InternationalNorm Dynamics and PoliticalChange MarthaFinnemore and Kathryn Sikkink

Normativeand ideationalconcerns have alwaysinformed the study of international politicsand are a consistentthread running through the life of International Organi- zation.When IO was founded,dominant realist views of politics,while rejecting idealism,were verymuch concerned with issues of legitimacyand ideology.The earlyCold War,after all, was not simplya positionalconflict among anonymous greatpowers: it was a war for "heartsand minds."The couplingof powerwith "legitimatesocial purpose"was centralto Americanforeign policy of thisperiod.' At the same time,international relations scholars were busy studyingtwo of the greatestsocial construction projects of the age: Europeanintegration and decoloniza- tion.Neofunctionalists, like therealists, were consciously trying to distancethem- selvesfrom "idealist" predecessors (in thiscase, David Mitranyand his colleagues), butthe complex web of technical tasks that they designed aimed at more than promot- ing materialwell-being; they aimed ultimately at ideationaland social ends.Spill- overwas supposedto do morethan create additional technical tasks; it was supposed tochange attitudes, identity, and affect among participants. Likewise, scholars recog- nizedthat decolonization was drivenby a profoundlynormative agenda and thatit explicitlysought to reconstitutethe identities of both the new states and their former colonizers,as wellas therelationships between them. Attemptsin the 1960s and 1970sto build a scienceof politics modeled on econom- ics or naturalscience never displaced these normative and ideationalconcerns com- pletely.They have surfacedconsistently in thestream of critiquesof thedominant state-centricparadigms that focused on materialpower. Scholars of transnational relationsin the 1970s called attentionto transnationalactors who weresometimes influencedby norms and ideas.2 The regimesscholarship of the early 1980s similarly

Wewould like to thank the three editors, two reviewers, and Emanuel Adler, Michael Barnett, Elizabeth Boyle,Audie Klotz, Jeff Legro, Richard Price, Thomas Risse, and Daniel Thomasfor their very helpful commentson variousdrafts of themanuscript. We are also gratefulto JamesFearon, Kurt Gaubatz, and JimMorrow for answering questions, sharing manuscripts, and making suggestions related to ourwork. 1. Ruggie1983b. 2. Keohaneand Nye 1971.

InternationalOrganization 52, 4, Autumn1998, pp. 887-917 ? 1998by The 10 Foundationand the Massachusetts Institute of Technology 888 InternationalOrganization emphasizedthe role of principles and norms in waysthat opened the door for a more sweeping"ideational turn" in thelate 1980s. Elsewherein thisissue JohnRuggie, James March, and JohanOlsen explorethe intellectualhistory of thisrecent "turn" and locateits proponentsin themore ab- stracttheoretical debates of social science.Building on theircontributions, we ad- dresstheoretical issues facing those of us interestedin empiricalresearch on social constructionprocesses and norminfluences in internationalpolitics. We are con- cernedwith such questions as How do we knowa normwhen we see one? How do we knownorms make a differencein politics?Where do normscome from? How do theychange? We areparticularly interested in therole norms play in political change- boththe ways in whichnorms, themselves, change and the ways in whichthey changeother features of the political landscape. Like othertheoretical frameworks in internationalrelations (IR), muchof themacrotheoretical equipment of constructiv- ismis betterat explainingstability than change. Claims that actors conform to "log- ics of appropriateness"3say littleabout how standardsof appropriatenessmight change.Such staticapproaches to IR areparticularly unsatisfying during the current era of global transformationwhen questions about change motivate much of the empiricalresearch we do. Lackinggood macrotheoreticguidance, our approach to thesequestions relies heavily on inductionfrom the extensive and growingbody of normsresearch that has sprungup, not just in politicalscience, but in law,econom- ics, sociology,and psychology. This varietyof conceptualand empiricalmaterial is usefulfor our inductive enterprise, but it also raisessome importantquestions for macrotheorythat we exploreat theend of the article. We use ourreview of scholarshipon normsand relatedideational phenomena in thisarticle to makethree arguments. First, the ideational "turn" of recentyears is actuallya returnto sometraditional concerns of the discipline, but it has notbrought us backto precisely the same place we began.Standards for good empirical research havechanged dramatically (and for the better) since the founding of IO, andapplying thesestandards to long-standingnormative issues has had realpayoffs. Second, we generatesome propositions about three aspects of norms-theirorigins, the mecha- nismsby whichthey exercise influence, and theconditions under which norms will be influentialin worldpolitics. Specifically, we arguethat norms evolve in a pat- terned"life cycle" and thatdifferent behavioral logics dominate different segments of thelife cycle. Third, we arguethat the current tendency to opposenorms against rationalityor rational choice is nothelpful in explainingmany of the most politically salientprocesses we see in empiricalresearch-processes we call "strategicsocial construction,"in whichactors strategize rationally to reconfigurepreferences, iden- tities,or social context.4 Rationality cannot be separatedfrom any politically signifi- cantepisode of normative influence or normativechange, just as thenormative con- textconditions any episode of rationalchoice. Normsand rationalityare thus intimatelyconnected, but scholars disagree about the precise nature of their relation-

3. Marchand Olsen, this issue. 4. See Kahler,this issue, for a historyof debates over rationality in internationalrelations. NormDynamics and Political Change 889 ship.We identifyfour focal points of debatewhere the relationship between norms and rationalityis leastunderstood and mostimportant, and we showhow thesede- batescross-cut research traditions in potentiallyfruitful ways.

The Returnto Norms

Normsand normative issues have been central to thestudy of politics for at least two millennia.Students of politics have struggled with questions not only about the mean- ingof justice and the good society but also aboutthe influence on humanbehavior of ideas aboutjustice and good. Our conclusions(or ourassumptions) about these is- suescondition every form of political analysis. Aristotle and Plato understood this in thefourth century B.C.E. just as E. H. Carrdid in thetwentieth century. Carr has becomecanonized in thediscipline as a debunkerof Wilsonianidealism, but this simplificationmisrepresents his message:"Political action must be basedon a coor- dinationof morality and power." Realism fails, in Carr'sanalysis, precisely because it excludesessential features of politics like emotional appeal to a politicalgoal and groundsfor moral judgment.5 This conviction that understanding social purpose and legitimacywas essentialto understandingpolitics continued through the 1950sand 1960s and is evidentin the pages of IO. Inis Claude's workon the legitimation functionof the United Nations deals preciselywith this issue.6 Early IR scholarship on certainissues-decolonization, human rights, education-recognized that much UN activityinvolved establishing norms, but it often failed to theorizethese norma- tiveprocesses.7 Certainly the work of theintegration theorists such as ErnstHaas was implicitly,and often explicitly, about creating shared social purpose.8 Even real- istslike Hans Morgenthauwrote extensively about the way in whichideational and normativefactors such as nationalism,morality, and internationallaw limitstates' exerciseof power.9 The "turn"away from norms and normativeconcerns began with the behavioral revolutionand itsenthusiasm for measurement. Normative and ideationalphenom- ena weredifficult to measureand so tendedto be pushedaside formethodological reasons.This tendency was reinforcedby the emerging infatuation of political scien- tistswith economic methods in thelate 1970sand 1980s.10Realists began recasting thepursuit of power as "utilitymaximization" and, following the economists, tended to specifyutility functions in materialterms only. Liberals drew on microeconomic analysesof collective action games (Prisoners' Dilemma, Stag Hunt) to reinvigorate theirlong-standing debate with realists and showthat cooperation, welfare improve-

5. Carr[1946] 1964,97, 89. 6. Claude 1966. 7. See Kay 1967;Henkin 1965; Jacobson 1962; and Ball 1961. 8. Haas 1961,1964a. For an argumentrelated to someconcerns of this article, see Haas 1993. 9. Morgenthau[1948] 1985,pt. 5, 6. 10. Fora discussionabout the causes and effects of the dominance of economic models and methods in thestudy of U.S. politics,see Lowi 1992; Simon1993a; Lowi 1993a,b;and Simon 1993b. 890 InternationalOrganization ment,and "progress"were possible even given some of realism's pessimistic assump- tionsabout self-seeking human nature. In fact,these "neos," both realist and liberal, mightmore appropriately be called "econorealists"and "econoliberals,"since what was newin bothcases was an injectionof microeconomic insights.'1 Althoughthe move to rationalchoice in no way requireda move to a material ontology,its proponents showed little interest in ideationalor socialphenomena, and studyof these issues languished during this period. However, when interest in these mattersrevived in the1980s, first with the regimes projectl2 and laterwith the con- structivistsled by JohnRuggie, Friedrich Kratochwil, Alexander Wendt, and oth- ers,'3the discipline to whichnorms returned had changedin importantways. Al- thoughthe behavioral revolution and the "economicturn" of the 1970s and 1980s mayhave neglected norms, they made important contributions by forcing scholars to thinkmuch more rigorously about issues of research design, theoretical clarity, disci- plinarycumulation, and parsimony. As contemporaryresearchers make their arguments about norms, culture, and ideas, theywill need to specifyideational causal claimsand mechanismsclearly, think seriouslyabout the microfoundations on which theoretical claims about norms rest, andevaluate those claims in thecontext of carefullydesigned historical and empiri- cal research.'4David Lumsdaine'sanalysis of therole of moralityin international politics,for example, is verydifferent from Carr's. Carr uses evidenceanecdotally to illustratehis argumentsabout the moral character of "the ordinaryman" and the politicalconsequences of thatmorality. The resultis a brilliantpiece of political thoughtbut one vulnerableto chargesof wishfulthinking, since Cafr provides no systematicevidence that human beings actually do behaveas his "ordinaryman"" does and notaccording to some otherconception of humannature. Lumsdaine, by contrast,offers systematic evidence that morality actually does playa significantrole in foreignaid byexamining predictions from alternative explanations and compiling extensiveevidence, both quantitative and qualitative,to arbitrateamong explana- tions.'5The same attentionto alternativeexplanations, rigorous standards of evi- dence,and social theoreticmicrofoundations has characterizedthe article-length re- searchon normsin IO on suchissues as theend of apartheid in SouthAfrica, the end of theCold War,prohibitions against certain kinds of weapons,the end of slavery, and otherprohibition regimes.'6 In a widevariety of issue areas,norms researchers have made inroadsprecisely because theyhave been able to provideexplanations substantiatedby evidencefor puzzles in internationalpolitics that other approaches hadbeen unable to explainsatisfactorily.

11. See also Ruggie,this issue. 12. Krasner1983b. 13. See Kratochwiland Ruggie 1986; Wendt 1987; Dessler 1989; Kratochwil1989; and Adler 1997. 14. Readersshould note that we use ideational"causation" here in a waythat recognizes that ideas and normsare often reasons for actions, not causes in thephysical sense of the word. See Ruggie,this issue. 15. Lumsdaine1993. 16. See Klotz 1995a;Risse-Kappen 1994; Price 1995; Ray 1989;and Nadelmann 1990. NormDynamics and Political Change 891

Normsare no easierto measuretoday than they were in the 1930s or 1960s,but conceptualprecision is essentialfor both meaningful theoretical debate and defen- sibleempirical work. In theremainder of this section we takeup threeissues where conceptualclarification seems most pressing and most possible: definitions, the rela- tionshipbetween domestic and international norms, and whether norms are agents of stabilityor change.

Definitions Thereis generalagreement on thedefinition of a normas a standardof appropriate behaviorfor actors with a givenidentity,17 but a numberof related conceptual issues stillcause confusionand debate.First, whereas constructivists in political science talka languageof norms,sociologists talk a languageof "institutions"to referto thesesame behavioralrules. Thus, elsewhere in thisissue Marchand Olsen define "institution"as "a relativelystable collection of practices and rules defining appro- priatebehavior for specific groups of actorsin specificsituations."18 One difference between"norm" and "institution"(in the sociologicalsense) is aggregation:the normdefinition isolates single standards of behavior, whereas institutions emphasize theway in whichbehavioral rules are structuredtogether and interrelate(a "collec- tionof practicesand rules"). The dangerin usingthe norm language is thatit can obscuredistinct and interrelated elements of social institutionsifnot used carefully. For example,political scientists tend to slip intodiscussions of "sovereignty"or "slavery"as ifthey were norms, when in factthey are (or were) collections of norms and themix of rules and practicesthat structure these institutions has variedsignifi- cantlyover time.19 Used carefully,however, norm language can help to steerscholars towardlooking inside social institutionsand consideringthe components of social institutionsas well as theway these elements are renegotiated into new arrangements overtime to createnew patterns of politics.20 Scholarsacross disciplines have recognized different types or categories of norms. The mostcommon distinction is betweenregulative norms, which order and con- strainbehavior, and constitutive norms, which create new actors, interests, or catego- riesof action.21 Some scholarshave also discusseda categoryof norms called evalu- ativeor prescriptive norms, but these have received much less attentionand, indeed, areoften explicitly omitted from analysis.22 This lack of attention is puzzling,since it is preciselythe prescriptive (or evaluative)quality of "oughtness"that sets norms apartfrom other kinds of rules. Because norms involve standards of "appropriate"or "proper"behavior, both the intersubjective and theevaluative dimensions are ines- capablewhen discussing norms. We onlyknow what is appropriateby referenceto

17. See Katzenstein1996b, 5; Finnemore1996a, 22; andKlotz 1995b. 18. Marchand Olsen, this issue. 19. Krasner1984, 1988, 1993; Thomson 1994; Strang 1991; Ruggie 1993; and Spruyt 1994. 20. Foran excellentdiscussion of these issues, see Jepperson1991. 21. Ruggie,this issue; Searle 1995; Katzenstein1996b; and Wendt forthcoming. 22. Gelpi 1997.See, forexample, the treatment in Katzenstein1996b, 5, fnl2. 892 InternationalOrganization thejudgments of a communityor a society.We recognizenorm-breaking behavior because it generatesdisapproval or stigmaand normconforming behavior either becauseit produces praise, or, in thecase ofa highlyinternalized norm, because it is so takenfor granted that it provokesno reactionwhatsoever.23 Thus, James Fearon arguesthat social normstake the generic form "Good peopledo (or do notdo) X in situationsA, B, C . . ." because "we typicallydo notconsider a ruleof conduct to be a social normunless a sharedmoral assessment is attachedto its observanceor non-observance.'"24 One logical corollaryto theprescriptive quality of normsis that,by definition, thereare no bad normsfrom the vantagepoint of thosewho promotethe norm. Normsmost of us would consider"bad"-norms aboutracial superiority,divine right,imperialism-were once powerful because some groups believed in theappro- priateness(that is, the "goodness") of the norm,and otherseither accepted it as obviousor inevitableor had no choicebut to acceptit. Slaveholdersand manynon- slaveholdersbelieved that slavery was appropriatebehavior; without that belief, the institutionof slaverywould not have been possible. Giventhis discussion, we canbegin to answerthe essential research question: how do we knowa normwhen we see one?We can onlyhave indirect evidence of norms just as we can onlyhave indirectevidence of mostother motivations for political action(interests or threats, for example). However, because norms by definition em- bodya qualityof "oughtness"and shared moral assessment, norms prompt justifica- tionsfor action and leave an extensivetrail of communication among actors that we can study.For example, the United States' explanations about why it feels compelled to continueusing land mines in SouthKorea reveal that it recognizesthe emerging normagainst the use of suchmines. If notfor the norm, there would be no needto mention,explain, or justify the use of minesin Korea at all. Notethat we separate normexistence or strengthfrom actual behavioral change in ouroperationalization. Becauseone central question of norms research is theeffect of norms on statebehav- ior,it is importantto operationalizea norm in a waythat is distinctfrom the state or nonstatebehavior it is designedto explain.25 Normsas sharedassessments raise the question of how many actors must share the assessmentbefore we call it a norm.In partthis is a questionof empiricaldomain. Normsmay be regional,for example, but not global. Even withina community, normsare "continuous,rather than dichotomous, entities.... [They]come in vary- ingstrengths" with different norms commanding different levels of agreement.26 We arguethat one wayto understand the dynamics of this agreement process is byexam- iningwhat we call the "life cycle" of norms.We show how agreementamong a criticalmass of actors on someemergent norm can createa tippingpoint after which

23. For a particularlygood discussionof theway in whichconventions produce judgments of social "oughtness"and morality, see Sugden1989. See also Elster1989a,c; and Sunstein 1997. 24. Fearon1997, 25, fnl8. 25. Legro 1997. 26. Ibid.,33. NormDynamics and Political Change 893 agreementbecomes widespread in manyempirical cases, and we providesome sug- gestionsabout common features of "criticalmass."

ConnectingDomestic and InternationalNorms In thisarticle we are concernedwith international or regionalnorms that set stan- dardsfor the appropriate behavior of states.27Domestic norms, however, are deeply entwinedwith the workings of international norms. Many international norms began as domesticnorms and becomeinternational through the efforts of entrepreneursof variouskinds. Women's suffrage, for example, began as a demandfor domestic change withina handfulof countriesand eventuallybecame an internationalnorm.28 In addition,international norms must always work their influence through the filter of domesticstructures and domesticnorms, which can produceimportant variations in complianceand interpretationof these norms.29 Even in situationswhere it might appearat first glance that international norms simply trump domestic norms, what we oftensee is a processby which domestic "norm entrepreneurs" advocating a minor- ityposition use internationalnorms to strengthentheir position in domesticdebates. In otherwords, there is a two-levelnorm game occurring in whichthe domestic and theinternational norm tables are increasingly linked.30 We arguelater, however, that all thesedomestic influences are strongestat theearly stage of a norm'slife cycle, and domesticinfluences lessen significantly once a normhas becomeinstitutional- ized in theinternational system. Recentwork in U.S. legal circlesalso suggeststhat there is moresimilarity in the way normsand law workdomestically and internationallythan IR scholarshave thought.IR scholarshave generallyassumed that the existence of a coercivestate able to enforcelaws madedomestic order very different from international order. A prominentgroup of legal scholarsat theUniversity of Chicago, however, now argue that,even withina domesticsetting, making successful law and policyrequires an understandingof thepervasive influence of social normsof behavior.This is a par- ticularlycompelling insight for IR scholars,since the international system is charac- terizedby law and normsoperating without direct punitive capacity. The processes throughwhich these legal scholarsclaim that norms work domestically-involving normentrepreneurs, imitation, "norm cascades," and "normbandwagons"-are en- tirelyconsistent with the research done on normsby scholarsin IR and suggestthat IR normsresearch might also learnfrom domestic analogies. For example,the nor- mativeand legal processthrough which Southern gentlemen in the UnitedStates stoppeddueling, examined by LawrenceLessig, may be relevantfor thinking about

27. For analysesof domesticnorms and theirinfluence on domesticpolitics, see Kier 1997; Johnston 1995a; Katzenstein1996a, 1993; and Berger1998. For a critiqueof thisemphasis on internationalas opposedto domesticnorms, see Checkel1998. 28. Dubois 1994. 29. See Risse-Kappen1995b; and Risse, Ropp, and Sikkink forthcoming. 30. See Putnam1988; andEvans, Jacobson, and Putnam 1993. 894 InternationalOrganization whatkinds of norms and rules could lead to a decreasein conflictin theinternational system.3'

StabilityVersus Change Macro-leveltheorizing has providedgood explanationsof theway normsproduce social orderand stability.Norms channel and regularizebehavior; they often limit therange of choiceand constrainactions.32 From a constructivistperspective, inter- nationalstructure is determinedby theinternational distribution of ideas.33Shared ideas,expectations, and beliefsabout appropriate behavior are whatgive theworld structure,order, and stability. The problem for constructivists thus becomes the same problemfacing realists-explaining change. In an ideationalinternational structure, idea shiftsand normshifts are themain vehicles for system transformation. Norm shiftsare to theideational theorist what changes in thebalance of powerare to the realist. JohnRuggie argues in thisissue that"having identified the possibility of system transformationinthe macro level, corresponding micro practices that may have trans- formativeeffects must be identifiedand inventoried."The followingsection is an attemptto identify these practices.

Evolutionand Influenceof Norms

In thissection we advancesome propositions about (1) theorigins or emergenceof internationalnorms, (2) theprocesses through which norms influence state and non- statebehavior, and (3) whichnorms will matter and under what conditions. We illus- tratethe arguments with material drawn from two major issue areas: women's rights, especiallysuffrage, and laws ofwar. International norms about women's rights often cameinto direct competition with strongly held domestic norms, and, typically, there was no self-evidentstate "interest" in thepromotion of suchnorms. Although topics relatedto gender and women have been absent from the pages of International Orga- nization,34the suffrage campaign led to theformal political participation of halfof theworld's population and therefore seems worthy of study. Laws ofwar allow us to discussthe impact of norms where we mightleast expect it-the traditionalsecurity

31. See Sunstein1997; and Lessig 1995.For an interestingjournalist's overview, see Rosen 1997. 32. See Katzenstein1996a, 3; and Sunstein1997, 40. Even Waltz,in his discussionof socialization, saysthat norms encourage conformity and that "socialization reduces variety." Waltz 1979, 76. 33. Wendt1992 and forthcoming. 34. In itsfirst fifty years International Organization has publishedonly one articleon anyissue related to genderor women,Craig Murphy's review essay on genderand internationalrelations; Murphy 1996. We suggestthat there may have been a well-internalizednorm (with a taken-for-grantedquality) that researchon genderand womendid notconstitute an appropriatetopic for international relations scholar- ship.Note that as withany well-internalized norm, this does notimply that the editors self-consciously rejectedarticles on gender-relatedtopics. To thecontrary, we knowa strongnorm is in effectwhen it does notoccur to authorsto writeon thetopic or submitarticles because it is notgenerally understood as an appropriatetopic. NormDynamics and Political Change 895 field,where such normslimit state discretion in an area perceivedas essentialto nationalsovereignty and security.

TheNorm "Life Cycle" Norminfluence may be understoodas a three-stageprocess. As shownin Figure1, thefirst stage is "normemergence"; the second stage involves broad norm accep- tance,which we term,following Cass Sunstein,35a 4normcascade"; and thethird stageinvolves internalization. The firsttwo stagesare dividedby a thresholdor "tipping"point, at whicha criticalmass of relevant state actors adopt the norm. This patternof norminfluence has beenfound independently in workon social normsin U.S. legal theory,quantitative research by sociology'sinstitutionalists or "world polity"theorists, and variousscholars of norms in IR.36The patternis importantfor researchersto understand because different social processes and logics of action may be involvedat differentstages in a norm's"life cycle." Thus,theoretical debates aboutthe degree to whichnorm-based behavior is drivenby choice or habit, specifi- cationissues about the costs of norm-violation orbenefits from norm adherence, and relatedresearch issues often turn out to hingeon thestage of thenorm's evolution one examines.Change at each stage,we argue,is characterizedby differentactors, motives,and mechanisms of influence. The characteristicmechanism of the first stage, norm emergence, is persuasionby normentrepreneurs. Norm entrepreneurs attempt to convincea criticalmass of states (normleaders) to embracenew norms. The secondstage is characterizedmore by a dynamicof imitation as thenorm leaders attempt to socializeother states to become normfollowers. The exactmotivation for this second stage where the norm "cas- cades" throughthe rest of thepopulation (in thiscase, of states)may vary, but we arguethat a combinationof pressure for conformity, desire to enhanceinternational legitimation,and thedesire of stateleaders to enhancetheir self-esteem facilitate normcascades. At the farend of the normcascade, norminternalization occurs; normsacquire a taken-for-grantedquality and are no longera matterof broad public debate.For example, few people today discuss whether women should be allowedto vote,whether slavery is useful,or whethermedical personnel should be granted immunityduring war. Completion of the "life cycle" is notan inevitableprocess. Many emergentnorms fail to reacha tippingpoint, and laterwe offerarguments aboutwhich norms are more likely to succeed.Internalized or cascading norms may eventuallybecome the prevailingstandard of appropriatenessagainst which new normsemerge and compete for support. Researchon women'ssuffrage globally provides support for the idea of thelife cycleof norms and the notion of a "tippingpoint" or threshold of normative change. Althoughmany domestic suffrage organizations were active in thenineteenth cen-

35. Sunstein1997. 36. See Sunstein1997; Meyer and Hannan1979; Bergesen1980; Thomas et al. 1987; andFinnemore 1993. 896 InternationalOrganization

Norm "Norm Internalization emergence cascade"

Stage 1 Tipping Stage2 Stage3 point FIGURE 1. Normlife cycle tury,it was notuntil 1904, when women's rights advocates founded the International Women'sSuffrage Association (IWSA), thatan internationalcampaign for suffrage was launched.In fact,rather than a singleinternational campaign for women's suf- frage,there were three or four overlapping campaigns with different degrees of coor- dination.37A quantitative analysis of the cross-national acquisition of suffragerights revealsa differentdynamic at work for early and late adopters of women's suffrage.38 Priorto a thresholdpoint in 1930,no countryhad adoptedwomen's suffrage without strongpressure from domestic suffrage organizations. Between 1890 and 1930,West- erncountries with strong national women's movements were most likely to grant femalesuffrage. Although some original norm entrepreneurs came fromthe United Statesand theUnited Kingdom, this was nota case of "hegemonicsocialization," sincethe first states to grantwomen the right to vote(New Zealand,Australia, Fin- land)were not hegemons, and the United States and the United Kingdom lagged ten to twentyyears behind. After 1930, internationaland transnationalinfluences be- comefar more important than domestic pressures for norm adoption, and countries adoptedwomen's suffrage even though they faced no domesticpressures to do so. Forwomen's suffrage, the first stage of normemergence lasted over eighty years: it tookfrom the Seneca Falls Conferencein 1848 until1930 fortwenty states to adopt women'ssuffrage. In thetwenty years that followed the tipping point, however, some forty-eightcountries adopted women's suffrage norms.39

Stage 1: Originsor emergenceof norms. Althoughlittle theoretical work has focusedexclusively on theprocess of "normbuilding," the accounts of norm origins in moststudies stress human agency, indeterminacy, chance occurrences, and favor- able events,using process tracing or genealogyas a method.40Generalizing from theseaccounts, two elements seem common in thesuccessful creation of mostnew norms:norm entrepreneurs andorganizational platforms from which entrepreneurs act.

NORMENTREPRENEURS. Norms do notappear out of thin air; they are actively built by agentshaving strong notions about appropriate or desirable behavior in their commu- nity.Prevailing norms that medical personnel and those wounded in war be treatedas neutralsand noncombatantsare clearlytraceable to theefforts of one man,a Gen-

37. See Dubois 1994;and Berkovitch 1995. 38. Ramirez,Soysal, and Shanahan 1997. 39. Ibid. 40. See Kowertand Legro 1996; andPrice 1995. NormDynamics and Political Change 897 evese Swiss bankernamed Henry Dunant. Dunant had a transformativepersonal experienceat thebattle of Solferinoin 1859 and helpedfound an organizationto promotethis cause (whatbecame theInternational Committee of the Red Cross) throughan internationaltreaty (the first Geneva Convention). The internationalcam- paignfor women's suffrage was similarlyindebted to theinitial leadership of such normentrepreneurs as Elizabeth Cady Stantonand Susan B. Anthonyin theUnited Statesand MillicentGarrett Fawcett and EmmelinePankhurst in England.Both of thesecases are consistentwith the description Ethan Nadelmann gives of "transna- tionalmoral entrepreneurs" who engagein "moralproselytism."'41 Legal theorist Lessig uses theterm "meaning managers" or "meaningarchitects" to describethe samekind of agencyin theprocess of creatingnorms and largercontexts of social meaning.42 Normentrepreneurs are critical for norm emergence because they call attentionto issuesor even "create"issues by usinglanguage that names, interprets, and drama- tizesthem. Social movementtheorists refer to thisreinterpretation orrenaming pro- cess as "framing."43The constructionof cognitiveframes is an essentialcomponent of normentrepreneurs' political strategies, since, when they are successful,the new framesresonate with broader public understandings and areadopted as newways of talkingabout and understanding issues. In constructingtheir frames, norm entrepre- neursface firmly embedded alternative norms and frames that create alternative per- ceptionsof bothappropriateness and interest.In thecase of theRed Cross,Dunant and his colleagueshad to persuademilitary commanders not to treatvaluable medi- cal personneland resources they captured as spoilsof war,to be treatedas theysaw fit.In thecase of women'ssuffrage and laterwomen's rights, norm entrepreneurs encounteredalternative norms about women's interests and theappropriate role for women.In otherwords, new normsnever enter a normativevacuum but instead emergein a highlycontested normative space wherethey must compete with other normsand perceptions of interest. Thisnormative contestation has importantimplications for our understandings of theways in whicha "logic ofappropriateness" relates to norms.44 Efforts to promote a new normtake place withinthe standards of "appropriateness"defined by prior norms.To challengeexisting logics of appropriateness,activists may need to be explicitly"inappropriate." Suffragettes chained themselves to fences,went on hun- ger strikes,broke windows of governmentbuildings, and refusedto pay taxes as waysof protesting their exclusion from political participation. Deliberately inappro- priateacts (such as organizedcivil disobedience),especially those entailing social ostracismor legal punishment, can be powerfultools for norm entrepreneurs seeking to senda messageand framean issue.Thus, at thisemergent stage of a norm'slife

41. Nadelmann1990. 42. Lessig 1995. 43. David Snow has called thisstrategic activity frame alignment-"by rendering events or occur- rencesmeaningful, frames function to organizeexperience and guide action, whether individual or collec- tive."Snow et al. 1986,464. 44. Marchand Olsen 1989,and this issue. 898 InternationalOrganization

TABLE 1. Stagesof norms

Stage 1 Stage 2 Stage 3 Norm emergence Normcascade Internalization

Actors Normentrepreneurs States, international Law, professions, withorganizational organizations, bureaucracy platforms networks

Motives Altruism,empathy, Legitimacy, Conformity ideational, reputation,esteem commitment

Dominant Persuasion Socialization, Habit, mechanisms institutionalization,institutionalization demonstration

cycle,invoking a logicof appropriatenessto explainbehavior is complicatedby the factthat standards of appropriateness are precisely what is beingcontested. Giventhe costs of inappropriate action and many of the persuasive tools they use, one has to wonderwhat could possibly motivate norm entrepreneurs (see Table 1). Obviouslythe answer varies with the norm and the entrepreneur, but for many of the socialnorms of interest to political scientists, itis verydifficult to explainthe motiva- tionsof normentrepreneurs without reference to empathy,altruism, and ideational commitment.Empathy exists when actors have thecapacity for participating in an- other'sfeelings or ideas. Such empathymay lead to empatheticinterdependence, whereactors "are interestedin thewelfare of others for its own sake, even if this has no effecton theirown material well-being or security."45 Altruism exists when actors actuallytake "action designed to benefitanother even at therisk of significantharm to theactor's own well-being."46 Kristen Monroe argues that the essence or "heart" of altruismis a "sharedperception of commonhumanity. . .. a verysimple but deeplyfelt recognition that we all sharecertain characteristics and are entitledto certainrights, merely by virtue of our common humanity."47 Ideational commitment is themain motivation when entrepreneurs promote norms or ideas because they believein theideals and valuesembodied in thenorms, even though the pursuit of thenorms may have no effecton theirwell-being. Of course,many norm entrepreneurs do notso muchact againsttheir interests as theyact in accordancewith a redefinedunderstanding of theirinterests. Suffragists,

45. See Keohane1984, chap. 7; Keohane1990a; and Mansbridge 1990. 46. Monroe1996. See also Olinerand Oliner 1988. 47. Monroe1996, 206. NormDynamics and Political Change 899 forexample, were working on behalfof a coherentconception of women'spolitical interests,but it was notan understandinginitially shared by thegreat majority of womenin theworld. Women had to be persuadedthat it was indeedin theirinterests to pursuesuffrage. Similarly, the Red Cross had to persuademilitary leaders that protectingthe wounded was compatiblewith their war aims.

ORGANIZATIONALPLATFORMS. All normpromoters at theinternational level need some kindof organizationalplatform from and throughwhich they promote their norms. Sometimesthese platforms are constructed specifically for the purpose of promoting thenorm, as aremany nongovernmental organizations (NGOs) (suchas Greenpeace, theRed Cross,and Transafrica)and thelarger transnational advocacy networks of whichthese NGOs becomea part(such as thosepromoting human rights, environ- mentalnorms, and a ban on land minesor thosethat opposed apartheid in South Africa).48Often, however, entrepreneurs work from standing international organiza- tionsthat have purposes and agendas other than simply promoting one specificnorm. Those otheragendas may shape the content of normspromoted by theorganization significantly.49The structureof the World Bank has beenamply documented to effect thekinds of development norms promulgated from that institution; its organizational structure,the professions from which it recruits,and its relationshipwith member statesand privatefinance all filterthe kinds of normsemerging from it.50 The UN, similarly,has distinctivestructural features that influence the kinds of normsit pro- mulgatesabout such mattersas decolonization,sovereignty, and humanitarianre- lief.51 The tripartitestructure of the International Labor Organization, which includes laborand businessas well as states,strongly influences the kinds of normsit pro- motesand the ways it promotes them.52 One prominentfeature of modernorganizations and an importantsource of influ- encefor international organizations in particular is theiruse ofexpertise and informa- tionto changethe behavior of otheractors. Expertise, in turn,usually resides in professionals,and a numberof empiricalstudies document the ways thatprofes- sionaltraining of bureaucrats in theseorganizations helps or blocks the promotion of new normswithin standing organizations. Peter Haas's studyof thecleanup of the Mediterraneanshows how ecologists were successful in promotingtheir norms over others'in partbecause they were able to persuadegovernments to createnew agen- cies to deal withthe cleanup and to staffthose posts with like-minded ecologists. Studiesof the World Bank similarly document a strongrole for professional training in filteringthe norms that the bank promotes. In thiscase, theinability to quantify manycosts and benefitsassociated with antipoverty and basic humanneeds norms

48. See Sikkink1993a; Keck and Sikkink1998; Klotz 1995a,b;and Price 1997. 49. See Strangand Chang 1993; Finnemore 1996a; Adler 1992; and Ikenberry and Kupchan 1990. 50. See Ascher1983; Miller-Adams 1997; Wade 1996b;and Finnemore 1996a. 51. See Barnett1995, 1997; McNeely 1995; and Weiss and Pasic 1997. 52. Strangand Chang 1993. 900 InternationalOrganization createdresistance among the many economists staffing the bank, because projects promotingthese norms could not be justifiedon thebasis of "good economics."53 Whatevertheir platform, norm entrepreneurs and theorganizations they inhabit usuallyneed to securethe support of stateactors to endorsetheir norms and make normsocialization a partof their agenda, and different organizational platforms pro- videdifferent kinds of tools for entrepreneurs todo this.54International organizations likethe UN and theWorld Bank, though not tailored to normpromotion, may have theadvantage of resources and leverage over weak or developing states they seek to convertto theirnormative convictions. Networks of NGOs and intergovernmental organizations(IGOs) dealingwith powerful states, however, are rarelyable to "co- erce" agreementto a norm-theymust persuade. They must take what is seen as naturalor appropriateand convert it intosomething perceived as wrongor inappro- priate.This process is notnecessarily or entirely in therealm of reason, though facts andinformation may be marshaledto supportclaims. Affect, empathy, and principled or moralbeliefs may also be deeplyinvolved, since the ultimate goal is notto chal- lengethe "truth" of something,but to challengewhether it is good,appropriate, and deservingof praise.55In thesecases, whatthe organizationalnetwork provides is informationand access toimportant audiences for that information, especially media anddecision makers. In mostcases, foran emergentnorm to reacha thresholdand move towardthe secondstage, it mustbecome institutionalized in specific sets of internationalrules and organizations.56Since 1948,emergent norms have increasinglybecome institu- tionalizedin internationallaw, in therules of multilateral organizations, and in bilat- eralforeign policies. Such institutionalizationcontributes strongly to thepossibility fora normcascade both by clarifying what, exactly, the norm is andwhat constitutes violation(often a matterof some disagreementamong actors) and by spellingout specificprocedures by which norm leaders coordinate disapproval and sanctionsfor normbreaking. Institutionalization ofnorms about biological and chemical weapons, forexample, has beenessential to coordinatingthe near universal sanctions on Iraq followingthe Gulf War and has enabledstates to coordinatean invasiveinspections regimeaimed at securingcompliance with those norms. Institutionalization is not a necessarycondition for a normcascade, however, and institutionalizationmay fol- low,rather than precede, the initiation of a normcascade. Women's suffrage was not institutionalizedin intemational rules or organizationsprior to thebeginning of the normcascade. The firstintergovernmental agency created to deal withwomen's is- sueswas a regionalorganization, the Inter-American Commission of Women (CIM), establishedin 1928.Although scholars locate the tipping point on women'ssuffrage around1930, the norm cascaded in similarways both in Latin America (where it was

53. See Ascher1983; Miller-Adams 1997; and Finnemore 1996a. 54. Paul Wapnerpoints out that there are exceptionsto thecentrality of the state in theseprocesses in environmentalpolitics where activists lobby polluting corporations directly to bringabout change (for example,the campaign against McDonald's clamshell containers for its sandwiches). Wapner 1996. 55. Fearon1997. 56. See Goldsteinand Keohane 1993b; and Katzenstein 1996b. NormDynamics and Political Change 901 institutionalized)and in otherplaces aroundthe world where women's rights were notsimilarly institutionalized.

TIPPING OR THRESHOLD POINTS. Afternorm entrepreneurs have persuadeda critical mass of statesto becomenorm leaders and adoptnew norms, we can say thenorm reachesa thresholdor tippingpoint. Although scholars have providedconvincing quantitativeempirical support for the idea of a normtipping point and normcas- cades,they have not yet provided a theoreticalaccount for why norm tipping occurs, norcriteria for specifying a prioriwhere, when, and how we wouldexpect it. We proposetwo tentative hypotheses about what constitutes a "critical mass" andwhen andwhere to expectnorm tipping. First, although it is notpossible to predictexactly how manystates must accept a normto "tip" theprocess, because statesare not equal whenit comesto normativeweight, empirical studies suggest that norm tip- pingrarely occurs before one-third of the total states in thesystem adopt the norm.57 In thecase of women'ssuffrage, Francisco Ramirez, Yasemin Soysal, and Suzanne Shanahanplace thethreshold point in 1930,when twenty states (or approximately one-thirdof thetotal states in thesystem at thattime) had acceptedwomen's suf- frage.58In case ofland mines, by May 1997the number of statessupporting the ban on anti-personnelland minesreached 60, or approximatelyone-third of thetotal statesin thesystem. After that point, a normcascade occurred, and 124 statesratified theOttawa land mine treaty in December1997. It also matterswhich states adopt the norm. Some statesare criticalto a norm's adoption;others are less so. Whatconstitutes a "critical state" will vary from issue to issue,but one criterionis thatcritical states are those without which the achievement ofthe substantive norm goal is compromised.Thus, in thecase ofland mines, a state thatdid not produceor use land mineswould not have been a criticalstate. By contrast,the decision in mid-1997by Franceand GreatBritain, both land mines producers,to supportthe treaty could well have contributed to thenorm cascade that happenedin late 1997.59 Securing support of those same two states was simplyessen- tialto producingthe norm cascade and near universal acceptance of the first Geneva Conventionin Europe130 yearsearlier. States may also be "critical"because they havea certainmoral stature. For example, the decision of South Africa under Nelson Mandelato supportthe land mine treaty was veryinfluential, especially with other statesin Africabut also globally.60Although cascades requiresupport from some criticalstates, unanimity among them is notessential. For example,after initially supportingthe norm of banning land mines, the United States refused to supportthe treaty,but the norm cascaded nevertheless.

57. Internationallaw has hadto wrestlewith this problem repeatedly, since many modern international normsare embodiedin treaties.Treaties implicitly recognize this concept of criticalmass by specifying thata particularnumber of countriesmust ratify for the treaty to enterinto force. Where treaties exist, theentry into force of thetreaty may be a usefulproxy for the critical mass necessary to say thata norm exists. 58. Ramirez,Soysal, and Shanahan 1997. 59. Price1998. 60. Ibid. 902 InternationalOrganization

Stage 2: Normcascades. Up to thetipping point, little normative change occurs withoutsignificant domestic movements supporting such change. After the tipping pointhas beenreached, however, a differentdynamic begins. More countriesbegin to adoptnew normsmore rapidly even without domestic pressure for such change. Empiricalstudies suggest that, at this point, often an internationalorregional demon- strationeffect or "contagion"occurs in whichinternational and transnationalnorm influencesbecome more important than domestic politics for effecting norm change.61 Contagion,however, is too passivea metaphor;we arguethat the primary mecha- nismfor promoting norm cascades is an activeprocess of international socialization intendedto inducenorm breakers to becomenorm followers.62 Kenneth Waltz sug- gestedsome of theways socializationin occurs:emulation (of heroes),praise (for behaviorthat conforms to groupnorms), and ridicule (for deviation).63 In thecontext of internationalpolitics, socialization involves diplomatic praise or censure,either bilateralor multilateral,which is reinforcedby materialsanctions and incentives. States,however, are not the only agents of socialization. Networks of norm entrepre- neursand international organizations also actas agentsof socialization by pressuring targetedactors to adopt new policies and laws and to ratify treaties and by monitoring compliancewith international standards. The InternationalCommittee of theRed Cross(ICRC) certainlydid not disappear with the signing of the first Geneva Conven- tion.Instead, the ICRC becameits chief socializing agent, helping states to teachthe newrules of war to theirsoldiers, collecting information about violations, and publi- cizingthem to pressure violators to confolrm. Socializationis thusthe dominant mechanism of a normcascade-the mechanism throughwhich norm leaders persuade others to adhere-butwhat makes socializa- tionwork? What are the motives that induce states opposed to the norm to adhere and adherequickly? We arguethat states comply with norms in stage2 forreasons that relateto theiridentities as membersof an internationalsociety. Recognition that state identityfundamentally shapes state behavior, and that state identity is, in turn, shaped by thecultural-institutional context within which states act, has been an important contributionof recentnorms research.64 James Fearon similarly argues that one's identityis as a memberof a particularsocial category,and partof thedefinition of thatcategory is thatall membersfollow certain norms.65 What happens at thetipping pointis thatenough states and enoughcritical states endorse the new norm to rede- fineappropriate behavior for the identity called "state"or somerelevant subset of states(such as a "liberal"state or a Europeanstate). To the degreethat states and stateelites fashion a politicalself or identityin relationto the international community, the concept of socialization suggests that the

61. See Ramirez,Soysal, and Shanahan 1997; and Whitehead 1996. 62. Socializationinvolves the "inductionof new members.. . intothe ways of behaviorthat are preferredin a society."Barnes, Carter, and Skidmore 1980, 35. Socializationcan thus be seenas a mecha- nismthrough which new states are induced to change their behavior by adopting those norms preferred by an internationalsociety of states. See also Risse,Ropp, and Sikkink, forthcoming. 63. Waltz1979, 75-76. 64. Katzenstein1996b. 65. Fearon1997. NormDynamics and Political Change 903 cumulativeeffect of many countries in a regionadopting new norms "may be analo- gousto 'peerpressure' among countries."66 Three possible motivations for respond- ingto such"peer pressure" are legitimation, conformity, and esteem. Scholarshave longunderstood that legitimation is importantfor states and have recognizedthe role of international sources of legitimation in shapingstate behavior. Claude,for example, described international organizations as "custodiansof the seals of internationalapproval and disapproval,"and emphasizedtheir crucial role in es- tablishingand assuringadherence to internationalnorms.67 Certainly there are costs thatcome with being labeled a "roguestate" in internationalinteractions, since this entailsloss of reputation,trust, and credibility,the presenceof whichhave been amplydocumented to contributeto Pareto-improvingeffects from interstate interac- tion.We argue,though, that states also careabout international legitimation because ithas becomean essentialcontributor to perceptionsof domestic legitimacy held by a state'sown citizens. Domestic legitimacy is thebelief that existing political institu- tionsare betterthan other alternatives and thereforedeserve obedience.68 Increas- ingly,citizens make judgments about whether their government is better than alterna- tivesby lookingat thosealternatives (in theinternational and regional arena) and by seeingwhat other people and countriessay abouttheir country. Domestic legitima- tionis obviouslyimportant because it promotescompliance with govemment rules andlaws; ruling by force alone is almostimpossible. Thus, international legitimation is importantinsofar as itreflects back on a government'sdomestic basis oflegitima- tionand consent and thus ultimately on itsability to stayin power. This dynamic was partof theexplanation for regime transitions in SouthAfrica, Latin America, and southernEurope.69 Conformityand esteemsimilarly involve evaluative relationships between states andtheir state "peers." Conformity involves what Robert Axelrod refers to as "so- cial proof"-statescomply with norms to demonstratethat they have adapted to the social environment-thatthey "belong." "By conformingto theactions of those aroundus, we fulfilla psychologicalneed to be partof a group."70Esteem is related toboth conformity and legitimacy, but it goes deeper, since it suggests that leaders of statessometimes follow norms because they want others to thinkwell of them,and theywant to think well of themselves.71 Social normsare sustained, in part, by "feel- ingsof embarrassment, anxiety, guilt, and shame that a personsuffers at the prospect ofviolating them."72 Fearon has arguedthat identity is basedon thoseaspects of the selfin whichan individualhas specialpride or fromwhich an individualgains self- esteem.73Thus, the desire to gainor defendone's prideor esteemcan explainnorm

66. Ramirez,Soysal, and Shanahan 1997. 67. Claude 1966.For morecontemporary arguments that intemational organizations continue to play thisrole, see Bamnett1997, 1995; and Barnett and Finnemore 1997. 68. Linz 1978. 69. See Klotz 1995a,b;and Whitehead 1996. 70. Axelrod1986, 1105. 71. Fearon1997. 72. Elster1989c. 73. Fearon1997, 23. 904 InternationalOrganization following.In thissense, states care about following norms associated with liberalism becausebeing "liberal states" is partof theiridentity in thesense of somethingthey takepride in orfrom which they gain self-esteem. The microfoundationsofboth the conformity and esteem arguments for individu- als arepsychological and rest on extensiveresearch on theimportance of self-esteem andconformity for individuals. The powerof conformity togroup norms is so strong in someexperimental situations that individuals will make statements that are objec- tivelywrong in orderto avoiddeviating from group judgments. In situationswhere, theobjective reality is ambiguous,individuals are even more likely to turn to "social reality"to formand evaluatetheir beliefs.74 Theories of cognitivedissonance may also provideinsights into the motivationsfor norm-conforming behavior. Studies suggestthat cognitive dissonance is arousedprimarily when people notice that their behaviorleads to aversiveconsequences that cannot be easilyrectified. The unpleas- antexperience of dissonanceleads actorsto reduceit by changingeither their atti- tudesor theirbehavior.75 We arguethat an analogto thisexists at thelevel of the state:state leaders conform to normsin orderto avoid thedisapproval aroused by normviolation and thus to enhancenational esteem (and, as a result,their own self- esteem).It is difficultto generalizeto thestate level from research on esteemdone at theindividual level, but norm entrepreneurs frequently target individual state leaders forcriticism. Because muchnorm advocacy involves pointing to discrepanciesbe- tweenwords and actions and holding actors personally responsible for averse conse- quencesof theiractions, one way to thinkabout norm entrepreneurs is that they providethe information and publicity that provoke cognitive dissonance among norm violators.In thearea of humanrights a bodyof empiricalresearch is emergingthat suggeststhat some state leaders care deeply about their international image as human rightsviolators and make significant policy changes in orderto changethat image.76

Stage 3: Internalization. At theextreme of a normcascade, norms may become so widelyaccepted that they are internalizedby actorsand achievea "taken-for- granted"quality that makes conformance with the norm almost automatic. For this reason,internalized norms can be bothextremely powerful (because behavior accord- ing to thenorm is notquestioned) and hardto discern(because actorsdo notseri- ouslyconsider or discuss whether to conform). Precisely because they are not contro- versial,however, these norms are often not the centerpiece of political debate and for thatreason tend to be ignoredby politicalscientists. Institutionalists in sociology, however,have made many of thesemost internalized norms the centerpiece of their researchprogram and have doneus theservice of problematizing and "denaturaliz- ing" manyof themost prominent Western norms that we takefor granted-such as thoseabout market exchange, sovereignty, and individualism.Instead of tryingto explainvariation in statebehavior, these scholars are puzzled by the degree of simi-

74. For a surveyof thisliterature, see Eagly and Chaiken1993; on compliancewith social norms, conformityin groups,and the normativeorigins of conformity,ibid., 630-34, 642-44; on self-esteem maintenance,ibid., 484; on ego defense,ibid., 480-81, 576-78; andon self-concept,ibid., 177-78. 75. Eaglyand Chaiken 1993. 76. See Risse,Ropp, and Sikkink forthcoming, esp. chap.by Sieglinde Granzer on Morocco. NormDynamics and Political Change 905 larityor "isomorphism"among states and societiesand how those similarities have increasedin recentyears. Their explanations for these similarities point to past norm cascades leadingto statestaking up new responsibilitiesor endowingindividuals withnew rights as a matterof course.77 Professionsoften serve as powerfuland pervasiveagents working to internalize normsamong their members. Professional training does morethan simply transfer technicalknowledge; it activelysocializes people to valuecertain things above oth- ers.Doctors are trainedto value lifeabove all else. Soldiersare trainedto sacrifice lifefor certain strategic goals. Economists, ecologists, and lawyers all carrydifferent normativebiases systematicallyinstilled by theirprofessional training. As statebu- reaucraciesand international organizations have become more and more profession- alizedover the twentieth century, we shouldexpect to see policyincreasingly reflect- ingthe normative biases of theprofessions that staff decision-making agencies.78 A numberof empiricalstudies have alreadydocumented a rolefor highly internalized normsheld by professionals determining policy. In additionto therole of economists at theWorld Bank mentioned earlier, Anne-Marie Burley's work shows a crucialrole forlegal professional norms in creating the post-World War II politicalorder, and her workwith Walter Mattli shows their importance in theEuropean Union.79 Anotherpowerful and relatedmechanism contributing to theconsolidation and universalizationof normsafter a normcascade maybe iteratedbehavior and habit. Politicalscientists have understoodthe power of thesemechanisms for years but havenot connected them theoretically to normsand socialconstruction debates. The coreof the neofunctionalist argument about integration in Europe,after all, was that frequentinteractions among people involvingjoint work on technicaltasks would ultimatelycreate predictability, stability, and habitsof trust.As trustbecame ha- bitual,it wouldbecome internalized and internalizedtrust would, in turn,change affectamong the participants. Changed affect meant changed identity and changed normsas empathyand identification with others shifted. Thus, the engine of integra- tionwas indirectand evolutionary.Diplomatic tools such as confidence-building measuresand track2 diplomacymay follow a similarlogic. Generalized, this argu- mentsuggests that routes to normativechange may be similarlyindirect and evolu- tionary:procedural changes that create new politicalprocesses can lead to gradual andinadvertent normative, ideational, and political convergence.80

WhichNorms Matter Under What Conditions? One ofthe common criticisms of norms research has been that it provides no substan- tivehypotheses about which norms will be influentialin worldpolitics and under

77. See Bergesen1980; Thomas et al. 1987; Scottand Meyer 1994; McNeely 1995; Meyer et al. 1997; andFinnemore 1996b. 78. See Haas 1989;Ascher 1983; Adler 1992; Miller-Adams1997; Finnemore1995; and Bamettand Finnemore1997. 79. See Burley1993; and Burleyand Mattli1993. These empiricalfindings are consistentwith theo- reticalarguments made by DiMaggio and Powell 1983. 80. See also Rosenau1986. 906 InternationalOrganization whatconditions they will be influential.Our review of the literature reveals a number of suchhypotheses that could be testedand elaborated in futureresearch.

Legitimation. We arguedearlier that an importantcondition for domestic receptive- nessto international norms is a needfor international legitimation. Iflegitimation is a mainmotivation for normative shifts, we mightexpect states to endorse international normsduring periods of domestic turmoil in whichthe legitimacy of elites is threat- ened.81If statesseek to enhancetheir reputation or esteem,we wouldexpect states thatare insecure about their international status or reputation to embrace new interna- tionalnorms most eagerly and thoroughly.Amy Gurowitz has argued,for example, thatJapan has beenmore open to endorsinginternational norms about refugees than has Germanybecause Japan is "insecure"about its international political leadership role.8'Dana Eyreand Mark Suchmansimilarly argue that developing states may adopthigh-tech weaponry out of statusconcerns rather than from a utilitarianwar- fightingcalculus.83

Prominence. Some domesticnorms appear more likely candidates for internation- alizationthan others. This could be due eitherto thequality of thenorm itself (dis- cussedlater) or to thequality of thestates promoting the norm. Drawing on natural selectiontheory, Ann Florini has argued that "prominence" is an importantcharacter- isticof normsthat are likelyto spreadthrough the system.84 Norms held by states widelyviewed as successfuland desirablemodels are thus likely to becomepromi- nentand diffuse. The factthat Western norms are morelikely to diffuseinternation- allywould seem to followfrom this observation. This fits the pattern of adoptionof women'ssuffrage norms, since almost all thenorm leaders were Western states (though the UnitedStates and Britainwere latecomernorm leaders, not earlyones). Jon Elster,however, suggests that prominence should be seenin cultural and economic as well as militaryterms since there are manyexamples of countrieslike Greeceand Chinathat were conquered but whose norms were assimilated by their conquerors.85

Intrinsiccharacteristics of the norm. Withinnorms research, there are several setsof claimsthat intrinsic qualities of thenorm itself determine its influence.We can dividethese claims between those stressing the formulationof thenorm (its clarityand specificity) and those stressing the substance of the norm and the issues it addresses(its content). Those stressingthe form of thenorm argue that norms that

81. See Ikenberryand Kupchan1990; and Ron 1997. 82. Gurowitz1997. 83. Eyreand Suchman 1996. 84. Florini1996. 85. Elster1989a. Some authorshave discussed"hegemonic socialization" in whichnorms will be influentialwhen they are embracedand espousedby the hegemon. Ikenberry and Kupchan1990. This is clearlyrelated to the prominence thesis but can onlybe a subsetof prominence, since states may be viewed as successfulwithout being hegemonic. NormDynamics and Political Change 907 areclear and specific, rather than ambiguous and complex,86 and those that have been aroundfor awhile, surviving numerous challenges, are more likely to be effective.87 Institutionalistsin sociology have also arguedthat norms making universalistic claims aboutwhat is good forall peoplein all places (suchas manyWestern norms) have moreexpansive potential than localized and particularisticnormative frameworks likethose in Bali describedby Clifford Geertz.88 Argumentsabout which substantive normative claims will be moreinfluential in worldpolitics have varied widely. Work by sociology's institutionalists suggests that normsabout issues congruentwith capitalism and liberalismwill be particularly powerful,but this formulation is too vague to be useful.Many norms (some of them conflicting)are congruentwith liberalism and capitalism,but only a subsetof such normshave powerfultransnational effects. John Boli and GeorgeThomas have re- finedthis observation somewhat and arguethat five principles are centralto world culture:universalism, individualism, voluntaristic authority, rational progress, and worldcitizenship. By implicationthey suggest that norms underpinned by these prin- cipleswill be moresuccessful internationally.89 Both James Lee Ray andNeta Craw- fordhave argued in a similarvein that there is a long-termtrend toward humanizing the "other,"or "moralprogress," that helps to explainboth the end of slaveryand the end of colonizationand could predictthe demiseof internationalwar in the future.90Margaret Keck and KathrynSikkink have advancedmore specific claims. Theyargue that norms involving (1) bodilyintegrity and prevention of bodilyharm forvulnerable or "innocent"groups, especially when a shortcausal chainexists betweencause andeffect, and (2) legalequality of opportunity are particularly effec- tivetransnationally and cross-culturally.91 Norm entrepreneurs must speak to aspects ofbelief systems or lifeworlds that transcend a specificcultural or political context. Althoughnotions of bodily harm are culturally interpreted, Keck andSikkink argue, theyalso resonatewith basic ideas of humandignity common to mostcultures. The notionthat norms about equality and protecting vulnerable groups from bodily harm will have moretransnational resonance than other norms explains why norm cam- paignsaround slavery and women'ssuffrage succeeded while a similar,powerful temperancecampaign organized by many of the same people failed to reach a critical massor tippingpoint. It mightalso explainwhy nolrms against drinking or smoking suddenlybecame more powerful when the effects on vulnerableor innocent bystand- ersof secondhandsmoke or fetal alcohol syndrome became more widely known. Argumentsthat the substantive content of a normdetermines whether it will be successfulimply that norm evolution has a cleardirection if not a finalendpoint and suggestthat proponents of sucharguments support some notion of "historicaleffi- ciency."It movesthese norms researchers out of the "history-dependent institution-

86. See Chayesand Chayes 1993; Legro 1997; and Franck 1992. 87. See Legro 1997;and Franck 1992. 88. See Meyer,Boli, andThomas 1987; and Geertz 1980. 89. Boli andThomas 1998. 90. See Crawford1993; andRay 1989. 91. Keck and Sikkink1998. 908 InternationalOrganization alism"box thatJames March and Johan Olsen discuss in this issue into some version of "functionalinstitutionalism." Not all researchersaccept functional arguments, however.In his workon thechemical weapons taboo, Richard Price rejects argu- mentsabout the intrinsic characteristics of theissue, arguing instead that chemical weaponsare notany moreterrible than other weapons, and yetthey are theonly weaponsthat have been thesubject of sucha prohibition.92Martha Finnemore has also emphasizedcontradictions among dominant global norms as a barrierto any sort of teleologicalargument about their effects.93 Price's recent work on theprohibition ofland mines, however, suggests that transnational norm entrepreneurs have success- fullyused graphic images of bodily harm as a meansof mobilizing a powerfultrans- nationalcampaign against these weapons.94 The speedwith which the ban on land minesmoved from norm emergence to a normcascade reinforces the idea thatnorms prohibitingbodily harm to innocentbystanders are amongthose most likely to find transnationalsupport.

ADJACENCY CLAIMS OR PATH DEPENDENCE. Therelationship of new normative claims to existingnorms may also influencethe likeliness of theirinfluence. This is most clearlytrue for norms within international law, since the power or persuasiveness of a normativeclaim in law is explicitlytied to the "fit"of thatclaim withinexisting normativeframeworks (discussed later). Robert Sugden makes a similarargument that"because prominence is largelya matterof common experience ... theconven- tionsthat are best able to spreadare those that are most susceptible to analogy.Thus we shouldfind family relationships among conventions."95 Political scientists also makearguments about adjacency, precedent, and fit. Price argues that the association ofchemical weapons with poison, which had already been prohibited, was important forsustaining the prohibition on chemicalweapons.96 Yet, as Priceand others recog- nize,the meanings of anyparticular norm and thelinkages between existing norms andemergent norms are often not obvious and mustbe activelyconstructed by pro- ponentsof new norms.Activists work hard to frametheir issues in waysthat make persuasiveconnections between existing norms and emergentnorms. Opponents of femalegenital mutilation made little headway, for example, when the practice was called "femalecircumcision," because male circumcision is oftena positivelyvalued practice.However, when they replaced the term circumcision with mutilation and campaignedunder the bannerof "violenceagainst women," the issue resonated muchmore strongly, and the campaign gained adherents. These activists clearly rec- ognizedthe power of adjacencyclaims and activelyworked to situatetheir issue in sucha wayas to makeit more difficult to dismissby tying it to thebetter-established bodyof human rights norms.97

92. Price1995, 1997. 93. Finnemore1996a, chap. 5 and 1996b. 94. Price1998. 95. Sugden1989, 93. 96. Price1995. 97. Keck andSikkink 1998. NormDynamics and Political Change 909

WORLD TIME-CONTEXT. Worldhistorical events such as warsor major depressions in theinternational system can lead to a searchfor new ideas and norms.98Ideas and normsmost associated with the losing side of a waror perceivedto have causedan economicfailure should be at particularrisk of beingdiscredited, opening the field foralternatives.99 This kindof explanationwould suggest that the end of theCold Warwould be sucha periodof majornormative growth and consolidation, based on theprinciples of the winning coalition in the"war." Notionsof "worldtime" are also presentin thearguments of thosewho see the currentperiod of globalization as onethat promotes dramatic expansion of new norms andcreates new opportunities for norm entrepreneurs. Although norms have always beena partof internationallife, changes in communicationand transportation tech- nologiesand increasing global interdependence have led to increasedconnectedness and,in a way,are leadingto thehomogenization of globalnorms. Although there is stillinadequate information to makea conclusiveargument, we suggestas an addi- tionalhypothesis that the speed of normative change has acceleratedsubstantially in thelater part of thetwentieth century. If we comparethe case of women'ssuffrage, wherenorm emergence took eighty years and the norm cascade anotherforty, to the issueof violenceagainst women, which moved from norm emergence to cascade in less thantwenty years, it appears that the speed of normative change is accelerating. The expansionof international organization (especially with the UN) is contributing tothis acceleration process by providing more opportunities to addressand negotiate on a broadrange of normative issues.

Norms,Rationality, and StrategicSocial Construction

The extensivebody of empiricalresearch on normsreveals an intimaterelationship betweennorms and rationality.However, there is littlegood theoreticaltreatment of thisrelationship, partly because scholarshave tendedto counterposenorms to rationalityin IR. The oppositionof constructivistand "rationalist"arguments thathas become widespreadin the disciplineimplies that the issues construc- tivistsstudy (norms, identities) are not rationaland, similarly,that "rationalists" cannotor do nottreat norms or identitiesin theirresearch programs.10? However, recenttheoretical work in rationalchoice and empirical work on normentrepreneurs makeit abundantly clear that this fault line is untenableboth empirically and theoreti- cally. Rationalchoice theorists have beenworking on problemsrelated to norm-based behaviorfor more that two decades now and have begunworking on identityprob-

98. Barkinand Cronin 1994. This is quitesimilar to explanationsin the "ideas" literatureabout failure, crisis,and disappointment leading to changingideas. See Odell 1982;and Hirschman 1982. 99. Hall makesthis kind of argument about shifts in economicideas. Hall 1989a. 100. Fora relatedargument, see Fearon1997, 28-29. 910 InternationalOrganization lemsas well.101 The factthat rational choice methods have beenappropriated in the pastby thosewith a materialontology has tendedto obscurethe fact that nothing aboutrational choice requires such an ontology.The utilities of actors could be speci- fiedas social or ideationalas easilyas theycan be material.By makingdifferent assumptionsabout social relationshipsand ideationalvalues, rational choice theo- ristsprovide interesting insights into the kinds of normativepatterns that may de- velop and be stable.For example,B. Douglas Bernheimascribes to actorsconcern aboutstatus and theopinions of othersand showshow differentpatterns of social conformityresult, including customs, fads, and subcultures.102 Sugden examines the evolutionof socialconventions and showshow someNash equilibriaare evolution- arilystable in iteratedgames, whereas others are not,thus yielding certain kinds of social conventions.His analysis,however, finds a largerole for"common experi- ence" in determiningthe "focal points"around which particular conventions will emerge,an argumentdovetailing nicely with constructivist arguments about the ways in whichsocial structureand normativecontext shape the actionsof agentsand, again,suggesting the need for more theorizing about the connection between strate- gic choiceand normative context.103 Similarly,empirical research on transnationalnorm entrepreneurs makes it abun- dantlyclear that these actors are extremely rational and, indeed, very sophisticated in theirmeans-ends calculations about how to achievetheir goals.104 They engage in somethingwe wouldcall "strategicsocial construction":these actors are making detailedmeans-ends calculations to maximizetheir utilities, but the utilities they wantto maximizeinvolve changing the other players' utility function in waysthat reflectthe normative commitments of thenorm entrepreneurs. The firsthalf of the processfits nicely in a rationalgame-theoretic framework. The secondhalf does not. This suggeststhat instead of opposinginstrumental rationality and social construc- tionwe needto findsome way to linkthose processes theoretically. FollowingSugden, the editors of this volume make the important observation that whatgame theoristshave called "commonknowledge" provides an openingfor conversationbetween rational choice scholarsand constructivists,and theyfurther suggesta two-stagedivision of labor between the frameworks: "Constructivists seek to understandhow preferencesare formedand knowledgegenerated, prior to the exerciseof instrumentalrationality."105 Although this suggested division of labor capturesone possibleinteraction of the two approaches, it does notexhaust the mul- tipleways in whichnorms and rationality interact. Specifically, this division ignores preciselywhat the empirical studies reveal-namely, that instrumental rationality and strategicinteraction play a significantrole in highlypoliticized social construc-

101. Akerlof1980; Jones 1984; Axelrod 1986; Elster 1989a,c; Molfow 1994b; Laitin 1995; Fearon and Laitin1996; and Fearon 1997. 102. Bernheim1994. 103. Sugden 1989. For a discussionof commonknowledge as "a pointof complementarity"across theoreticalperspectives, see Katzenstein,Keohane, and Krasner, this issue. 104. See Klotz1995a,b; Price 1997; Sikkink 1993a; Keck and Sikkink 1998; Finnemore 1996a; Thomas 1997;and Nadelmann 1990. 105. Katzenstein,Keohane, and Krasner, this issue. NormDynamics and Political Change 911 tionof norms,preferences, identities, and commonknowledge by normentrepre- neursin worldpolitics. It ignoresthe "strategic"character of social construction. One of the consistentfeatures of theempirical research reviewed here is thatthe commonknowledge (or whatconstructivists would call intersubjectiveunderstand- ings)informing actors' calculations is notstatic nor is itjust "outthere" accreted by historyand experience in someautomatic way like the prominence of Grand Central Station(in Schelling'saccount) or drivingon theleft side of theroad (in Sugden's account).106In manyof themost politically salient strategic interactions, it is pre- ciselythe changing contours of commonknowledge that are the object of the game, atleast for some sets of players. Common knowledge about who is a politicalpartici- pant(suffrage), what the rules of war are, and even who is a person(slavery) has been createdby strategicactors in highly contested processes that are central to ourunder- standingof politics. We realizethat simply pointing out thecomplexity of thisrelationship between normsand rationalityis nothelpful by itself.We also understandthat bracketing segmentsof a complexprocess like thisone can be a usefulway to understanda largerwhole. After all, constructivistsfrequently bracket structure, then agency, to understandtheir mutual constitution. Our pointis simplythat processes of social constructionand strategicbargaining are deeplyintertwined, and pendinga better theoreticaltreatment of thisrelationship we suggestthat, at a minimum,a staged analysisof therelationship between rationality and social contextcould runeither way:one couldmodel rational choice as producingsocial knowledge as easilyas one could modelsocial contextas a backgroundfor rational choice, depending on the empiricalquestion being researched. Rationalityplays a rolein virtuallyall researchon norms.Even the institutionalists in sociology,whose work is perhapsmost analytically distinct from rational choice, giverationality pride of place in theWeberian world culture that drives behavior and emphasizethe ways in whichrational action is prescribedand celebratedin that culture.107However, although all theseresearch programs recognize rationality and linknorms to rationalbehavior in importantways, they disagree strongly about the natureof thatlink. Our pointis notthat all scholarsagree about how to research norms.They do not.Rather, our point is thatthe fights are not about (or shouldnot be about)whether rationality plays a rolein norm-basedbehavior. The fightsare about thenature of thelink between rationality and norm-basedbehavior. By unpacking thatconnection between rationality and normswe can see moreclearly the issues at stakein contemporaryresearch and understand better some of the debates driving it. Fourissues stand out and create broad lines of cleavage and debate among scholars: materialism,utilitarianism, choice, and persuasion.

106. Architecturaland social historians no doubtcould provide an accountof the active construction of thesebits of common knowledge. 107. In fact,the institutionalists could make a subsumingargument: If we see somethingthat looks like rationalchoosing going on inhuman affairs, institutionalists would argue, that is in largepart an artifactof ourculture that tells us thatutility-maximization is the "right"way to act (as opposedto, for example, actionaccording to dutyor social hierarchyor divinewill. In fact,the success of rationalchoice theory withinthe social sciencesis, itself,a logicaloutgrowth of world culture, in institutionalistterms. 912 IntemationalOrganization

One problemwith claims in empiricalwork that norms are a cause ofbehavior is that,in fact,such claims do notreally tell us verymuch. There are lotsof possible reasonsto conformto a norm,and scholarsdisagree about the motivationsthey imputeto actorsin theiranalyses.108 One set of disagreementsinvolves the prefer- ences we imputeto actorsand whethernorm-based behavior can be explainedby preferencesthat are entirelymaterial. Certainly, some normconformance may be drivenby material self-interest. Neoliberal institutionalists and regimes scholars have been investigatingnorms flowing from this type of motivationfor many years.109 More recently,however, scholars have begundebating the adequacy of an entirely materialistontology and have emphasized nonmaterial and even other-regarding pref- erencesin theiranalyses of norm-basedbehavior. Note, however, that there are no obviousmethodological consequences to this particular debate over the specification ofpreferences, since rational choice can specifya utilityfunction that includes reli- gious,ideological, or altruisticconcerns, just as moreinterpretivist accounts could focuson materialrather than social facts. Note, too, that frequently heard arguments aboutwhether behavior is norm-basedor interest-basedmiss thepoint that norm conformancecan oftenbe self-interested,depending on how one specifiesinterests andthe nature of the norm. Anotherdebate among those studying norms that does, by contrast, have profound methodologicaland theoretical consequences involves the behavioral logic that schol- ars believedrives norm-conforming behavior. The behaviorallogic underlyingthe economisticand rational choice approaches to norm analysis is utilitymaximization. Actorsconstruct and conform to norms because norms help them get what they want. An alternativeapproach to understandingnorm conformance is whatMarch and Olsencall the"logic of appropriateness,"in which actors internalize roles and rules as scriptsto whichthey conform, not for instrumental reasons-to get whatthey want-butbecause they understand the behavior to be good,desirable, and appropri- ate. Habit,duty, sense of obligationand responsibilityas well as principledbelief mayall be powerfulmotivators for people and underpin significant episodes of world politics.110 The debateover behavioral logics is thefocus of bothRuggie's and Marchand Olsen's articlesin thisissue and does notrequire extended treatment here. As those articlesmake abundantlyclear, which logic one invokesto explainbehavior has significantmethodological and theoreticalconsequences. Most obviously,the two logicslead oneto differentsides of the structure-agent debate that has beenbubbling throughthe field for some time. The utilitarianand instrumental approach is entirely agent-driven.It is compatiblewith rational choice and methodologicalindividual-

108. We areindebted to Fearonfor helping us formulatethis point succinctly. 109. See Keohane1984; and Krasner 1983b. 110. On thelogic of appropriatenessand itscontrast with a "logic of consequences,"see Marchand Olsen 1989,chap. 2; and thisissue. For moreon the social psychologicalunderpinnings of scriptand schemasand the ways these may override utility-maximizing choice, see Fiskeand Taylor 1994; Nisbett andRoss 1980;Gilovich 1991; and Wilcox and Williams 1990. For a discussionof "habit-drivenactors" in worldpolitics, see Rosenau1986. NormDynamics and Political Change 913 ism,which, in turn,have long-standingrelationships with realism and liberalismin thefield of IR. The logic of appropriateness,however, necessarily has a structure- drivencomponent. What channels and directs behavior in thisview are components of social structure-normsof behavior, social institutions, and the values, roles, and rulesthey embody. Under a logicof appropriateness,notions of duty,responsibility, identity,and obligation(all social constructions)may drive behavior as well as self- interestand gain.Theoretically, this logic focusesattention on social construction processesthat are notwell explainedby IR theoriesin theircontemporary incarna- tionsand has led scholarsback into political theory writings dealing with the geneal- ogyof morality(Price on Nietzsche),the social constructionof rationality(Michael Barnettand Finnemore on Weber),the politics of communication and argumentation (Crawfordand Thomas Risse on Aristotleand Habermas),symbolic interaction and the"presentation of self" (Wendton Meade and Barnetton Goffman),and theori- ginsof individualism and humanitarianism in liberal thought. Methodologically, this grouphas invokeddifferent kinds of structuration, process tracing, text analysis, and verstehen.Unfortunately, debate in thefield on thisissue has notbeen overwhich logic appliesto whatkinds of actorsunder what circumstances. Rather, the debate has beenover whether a "logic of appropriateness"exists at all or whetherone can adequatelycapture all politicallysalient normative effects with a utilitarianand instru- mentalapproach. A thirdissue on whichthere is disagreementand, we think,some misunderstand- ing,is therole of choice and its converse, determinism, in norm-based behavior. For rationalchoice scholars, actors conform to norms(not surprisingly) out of choice; choicesmay be constrained,at timeshighly constrained, but the focus of the analysis is on thechoice. Other researchers, however, focus almost exclusively on theway normsare "internalized"in actors,which raises important issues about the locus of causalityin normsarguments. When norms become internalized in actors,actors are no longerchoosing to conformto themin anymeaningful way. For institutionalists in sociology,many norms, including some of themost powerful ones, have been so internalizedthat we no longthink seriously about alternative behaviors. In thisview, actorsno longerthink seriously about whether "the state" is thebest or most efficient formof political organization (it almostcertainly is not).They just setup moreand morestates to theexclusion of otherpolitical forms. They no longerthink seriously aboutwhether international institutions are the best way to solveinternational prob- lems (again,a mixedbag). Theyjust set up moreand moreinternational institu- tions.111Institutionalists might argue that "choice" is nota particularlyrobust ana- lytictool, since muchof the mostbasic humanbehavior is not chosen in any meaningfulsense. Rather, it is suppliedto agentsby thelarger social and cultural environment.Social structure,not agent choice, institutionalists would argue, is where thereal explanatory action lies. In thisunderstanding, norm conformance driven by a logic of appropriateness startsto look deterministic.Yet, as we surveythe norms research that emphasizes

111. Barnettand Finnemore 1997. 914 InternationalOrganization appropriatenesslogic in IR, verylittle of it looks deterministic.Indeed, most of it emphasizesthe highlycontingent and contestednature of normativechange and normativeinfluence. We see tworeasons for this. First, IR scholarsapplying a logic ofappropriateness in their analysis have never been imperialistic in theirclaims. The argumenthas neverbeen thatother logics of actiondo notever drive behavior; the argumenthas beenthat appropriateness is a powerfuland importantmotor of politi- cal behaviorworthy of investigation.Second, and moreimportant, even withina logicof appropriateness there is oftensubstantial room for agent choice. Actors may facevaried and conflicting rules and norms all makingclaims for different courses of action.Indeed, most significant political choices are significantand difficultpre- ciselybecause they involve two or moreconflicting claims for action on a decision maker.Actors must choose which rules or normsto followand whichobligations to meetat the expense of others in a givensituation, and doing so mayinvolve sophisti- cated reasoningprocesses. These processes,however, involve a differentkind of reasoningthan that of utility maximization. Actors may ask themselves,"What kind of situationis this?"and "Whatam I supposedto do now?" ratherthan "How do I getwhat I want?"Actors often must choose between very different duties, obliga- tions,rights, and responsibilitieswith huge social consequences,but understanding thechoice depends on an understanding,not of utilitymaximization, but of social normsand rules that structure that choice."12 A finalissue that separates different approaches to normsresearch is whetherand howthey treat persuasion. Persuasion is centralto mostof the empirical case studies aboutnormative influence and change.It is themission of normentrepreneurs: they seek to changethe utility functions of otherplayers to reflectsome new normative commitment.Persuasion is theprocess by whichagent action becomes social struc- ture,ideas become norms, and the subjective becomes the intersubjective. It is essen- tialto muchof theprocess tracing that scholars are currentlydoing, yet we have no good way of treatingit theoretically.Rational choice can modelthe ways in which transfersof information may change actor strategies, but changes in underlying pref- erencescannot be endogenized.Institutionalists in sociology can talkabout applying culturalmodels of action to new situations and the consequent unfolding or elaborat- ingof globalculture, but the process by whichsome models appear compelling and othersdo not,why some cultural innovations are persuasive and others are not, is not wellexplained. These approaches thus gloss over this problem in differentways, but persuasionis centralto politics of all kinds,and we needa goodtheoretical apparatus forunderstanding it. Lookingto otherdisciplines, we see twotypes of arguments about persuasion that mightbe usefulin politicalscience: one structuraland logical, the other psychologi- cal andaffective. International legal scholars draw on a complexstructure of rules to craftarguments about and adjudicateamong competing normative claims. The per- suasivenessof a normativeclaim in law is explicitlytied to the "fit"of thatclaim

112. Marchand Olsen emphasizechoice withinappropriateness much more than do Meyerand his colleaguesin theirwork. March and Olsen 1989. NormDynamics and Political Change 915 withinexisting normative frameworks; legal arguments are persuasive when they are groundedin precedent, and there are complex rules about the creation of precedent- suchas whichjudgments trump which and how theaccretion of judgments is to be aggregatedover time.Since normativecontestation in law is so explicitand well documented,and since much of contemporary norm politics in the world has a strong legalcomponent, we believean examinationof legal mechanisms for norm selection anddissemination will be instructivefor IR scholars."13 Approachesin psychologyemphasize very different factors in bringing about per- suasion.In thesearguments both cognition and affect work synergistically toproduce changesin attitudes, beliefs, and preferences. Most of this work stresses communica- tiveprocesses that happen through argumentation, but, unlike the legal approach, logicalone does notdictate the result, since appeals to emotionmay well be usedto strengthenor underminelogical extensionsof norms."14German IR scholars,for example,are beginning to applyHabermas's theory of communicative action (which requiresempathy) as a meansof understandingpersuasion, accommodation, and arrivalat mutualunderstandings in international politics. '15 Debatesover these four issues-materialism, utilitarianism, choice, and persua- sion-will continueto shape the ways we understandnormative influence and norma- tivechange in worldpolitics. Our concern here has notbeen to settlethese debates, butto clarifywhat, exactly, is beingdebated and whatthe stakes of thedebates are. These are notdebates about rationality, although rationality certainly plays a rolein all of them.The debatesalso do notdivide norms researchers into two tidy camps. Researchersmay marry ideational ontologies with rational choice; " 16they may exam- ine reasonedchoice among conflicting "appropriate" behaviors;"I7 they may exam- ine highlyinstrumental and strategicinteractions designed to constructnew stan- dardsof appropriateness,as most studies of normentrepreneurs do;"18 and they may generallyfind themselves cross-cut in waysthat are refreshing and, we hope,stimu- latenew kinds of conversations.

Conclusions

The "return"to norms holds immense promise for shaking up theIR researchagenda andopening up excitingnew avenues for inquiry-and not just because it offers new (or previouslyforgotten) subject matter to study.More interesting,to ourminds, is theway normsresearch opens up conversationswith theoretical traditions that IR scholarshave ignored in recentdecades. The evaluativeand prescriptive character of

113. See Franck1990; Burley and Mattli 1993; and Burley 1993. 114. See Eaglyand Chaiken 1993; and Chaiken, Wood, and Eagly 1996. 115. Risse 1997. 116. See Laitin1995; Fearon and Laitin 1996; and Fearon 1997. 117. Marchand Olsen 1989. 118. See Klotz1995a,b; Price 1997; Sikkink 1993a; Keck and Sikkink 1998; Finnemore 1996a; Thomas 1997.and Nadelmann 1990. 916 InternationalOrganization normsopens the way for a long-overdueconversation with political theory and eth- ics. For decades now IR researchhas been divorcedfrom political theory on the grounds(implicitly, ifnot explicitly, articulated) that what "is" inthe world and what "oughtto be" arevery different and must be keptseparate, both intellectually and in policy.However, contemporary empirical research on normsis aimedprecisely at showinghow the "ought"becomes the "is." Empiricalresearch documents again andagain how people's ideas aboutwhat is goodand what "should be" in theworld becometranslated into political reality. People withprincipled commitments have made significantchanges in thepolitical landscape: slavery as a legal institutionof propertyrights has been abolishedeverywhere on theplanet for the firsttime in humanhistory; women, more than half the world's population, have fullformal po- liticalparticipation in moststates of theworld; and thoughwar continuesto be a horriblehuman practice, there is no doubtthat it is less horribleas a resultof efforts by humanitariansto curbthe most awful weapons and practices.At thesame time, principledcommitments and notionsof what"should be" have fueledxenophobic nationalism,fascism, and ethnic cleansing. Understanding where these "oughtness" claimscome from, how they are related to one another,which ones will be powerful, andwhat their implications are for world politics is vital,but it is an inquirythat can onlybe undertakenat thenexus of political theory and IR. Similarconnections exist between norms research and other fields of study.Inter- nationallaw, like philosophy and ethics, has been ignored by IR scholarsfor decades, yetcustomary international law is norms,and empiricalresearch in IR is, again, demonstratingthat these legal norms have powerful behavioral effects. Legal norms arealso boundup inextricablywith the workings of international institutions, which havebeen a centralfocus of virtually all typesof IR researchin recent years. Further, theselegal normsare structuredand channelbehavior in waysthat create precisely thetypes of patterns political scientists seek to explain.Understanding which norms will becomelaw ("soft" law as well as "hard"law) and how,exactly, compliance withthose laws comesabout would seem, again, to be a crucialtopic of inquiry that lies at thenexus of law andIR. Microfoundationsfor norm-based behavior might be improvedby payingmore attentionto studiesin psychology,particularly work on theroles of affect,empathy, conformity,and esteem.Like law and philosophy,affect and empathyhave been sweptunder the carpet in recentdecades. Ideational phenomena have been treated as "information,"which reduces uncertainty or providesnew strategiesto maximize utilities.The resultis politicswithout passion or principles, which is hardlythe poli- ticsof the world in whichwe live.Emotions can be politicallydangerous and unde- sirablein politics;hate, after all, is affect,too. But to pretend that affect and empathy do notexist is to missfundamental dynamics of politicallife, and we have triedto suggesthow attempts in psychologyto wrestlewith these issues may be helpful. Finally,we havetried to showhow normsresearch cross-cuts our own discipline in waysthat are refreshingand stimulatenew conversations.Contrary to whatwe perceivedto be thepopular impression, rationality is notthe issue thatdivides IR scholars.Rational choice theoristscan and do have a greatdeal to say aboutide- NormDynamics and Political Change 917 ationalphenomena and how norms work, just as empiricalstudies of social construc- tionand norm emergence repeatedly reveal highly rational strategic interaction. Schol- ars are divided,however, about the role that choice plays in norm-basedbehavior, aboutwhat motivates choice, and aboutthe role persuasion plays in normativepro- cesses andhow to treatit. No schoolof thought in thediscipline is entirelycomfort- able withits answers to these questions at the moment. We believe this is a goodstate ofaffairs, one thatwill encourage scholars to venture beyond narrow methodological commitmentsto thinkmore broadly about these issues.