1. Scientific Realism and International Relations

Colin Wight and Jonathan Joseph

In the mid-1980s international relations (IR) moved into what is known as its post-positivist phase. Underpinning many of the major theoretical debates that have occurred since then has been a concern to reconsider the manner in which research into international phenomena can best be conducted. At the heart of many of these debates was, and still is, a deep dissatisfaction with the positivist model of IR. Problems with the positivist model of social scientific research had been identified for some time in cognate disciplines long before they came to the attention of the IR community of scholars. Indeed, in the philosophy of science, positivism had been rejected as an adequate model of science in the 1950s (Kolakowski, 1969). When the rejection of positivism and the search for alternative models of social inquiry finally produced a range of alternative positions in the discipline of IR, they were united in their rejection of the positivist model there but had little else in common. Steve Smith identifies five distinct ‘post-positivist’ approaches in IR: constructivism, postmodernism, feminism, critical theory, historical sociology (Smith, 1995). Smith also identifies scientific realism (SR), which underpins Alex Wendt’s version of constructivism, as a potentially fruitful epistemological position to adopt in opposition to positivism.

Notwithstanding Smith’s problematic understanding of SR as an epistemology, there is no doubt that it, and associated positions, such as critical realism (CR), have begun to make inroads into the way IR scholars conduct research. Often, however, the influence of SR remains hidden due to the manner in which it plays an ‘underlabouring’ role for research, rather than providing new and innovative theories of IR. As such, SR/CR works at the level of philosophical critique, challenging the philosophical assumptions of much contemporary IR theory and in so doing it also introduces important epistemological and ontological insights in its own right. In this respect, SR is less visible than many of the other post-positivist positions that have gained prominence. Arguably, however, out of all of the post-positivist approaches SR has had the greatest, albeit unacknowledged, influence on the discipline, due to its role in Wendt’s version of constructivism, which has itself become a major new approach within IR, and the manner in which SR has begun to reshape understandings of science within the discipline. However, it is wrong to think that SR can provide an alternative to mainstream approaches to IR such as neo-realism and neo-liberalism. SR simply does not engage at that level. As an underpinning philosophy, there is no reason why SR cannot be used to support a range of substantive theories of IR. But it cannot support them all and some theories of IR are not compatible with SR/CR (see for example the later discussions of Waltz). The aim of this book is to bring together scholars with a broad interest in using SR in order to try and draw out exactly how SR and CR can be used within the field of IR, to see how philosophical enquiry raises questions over our understanding of key theoretical issues, and to look at how scholars in IR use SR and CR in different ways to help guide and shape their research practices.

SR is a philosophy of and for science. It provides one way of thinking about the practice of science that runs counter to broadly conceived positivist approaches. CR is a specific development of SR that attempts to show how these issues might impact on social scientific practice. Neither SR nor CR provides a substantive theory of either society or IR. As such, their impact on the discipline will be fed through issues surrounding the philosophy of social science. No discussion of the philosophy of science in IR could avoid situating itself in relation to positivism. Positivism constitutes not only the standard around which the mainstream is said to converge, and the focal point around which non-mainstream approaches situate their criticisms, but also, perhaps, the very definition of science itself. This is a serious error, and throughout this book all the authors will attempt to demonstrate how the positivist vision of science is flawed and how SR/CR might help correct some of the persistent misunderstandings concerning science and show how a non-positivist account of IR can accommodate many of the so-called ‘post-positivist’ criticisms of positivism without regressing into a debilitating, and potentially relativistic, anti-science stance.

CR is a very specific development of SR within the human sciences. It is normally associated with the work of Roy Bhaskar and a group of social researchers that have taken his ideas and developed them in a range of social sciences. It is possible, however, to be a scientific realist about the social sciences, but not a critical realist. Equally, one could extend the notion of CR beyond the narrow confines of Bhaskar’s particular framework. Indeed, there has been something of a ‘realist turn’ in the philosophy of social science that in some ways mirrors the turn to realism within the philosophy of science. We would not go as far as to claim that realism is now the dominant approach taken to the philosophy of social science, but it does represent an important and identifiable strand (Mantzavinos, 2009; Manicas, 2007).

Most realist philosophies of social science accept the following basic principle concerning social science. They maintain (or at least do not deny) the view that there is a social reality which consists of multiple forces that condition individuals’ lives; that some of these forces may well be unobservable, but nonetheless real; that these forces are structured by forms of internal and external relations, power structures and social roles; that the social sciences cannot capture the nature of causal forces in the social sciences merely through empirical investigation; that social and political sciences are fundamentally social and political in nature and reflect, in part, the inquirer’s position in social reality; and that the interaction of agents and structures and material and ideational forces is an important question to be settled empirically and not by theoretical fiat.

Before proceeding to outline some of the basic principles of SR/CR, however, we need to address some of the persistent misunderstandings that surround its place and application in the discipline. First, as already indicated, SR/CR is not a theory of IR per se, or even a theory of the social. Certainly some authors working within an SR/CR perspective have developed particular insights concerning IR or other aspects of the social, but these do not amount to anything approaching a critical realist theory of IR. This is important in terms of Chris Brown’s claim that CR is basically reducible to a form of Marxism (Brown, 2008; cf. Yalvaç in this volume). Brown argues that SR and CR are best understood and most fruitfully employed when associated with Marxist or historically materialist concerns. He prefaces his argument by tracing the historical and sociological conditions of the rise of critical realist thought, notably the motivations behind Roy Bhaskar’s thought (pp. 410–12). Despite the overall veracity of Brown’s description of the emergence of CR, his attempt to describe SR/CR as reducible to a form of Marxism is problematic. First, Bhaskar’s account of SR/CR is not the only one and it is possible to develop an account of SR that draws on a wide, and differing, range of intellectual resources that go well beyond Bhaskar, but which are all generally compatible with a realist view of social science. To reduce SR/CR to simply Bhaskar’s work is an error. Of course, one might legitimately argue that Bhaskar’s work has varying affinities to Marxism (his earlier work much more so than his contemporary position), but it does not follow that CR, as a body of thought, must be Marxist. It is certainly correct that many critical realists are indeed Marxists and historical materialists, but it is equally true that many Marxists are deeply critical of CR.

Those Marxists who do find CR attractive are drawn to its emphasis on the non-positivist study of the social world that emphasises unobservable and materially embedded social structures and processes as opposed to the lawlike empirical regularities foregrounded under a positivist account. Critical realism also advocates the use of explanatory critique by linking certain theories to the reproduction of social reality, a view consistent with Marxism, feminism and critical theoretical strands of IR. But unlike some constructivist and post-structuralist approaches, CR avoids the conflation of theory with the world itself, or of embracing a relativistic account of the relationship between power and knowledge. This all helps in providing philosophical support to a Marxist position that grounds itself in the study of exploitative social relations, but to assume ‘critical realism equals Marxism’ is erroneous and unhelpful (even to Marxism).1 Firstly, SR emerged as an account of science that can be applied to biology or geography, just as much as to politics or IR. Indeed, many scientific realists have not tried to develop the implications of this account of science in terms of social science, and many of those that have, have concluded that SR cannot be easily incorporated into the social sciences. Nonetheless, there are many who believe that SR can provide an adequate account of social scientific practice. However, not all of those who take this position agree with either CR or Marxism. Mario Bunge, for example, is a scientific realist and also committed to the role SR can play in the social sciences, yet he never cites Bhaskar, or even acknowledges CR, and moreover he is a deep critic of Marxism. Equally, others, such as Roberto Unger, Manuel DeLanda and Pierre Bourdieu, have explicitly advocated a realist approach to social science, and none, we think, can be described as Marxist in the traditional sense of the term. Hence while SR/CR may delimit what kind of social theory can be advanced, it does not determine substantive theoretical choices in the way Brown suggests. Indeed, as Brown recognises, CR does not aspire to be a theory of social reality and it steps aside to allow social theorists to get on with their business. There is no reason to suppose that CR would a priori rule out non-Marxist approaches to IR, and it is more than possible to contend that certain arguments – be they feminist, English School, realist or even post-structuralist – might be consistent with critical realist meta-theoretical premises. Indeed the most influential use of SR in IR is that of Wendt and his position can hardly be described as Marxist.

A second important issue to note is that it is vital to differentiate SR from CR. SR is a philosophy of science that arguably is the dominant approach now taken within the philosophy of science. In fact, the only serious contender to SR is Bass Van Fraassen’s ‘constructive empiricism’, but this approach is fighting a rearguard defence of an empiricist account of science; something that seems to most to be a lost cause. However, while the fact that SR is currently dominant in the philosophy of science does not make it correct, it is surely worthy of consideration for any discipline that aspires to the label ‘science’. Fred Chernoff’s critique of SR is to a large extent commensurate with Van Fraassen’s ‘constructive empiricism’ and in effect differs little from the position staked out by Kenneth Waltz in Theory of International Politics (Chernoff, 2005, 2007b). There is, however, a fundamental difference in the relationship between the philosophy of science and the practice of science that stands at the heart of our position when compared to that of Chernoff. Chernoff considers the aim of the philosophy of science to be concerned with the provision of a logical account of what science is, that is prior to and privileged in relation to the practice of science. Thus for Chernoff, the philosophy of science takes precedence over the practice of science. Accordingly, Chernoff seems to suggest that if the practice of science is not consistent with the philosophy of science then it is the practice of science that needs to alter to bring it into line with the philosophy. We think this would be disastrous. The aim of the philosophy of science according to SR is to reflect on the practice of science, and in so far as possible, provide an account of what that practice is. If the practice of science conflicts with the philosophy, then it is the philosophy that will need to be amended. After all, this is a philosophy ‘of’ science, not a science ‘of’ philosophy. In the social sciences, however, the situation is different. The natural sciences are often said to have only fully matured when they left philosophy behind and established themselves as autonomous modes of knowledge generation. The social sciences, on the other hand, have always turned to philosophy for guidance and have constantly attempted to assert their scientific status through appeals to the philosophy of science. Moreover, the social sciences have not yet managed to replicate the practical success enjoyed by the natural sciences. Equally, there are good reasons (discussed below) as to why the social sciences must engage in philosophical reflection.

In our view philosophical inquiry in IR, while clearly not the primary aim of IR scholarship, has an important role within it. This is because any discussion over the meaning of science, the nature of social ontology, the balance of agency and structure, or the study of causality is fundamental to all IR accounts: there is simply no avoiding these meta-theoretically loaded areas, nor engaging in (explicit or implicit) philosophy of social science argumentation to justify one’s particular approach to such matters (see e.g. Kurki, 2007; Wight, 2007: 13–33). Whether one argues that IR is characterised by laws, such as democratic peace, or structural conditions of anarchy, or socially constructed norms and rules, or ethical and normative decisions, one is assuming something about the nature of the social sciences as a discipline, its ontological objects of study and the nature of the epistemological claims made about it.

However, contrary to Nuno Monteiro and Keven Ruby (2009), we do not see SR/CR as attempting to construct robust foundations for social scientific practice. We do not believe such foundations are possible. What SR/CR does hope to achieve, however, is the much less ambitious aim of ensuring that reflection on the philosophical claims of alternative approaches is grounded in appropriate reflection on the prevailing wisdom in the philosophy of science. The aim is not to provide foundations as such, but merely to use CR to reflect on the deep ontological, epistemological and methodological assumptions that are integral to all research even if these assumptions remain largely hidden. In fact, given the SR/CR focus on ontology then the idea of foundations is a non-starter, since differing ontological domains may well necessitate differing approaches to the problem at hand. Of course, there are many in the social sciences that find philosophical debate concerning the nature of social science to be an irritating distraction and argue that we would be better advised in simply engaging in the practice of ‘doing’ social science, rather than introspectively discussing ‘how to do’ it (Wallace, 1996; Gunnell, 1998). One can undoubtedly sympathise with such views. After all, one major aim of political science is to understand, and perhaps explain, those sets of practices we call politics. Surely we can just get on with the task of producing substantive theories of our chosen realm of study in the manner of the natural sciences and leave philosophy to the philosophers. Attractive as this might seem, however, there are historical, ontological and ethico-political reasons why this is simply not possible, at least for the social sciences.

Historically we cannot simply leave philosophy behind because the discipline already has certain philosophical narratives about what constitutes good science which inform its methods, theories and discourses. To leave philo sophy behind at this stage of disciplinary history would be to leave in place a certain form of political analysis that is already deeply embedded within particular philosophical assumptions that may be problematic. Moreover, the philosophies of science that dominate in the discipline can only be tackled at the appropriate level, and that is at the level of philosophy. Like it or not, even if the idea of a political science without philosophy was attractive, it could only be achieved through more not less philosophy; at least temporarily. But perhaps this concern with philosophical groundings is just what social scientists do anyway, and any critique of this practice could only emerge from an alternative philosophical position that seemingly holds the truth of what ‘real’ social science should be (i.e. no philosophy), but currently is not.

Another important reason for maintaining a role for the philosophy of science and the philosophy of social science in the social sciences is the ontology of social practices (what they are and what they involve). Philosophy forms part of the object of study for the social sciences because the practices of the actors in that world are already embedded within philosophical positions. Indeed Peter Winch thought social science was a form of philosophy (Winch, 1958) and used Wittgenstein’s theory of language games to justify philosophy’s role in shaping human practice. And by this we do not simply mean to highlight the manner in which certain political leaders are believed to be guided in their practices by an explicit engagement with the philosophies of Plato, Machiavelli, Locke and Hume, and so on. Rather, it is the recognition that even a social actor with little or no awareness of political philosophy, metaphysics, the philosophy of social science, ethics, epistemology, or the whole range of philosophical discourses will, and can only, act on the basis of some or other account of the world: What is truth? What is a fact? How do I know anything? How should I live? What is a cause? What is an explanation? These are very basic philosophical questions that play a role in almost all social practices, even if those engaged in those social practices never stop to reflect on such questions. In many respects, unreflective social practices are governed by answers without due consideration of the questions. Indeed, we see one important function of any properly conceived social science to be the uncovering of just those assumptions that underpin social practice, particularly in those instances where the actors involved are not aware of the role and function of those philosophical assumptions (i.e. that the truth is self-evident, that to believe is to know, that social wholes do not exist, etc.). The objects of the natural sciences, on the other hand, have no such philosophical issues that govern their practices; or at least, as far as we can tell they do not.

At the ethico-political level the need for reflection on the claims of social science is also clear. Social scientific results are not simply descriptive but normative. Given this, and the particular place of science in the contemporary world, we believe it is wholly appropriate to examine the claims produced by the social sciences in terms of their claimed scientific status. Of course, the claims of the natural sciences should also be subject to both internal and external scrutiny, but this issue is much more pressing in the social sciences, since their status is already highly contested and the potential implications in terms of social action are more immediately apparent. The claim that a certain gene should not reproduce is meaningless (yet still scientific); yet to imply, for example, that single mothers and crime rates are related is an altogether different claim, and carries with it a normative element. The social sciences study human beings and social scientific outcomes are concerned with explanations of the practices of human beings. The natural sciences, on the other hand, can have an impact on human practices, but they do not necessarily involve a critical comment on those practices. The social sciences always do.

This introduction now goes on to provide a broad understanding of SR and CR and hence situate the approach that unites all the authors. Obviously not all of the authors will accept all aspects of this account, and SR is a broad movement with many differences within it concerning the key principles and implications of it for research.

The approach to philosophy

It is important to note that the form of SR advocated here begins by placing some rigorous limits on its scope of application. These limitations are derived from the fact that the conclusions of the philosophy are based upon empirical aspects of the practice of science. For Roy Bhaskar, the philosophy of science examines a specific set of existing practices, in this case science, and as such, it provides no transhistorical transcendent truths. The approach is transcendental in the sense of examining the conditions of possibility for a given set of practices; in this instance science.2 Because philosophy has as its object of inquiry a set of actually existing social practices – again in this case science – it is always a historically located enterprise. The most it may aspire to is the possible enunciation of insights pertaining to those practices from which the arguments are derived. There can be no philosophy of science in general, but only the philosophy of particular, historically determinate, specific forms of scientific practice. Thus scientific practice forms the object of concern for the philosophy of science. This both historically embeds it and establishes its particular character. Hence no philosophy of science can consider itself privileged by some special (high) subject matter or (superior) mode of truth (Bhaskar, 1986: 12).

However, although the practice of science constitutes the object through which SR is developed it does not follow that it is bound to particular scientific theories, but rather, SR makes the more fundamental claim that science itself must be of a certain form if the practice is to be intelligible. SR is derived from the practice of science, not particular existing scientific theories; such theories are always in principle revisable and/or refutable.

This is an important point, since many of the rejections of SR that have emerged within IR suggest that since some of the theoretical entities postulated in theories have turned out not to exist then SR is problematised (Kratochwil, 2000). This view rests on a mistaken understanding of the claims of SR. SR does not suggest that the theoretical entities postulated in all scientific theories must have existed at some point in time, it only suggests that science progresses because in their practices scientists set out to test the supposed reality of the proposed mechanisms. Some may exist, and some may not, but it is the attempt to differentiate between the two that drives the scientific enterprise on. SR makes intelligible what scientists do, it does not entail that all scientific claims relating to theoretical objects are necessarily correct. Indeed, no matter how much epistemological support there may be for the existence of a given theoretical entity, the ontological question of whether or not it exists is independent of the epistemological claims.

This approach to the philosophy of science in particular and philosophy in general indicates a reversal of a long-standing philosophical orthodoxy and a turn away from epistemological concerns to those of ontology. For realist philosophies of science, to be is more than to be perceived and the nature of the object itself may well place limits on what we can know of it and how we might come to know it (Outhwaite, 1987). The knowledge we might gain of subatomic particles, for example, will differ from the knowledge we might gain of societies because subatomic particles and societies are differing types of thing. The attempt to specify epistemological and/or methodological criteria for science in advance of ontological commitments is an act of unnecessary closure. Any discourse on epistemology or methodology is bound to be more or less arbitrary without a prior specification of an object of inquiry. Epistemological questions cannot be settled in advance of, or distinct from, ontological questions.

The realist move to ontology provides a more open, less dogmatic, approach to epistemological matters. This is not to say that ontological statements can be dogmatically asserted. All ontological assertions will require some form of epistemological support. But the scientific realist approach to epistemological concerns is to reject ‘epistemological monism’ in favour of epistemological opportunism.

From the analysis of scientific practice SR concludes that science can be understood as the attempt to specify explanations that make sense of the deep underlying structures and mechanisms that make experience possible. This means that it is committed to: ontological realism (that there is a reality independent of the mind(s) that would wish to come to know it); epistemological relativism (that all beliefs are socially produced); and judgemental rationalism (that despite epistemological relativism, it is still possible, in principle, to choose between competing theories).

Ontological realism

Every theory of knowledge must logically presuppose a theory of what the world is like (ontology), for knowledge (epistemology) to be possible. In which case, all philosophies, discourses and practices presuppose a realism – in the sense of some ontology or general account of the world – of one kind or another. The question is not of whether to be a realist, but of what kind. In the social sciences, however, realism does face some major problems. Realism normally implies that objects have a mind-independent existence. Social objects, however, depend upon minds for their existence. This has led all manner of positions to declare that ‘reality is a social construct’ (Berger and Luckmann, 1967), or, that ‘there is nothing outside of discourse’ (Campbell, 2001).

Contemporary social scientific anti-realism can be traced, on the one hand, to the rejections of, and attempts to transcend, positivism, which is wrongly believed to adhere to philosophical realism; and, on the other hand, to the twentieth-century fascination with linguistics. As should be clear, the first is simply mistaken. Positivism has always been an anti-realist philosophical position. Indeed, as Martin Hollis has argued, all empiricist theories of knowledge are anti-realist at bottom, even though they try to finesse their way to speaking of an independent world (Hollis, 1996: 303–4). At best, when positivists overtly espouse realism it is an empirical realism as opposed to the depth realism enunciated by SR.

Of course, we can only know things under certain descriptions. But there is no inference from ‘there is no way to know a thing except under a particular description’ to ‘there is no way to know that that thing exists (and acts) independently of its particular description (and descriptions in general)’ (Bhaskar, 1991a: 24). Empirical realists and contemporary linguistic/conceptual realists both commit two closely related philosophical errors. The first consists in the way both positions define the real. Whereas the former define the world in terms of our experience of it, the latter define it in terms of our theories and/or linguistic conventions. Both give an epistemological category an ontological task. This is an error that Bhaskar calls the ‘epistemic fallacy’: the transposing of epistemological arguments into ontological ones (Bhaskar, 1997: 36–8). This is most clear in positivism where all that we can know is what we can experience and what we cannot experience has a mere methodological function, but no ontological import. But statements about being cannot be reduced to, and/or analysed solely in terms of, statements about our knowledge of being.

The second error consists in the belief that the possibility of being experienced or the possibility of being conceptualised, and/or talked about, is an essential feature of reality itself. This assumption is unwarranted. There is no reason to assume that all of existence might be susceptible to human cognition. What empirical realism and linguistic conceptual realism both overlook is the role of a causal criterion for the ascription of something as real. On a realist ontology, on the other hand, being is independent of human experiences and/or expressions of it. Indeed, ‘all science would be superfluous if the outward appearances and essences of things directly coincided’ (Marx, 1971: 817). Science, then, is driven by a commitment to a deep realism, and not just an empirical realism. The understanding of science as a practice requires the depth realism suggested by SR.3

The empirical realist error is the conflation of three domains, or levels of reality, into one – that of the empirical. In contrast to this, scientific realists argue that in order to make sense of the scientific enterprise we need to distinguish between the domains of the empirical (experiences and impressions), the actual (events and states-of-affairs, i.e. the actual objects of potential direct experience) and the real or non-actual (the deep structures mechanisms and tendencies) (Bhaskar, 1979: 56–62). If the word ‘empirical’ is to have a particular meaning in human knowledge, it must be restricted to denote only that which we humans experience at particular locations in space and time. This should not be taken to imply a rejection of the importance of the empirical to human knowledge production. Any coherent epistemology requires an empirical component. In order to say things are, or are not as they seem, we need to look, to experience, to examine. We do not need, however, to make knowledge, in its entirety, dependent upon experience; this is the error of empiricism. The empirical is not exhaustive of reality but it does constitute a crucial element in any epistemology. SR is not opposed to empirical research and can incorporate it into its account of science without reducing scientific practice to nothing but an exercise in empirical data gathering. The empirical forms an important core to our understanding of the world around us, but it does not exhaust it.

A commitment to depth realism presupposes that there are things, entities, structures and/or mechanisms that operate and exist independently of our ability to know or manipulate them. The laws of nature, the entities, structures or mechanisms, which are often not empirically ‘observable’, are what Bhaskar terms the ‘intransitive objects of knowledge’ which exist independently of us and are separable from our ability to know them. Science is possible, then, because the world consists of ‘intransitive’ objects which form the focus of scientific discourses, with the aim of science, in particular, being the production of knowledge of mechanisms that in certain combinations produce the phenomena that are actually manifest in experience/ appearance. Likewise, the causal laws that science has discovered must be transcontextual – that is must operate in open and closed systems alike – if we are to make any sense of the application of science outside of the confines of tight experimental closure. The clarity and order that appear to accompany law-like regularities often disappear when we move from the laboratory and attempt to explain outcomes in the open world of everyday life. Thus, the objects of scientific inquiry are not only constant conjunctions or other such empirical regularities, but also the structures or mechanisms that generate such phenomena. This suggests that the Humean account of cause as ‘constant conjunctions’ is inadequate (Kurki, 2008).

A scientific realist causal account would involve the theoretical identification of these structures and mechanisms and their causal powers. The explanation, identification and cataloguing of the powers, effects and liabilities these entities possess and produce, not prediction, becomes the object of scientific knowledge. In effect, many of the crucial questions in science are concerned not with the mapping of causal relations in time, but with the causal power of entities, structures and mechanisms. This process of cataloguing is what Wendt calls ‘constitutive’ theorising (Wendt, 1998).

Surface forms or phenomena, and our experiences of them, then, do not exhaust the real. What we experience is the result of a complex interaction of these structures and mechanisms, which in controlled experimental conditions produce law-like regularities. If we are to make sense of how this experimental knowledge is then put into practice in open systems, then these same entities must also generate effects in the world beyond the laboratory; but devoid of the interventions of scientists, not in the same clean, recurrent stream of cause and effect. Outside of the human-induced effects of experimental closure they are part of a natural interactional complexity that sometimes results in particular causal relations, while at other times in the suppressing or complete neutralisation of the generative effects in question.

The arguments for depth realism suggest that the universe (material matter) existed prior to the emergence of humanity (biological matter) and that what living organisms there are, are composed of, surrounded by, and dependent upon, matter. In this very limited sense, matter may be said to be more ‘basic’ than life and to be necessary for life; life can be said to be more ‘basic’ than rationality and to have preceded it; and, matter, life and rationality can all be said to be more ‘basic’, and to have preceded, human society and history. This is what it means to talk of reality as being stratified. A superficial understanding of this stratification would suggest that those sciences that explain a more basic layer could claim explanatory primacy over those explaining a less basic layer. Such explanations are common. Modern genetics seems to suggest all explanations of human behaviour can be traced to the one level, that of genetics. Scientific realists reject this reductive approach and advance the idea of an emergent ontology, in which, although the more basic sciences might be able to explain something about the mechanisms of the less basic ones, they cannot explain them away. The laws discovered and identified at one level are irreducible to those at other levels. Each level has its own emergent powers that although rooted in, emergent from and dependent upon, other levels, cannot be explained simply by explanations based at the more fundamental levels. The emergent levels, then, have powers and liabilities unique to that level.

Moreover, when a mechanism at a particular level has been identified, described and used to explain some phenomena, it itself then becomes something to be explained; and often this explanation will be incomplete without recourse to other mechanisms located both horizontally and vertically in relation to the explained mechanism. Scientific progress is a process where our knowledge of nature is deepened and underlying each mechanism, or level, there are always other levels waiting to be explained. This ontological commitment to depth when combined with the theory of emergence has important consequences for all forms of scientific explanation. It entails that reductive explanations, in either upwards or downwards directions, will not suffice. Some entities, humans for example, will be subject to laws operative at more than one level. Hence, the identification and distinguishing of vertical and horizontal explanations is necessary. Emergence means that although the more complex levels of reality, for example, societies, presuppose the more basic or less complex levels, for example, people, explanations of them are not reducible to the other.

Epistemological relativism and judgemental rationalism

The commitment to depth realism also has implications for how we understand the process of knowledge construction. For if we reject the view that descriptions of reality are directly given to us in experience (the world of experiences/appearances does not exhaust the real) then we are left with a problem making sense of how our descriptions do come about and also explaining why some seem to be better than others at capturing various aspects of the world. Knowledge emerges through a transformation of pre-existing knowledge; a set of antecedent materials; theories, paradigms, models, facts, speculations, linguistic conventions, beliefs, hunches, hypotheses, guesses, symbolic gestures, and so on. Knowledge is a social product, dynamically produced by means of antecedent social products – albeit on the basis of a continual engagement, or interaction, with its (intransitive) object.

Since a mode of knowledge production is concerned with something other than itself, science can be mistaken about its objects. Hence science is an inherently fallibilist enterprise. Because it is fallibilist it is also necessarily continuously critical of its own knowledge claims. SR recognises that the science of any given time can be wrong about its object. Indeed, the continuous possibility of being wrong drives science on. The production of scientific knowledge should be seen as a specific social practice; a form of work that takes existing scientific theories as its starting point, even if only to reject them and/or transform them into ever deeper knowledge of the world. Science is an active intentional examination, and possible intervention, into nature or some aspect of the world.

Equally, since it is difficult to conceive of our minds being formed outside of the influence of specific societies, non-scientific knowledge is also historically specific. Different societies will instil different ideas and practices. Given the historically specific nature of knowledge, we have to accept the fact of epistemological relativism; namely, that all beliefs are socially produced, so that knowledge is transient, and neither truth values nor criteria of rationality exist outside of historical time. It is conceivable that all of our current stock of knowledge could be overturned at some point in the future. To recognise that science is the social production of knowledge by means of knowledge means no more than to place it within, rather than outside of, history. That is, that ‘whenever we speak of things or of events, etc. in science we must always speak of them and know them under particular descriptions, descriptions which will always be, to a greater or lesser extent, theoretically determined’ (Bhaskar, 1975: 249).

Bhaskar introduces the distinction between the transitive and the intransitive in order to account for the relationship between the production of knowledge and the world that this knowledge is about. Our knowledge of the world is embodied in the social practices that produce transitive objects of knowledge whereas the intransitive refers to that which this knowledge is about. These intransitive objects (real things, structures and mechanisms) are generally independent of our (transitive) knowledge of them. SR is epistemologically relativist, that is, relativist about the transitive object (the knowledge object), not ontologically relativist (about the entities this knowledge explains). And because it is knowledge of an intransitive object, some knowledge claims may be better than others. Hence there is at least the possibility of judgemental rationalism. This means that fallibilism can be embraced without endorsing a debilitating epistemological nihilism.

From science to social science

According to SR then, the positivist account of science does not, epistemologically, methodologically or ontologically, provide an accurate model of the actual practices of scientists. Insofar as science is successful in understanding, explaining and manipulating nature, both in closed laboratory situations and/or in the open world, it is because it operates with, and utilises, a multi-layered realist conception of the world, even if scientists rarely make this explicit. According to this realist model of science the world is seen to be an ensemble of powers, propensities and forces which are a result of the ways in which things and entities are composed, structured and related to each other within systems. These powers, propensities and forces can be given abstract formulations as laws, but these laws are used to help refer to and explain real complex situations, processes and events. It is the discovery of the real powers, propensities and forces of the world that gives science its explanatory power.

Furthermore, new levels of reality may emerge from the combination of particulars into systems, and these new levels will possess their own emergent powers. Thus, science has to construct explanations of causation on several levels without always attempting to make reductions to lower levels. Given that reality consists of these complex structured entities, possessing their own powers, propensities and forces, the problem of epistemic access takes on a different form than that suggested by positivism. The practising scientist does not search for constant conjunctions of observable events, but rather is involved in a process of modelling hypothetical mechanisms and inferring their necessary existence from their effects within emergent structured systems. Once this non-positivist account of science is accepted, any argument about the possibility of a science of society that bases itself on positivism is bound to be misleading. If positivism cannot be assumed the correct account of the method of natural science, the question of naturalism versus anti-naturalism must be re-examined.

Ontologically this re-examination takes the form of establishing three important factors about societies. First, societies are irreducible to people; social forms are a necessary condition for any intentional social act. Second, their pre-existence establishes their autonomy as possible objects of study. Third, their causal power establishes their reality. All social activity presupposes the prior existence of social forms. ‘Speech requires language; making materials; actions conditions; agency resources; activity rules’ (Bhaskar, 1979: 43). Put simply, these can be seen as arguments for the reality of social forms that are not explainable solely in terms of individuals. Of course, societies are (in part) made up of people who have causal powers that can be brought to bear upon the material world, but people do not exhaust the social.

Thus the social ontologies of individualist and structuralist theories are deficient in one or other area. On an individualist reading there are actions but not the conditions that make action possible, whereas on a structuralist account, there are the conditions that make action possible but no actions. Social action never occurs outside of a social setting, but social settings, or as more commonly put, social structures, ‘don’t take to the streets’, that is, do not in themselves act. Societies are an ensemble of structures, practices and conventions that individuals reproduce or transform, but which would not exist unless they did so. Societies do not exist independently of human activity (the error of reification); but nor are they the product of it either (the error of voluntarism).

Society is both the ever-present condition, that is, the material cause, and the continually reproduced outcome of human agency. All social practices have an action and a structural aspect that is integral to the practice. The crucial ontological issues in terms of the agent–structure problem are those concerned with the nature of these action and structural aspects and their relationship. The study of social objects, then, is concerned with the persistent relations between individuals (and groups), and with the relations between these relations (and between such relations and nature and the products of such relations). According to this relational model of societies, one is what one is, in virtue of the relations within which one is embedded. At any particular moment in time an individual may be implicated in all manner of relations each exerting its own peculiar causal effects. This ‘lattice-work’ of relations constitutes the structure of particular societies and endures despite changes in the individuals occupying them. Thus, the relations, the structures, are ontologically distinct from the individuals who enter into them.

At a minimum, then, the social sciences are concerned with two distinct, although mutually interdependent, strata. There is an ontological difference between people and structures, ‘people are not relations, societies are not conscious agents’ (Collier, 1994: 147). Any attempt to explain one in terms of the other should be rejected. If there is an ontological difference between society and people, however, we need to elaborate on the relationship between them. Bhaskar argues that we need a system of mediating concepts, encompassing both aspects of the duality of praxis into which active subjects must fit in order to reproduce it: that is, a system of concepts designating the ‘point of contact’ between human agency and social structures (Bhaskar, 1979: 51). This is known as ‘positioned practices’. In many respects, the idea of ‘positioned practices’ is very similar to Pierre Bourdieu’s (1977) notion of habitus:

Thus social practices are produced in and by the encounter between: (i) the habitus and its dispositions; (ii) the constraints and demands of the socio-cultural field to which the habitus is appropriate or within; and, (iii) the dispositions of the individual agents located within both the socio-cultural field and the habitus.

Society has a specific shape and form and it has material and potentially empirically perceivable effects, despite the fact that it would seem to be in principle unobservable. In this respect, we would be justified in saying that social power, for example, is real if it results in observable human action, utterances and perhaps, institutionally organised patterns of behaviour. Thus, and preserving the insight of hermeneutic theorists, it is argued that social action occurs in large part due to the knowledge and beliefs about social situations that are shared by groups of people. However, it is equally important to note that roles, rules and relations structure behaviour in ways that are sometimes opaque to consciousness, decisions or choices. Hence the conceptual aspects of society do not provide an exhaustive account.

On this view, society does not simply consist of individuals and/or groups and their activity, but rather, is the sum of the relations within which individuals and groups stand. Thus the raison d’être of the social sciences consists in the move from the specification of manifest phenomena of social life, as conceptualised in the experience of the social agents concerned, to the uncovering of the social relations that necessitate and regulate such experiences and phenomena. This gives social science a critical impulse insofar as the agents, whose activities are necessary for the reproduction of these relations, may be unaware of the social relations which (in part) explain their activities. It is through the capacity of social science to illuminate such relations that it may come to play an emancipatory role.

How does scientific realism relate to international relations?

SR, in contrast to the different theories of IR, is a philosophy not a theory, producing second-order, conceptual or meta-theoretical claims. While not being a theory of IR in its own right, it can criticise some of the assumptions of different theories by examining their philosophical underpinnings, while offering support to alternative theories by strengthening their ontological and epistemological claims. SR is useful to those who want to study IR in the sense that it can offer a critique of the philosophical underpinnings of mainstream theoretical positions, and point to the way that the assumptions these theories make hold back a proper analysis – particularly where power relations, inequalities exploitation and other issues are concerned.

For example, Waltz’s Theory of International Politics clearly makes assumptions of a positivist nature – that scientific laws are based on the relations between variables along the lines of if a then b with probability x (Waltz, 1979: 1). Waltz adds that theories differ from laws in showing why these associations obtain; in other words, theories explain laws (ibid., 6). The ontological consequences of Waltz’s views can be found in the familiar statements that ‘The texture of international politics remains highly constant, patterns recur, and events repeat themselves endlessly … The enduring anarchic character of international politics accounts for the striking sameness in the quality of international life through the millennia’ (ibid., 66). Like other positivists, Waltz embraces assumptions about states being the basic units and ‘structure’ being comprised of external relations between units. For all the talk that one should focus on the relation of unit to structure rather than interactions among the units, the structure is ultimately nothing more than precisely these interactions.

For Waltz, structural questions are about the arrangements of the parts of a system. In the international system each part is said to be formally the equal of another (ibid., 88).

The effect of this analysis is typical of any number of positivist approaches – that is to say, to naturalise a reified view of the social world (in this case its international system) and to hide the deeper structures of the international system by focusing on recurring relations between formally equal units. This can tell us nothing about the specificity of North–South relations, for example, or the reasons why such relations are exploitative. To do so, we would have to go to the underlying conditions that generate the unequal relations between North and South, something that positivist approaches rule out. For Waltz, international politics is about discovering and explaining law-like regularities (1979: 116). Even if we were to accept the argument about law-like regularities (which we should not) there is no possibility of conceiving of any underlying processes that produce them. Put in scientific realist terms, Waltz embraces a philosophy that focuses at the level of events, but ignores the level of the underlying structures – those unobservable social structures, causal processes and generative mechanisms that produce the events. Unless IR theory is to switch attention to the level of underlying structures, there is little possibility of discovering the reasons why North–South relations take on the unequal form that they do.

The ontological implications of positivist assumptions can be seen in most aspects of realist, neo-realist and other ‘rationalist’ theories of IR which are underpinned by positivist assumptions about unitary actors (states) engaging in rational behaviour, and engaging with one another in the manner of a billiard-ball model of interaction, which emphasises external connections producing regularities and predictable outcomes. This results in a reified social ontology that excludes underlying structures, causal mechanisms or constitutive processes. In other words, the very things that might explain the nature of IR and get at the exploitative basis underlying these (relating to the international capitalist system) are already excluded from analysis. As an alternative, SR can bring a clearer understanding of the ontological implications of positivism and other philosophical positions – something that is often denied or obscured in mainstream work. It is common to see such issues as ‘methodological’ such as is the case with the Second Debate, when in fact such debates have important ontological implications about what the world itself is like (as well as how it should be studied). Often positivism is talked of by IR theorists as if it is merely a methodology to be used in analysing the world rather than having its own ontological assumptions about what the world itself is like. Positivism is thus more than an epistemology or methodology; it is a philosophical position that makes ontological assumptions about the world as comprised of patterns and regularities that can be explained by the correct methodology. Yet positivism conflates these issues so that ontological matters are disguised as questions of epistemology and methodology.

An ontological approach such as suggested by SR then moves beyond questions of knowledge to ask what the world itself must be like for this to be the case.

The distinction between the knowledge we have of the world and the world that this knowledge is of is captured by Bhaskar’s distinction between the transitive and the intransitive. The chief error of alternative approaches within IR – whether constructivist, critical theory or post-structuralism – is to deny this distinction. Indeed, even the positivists, for all their talk of an objective reality, commit the epistemic fallacy of conflating their own identification of scientific laws (transitive) with the intransitive real processes they attempt to explain. Having said this, knowledge takes place within a social context which influences and shapes it. This is particularly clear in the discipline of IR given the way that the socio-historical context of the Cold War has shaped the manner in which IR problems have been posed and addressed. Theories are a product of the same society they seek to explain. This should not mean that we should abandon attempts at explanation because they are historically conditioned, but we should strive for explanatory adequacy or the best possible explanation. This would include explaining why it is that certain theories like neo-realism come to dominate thinking during particular historical periods, and how alternative theories can provide both a better explanation of real conditions, and account for the way that these conditions influence alternative explanations.

So we now move to what is distinctive about the kind of social ontology a scientific realist approach advocates. IR theories underpinned by positivism – realism, neo-realism, neo-liberal institutionalism, interdependence theory and so on – exhibit a shallow, surface or flat ontology that operates at the level of events without examining underlying causes. By focusing on the constant conjunctions of these events, or empirical regularities, mainstream theories of IR reduce ontology to behaviour and its outcomes, or conflate reality itself with their own theories of law-like regularities. Yet as Steve Smith notes, rejecting the positivist insistence on the observable and positing the idea of non-observable structures gives us the best chance of explaining social action (Hollis and Smith, 1990: 207). Following Bhaskar we should say that these structures ‘are non-empirical but empirically identifiable, transfactually efficacious but only contingently manifest in particular outcomes and they form the very real ground of causal laws’ (Bhaskar, 1986: 106).

However, and against those versions of Marxism or world-systems theory that invoke economic reductionism to explain the determining influence of one particular structure – the economy – on all others, it should be argued that each structure has its own specificity. Even if a structure, process or event is dependent on an underlying structure like the economy out of which it might be said to emerge, it cannot be reduced to this base level and we cannot explain the goings-on at one level simply by invoking another. Political processes can no more be reduced to the economic structure of production than mental processes can be reduced to physical ones, even if the former, in both cases, are rooted in the latter. Political and mental processes have their own specificity that is emergent out of the ‘lower order’ structure, but not reducible to it. Instead, we should conceive of these relations in terms of a stratified and differentiated totality of social relations. These different structures, processes and mechanisms are overlapping and co-determining. It is difficult to single out any one structure or process as autonomous in the way reductionist Marxism might. Structures are both externally and internally related. However, it should still be possible, within this, to identify which structures and tendencies are the more powerful or influential. Indeed if we are to come to an adequate understanding of IR we have to be able to do this and to highlight the power of economic processes within a complex set of structural relations. Marx’s own approach recognises this, implicitly at least, with its advocacy of abstraction to identify the dominant social processes while also stressing that ‘the concrete is … the concentration of many determinations’ (Marx, 1973a: 101). This points to the important tasks facing the social or political theorist. To identify what might be the most significant social relations, to engage in a process of abstraction to examine how these relations operate, while also recognising that in the open context of the real world such relations interact with other relations. The powers and liabilities present at the level of the real may or may not be exercised depending on a particular set of circumstances – SR, unlike positivism, does not reduce powers to their exercise. These powers exist as tendencies, rather than as fixed laws. The domain of the real is the domain of the possible, rather than the actual. Tendencies, if they exist, may have counter-tendencies. Processes may be the product of a coincidence of a number of different tendencies and causal powers. They may, in turn, act back upon the causal process and change it. In other words, SR points to an open world comprised of various complex sets of interrelations that produce a structured but differentiated social totality.

The social world is different from the natural world in that social structures are bound up with the activities that they govern while society is not independent of agents’ conceptions and, therefore, society has an essential conceptual element. This has led to constructivists in IR seizing on the structure–agency issue as indicative of their unique contribution to a social ontology of IR. Constructivists view structure and agency as mutually constitutive and have drawn on a range of philosophers from the hermeneutic and interpretivist traditions to support this view. This section will argue against such a conflation and will emphasise a clear distinction between the properties of structures and the properties of agents, giving rise to an ontological hiatus between the two (Bhaskar, 1989: 37). It is important to conceive of this relationship correctly if we are to then identify how agents must act in order to transform structures.

Wendt’s contribution to the structure–agency debate starts by comparing two of the influential theories in IR that have been mentioned earlier, namely neo-realism and world-systems theory. Wendt argues that the neo-realists reduce the structure of the states system to the properties and interactions of its constitutive elements (states), while world-system theorists reduce state agents to the effects of the reproduction of the world (capitalist) system (1987: 339). Wendt believes that the correct response is to show how structure and agency are mutually constitutive. Although he flirts with the ideas of Roy Bhaskar and CR, citing Bhaskar’s argument that ‘Society is both the everpresent condition (material cause) and the continually reproduced outcome of human activity’ (Bhaskar, 1989: 34–5), Wendt’s position really owes its argument to Anthony Giddens and structuration theory. The term Giddens uses to describe the relationship between structure and agency is ‘instantiation’. For Giddens, social structure and the activity of agents are mutually constitutive, with the structure only ever in existence (or instantiated) in the present moment of its renewal (see Rivas’s chapter).

By contrast, scientific realists like Bhaskar, Margaret Archer and Douglas Porpora reject this conflation of structure and agency. Archer’s alternative to structuration theory develops a position where ‘the social relations upon which they [agents] depend are held to have independent causal properties rather than being mere abstractions from our repetitive and routinised behaviour … these relations which constitute structures pre-date occupants of positions within them, thus constraining or enabling activity’ (Archer, 1995: 106). Structures and agents possess distinct properties. As Carter and New have argued, among the unique properties of structures are firstly anteriority – such things as legal systems and linguistic systems are preexisting features of the world we engage with, secondly endurance over time, and thirdly that structures possess powers of enablement and constraint – for example, existing allocations of wealth enable some and constrain others. Meanwhile, among the sui generis properties of agents are self-consciousness, reflexivity, intentionality, cognition and emotionality. Such properties of agency, and the properties of structure, are irreducible to one another (Carter and New, 2004: 5). Therefore, the relationship is one of ‘pre-existent structures, possessing causal powers and properties, and people, possessing distinctive causal powers and properties of their own, result[ing] in contingent yet explicable outcomes’ (ibid., 6).

This form of SR argues against instantiation by insisting that structures and agents have distinct properties and powers and that social structures are deeper and prior to the social activities that they govern as well as the conceptions that agents may have of this activity. One way to look it this is to focus on the way that while structures may depend upon human actions for their reproduction, these actions are already conditioned by the structures in a way of which the actors are seldom aware. It is accepted that agents act consciously, but their awareness is connected to the practices that they engage in rather than the deeper structures that these practices tend to reproduce. The model proposed here, then, is that agents act consciously within (positioned) practices, but that the effect of this is generally the unconscious or unintended reproduction of deeper social structures that the agent is largely unaware of. Agents may believe that they are acting within a particular context – like earning a living, getting married or generating development aid – but the consequences may well be to contribute to the reproduction of a wider set of structural relations – various aspects of the capitalist system, to put it crudely – beyond any intention that the agent may have. We can see how this model clearly goes beyond the social ontology presented by constructivism that focuses on practices and rule-governed behaviour, but does not situate these practices within a deeper social context (and thus find it impossible to explain what generates the rules and norms that agents follow in their practices).

A scientific realist ontology should therefore insist on the irreduciblity of structure to human actions or intersubjective relations. While social structures may depend on human activity for their reproduction, this reproduction is normally the unconscious rather than intentional product of such activity. Societies pre-exist the human agents who live in them and structures have prior causal power over agents. It is the prior existence of structures and their relatively enduring nature that provide the very conditions within which agents may act. This has the effect of defining and limiting the parameters of this activity. David Dessler, in his contribution to the structure–agency debate, criticises the neo-realist account of structure in IR, arguing that it refers to the spontaneously formed unintended consequences of action, as a by-product rather than a consequence of interaction (1989: 450). Actually structure is largely the unintended consequence of action. This is what distinguishes a structural approach from an intersubjective one. Social structures have an existence independent of the agent’s (proper) conception of them.

However, this caution about the role of agents is important if we are to correctly understand precisely what powers agents do have. Since structures are the reproduced outcome of human activity then a possibility does exist for agential activity to go beyond mere reproduction and engage in transformation. Such a transformation or development needs to be carefully located within the structural conditions, but it nevertheless exists. Under particular conditions or in particular circumstances agents may act consciously to change or transform these conditions, albeit within structural limits. This points to four important factors – context, position, limitations and possibilities.

The political scientist who wants to look at the possibilities for social transformation – and the analyst of IR should surely be one of these – needs to look at which social groups have the power or potential to act in a transformative way. These powers and liabilities derive from the structural location of these groups and the powers and potentialities conferred upon them. Usually social transformation only takes place under exceptional conditions when structural crisis occurs. Moments of structural crisis tend to throw the process of unconscious reproduction into question and agents become more aware of the situation confronting them and the possibilities open to them. At the same time, if some agents may become aware of their transformative capabilities, then certain other agents will attempt to resist this. The process of transformation encounters resistance from those who have an interest in maintaining the structure as it is. They will engage in an act of conservation. Thus struggles break out over transformation and conservation of structures. These struggles can be called hegemonic struggles where agents, through their practices, act to preserve or transform a given set of relations (Joseph, 2002: 38–40).

From explanatory critique to emancipatory critique

We can now put the various sections together to show why SR is a holistic philosophy with radical consequences. First, scientific realism insists on the independent nature of the intransitive object in opposition to those attempts – including many alternative theories in IR – which blur the boundary between reality and the knowledge we have of it. This separateness is the very thing that makes knowledge intelligible. Without this, knowledge becomes the product of the subject, the collective, the group or community or else is bound up with some sort of worldview, discourse, text or language game. Whether we take a hermeneutic, idealist, constructivist or post-structuralist stance on this, the end result is a relativisation of knowledge and a denial of either a necessary relation to an object or else a denial of ‘objects’ themselves. This does not mean that we should not recognise the impact of such things as communities or discursive contexts as necessary features of the transitive production of knowledge. Knowledge is of an object, but at the same time it is produced within a specific social and historical context and within particular communities. Knowledge must be understood in relation to the transitive conditions of its production and in relation to the intransitive object. Against post-structuralism and constructivism, SR insists on the separateness of knowledge and the objects of knowledge; against positivism it insists on the social and historical nature of knowledge. It also insists on its critical nature.

Social science – political science perhaps more than all others – is implicitly critical in nature, criticising other ideas and the claims that they make and the inadequacies of their explanations. In drawing attention to these inadequacies, critical social theory should attempt to explain why these inadequacies occur and what it might be that produces them. It must ask the question as to whether it is something about the object of inquiry itself that gives rise to false theoretical beliefs. If a connection can be made between social objects themselves and the ideas we have of them, then a critique of these objects – as well as of the theory – follows. An explanatory critique does not simply criticise the theory at the level of ideas, but examines the relationship between the theory and that which it is trying to explain and may, ultimately, critique that which theory is trying to explain for helping to generate such false ideas. Such is the case, for example, with Marx’s critique of capitalism and his linking of this to the failings of classical political economy. The same point might be made in relation to IR, and theoretical approaches to IR such as realism or neo-liberal institutionalism.

Explanatory critique flows from the distinctive social ontology developed by SR. It is radical in the sense of identifying the dominant social structures and mechanisms, examining how they operate and what they do. In the social world the question of how these structures are maintained and reproduced is important. This also raises the subsequent question of whether such relations might be other than what they are and whether they might be transformed. An explanatory critique becomes an emancipatory critique when it points to the wider consequences of social activity, looking at how structures operate, the effects they have and how social structures generate and distribute on agents certain causal powers, identities and interests. It points to the social context of agents and their activities and the range of possibilities open to them.

SR does not offer radical political solutions to the questions it raises. It is radical by raising these questions in the first place, when other philo sophies hide behind the neutrality of the status quo. It opens up social processes to scrutiny while putting social theories and ideas in their appropriate social context. The partisan stance of SR comes not from offering political solutions, but from criticising the assumptions of social theories and the way these might be linked to dominant social structures and power relations – for example, showing how mainstream IR theory reinforces a reified view of structures and agents that is consistent with existing power relations.

Summary: How to study international relations

SR is a meta-theory. But in social analysis, due to the conceptual nature of that which we study, it is difficult for meta-theory to remain as such. This has led to confusion with SR being claimed as a theory of IR. Rather than trying to do this, it should instead provide a critique of existing theories of IR, while making enough ontological claims to provide assistance to those approaches best placed to account for IR. While we have seen Brown tries to claim that SR equals Marxism, this does not necessarily have to be the case, and we would wish to argue that even if SR is put to work as an underlabourer to Marxist IR theory, SR would always aim to question the claims that such theories make, in particular warning against Marxism’s tendencies towards reductionism, economic determinism, teleology and other potential problems.4

If SR can offer anything to an analysis of IR, then it is a critique of the assumptions of dominant theories, especially of IR, which prevent depth enquiry and hide the real nature of things. The empirical realism upon which such approaches are founded ‘establishes at once the destratificatification of being in ontology and the dehistoricisation of knowledge in epistemology’ (Bhaskar, 1986: 96). For example, Branwen Gruffydd Jones (2001) has looked at how the World Bank’s analysis of poverty reduces the issue to those things that can be observed and measured without any reference to underlying relations of production or those deep-rooted causes of the problem. In particular, such approaches lack sufficient social context, while at the same time denying the social nature of their own assumptions by hiding behind claims to neutrality and scientific objectivity. Such approaches exemplify the mystification of the fact-form which, like Marx’s commodity-form, hides from view the underlying social relations. A scientific realist approach should begin by raising the issue of a depth ontology that deals with the question of dominant social structures and generative mechanisms.

Rival philosophies like constructivism also claim to deal with such ontological questions. But SR goes beyond constructivism’s intersubjective social ontology in that the social world cannot be reduced to the interactions between people, shared norms, rules and understandings. SR argues for a deeper sense of structure that can explain the context within which rules might operate. Human action, meanwhile, might be intentional at the level of human practices, but in the main, it unintentionally reproduces pre-existing, deeper structural relations.

Why should this social ontology be good for IR theorists? First, because it is argued that it is the structured nature of the world itself that makes knowledge possible and meaningful. Against postmodernism and other forms of post-positivist IR, this allows us to make clear statements of an ontological nature. Our understanding of North–South relations should therefore be grounded in a social ontology based on underlying social relations. This moves us away from the positivist world of events and states of affairs where discussions such as that of the World Development Report on Poverty can get no further than, at best, suggestions that amount to an amelioration of affairs. It also avoids rooting discussion in liberal discourses about rights. To have rights, we need social transformation. To have social transformation, we need to base our approach on a social ontology that normative theory largely avoids.

Next, this type of approach is critical of other theories and develops this critique by looking at the relationship between such ideas and the social and historical context within which they are located. The weaknesses of realist IR or the statements of the World Bank must be related to actual power structures within society if the arguments of such theory are to be fully explained. Once this is done, the question of the status of such structures, practices and institutions itself becomes a matter for discussion. In relation to North–South questions, as with relations of explanation in general, we might say that the exploited have an interest in knowledge that the exploiters do not have. Wrong social theories simply do not have the ability to explain real structures, social processes and contradictions. Thus to truly tackle issues like the North–South divide is to already engage in a partisan form of politics. From here, we might go to the matter of social transformation. The structure–agency question poses the possibility of transformation and it is the task of the social theorist to identify such things as what is possible, what is not possible, who has the power to transform and what obstacles need to be faced. Again, it needs to be stressed that SR itself does not perform this task. Instead its radical nature comes from the way it conceives of the connectedness of these issues, its advocacy of a totalising critique, and the fact that unlike positivist, constructivist and post-structuralist approaches, it has led us to this conclusion.

Structure of the book

We start the book with two chapters that are more focused on analysing the discipline of IR, albeit in a critical way. Together they challenge the way the discipline of IR sets up its ‘levels of analysis’ and which instead explore the idea of emergence. David Leon’s chapter focuses on the scientific realist idea of emergent properties, i.e. causally efficacious properties that are neither ontologically reducible to, nor analytically exhausted by, their constituent parts. He notes how in recent years, there has been a growing recognition among philosophers of science, theoretical biologists and some social scientists that emergence can provide the basis for strong anti-reductionist arguments. Contemporary with this interdisciplinary development, however, is a demand within IR for explanations to be grounded in microfoundations, often cast in terms of opening up ‘black boxes’ of states and integrating domestic and international politics. Leon’s chapter first shows that the importance attached to the micro-level rests on notions of scientific explanation that are philosophically problematic. Then, drawing on the theory of emergence and combining it with a scientific realist stance on the ontological status of causal entities, this chapter argues that there are good grounds for rejecting empiricist and reductionist arguments in favour of those which take full account of emergent properties in the international system.

Like Leon, Jonathan Joseph challenges the way that IR has traditionally set up its field of analysis. According to Joseph, while post-structuralists in IR have indicated the problematic nature of such oppositions as inside/outside, anarchy/sovereignty and self/other, his concern is to illuminate how these issues are themselves largely underpinned by an unproblematic acceptance of a framework structured by notions of system and unit. This framework, Joseph argues, has been reinforced by the work of neo-realism and the levels of analysis debate to the point where it is accepted even by critical IR scholars. By and large the international has come to be defined as a structure or a system, while the units may range from individuals to states. This distinction runs parallel to other (perhaps more methodological) oppositions such as holism/individualism and macro/microtheory. While there are many debates as to where to place the emphasis, there are not so many debates about such distinctions themselves. Joseph’s chapter is concerned with showing how we might question such a way of seeing things and examines the Marxist tradition to see how things might be conceptualised differently.

The next two chapters raise important questions about how IR theorists should understand the state. Marjo Koivisto’s chapter addresses important questions about the nature of states as entities. Scientific and critical realist contributions on the state in IR have posited that instead of being mere fictions or like units of a system, states are real entities in world politics in possession of specific causal properties. Causal powers of the state have been suggested to arise from states being people-like social organisms, or from agents acting in state roles. Yet as existing accounts take IR theories as their starting point to examine the state in the international, they leave open many questions concerning the range of social systemic and structural emergent properties of the state. Starting from Weberian and Marxist theories of the state in historical sociology and global political economy of IR, Koivisto’s chapter advances a framework for a realist analysis of the historically specific causal properties of the state. She argues that realist analysis can further existing theories of transformations of contemporary capitalist states (including processes of privatisation and norm socialisation) in world politics, and thus offer a powerful analytical alternative to rational choice and constructivist theories of the state for the field of IR.

In his chapter Doug Porpora argues that it is important for IR to better understand how states as complex collectives come to act on the world stage. Toward that end, the chapter examines how the United States reached a collective decision to attack Iraq and what it means and does not mean to call that decision a collective one. In the process, the chapter illustrates how a critical realist perspective permits a synthesis of two approaches normally considered opposed: political economy and discourse analysis. The decisionmaking process will be shown to have been a decentred one, dependent on rhetorical manoeuvres among elite actors within a field of structured power relations and against a background of public inattention. The account Porpora develops will hardly accord with the traditional IR conception of nation states as simple rational actors pursuing clear national interests. Called into question as well is the traditional Kantian confidence in the moral behaviour of democracies.

Recent work on the theory of international law is addressed in Bill Bowring’s chapter. Although the next few chapters cover a diverse range of topics, they follow on from Porpora’s chapter in taking normative questions as a key concern. Bowring is concerned with the way that social constructivism has colonised many areas of the field precisely because it offers what he considers to be a false normative allure. His chapter explores the genesis and trajectory of social constructivism in IR theory through the work of Christian Reus-Smit (2004), Antony Anghie, Dino Kritsiotis and David Wippman, arguing against the idea that their methodology is a progressive one. The social practice of law, it is argued, is better seen through the methodology proposed by the Bhaskar’s early work (1979, 1989). A rich variety of social structures, displaying causal powers, existing across human generations, have their own laws and dialectic. However, international law and human rights have commonly been analysed as normative discourses, ignoring the material conditions that make these discourses possible.

In looking at the construction of responsibility, blame and sanction in loans to states in the course of dealing with debt problems since the 1970s, Tomohisa Hattori also examines important normative issues. His chapter argues that this process can be best described as the gradual institutionalisation of the assignment of responsibility, blame and sanction based on the juridical construction of knowledge. Because juridically constructed causes often contradict real social causes – which include, at a minimum, the agency of creditors and indebted states, institutional settings and structural conditions – they are inaccurate and inadequate bases for ethical inquiry. The chapter provides what critical realists call an explanatory critique that examines how the dominant ethical discourse about a social practice, such as the debt reduction practice, not only justifies the existing order and power relations surrounding the practice but also contradicts a scientific explanation that clarifies the causal mechanism of this practice.

Milja Kurki’s chapter aims to demonstrate how a critical realist approach to research can contribute to the analysis of processes of democratisation. She argues that many existing frameworks of democratisation have tended to work with empiricist, or quasi-empiricist, assumptions regarding the law-like logic of democratisation and methods appropriate to study democratisation. She attempts to show how in the adopting of procedural understandings of democracy they have often failed to study democratic political systems as conditioned by complex social and economic contexts. She also demonstrates that many researchers do not recognise that deeply held normative underpinnings often inform the seemingly neutral social scientific understandings of democracy and democratisation in the field. The chapter does not seek to develop a new theory of democratisation, but rather only seeks to use CR to highlight the kinds of considerations that researchers in IR should keep in mind in approaching an empirical field of study.

In his chapter Heikki Patomäki uses CR to explore the relationship between closed and open systems and then examines the implications of this for important research into possible futures. According to Patomäki, it has become increasingly evident that systems are always open and closed only to a degree. Equally, the allegedly radical asymmetry between explanation and prediction does not hold and the future can also be analysed in terms of more or less likely conditional possibilities of becoming. This entails that we should therefore try to overcome the prevailing past-orientation of human and social sciences and build a methodology for futures studies. The point of his chapter, then, is to demonstrate how futures studies can work in practice, rather than by deploying abstract conceptual arguments. As examples he engages in a critical discussion of various scenarios of global futures until 2045 and shows how these scenarios should be reassessed and developed further in the light of new evidence. New evidence, he argues, can assume two forms, concerning either (i) our explanatory models of the past and present or (ii) how world history unfolds over time as a possible future is becoming present. In particular, his attention is focused on the implications of the global financial crisis of 2008–9 and on the new evidence about the acceleration of the process of global warming.

The next two chapters deal more explicitly with Marxist theory. Faruk Yalvaç, in particular, is concerned to look at the relationship between Marxism, CR and IR theory. His chapter argues that a Marxist social theory enriched by the principles of CR can provide a radical alternative to both the positivist and constructivist accounts of IR, which ultimately consolidate and reproduce the worldwide system of capitalism, along with its structured inequalities. In this current conjuncture of IR theorising, Marxism, enriched and developed by CR, it is argued, provides the best way to overcome the current stalemate brought about by the rationalist/reflectivist or positivist/ post-positivist divide.

Bob Jessop’s chapter considers how theorists from different perspectives have been investigating the dynamic of the world market, the possibility of a world state or global governance, and the emergence of a distinctive world society. His contribution subjects each of these concepts to a critical interrogation, paying particular attention to the emergent logics of the world market, the prospects of a world state, and the reality of world society. In each case it explores the spatio-temporal as well as material–institutional and discursive–strategic dimensions, paying particular attention to the role of spatio-temporal fixes in displacing and deferring contradictions and conflicts as the basis for the temporary stabilising of interstate relations, and a shrinking world society. It is emphasised that there is no single dynamic to world society but a series of conflicting dynamics with no overall logic and that this poses a series of problems, both theoretical and practical, on how meta-stability, if any, can be secured on a global scale. Some reflections on recent developments in systems theory are introduced to address these problems and they are linked to reflections on the ecological dominance of the world market within the overall development of world society.

The final chapter by Jorge Rivas returns to some of the main themes of this volume and provides an overview of the benefits an SR approach brings to IR. It does this by comparing the possibilities offered by SR with the problems presented by constructivist approaches to IR. Above all, he addresses the hugely influential work of Alexander Wendt, and is particularly concerned with the detrimental effect of Wendt calling his approach SR. He suggests that in making the case for the scientific study of ideas in international politics, Wendt’s version of SR preserved many vestiges of the positivist model of science. In order to establish scientific validity within a discipline whose philosophical vision has not yet discarded the mistaken conflation of science and positivism, Wendt constructed a middle way between the alleged positivist/interpretivist dichotomy, rather than rejecting both altogether and turning to a fundamentally different alternative offered by philosophical realism. This chapter explains how the failure to complete the break with the positivist model of science has distorted and undermined his presentation of SR, created confusion around the content and implications of SR, and compromised the valuable philosophical potential of his wider project. Given the great importance and widespread impact of Wendt’s contributions in IR to the understanding of SR in particular and knowledge of philosophy of science in general, this chapter will refine and, where appropriate, correct Wendt’s presentation of the realist positions on ontology, epistemology, material–ideational dualism and the agent–structure relationship, with the intention of maintaining and furthering Wendt’s philosophical objectives while overcoming the limitations inherent in his presentation, and thus clarifying these vital theoretical issues for our discipline.

Notes

1. Despite their apparent closeness on some issues, CR is useful to Marxism precisely because it is different from Marxism and can therefore act as a critical tool that can subject Marxist claims to scrutiny in a way that traditional forms of Marxist philosophising cannot. There is much within the Marxist tradition that a CR approach would question, ranging from economic determinism to structural Marxism to praxis theory. And above all, CR would criticise the idealist view that Marxism can be a self-contained theory that does not need to engage with and incorporate non-Marxist approaches (see Joseph, 2001).

2. Bhaskar labels his approach transcendental to indicate the close links to Kant. This is not identical with Kant’s transcendental approach. For Kant, transcendental logic is the apparatus of concepts and principles, common to all rational minds, that organises experience and is thus logically prior to it. Thus for Kant, a transcendental inquiry would be one in which a critical philosophy works out the presuppositions of our knowledge. In effect, the question: ‘What must be the case for knowledge to be possible?’ Having established these presuppositions they must be considered ahistorically, existing as they do prior to experience. It is for this reason that Kant labelled his philosophy ‘Transcendental Idealism’. Bhaskar, on the other hand, asks the question: ‘What must be the case for science to be possible?’ Now, insofar as science is a human practice with a discernible history Bhaskar’s conclusions cannot be ahistorical; that is, they provide no timeless routes to secure knowledge. Moreover, since Bhaskar’s concern is with both the ontological and the epistemological conditions of possibility for science it is simply incorrect to view him as giving ontology priority over epistemology. Science makes no sense without both dimensions. For Bhaskar, the very conditions of possibility for science require both dimensions.

3. Chernoff (2002) argues that science does not need this depth realism. However, while his points concerning the conventionality of all science are well taken – indeed SR would insist on it – the realism issue refuses to go away. The role of conventionalism in science is important, but the practice of science suggests that it is not the core of science as an inherently sceptical enterprise. No matter how much general agreement there might be concerning core issues of any science, the practice of scientists is to take any and every claim as potentially subject to doubt. But why do they do this if we have agreement and modes of validation, perhaps related to the practical implications of theories for practice? Only the commitment to realism can explain this constant process of critical enquiry. Indeed, a conventional account of science would tend towards conservativism.

4. For more detail on the relationship between SR and Marxism see Joseph (2006).