My Foucault, my genealogy

In terms of the broader debate on the status of socio-historical knowledge, Foucault can be read in two ideal-typical ways. In the first, Foucault comes off as an arch-opponent to epistemology, a slash-and-burn anti-foundationalist, a ‘denier’, a ‘prophet of extremity’ and a ‘nihilist’.Footnote ⁹ In nominalist ontology, any research tool is always a social item and therefore subject to genealogy, rather than to some definition. Here, the point behind genealogy is therefore not a codification of distinctive techniques of inquiry, but an attempt to open the intellectual and political space for resistance to the dominant regimes of truth and the emancipation of marginalised forms of knowledge. The proponents of this reading point out how Foucault himself described his research tools as ‘non-defined method’ or ‘anti-science’.Footnote ¹⁰ Consequently, genealogy can thus never be a methodology, only an ‘analytics’ aimed at liberation.Footnote ¹¹

The ‘antiscientific’ reading of genealogy appears plausible: if there is no epistemic truth (or inherent truth's value for that matter), only the regime or politics of truth, then it follows that all judgments are political and/or ethical. Geneology is by necessity a political-ethical critique of values and Foucault's own genealogies formulated no shortage of significant and subtle political and ethical lessons.Footnote ¹² But what matters here is that Foucault drew these lessons on the basis of certain epistemic truth-claims, not on the basis of some unfettered, subjective opinion. In this sense it is misleading to say that Foucauldian genealogy rejects epistemological universalism in favour of political and/or ethical expressions. In the rest of this section I want to argue that genealogy not only accepts the possibility and utility of historical representation, but that it also stands in line with traditional social science and its methodologies.

Let us begin with Linda Alcoff's ‘Foucault as an epistemologist’ perspective.Footnote ¹³ In this interpretation, while Foucault denied that there can be such thing as a scientific truth defined as propositional correspondence with reality, he nonetheless accepted that there can be a scientific truth that explicitly recognises the temporal, spatial and cultural domains in which the regimes of truth are made. To paraphrase one of Felipe Fernández-Armesto's titles, truth has a history, too.Footnote ¹⁴ It is here that we can situate Foucault as a minimalist, local or commonsense foundationalist, ‘suspicious’ and a ‘modest’ epistemologist and, last but not least, as a self-declared ‘happy positivist’.Footnote ¹⁵ Quite apart from politics, the aim of this version of genealogy is to produce what can be called the truth with adjectives – ‘local’, ‘contextual’, ‘situational’, ‘secondary’, ‘regime’, ‘relative’, ‘procedural’, ‘working’, or ‘small-t’.

From Foucault the epistemologist there is only a short step to Foucault the methodologist. To drive home his favourite point that research should be problem-driven, not method-driven, Foucault frequently insisted that he only wrote about methodology, never a methodology.Footnote ¹⁶ It was in this context that his toolbox metaphor gained fame and commonplace currency in history and social sciences: ‘I would like my books to be a kind of toolbox that people can rummage through to find a tool they can use however they wish in their own area […] I don't write for an audience, I write for users, not readers.’Footnote ¹⁷ If anything unifies this toolbox it is the concept of discourse – an anonymous collective consciousness within the human language through which meaning is given to subjects and objects.Footnote ¹⁸ In turn, virtually all Foucauldian research tools embody and the idea that history is the recovery of discourses and discursive contexts.

Let us briefly sketch the two main elements of this historical-philosophical perspective. First, discourses constitute subjects – subjectivities, subject positions, identities – in two continuous ways: they enable/constrain what is ‘thinkable’ in a given discursive context and they reward/punish ideas, institutions, and practices that are congruent/deviant with the pre-set political boundaries of inclusion/exclusion. The thinkability thesis is not Foucault's alone – think of the earlier histories of geocentrism, the platypus and the like – but it Foucault who linked thinkability to subjectivity. His histories of European asylums, sexuality, prisons, medicine and policing showed not only how discourses materially restricted reality and knowledge, but also how they ordered and re-ordered citizen populations and other subjectivities.Footnote ¹⁹ Equally important in this approach are the notions that discourse and its contexts are mutually constituted and that power implies resistance. Here, discourses might be cumulative and aggregating, but new developments in the discursive and extra-discursive contexts can and do impose on and, to varying degrees, transform old meanings. Subjectivising power is never final because each context is characterised by multiple discourses – some discourses are dominant (hegemonic, governing, ruling), while others as challenging (counter-hegemonic, resisting, alternative or subaltern). This multiplicity means that discourses are open to mutual contestations, critiques and, ultimately, change; for one, every discourse implies expectations about future courses of action, which are, in principle, always open-ended.Footnote ²⁰

This particular historical-philosophical perspective is the context for the well-known ‘three E’ definition of genealogy – a reference to episodes, examples and effectiveness. The first two features are common. Episodes – also known as historical series, epochs, events or moments – are meant to perioditise the history of an object under study. Each episode is narrated though examples, which are themselves reconstructed in a discourse analysis of historical documents ranging from diaries left behind by marginal authors to programmatic or pioneering statements on how to maintain social order. In Discipline and Punish, Foucault offers that a key episode in the development of the contemporary ‘carceral’ society is the eighteenth century debate between two discourses of punishment disciplinary/corporal vs. reformist. The famous example analysed within this episode was Jeremy Bentham's Panopticon – an essay which prescribed an efficient disciplinary penal regime. The Panopticon refers to a prison where individual cells surround a central control tower producing a sense of constant surveillance and control – the panoptic ‘gaze’. In Foucault's analysis, these disciplinary micro-tactics represented a distinctly modern type of power, which modified the behaviour of society without the use of physical force.Footnote ²¹

The rationale behind using these particular episodes and examples should be familiar to qualitative researchers versed in the theory of case studies. Though Foucault never explicitly offered selection criteria for his episodes and examples, some of the users of his toolbox have concluded that such criteria must follow more-or-less standard social science casing techniques, even if only implicitly. For one, there is no logical distinction between the word case and the word episode as both parcel history into discrete moments.Footnote ²² More important, to the extent that one purpose of genealogy is to demonstrate the diversity and specificity of battles between different interpretations of social items, purely random selection of episodes and examples is self-defeating. In short, to the extent that genealogy serves to show how the reality could have been different if certain social and political constructions in history had been different, then it must rely on a comparative method.

As numerous readers have noted, Foucault's own episodes and examples are not randomly chosen, but as hard/least likely, easy/most likely, maximum variation or anomalous/extreme/deviant/outlying or paradigmatic.Footnote ²³ The judiciousness of empirical shifts the classical to modern episodes or from the examples of Bentham's Panopticon to a prison warden's private diary suggest that Foucault paid a great deal of attention to the variability in the scope, depth, and duration of social phenomena under investigation. Drawing from Foucault's own work, it follows that specifying or even standardising the selection criteria for episodes and examples does not necessarily clash with genealogy's critical aims. The mechanics of writing genealogy are similar to an interpretative study based on historical case studies: episodes break into examples which in turn relate back to the overall analysis. When it comes to case selection writ large, the between genealogy and other qualitative-interpretative social science methods can be seen as practical, rather than principled.Footnote ²⁴

The last of the ‘three E’ – effectiveness – is not as straightforward from the perspective of mainstream social science methodologies. In brief, effective genealogies are those that focus on a ‘problem’ – a social phenomenon that appears (seems, feels) normal or true (commonplace, natural, intuitive) and then turns it into a question, that is, it asks how it came about in the light of contingency and power.Footnote ²⁵ Here, the purpose of research is not ‘accurate’ correspondence between history and objective reality, but ‘problematization’ of the socially constructed reality and knowledge – the ‘disruption’, ‘de-stabilization’, ‘devaluation’ or ‘deciphering’ of what we know or, to paraphrase the first epigraph, our belonging to history. Put differently, the effectiveness of genealogy depends on two factors: first, the ability to foreground how the conventional assumptions about the world have implications for our present-day reasoning and, second, to suggest, directly or indirectly, alternative ways to constitute the aspect of humanity under study. In turn, each task hinges on the ability to analyse and critique power relations.Footnote ²⁶

So defined, effectiveness was Foucault's main goal in his many attempts to ‘intervene’ in the past to ‘diagnose’ the present.Footnote ²⁷ His historical analyses of clinics, prisons, asylums and hospitals in European history can all be seen as effective because they each clearly demonstrated how and why some subjects and social items were brought about and not others, what became forgotten and with what consequences for the present. It was in Discipline and Punish, his first book-length genealogy, that Foucault discovered how ‘power and knowledge directly imply one another’.Footnote ²⁸ His argument that knowledge is not simply socially constructed but also co-constituted with power became known simply as ‘power/knowledge’ (pouvoir-savoir), Foucault's main claim to fame. In the same study, Foucault identified genealogy as a ‘history of the present’, a definition that led to much confusion, grammatical and otherwise, but the phrase implied that genealogy serves to show the contingency of the present, against biological necessity, rational design, teleology, liberal progress narratives and similar schemas once present in the putative traditional history.Footnote ²⁹

In The History of Sexuality, Foucault noted that classical Greek civilisation recognised only sexual acts, not sexual identities, and that the subsequent evolution of human sexuality followed the vagaries of history, not biology.Footnote ³⁰ Here, Foucault borrowed his theory of history from Nietzsche, who famously argued that ‘origins’ are never metaphysical (origin-as-Ursprung), but always contingent (origin-as-Herkunft/Entstehung), such that subjects and social items evolve in and through circumstances that come together in particular times and contexts.Footnote ³¹ In short, while a traditional historian might have been concerned with ‘what happened and why?’, a genealogist is expected to ask ‘how did X get here?’ or ‘how did Y become possible?’

It is important to clarify that in genealogy, contingency does not imply that anything goes. So while a subject or a social item can be genealogised through a series of shifts that do not add up to a progression of the origin-rise-decline-fall sort, each subsequent shift logically entails legacies from the past, otherwise there would be no historical continuities and no need for genealogy. Discourses, after all, make some, not any thoughts and actions possible. So viewed, effective history departs little from Weberian social science, specifically, the emphasis on counterfactuals and contingencies in historical-institutional analysis. Counterfactuals refer to the ‘what ifs’ and ‘might-have beens’ of history – the idea that history could have been and could be otherwise. Traditionally, counterfactuals were dismissed as purely speculative guessing-games; today, few historians would deny that they project alternatives to events and developments in order to better explain chronological patterns. In social science, too, counterfactuals are now regarded as inevitable tools for making causal judgments (outside the experimental setting, at least). Max Weber explained it a century ago: a cause of an event cannot be determined without imagining the effect of the absence of conditions that lead to it in the first place.Footnote ³²

Historical-institutional analysis is most directly related to genealogy through Weberian research tradition on the role of ‘unintended consequences’.Footnote ³³ This is obvious in virtually of institutionalist analysis. In this research community, institutions are seen as ‘rules of the game’ which are subject to path dependence and critical junctures and the purpose of the analysis is to how institutional equilibria endure and change in interaction with other institutions, exogenous shocks and, importantly, with their own unintended consequences.Footnote ³⁴ At the time when he wrote, Foucault rejected institutionalism because of its narrow conceptualisation of power.Footnote ³⁵ He was right. Most institutionalists then, not unlike many institutionalists now, focus on who gets what over whom, when, where and in what proportion. While these questions are central to political analysis, they do not necessarily lead the genealogist to the productive dimensions of power.

Luckily, institutionalist research programmes have greatly evolved since Foucault wrote such that there is now no shortage of those who argue that institutions, like discourses, also impact the ‘who’ in ‘who gets what?’ The emergence of discursive and constructivist institutionalisms in recent years can be seen as a welcome rapprochement between Weberian institutionalism and Foucauldian ideas on the social and political construction of reality and knowledge. According to Colin Hay, these new institutionalisms are desirable because they promise to ‘interrogate and open up the often acknowledged and yet rarely explored question of institutional dynamics under disequilibrium conditions’.Footnote ³⁶ The awareness that all knowledge is fundamentally uncertain and politically consequential has become more salient since Foucault, but it can also be found elsewhere the history of humanities and social sciences. Particularly well-documented is the way in which Weber build on Nietzsche's insight that history is stained with an idealised portrayal of the present such that historical discontinuities are silenced or assumed away. Along the same lines, Weber's idea of ‘interior authority’ recognised the links between the social and institutional which later became the mainstay of sociological constructivism, to say nothing of its constructivist and discursive cousins.Footnote ³⁷ Now let us review how genealogies are executed and evaluated in IR from the perspective of Foucault the methodologist.

Reading genealogy in IR

In IR, genealogy tends to be associated with two broad schools of thought, post-structuralism and constructivism.Footnote ³⁸ Without going into overly detailed (and hackneyed) moments of disciplinary identity politics, let us posit that post-structuralists and constructivists share a basic ontology (in which reality and knowledge are socially constructed) and empirical research practices (centred on the interpretation of texts and the observations of lived practices). What they do not share is epistemology: while both constructivists and post-structuralists agree that the promise of positivism is false and argue in favour of interpretation (in the sense of Giddensian double hermeneutics), many constructivists are happy to appropriate some broadly positivist standards for making truth-claims about the political world; specifically: operationalism, inference to best explanation and causation. In other words, post-structuralists and constructivists agree that objects of social science are also subjects, yet they disagree on theories of knowing. Different epistemologies, in turn, lead them to different theoretical and empirical foci: the former tend to privilege the political construction of knowledge (power/knowledge), while the latter tend to focus on the social construction of reality.Footnote ³⁹ When it comes to executing and evaluating genealogy, I will now argue, the difference between constructivist and post-structuralist IRs tends to be overdrawn.

Let us begin with a brief and stylised review of genealogical arguments made in IR, beginning with Jens Bartelson's Genealogy of Sovereignty, a book still regarded as a how-to guide for making genealogical arguments in IR. Bartelson famously declared that his methodology must not be taken ‘too seriously’, but then he still went on at some length to discuss the rationale for selecting his episodes and generating his archive.Footnote ⁴⁰ Next consider David Campbell's Writing Security. In this classic on the mutual determination between the state and foreign policy, the author began by describing genealogy as an ‘attitude’ and then passionately defended his US case study as paradigmatic in the study of International Relations: ‘No state possesses a prediscursive stable identity […] Yet for no state is this condition as central as it is for America.’Footnote ⁴¹ Similarly seminal was Richard Price's Chemical Weapons Taboo. Here, the author began by asking what made chemical weapons ‘morally illegitimate’, then, using a combination of counterfactual reasoning and historical analysis of key episodes, he showed – against two alternative explanations – how the chemical weapons taboo was a product of the constructions of identity and order at various points in history.Footnote ⁴²

By mid-2000s, methodological discussions of genealogy became de rigueur in the post-structuralist tradition. In Civilizing the Enemy, Patrick T. Jackson showed how the Marshall Plan and NATO were made possible by a discourse of a ‘Western civilization’ which identified Germany as a friend, as opposed to a vanquished enemy. Methodologically, Jackson's genealogy is remarkable for the way its author first establishes the dominance of this discourse in American and West German policy circles and then moves on to trace its productivity at a set of specific policy debates in the 1944–1955 period.Footnote ⁴³ In the same sense even more impressive is Lene Hansen's Security as Practice, a genealogy of the Western debates on the Balkans from the 1870s to 1995. In order to strengthen her narrative on how different representations of Bosnia and the Balkans produced different Western interventions in the 1992–1995 Bosnian War, Hansen commits an entire chapter, and then some, on methodological trade-offs in genealogical scholarship.Footnote ⁴⁴

From the perspective of this review, and in line with my argument in the first section, a rejection of an abstract methodological template does not necessarily imply some sort of free rein. Genealogies, whether authored by Foucault or IR scholars, are historical accounts justified by historical evidence which must be validated in some way. The nature of that evidence of course depends on the specific research question, but what is common is a need to develop a genealogical argument in a systematically coherent and logically consistent way, such that the conclusions follow from the overarching explanatory logic.Footnote ⁴⁵ What is more important for my argument, some form of operationalism and some anticipation of alternative explanations was present in every genealogy reviewed above. Now I want to propose that genealogy can be seen as a tool for making standard social science causal inferences, broadly construed.

Foucauldian purists rarely fail to point out to Foucault's own definition as contingency as anti-causation and argue, on this basis, that genealogy simply does not do ‘ideas as causes’.Footnote ⁴⁶ Here, too, my argument hangs on a local/minimal foundationalist, methodological reading of Foucault, which can, in principle, accommodate a broader idea of causation.Footnote ⁴⁷ Consider Price's and Jackson's books again. As a genealogy, Price's Chemical Weapons Taboo was remarkable because its stated aim was to resolve a puzzle for IR theory: if states seek to survive at any cost, why, then, did they discriminate against the use of lethal gas as a weapon of war in World War II?Footnote ⁴⁸ For one, why did Hitler – he who ordered the use of cyanide gas against civilian prisoners in Auschwitz – refused similar chemical weapons against the Allied onslaught in 1945? In answering these questions, the author first shelved the conventional wisdom on the general non-use of lethal gas in war – the arguments that poisonous gas is ineffective on the battlefield or that it is a result of some genetic revulsion towards asphyxiation shared by all humankind – and then showed how chemical warfare became delegitimised by the rise of the specific jus in bello norms emergent from World War I. Jackson was even more upfront about the causal claims he made. His Civilizing the Enemy centred in a triple ‘why?’: why was a German state reconstructed after 1945? Why did the US lead the reconstruction? And why was West Germany treated as an equal in the Marshall Plan and NATO? The answer to each had to with the invention of a ‘Western civilization’ that included West Germany. This framing ‘made possible’ not only the creation of the Federal Republic of Germany, but also its integration into the Euro-Atlantic institutional mainstream.Footnote ⁴⁹

In their genealogies, both Price and Jackson used the language of causation and embraced causation as the matter of ontological principle. As openness to the ontological complexity of causal relationships is not a characteristic of many social science research tools, this feature could be genealogy's next claim to fame. Before we get to the conceptualisations of causation, let us briefly reconsider the re-occurring Methodenstreit, the old epistemological debate between explanation and understanding, which has long been familiar to IR.Footnote ⁵⁰ The ‘positivist’ camp typically defines causation as ‘invariance under treatment’ – a mechanistic sequence of empirical occurrences which can be explained in terms of law-like patterns. In this view, to the extent that the researcher can meet the Humean requirements of independent existence, covariance, temporal asymmetry and counter-factuality, she can also draw causal inferences about the necessary and sufficient conditions under which empirical occurrences obtain. The opposing, ‘post-positivist’ camp typically argues that that causation is constitutive, not ‘direct’, such that the primary research questions are the ‘how-possible?’ and ‘what?’, and not ‘why?’ and ‘how?’ Here, the task of causal analysis is to understand how human actors think and act in given situations.

The Methodenstreit is still alive in the philosophy of social science, having weathered numerous attempts to synthesise the explanation-understanding distinction and/or to subsume one vision with another. In IR, the latest manifestation of this debate is the question whether constitutive theory can or should be seen as a form of explanation. Among the various affirmative answers, the most famous and the most controversial is Alexander Wendt's ‘via media’.Footnote ⁵¹ In this approach, constitutive theory is said to contribute to explanation by helping identify the conditions of possibility for social phenomena. Knowing what X is or how X is possible in the first place greatly facilitates how or why X causes Y. So far, post-structuralist theorists have rejected this reading as an attempt to further positivise IR.Footnote ⁵² Causation, epistemologically speaking, runs directly counter to constitution and all social phenomena, including the putatively causal effects of taboos and rhetorical coercion, can be seen as the manifestations of constituted realities.Footnote ⁵³ The critique in fact goes back to Foucault and his bid to overthrow representational metaphysics; for Foucauldian purists, therefore, the goal of political inquiry is not causation, but the ‘power analysis’ of the ways in which subject positions are constituted, starting with IR and its claim to scientism.

And yet, the relationship between causation and power analysis is more complex than mere incommensurability. Recall that in a local/minimalist foundationalist reading of Foucault, the subject remains capable of resisting and choosing among the discursive practices that constitute it, if ever so slightly. The conditions under which this capability (contingency, creativity, etc.) varies in time and space cannot be assumed away but must be subjected to empirical scrutiny. Herein lies the potential for making causal claims in interpreting history. For ontological reasons, these claims are not likely to be Humean – his criteria for causal inferences are impossible in the socially constructed world. But the philosophy of causation offers other criteria – Aristotelian, Cartesian, Gallilean, Kantian and others. In his genealogy, Jackson, advocated the Weberian school of ‘adequate causation’. Here, causal factors are not variables in the sense of necessary-and-sufficient conditions, but ‘reasons’ in the sense of historically situated ‘configurations’ which ‘simultaneously give rise to both actions and the actors that carry them out’.Footnote ⁵⁴ In short, Jackson's book, like Price's, shows how genealogies can make causal claims, too.

So we come back to epistemology again: while it is of course correct to argue that any criteria are always formulated within certain regimes of truth, be they positivist, post-positivistic, political, artistic, religious and so on, it is equally correct to point out that criteria are always necessary in order to halt the infinite regress and the paradox of denying all truth-claims that constitute knowledge. As the epigraph to this article says, we cannot help but make truth-claims about the world. This point can in fact be restated is the nature-of-language terms: in order to talk about reality at all, we not only have to reduce it to symbols expressed as words, but we also have to assign certain chronological and geographical stability to each of these words.Footnote ⁵⁵ Such stabilisations are an inescapable part of our language – and so of our historical interpretations – in anything from conceptual labels to full-blown formalisation of empirical occurrences in terms of time, space, intensity, magnitude, presence/absence, similarity/difference, continuity/change, and, indeed, cause and effect.

In post-structuralist IR, such stabilisations usually appear as mere ‘analytical shortcuts’, ‘acts of faith’, ‘arrestations’, ‘temporary reifications’ or ‘strategic essentializations’ and other seemingly practical, pragmatic and pedagogical moves.Footnote ⁵⁶ Not everyone of course likes it: a review of Hansen's Security as Practice protested such stabilisations as unnecessary concessions to mainstream IR.Footnote ⁵⁷ My view is exactly the opposite. These concessions are indeed necessary: while genealogy makes no stabilisations in principle, in practice it must, regardless its claims of frivolity.Footnote ⁵⁸ Even Foucault understood that in order to inquire about some inter-subjective systems of meanings, one had to assume the stability other inter-subjective systems of meanings. To go back to a point made earlier, the human language is successive, not simultaneous, and for that reason no meaningful analysis can make all meaning unstable at the same time.

Let me now bring the epistemological and ontological propositions together. In my reading, the genealogies authored by Price and Jackson demonstrate how the same research design, based on common evidentiary base can make both constitutive (for example, the constitution of subjectivity in the nexus of technology and morality; the constitution of the Western civilisation) and causal inferences (for example, Hitler's non-use of chemical weapons against Allied soldiers, the failure of isolationism in the post-1945 US). While constitutive theory tends to more systematically reflect back on the meaning of any stabilisation made by theory, both causal and constitutive theories stabilise meanings since their foremost ontological task is to account for historically specific conditions in which subjects and objects are constituted. Once the theorisation of being is stabilised, research can move onto epistemologically varied pursuits. For post-structuralists, this is typically a theorisation of the mutually constitutive relationship between agents and structures (or agents and agents). For constructivists, this is typically a theorisation of the causal conditions present in such relationships. Whether these conditions are seen as necessary-and-sufficient or as something else (‘permissible’, ‘productive’, ‘generative’, ‘adequate’) entirely depends on the researcher's conception of causation.

As I said earlier, the conceptualisation of causation in terms of necessary-and-sufficient conditions is usually associated with the Humean successionist theory of causation. Constructivists and post-structuralists agree that this conceptualisation of causation is too ‘rigid’.Footnote ⁵⁹ This anti-Humean revolution is welcome, but, as always, revolutionaries should be careful. For one, there is no such thing as a single Humean model of causation. The ‘invariance under treatment’ theory has indeed dominated, via positivism, the mainstream social science for many decades. But there are several ‘new Hume’ readings of Hume which question the invariance under treatment theory.Footnote ⁶⁰ More important, the revolution might lead to ironical outcomes: a post-structuralist call to abandon Hume in conceptualising causation might in fact hurt post-structuralism more than the putative positivist mainstream. Put differently, causal pluralism has a potential to render a post-structuralist claim that constitutive theory is strictly non-causal less defensible or even paradoxical.

Consider the Aristotelian model, which is currently the main significant ‘Other’ to the Humean conception of causation in IR. For the Aristotelians, causal questions are primarily ontological, not epistemological, which implies that constitutive analysis is simply one ‘form’ of causal analysis, but that is ‘inseparable’ from causal analysis.Footnote ⁶¹ The idea of the productivity of discourse is a case in point. To say that discourses make the boundaries of social identity by rewarding/punishing actions that are congruent/deviant with those identities as well as by enabling/constraining what is possible is to imply that constitutive discourses are in some way causal. In this reading, every time one writes that a discourse X ‘produces’ – or even ‘(re)produces’ – an identity Y, one is in fact making a causal truth-claim. The same goes for other ways of linking X to Y: ‘rewards/punishes’, ‘enables/constrains’, ‘implies’, ‘directs’, ‘shapes’, ‘favors’, ‘X has consequences/implications for Y’ or ‘Y is dependent on X’. Some self-identified constructivists – as well as some self-identified post-structuralists – accept that such relationships can be seen as causal or at least ‘quasi-causal’.Footnote ⁶²

From a post-Humean perspective on causation, this is a sensible position. But it is also logical from the perspective of research practice: as per ontological assumptions, X and Y are always mutually constituted, but empirical analysis usually focuses on one side of the relationship, while temporising or bracketing the other. As I said earlier, no matter how ontologically tentative X and Y, once we write them down we always invest them with analytically operational (that is, stabilised) identities and interaction patterns. Another reason to refuse calling the X-Y relationship non-casual lies in the impossibility of falsifying the null hypothesis – that the relationship is, in fact, causal in the Humean sense. A strictly non-causal claim would be X (probably) means Y, but (probably) not Z.Footnote ⁶³ The last irony is the possibility that greater causal pluralism in IR might lead to a formalisation of constitutive analysis. How such formalisation may develop is another subject entirely and I do not wish to argue that constitutive analysis should be formalised. But consider the following logic: because there are multiple legitimate social scientific questions to pose about a single evidentiary base, even traditional causal claims should not be denied, put aside or even temporised. So while post-structuralists are right to argue that explaining causation cannot be divorced from understanding meaning, they should also recognise the flip side of their argument – any understanding of the meaning-making processes cannot proceed without causal inferences.

The very word ‘because’, H. Stuart Hughes once noted, ‘gives warning that causal explanation is at hand’.Footnote ⁶⁴ Causal claims might indeed be frequent and familiar, but what constitutes a causal explanation continues to be debated in the philosophies of history and social science. In recent decades, promising in both history and social science are explanations based on ‘causal mechanisms’ – ‘frequently occurring and easily recognisable causal patterns that are triggered under generally unknown conditions or with undeterminate consequences.’Footnote ⁶⁵ From this perspective, the goal of a causal explanation is to account for the workings of human agency located within historically specific structures and processes. In principle, the idea of causal mechanisms can be regarded as compatible with multiple, non-naturalistic models of causation, which, for good or ill, could prove to be a major advantage to the advancement of epistemic science.Footnote ⁶⁶

One general drawback to thinking in terms of binary oppositions such as causal vs. constitutive theory, understanding vs. explanation or, for that matter, constructivism vs. post-structuralism is the lack of sensitivity to possible complementarities. Given the ontological complexity of the social world, any theory should prefer the pluralistic ‘both-and’ over the dualistic ‘either-or’. And while no single research design should ever be expected to answer all ontologically relevant questions concerning any given phenomenon – the whys, the hows, the whats, and the how-possibles? – a composite methodology which seeks to account for constitution and causation should be seen as both possible and useful. Based on the preceding arguments, genealogy may be exactly one such methodology. Because it is able to shift emphases among different kinds of causal inferences, genealogy can be applied anywhere from highly contextualised understandings of constitutions to historical causal explanations. Arguably, epistemologically varied truth-claims may come to yield an explanatory whole greater than the sum of its parts.

In lieu of summing up this section, let me refer to Jan Selby's recent argument that Foucault's ideas had long been ‘pigeonholed’ by post-structuralist IR, ‘while being generally ignored or critiqued by those of other theoretical persuasions’.Footnote ⁶⁷ The argument was prompted by Selby's own research on Israeli water policy, where the author discovered many complementarities between Foucauldian and Marxist approaches. One of these is a division of research questions: Foucauldians could get the how-possibles, while Marxists get the whys. Another mutually enriching convergence is methodology: ‘both the Marxist tradition and Foucault share a sensitivity to historical disjunctures and transformations, and a corollary hostility to a historical and positivist modes of analysis, which set them apart from orthodox IR’.Footnote ⁶⁸ Selby’s argument is likely to spark debate on Foucault and IR, which is good for the discussion of genealogy.Footnote ⁶⁹