You are on page 1of 37

The Agent-Structure Problem in International Relations Theory Author(s): Alexander E. Wendt Source: International Organization, Vol. 41, No.

3 (Summer, 1987), pp. 335-370 Published by: The MIT Press Stable URL: http://www.jstor.org/stable/2706749 . Accessed: 04/06/2013 01:41
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

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

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

http://www.jstor.org

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

The agent-structure problem in international relations theory


E. Wendt Alexander

Two theories, neorealism and world-system theory, strongly influence contemporary academicdiscourse aboutinternational relations. Bothclaimto provide "structural" explanations ofhowstates behaveintheinternational system. Despitetheir common commitment to structural analysis, however, their understanding ofsystem "structure," andtherefore ofstructural explanation, is quitedifferent. Neorealists define international system structures in terms oftheobservable attributes oftheir states(the"distribumember tionofcapabilities"), and as a result, they understand theexplanatory role ofthosestructures in individualist terms thechoicesofpreas constraining stateactors. existing interWorld-system theorists, on theother hand, define interms national system structures ofthefundamental organizing principles of thecapitalist worldeconomy and constitute whichunderlie states,and thus they understand theexplanatory roleofstructures instructuralist terms as generating stateactorsthemselves. These differences, and their implicahave yetto be explicated in theinternational tions, In relations literature.1
of theInternational at the 1986meeting was presented of thisarticle version An earlier Jeffrey Duvall, Raymond Richard Ashley, Alker, Hayward I want tothank Studies Association. Weldes,and two PeterManicas,David Sylvan,Jutta Isaac, BrianJob,StephenKrasner, drafts. on previous and suggestions comments helpful fortheir referees anonymous in theory" and uses of "structural of themeanings of discussions 1. Thereare a number ordiffernoneexplicitly compares butas faras I know, theory, neorealism andworld-system On analysis. and structural to structure approaches and world-system entiates theneorealist Mass.: (Reading, Politics ofInternational Waltz, Theory Kenneth see,for example, neorealism Realism Structural Politics: ofWorld Keohane,"Theory and Robert 1979), Addison-Wesley, Ashis Richard ofstructure conception ofneorealism's D.C.: APSA, 1983).The bestcritique pp. 225-86. 1984), 38 (Spring Organization International ofNeorealism," ley's "The Poverty DemiseoftheWorld "The RiseandFuture Wallerstein, see Immanuel On world-system theory and in Society Studies forComparative Analysis," Comparative Concepts System: Capitalist Rubinson, andRichard Chase-Dunn andChristopher pp. 387-415, 1974), 16(September History 7 (no. 4, 1977), and Society Politics on theWorld-System," Perspective "Towarda Structural in this thatcomesclosestto myconcerns theory of world-system pp. 453-76. The critique and A Theoretical World System: Capitalist ThedaSkocpol's"Wallerstein's is probably article 82 (March1977), Journal American Historical pp. 1075-90. ofSociology Critique,"
InternationalOrganization41, 3, Summer 1987 ed., Political Science: The State of theDiscipline (Washington, and Beyond," in Ada Finifter,

ofTechnology Institute andtheMassachusetts Peace Foundation C) 1987bytheWorld

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Organization 336 International I wantto beginto clarify and contrast thenature of structural thisarticle, however, is to interest, Myprimary ineach ofthesetwotraditions. analysis and to theconceptions of structural found in each ofthem, theory critique to structural thedevelopment ofa newapproach use this tomotivate critique from thework of"structurarelations adapted aboutinternational theorizing agendaitimin sociology.2 and theresearch Thisapproach tiontheorists" in realistphilosophy of science (or plies, in turn,requirea foundation of in thephilosophy the "new orthodoxy" arguably "scientific realism"3), scientists.4 bypolitical unacknowledged natural science,butas yetlargely and worldneorealism relations, theories of international As structural ofdimenandthus be compared, alonga number might theory differ, system among power,scope,and parsimony, sions:substantive claims,predictive strongly are important, theyare, I think, others.Whilethesedifferences emneorealism difference ofontology: conditioned by a morefundamental whileworld-system embodiesa theory bodies an individualist ontology, of this and implications the nature holistic one. A usefulway to capture asin terms of their underlying difference is to evaluatethetwo theories structures to humanagents. about the relationship of system sumptions than"agentic"theorizing, rather commitment to "structural" Despitetheir is being some theory ofwhat likeall structural both presuppose theories they
2. The term "structuration theory" is sometimes narrowly identified with thework ofAnthony Giddens, who has articulated its basic problematic in his Central Problems in Social lineoftheTheory ofStructuration (Cambridge, U.K.: Polity Press,1984). In "On theDetermination of Social Action in Space and Time," Society and Space 1 (March1983), pp. 23-57, however, NigelThrift uses the term morebroadly as a generic label fora groupof social theories which sharecertain fundamental assumptions abouttheagent-structure relationship; thisgroupincludes, butis notlimited to, Pierre Bourdieu, Outline of a Theory of Practice (Cambridge: Cambridge University Press,1977), Roy Bhaskar, ThePossibility ofNaturalism (Brighton, U.K.: Harvester Press,1979), andDerekLayder, Structure, Interaction, andSocial Theory (London:Routledge & KeganPaul, 1981).Since mypurpose in thispaperis less to advanceGiddens's ideas (indeed, I willrelymoreon Bhaskar thanGiddens) than to demonstrate therelevance oftheoverall problematic forinternational relations theory, I shallfollow Thrift's moreinclusive use oftheterm. 3. Scientific realism (or simply "realism")is notrelated to political realism orneorealism in international relations. of natural 4. Whether in thephilosophy or notscientific realism is the "new orthodoxy" butithasinanycase science is undoubtedly a contentious issueamong realists andempiricists, forthe Philosophy made sufficient inroadsthatthe Minnesota Center of Science,longan in 1985/86 bastionof empiricism, helda year-long institute important which, amongother seemto be American scientists things, focused on that explicitly question. political generally for science. oforuninterested inthis anditspotential unaware debate implications To political are British: in international relations theonlydiscussions of scientific realism myknowledge, andtheStudy ofInternational ofChange John Maclean,"Marxist Epistemology, Explanations in Barry BuzanandR. J.Barry Relations," Jones, eds., ChangeintheStudy ofInternational Relations:The Evaded Dimension (London:FrancesPinter, 1981),pp. 46-67, and Richard American Perspectives (Oxford: Blackwell, 1985), pp. 79-91.
Little, "The Systems Approach," in Steve Smith, ed., InternationalRelations: Britishand Theory(Berkeley: University of CaliforniaPress, 1979) and The Constitution of Society: Out-

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Agent-structure problem 337 human or organizational structured, agents, and oftheir relationship to soPutmore cial structures. generally, all socialscientific theories embody an at least implicit solutionto the "agent-structure problem,"whichsituates and socialstructures inrelation agents to one another. These solutions help determine a theory's understanding of,and therelative explanatory importance it attachesto, structural analysis.Whilegenerating verydifferent understandings ofstructural theory, however, I shallargue that theneorealistandworld-system solutions totheagent-structure problem are,inat least one respect, and that verysimilar, thissimilarity createsa commonfundamental weaknessin thesetheories as "structural" approaches to international relations. Structuration theory, inturn, is a response to thiscommon weakness whichbothsubsumes and points beyond neorealism and worldsystem theory. In Section1, I examine thenature oftheagent-structure "problem"and theprincipal inSection briefly identify kinds ofsolutions to it.I argue 2 that and world-system neorealism theory embody two of thesesolutions, the methodological individualist and structuralist ones, respectively. Despite important differences between them, each of theseapproaches solves the agent-structure problem by making either stateagents or system structures units. The resulting and worldontologically primitive effect on neorealism is an inability theproperties of system theory to explain and causalpowers their units ofanalysis, undermines their a weakness primary which seriously of stateaction.This situation potential explanations can be prevented by an approachto theagent-structure adopting problem whichdoes notpreclude a priori bothagentsand structures or "demaking "problematic" pendentvariables." In Section 3, I describethis third, structurationist and its foundations in realistphilosophy of science. Since the approach, forinternautility of structuration theory as a meta-theoretical framework tional relationsultimately depends on its abilityto enrichsubstantive andconcrete be convincingly theorizing empirical research, itsvaluecannot in a programmatic demonstrated articlesuch as thisone. It is possible, however, to indicate someofthechanges which a structurationist perspectivesuggests inthecontemporary are necessary research agendaininternaTowardsthisend, in Section4, I examinesome general tionalrelations. and theoretical of structuration forthe epistemological implications theory I return ofstateaction.In theconclusion, explanation to someimplications of scientific realism forsocial scientific research. 1. The agent-structure problem The agent-structure problem has itsorigins in twotruisms aboutsociallife whichunderlie most social scientific inquiry: 1) humanbeingsand their

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Organization 338 International helpreproduce or transare purposeful actorswhoseactions organizations the societyin whichtheylive; and 2) societyis made up of social form acthesepurposeful between theinteractions structure which relationships, thathumanagentsand social suggest thesetruisms tors.Taken together or mutuinterdependent theoretically are,inone wayor another, structures Thus, the analysisof actioninvokesan at least entities. ally implicating (or "rules of the of particular social relationships understanding implicit of social structures theactionis set-just as theanalysis game") in which makeup the oftheactorswhoserelationships invokes someunderstanding that of It is then theproperties stepto believe a plausible structural context. of are bothrelevant to explanations agentsand thoseof social structures and invery different Andinfact, although socialbehavior. ways,neorealism interests) (powers, do use theproperties bothofstates theory world-system to explain relations exchange) ofunequal structures (polarity, and ofsystem The "problem"withall thisis thatwe lack a self-evident statebehavior. Thisabsenceofa relationship. andtheir theseentities wayto conceptualize relation of the agent-structure conception immediately compelling single, acrossthe of therelationship of conceptualizations has spawneda variety and practical philosophical theparticular each reflecting social sciences,5 of the of itsparent theoretical discourse. (My own adoption commitments is nottheory-neutral.) therefore, of "agents" and "structures," language "partshowever, the "agent-structure," Despite theirmanydifferences, thesame all reflect problems and "micro-macro" whole,""actor-system," need to adopt,forthe purposeof eximperative-the meta-theoretical and exoftheontological someconceptualization socialbehavior, plaining
ofa as something emerged recently guises, problem has, in various 5. The agent-structure see ingeography, ofthis work: Fora sampling thesocialsciences. throughout industry cottage of Transactions of theInstitute and HumanGeography," "HumanAgency DerekGregory, andJohn eds., Social Urry, pp. 1-18,andDerekGregory 6 (no. 1, 1981), Geographers British
Relations and Spatial Structures(London: MacMillan, 1985); in sociology, in additionto the Tom Bottomore and Robert Nisbet, eds., A History of Sociological Analysis (London:

ofSocialAction,"in see AlanDawe, "Theories cited, already and Bhaskar ofGiddens work and AaronCicourel, eds., Advancesin Social 1979),and KarinKnorr-Cetina Heinemann, UniverHistorical (Ithaca:Cornell see Philip Sociology Abrams, in socialhistory, Paul,1981); Blackwell, in Social History (Oxford: Lloyd,Explanation sityPress,1982),and Christopher and see John O'Neill,ed., ModesofIndividualism ofsocialscience, 1986);in thephilosophy TheMetaphysics ofthe andDavid-Hillel Ruben, 1973), (New York:St. Martins, Collectivism Thompsee Edward inMarxist theory, & KeganPaul,1985); (London:Routledge Social World debate on this andthecommentaries ofLouis Althusser, Marxism thestructural against 1978) Marxism (London:Verso, 1980),and Nicos within English Arguments Anderson, by Perry andininternaTelos45 (Fall 1980), pp. 173-85; Theory," inMarxist "Reductionism Mouzelis, Rosenau,"Before Politics,and James of International see Waltz,Theory tionalrelations, International inWorld Actors Politics," andHabit-Driven Regimes, Hegemons, Cooperation: pp. 849-94. 40 (Autumn 1986), Organization
son's polemic in The Povertyof Theoryand OtherEssays (New York: MonthlyReview Press, of Micro and Macro-Sociologies (London: Routledge& Kegan Theory:Toward an Integration

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

problem 339 Agent-structure (inthiscase, states)6 planatory relationship between socialactorsor agents system). and societalstructures (in thiscase, theinternational problems, one onThe agent-structure problem is really twointerrelated and the otherepistemological. and morefundamental, tological The first, issue concerns thenature ofbothagents and structures and,because they are in some way mutually interrelationship. In other implicating, of their are words, what kind ofentities arethese(or,inthecase ofsocialstructures, Thereare two basic theyentities at all?), and how are theyinterrelated? ofanalysis ontologically waysto approach thisquestion: bymaking oneunit irreducible ontological primitive, or by giving themequal and therefore theseapproaches genstatus. Depending on which entity is madeprimitive, eratethree whichI willcall possibleanswersto theontological question, Neorealism and worldindividualism, structuralism, and structurationism. both thefirst of positions, system theory embody, respectively, twoofthese Thus,neorealists which ultimately reduceone unit ofanalysis to theother. andinteractions of reduce thestructure ofthestatesystem to theproperties theorists reducestate its constituent elements, states,whileworld-system requirements of thecapi(and class) agentsto effects of thereproduction on theother hand,tries talist worldsystem. The structurationist approach, ofindividualism to avoidwhatI shallargueare thenegative consequences and structuralism status. equal ontological by giving agents and structures Farfrom worlds," however, being a mindless synthesis ofthe"bestofboth thestructuration conceptualization ofthe project requires a very particular the us to rethink forces agent-structure relationship. Thisconceptualization In turn, it fundamental structures. properties of (state)agentsand system permits us touse agents andstructures toexplain someofthekeyproperties as "coof each as effects of the other,to see agentsand structures determined" or "mutually constituted" entities. issues theseontological The manner in whicha social theory addresses conditions itsapproach to theepistemological aspectoftheagent-structure within ofexplanations thechoiceandintegration problem, ofdifferent types theories ofsocialbehavior. raisestwoepistemological Thisproblem actually issues. The first is the choice of the form of explanation corresponding on the to agentsand structures. respectively This choicedependslargely havebeendeemed kinds ofproperties and structures that ofagents causally
6. Recent theoretical work has conceptualized thestate both as an agent andas a structure; see,for example, Roger Benjamin andRaymond Duvall,"The Capitalist StateinContext," in Roger Benjamin andStephen Elkin, eds., TheDemocratic State(Lawrence, Kans.: University that ofKansasPress,1985), pp. 19-57.Forpurposes ofthis paper, I assume with neorealists the state is an agent, a movewhich can bejustified inpart becausetheorganizing principles ofthe state for their system constitute states as individual choice-making units which areresponsible constitute states actions. My subsequent arguments aboutthewayinwhich system structures as agents should not,however, be seenas excluding a conception ofthestate as a structure of political authority in which governmental agents are inturn embedded.

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Organization 340 International bethatconceiveof human to social inquiry Thus,approaches significant. choice theory, subjects,such as rational goal-directed ingsas reflective, "interpretive"thatare, broadlyspeaking, agent-explanations generate ofagents. and self-understandings ofthegoals,beliefs, that is, castinterms thatconceiveof human beingsas nothing On theother hand,approaches stimuli-suchas behaviorismprocessing morethancomplexorganisms causal in form. thatare moremechanistically agent-explanations generate Social explanations. withrespectto "structural" The situation is similar of individuals to the properties that reduce systemstructures theories as one of constrainrole of structures the explanatory usuallyconstrue system while conceptualize thosethat agents, ingthechoicesofpre-existing understand agentstypically entities underlying as irreducible structures The second agentsthemselves. structures as generating or explaining of agent-explaissue concernsthe relativeimportance epistemological This type, in socialtheory. ofwhatever and structure-explanations, nations and in thisarticlebecause neorealists importance issue is of secondary theory relations an adequateinternational agreethat theorists world-system thisrequireTheyunderstand thanagent-oriented. must be morestructurewhichI willlater in verydifferent ment ways,however-a disagreement dimension of the theontological to theway theyapproach showis linked problem. agent-structure 2. Reductionism and reification in international relationstheory theory of"structural" In thissection I wantto: 1) compare theconceptions and 2) show that,despite and world-system theory; foundin neorealism to the approach theseconceptions sharea common differences, important of thisapproach an explanation andthat precludes problem, agent-structure units.This inability primitive the essentialproperties of theirrespective fountheoretical that arewithout units aboutprimitive leadsto assumptions ofstate inturn explanations undermines thetheories' a movewhich dation, limitation in theinternational I shallarguethat thiscommon system. action logicof each theory's and internal is a function of thebasic assumptions cannot and thattheytherefore problem, approachto the agent-structure and epistemological it within theterms oftheir basic ontological overcome commitments.
a. Neorealism

and antihave strong structural On the surface,at least, neorealists and of systemic the nature of In his discussion reductionist commitments.

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Agent-structure problem 341 reductionist theories, Kenneth Waltzdefines the latter as theories which explain theforeign policy behavior ofstates exclusively interms ofcausesat thenational levelofanalysis.7 is Lenin'stheory ofimperialism, for example, reductionist because it explainsexpansionist behavior in terms of theaccumulation Waltzcriticizes dynamics of national capitalism. suchtheories, I think, forignoring theintervening correctly role playedby international in the translation structures intoforeign system of domestic imperatives behavior. the policy Neorealists avoidthis type ofreductionism byadopting systemic (but, I shall argue,not "structural") logic and conceptual apparatus of micro-economic theory.8 This movepermits neorealists to intewithin a coherent grate thestate-centric of theoretical framework approach classicalpolitical realism withthe systemic approach of international systemstheory, and thusto developa conception relaoftheagent-structure in international tionship thecausal roleofboth relations which recognizes and system stateagents structures. In viewofneorealists' desire to avoidmicro-level reductionism, however, itis ironic that their solution to theagent-structure problem is, ina different anddeepersense,reductionist. The kindof "reductionism" which neorealistsopposeis defined as theory which tries to explain behavior in terms of strictly agent-level properties. ofwhat be calledexplanThisrejection might atoryreductionism does notinitself, however, impose anyparticular restrictionon theontological issue of how system structures shouldbe defined, sincean opposition toagent-level of explanations is analytically independent how system structures, once recognized as causallysignificant, shouldbe theorized. Thus,neorealists' individualist definition ofthestructure ofthe international system as reducible to theproperties of states9-tothedistribution of capabilities-isperfectly withtheimportant rolethat consistent
Princeton Politics 8. Robert Gilpin, Warand Changein World (Princeton: University Press, StudiesQuarterly 27 1981);Richard Ashley,"ThreeModes of Economism," International
7. Waltz, Theoryof International Politics, p. 18.

PoliticalEconomy the World (Princeton: Princeton University Press,1984);DuncanSnidal, 9. Ashley thoroughly critiques theindividualist (andempiricist) foundations oftheneorealist conception of international system structure in his "Poverty of Neorealism," especially pp. 238-42.Itis important tokeepinmind, however, that inTheory ofInternational Politics, Waltz starts outwith three defining features ofpolitical structures: 1)theprinciple according towhich they areorganized, 2) thedifferentiation ofunits andtheir functions, and3) thedistribution of acrossunits. Thisdefinition can be usedto support a generative to struccapabilities approach tural theorizing, as John Ruggie showsin hisDurkheimian reconstruction ofWaltzin "Continuity andTransformation intheWorld Polity: Toward a Neorealist Synthesis," World Politics 35 (January 1983),pp. 261-85.Despitethispromising beginning, however, Waltzand other neorealists argue that thefirst twofeatures ofthis definition don'tapply tointernational political us inpractice structures, leaving with an individualist conception ofstructure as thedistribution ofcapabilities. links to For an argument that thisresult to a lingering neorealist commitment see Little, "The Systems positivism, Approach."
"The Game Theoryof International Politics," WorldPolitics 38 (October 1985), pp. 25-57.

(December 1983),pp. 463-96; RobertKeohane, After Hegemony: Cooperationand Discord in

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Organization 342 International Indeed, behavior. ofstate explanations playin neorealist structures system likemicroneorealism, versions'0 and game-theoretic in bothits decisionby a modelof determinism," by "situational is characterized economics, by the or evendetermined is conditioned behavior rational actionin which engagein The point is notthatneorealists of choicesituations.11 structure definition that their do not), butrather they (which reductionism explanatory Thisdefinireductionism. is characterized byontological structure ofsystem the as onlyconstraining structures ofsystem tionleads to an understanding and structurarather than, as in world-system states, ofpreexisting agency inevitably Thisfollows themselves. as generatingstateagents tiontheory, system structure: of system definition reductionist its (ontologically) from of cannotgenerate agentsiftheyare reducedto theproperties structures neorealfocus,then, in thefirst systemic place. Despiteitsstrongly agents is decidedly stateor structures roleofsystem ism'sviewoftheexplanatory inwhich in themanner they appear It sees system structures agent-centric. thanas actions-rather on their constraints to states-as given,external forstateaction. ofpossibility conditions themost important theory, relations theperspective ofinternational From probtotheagent-structure approach individualist ofneorealism's weakness ofthe theory an explicit a basisfordeveloping itfailsto provide lemis that abouta state.It is nothardto see whythismustbe thecase. Theorizing oftwoforms likethestate, can takeeither kind ofindividual unit, particular in is to explaintheindividual "reductionist" strategy (or both).The first, Whilethisapproachmay structure.12 organizational of its internal terms it neglects theirof theindividual, explainsome of the causal properties Thus,while predicates.'3 individual-level of many social content reducibly somedisposimayexplain structure ofa capitalist theinternal physiological as that ofa "capiwe cannot explain hisor herbehavior tionsand actions,
Press,1981); The WarTrap(New Haven:Yale University 10. BruceBuenode Mesquita, between differences Politics."Despiteimportant ofInternational Snidal,"The GameTheory both relations, ininternational ofchoicesituations overtheconceptualization thetwoversions as the system of the international of the structure definition are based on an individualist ofcapabilities. distribution forthePhiJournal British inEconomics," Determinism 11. See SpiroLatsis,"Situational "Situational byFritz Machlup, andthereply pp. 207-45, 1972), ofScience23 (August losophy pp. 271-84. 1974), Allision, Graham include might relations ininternational ofsuchan approach 12. Examples The Cybernetic Steinbrunner, Essence of Decision (Boston:LittleBrown,1971)and John Press,1974). University Princeton ofDecision(Princeton: Theory of program in the individualist citedproblem the mostpersistently 13. This is probably interactions. or their ofindividuals to theproperties explanations all socialscientific reducing 6 Journal of Sociology "Societal Facts," British MauriceMandelbaum, See, forexample, of Journal British Reconsidered," Individualism (1955); Steven Lukes, "Methodological andIndividExplanation, "Reduction, 19(June 1968), pp. 119-29;HaroldKincaid, Sociology 1986), pp. 492-513. Philosophy ofScience53 (December ualism,"
Determinismin Economics," BritishJournalfor the Philosophy of Science 25 (September

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Agent-structure problem 343 to other agents. social relations of theindividual's talist"exceptin terms theserelationally or andtheorize then, is to explicate The secondapproach, of the a social theory that to develop is, properties, structurally defined abstracts away that neorealism theproperties state.Yet theseare precisely of of properties to a distribution structures whenit reducessystem from of conceptualization individualist The neorealist's individuals. preexisting a socialtheory ofthestate: is therefore tooweakto support structure system in exclusively defined if they are agents cannot generate structures system the place. The consequenceof making of those agentsin the first terms thesocialrelations is that words, inother primitive, individual ontologically with particular kind of agent particular individual is a that invirtue ofwhich and untheorized. opaque mustremain forever causal properties that thedevelopas Waltzdoes,"4 be to argue, to thismight One response of the development integral to is not an of the state theory ment of explicit model of the some clearly Yet international relations. theories of systemic and relations, theories of international to buildsystemic stateis necessary Thus,to argue thecontent ofthosetheories. affect thismodelcan strongly withrespectto states'interaction thatthe structure of the industrialized free andthat Dilemma, Prisoner's iterated international tradeis an n-person of understanding a certain requires alwaysproblematic, tradeis therefore is notwhether and powers.The issue,then, interests thosestatesand their theories (it to buildsystemic someunderstanding of the stateis necessary in a grounded a theory, follows from thatunderstanding is), but whether to reality, or simply with setofpropositions coherent somecorrespondence orideology. inintuition grounded assumptions, from a setofpre-theoretical a reliance on in terms convenience, Whatever its advantages of analytical leaves us unableto justify terms aboutprimitive assumptions untheorized and leads, therefore, situations of interaction conceptualizations particular 15 Thus,withtheory building. to an untenable to systemic "as if" approach in international of the state'spowersand interests out an explicit theory be determined a theory without ofthe"rulesofthegame,"itcannot trade, as some than, rather Dilemma is a Prisoner's gamereally ornotthis whether
inRobert to MyCritics," A Response Politics: on Theory ofInternational 14. "Reflections Press,1986),p. University (New York:Columbia and Its Critics Keohane,ed., Neorealism 340. thatthesocialworld builton theassumption of theories 15. The debateoverthevalidity is thatsuchreasoning weretrueis a longone, so mysaying "as if" certain things operates by Milton defined ofthedebatewerefirst The terms is, of course,contentious. "untenable" Economics in hisEssaysinPositive Economics," ofPositive "The Methodology Friedman's Paul debatewith a lively initiated Press,1953),a piece which (Chicago:ChicagoUniversity Fora intheearly1960s. Review Economic inthepagesoftheAmerican andothers Samuelson thelogical empirievenwith is inconsistent that "as if" reasoning argument cogent particularly see seminal contribution, Friedman's thatinformed explanation of scientific cistconception Journal of Political Models," American Statusof Rational Moe, "On the Scientific Terry 1979), pp. 215-43. Science23 (February

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

344 International Organization


16 And,without neo-Marxists might argue, a purecooperation a comgame. to that we cannot pelling argument effect, knowifa failed is due prediction inoursystemic to an error theory orto a misspecification ofthe(pretheoretical) rules of the game. Without a theory of the state,in otherwords, neorealists' efforts to buildcompelling systemic theories of international relations are seriously compromised. This consequencedoes not mean thatthe actual,micro-economic, asneorealists sumptions make about the stateare wrong or misleading (althoughI thinkthey probablyare)-just that they lack a theoretical foundation with somedemonstrated to reality. The result correspondence is a tendency to arguethatstatesact "as if" theymaximize, forexample, and a corresponding powerand wealth, to buildcredible inability systemic theories ofinternational ofthestate relations. Waltz'sseparation oftheories and of international ifneorealists wantto avoid relations notwithstanding, these problems must a socialtheory ofthestate, that they ultimately develop is, theymustmakethe statetheoretically This wouldre"problematic." quirean attempt to theorize directly aboutthegenerative structures of the world and domestic political-economy which constitute statesas particular kinds ofagents with certain causalpowers andinterests. Neorealists already havean implicit theory ofthesesocialrelations (ifthey didnot,they could notattribute anycausalpowers orinterests to state agents), butthey cannot makethistheory explicit, and therefore falsifiable, as longas they treat the stateas ontologically primitive. Since thesocialrelations which constitute states as states willbe potentially unobservable andirreducible to thepropertiesof statesthemselves, however, such a theoretical reorientation will requirea non-individualist and non-empiricist understanding of system and structural structures analysis, an understanding of structure as something morethana distribution ofcapabilities.

b. World-system theory offers suchan understanding andthus, at ofstructure World-system theory to itsconceptualization and structural ofstructure leastwith respect analyIn one shift over neorealism. sis, can be seen as a progressive problem the neorealist crucialrespect,however, theorists world-system duplicate
to characterize language has used game-theoretic 16. As faras I know,no neo-Marxist But clearly, countries. industrialized between theadvanced relations economic international are scholars neo-Marxist ofthestate, understanding theoretical very different becauseoftheir politically andtherefore inmercantilist, to see thoserelations less likely than neorealists much of Capitaland the "The Internationalization RobinMurray, see, forexample, terms; fragile, "The 1971), pp. 84-109, andJohn Willoughby, 67 (May-June Review NewLeft Nation-State," 6 (June Cambridge Journal ofEconomics Economy," intheWorld RoleofProtection Changing but correct, viewis actually is notwhich ofcourse, article, Theissueinthis pp. 195-211. 1982), at leastthe ensures which problem to theagent-structure howto developan approach rather in internaofstates a theory that is, ofdeveloping which is correct, ofdetermining possibility structures. tional economic

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Agent-structure problem 345 makeone approach to theagent-structure problem: theyat leastimplicitly in thiscase thestructure andthen oftheworld system, try entity primitive, The to its effects. to reduceother entities, suchas stateand class agents, I shallargue, the theorists reify result ofthisstrategy, is that world-system likeneorealists, are unableevenin structure oftheworldsystem and thus, A socialstructure is reified when principle to explain itsessential properties. "it is treated oftheactions bywhich it as an objectanalytically independent A solution in is produced.",17 then, engages to theagent-structure problem, without recognizing that only reification whenitobjectifies socialstructures thosestructures. I human actioninstantiates, reproduces, and transforms atleastan implicit should that reification conception emphasize presupposes have reproducoftherelationship ofagents structures to social structures: tiverequirements passively implement. which, forwhatever reason,agents theinclusion or The problem withreification, therefore, does notconcern theories (sincethey must be exclusion ofagents perse from socialscientific intothosetheories. included), butrather theterms oftheir inclusion problem has the Immanuel Wallerstein's solution to theagent-structure samegeneral and thusthesame strengths and weaknesses, as Louis form, 18 Like Althusser's insistence on the Althusser's structural Marxist solution. the core of "absoluteontological priority of thewholeovertheparts,"19 Wallerstein's unitof thattheonlymeaningful approachis theproposition analysisin comparative or international politicaleconomyis the whole anddiscourse of world system. Moreover, bothscholars accepttheconcept laws, "totality,"20 ofsocialwholes that areirreducible, evenbycomposition likeAlthusser, conto their constituent elements. As a result, Wallerstein, in structuralist terms rather thanin ceptualizes "structure" or generative oftheobservable indiof,primitive terms relations between, or properties as viduals. Thatis, incontrast totheneorealist definition ofsystem structure
McNall and Gary Howe, eds., CurrentPerspectives in Social Theory,vol. 1 (Greenwich,

inScott ofSocialStructure," "On theReification Wilson, andThomas 17. DouglasMaynard

p. 287. Conn.:JAIPress,1980), in Louis is discussed to theagent-structure problem Marxist approach 18. The structural pp. 180-81, Books,1970), Capital(London:NewLeft Reading Balibar, andEtienne Althusser pp. 192-200. Press,1984), University (Ithaca:Cornell Althusser Reading Smith, andinSteven and theory world-system between thesimilarities thatdespite be noted,however, It should they relationship, oftheagent-structure understandings totheir with respect Marxism structural modeof ofthecapitalist issues,suchas theconceptualization inimportant wayson other differ Imperialism, Economy, GaryHowe and AlanSica, "Political See, forexample, production. in in McNalland Howe, Current Perspectives Theory," of World-System and theProblem is, aboutwhether that ofthisterm, abouttheexactmeaning however, 20. Theydisagree, "ConBurawoy, see Michael differences, Onthese or"structured." are"expressive" totalities in Perspectives Theory,"in ScottMcNall,ed., Theoretical in Marxist Currents temporary ItsApplicaandHarvey Kaye,"Totality: pp. 16-39, 1979), Sociology (NewYork:St. Martins, 6 Reflections andGenovese,"Historical byWallerstein and SocialAnalysis tionto Historical 1979), pp. 405-19. (Winter
Social Theory,pp. 235-86. 19. Smith,Reading Althusser,p. 177.

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

346 International Organization thedistribution ofcapabilities acrosspreexisting states, world-system theothe structure ristsdefine of the worldsystem in terms of the underlying oftheworld andinparticular organizing principles economy, oftheinterna*21 tional division oflabor, which constitute orgenerate state andclassagents The existence and identity as agents, andtherefore ofagents oftheir causal is produced, and therefore powersand real interests, explained, by their relation to thetotality ofthecapitalist worldsystem. Thus,stateagents are of the structure of theworldsystem in muchthe same sense that effects areeffects ofthestructure ofthecapitalist capitalists modeofproduction, or slavesare effects of thestructure ofmaster-slave relationships. Thisgenerative reading of world-system theory presupposes an ontological and explanatory distinction between"internal" and "external"relarelations are necessary tions.22 Internal inthe relationships between entities sensethat theentities for their Standepend upontherelation very identity. dard examplesof internal relations are parent-child and master-slave; is conceivable theexistence Thisimplies neither without oftheother. entity an internal that relation cannot be reduced to theproperties or interactions of its member on the contrary, the relationship itself elements; explains essential of each entity, and thusthecharacter properties of their interaction.External relations, on theother hand,are contingent relationships or between eachofwhich theother. The interactions entities, can existwithout that fact a peace,for twostates is notessential to go to waror sign example, their as states.External are important relations forexplaining what identity happens to entities inthecourseoftheir interaction, butthey do notexplain theessential characteristics ofthoseentities themselves. approach totheorizing aboutthestructure oftheinternational theresystem, thestateas an effect is to understand ofitsinternal fore, relations to other statesand social formations in the worldpolitical-economy, rather than as an untheorized purely cause of international of a events.The strength to generative approachto structural theorizing, then,is thatin contrast
is characofworld-system ofstructure theory's conceptualization reading 21. Mygenerative schoolofworldapparently theminority) (and at thispoint, teristic onlyofthe"qualitative" the Hopkins. Actually, andTerence for example, byWallerstein represented, system theorists theorists is an interesting world-system and quantitative qualitative recent debatebetween realist between scientific community within a single research explicit tension example ofa quite On this analysis. of structural and methodology conceptions of theontology and empiricist ed., InterinSteveSmith, Approach," Little, "The Systems example, Richard debatesee,for
national Relations, Britishand American Perspectives (Oxford,U.K.: Blackwell, 1985), pp.

Generativestructures are sets of internal relations.To adopt a generative

and Bhaskar, The Possibilityof Naturalism,pp. 53-55.

Some Methodological "The Poverty of International Comparisons: 71-91,and PeterTaylor, ofNewcastleofGeography, University Analysis" (Department World-Systems Lessonsfrom 1985). upon-Tyne, Marx'sConcepinclude Bertell Alienation: Ollman, ofthis distinction discussions 22. Useful University Press,1976), pp. 26-40, Cambridge (Cambridge: tionofMan in Capitalist Society

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Agent-structure problem 347 neorealism's individualist approachit is able, in principle, to explainthe causalpowers andinterests ofstate andclassagents, tomakethesetheoreticallyand empirically problematic. A generative approach to structural analysis does not,however, require thatsystem structures be reified. World-system theorists beginto reify social structures in their whentheyassert,or imply concrete research, not certain onlythat socialrelations are irreducible and constitute thestateand class agentswhichare their elements, butthattheserelations are analytically independent of,andontologically prior to,those agents. Sucha viewis implied by the tendency of world-system theorists to followAlthusser in treating stateand class agentsas no morethanpassive "bearers"of systemic a tendency in a reliance imperatives, whichmanifests itself on functional explanations of state behavior.23 Functionalism is evident,for in world-system example, in theorists' explanation ofgeneral warsdirectly terms ofthereproduction requirements oftheworld-system,24 requirements which becometranslated (or translate intobellicosestateinterthemselves) ests,as wellas in theinterpretation oftheriseof socialist statesin sucha way thatit is consistent withthe reproduction requirements of theworld system.25 Thisis notto say thatworld-system wouldconsciously theorists argue that the reproduction of the world-system occurs without state in manyexplanations agency-theyprobably wouldnot. But nonetheless theworld system ineffect seemsto callforth itsownreproduction bystates; that at leastin their thissuggests concrete research, world-system theorists treatthe world-system as at some level operating of state independently that in theworld-system. While this result action, practice they reify maybe I it On it inevitado notthink is accidental. thecontrary,follows unintended, thefundamental thewholeis ofWallerstein's holism-that blyfrom premise to itsparts. ontologically prior solution to theagent-structure The principal weaknessof a structuralist but behavioral conis it problem that,because cannot"explainanything 26 it to a basis for to fails formity structural demands,' ultimately provide It for exthe of structures themselves. explaining properties deep maybe, structural of theworldsystem intothreedistinct ample,thatthe division
23. Thistendency is oneofthemost persistently cited ofatleasttheearly in criticisms work world-system theory. See, for example, Robert Duplessis, "FromDemesne to World-System: A Critical ReviewoftheLiterature on theTransition from Feudalism to Capitalism," Radical History Review 3 (Fall 1976), pp. 3-41,andSkocpol, "Wallerstein's World Capitalist System." 24. Christopher Chase-Dunn and JoanSokolovsky, "Interstate Systems, World-Empires A Responseto Thompson," and theCapitalist World-Economy: International StudiesQuarterly 27 (September 1983), pp. 357-67. 25. Christopher Chase-Dunn, "SocialistStatesin theCapitalist in his World-Economy,"
Socialist States in the World-System (Beverly Hills: Sage Publications,1982), pp. 21-56. reticalDiscourse," BritishJournalof Sociology 30 (June 1979), p. 150.

26. DerekLayder, "Problems inAccounting for theIndividual TheoinMarxist-Rationalist

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

348 International Organization positions (core,semi-periphery, periphery) thereproduction is functional for but this does not explainwhythe system of capitalism, developedthat particular structure, nordoes itguarantee thatstructure willendure.27 that Becauseoftheir passiveconception ofstateandclass agency, world-system theorists tendto fallintoan historical determinism which, byignoring other possiblehistorical assumesthattheevolution ofthe trajectories, implicitly worldsystem couldnothave turned out anyother way. The limitations of world-system theory's solution to theagent-structure problem, and also its ofstructural similarities to that arein this no moreapparMarxism, respect entthan inWallerstein's ofa fundamental explanation structural like change thetransition from feudalism to capitalism.28 of the Without a recognition ofsystem structures on state andclass agents, ontological dependence Wallerstein is forced intoan explanation ofthat interms transition ofexogenous of an immanent shocksand theteleological modeof imperatives capitalist Thisexplanation production. Brenner's opposesRobert moreor less strucin terms turationist one, whichdescribesthe transition of a dialecticof class struggle and structural This explanation endogenous conditioning.29 reflects difficulties remarkably similar to thosestructural havehad Marxists in theorizing thetransition from one modeofproduction to another.30 likeneorealists, treat their World-system theorists, then, primitive units, in thiscase thestructure oftheworldsystem, as givenand unproblematic. This treatment leads themto separatethe operation of system structures theactivities from ofstateandclass agents-in other words, toreify system ina waywhich structures leadsto static andevenfunctional of explanations stateaction.The worldsystem is nottreated as an historically contingent, andtherefore andrecreation creation ofstate continuously problematic, and thegreater class agents.I think attention Wallerstein's laterworkgivesto of agencyindicates an awarenessof thisdifficulty,31 problems and these efforts have helpedto moveworld-system theexcessive theory awayfrom in his earlycontributions functionalism evident in myporand, perhaps,
inTheRulesofSociological (Chicago: makes this Method point exactly 27. EmileDurkheim is notto is useful "to showhowa fact he saysthat Press,1938), p. 90,when University Chicago thespecific itserves itis. Theuseswhich presuppose orwhy itis what howitoriginated explain cannot give The needwe haveofthings it,butdo notcreatethem. characterizing properties sort Itis to causesofanother specific nature uponthem. their existence, norcan itconfer them owe their existence." that they Press, TheModern I (NewYork:Academic World-System Wallerstein, 28. See, for example, chap. 1. 1974), especially of Neo-Smithian MarxA Critique "The Origins of Capitalist Development: 29. Brenner, 1977), pp. 25-92. ism,"New LeftReview104(July-August of Transitional "The Theory Wolff, Resnickand Richard Stephen 30. See, forexample, Review ofRadicalPolitical Feudalism to Capitalism," from and theTransition Conjunctures Gintis, "On inthesameissuebyHerbert 11(Fall 1979), pp. 3-22,andtheresponse Economics pp. 23-31. Conjunctures," ofTransitional theTheory CamThePolitics (Cambridge: I. Wallerstein, oftheWorld-Economy 31. See, forexample, pp. 112-46. University Press,1984), bridge

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Agent-structure problem 349 of tothecomplete neglect Indeed,incontrast theory. trayal ofworld-system seem to be theorists world-system by neorealists, structures generative problem that to theagent-structure at a solution awareoftheneedto arrive and stateand class agency.World-system structures generative integrates towards a greater focuson agency evolution theory's thus,in someways, Marxism in thelaterworkofNicos of structural thedevelopment parallels to see howagency itis difficult however, Poulantzas, As with Poulantzas.32 structures system parwith andexplanatory canbe brought on an ontological reification holismand structural the strict jettisoning explicitly without probto the agent-structure approach the structuralist whichcharacterize and theexistence ifonlybecause it at least recognizes lem. Nonetheless, to the solution theworld-system structures, roleofgenerative explanatory theory ofstructuration closertothat is considerably problem agent-structure solution. thanis theneorealist
c. Summary

the between I havetried to identify differences important In thissection, explanation, of "structural" understandings and world-system neorealist I havealso to their different socialontologies. and to linkthesedifferences and worldneorealism to show that,despitethesedifferences, attempted to theagent-structure approach underlying sharea common, theory system into primitive agents orstructures attempt tomakeeither they both problem: of those whichleaves each equallyunableto explaintheproperties units, and explanatory claimsabout and therefore to justify its theoretical units, state is thatneither of thisargument stateaction.The obviousimplication constitute structures which andinternational system northedomestic agents theories ofinteror primitive units; them be treated alwaysas given should on leverage explanatory relations shouldbe capable of providing national cannottake endeavor research both.This does notmeanthata particular It does has to start somewhere. as primitive: scientific practice somethings be at must in one research endeavor whatis primitive that mean,however, variable")in as a "dependent (or function least potentially problematic units.Notwithneed theories of theirprimitive another-thatscientists of international to be generaltheories theirapparent aspiration standing and of neorealism ontologies the individualist and structuralist relations, In contrast, ofsuchtheories. thedevelopment preclude theory world-system us wouldpermit problem a structurationist to theagent-structure approach structures and system accountsofbothstateagents to developtheoretical or reification. in either reductionism without ontological engaging
32. State, Power, Socialism (London: Verso, 1978); see also Bob Jessop,Nicos Poulantzas: Marxist Theoryand Political Strategy(New York: St. Martins,1985).

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

350 International Organization

3. An alternative approach to theagent-structure problem


is a relational solution Structuration theory to theagent-structure problem thatconceptualizes agentsand structures as mutually or coconstituted entities. Whatthismeanswillbecomemoreapparent determined but later, ofstructuration I want foundations to consider thephilosophical in first theory andthecurrent debateofthelatter with scientific realism, inthe empiricism philosophy ofscience.Thisdigression is important to myargument because of the continuing of empiricist discourseon social scientists' hegemony of "science," and therealpossibility conceptualization thatskeptics might use that discourse to write off structuration theory's generative approach to structural as "metaphysical."33 In contrast to empiricism, theorizing scientific realism can,inprinciple, call an ontology "scientific" evenifitincludes unobservable generative structures. Whilescientific realism does notmandatesuchan ontology for sociallife that solution (or,for matter, anyother to theagent-structure itis a necessary problem), condition fortheontology of structuration theory. a. Scientific realism The philosophy ofsciencecommunity is currently inthemidst ofa wideranging debatebetween empiricists and scientific realists aboutwhatmight ofscience."34 At issuein thedebatearefundamental be calledthe"theory questionsof ontology, epistemology, and the rational of rejustification search inboththenatural andsocialsciences.Rather practices than attempt to reviewthe entire debate,I will concentrate on contrasting the "hard
33. Thiskindof dismissal is an old individualist May Brodbeck's move;see, forexample, juxtaposition of methodological individualism with"metaphysical" holism in her"MethodologicalIndividualisms: Definition and Reduction," in O'Neill,Modes ofIndividualism and this haveresurrected Collectivism, pp. 289-90.Morerecently, "analytical Marxists" argument tomotivate a reconstruction ofMarxist theory on "micro-foundations"; see Jon Elster, Making SenseofMarx(Cambridge: Cambridge University Press,1985), pp.3-8. In this latter context, it thata number is perhaps worth noting of social scientific realists have argued thatMarxist is bestunderstood inrealist, theory rather than empiricist, terms andtherefore doesnotneedto be reconstructed on microfoundations to be "scientific"; see RussellKeat and John Urry, as Science(London:Routledge & KeganPaul, 1982), Social Theory pp. 96-118,and James Farr,"Marx's Laws," Political Studies 34 (June 1986), pp. 202-22. 34. The terms "empiricist" and "scientific realist"are thelabelstheparticipants in this most ofwhom arephilosophers debate, ofnatural science, use todescribe themselves. Someof the important contributions and overviews are HilaryPutnam, Mathematics, Matter, and Method(Cambridge: Cambridge University Press, 1975);Bas van Fraassen,The Scientific and Intervening Clarendon Image (Oxford: Press, 1980);Ian Hacking, Representing (CamoftheIssue Richard Status bridge: Cambridge University Press,1983); Boyd,"On theCurrent Erkenntnis 19(May1983), A Realist PhilosofScientific Realism," pp.45-90;Jerrold Aronson, Realism Jarrett ophy ofScience(New York:St. Martins, 1984); Leplin, ed., Scientific (Berkeand the of California ley: University Press, 1984);WesleySalmon,Scientific Explanation Princeton andPaulChurchCausal Structure oftheWorld (Princeton: University Press,1984); land and Clifford Hooker,eds., Images of Science: Essays on Realismand Empiricism (Chicago: ChicagoUniversity Press,1985).

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Agent-structure problem 351 core" empiricist and realist are relevant positions on twoissuesthat to the agent-structure problem and to structuration theory in particular: 1) the legitimacy of ascribing ontologicalstatus to unobservable like entities generative structures, and2) thenature ofcausalclaims andscientific explanation.If the realistpositions on these issues seem upon reading to be unexceptionable, that is becausethey are: oneoftheprincipal arguments for scientific realism is that itclaims to makebetter sensethan empiricism ofthe actualresearch practices ofnatural socialscientists. and,to a lesserextent, In other words, realists assumethat scientists, notphilosophers, arethefinal arbiters ofwhatis "scientific." Thiscontrasts with theempiricist position, which is quiteexplicitly an artificial reconstruction ofwhatscientists are or shouldbe doing.Indeed, it could be arguedthatneorealists and worldsystem theorists are,at leastin somerespects, "closet" scientific realists.35 The explicit metatheoretical statements ofbothremain within an empiricist discourse,36 however,and thus theirresearchpracticedoes not follow through on themethodological implications of the scientific realist model. Thisfailure suggests an ironic twist on theold behavioral argument that the socialsciencesare "immature" becausethey are not"scientific" enough: a realist might arguethat, farfrom beingpartof thesolution, theempiricist ofnatural conception scienceuponwhich is based mainstream socialscience is partofthecause of itstheoretical impoverishment. The first axis ofdebatebetween empiricists andrealists is theontological ofunobservables. status tendto "equatetherealwith theexpeEmpiricists knowable"in thesensethat rientially they are unwilling to saythat entities existifwe cannot, at least in principle, have direct of sensory experience them. Theyarguethatwe shouldremain, at most, agnostic abouttheexistenceofunobservable likequarks, entities orgenerative utilities, structures, and thatwe shouldinsteadinterpret thetheoretical terms such describing and thetheories in which entities, thoseterms are embedded, "instrumenthan"realistically."37 tally"rather andtheoretical terms areuseful Theories
35. Neorealists might be seenas scientific realists to theextent that they believethat state interests orutilities arerealbutunobservable mechanisms which generate state behavior, while world-system theorists wouldbe realists to theextent that they believe that thestructure ofthe world-system is a realbutunobservable entity which generates agents. 36. Themost explicit recent discussion ofthephilosophy ofscience underlying of neorealism which I am awareis thesymposium around BruceBuenode Mesquita's "Towarda Scientific Understanding ofInternational Conflict: A Personal View,"International Studies Quarterly 29 (June 1985), pp. 121-36. Buenode Mesquita's emphasis ondeductive analysis andlogical proof, rather than theidentification ofpotentially unobservable causalmechanisms, as thefoundation of scientific a clearly explanation displays empiricist The explicit epistemological orientation. statements on philosophy ofsciencebyat leastthequantitative schoolofworld-system theoristsshowa similar reliance on empiricist arguments; see, forexample, Christopher ChaseDunn,"The Kernel oftheCapitalist World-Economy: Three inThompson, Approaches," ed.,
ContendingApproaches, pp. 55-78. sen, The Scientific Image.

37. The bestrecent ofinstrumentalism defense more is vanFraasandempiricism generally

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Organization 352 International experience, butthere is too much andpredicting instruments fororganizing an "abductive inference,"38 falseentities tojustify epistemic risk ofpositing and entities terms refer torealbutunobservable an inference that theoretical ontology to epistemolthen,in effect subordinate processes.Empiricists, In experientially.39 of whatcan be known ogy-what existsis a function realists scientific inference, contrast to empiricists' rejection of abductive in question iftheentity justified are,inprinciple, arguethat suchinferences or ifitsmanipulation us to interpermits can produce effects,40 observable Thus,thefactthatwe can use world.41 venewitheffect in theobservable structure of atomsto buildnutheories internal aboutthe (unobservable) to clearweaponswhichcan destroy citiesis a good reasonfortherealist them today.Thisthesis believethat suchstructures exist, as we understand talk to empiricism, theory because,incontrast is important to structuration can be scientifically legitiofunobservable and irreducible socialstructures inthis haveobservable oraremanipulable mate view.As longas they effects aboutthe"realwe can, in principle, speakmeaningfully byhuman agents, in other "Generative social structures. structure," ity" of unobservable concept. scientific rather thanmetaphysical words, is a (potentially) in favor ofabcommonly adducetwobasic arguments Scientific realists status ofunobservand,more generally, oftheontological ductive inference ables.42 Proponents argument" arguethata realist of the"indispensability construal is necessary to makesenseof theactualreterms of theoretical search Physicists wouldnotposit andsocialscientists.4 practices ofnatural andbuild wouldnotposit andbuild quarks, andsocialscientists tests around if theythought thatthese or modes of production, testsaroundutilities efficacious. unobservable, werenotrealand causally entities, despite being by go one step further in turn, Proponents of the "miracleargument," that pracnecessary to makescientific notonlyis scientific realism arguing theinstrumento explain ticesrationally intelligible, butitis also necessary
in arefound Useful ofabduction discussions is also known as "retroduction." 38. Abduction KrimerinLeonard Discovery," andtheLogicofScientific Hanson,"Retroduction Norwood

23 (July1986), pp. 287-97. Philosophical Quarterly "RealistPrinciples," Hellman, p. 22. Geoffrey 43. Bhaskar, of Naturalism, ThePossibility Philosophyof Science 50 (June 1983), especially pp. 231-32.

ofScientific Realism," especially pp. 72Status 1969), pp. 73-83,andBoyd,"On theCurrent inthesocialsciences illustration ofabductive andexplicit detailed reasoning 89. Anunusually scientific are practicing that somesocialscientists (andthussupporting myearlier suggestion PublicChoice theStudy ofInstitutions," "An Agenda for Ostrom's inElinor realists) is found 48 (no. 1, 1986), p. 19. 39. Aronson,A Realist Philosophyof Science, p. 261. 40. Bhaskar, The Possibilityof Naturalism,p. 16. "The ThomasCook and Donald Campbell, and Intervening; 41. Hacking, Representing 68 (July especially 1986), pp. Practice," Synthese CausalAssumptions ofQuasi-Experimental 169-72. Spiral,"American Realism:The Ascending forScientific 42. AlisonWylie,"Arguments

man, ed., The Nature and Scope of Social Science (New York: Appleton-Century-Crofts,

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Agent-structure problem 353 Ifmature theories scientific talsuccessofscienceincontrolling theworld.44 of reality, the did not at least partially correspond to the deep structure bothof "miracle."Certainly successof sciencewouldbe an unexplainable thesearguments force,and thus dependon "maturetheories"fortheir be less convincing context scientific (almight in the social thannatural theory might be candidates for though micro-economic theory and Marxist But therelative theories sucha status). ofextant socialscientific weakness thescientific basicpoint: itis a well-established does not jeopardize realist's and perfectly to positunobservable entities to legitimate scientific practice ofthis practice does notimply account for observable behavior. Acceptance or indirect thatanypositis a good one; scientists muststilladducedirect evidenceforthe validity claims,and thisevidenceis of their ontological not philosophers of alwaysrevisable.But by the same token,scientists, science,are thearbiters ofthatevidence. realists overontology The differences between empiricists and scientific of scientific fueldebate on a second axis, the natureand requirements idealsof scienexplanation. Traditionally, there have been twocompeting viewthatexplanation intific explanation: theempiricist or "nomothetic" andthe under a lawlike regularity; volvesthesubsumption ofa phenomenon realist theidentification oftheunderor "retroductive" viewthat itinvolves The generated thephenomenon.45 lying causalmechanisms which physically realist viewcoincideswithreresurgence in recent yearsof the scientific newedattacks which theemsupports on theHumeanmodelofcausation, model, a causalrelation is a piricist account ofexplanation.46 In theHumean ''constantconjunction" of temporally sequencedobservedevents that and standsin a relation initial conditions to certain of logicalnecessity to experience causalmechanisms laws.47 On theHumean view,ourinability us from necessity to causalrelations. directly prevents imputing anynatural and Scientific realistscriticize thismodelbecause constant conjunctions and argueinsteadthatto generalizations are not theirown explanation,
44. See, forexample, Putnam, Matter, Mathematics, and Method; Boyd,"On theCurrent Status oftheIssue ofScientific Realism";Richard Schlagel, "A Reasonable Reply to Hume's

45. See Ernan "Two IdealsofExplanation inNatural McMullin, Science,"inPeter French, et al., eds., Causationand Causal Theories of Minnesota (Minneapolis: University Press, 1984),pp. 205-20,and thethree-way debatebetween Philip Kitcher, Bas van Fraassen, and WesleySalmonin "Approaches to Explanation," TheJournal ofPhilosophy 82 (November
1985), pp. 632-54. 1975); Salmon, Scientific Explanation; Schlagel, "Hume's Skepticism."

Skepticism,"BritishJournalforthe Philosophyof Science 35 (December 1984), pp. 359-74.

andLittlefield, N.J.:Rowman 46. RomHarre andEdward CausalPowers Madden, (Totowa,

47. Hencebehavioral socialscientists' emphasis onquantitative analysis todiscover law-like regularities, rather thanqualitative analysis andtheory to identify causalmechanisms. On the we cannot empiricist model, havesciencewithout (relatively) "constant" conjunctions. Fora useful more orless realist critique ofthis model ofcausation as itrelates to socialscience, see DanielHausman, "Are ThereCausal Relations among Dependent Variables?" Philosophy of Science50 (March1983), pp. 58-81.

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

354 International Organization makea trueexplanatory claim,it is necessary to identify the underlying causalmechanisms makean event which naturally necessary. The disagreement hereultimately on thelegitimacy hinges ofabductive inference. Realistsarguethat ifwe can explain thephysical dispositions andcausalpowers we can make a legitimate of unobservable entities, abductive inference abouttheexistence of naturally necessary relations between cause and efandthereby fect, transcend Hume'sskepticism aboutcausation.48 In effect, therealist's focuson causalmechanisms is an attempt to explain theempiricist'sconstant conjunctions. Thesedifferent models ofcausation, then, generatevery different modelsofscientific explanation. Whereas theempiricist explains by generalizing aboutobservable behavior, therealist explains by how the(often showing unobservable) causal mechanisms which makeobservableregularities possiblework.The scientific realist, in otherwords, arguesthat"answersto why-questions require answers to how-and whatquestions."49 I wantto conclude this brief debatewith overview oftheempiricist/realist three aboutitsimplications forstructuration points theory and,more generresearch scientific realism ally,forsocial scientific practice. First, attempts natural and social scientists in factdo, to make sense of whatpracticing than rather on thelegitimacy ofcertain research versus prescribing practices and the de factopositing of unobservable causal proothers.Abduction cesses and entities, whether or modesofproducthoseare quarks, utilities, in scientific and scientific see no realists tion,go on constantly research, reasonto write thispractice off as unscientific. For thisreason,and thisis the second point,in contrast scientific realismcan make to empiricism, that are scientific senseofunobservable ofstructures generative structures, and their a irreducible to elements. Thisoutflanks keymotivation generate that structural ofthegenerative for individualism, namely theorizing variety there are is necessarily or "unscientific." "metaphysical" Finally, although in translating the protocols and discourseof natural important problems to thesocial sciences-whatRoy Bhaskarcalls scientific practice directly 50-the basic realist idea thatscientific "limits to naturalism" explanation
ofHarreandMadden, theaccount andparticularly ofcausation, accounts 48. Somerealist ofobservable -the explanation "essentialism" an Aristotelian havebeenaccusedofimplying David Miller, "essences"; see, forexample, in terms ofoccultand impenetrable phenomena Philosophy Concepts," on Analyzing Dispositional "HarreandMadden 78, andFredWilson, thatthis emphasize however, of Science 52 (December1985),pp. 591-607.Otherrealists, and properties in terms ofthephysical causal powers can be vitiated by explaining objection Schlagel, "Hume's Skepticism." which underlie them; socialrelations & Kegan as Science(London:Routledge Urry, Social Theory 49. RussellKeat and John p. 31. Paul, 1982), theroleofhuman this translation concern inmaking themost difficult problems 50. Perhaps ofthe andtheambiguity in socialscientific explanations, and self-understandings motivations scientific debate insociallife. Fora sample oftherecent among ofcausal"mechanisms" notion of in the social sciences,see Bhaskar,The Possibility of naturalism on the limits realists
JournalforthePhilosophyof Science 23 (February1972), pp. 69"Back to Aristotle,"British

Naturalism, and Keat and Urry,Social Theoryas Science, especially the postscript.

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Agent-structure problem 355 in rather than causalmechanisms ofunderlying consists intheidentification regularities does applyto the social sciaboutobservable generalizations implications forthe ences, and its adoptiontherewould have important moreprovocatively, and perhaps explanation of social action.Moreover, practicemustbe "critical"to be thatsocial scientific thisidea suggests thenature andimplicatoconsider "scientific." Therest ofthis paperbegins on the perspective structurationist a specifically tionsof thispointwithin problem. agent-structure
b. Structuration theory

a generative a philosophical approach basisfor Scientific realism provides a itprovides andin so doing, theorizing in thesocialsciences, to structural about out theimplications of one of theintuitions foundation forworking probof theagent-structure sociallifewith whichI openedthediscussion areinsomeway agents thecapacities andevenexistence ofhuman lem:that are inseparable they related context-that to a social structural necessarily can be however, humansociality. The implications of thisinsight, from the social relations ways-one reifies worked out in at least two different that earlier world-system and one does not.I argued agents, that constitute that problem totheagent-structure a structuralist approach theory embodies to and determinism. attempts theory Structuration is proneto reification whiletaking thegenerative and relational aspectsofstructuralism preserve theanalytical sepand methodological stepsto prevent explicit conceptual and practices theself-understandings of generative structures from aration to prevent structural reification. ofhuman agents on what withsomecomments It maybe useful thediscussion to preface is an "anaas a theory, theory is about.Structuration theory, structuration in thesensethatit is aboutthe than"substantive" theory, lytical"rather Structuration theory rather than thesubstance ofthesocialworld.51 analysis and there areinthesocialworld aboutwhatkinds ofentities sayssomething a and as suchit provides shouldbe conceptualized, howtheir relationship aboutreal worldsocial or meta-theory forthinking framework conceptual kindsof agentsor what but it does not tell us whatparticular systems, socialsystem. kinds concrete ofstructures to expectinanygiven particular withneorealism or Structuration theory, then,does notcompetedirectly and structuralist individualist theory, but insteadwiththeir world-system social onis, withtheir to the agent-structure problem-that approaches does have structuration theory however, tologies.As a social ontology,
A Theory: of Structuration in "The Status distinction 51. Ira Cohenmakesthisparticular pp. 123-34.NigelThrift 3 (no. 1, 1986), and Society Culture, Replyto McLennan,"Theory, in thantheory is moremeta-theory theory thatstructuration arguing point, makesa similar of Social Theory," Giddens'Reconstitution "Bear and Mouse or Bear and Tree? Anthony
Sociology 19 (November 1985), pp. 609-23.

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Organization 356 International about realtheories of substantive content forthe potential implications should thatsocialscientists and forthemethodology worldsocial systems, a research in turn define These implications use fo studythose systems. and I shallsuggest in Section4 how neorealism agendaforsocial inquiry. some of agenda,and indicate relateto thisresearch theory world-system that emerge from research relations international thegapsin contemporary thiscomparison. approach as a conceptual character with theory's structuration In keeping theoof "structuration theory, the group than a substantively-defined rather as be rists"is also quitediverse and, indeed,mayonly recognizable such identifies example, Thrift, for problematic. thestructurationist from outside Pierre Bourdieu, Abrams, Roy Bhaskar, (Philip fivemajorsocial theorists eventhough most as structurationists, Giddens, andDerekLayder) Anthony label(perhaps becauseit thestructurationist resist ofthesewouldprobably theyall share however, internal differences, is Giddens's).52 Despitetheir the"hardcore" that can be seenas defining basicanalytical objectives four research ofthestructuration program.53 and explanathey acceptthereality to individualists, 1) In opposition social unobservable ofirreducible and potentially importance tory that agents. structures generate and stress they opposefunctionalism to structuralists, 2) In opposition that "the needfora theory reasonand consciousness ofpractical and motivation."54 can account intentionality forhuman in and structures are reconciled agents byjoining 3) These oppositions of one a "dialecticalsynthesis" thatovercomesthe subordination and strucwhich is characteristic ofbothindividualism to theother, turalism.55 from are inseparable spatial they arguethatsocial structures 4) Finally, be timeand space must therefore and that and temporal structures, and concrete soand explicitly intotheoretical directly incorporated cial research.56
in Space andTime,"p. 30. ofSocial Action "On theDetermination 52. Thrift, ibid.,pp. 28-32. from 53. Adapted 54. Ibid.,p. 30. thatcan linkstructure concepts of mediating thedevelopment requires 55. This synthesis thekeysourceof disagreement and as suchis probably in concrete situations, and agency a "positionis established thislinkage through But whether theorists. amongstructuration nexus(Gidor a "system-institution" a "habitus"(Bourdieu), (Bhaskar), system" practice agents inconcrete binding function namely, research, all servethesametheoretical dens),they roles. andexplanatory ontological implicating intomutually and structures to the to be sensitive forsocial scientists admonition 56. This pointis morethana ritual must be about "socialtheories substantive subjects: oftheir context andgeographical historical "On theDetermithestart."(Thrift, from right ofsocialstructures constitution thetime-space in original.) ofSocial Action,"p. 31, italics nation

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Agent-structure problem 357 The following discussion thenaelaborates thesepoints by first discussing ture ofsocialstructures, then ofagents, andfinally oftheir interrelationship. which My accountreliesprimarily on Bhaskar'swork, ofthefive theorists displays themostexplicitly scientific realist orientation.S7 Structuration theorists startout much like structuralists by defining "structure" ingenerative terms as a setofinternally related elements.58 The elements ofa social structure couldbe agents, practices, technologies, territories-whatever can be seen as occupying a position within a social orThe factthattheseelements ganization. are internally meansthat related they cannot be defined or evenconceived in independently oftheir position thestructure. Thus,in contrast to theneorealist definition of international systemstructures as consisting of externally related,preexisting, state agents, a structurationist approach to thestatesystem wouldsee statesin relational terms as generated orconstituted byinternal relations ofindividuation (sovereignty) and,perhaps, penetration (spheres ofinfluence). In other words, states arenotevenconceivable as states apart from their position ina global structure of individuated and penetrated The political authorities. natureand configuration of the internal relations thatcomprise a social inturn, define structure, a setofpossible of transformations orcombinations itselements. As a set ofpossibletransformations, socialstructures are,by a structure's definition, notreducible to therelationships between elements thatare observedin a givenconcrete make a given context.Structures combination or instantiation of elements possible,but theyare not exhausted by whatever particular manifestation is actual. Structuration theorists arguethescientific realist thesis that because social structures generate agentsand their behavior (in the sense thatthey makethe latter possible),thatbecause social structures have observable we can potentially effects, claimthattheyare real entities despitebeing possibly unobservable. This thesisraisestheissue of whenwe can legiticlaimthata social structure mately ofabductive exists.The keyweakness inference ofcircular we assert is thedanger andself-confirmation; reasoning that a structure we posited existsbecauseithas theobserved which effects forthestructure in thefirst at theheart place. Thisweakness of is, I think,
1982),Giddens indicates (p.14) thathe also acceptsa realist conception of science,buthis realism is generally less explicit andthusmoreattenuated thanBhaskar's. A more important reason for relying on Bhaskar rather than Giddens, however, is thelatter's weaker conception ofsocialstructure as rulesandresources rather than as a setofrealbutunobservable internal relations, a conception whichis arguably ultimately voluntarist in its implications; see for example, AlexCallinicos, "Anthony Giddens: A Contemporary andSociety Critique," Theory 14 (March1985), pp. 133-66.
58. See, for example, Bhaskar, The Possibilityof Naturalism,especially pp. 47-56; Peter Manicas, "The Concept of Social Structure,"Journalfor the Theoryof Social Behavior 10 Method in Social Science: A Realist Approach (London: Hutchinson,1984), pp. 80-87. 57. In his Profilesand Critiquesin Social Theory(Berkeley: University of CaliforniaPress,

(July 1980),pp. 65-82; Keat and Urry, Social Theory as Science,p. 121; Andrew Sayer,

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Organization 358 International who claim social scientists, of mainstream heardcomplaint thefrequently in socialscienceare,inprinciple, theories structural Marxist and other that eviis to find to thisproblem response The appropriate "non-falsifiable." of thethatare independent or workings existence dencefora structure's andtorecthestructure was abduced,59 from which observations particular of social systems.60 fortheradicalopenness to control ognizeand attempt is all about: scientists practice This is, of course,whatnaturalscientific entity or to a postulated from an observedphenomenon workbackward to and thenthey to developtestsin closed systems try causal mechanism, Clearlysuch is justified. whether the inference independently determine butthis does inthesocialsciences, toobtain is harder evidence independent and causal structures the basic idea of usingunobservable notinvalidate Indeed,in viewoftheindividto account forsocialbehavior. mechanisms that itis ironic as "metaphysical," structures ofgenerative ualist'srejection ofstructure closertotheconceptualization thegenerative is much approach definition andthenatural is theindividualist's sciencesthan inmathematics properties.61 ofunit-level as a distribution ofstructure is identical thestructurationist understanding point, Up tothis as in combinatorial terms ofstructure Each conceives to thestructuralist's. andtheir transpossible that"generates" itselements an irreducible entity in however, theorists diverge from structuralists, formations. Structuration from inatleasttwofundamental respects differ that socialstructures arguing to is essential a recognition ofthesedifferences natural and that structures, The ofstructuralism. characteristic avoidthereification ofsocialstructures do not structures, unlike natural difference is that"social structures, first Whileit may make theygovern."62 existindependently of the activities thebehavior a natural from apart has an existence senseto saythat structure
Social Theory as Science,postscript. 59. Keat and Urry, do notobtain. is oneinwhich invariant constant conjunctions Although 60. Anopensystem decisive tests of limit thepossibilities for ofopensystems thecomplexity andopen-endedness this problem ofNaturalism, pp. 164-65), (see Bhaskar, ThePossibility socialscientific claims and refer to unobservable entities. For an interesting which afflicts notonlythosetheories ina way might, in somecases,be studied abouthowopensystems explicitly realist argument "Quasi-Experimental controlled tests,see Cook and Campbell, thatwouldpermit relatively Practice." "On the MarcBarbut, of structure in mathematics see, forexample, 61. On thedefinition A Reader in Mathematics," in M. Lane, ed., Structuralism: Meaning oftheWord'Structure' as a Scienceof Patterns: "Mathematics Cape, 1970),MichaelResnick, (London:Jonathan Shapiro, and Reference," Nous 15 (November 1981),pp. 529-50,and Stewart Ontology 1983), pp. 523-48.Modern andReality," Philosophy ofScience50 (December "Mathematics which is inturn, ofbinary systems), theory is basedongroup theory (themathematical physics, I should howandpossibilistic initsviewofstructure. note, probably combinatorial explicitly of I emphasize thissimilarity in socialand natural scientific conceptions ever,thatalthough tojustify I amnotsaying be socialphysics. I amonly trying that socialscience should structure, outthat itpervades the kind andexplanation in socialscience bypointing a certain ofthinking natural as well. sciences natural between 62. Bhaskar, ThePossibility ofNaturalism, pp. 48-49; on thedifferences (London: HutchTheory andsocialstructures, see also Giddens, Studies inSocialandPolitical inson,1977), pp. 118-19.

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Agent-structure problem 359 of its elements, social structures are onlyinstantiated by thepractices of agents.The deep structure ofthestatesystem, forexample, existsonlyin virtue of the recognition of certain rulesand the performance of certain practices by states;if statesceased suchrecognition of performances, the statesystem as presently Social constituted wouldautomatically disappear. structures, then,are ontologically dependent upon (although theyare not reducible to) their elements in a way thatnatural structures are not. The seconddifference is that"social structures, unlike natural structures, do not exist independently oftheagents' conceptions ofwhat they aredoing intheir activity."63 In other words,social structures have an inherently discursive in thesensethat dimension they are inseparable from thereasonsand selfunderstandings thatagentsbring to their actions.This discursive quality does notmeanthatsocialstructures arereducible to whatagents think they are doing,since agentsmaynotunderstand the structural antecedents or implications oftheir actions.Butitdoes meanthat theexistence andoperationof social structures are dependent uponhuman self-understandings; it also meansthatsocial structures acquiretheir causal efficacy onlythrough themedium ofpractical consciousness and action. as social structures Just are ontologically dependent uponand therefore constituted and self-understandings of agents, the causal by the practices and ofthoseagents, in their are constituted powersand interests ownturn, are therefore explained by structures. The structures that constitute agents oftwodistinct kinds:external, or social,structures; andinternal, or organizational, structures. Each typeexplains a distinct set ofthecausal powers and interests of agents-social and intrinsic ones, respectively. Thus, all invirtue internal agents possessthree intrinsic capacities or powers oftheir understructure or "anatomy":64 organizational 1) to have a theoretical in the sense thatit could of its activities, standing (howeverinaccurate) and potentially monitor supplyreasonsforits behavior;2) to reflexively and 3) to makedecisions.These causal powersdifferadaptits behavior; entiate the non-sapient thatcomprise natural strucagentsfrom elements and to theextent of tures, thatstatescan be considered goal-directed units Internal action, theycan be considered agents by thisdefinition. organizational structures arealso important, for thesubjectively however, explaining andorganizational interests ofagents. Individual perceived decision-making in the state,forexample, pathologies maybe crucial fordetermining how social structural or objectiveimperatives forcompetent statepracticea state's "real interests"-translate into subjective interests and actual
63. Bhaskar, The Possibilityof Naturalism,pp. 48-49. of Society, pp. 5-6. 64. Adapted fromGiddens, The Constitution

performance.65

and role of way the nature have yetto tacklein a sustained theorists 65. Structuration inclined materialistically someofthemore Although in socialscientific explanations. interests to I aminclined altogether, use ofinterests theexplanatory theorists might reject structuration

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Organization 360 International structures notwithstandof their internal organizational The importance ofagents aredepencausalpowers andtherealinterests ing, however, other or social structural onlyby the external dentupon and thusexplainable in whichtheyare embedded. the causal powersof context For example, a surplus anddisinvest capital, to extract to invest capitalists (for example, and so forth) are a profits, from realinterests (to maximize labor)and their modeof production of the organizing principles of the capitalist function Similarly, the as a capitalist's. whichdefine theirpositionand interests over the resourcesand control causal powersof the state-to maintain ina given environment territory, to actinan international violence potential and so on-are conferred uponitbythedomestic free oflegalcompulsion, in virtue itis a statein thefirst and international social structures ofwhich princias theorganizing relations maybe as general place. These structural which sovereignty, penetration) system (forexample, ples oftheinterstate localizedorganizconstitute as such,or they ofthemore mayconsist states likethebalanceofpower, international systems, ingprinciples ofconcrete which kinds ofstates. Thus,the"balancer"ina balance-ofdefine particular world has certain or a core statein thecapitalist economy, powersystem, invirtue ofits andinterests which itpossessesonly powers, responsibilities, theconditions socialstructural Social structures, then, constitute position.66 of statesand stateaction;indeed,without social structuring of existence building one could nottalkmeaningfully aboutthefundamental principles polblocksof international relations: "states," "state powers," "foreign anddomestic structures Putinanother way,international icy," andso forth. the "rules of thegame" (broadly defined to includestateagents generate within whichstatesinteract. themselves) as mutuagents and structures Structuration theory, then, conceptualizes entities. Each is in somesensean distinct allyconstitutive yetontologically effect of the other;theyare "co-determined." are the Social structures justas ofhuman action, oftheintended andunintended consequences result thoseactionspresuppose or are mediated conby an irreducible structural relationship is madepossible text. oftheagent-structure Thisunderstanding each from thestart uponthe dependent as ontologically by conceptualizing oftheinternal interms that define relations agents other, byconceptualizing themas such, and by conceptualizing only as existing social structures Thisis that they constitute. themedium oftheagents andDractices through
between distinction at leastan implicit presupposes framework that agent-structure their think of conceptualizations of the various The best overview "subjective"and "real" interests. Connolly's stillWilliam is probably interests, in explaining "interest," and ofthedifficulties Princeton of PoliticalDiscourse(Princeton: "Interests in Politics,"in his book, The Terms pp. 45-84. Press,1974), University ifnotin its in itssubstance, is consistent that ofthebalanceofpower 66. For a discussion ofNeoreal"The Poverty see Ashley, I suggest, with theinterpretation rationale, philosophical ism,"pp. 276-79.

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Agent-structure problem 361 whatGiddensmeansby the "dualityi of structure," that"the structural and outcome ofthepracproperties ofsocial systems are boththemedium 67 Structuration is therefore ticeswhichconstitute thosesystems." theory a greater andagency more than an attempt to introduce balanceofstructure in socialtheory thanis found in individualism and structuralism. Its social of agents the fundamental ontology radically reconceptualizes properties and social structures in sucha wayto makethem ontologically interdepenof thisreconceptualization thatthe "errors"of dent,and it is onlyvirtue reduction and reification characteristic of individualism and structuralism are avoided. As I shall indicate in the nextsection,the ontological and in structuration ofagents and structures conceptual interdependence theory has important implications forthe explanation of social action.Put very ingenerally, it forcesus to see agentsand structures as simultaneously volvedin theproduction of socialphenomena. In Bhaskar'swords: Thussociety is nottheunconditioned creation ofhuman agency (voluntarism), butneither does it existindependently ofit (reification). And individual actionneither completely determines (individualism) noris completely determined by (determinism) socialforms.68 4. Implications forinternational relations theory The discussion ofstructuration theory so farhas focused on itssocialontoland relationship of humanor ogy,on its conceptualization of the nature does and socialstructures. Whilestructuration organizational agents theory international notby itself generate claimsor hypotheses aboutparticular or thecauses of stateaction,therealist/structurationist structures system andtheoretical for problematic does havebothepistemological implications thestudy theofinternational relations. Thus,on theonehand, structuration to theexplanation of itsapproach ory'ssocialontology strongly conditions to stateaction.This idea is consistent withtheeffort realists of scientific reverse thesubordination which ofontology to epistemology, is characteristic of empiricism, and insteadto maketheform of scientific explanations thisgendependent on thenature and causal properties ofentities. Beyond eralconcern withtheform of explanations, structuration however, theory also has implications relations forthecontent of substantive international theories or,perhaps more precisely, for thenature andscopeoftheresearch In particular, agendaswhichunderlie thosetheories. structuration theory whileneorealism andworld-system suggests that, theory provide important leave insights intothestructure anddynamics ofinternational systems, they
and Emancipation," Explanation, in Paul Secord,ed., Ex"Emergence, 68. Roy Bhaskar, Hills:Sage Publications, 1982), p. 286. (Beverly HumanBehavior plaining
67. Giddens, CentralProblems in Social Theory,p. 69.

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

362 International Organization blocks ofinternaimportant gapsinthetheorization ofthetwobasicbuilding system structures. tional relations theory, statesand international
a. Epistemological implications

informed by strucRelatively little empirical research has beenexplicitly for theexplanation of turation theory, which might illustrate itsimplications In this I cannot action.69 an extended illustrastate empirical article, develop I shalladaptmaterials from thefewstructuration ion of myown; instead, andrealist whohavetackled issues theorists70 philosophers ofsocialscience inorder ofsocialscientific to maketwogeneral epistemological explanation and and agent-based analyseshave distinct arguments: 1) thatstructural are in theexplanation butthat 2) they irreducible functions ofsocialaction, of social action.These bothnecessary elements of a complete explanation of the two arguments have important forour understanding implications I shallcallhistorical as nature andlimits andwhat explanations, ofstructural analysis. wellas fortheir integration in "structural-historical" areanswers What counts as an Explanations to certain kinds ofquestions. on adequateexplanation therefore dependson theobjectof thequestion, perspective, two whatis taken to be problematic.71 Froma structurationist of socialackindsof questions are particularly relevant to theexplanation Y?" X possible?"and "WhydidX happen rather than tion:"How is action we thekinds ofanswers The domains ofthesetwoquestions, andtherefore are concerned with thedowouldexpect,are different. "How-questions" withthe mainof the possible,whereas"why-questions" are concerned and limits ofstructural domain oftheactual.To remain clearon thenature an explicit must andmethodological distinction explanation, epistemological be maintained between analysis thelogicof thesequestions:"structural" thepossible,while"historical" theactual.Hisexplains explains analysis andthus torical or willhappen, analysis focuseson whatactually happened Actual takesas unproblematic thepossibility that can happen. thoseevents rather thantherangeofpossiblebehaviors, is theexplanandum. behavior, While thishistorical ofconjunctural causesis an essential component analysis
inempirical research perspective structurationist extensive use ofan explicitly 69. Themost U.K.: Polity Press,1986). (Cambridge, and Structure AllanPred, Place, Practice, is probably thework ofa number (p. 306)that however, Lloydargues inSocial History, In hisExplanation forexincluding, "structure," structurationist theorists has a distinctly of prestructuration andDemocracy theworks ofBarrington Moore, Social Origins ofDictatorship (Boston: ample, (Chicago:Chicago of Society TheSelf-Production Beacon Press,1966),and AlainTouraine, Sociology. and Abrams, Historical Press,1977), University Methodology. andGlassner, A Rationalist 70. Sayer,Method in Social Science;Sylvan different kinds ofquestions between distinctions 71. Theimplications oftheepistemological (New Haven: Yale in Alan Garfinkel, FormsofExplanation are brought out systematically anti-realist ontological pp. 21-48. Despiteits explicitly Press, 1981),especially University Imageis also quitegoodon thelogicor "pragmatvan Fraassen'sTheScientific perspective, types ofexplanations. ics" ofdifferent

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

problem 363 Agent-structure nota it is onlya necessary, of stateaction,however, of the explanation realist a scientific episWithin for scientific explanation. condition sufficient, that is, to exanswers to how-questions; require why-questions temology, Y, we need to knowhowthatstateand its plainwhya statedid X rather to makethe necessary inthefirst place. It is therefore choiceswerepossible "problematic" of statesand statesystems and properties actualbehavior words, in other as given.It is necessary, them accepting rather thansimply ofstatesin thecausal properties to explain in structural analysis to engage ofstrucdiscussion arepossible.The following their actions ofwhich virtue interdebutnonetheless on thedistinct, andhistorical expands analysis tural ofexplanation. rolesofthesetwoforms epistemological pendent, and by a prohistory, with actualevents, with starts Structural research existforthose whatmust is, byasking and abduction-that cess ofcritique strucorganizational to thesocialand internal events to happen-abstracts thenbe might tureswhichmakethose eventspossible.These structures orgrammatical) (mathematical techniques with qualitative formally modeled alevents,72 to observable relationship whichdescribetheirpossibilistic In any maynotalwaysbe possiblein opensystems. though suchmodeling of observable to the explanation contribute explanations case, structural of ofthepossiblewaysofacting that are instances they events by showing deteraredefined bythestructurally where thosepossibilities socialagents, way,strucPutanother ofthoseagents. mined causal powersand interests or "rulesofthegame" ofexistence tural revealtheconditions explanations "critical," is necessarily theory of social action.In thissense structural social to theunderlying appearances given sinceitforces us to lookbeyond forms. sense) phenomenal thatgenerate(in a possibilistic relationships of actual explanation Whilestructural analysesare thuspartofa complete theyonlyantheydo notexplainthoseeventsdirectly; however, events, or transare possible, ofwhatcombinations swerthequestion ofhowthey withits organizing are consistent of a structure's elements formations analysesmay uncover"tendencies"for structural Although principles. norpoint in certain generalization to be actualized ways,neither structures andanyattempt explanations, is an important aspectofstructural prediction would events ofparticular theproduction for directly touse them to account domain.73 explanatory proper them beyondtheir riskoverextending andobjects hand,"studiesactualevents on theother research, Historical and each ofwhich havebeenisolated ofdiverse determinations,' as 'unities "74 Historicalexplanaresearch. examinedthrough abstract[structural]
couldbe usedingenerative that methods to someoftheformal introduction 72. Anexcellent A Rationalist analysesis foundin chaps. 5 and 6 of Sylvanand Glassner, structural Methodology. these to recognize 73. In Methodin Social Science,Sayerargues(p. 217) thata failure he calls"pseudoorwhat thedeterministic, for is responsible analysis ofstructural limitations research. Marxist ofmuch quality concrete," in Social Science,p. 216. Method 74. Sayer,

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Organization 364 International (orreconstruct as given ofagents andcausalpowers taketheinterests tions75

particular attempt to explain them), andthen trying to explain them without are affected by the on how thosepowersand interests eventsby focusing the at thislevel;it stipulates proceeds facing actors.Neorealism incentives and theinterests and causal powersof agentsand then context structural thanZ?" attempts to answerthequestion"Whydid stateX do Y rather or it is important to notethatone of theintended Beyondthis,however, system strucor reproduce effects of stateactionis to produce unintended to explainthe emeranalysisis necessary historical tures;consequently, the whichconstitute conditions of the structural gence and persistence andconditions quality medium ofpossibility forstateaction.Thisrecursive their "diverandhistorical is the"unity"beneath ofstructural explanations interdepenepistemological basis of their sity,"and thusit is theultimate and between thedistinction to maintain necessary dence.It is nonetheless theproperties explains mode:eachultimately autonomy ofeachexplanatory ofthecentral objectsoftheother. analand historical limitations of structural explanatory The respective a complete that ofstateaction-thatis, explanation however, ysessuggest, bothhowthat andwhythat possibility one that action was possible explains ina particular havetocombine moment-will was actualized form at a given or "dialectical"analythesemethodologies intoa "structural-historical" to theorize This combination structural analysis sis.76 willrequire abstract and conand explainthecausal powers, and interests of states, practices, sequenceofchoices crete historical to tracethecausallysignificant analysis of and interactions lead to particular (andto thereproduction events which reand historical ofdoingstructural social structures). Giventhedifficulty analysismayrequire"brackstructural-historical searchsimultaneously, first one andthen theother that is, taking social mode,77 explanatory eting"
to I do notmean form ofexplanation, todescribe this "historical" 75. Bymyuse oftheterm of theresearch practice use,orthat always is theexplanatory modehistorians that this suggest justas itseemstomethat On thecontrary, oratheoretical. astructural is necessarily historians to I am onlytrying andtheoretical. is structural goodhistory goodsocialscienceis historical, butinterdedistinct areepistemologically explanations and"structural" "historical" that argue ofwhouses them. regardless forms ofinquiry, pendent DepenCardosoand Enzo Faletto, Fernando is from "structural-historical" 76. The term pp. 154Methodology, A Rationalist andGlassner, Sylvan is from ix-xiv,while"dialectical" inSayer's and"concrete" research "abstract" between therelationship terms parallel 59; both PeterManicas of theseterms, he does notuse either Methodin Social Science. Although of inhiscritique ofinquiry form ofthis ofthelogicandimplications a goodillustration provides Press,1979); University Cambridge (Cambridge: ThedaSkocpol'sStateandSocialRevolutions pp. 204-18. inHistory 20 (no. 2, 1981), and Theory see hisreview is a of"bracketing" pp. 80-81.Thisnotion 77. Giddens, inSocial Theory, Problems Central Margaret see, forexample, ofstructuration theory; ofsomeofthemajor critiques focalpoint British and Action," Structure On Combining versusStructuration: "Morphogenesis Archer, and "On Duality Gregson, pp. 455-83,and Nicky 1982), 33 (December of Sociology Journal inHumanGeogProgress andTimeGeography," Theory The Case ofStructuration Dualism: pp. 184-205. 1986), 10(June raphy
of CaliforniaPress, 1978),pp. dencyand Developmentin Latin America (Berkeley: University

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

problem 365 Agent-structure the to examine and agents in turn as temporarily structures givenin order difficulty, however, of theother.This methodological explanatory effects of structural and interdependence shouldnot obscurethe epistemological rolesof agents explanatory thefactthattherespective historical analysis, interrelationtheir apartfrom cannotbe understood and social structures the ontology of structuration from followsdirectly ship.This conclusion their inthesensethat are inseparable from socialstructures theory. Agents and social structures of thosestructures, actionis possibleonlyin virtue as theyare instantiated by cannothave causal significance exceptinsofar of both by theproperties agents.Social action,then,is "co-determined" and social structures. agents
b. Theoreticalimplications

Whilethisdiscussion of structuraof some epistemological implications tothescope relevant very general, itis nonetheless tion theory is admittedly andcontent relations theories. A keyimplication international ofsubstantive in Section2 abouttheagent-structure was that oftheargument relationship ofboth havefoundations in theories relations must theories ofinternational and system structures). Such (stateagents their principal unitsof analysis or desirable: theyare necessary convenient theories are morethansimply the bothfrom followsdirectly to explainstateaction.This requirement scientific realist'sconception of explanation as identifying causal mechaaboutthe theory the ontological claimsof structuration nisms,and from ofstatesand system and structures. If theproperties relationship ofagents in theinternastructures are boththought to be causally relevant to events then theointerrelated, are somehow tional system, and ifthoseproperties to explainstate reticalunderstandings of boththoseunitsare necessary to the ofthestateis notintegral action.Waltz'ssuggestion that thetheory must therefore relations taskofdeveloping systemic theories ofinternational and methodological a conceptual be rejected. Structuration theory provides a research and as suchit defines framework to overcome thisseparation, in structures and thesystem aboutbothstateagents agendafortheorizing The core ofthisagendais theuse of structural which theyare embedded. andtheuse of ofstate agents, totheorize theconditions ofexistence analysis of social struchistorical analysisto explainthegenesisand reproduction ofsuch aboutthepossible content evenpreliminary remarks tures. Although I can indicate theories wouldrequire someofthedirections another article, agenda. andbodiesofresearch which to sucha research might be relevant the state" implies whichseeks to dea research endeavor "Theorizing understanding ofthecausally andempirically grounded velopa theoretically as an interests, practices) ofthestate (suchas powers, significant properties woulddefine exhausIdeallysucha theory organizational agentor entity. deterthepossible than agents, rather generate waysofacting ofstate tively

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Organization 366 International The possibleways of statebehaviors. aboutparticular minate predictions by the social explained and therefore actingof an agentare constituted organizational inwhich andbyitsinternal itis embedded structural context Thus, explanation. to structural and as such theyare amenable structure, organizing thesocial structural can be used to explicate analysis structural kindofsocialactor,that thestateas a particular which generate principles place. Thisuse would of whichthestateis a statein thefirst is, in virtue as a Hobbesian rather than socialentity, as an inherently thestate recognize organitheinternal analysis couldalso reveal Structural individual. primitive and reits perceptions of the state whichcondition zationalstructures International and opportunities. imperatives sponses to social structural of suchinternal theimportance to discount have tended relations theorists but giventhatthese stateaction,78 in explaining structures organizational arelikely to they willbe theproximate cause ofanystateaction, structures of statebehavior. in theproduction mechanisms constitute important constitute might socialstructures a little at leastfour further, To elaborate international-economic, domestic-political, states: domestic-economic, withthe Recentworkconsistent structures.79 and international-political all these has examined orientation ofstructuration theory critical-structural with and unintegrated muchremains quitepreliminary although structures, analysesof the research.The mostsustained otherbodies of structural within theneo-Marxist are found of existence conditions state'sstructural Whilethe withWeberians.80 thedebatesof thelatter and within tradition of the withthe relationship primarily is concerned neo-Marxist literature theorists to economic domestic world-system structures, state (capitalist) structures. economic on thestate'srolein international have concentrated of globalcapiof the structure the world-system understanding Although
pp. 7 (Summer 1972), Policy Foreign Important?" "AreBureaucracies Krasner, 78. Stephen Policy A Critique," Policy: Foreign andAmerican Politics Art,"Bureaucratic 159-79;Robert pp. 467-90. 1973), Sciences4 (December be called structural of whatmight a rejection implies of structures 79. This multiplicity suchas principles, organizing one setofunderlying there is only that is, theviewthat monism, of constitutive andtherefore ingenerative terms can be explicated that thoseoftheeconomy, by Marxism developed thecritique of structural with is consistent Thisanti-monism agents. andSocialFormainModeofProduction andPaulHirst Hindess likeBarry post-Althusserians and Mouffe in Hegemony Laclau and Chantal 1977),and Ernesto tion(London:MacMillan, totheproblem solution Buttheir discourse-theoretic Strategy (London:Verso,1982). Socialist theories that we build opposesmysuggestion inmany monism waysfundamentally ofstructural realism. on thebasisofscientific socialstructures ofmultiple and Sol PicHolloway John include statetheory of neo-Marxist examples 80. Prominent Poulantzas, 1978); Arnold, A Marxist Debate(London:Edward eds.,Stateand Capital: ciotto, It Rules? ClassDo When Does the Ruling What Therborn, andGoran Socialism; Power, State, ReThedaSkocpol,"Political include critiques (London:New LeftBooks, 1978).Weberian oftheStateandtheCase oftheNew Deal," Theories Crisis:Neo-Marxist sponseto Capitalist and MichaelMann,"The Autonomous Politicsand Society10 (no. 2, 1981),pp. 155-201, 25 (no. 2, 1984), pp. 185-213. Journal ofSociology PoweroftheState,"European

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Agent-structure problem 367 talism is arguably marred by an overemphasis on exchange relationships,81 offer and world-system literature boththe neo-Marxist important insights of thestate,and therefore some of existence intotheeconomic conditions Less research, I think, has been done of its causal powersand liabilities. thatmight on thepolitical structures from a critical-structural perspective at the domestic the state.Nonetheless, work constitute level,innovative ofthe takesas itsstarting a Gramscian is beingdonewhich point rejection to theorize of mostneo-Marxist and insteadattempts economism theory in critical-structural This line of analysisis now forms terms.82 political of scholarswho beingextendedto the international level by a number have focusedon thenature and implications of suchfundamental organizingprinciples ofthestatesystem as sovereignty, thebalanceofpower, and hegemonic domination forthe conceptualization of the stateand explanationof stateaction.83 Purelyschematic though theseremarks are, I think all thesebodies of research wouldpotentially contribute to a single overall problematic generatedby a structurationist approach to international relations-the development ofa critical theory ofthecausalpowersand interests ofthestate.An implication of a rejection of structural monism, however, is thatit willbe necessary totheorize therelationships or "articulations" between thedifferthe state. In otherwords,a "structureent structures whichconstitute structure" problememergesfromthe structurationist problematic. This is at the core of the literature problem on the "articulation of modesof production,"84 andis implicit inmuch oftherecent "post-Marxist" debate. Ironically, thisproblem is strongly reminiscent of J. D. Singer'soriginal
ofworld-system "The 81. Examples ofthe"productionist" critique theory include Brenner, and Origins ofCapitalist Development," andHoweandSica, "Political Economy, Imperialism, of World-System theProblem Theory."The alternative conceptualization ofthestructure of has beenmostfully thecapitalist worldeconomy (in terms of a globalmodeof production) see developed by the "internationalization of capital"schoolof Marxist political economy; Christian Palloix, "The Self-Expansion ofCapital on a World Scale," Review ofRadicalPolitical Economics 9 (Summer 1977), pp. 1-28. 82. Poulantzas, State,Power,Socialism;Nicos Mouzelis,Politicsin theSemi-Periphery (NewYork:St. Martins, 1986); Samuel BowlesandHerbert Gintis, Democracy andCapitalism (New York:Basic Books, 1986). 27 (July 83. BruceAndrews, "Social RulesandtheStateas a SocialActor,"World Politics AnEssayin 1975), pp. 521-40;Robert Cox, "Gramsci, Hegemony, andInternational Relations: 12 (Summer Method,"Millenium 1983),pp. 162-75;Ruggie, "Continuity and Transformation";Ashley, "The Poverty of Neo-Realism," and "Social Willand International Anarchy: theDomestic Alker in theStudy in Hayward Beyond Analogy ofGlobalCollaboration," and Ashley, Anarchy, Power,Community: Understanding International Cooperation (forthcoming).Despitethepotential usefulness ofthis research to thestructurationist problematic, however,someof thesescholars wouldprobably rejectassociation with thattheory, especially insofar as itis grounded in realist philosophy ofscience. 84. AidanFoster-Carter, "The Modes ofProduction Controversy," New LeftReview107 (January-February 1978), pp. 47-77;Harold Wolpe, ed., TheArticulation ofModesofProduction(London:Routledge & KeganPaul, 1980).

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

368 International Organization discussion ofthelevelsofanalysis thelatter wouldbe problem,85 although seenhereas one instance ofa moregeneral theoretical and methodological problem of apprehending therelationships between different structures of whatever type (political or economic) orlevelofanalysis (domestic or interas a theoretical national). and methodological Although issuethe"structurestructure problem" points beyond structuration itstreatment theory, clearly has implications fortheattempt to buildsatisfactory theories and explanaofstate tions action.Each structure inwhich thestate is embedded willhave itsownlogic,reproduction andthus requirements, prescriptions for competentstatepractice.Developing a structural theory of the stateand state action, involves morethansimply then, thedifferent explicating structuring principles which generate states;italso requires isolating and assessing the causalroleofandinterrelationships different andsometimes among competingstructural determinations of stateaction. The needfora theory ofthestateininternational relations is mirrored by theneedfortheories ofthesystem structures which thestate.In constitute thesetheories wouldhave at leasttwomainelements: general terms, 1) a modelof theorganizing synchronic reprinciples, logic,and reproduction of thestructure in question, quirements and 2) an historical account ofthe of thestructure. genesisand reproduction Structural have genertheorists or unableto graspthe contingent ally been reluctant nature of structural and have tendedinsteadtowards genesisand reproduction, functional or ofthat teleological can be corrected readings process.Thistendency bythe of structuration on the historical emphasis and contintheory specificity ofthestructuring ofsocialstructures. gency Giventhestructurationist conofsocialstructure as onlyinstantiated ceptualization andreproduced bythe activities of social agents, an historical of social structuring must analysis beginwiththeintended and unintended consequences of stateaction(and theactionof otheragents).Although therelevant methodological toolsto suchan analysis are potentially quitevaried, thegame-theoretic methodolofrecent work inneorealism ogycharacteristic is potentially to well-adapted thisanalytical task. The analysisof iterated games and the "new institutionalism" in thestudy of political in particular have proven institutions useful ingenerating theemergence insights into ofandreproduction ofsocial institutions as theunintended of strategic consequences and interactions,86 is no a priori there reasonwhywe cannot extend thelogicofsuchanalyses to theanalysis ofgenerative We must structures. recognize, however, that
85. J.D. Singer, "The Levels of Analysis Problem in International Relations," in Klaus Knorrand SidneyVerba,eds., The International System:Theoretical Essays (Princeton: Princeton University Press,1961), pp. 77-92. 86. MichaelTaylor, Anarchy and Cooperation (New York:Wiley, 1976);Robert Axelrod, TheEvolution ofCooperation (NewYork:Basic Books,1984); James March andJohan Olsen, "The New Institutionalism: Organizational Factors in Political Life,"American Political ScienceReview 78 (September 1984), pp. 734-48.

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Agent-structure problem 369 game-theoretic models focusattention on thetechnical of decision problems

structure of social interactions constitute or empower thoseagentsin the first place. The use ofgametheory to developan historical of understanding the emergence of social structures, therefore, wouldhave to be complemented by a generative understanding of the construction of agentsand situations of strategic interaction. These remarks on theimplications of structuration theory forthescope and content of international relations theories are obviously purely schematic and are intended onlyto illustrate thekinds ofresearch that be might relevant to a structurationist approach to international relations. Indeed,I should that emphasize structuration theory byitself cannot generate specific theoretical claimsaboutinternational relations. The theory has epistemologicalimplications for theform which explanations ofstate action should take, and it suggests a research agendaforsubsequent theorizing, butitdoes not makea direct contribution to oursubstantive understanding ofinternational relations perse. Thispoint raisestheissueofthecriteria bywhich structurationtheory should be evaluated byscholars ofinternational relations. Given that its analytical or meta-theoretical quality prevents an empirical assessmentof the theory, it seems to me thatstructuration theory shouldbe evaluated on pragmatic grounds, on itsability in existing to solveproblems substantive theories, to suggest new areas of theoretical and empirical inor to integrate quiry, different I think bodiesofresearch. By thiscriterion, structuration theory clearly impoveson its principal competitors, individualism and structuralism. It provides a framework for explaining theproperties of both state agentsand systemstructures whichis deniedto the individualist and structuralist ontologies of neorealism and world-system theory, anditdefines a research agendafor international relations that organizesand subsumes under a single problematic a potentially greater variety of extantsocial scientific research. The potential contribution to international relations research is there, butwe cannot ofthe assess theimportance contribution until theorists attempt to use a structurationist to perspective ground and inform their theoretical and empirical research. Conclusion Rather thantrying to summarize a longargument abouttheagent-structure and structuration I wantto conclude thisarticle problem theory, bypicking the implications realism forsocial of scientific up its othermainthread, scientific research. a satisfacWhether or notstructuration theory provides resolution to theagent-structure of socialscientists' tory problem, adoption a scientific realistperspective could have on ontology and epistemology and empirical theoretical fortheir potentially revolutionary consequences

given agents, and thatthey therefore tendto neglect thewaysin which the

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

Organization 370 International in social sciencehas led discourse of empiricist The hegemony research. between"science" (thatis, dichotomy intoan apparent social scientists of herparadigms science) and the allegedly"un-scientific" empiricist in the to naturalism the limits theory.87 Whatever and critical meneutics bychallenging thisdichotomy undermines realism scientific socialsciences, science of natural theinterpretation argument, thecore of theempiricist's of "science" rests.Scientific of themantle uponwhichherappropriation inthePositivispositions an alternative to thestandard then, offers realism, about "abductively" to think enjoinssocial scientists one which musstreit, lawto find of trying instead to buildtheir theories, "causal mechanisms" themoreimporAmong regularities. aboutobservable likegeneralizations shift is a and epistemological tantconsequencesof such an ontological or relational in thegenerative theorizing forstructural scientific motivation "critical"sinceit reis inherently prescription sense.Thismethodological social forms totheunderlying ofobservable andpenetration a critique quires then, ofscientific realism, An implication them.88 which generate structures of tothedevelopment (ina broadsense)is essential theory" "critical is that as a "science." relations, international social science,and by extension
Practice 1975). and Political 87. BrianFay,Social Theory (London:Allen& Unwin, RadicalPhilosophy "Scientific Explanation andHuman Emancipation," 88. Roy Bhaskar, The Critical "Theory andPractice in Sociology: Dandeker, 26 (1980),pp. 16-26;Christopher 1983), pp. 195-210. theTheory ofSocialBehavior 13(July Journalfor Imperatives ofRealism,"

This content downloaded from 202.92.128.28 on Tue, 4 Jun 2013 01:41:02 AM All use subject to JSTOR Terms and Conditions

You might also like