'Causation' has been a deeply divisive concept in I nternational Relations theory. Many post-positivist theorists have rejected the aims and methods of causal explanation in favour of 'constitutive' theorising.
'Causation' has been a deeply divisive concept in I nternational Relations theory. Many post-positivist theorists have rejected the aims and methods of causal explanation in favour of 'constitutive' theorising.
'Causation' has been a deeply divisive concept in I nternational Relations theory. Many post-positivist theorists have rejected the aims and methods of causal explanation in favour of 'constitutive' theorising.
Review of International Studies (2006), 32, 189216 Copyright
British International Studies Association
doi:10.1017/S026021050600698X Causes of a divided discipline: rethinking the concept of causein I nternational Relations theory MI LJ A K URK I * Abstract. During the last decades causation has been a deeply divisive concept in I nter- national Relations (I R) theory. While the positivist mainstream has extolled the virtues of causal analysis, many post-positivist theorists have rejected the aims and methods of causal explanation in favour of constitutive theorising. I t is argued here that the debates on causation in I R havebeen misleadingin that they havebeen premised on, and havehelped to reify, arather narrowempiricist understandingof causal analysis. I t issuggested that in order to move I R theorising forward we need to deepen and broaden our understandings of the concept of cause. Thereby, wecanradicallyreinterpret thecausal-constitutivetheorydividein I R, as well as redirect thestudy of world politics towards moreconstructivemulti-causal and complexity-sensitiveanalyses. Introduction I n Explaining and Understanding International Relations Martin Hollis and Steve Smith famously argued that there are always two sorts of stories to tell 1 in the discipline of I nternational Relations (I R): one can explain international politics throughcausal analysisof international processesor seek to understand international politics through inquiringinto themeanings of, and thereasons for, theactions of world political actors. 2 Theassumption elicited by Hollis and Smith that thereis a fundamental dichotomy between causal and non-causal approaches to the social world, has cometo permeatethedisciplineof I R in thelast decadeor so. Whilethe so-called scientic theorists have advocated systematic causal analysis in I R 3 , the so-called reectivist constitutive theorists have maintained that causal analysis is neither a necessary, nor a desirable aim in understanding world politics. 4 As a * Theauthor would liketo thank Alexander Wendt, Colin Wight, J onathan J oseph, Hidemi Suganami and an anonymous reviewer for their comments on earlier drafts of this article. 1 Martin Hollis and SteveSmith, Explaining and Understanding International Relations (Oxford: Clarendon Press, 1990), p. 1. 2 Hollis and Smith, Explaining and Understanding, pp. 17. 3 G. King, R. O. Keohaneand S. Verba, Designing Social Inquiry; Scientic Inference in Qualitative Research (Princeton, NJ : Princeton University Press, 1994). 4 David Campbell, Writing Security: United States Foreign Policy and Politics of Identity (Manchester: Manchester University Press, 1998), p. 4. 189 consequenceof their disagreements thetheorists in thedierent camps havetended to eschew talking to each other at least constructively in providing accounts of world political processes. 5 Curiously, despite the fact that the contentions between these camps have centred around theconcept of cause, therehas been no real engagement in I R with the source from which this dichotomisation of causal and non-causal forms of theorising arises. This article argues that an uncritical acceptance of a so-called Humean conception of causation is at theroot of thedisciplinary divisions in I R. A Humean discourseof causation a set of assumptions deeply embedded in modern philosophy has, it is argued, deeply informed most I R theorists engagements with causation. Theimplicit inuenceof Humean assumptions has brought with it many prejudiceswithregardtothewayinwhichcausesarediscussedinI R. Thus, it isoften assumed that causes or causal analyses imply determinism, laws and objectivism. I t is also assumed that causes refer to pushing and pulling forces. I t issuggestedherethat weneednot conceptualisecausationaccordingtoHumean assumptions: we can, in fact, think of causation as a common-sensical intuitive notion with a multiplicity of dierent meanings, none of which entail laws or determinism. Wecanalso understandsocial scienticcausal analysisasepistemically reective, methodologically pluralist and complexity-sensitive. I t is seen that if we drawon philosophies of causation with much richer and broader understandings of the notion of cause we can start doing away with the prejudices that the Humean discourseof causation has promulgated in I R. I f theconcept of causeis reconcep- tualised on thelines suggested here, it emerges that, not only is theempiricist form of causal analysisthat dominatesinI R problematicasamodel of causal analysis, but alsothat causal analysisinthewider reconceptualisedsenseis, infact, somethingthat all I R theorists, includingconstitutivetheorists, engagein. I t isargued, then, that the deeper and broader conception of causal analysis advanced here can, not only improve causal analysis in I R, but also help forge constructive links between theoretical camps in thedivided discipline of I R. Thereconceptualisationof causationisadvancedinvesections. To contextualise thediscussion, therst section will tracehowtheHumean philosophy of causation hascometo inuencethemodernphilosophyof scienceandsocial science. Thefocus will then shift to the examination of Humean assumptions in I R debates on causation. Then, the concept of cause will be reconceptualised in two interrelated sections. First, it is argued that we need to challenge the inuence of Humeanism through adoptinga philosophically realist deeper conception of cause. Thesection that followsarguesthat thismoveneedsto beextended through thedevelopment of abroader meaningfor theconcept of cause. I t issuggestedthat I R theorisingreturns to theinsights of theAristotelian conception of causation. Therecongured deeper and broader model of causal analysis has many metatheoretical, theoretical and methodological implications for I R theorising and the disciplinary debates. These will bereected on in thenal section. 5 Therehas been a tendency to seetheseapproaches as mutually exclusive, incommensurableaccounts as Hollis and Smith portrayed them. Hollis and Smith, Explaining and Understanding, pp. 196216. 190 Milja Kurki Thedominanceof Humeanismandthedeclineof causation The concept of cause has been one of the central but also one of the most controversial conceptsinthehistoryof philosophyandscience. However, duringthe last threehundred years a particular set of assumptions, arisingfromtheempiricist philosophy of David Hume, have come to dominate the way in which causes and causal analysishavebeenunderstood. I norder to gainadeeper understandingof I R debates on causation, it is important to grasp thenatureof theHumean legacy in modern philosophy and science. The rise of Humeanism: the narrowing down and emptying out of the notion of cause I nancient Greek philosophy, whichrst formulatedtheconcept of cause, thenotion referred to that which brought somethingabout or contributed in any way to the existenceof objects, or to changein or between them. I t was accepted that nothing in theworld comes fromnothing and that thenotion of aition (cause) provided an openmetaphor that referredtothoseforcesthat werebehind other things. Thereare two aspects that areworth highlightingwith regard to theseearly understandings of causation. First, for classical ancient thinkers, notably Aristotle, causes wereunderstood as ontologicallyreal. Althoughtheconcept of causewasseenasaman-made concept designed to help us to understand why the world works as it does, this concept was conceived to havean ontological referencepoint in thereal causal powers of nature. 6 Second, theconception of causein ancient philosophy wasplural. Aristotle, withhisfour causes account, recognisedfour maintypesof causes: material, formal, nal and ecient causes. 7 Material and formal causes referred to theroleof matter and ideas in shaping reality, ecient causes to pushing and pulling moving causes, andnal causesto endsor purposesascauses. Crucially, thesedierent types of causes which wewill examinein moredetail in thefourth section wereseen as interlinked: anycausal analysiswouldhaveto understand, not onlydierent typesof causes on their own, but also their complex interplay in concretesituations. I nterestingly, in modern philosophy theAristotelian ontologically grounded and broad conception of cause has been sidelined in favour of a very dierent under- standing of causation. During the sixteenth century important shifts took place in theorisingcausation. 8 Descartesinitiatedanarrowingdown of theconcept of cause by concentrating on the mechanical meaning of the term cause: the notion of ecient cause. 9 Descartes rejected as vagueand unsubstantiated thewider material, formal andnal causemeaningsof thenotionof causeandcametoseecausesstrictly 6 For descriptions of early accounts of cause, seefor exampleR. J . Hankinson, Cause and Explanation in Ancient Greek Thought (Oxford: Clarendon Press, 1998); J onathan Lear, Aristotle; the Desire to Understand (Cambridge: CambridgeUniversity Press, 1988). 7 Aristotle, Metaphysics (Harmondsworth: Penguin, 1998), p. 115. 8 For an account of thehistorical changes in theconcept of causein themodern period seeW.A. Wallace, Causality and Scientic Explanation, vol. I (Ann Arbor, MI : University of Michigan, 1972), and K. Clatterbaugh, Causation Debate in Modern Philosophy 16371739 (NewYork, London: Routledge, 1999). 9 ReneDescartes, Key Philosophical Writings (Ware: Wordsworth Classics of World Literature, 1997). The concept of cause in IR theory 191 as pushing and pulling, so-called ecient, forces. He argued that, rather than attributing objects unobservable occult qualities (material, formal or nal causal powers), themost useful waytothink of causeswastotracethepushingandpulling relations between things. 10 Having concurred with the Cartesian narrowing down of the meaning of the concept of cause, David Humes eighteenth century philosophy initiated the emptying out of the notion. Hume advanced a radical empiricist critique of metaphysics according to which, in the search for reliable knowledge, human perceptionsshouldtakeprecedenceover anyspeculationabout thenatureof reality. Throughfocusingtheacquisitionof knowledgeontheobservable, Humesempiricism sought to challenge the ontological reality of causes. Hume had a profound disagreement with all the philosophers before him who tried to dene causes as (ontologically) naturally necessary. He rejected all eorts to dene causes on the basisof ecacy, agency, power, force, energy, necessity, connexion, and productive quality. 11 Humecontendedthat all wecansayabout causationmust bebasedonour experiencesof theempirical world. ThisledHumeto theconclusionthat, sincethere is nothing that can be observed directly about causal connections, we cannot attributecausal relations any reality beyond our observations. The notion of cause, for Hume, arose simply from human observations of constant conjunctions of events. Heargued that when regular successions of types of events havebeen observed, themind through custom comes to associatethese eventsinsuchawayasto createtheillusionarybelief inacausal connection. When wehaveobserved that billiard ball B has regularly moved after ball A has hit it, we can say that A is the cause of Bs movement. There is nothing more to the causal connection between theseobjects, however, or wecannot say whether thereis, since wehaveno observational evidenceof deeper causal connections. I t followsthat for Hume[causal] necessity is something that exists in themind, not in objects. . . or necessity is nothing but that determination of the thought to pass fromcauses to eects and fromeects to causes, according to their experiencd union. 12 Humes inuential solution to theproblemof causation entails certain important assumptionsthat needtobedrawnout. TheHumeanphilosophyof causation, which has been deeply entwined with the empiricist tradition in modern philosophy, has entailed thefollowing assumptions. 1. Causal relationsaretied to regularities and causal analysisto ndingassociations between patterns of regularities. 2. Causal relationsareregularity relationsof patternsof observables. Asempiricism dictatesthat onlyobservableevents/thingscanbethebasisof knowledge, causality has been reduced to a relation of observables. 3. Causal relations are seen as regularity-deterministic. 13 Most Humeans have assumed that, given certain regularities have been observed in the past, we can makewhen A, then B statements about therelations of certain types of events (given regularities wecan assumetheexistenceof closed systems). 10 E. Chavez-Arvizo, I ntroduction, in Descartes, Key Philosophical Writings, p. ix. 11 David Hume, A Treatise of Human Nature (Oxford: Clarendon Press, 1978), p. 157. 12 Hume, Treatise, pp. 1656. 13 Thetermregularity-determinismwas coined by Roy Bhaskar. Roy Bhaskar, A Realist Theory of Science (Hassocks: Harvester Press, 1978), p. 69. 192 Milja Kurki 4. Beyondthesestrictlyempiricist assumptions, it hasalso beenassumedthat causes refer to moving causes that push and pull, that is, so-called ecient causes. Theseassumptions arereferred to hereas Humean assumptions. I t will beseen that they underlie many contemporary engagements with causation in philosophy of science, social scienceand I R, even if often unsystematically or inadvertently. Humes legacy in the philosophy of science and the philosophy of social science The most inuential advocates of Humeanism in twentieth century philosophy of science can be found within the so-called positivist tradition. I ndeed, at the beginning of the twentieth century the so-called logical positivists transformed the Humeanpremisesintoaphilosophyof sciencegearedaroundthenotionof laws. By basing scienceon theanalysis of laws (based on observed event regularities), it was believed that sciencecould get rid of all metaphysical speculation. 14 Poppers and Hempels later deductive-nomological (DN-) model explanation followed the same Humean lineof thought: although falsication and deductivemethods werepriori- tised over verication and inductive methods, the basis of scientic theories was conceived to lie in generalised patterns of observables. Without generalisations to back it upacausal account wasseenasunscienticandasmerespeculation. 15 I ndeed, in the course of the twentieth century causal explanation has become closely tied up with analysis of general laws: science has come to be understood to be about nding falsiable, predictive, observation-based regularities, or generalisations. I nterestingly, Humean assumptions havebecomeso widely accepted that they have been increasingly taken for granted in most philosophy of sciencedebates. 16 I t isimportant to notethat Humeanassumptionshavebeendominant, not just in thenatural sciences, but also in thesocial sciences. TheHumean assumptions were givenrst trulysystematic guisebythebehaviourist social scientists: causal analysis inthesocial sciencesbecameequatedwithlookingfor associationsamongst patterns of observed behaviour. 17 Since the 1960s many social scientists have criticised the strict regularity assumptions and quantitative methods of the 1960s behaviourists. Yet, arguably the key assumptions of Humeanism still hold sway in many social sciences through the inuence of so-called post-behaviourist positivism. Social 14 Humean assumptions wereaccepted, however, with a distinct anti-causal twist. Logical positivists argued that sincewecan only legitimately talk of regularities of events, weshould refrain from using causal terminology. I nstead, they talked of functionally determinate relations between laws. See, for example, A. J . Ayer, Logical Positivism (Glencoe, I L: TheFreePress, 1959), R. Carnap, Logical Foundations of Probability (London: Routledge& Kegan Paul, 1950). 15 K.R. Popper, The Logic of Scientic Discovery (London: Hutchinson of London, 1959), C. G. Hempel (ed.), Aspects of Scientic Explanation and Other Essays in the Philosophy of Science (New York: FreePress, 1965). 16 Theinuential growth of knowledge debates on philosophy of sciencewere, for example, implicitly underpinned by Humean assumptions. Although thelogical positivist and Popperian models of scientic progress havecomeunder criticismfromphilosophers such as Thomas Kuhn, I mre Lakatos and Paul Feyerabend, theseattacks havenot challenged theHumean notion of cause embedded in theseaccounts of scientic progress. 17 For a classical logical positivist/behaviourist viewof social sciences seeMarieNeurath and Robert S. Cohen (eds.), Otto Neurath: Empiricism and Sociology (Dordrecht, TheNetherlands: D. Reidel Publishing, 1973). The concept of cause in IR theory 193 scientistsnowaccept thelegitimacy of qualitativemethodsand data, yet most social scientists are still adamant that only careful observation of regularities (even if of localised regularities) can giveus an adequateunderstandingof human action and society. 18 Of course, the hermeneutic tradition in the social sciences disagrees with both behaviourist and post-behaviourist forms of positivism and argues that mere descriptionof patternsof behaviour explainsnothingabout whypeopledowhat they do: what weshould do isanalysemorecarefully howpeoplecometo understand the meaning of social situations. Causal descriptions of the social world have been rejected by thehermeneutictheoristsasinvalid in theinterpretiveunderstanding of subjects. Social science, as Peter Winch famously argued, is about studying the reasons, not causes, of actions. 19 Theinternal relations between meanings, rules, reasonsand actions, it isargued, cannot betreated in thesameway astheexternal relations of events. I t follows that social relations do not lend themselves to generalisation and prediction in the same way as the relations that the natural sciences study. 20 Assessing human behaviour from the point of view of general patterns of behaviour misses out the crucial role that rules and reasons play in constituting themeaningful context of social action. Thedisagreements over thelegitimacy of thenotion of causehavegiven riseto a sharp dichotomisation of reasons and causes, understandingand explaining, as well as causal and constitutive forms of inquiry in the social sciences. However, it is crucial to noticethat thehermeneuticrule-followingaccountsof thesocial worldare also basedonaHumeanunderstandingof causation. Rejectingacausal approachto the social world has been relatively easy for these theorists because they have uncritically accepted the positivist Humean understanding of causation as charac- teristicof causal analysis. Winch, for example, rejectscausesbecauseassumptionsof lawfulness andwhenA, thenB (regularity-determinism, ecient causation) donot seemtoapplyinthesocial world. 21 I ndeed, oneof thekeyproblemsinthephilosophy of social sciencesisthat, becauseof the(ofteninadvertent) acceptanceof theHumean assumptions as thebaselinefor evaluatingcausal approaches, therehas been little engagement with alternative ways of thinking about causation. As we will see, the disciplineof I R has reproduced theseproblems by drawing on theterms of debate between positivist and interpretive approaches to justify the present divided disciplinary self-image. HumeanisminInternational Relationstheory Thegoal of thissectionis, rst, toidentifytheHumeanassumptionsoperatingwithin contemporaryI R theory. Humeanassumptions, it isseen, characterisemost contem- porary I R theorists understandingsof causation. However, it isalso recognisedthat 18 C. Frankfort-Nachmias and D. Nachmias, Research Methods in the Social Sciences (London: Edward Arnold, 1992); King, Keohaneand Verba, Designing Social Inquiry. 19 Peter Winch, The Idea of Social Science and Its Relation to Philosophy (London: Routledge, 1990). 20 Charles Taylor, Philosophy and the Human Sciences (Cambridge: CambridgeUniversity Press, 1985), p. 56. 21 Winch, Idea of Social Science, p. xii. 194 Milja Kurki evenwhenHumeanassumptionsareaccepted, morecommon-sensical non-Humean causal terminology can bedetected in I R theorising. Thelatter sections will seek to clarify themeaningof theimplicit non-Humean causal languagethrough opening up thenotion of cause. Humeanism and scientic causal analysis in IR Arguably, a particular kind of orthodoxy has dominated the way in which causal analysis has been thought about in contemporary I R theory: orthodoxy dened by the advocates of a positivist, or as some term it, an empiricist, model of social science. 22 I t is argued herethat thepositivist/empiricist mainstreamin I R is deeply informed by Humean assumptions, even though it does not necessarily advocate Humeanismin thehard form(exemplied by themorebehaviourist approaches). 23 I t is seen that the acceptance of Humeanismcreates some problems and inconsist- encies in themainstreamapproaches to causal analysis in I R. To understand howHumeanism functionsin I R at present, King, Keohaneand Verbas methodological thesis Designing Social Inquiry will beexamined here. This book has not only outlined the premises of social scientic causal analysis in an admirably systematicmanner, but hasalso becomevery inuential asaguidinglight of causal analysis in political science and I R. I ndeed, most I R theorists and researchers in theAmerican mainstreamseek to followprecepts for causal analysis that areinlinewithKing, KeohaneandVerba: not simplytherationalist neorealist and neoliberal theorists that haveused an empiricist framework for sometime, but also researchers driven by morehistorical interests. 24 King, Keohane and Verba attempted to bring cohesion and order into social scienticinquirybyoutlininghowweshouldconduct validcausal analysis. Causality for King, Keohaneand Verba is measured in terms of thecausal eect exerted by an explanatory variableon a dependent variable. They proposethat wemeasure causal eect as thedierencebetween thesystematic component of observations madewhenanexplanatoryvariabletakesonevalueandthesystematiccomponent of 22 Positivismis understood hereto refer to thoseapproaches that (1) believein a scientic method that is applicableacross sciences and hence(2) assumenaturalism, (3) empiricism, (4) believein value-neutrality of scientic method and (5) emphasisetheimportanceof instrumental (predictive) knowledge. Gerard Delanty, Social Science: Beyond Constructivism and Realism, Concepts in Social Sciences (Buckingham: Open University Press, 1997), p. 12. Empiricismis morenarrowly an epistemological approach to theconstruction of knowledge(through empirical observation). However, empiricist epistemology is understood to bea crucial ingredient of a positivist approach to science. 23 Examples of theharder Humean approach can beseen advocated explicitly in American journals, such as theJournal of Conict Resolution. Many democratic peacetheorists can beseen as examples of hard Humeanismbecauseof their statistical approach. See, for example, Z. Maoz and B. Russett, Normativeand Structural Causes of Democratic Peace19461986, American Political Science Review, 87:3 (1993), pp. 62438; R. I . Rummel, Democracies AreLess WarlikeThan Other Regimes, European Journal of International Relations, 1:4 (1995), pp. 45779; Russett, Grasping the Democratic Peace; Principles of Post-Cold War World (Princeton, NJ : Princeton University Press, 1993). 24 Theneo-neo contenders sharea common, arguably, empiricist conception of scienceas highlighted by Baldwin. D. A. Baldwin (ed.), Neorealism and Neoliberalism: The Contemporary Debate (New York: Columbia University Press, 1993), p. 9. This conception is arguably largely compatiblewith King, Keohaneand Verbas precepts. For an application of King, Keohaneand Verba in a more historical inquiry see, for example, Randall L. Schweller, Deadly Imbalances: Tripolarity and Hitlers Strategy of World Conquest (NewYork: Columbia University Press, 1998). The concept of cause in IR theory 195 comparableobservationswhentheexplanatoryvariabletakesonanother value. 25 I n other words, whenweassesscausal relationswemeasuretheaverageeect that the changing value of an explanatory variable has for the dependent variable. King, Keohane and Verba acknowledge that we can never completely securely rerun explanatory variables against dependent variables (as if in controlled experiments) but argue that through a careful observation of some central rules of causal inference falsiability, consistency, careful selection of dependent variables, max- imisation of concreteness and of encompassing qualities of theories 26 we can minimise disturbances in causal explanations of the social world. I mportantly, King, KeohaneandVerbaarguethat therulesandlogicof causal inferencethat they advanceapply equally to quantitativeand qualitativeinferences. King, Keohane and Verbas model of causal analysis is steeped in Humean assumptions, althoughthismight not beimmediatelyobvious. First, causal relations are seen as relations between observables or patterns of observables. I mportantly, King, KeohaneandVerbawarnagainst utilisingconceptsthat cannot beempirically operationalised in testingof theories. This raises thequestion; what is thenatureof thecausal relation for King, Keohaneand Verba? Theseauthors do not talk about real (ontological) relationsbetween things: causality for themisan epistemological concept, and relations between observables (events, things) logically, rather than naturally, necessitating relations. As for the key Humean assumption of regularity, although strict behaviourist requirements of quantiability have been left behind, King, Keohane and Verbas account still works on thebasis of theexpectation that thequalitativevariables will beexpressedinquantiableterms. Also, it isarguedthat thelarger thesamples, even in qualitative inquiries, the better the reliability of the inquiry. 27 Further, generali- sation isprioritised over theparticular: too much concentration on thecomplex and theunique, it is argued, dampens theeciency of theexplanation, and accounting for too many contributory factors lowers the mean causal eect of the key variable. 28 I mportantly, while King, Keohane and Verba acknowledge that social scientists can, and sometimes do, concentrate their study on the so-called causal mechanisms of social life, it is argued that accounts of causal mechanisms must always bepremised on theidentication of appropriateempirical variables. 29 The wide acceptance of King, Keohane and Verbas understanding of causal analysis in themainstreamof I R, and hencethereproduction of Humean assump- tionsinthediscipline, hasthreeimportant implicationsthat must bedrawnout. First, the acceptance of Humean causal analysis has led to an empiricist formof causal analysis being advocated as thenorm. Methodologically, this has entailed a certain degreeof rigidity: King, KeohaneandVerbasconceptionof causal analysisdoesnot allowfor amultiplicityof dierent typesof evidencetobeappreciated. Qualitativeor historical data, for example, arenot evaluated on their own terms but aremadeto conform to the regularity criteria. On the other hand, methods such as discourse analysis aresidelined as they cannot bebent to t in with theempiricist regularity- drivenassumptions. Besidesgeneratingarather methodologicallyrigidconceptionof 25 King, Keohaneand Verba, Designing Social Inquiry, pp. 812. 26 I bid., pp. 99114. 27 I bid., pp. 20830. 28 I bid., pp. 104, 1823. 29 I bid., p. 86. 196 Milja Kurki causal analysis, theempiricist criteria for scientic causal analysis put forward by King, Keohaneand Verbaalso imply theepistemological superiority of thisformof gaining knowledge, an assumption that has been strongly criticised by many post-positivists and onethat is, indeed, far fromstraightforward. Moreover, Humeanism has also entailed adoption of particular ontological assumptions. The mainstream empiricists have often not been able to focus on explaining why event-regularities come about. Crucially, because explanation is conceived to take place through the analysis of the logical relations of observable variables, thesetheoristshavenot been very interested in formingunderstandingsof thedeepontological structures, processesandconditions(theunderlyingunobserv- able causal powers), which would provide so-called depth explanations of the patterns of observables identied. Thus, empiricist explanations of, for example, democratic peace have mostly been focused on the analysis of the observable independent variables and their logical relations (given patterns in quantitativeor qualitative data), rather than on conceptualising the complex deep ontological social relations that underlietheempirically observed sets of variables. 30 Also, as a result of prioritisingobservability, theempiricists social ontologieshavetendedtobe atomistic, that is geared around methodological individualism. 31 Third, it isalso important to notethat thereissomethingof aparadox within the empiricist approaches. Whiletheempiricistsclaimthat causation can only betalked of when strict epistemological and methodological rules of the positivist model of scienceareemployed, causal languagein a much more(for want of a better word) common-sensical manner can also bedetected in most mainstreamcausal analyses. I ndeed, if we start paying attention to causal terminology in a wider everyday sense to words such as because, leads to, produces, makes, enables and constrains we can see that a great deal of broader (but only implicitly causal) terminologyisat work inempiricist I R theorising. Althoughvalid causal theorising is seen as that backed up by observed regularities, I R theorists also make more common-sensical assumptions about theproductiveconnections between things. I t is seen in the latter part of this article that these common-sensical causal statements demonstrate that causal analysis in a deeper non-Humean sense is possibleand inltrates even theHumean frameworks. I n this sense, themainstream scientic accounts of causal relations are not necessarily the nal word on causation, nor necessarily as internally consistent as is often thought. Constitutive theorists: another case of Humeanism The constitutive, sometimes also called reectivist, theorists in I R are, as opposed to mainstream scientic causal analysts, wary of causal terminology. Causal 30 Especially evident in Maoz and Russett, Normativeand Structural Causes, pp. 62438. 31 Waltzs work, for example, is geared around methodologically individualismand, also, closed system regularity-deterministic logic. Regularity of war is logically deduced fromassumptions about structure premised on an individualistic understanding of states as actors. Kenneth Waltz, Theory of International Politics (London: McGraw-Hill, 1979). For an excellent critiqueof Waltzs individualismseeAlexander Wendt, Agent-StructureProblemin I nternational Relations Theory, International Organization, 41:3, (1987), pp. 33570, Alexander Wendt, Anarchy I s What States Makeof I t, International Organization, 46:2 (1992), pp. 391425. The concept of cause in IR theory 197 descriptionsareseenasproblematicbythesetheoristsbecausecauses, for them, tend to imply deterministicand materialisticexplanations. Thesetheoristsliketo conduct constitutive theorisingand research instead. Thetermconstitutive has a rangeof meanings. I t is argued, for example, that ideas constitute themeaning of material forces. Others seerules, norms and discourses as sets of constitutive forces in that they providetheconstitutive framework within which human actorsthink and act. Beyond this, it is also accepted by many constitutive theorists that our theories about theworld do not simply reect theworld but arealso constitutive of social reality. Byopeningupthisnewwayof talkingabout thesocial worldtheconstitutive theorists have opened up new important avenues of inquiry in I R. However, it is important to note that causation is rejected by these theorists on curious grounds. Reectivists tend to favour constitutive descriptions over causal ones becausethey tend to associatecausal analysis with theempiricist Humean viewof social inquiry. When critical theorists, for example, refer to causes they do so only when criticising positivist causal analysis. Robert Cox, for example, argues that the concept of cause is applicable strictly to the positivist framework and that his historical explanation cannot be equated with causal explanation since causal explanation cannot capturethecomplexityof thesocial worldasthehistorical mode of analysis can. 32 Causal analysis is associated with the ahistorical neorealist frameworks and the scientic claims of objectivity of the mainstream. Causal analysis, then, is understood in accordance with Humean assumptions and, as a result, rejected altogether. However, aswewill see, weshouldresist thesimpleconclusionthat Coxsaccount is void of causal concerns. I t should be kept in mind that Coxs account of world politics seems to bebased on careful outlining of forces material, ideational and institutional that produce and shape theworld order and agents actions within it. However, Cox describes the layered and interacting structural forces, not as causes, but aspressuresandconstraints. 33 I t couldbearguedthat to theextent that thisterminologyimpliesaproductive meaning, andisdrawnuponto explainwhy thingshappenincertainwaysrather thanothers, Coxismakingcommon-sensically or implicitly causal claims. However, becauseCox associates causation with positiv- ism, he does not recognise his own implicit interest in causal forces that shape the world. Poststructuralists also harbour a deep dislike of causation. On the basis of the poststructuralist critiqueof knowledge, J enny Edkins, for example, argues that the notions of causeand eect areuntenable. 34 Moreover, sheargues that lookingfor causes hasresulted in inadequateresponsesin thedealingwith particular problems inworldpoliticsand, hence, talk of causesshouldbeavoidedfor practical reasonsas well as philosophical ones. Processes of technologization and depoliticization can beseen in international politics itself, as well as in thedisciplinethat studies it. Oneexampleof this is found in responses to famines, humanitarian crises, or complex political emergencies. Agencies and governments outsidethecrisis area do not takeaccount of thepolitical processes that areunder way, of 32 Robert Cox, Realism, Positivismand Historicism, in Robert Cox and Timothy Sinclair (eds.), Approaches to World Order (NewYork: CambridgeUniversity Press, 1995), p. 51. 33 Cox, Social forces, states, and world orders, in Cox and Sinclair (eds.), Approaches, p. 98. 34 J enny Edkins, Poststructuralism and International Relations: Bringing the Political Back In (London: LynneRienner, 1999), p. 15. 198 Milja Kurki which thecrisis is a symptom. I nstead, they rely on interventions derived fromabstract, technical analysis of thesituation, onethat looks for causes, not political reasons or motivations. 35 While Edkins voices a fair criticism of some political science approaches to humanitarian emergencies, her statement gives an unfair portrayal of thenotion of cause. Causal analyses can, of course, have adverse consequences for the way in which concrete problems are dealt with. However, it must be noted that Edkinss assessment itself depends on an implicitly causal understanding of the situation: presumably the political processes of which the crisis is a symptom, and the contextual political reasons or motivations, areinfact the(real) causesthat should beaddressed in order to deal with thesituation. Becauseof her seemingly positivist understanding of causation, Edkins rejects the concept too swiftly and, thereby, ignores her own implicit causal claims. The same paradox characterises David Campbells work. I n Writing Security Campbell declaresthat theinterpretivepositionheassociateshimself withisopposed to cataloging, calculating and specifying the real causes ; 36 instead, Campbell maintains that his poststructuralist theory aims to inquire into the political consequences of adopting one mode of representation over another. 37 While appearing anti-causal, his statement evidences an implicit causal commitment: representations matter precisely because they produce certain consequences. This understandingof representationsand discoursescan beseen ascausal, even if not in a when A, then B manner. 38 Because Campbell, as other reectivists, associates causationwiththemainstream HumeanisminI R, hedoesnot recognisetheimplicit causal claims in his own work. Constructivist oscillation Constructivists have not rejected the notion of cause as readily as many critical theoristsandpoststructuralists. Thereisatendencyinconstructivist work tooscillate between reasons and causes accounts, and between causal and constitutive theorising. However, the terms causal and constitutive seem to lack coherent meaning for many constructivists. Nicholas Onuf provides a good example. Onuf thinks there is something to the notion of cause, but he also wants to resist accounting for social action in merely causal terms. Onuf wants to givea special meaning to theintentionality, rules and constitution of the social world that cannot, for him, be understood through a causal approach. 39 However, neither the concept of causation, nor the notion of 35 Edkins, Poststructuralism, p. 910. 36 Campbell, Writing Security, p. 4. 37 I bid., p. 4. 38 I n National Deconstruction, for example, Campbell argues that theontopology of binding together of territoriality, statismand mono-culturalismin Western liberal discourses has had somecrucial implications on howtheWest viewed and dealt with thesituation in Bosnia. David Campbell, National Deconstruction: Violence, Identity and Justice in Bosnia (London: University of Minnesota Press, 1998). 39 Seehis discussion of Bhaskars reasons as causes account. Nicholas Onuf, World of Our Making: Rules and Rule in Social Theory and International Relations (Columbia, CA: University of Southern California Press, 1989), p. 49. The concept of cause in IR theory 199 constitution, areclearly dened and causal concerns, in theend, largely drop o the agendaasthenotionof constitutiverulesisgivenpriorityinanalysinghowrulesand norms work. Similar trends can be detected in the work of other constructivists such as Friedrich Kratochwil. Kratochwil attacks the mono-causal neorealist theorising and its inappropriate concept of causality. 40 However, despite his rejection of outright Humeanism, what alternativeassumptions about causation entail remains unclear and, as a result, causation, again, disappears fromthe theoretical agenda. Norms, for example, arenot seenascausal but rather asconstitutive. 41 Ruggie, too, despiterecognising ideational causation, continually contrasts causal explanations with the so-called constitutive non-causal explanations. 42 Yet, it remains unclear what hends causal about causal explanations and non-causal about constitutive explanations. While they do not reject the concept of cause outright, these constructivists arguably continue to attach certain deterministic and materialistic connotations to the notion of cause and, as a result, remain unclear about its role in their own explanations. Oneof theonly openly constructivist theorists to takesteps towards a clearer understandingof causationhasbeenAlexander Wendt, whoseeortstobuild a deeper account of causewill bediscussed in thenext section. The failure of many I R theorists empiricist, reectivist and constructivist to giveadequateemphasisto theconceptualisationof thenotionof causehashadsome crucial eects on thedisciplineof I R. Crucially, theHumean discourseof causation has had an overwhelmingly inuential rolein directingtheassumptions attached to thenotion of cause. Thewideacceptanceof Humean assumptions on causation has ledto theempiricist formof causal analysisbeingtreatedastheonlyacceptableform of causal inquiry in thediscipline. This, in turn, has resulted in thedichotomisation of scientic causal and reectivist constitutive(non-causal) approaches in I R. As theempiricist scientists haveinsisted on theneed for systematic causal analysisas dened by them(on Humean terms), thepost-positivist constitutivetheorists have rejected thevalidity of causal analysis altogether in an eort to avoid being forced into astraightjacket conceptionof howto analysesocial aairs. Also, asaresult of the acceptance of Humeanism, theorists have not given adequate attention to the manycommon-sensical assumptionsat work intheir theorising. I t isarguedherethat thesecommon-sensical assumptionsreveal that causation can, and in fact should, be thought of inamuchdeeper andbroader waythanisrecognisedbythecontemporary causal or non-causal theorists. I n order to solve the tensions and confusions that contemporary I R, with its Humeanframingof causation, isweddedtowemust rethink andopenuptheconcept of causealtogether. Thenext section arguesthat weshould accept thedeeper notion 40 R. Koslowski and F. Kratochwil, Understanding Changein I nternational Politics: TheSoviet Empires Demiseand theI nternational System, in T. Risse-Kappen et al. (eds.), International Relations Theory and the End of the Cold War (NewYork: Columbia University Press, 1995), p. 136. Seealso F. Kratochwil, Constructing a NewOrthodoxy? Wendts Social Theory of I nternational Politics and theConstructivist Challenge, Millennium, 29 (2000), pp. 73101. 41 Koslowski and Kratochwil, Understanding Change, p. 137. 42 J . G. Ruggie, Constructing the World Polity: Essays on International Organization (London: Routledge, 1998), p. 34. Seealso J . G. Ruggie, What Makes theWorld Hang Together? The Neo-Utilitarianismand theSocial Constructivist Challenge, in Krasner et al. (eds.), Exploration and Contestation in the Study of World Politics (Cambridge, MA: MI T Press, 1999), p. 229. 200 Milja Kurki of causeadvocated by thephilosophical realists. I t is then argued that wemust also broaden the notion of cause away from the pushing and pulling ecient cause conception of causation. By taking these two steps we can comprehensively avoid Humean assumptions and radically recongurecausal analysis in I R. Philosophical realismandadeeper conceptionof cause This section argues that to escape from the problem-eld that has plagued I R theoristson all sideswith regard to causal analysis, weneed to rst adopt adeeper ontologically grounded conception of causation advanced in the philosophy of science, and in I R, by theso-called philosophical realists. Philosophical realists aim to put forward a new ontological framework for thinking about the objects of science, which in turn necessitates a reconguration of the epistemological and methodological parameters of scientic causal analysis. Thephilosophically realist literaturehas already been drawn on in I R by certain key gures such as Alexander Wendt, David Dessler, Heikki Patomaki and Colin Wight. This section seeks to explicate why the philosophically realist turn is important in redirecting causal analysis, whilealso pointing out why weneed to go beyond thedeepening of the meaning of causation. Towards a deeper conception of cause Theturn towards philosophical realismhas been an important development in the recent philosophy of science. I t has had wide ranging implications within the philosophy of science, social science and I R. This is because philosophical realism aims to, and by and large succeeds, in solving a number of seemingly intractable problems and debates in modern philosophy and social science. What does philosophical realismas a general philosophy of sciencecontributeto our understanding of causation? First, philosophical realismhas been important in that it hasallowedustoreclaimanontological conceptionof causationthat hasbeen lost for threehundred years or so. Realist philosophies of scienceand social science havehad as their aimtherefocusing of debates in thecontemporary philosophy of scienceand social scienceon ontological questions. 43 As a result, thephilosophical realists, importantly, advancearadically anti-Humean ontological understandingof causation: causes, therealists argue, can be, or indeed, must beassumed to exist as real ontological entities, that is, they arenot merecreations of our imagination, but have real existence in the world outside our thought and observations. 44 Causal analysis, then, is about analysing causes out there (outside what we think or observe), an assumption rejected by both Humean empiricists and many reectivist sceptics. 43 Notethat philosophical realismshould not beequated with theI R tradition of realism, which being based on empiricist assumptions in many cases is, in fact, largely anti-realist. 44 RomHarreand Edward H. Madden, Causal Powers: A Theory of Natural Necessity (Oxford: Blackwell, 1975). The concept of cause in IR theory 201 Second, in reclaiming the ontological meaning of the concept of cause, the philosophically realist conception of cause allows us to transcend the regularity- dependence of Humeanism. For philosophical realists, regularities, although accepted as possibly indicative of underlying causal structures, are deemed neither necessary nor sucient for establishing a causal explanation. 45 To grasp the real underlyingcauses(why somethinghappens), realistsargue, weneedknowledgebased on various types of evidenceand, importantly, a conceptual framework that allows usto conceptualisethereal (ontological) unobservablecausal powersthat arebehind observableevents(or regularitiesof events). Thus, whileobservedregularitiesarenot thrown away altogether, they aregiven aradically dierent rolein thisnon-Humean approach to causal analysis: they areonly oneformof data amongst many and in themselves cannot providescientically objective causal analysis. Third, philosophical realists also challenge the regularity-determinism of the Humean empiricist model of causation. The realists emphasise that causes exist outsideclosed systems and that theworld, in fact, consists of open systems, where multiple causes interact and counteract each other in complex and, importantly, unpredictableways. Thus, thecentral focus of causal analysis is not theanalysis of isolated independent variables (through statistical methods), but rather understand- ingthecomplex interaction of a variety of dierent kinds of causal factors (through thebuilding of conceptual frameworks). Furthermore, causation is dened much moreopenly, or common-sensically, by thephilosophical realists. Causes aredened rather loosely as all thosethings that bringabout, produce, direct or contributeto statesof aairsor changesintheworld. This allows us to reclaimthediversepragmatic causal languagein common useand reectstheubiquity of causal analysisin our everyday lives. Causal analysis, then, is not something that is uniquely abstract and scientic: rather scientic causal analysis is arenement and extension of what wedo in thepractical functioningof everyday life. 46 These philosophically realist arguments have important implications for the analysisof thesocial world. Philosophical realistswho concentrateon social inquiry (often called critical realists) 47 reject theterms of debatein much of thephilosophy of social science by arguing that the philosophy of social science has been deeply informed by a misleading positivist stanceon science. Thecritical realists, in linewith thegeneral philosophically realist critiques, reject the applicability of empiricist observation based scientic inquiry and the closed system model of explanation. As a result, they aim to recongure radically philosophy of social sciencedebatesaway fromthepositivist vs. hermeneutic theory dichotomy. Notably, the reasons vs. causes debate is recongured by the critical realists. Thecritical realistsarguethat, whenwedisentanglethenotionof causefrom theHumean regularity-deterministic model of causation, wecan accept that reasons are, in fact, a type of cause. Critical realists argue that just because humans are intentional, meaningful and human action reasoned this does not mean that our 45 Roy Bhaskar, A Realist Theory of Science (Hassocks: Harvester Press, 1978), p. 12. 46 J . Lopez and G. Potter, After Postmodernism: TheMillennium, in J . Lopez and G. Potter (eds.), After Postmodernism: an Introduction to Critical Realism (London: AthlorePress, 2001), p. 9. 47 Short for realist critical naturalism Roy Bhaskar, Possibility of Naturalism: A Critique of the Contemporary Human Sciences, 3rd edn. (London: Routledge, 1998), p. 28. 202 Milja Kurki actions, and the rules and reasons that inform our actions are non-causal (or uncaused) . 48 Onthecontraryideas, meaningsandreasonsareimportant inthesocial world precisely becausethey arecausal. 49 I t follows that interpretivemethods, far frombeing anti-causal or non-scientic, areseenasnecessaryinorder toconduct social scienticcausal analysis: hermeneutic, historical and qualitative methods are seen as fundamentally important in getting to grips with the complex nature of social relations. Critical realists advance a methodologically pluralist approach to social scienceaccepting thevalidity of both extensive statistical methodsand intensive qualitativeand interpretivemethods. 50 Critical realists also avoid the epistemologically objectivist tendencies of the empiricists. Although realists accept that theworld is characterised by ontologically real things and processes, they accept that in coming to knowthoseforces, wewill always be inevitably informed by the social and political context that we inhabit. I ndeed, all knowledge about the world is deeply constrained and enabled by the linguistic conventions, conceptual systems and thesocial-political backgrounds that weknow within. Thismeansthat scienceisnever purelyobjective. However, science is not relativistic either becauseour knowledgeis always of something, that is, our accounts of theworld arenot merely imagined but makeprojections about really existing ontological objects, relations and processes. 51 Philosophical and critical realismallowus to deeply challengethedominanceof thetraditional positivist model of science, and thetaken-for-granted natureof the Humean model of causal analysis attached to it. They introducea new ontological approach to causal analysis and, in so doing, transcend the epistemological and methodological deadlocks betweenthetraditional contendersinthesocial sciences. Theadvancement of realist ideas in I R has been very important in redirectingI R theory. Theworks of Wendt, Dessler, Patomaki and Wight haveopened important new theoretical and empirical avenues in I R. 52 Wendt has demonstrated that the philosophically realist logic can be used to bridge the gap between rationalist and reectivist theorising in I R, while Dessler has demonstrated that philosophical realism directs us towards more integrative analysis of world political processes. Patomaki andWight, on theother hand, havedemonstratedthedeepembeddedness of I R theoretical approaches in an anti-realist problem-eld that has weakened I R theorising ontologically, epistemologically and methodologically. Notably, all these theorists have challenged the taken-for-granted conception of science in I R and, hence, have directed I R theorisations towards more ontologically and epistemologically reectiveand methodologically pluralist frameworks. Thesetheorists havealso madeimportant contributions in rethinking causation: Wendt and Dessler have emphasised the analysis of causal mechanisms over 48 Paul Lewis, Agency, Structureand Causality in Political Science: A Comment on Sibeon, Politics, 22:1 (2002), pp. 1723. 49 AndrewSayer, Method in Social Science: A Realist Approach (London: Routledge, 1992), p. 111. 50 See, for example, Sayer, Method in Social Science. 51 Heikki Patomki, After International Relations; Critical Realism and the (Re)Construction of World Politics (London: Routledge, 2002), p. 79. 52 See, for example, Heikki Patomki and Colin Wight, After Post-Positivism? ThePromises of Critical Realism, International Studies Quarterly, 44 (2000), pp. 21337; Patomki, After International Relations; Alexander Wendt, Social Theory of International Politics (NewYork: CambridgeUniversity Press, 1999); David Dessler, Beyond Correlations: Towards a Causal Theory of War, International Studies Quarterly, 35:3 (1991), pp. 33755. The concept of cause in IR theory 203 regularitiesinanalysingworldpolitics, andhaveraisedconcernsabout therigidityof thecausal-constitutivetheory dividein I R. Patomaki and Wight, on theother hand, havearguedfor moremulti-causal inquiriesthroughadvancingthenotionof causal complex. However, more could be done to develop the philosophically realist conception of causation. I ndeed, the next section will point out that we should go beyondadvocatingdeeper analysisof causationby coherently broadening out the meaning of causation. This provides the philosophically realist approaches with a sharper focus in analysing themultiplicity of social causes. Beyond the deeper conception of cause? Whileprovidingauseful correctiveto thedominant positivist model of scienceinI R, and, hence, enablingtheopeningupof newconceptual andmethodological avenues, the philosophically realist critiques have not gone far enough in challenging the Humean discourse of causation in philosophy of science, social science and in I R. Thisisbecausethephilosophical realistsseemto havebeen unnecessarily wedded to theecient causeunderstandingof causation. Aswill beseen, morecanandshould be done to open up systematically the meaning of the notion of cause beyond the pushing and pulling ecient causemetaphor. When we examine the philosophically realist accounts more closely, we can see that many of them retain a belief in a pushing and pulling understanding of causation, the so-called ecient cause understanding of causation. For example, Harreand Madden, thetheorists behind theturn towards thestudy of ontological causal powers, denecausation and causal powers squarely through themetaphor of ecient cause. Causation, for them, always involves a material particular which producesor generatessomething, that is, [powerful] particularsareto beconceived ascausal agents. 53 Thisfollowscloselythepost-Cartesianassumptionthat, whenwe talk about causes, we only talk about pushing and pulling causes: causation is dened by the ability of objects to bring about change through agential action. I mportantly, because they accept this assumption of causal powers as agential movers, HarreandMaddenhaveacceptedthat inthesocial worldtheonlyimportant causal forceis theactive human action by individuals. 54 Somecritical realists in Roy Bhaskars tradition havechallenged this reduction of social causalityto activehumanaction becausethisisperceivedto leadto methodo- logical individualism. As a result, some Bhaskarian critical realists have started to open up themeaningof thenotion of causeaway fromtheecient causeconnota- tions. First, theseBhaskarian critical realists haveargued for a wider conception of ecient cause, onethat encapsulatesnot onlyhumanaction, but also ideas, rulesand reasons: thesetoo cause statesof aairs, althoughnot necessarilyinawhenA, then B manner. 55 Second, however, thecritical realists havealso cometo arguethat the onlywayinwhichwecangraspthecausal natureof factorssuchassocial structures is by accepting that social structures do not necessarily push and pull, rather they 53 Harreand Madden, Causal Powers, p. 5. 54 Paul Lewis, Realism, Causality and theProblemof Social Structure, Journal for the Theory of Social Behaviour, 30: 3 (2000), pp 2557. 55 Bhaskar, Possibility of Naturalism, p. 89, Sayer, Method in Social Science, p. 111. 204 Milja Kurki constrainandenable. Toconceptualisethesenon-pushingandpullingtypesof causes the Bhaskarian critical realists have utilised the Aristotelian metaphor of material cause. As Paul Lewis explains: J ust as a sculptor fashions a product out of therawmaterials and tools availableto him, so social actors producetheir actions out of pre-existing social structure. Likethemediumin which thesculptor works, pre-existing social structurelacks thecapacity to initiateactivity and makethings of its own accord social actors aretheonly ecient causes or prime movers in society but it does aect thecourseof events in thesocial world by inuencing theactions that peoplechooseto undertake . . And by inuencing thebehaviour of social actors, pre-existing social structuremakes a dierenceto and henceexerts a (material) causal inuenceover social life. 56 The opening up of the possibility that there might be other types of causes than ecient causes in social life is important and promising. However, the broader conception of causation has not been developed fully by thephilosophical realists. The use of the Aristotelian material cause analogy, for example, is not adequately developed as many would question thesimilarity of material and social structural causes. Also, thereis no explicit acceptanceamong thephilosophical realists of the general principle that when we talk about causation we are actually talking about many dierent types of causes, nor real willingness to develop broader categorisa- tions of dierent types of causes. Thus, thedierent ways in which ideas, discourses and reasons, for example, cause are not examined but subsumed under the now rather broad ecient causeheading. I f therearedierent types of causes at work in social life, why should wethink only in terms of material and ecient causes? I t is argued in thefollowing section that it is, indeed, useful to drawon theAristotelian account to develop a more pluralistic understanding of causation, but that this broadening out should bedonemoreconsistently and holistically. Furthermore, theimplicationsof awider conceptionof causeshouldbedeveloped in more detail in the I R theoretical context. While the followers of philosophical realismin I R haverethought causation on deeper lines, and havebeen sceptical of mono-causal explanations, thesetheoristshavenot sofar focusedondrawingout the implications of broadening out thenotion of causein I R context. AninterestingexceptioninthisregardisAlexander Wendt who inhisrecent work hasshowninterest inexploringthebroader conceptionsof cause. I mportantly, inhis article Why the World State is I nevitable, Alexander Wendt turned to the Aristotelian notion of cause in order to elucidate a teleological logic for the development of theworld state. 57 WhileWendt focused on developingthenotion of nal cause, he also pointed out that parallels can be drawn between constitutive analyses in I R and theAristotelian causal categories. 58 Thus, Wendt has opened up the possibility of broadening out the notion of cause for the purposes of I R theorising. The following section seeks to take further Wendts reections by systematically exploring theimport of theAristotelian philosophy of causation for thepurposes of I R theorising. 56 Lewis, Agency, pp. 2021. 57 Alexander Wendt, Why a World Stateis I nevitable, European Journal of International Relations, 9:4 (2003), pp. 491542. 58 Wendt, Why A World State?, p. 495. The concept of cause in IR theory 205 Aristotlerevisited: broadeningtheconcept of cause Whilebeingtheoldest andmost famousaccount of causation, Aristotlesphilosophy of causationhasbeenlargelyforgottenduringthelast centuries. Thissectionseeksto showthat if werevisit thebroader Aristotelian logic of causal explanation wegain a radically recongured understandingof causal analysis philosophically and for the purposes of I R theorising. I t is accepted here that, contrary to what Hume, the empiricists and even many philosophical realists assume, causation is not a single, monolithicconcept 59 and, hence, causal analysisinvolvesthecareful identicationof various dierent types of causes and understanding their complex interactions. Aristotles four causes account As was seen in therst section, theoriginal meaning of theword cause, theGreek wordaition, didnot haveaprecisemeaninginthesensethat modernphilosophyhas tried to establish. An aition was anything that contributed in any way to the producingor maintainingof acertainreality, or whatever onecouldciteasananswer to a why-question. 60 Crucially, for Aristotle, dierent causes material, formal, ecient and nal cause in dierent ways. Aristotlesawecient causes (by which something is made) and nal causes (for the sake of which something is made) as active or extrinsic causes that cause by lending an inuence or activity to the producingof something. Ontheother hand, anintrinsiccause, for Aristotle, wasthat which causes through constituting an object or thing. 61 Within his framework of four causes (constitutivecauses of reality) could bethought to consist of material causes (material out of which something is made) and formal causes (ideas or relations accordingto which somethingis made). Aristotlesawtheworld as shaped through the complex interaction of all these dierent types of causal forces. To explainwhyanychangeor thinghascomeabout onewouldneedto refer to all these dierent categories of causeand therelations between them. TheAristoteliancategoriesof thinkingabout themeaningof causationallowusto open up themeaningof thenotion of causeand to exploretheplurality of meanings of the concept, something that has been pushed aside since Descartes narrowing down of the concept of cause. What do the categories mean and how can we use themto understand thesocial world in better ways? Material causes, for Aristotle, werea fundamental part of any explanation in the sensethat all accountsof theworld would haveto refer to thematter out of which things come to be. Material causes simply referred to the passive potentiality of matter as a type of cause that enables and delimits possible ways of being or changing. I mportantly, in the Aristotelian framework the notion of material cause has dierent meanings in dierent explanatory contexts: thus, whilethings such as a table or a gun, can be treated as material causes in one instance, these things 59 Nancy Cartwright, Causation: OneWord, Many Things, Philosophy of Science, 71 (2004), p. 805. 60 Lear, Aristotle, p. 6. 61 S. Waterlow, Nature, Change and Agency in Aristotles Physics (Oxford: Clarendon Press, 1982), p. 11. 206 Milja Kurki can also be understood to have material causes themselves in the constitution of substances (wood, metal). 62 What does this notion of material cause contribute to our understanding of causation?Therst contributionof theAristotelianunderstandingof material causes is that it points us to recognise that material causes are fundamental in any explanation. I t directsusto accept that without accountingfor material potentiality, and thevarious forms that matter takes, any account of theworld is limited. At the sametime, theAristotelian notion of material causealso allows us to usematerial causesasaexiblecategorythat refersto awiderangeof material substances, things andresourcesandallowsustoconceptualisethesematerial resourcesasconstraining andenabling causes, not asmechanical pushingandpulling causesoftenimpliedin modernmaterialist accounts. Thisframingisuseful inthesocial sciencesasit getsus away fromcompleterejection of material factors (exemplied by idealist strands of thinking) as well as the deterministic overtones often attached to more materially based explanationsof thesocial world. To givean examplefromI R context, wecan recognise that the availability of guns in a crisis situation is an important causal factor that conditionstheconict, whilerealisingthat thismaterial causeinitself does not determine outcomes, nor does it provide an adequate explanation in and of itself. I ndeed, in order to understand thenatureand roleof material causes weneed to consider threeother types of cause. Formal causes, for Aristotle, referred to that which shapes or denes matter. For Aristotle, aformal causeisthat which makesor denesagiven thing, itsstructure, its qualities and its properties. I n modern discourse formal causes are often understoodideationally (inthePlatonicsense), that is, aformistakentorefer tothe idea of athing. Whilethisis, indeed, avalidinterpretation, it isuseful to remember that Aristotelian formal causes werenot dened by ideationality alone, but rather by relationality (which ideas can reect): formal causes describe and dene the structure or internal relations that give meaning and being to things. I f the material cause of a table is the wood it is made of, the formal cause of it is the structure (embodied in the idea of a table) that denes the relationship between pieces of wood to makeit into a table. Why are formal causes useful for our understanding of causal relations? First and foremost, because it seems that in the social world, ideas, rules, norms and discourses often interpreted as non-causal constitutive forces can usefully be understood through the notion of formal cause. Rules, norms and discourses, are causes, then: asformal causes, they deneand structuresocial relations, that is, they relateagentstoeachother, their social rolesandthemeaningsof their practices. They describe the rules and relations that dene social positions and relationships, and hence can be seen as that according to which social reality works. Crucially, the Aristotelian conception of formal cause allows us to understand rules, norms and discoursesasconstrainingandenabling causes, andgetsusawayfromthepushing and pulling model of framingthecausal roleof ideas, rules, normsand discourses themodel that thereectivist constitutivetheoristsinI R havealwaysbeenwaryof. Crucially, both material and formal causes break the mould of modern causal analysis in thesensethat they do not conformto thecommonly elicited assumption 62 F. A. Lewis, Aristotleon theRelation between a Thing and its Matter, in M. L. Gill (ed.), Unity, Identity and Explanation in Aristotles Metaphysics (Oxford: Clarendon Press, 2001), p. 248. The concept of cause in IR theory 207 that causes should temporally precede and beindependent of eects. I t could be argued that these two criteria often advanced by Humeans, as well as by some philosophical realists, 63 confuse more than they clarify in the light of the analysis advanced here. This is because these criteria, which arguably have their origins in Humeanqualicationsfor howto distinguishbetweencausesandeects, 64 restrict us from accepting as causal certain important conditioning or constitutive causal powers. The key contribution of the Aristotelian notion of constitutive causes, exemplied by material and formal causes, is that it directs us to accept the constraining and enabling conditions of social life as real and as causal, thus exposing a deeper level of causality in the social world than is often recognised, especially by thosefocused on analysing merely observablepatterns of behaviour. However, there is also place for the traditional active causes within this framework. The active causes, as opposed to conditioning causes, for Aristotle, were ecient and nal causes, as these causes, through their activity, go towards producingchange. Aristoteliannotionof ecient cause refersto aso-calledprimary mover, or a sourceof change, for example, a carpenter as themaker of a table. I mportantly, Aristotelian ecient causality does not have modern mechanistic overtonesasecient causes, for Aristotle, werefundamentallyembeddedwithin, and in relation to, other types of causes and could not in and of themselves explain anything. Conceiving of the causal actions and causal conditions of agency in the Aristotelian manner is useful in that the Aristotelian conceptualisation of ecient causesgetsusaway fromthemechanisticwaysof thinkingof agency aswell asfrom theindividualist tendencies to isolateagents as theonly typeof causein thesocial world (agency is conceived as embedded in a complex causal social environment, material and formal). Also, it is useful because it allows us to link ecient causes closely with so-called nal causes. Final causes, for Aristotle, referred to the ends and purposes that go towards makingthingshappen, tothat for thesakeof which somethinghappensor isdone. Final form of causality was, for Aristotle, an irreducible form of causality in the social aswell asthenatural world. 65 Manywoulddoubt theapplicabilityof teleology in natural sciences. One might also doubt the kind of teleological explanations of social processesasoutlined, for example, byWendt. 66 However, whether oneaccepts these forms of nal causality or not, in one simple sense nal causes seemlike an inherently important type of cause in the social world. Social action, even when unplanned and spontaneous, is inevitably premised on the intentionality of human agency, which in turn can beseen asaformof nal causality. 67 When wetalk of the intentions, motivationsor, incertaincontextsreasons, that direct actors, weareinfact referring to thenal causes becauseof which certain (ecient) actions aretaken. 63 Wendt, Social Theory, pp. 7788. 64 For Hume, causes had to beobserved independently fromeects, and causes wereidentied as dening themas thoseobservables that wereobserved beforetheeects. 65 E. Gilson, From Aristotle to Darwin and Back Again: A Journey in Final Causality, Species and Evolution (I ndiana, I N: NotreDamePress, 1984), p. 5. 66 Wendt, Why theWorld Stateis I nevitable. 67 D. V. Porpora, On thePost-Wittgenstein Critiqueof theConcept of Action in Sociology, Journal for the Theory of Social Behaviour, 13 (1983), pp. 12946. For a similar account of reasons as nal causes seeRuth Gro, Critical Realism, Postpositivism and the Possibility of Knowledge (London: Routledge, 2004). 208 Milja Kurki Crucially, acceptance of the Aristotelian conception of nal causality does not necessitateamechanistic or deterministicunderstandingof causality. I ntentionsand reasons should not be conceived to push and pull in the same sense as ecient causes: instead, they are causal in the sense that they signify a contributory cause, that for thesakeof which something is done. I t must also benoted that, against commonmisconceptions, acceptingthenotionof nal causalityinthissensedoesnot downgradeother typesof causality. Final causalitypresupposesmaterial causalityof themind as well as a material world to act upon. I t presupposes a formal relational social context (rules, norms, discourses) that constrainsand enables theformingof intentions. An(ecient) agent andactionsarealso requiredto actualise intentions/ purposes/goals. Towards causal holism and explanatory pragmatism The Aristotelian categories, arguably, provide us with interesting new ways to describe and analyse causes. I mportantly, the rethought conception of cause advancedhereallowsustoget awayfromHumeanandpushingandpulling modern conceptions of causation by outliningdierent ways of causing(constitutiveas well as active). They also allow us to position and assess causes as complex and interacting. I ndeed, Aristotleimportantly stresses that in inquiring into any change or thing, wemust alwaysask manydierent kindsof why-questions: inquiringmerely into singular causes tells us littlein most cases, sincecauses never exist in isolation fromeach other. I t follows that thebroadening of our understanding of causation allowsusto advancean ontologically holistic framework for causation. By allowing usto look into avarietyof dierent kindsof causal factorsthisrethought framework of causal analysis allows us to ask much moreopen and plural questions about the social worldandabout theinteractionof dierent kindsof actors, objects, discourses and structures. The Aristotelian system also allows us to advocate explanatory pragmatism. Social life, aswell asnatural life, can, rst, beconceivedasworkingthroughmultiple cycles of causes. Second, the Aristotelian system recognises that the multiple cyclesof causescan betreated fromdierent anglesdependingon onesexplanatory interests: what we assign as causes, and which types of causes, can be seen as a questiontiedtoour pragmaticexplanatoryinterests. Thus, onemight beinterestedin explainingtheformationof anorm, whichwouldentail inquiryinto thespeechacts of actorsin aparticular social (material andformal) context, but thesamenormcan in a dierent context betreated as theformal cause(of an action, for example). 68 I t canalsobeacceptedthat eventhoughcausesarereal andubiquitous, our causal accounts do not need to be treated as objective or xed. This ts in with the requirements of philosophical realism according to which we must accept that all 68 However, importantly, simply becauseweassign certain things as causes for our explanatory interest does not makefactors outsideour accounts non-causal. Wemerely designatethemas unimportant background causes for our explanatory interests. This is similar to what themanipulability theorists argue. SeeR. G. Collingwood, Essay in Metaphysics (Oxford: Clarendon Press, 1940). For an interesting manipulability account of causation in I R, seeSuganami, On the Causes of War (Oxford: Oxford University Press, 1996). The concept of cause in IR theory 209 knowledge claims are socially embedded and dependent on the conceptual and linguistic categoriesweinherit. I t can, then, beaccepted that claimsto epistemologi- cal objectivity aremisleading. However, if weaccept philosophical realism, weneed not accept relativist conclusions: we do not have to accept that all accounts are equally valid since we have to, and do, make evaluations between dierent conceptual/theoretical systems on the basis of their ability to account for evidence and to put forward plausibleontological projections. Also, theaccount advanced hereis commensuratewith methodological pluralism as emphasised by philosophical realists. The methodological tools used in social science should remain non-specied: we need many dierent ways of studying the ontologically complex social world. I t is accepted that quantitative methods can point to someinterestingpatternsof observabledata. However, in order to explain thesepatternsof data, qualitativeanalysisisnecessaryasthisdataallowsustoaccess theactual causal processes that takeplacein morenuanced ways. Theemphasis of therethought deeper and broader account of causeis on asking many typesof causal questionsand refusingto delimit methodsand epistemological approaches a priori. This points causal theorising in I R towards a new direc- tion which, in turn, has important implications for the disciplinary self-images reproduced in I R. Implicationsfor IR: beyondHumeanism, beyondthecausal-constitutivedivide The simultaneous deepening and broadening of the notion of cause has important implications for thedisciplineof I R. First, it allows us to rethink theway in which wethink about, andconduct, causal analysisinI R. I t alsodeeplychallengesthelogic of thecausal-constitutivetheory dividein I R. Furthermore, in so doing, rethinking causation can help direct I R theorisations away from theoretical reductionism towards more constructive holistic understandings of concrete world political processes. Beyond Humean causal analysis in IR First of all, therethinkingof causation advanced hereremindsusthat theempiricist modeof causal analysis is not theonly way to framecausal analysis and, in fact, is methodologically, epistemologically and ontologically problematic in important ways. Theapproach defended heremaintains that theessenceof causal explanation is not thegatheringof regularities, but conceptual explanation of thevariety of forces that bring about regularities of observables. I t follows that analysis of data is methodologicallypluralist, not gearedaroundspecicobservation-basedmethods, or identication of regularities. Emphasis is on combining empirical data sets of various kinds and, through them, thedevelopment of reectivecomplexity-sensitive conceptual (ontological) frameworks. I t also follows that more holistic integrative causal explanationscan beprovided. Whereasvariableswereisolated and compared 210 Milja Kurki against eachother intheempiricist Humeanmodelsof causal analysis, nowtheycan bebrought together into integrativeexplanatory frameworks. I n many ways themodel of causal analysis advanced hereis commensuratewith the more common-sensical causal explanations and descriptions we give. Since causal analysis is not dened by certain types of systematic observation and generalisationbut rather throughreexiveuseof avarietyof data, it isacceptedthat words such as produce, enable, constrain, bring about, push and direct usefully describethemany dierent kindsof causal connectionsthat therearein the world. Hence, to reducecausal explanation to mechanisticmetaphorsor relationsof independent and dependent variables is to overly restrict our understandingof the complex social reality around us. I t also follows that causal explanations in I R cannot simply rely on an empiricist conception of scientic methods to deliver an objectivescientic truth on causal relations. Wehaveto besceptical of theepistemological condenceof theempiricist causal theorists in the superior objectivity of observational methods and, indeed, recognise that the positivist model of social science is in itself limited in its ontological, epistemological and methodological scope. Crucially, the approach to causal analysisadvancedhereallowsustoaccept that all causal accountsof theworld are, asCoxandthepost-positivistswouldhaveit, alwaysfor someoneandfor some purpose. 69 Also, the approach advanced here is important because it challenges the tradi- tional ontological framings of world political processes. Against the mainstream followersof King, Keohaneand Verba, observability isnot theonly, nor necessarily a useful, benchmark for what matters in causal analysis in I R: as the reectivists havepointed out, only so much can beexplained through thestudy of measurable variables. On thebasis of thediscussion hereit can beaccepted that unobservable objects are real and causal and can be got at through careful conceptualisation. Reasons and motivations as well as rules, norms and discourses can beconceptual- ised as real and as causal within this framework and, hence, can beaccepted as legitimateobjectsof social scienceinquiry, evenif theyarenot directlyobservableor stable in termsof empirical outcomes. Theapproach here, then, challengesnot just the empiricist approach but also reectivism. I t emphasises that when Humean criteriafor causal explanationarerejected, wecanseethat thereectivistsare, infact, involved in making a number of causal claims. Beyond the causal-constitutive theory divide I t can beargued on thebasis of thereconceptualised conception of causeadvanced herethat thecausal-constitutivetheorydichotomythat manyI R theoristshavecome to accept is misleading and could, in fact, be radically reinterpreted. Constitutive theorising can, in the light of the present analysis, be interpreted as a form of causal theorising. I t follows that it is not useful for I R theorists to reify the traditional causal vs. constitutivetheory logic in understandingdierent approaches to I R theory. 69 Robert Cox, Social Forces, in Cox and Sinclair (eds.), Approaches, p. 87. The concept of cause in IR theory 211 I f weaccept that causes, not onlypushandpull, but condition or constrainand enable, we need to start recognising that there are some serious problems in the causal vs. constitutive theory self-image in I R. This is because we have to start recognisingthat constitutiveanalysisis, moreoftenthannot, engagedinthestudyof the(formal) conditioning causes of social life. When, for example, theconstructiv- ists talk of theconstitutive norms and rules because of which shifts takeplacein worldpolitics, theyareengagingincausal analysisinthat theyarecontextualisingthe agents actions within a formal causal context which shapes theagents perceptions and thinking processes. 70 Equally, when poststructuralists highlight the role of discourseor theories in constituting social life, they do so becausethesediscourses or theories, through constituting agents perceptions and reasoning, have conse- quences for how agents perceive the world, themselves, others and, hence, their actions. When Campbell, for example, studies the Bosnian war, the constitutive discourses referredto (for exampleterritorial discourse) do not merely matter inthe senseof deningmeanings but inthesensethat theyconstrainandenable courses of action 71 constitutivediscoursesare, in fact, causal. AsCampbell himself admits, discursivesystems havepolitical consequences. 72 I n the view of the model of causal analysis advanced here, causal relations and causal analysis should not, in fact, be separated from constitutive relations or constitutiveanalysisat all. I t followsfromour reconceptualisation of causation that wewould bebetter placed to deal with thesocial world and its complex causes and causal conditioningif wesawconstitutivetheorisingasan inseparable part of causal theorising. Reectivist criticsmayndthisargument disturbing. After all, for them, inquiring into constitutive relations is very dierent fromcausal analysis becauseit inquires into relations of meaning not relations of change. On one level this criticism is correct. Of coursewecan say that when concepts deneeach others meaning they are non-causally related. Also, we can, of course, ask non-causal questions of meaning (such as what does X mean?). However, in the view of the approach accepted here, weneed to recognisethat thestudy of thesenon-causal conceptual relations is never really an end in itself: 73 conceptual relations that makeup rules, discourses and norms matter ultimately precisely because they are causal, that is becausetheyproduce, contributeto, direct or constrainandenableontological social realities. I n thenal analysisit seemsthat theconstitutiverelations matter because conceptual relations causally condition thoughts, actions and relations. I t follows that we must recognise that our inquiries are never limited to mere non-causal understandingof meanings: most theorists, includingthepoststructuralists, want to, andinevitablydo, account for howthosemeaningsweremade, reproducedor reied and how they shape, inuence and condition other meanings/discourses/ideas and social lifemorewidely. The unique contribution of the Aristotelian framework is that it allows us to accept that constitutive claims areessentially inseparablefromcausal claims. This also means that the causal-constitutive theory divide in I R no longer stands as a fundamental divide as both sides of the divide can, in the light of this analysis, be 70 Koslowski and Kratochwil, Understanding Change, p. 127. 71 D. Campbell, National Deconstruction, p. 84. 72 Campbell, Writing Security, p. 4. 73 Wendt, Social Theory, p. 86. 212 Milja Kurki interpreted asbeingengaged in causal analysis even if causal analysisconceived of in a very dierent sense. The traditional causal theorists must be recognised to be engagedwithinaparticular, insomewayslimited, formof empiricist causal analysis (focused on patterns of observables), whereas the constitutive theorists can be understoodtobeengagedindeeper andbroader realist causal analysis(oftenfocused on theanalysis of discursive/ideational causes). The Aristotelian account builds bridges in another sense too. I t is important to notethat constitutive conditioningcausesconcern, not onlyconstitutivetheorists, but also those explicitly interested in the role of material conditions. I n I R realist theorists have traditionally explained world politics through resort to material factors. However, often thewrongkind of causal logic has been applied to material forces, just as to ideational forces. I nstead of framing material forces as akin to ecient pushing and pulling causes, as has arguably been thecasein many realist explanations, 74 it is implied herethat they should beseen as forms of conditioning causes in social life, whoseroleneeds to beunderstood in relation to other types of causes in any given context. Of coursematerial resources matter, for they condition much of international politics, but material resources must be recognised to be constitutedthroughsocial processesinvolvingsocial actorsandsocialisingprinciples (formal causes) and, indeed, to lend their inuence dierently in dierent causal contexts. Manyrealistshavebeenunwillingto examinetheroleof material resources as causally complex. As a result, they have failed to account for how the material causesaredeterminingof outcomes(howtheyinuenceideas, rulesandnorms). The Aristotelian framework that sees material resources as important and as causal, but as causally conditioning (constitutive) and as intertwined with other causal factors, canbeusedtoovercomesomeof thereductionist materialist tendenciesinI R theorising. Towards new kinds of causal explanations of world politics? The conceptualisation of causation advanced here radically opens up how we should think about causal analysis and the role of constitutive factors in our analysis. This insight poses a deep challenge to the divisive causal vs. constitutive theoryself-image perpetuatedbymanyempiricists, reectivistsandconstructivist in I R, aswell astheontologically, epistemologically and methodologically reductionist tendencieswithinthesetheoretical approaches. Theimpact of thereconceptualisation of causation is not merely metatheoretical, however. This is because meta- theoretical framings of explanatory frameworks have direct eects on the kinds of explanations we advance for concrete world political processes: indeed, theoretical and conceptual lenses constrain and enable (causally direct) the kinds of explanations wecan construct. Theconceptual lenses advanced herearemoreopen andholisticthanmanyof thoseadvancedbyI R theoretical camps. I t followsthat the approachheredirectsI R theorisationstowardsnewmoreopenandholisticavenues. First, it directs empiricist researchers in I R away from mere statistical and observational analysis towards the construction of integrative and holistic 74 As is implied in Waltzs regularity-deterministic account for example. Seefn. 31. The concept of cause in IR theory 213 explanatory systems. Thus, it emerges that the study of democratic peace, for example, should not beconducted merely on thebasis of traditional taxonomical (procedural) understandingsof democraciesand through measuringtherelevanceof observed variables (democracy, wealth, alliances, culture) against each other. Rather, the focus becomes the construction of holistic and integrative frameworks where many types of conditioning causes, from material constraints of capitalist social relations to the ideological congruence of Western cultures, can be brought together to provide explanations of social and historical dynamics. Because the questions asked aremoreholistic and themethods used morepluralist, wecan, not only devisemoreholistic explanations, but also uncover newlevels/areas of relevant social realities. For example, proxy wars or patriarchal power relations within states cannot simply be dened away as irrelevant variables, but can be seen to providean important part of aholistic structural understandingof democraciesand their embeddedness in thewider world system. 75 Also theI R theoretical camps exclusionary approachto social explanationscan be countered. Too often I R theorists have avoided engaging with each others theoretical frameworksinexplainingconcreteworldpolitical processes. Think of the explanations of theend of theCold War, for example. This complex world political process has been explained in rather reductionist and theoretically incommensura- ble terms by the realists and the constructivists, one emphasising material and structural determinants, the other normative ones. 76 The key to providing better explanations of theend of theCold War, and to initiatingmoreconstructivedebate between theoretical schools, lies with the abandonment of the beliefs that single ontological factors(ideas, material concerns, agents, structures) explainanevent, and that causal factors are independent. 77 The approach to causal analysis advanced hereenablesmuch moreopen and multi-causal questionsto beasked, which in turn necessitates a turn away fromtheoretically reductionist explanations. Theorists and researchers are, instead, directed towards providing accounts where the complex interaction of norms and material constraints are analysed in a holistic and historically attuned manner. Themodel of causal analysis advocated here, by advancingcausal pluralismand by rejecting themetatheoretical persuasiveness of thecausal vs. constitutivetheory divide, holdsopenthepossibilityfor newkindsof integrativeandholistictheoretical engagements with world political processes. Although the deeper and broader conceptualisation of causation does not in itself solve any of the concrete causal puzzles in I R, it opens up newlines of inquiry and newways of dealing with them. As well as justifying theuseof a variety of methodological tools and accepting the epistemological reectivity of scientic causal analysis, it also forces I R theorists to ask morepluralistic questions about dierent types of causal forces and about their 75 For an appreciation of theimportanceof thesewider analyses see, for example, Tarak Barkawi and Mark Laey (eds.), Democracy, Liberalism and War: Rethinking the Democratic Peace Debate (London: LynneRienner, 2001). 76 Compare, for example, Koslowski and Kratochwil, Understanding Change and R. D. English, Power, I deas, and theNewEvidenceon theCold Wars End, International Security, 26: 4 (2002), pp. 7092, to S. G. Brooks and W. C. Wohlforth, Power, Globalization and theEnd of theCold War; Reevaluating a Landmark Casefor I deas, International Security, 25:3 (2000), pp. 553. 77 A point also madeby Gaddis. J . L. Gaddis, On Starting All Over Again: A NaveApproach to the Study of Cold War, in O. A. Westad (ed.), Reviewing the Cold War: Approaches, Interpretation, Theory (London: Frank Cass, 2000), p. 28. 214 Milja Kurki interaction. Thus, philosophical rethinking of causation is not just a fanciful meta-theoretical exercise but has the potential to redirect the concrete study of world political processes in signicant ways. Conclusion I n 1998 David Campbell famously attacked theparochial and hegemonic main- stream I R and its eorts to suppress critical work, either by denouncing it as anti-scientic or by coopting identity issues within themainstreamcausal analysis that reduces identity issues to study of independent variables. 78 Campbell argued that it was thescientic conception of causal analysis and theeorts to extend this formof inquiry to all questions that is at the core of the problems of the divided discipline. Thisassessment isuncannilycorrect: not onlywasCampbell correct topoint tothe deep and politicised nature of the disciplinary divisions in I R, but he also rightly identied the conception of causal analysis as a problem in the division of the discipline. Yet, Campbell himself didnot seek awayout of thispredicament. I nstead, being caught up in the Humean discourse himself, Campbell through his critique perpetuated thedichotomies in thediscipline. Thedisciplinary politics of I R can be criticised but not through criticisingthemainstreamapproach to causal analysisin isolation, or supercially, by rejecting causation. A morecomprehensivecritiqueis attained through addressing thewider metatheoretical groundings of I R 79 in which theHumean conception of causeplays a crucial role. The reconceptualisation of causation advanced here takes steps towards doing away with someof themisleadingand unhelpful ways of theorisingthesocial world that havebecomeuncriticallyacceptedincontemporaryI R and, thereby, holdsopen thepossibilitythat I R theoristscanstart talkingto, rather thanpast, eachother. The belief that causal and constitutive approaches are fundamentally dierent is an illusion created by theacceptanceof a Humean empiricist concept of cause. I n the light of theconceptionof causeadvancedhere, theempiricist conceptionof causation is ontologically, epistemologically and methodologically overly restrictive. I t is also seen to beconstrained within straightjacket of theecient cause metaphor. The reectivists and constructivists, on the other hand, have been equally constrained by Humean and ecient cause assumptions, in that they reject causal analysis on thesegrounds. However, when interpreted through a wider lens wecan see that, rather than being a-causal, they are, in fact, deeply engaged in causal analysis: although causal analysis concentrated on examining theconstraining and enabling inuence of norms, rules and discourses. The metaphor of constitution has been applied in such a way in I R as to hide the causal nature of ideas, rules, norms and discourses. I t has been argued here that, whether we consider ideas constitutiveof our identities, theory as constitutiveof practice, rules as constitutive of action or structures constitutive of things, constitutive relations are intimately tied up to causal relations. 78 Campbell, Writing Security, pp. 20727. 79 SeePatomaki and Wight, After Post-Positivism?, pp. 21337. The concept of cause in IR theory 215 I nadisciplineconstructedonthebasisof themetatheoretical groundingadvanced here, political and theoretical dierences will no doubt remain, but theoretical insulation between causal and constitutive approaches, and between dierent theoretical campsfocused on dierent kind of causal factors, becomesmoredicult to justify. I f weseecausal analysis as pluralistic, complexity-sensitive, epistemologi- cally reective and methodologically open, the constraints and prejudices that the Humeandiscourseof causation, alongwiththepositivist model of science, hasplaced uponI R theorisingcanbelifted. Onthebasisof areconceptualisedconcept of cause wecan, arguably, openuppathstowardsmoreconstructiveconsistent andpragmatic debates on causation and, indeed, on world politics. 216 Milja Kurki
Critiquing Anthropological Imagination in Peace and Conflict Studies From Empiricist Positivism To A Dialogical Approach in Ethnographic Peace Research
Ψυχήas Differentiated Unity in the Philosophy of Plato Author(s) : Robert W. Hall Source: Phronesis, Vol. 8, No. 1 (1963), pp. 63-82 Published by: Stable URL: Accessed: 15/08/2013 18:25