencesover the epistemologicalstatus of social science. The state of the social sciences and, in particular,of internationalrelations, is such thatepistemological prescriptions and conclusionsare at best premature.Different questions involve different standards of inference; to reject certain questions because their answers cannot conformto the standardsof classical physicsis to fall into the trap of method-drivenrather thanquestion-driven social science.By the same token,however, giving up the artificialrestrictions of logical positivistconceptions of inquirydoes not force us to giveup on "Science." Beyondthis, there is littlereason to attachso much importanceto epistemology.Neither positivism,nor scientificrealism, nor poststructuralismtells us about the structureand dynamicsof international life. Philosophiesof science are not theoriesof internationalrelations. The good news is that strongliberals and modernand postmodernconstructivists are asking broadly similar questions about the substance of international relationsthat differentiate both groups from the neorealist-rationalistalliance. Strongliberals and constructivistshave muchto learn fromeach otherif they can come to see thisthrough the smokeand heat of epistemology.