Social science and the philosophy of science

There are several varieties of philosophical realism.Footnote ² As the philosopher Hilary Putnam once suggested, it has displayed many ‘faces’,Footnote ³ and he himself, in the course of his career has exemplified some of those countenances. Although these visages are not all easily differentiated, it is possible to make some basic distinctions. Ontological or metaphysical realism involves the broad, and sometimes ambiguous, claim that there exists, what is often referred to as, a ‘mind-independent’ reality which is the basis for empirical truth-claims.Footnote ⁴ Most of the individuals who subscribe to the specific label of scientific realism accept a version of this assumption and argue that scientific theories, including statements about unobservable entities, are true or false by virtue of the extent to which they correspond to such a ‘reality’. Some philosophers, such as Michael Devitt, stress the difference between the existence of scientific (as well as common sense) objects and the way in which they are conceptualised, but Devitt is also a reductionist who claims that all real objects are physical in nature and ultimately explained by physical laws.Footnote ⁵ Although it is admitted that this ‘world’ of objects cannot be conceptually articulated apart from the claims of science, as set forth at any particular time, it is posited as a transcendental standard which is viewed as more than a regulative ideal. Most realists embrace both semantic realism, and the idea that truth is a matter of correspondence between language and reality, as well as methodological realism, or the assumption that the basic goal of science is truth rather than ‘saving the appearances’ by simply accounting for and predicting observable phenomena. The viewpoint of many natural scientists may very well appear to be similar to scientific realism, but probably as many scientists, if pressed to give a meta-theoretical account of their activity, would offer something that might appear closer to some version of anti-realism or the philosophical thesis that theories are basically those propositions that science takes as justified belief in some specific context and which make sense of what is accepted as factual data.Footnote ⁶ Despite some obvious resemblances, it is also important not to confuse the meta-scientific reflections, such as the conflicting positions of scientists like Albert Einstein and Werner Heisenberg on matters such as quantum mechanics, with formal philosophical positions such as realism and anti-realism.

The turn to philosophical realism in International Relations is a relatively recent tributary of an antecedent relationship between realism and social theory, which extends back at least to Lenin's critique of positivism, and it is only the latest chapter in the more inclusive story of the relationship between philosophy and social science.Footnote ⁷ Although much of this work has in part been directed toward countering orthodox empiricist and positivist images of scientific explanation in mainstream social science, it has perpetuated misconceptions about the relationship between philosophy and social science. Whether social scientists have wished to emulate, or distance themselves from, the methods of natural science, they have characteristically, for at least half a century, drawn not upon a study of the practice of natural science but upon the literature of the philosophy of science for an account of the nature of scientific explanation. The philosophy of science, however, has hardly been a stable and uncontentious literature, and its relationship to the practice of science is highly problematical and a matter of dispute among philosophers themselves. During the nineteenth and early twentieth centuries, images of explanation in natural science were somewhat vaguely depicted in the literature of both philosophy and social science. By the middle of the twentieth century, however, logical positivism and logical empiricism (L P/E), represented by philosophers such as Rudolf Carnap, Carl Hempel, and Herbert Feigl, had been transplanted from Europe to the US where this school of thought not only contributed to the institutionalisation of the philosophy of science as a distinct field of study but established a philosophically dominant reconstruction of the logic and epistemology of natural science. This account, which stressed the unity of scientific method, became authoritative for disciplines such as political science which, for various reasons, were insecure about their scientific identity. If political scientists, during the behavioural era, had repaired to the philosophy of science merely for rhetorical validation, it might not have been any more significant than the potted accounts of scientific method that are typically contained in introductions to textbooks in the natural sciences. Social scientists, however, often viewed the philosophy of science as not only a description of scientific practice but as the basis of methods and precepts involving, what was often referred to as, ‘theory-construction’. Few social scientists had either an adequate grasp of the nature and origins of the philosophies to which they subscribed or more than a tertiary acquaintance with the primary literature. These images functioned largely as the basis of a methodological ethic demanding, for example, that explanation be couched in terms of law-like generalisations which could be tested by factual observation. Despite the rejection of positivism by social theorists who have turned to realism, the notion that the philosophy of science is a guide to scientific practice has persisted.

Meta-theories such as realism and anti-realism, like the philosophy of science in general, are not the foundation of scientific inquiry but, instead, post-hoc reconstructions of what are taken to be the logic and epistemology that underlies the practice of science. Often, in contemporary philosophy, these have been quite divorced from any grounding in the practice of science but have been extrapolated into transcendental normative claims about the standards of inquiry. This form of discourse began as an indigenous rhetoric which, in the work of Descartes, Newton and others, was entwined with substantive scientific claims and designed both to valorise certain scientific theories and to justify claims to scientific knowledge in opposition to rival, and often religious, authorities. When, in the late nineteenth and early twentieth centuries, these epistemological arguments began to differentiate and take shape as an autonomous discourse and eventually as an academic practice, they were still often tied to specific scientific theories. In the case of Francis Bacon, August Comte, and John Stuart Mill, these accounts of science were deployed as justifications for social programmes, and it is especially ironic that political scientists have failed to recognise the ideological background and character of the philosophy of science. Positivism, as expressed in the 1929 Manifesto of the Vienna Circle,Footnote ⁸ was devoted to challenging traditional intellectual and social authority and to justifying what was referred to as the ‘scientific view of the world’ as well as the liberal and socialist political ideologies for which it was viewed as the vanguard. This modernist programme was also in part an attempt to re-secure the philosophical foundations of science after the decline of the assumption that Newtonian theory was a ‘realistic’ description of the world. Nineteenth century philosophy and science were already wary of speculative thought, and the subsequent crisis in physics, precipitated in large part by the work of Einstein, contributed to the positivist idea, already embraced by Ernst Mach, the intellectual father of positivism, that while theories were ephemeral and transient instrumental constructs, facts were constant and in some manner open to immediate perception.

Although L P/E stressed the nomological and deductive character of explanation, it posited observation both as the foundation of scientific knowledge and as the basis for verifying or falsifying empirical generalisations. Theories were, in the early years, largely construed as somewhat heuristic or instrumental cognitive devices for generating descriptive and predictive statements about observable phenomena rather than as claims which, themselves, could be judged as true or false. Some philosophers even toyed with the idea that theories, once they had functioned as devices for accessing and economically organising and describing facts, were eliminable, but what Hempel termed the ‘theoretician's dilemma’,Footnote ⁹ that is, the practical need for theories despite their diminished cognitive status, would be a persistent problem. The exact nature of what was taken to be the observational domain was contested, elusive, and ranged from phenomenal sense-data to gross physical objects. The attempt to specify such a domain drew positivism to the work of Bertrand Russell and especially to the early philosophy of Ludwig Wittgenstein, but it was in large measure the failure to determine the character of this putative theoretically untainted foundation and the shift in Wittgenstein's position that, beginning in the middle of the twentieth century, led to the decline of L P/E.

By this point, however, some of the basic doctrines of L P/E had become deeply embedded in the practices of social science. Although behaviouralism is often viewed as having mounted an attack on the subfield of political theory and qualitative approaches to political inquiry, the actual situation was somewhat the opposite. In the post-war years, there was an unprecedented attack by European émigré political theorists, ranging from Leo Strauss to the Frankfurt School, on what had been the virtually unchallenged commitment of political science to some version of empiricism and the unity of scientific method. The émigré scholars who propagated logical empiricism came to prominence in the philosophy of science at the same point in time and offered a schematic image of scientific explanation which was embraced by many political scientists as a basis for validating the scientific identity of political science. In a peculiar way, the terms of the debate about behaviouralism were to a large extent the reflection of a conflict between transplanted European philosophies.Footnote ¹⁰ At the very apex of its influence, the basic tenets of L P/E were challenged by individuals as diverse in some respects as W. V. O. Quine, Karl Popper, Thomas Kuhn, and Paul Feyerabend. These criticisms, which focused on the character of scientific theory and called into question what Quine referred to as the ‘dogmas of empiricism’, were embraced by some opponents of the behavioural programme in mainstream political science. Although the remnants of L P/E remained sedimented in much of the language of social science, it had, by the last part of the twentieth century, been largely discredited in philosophy, and social scientists, who had taken their cues from L P/E, suffered a new identity crisis. Questions were raised not only about the doctrines of L P/E but about the extent to which the philosophy of science could, in principle, be a resource for the judging and prescribing the practice of either natural or social science. The collapse of the hegemony of L P/E in philosophy was not followed by the emergence of another clearly dominant school of thought, but scientific realism became one of the principal contenders.

Philosophical realism

Although empiricism, because of its emphasis on observable facts, has sometimes been equated with realism, contemporary versions of scientific realism, as well as nearly all forms of post-positivism, are defined in part by a challenge to the empiricist account of scientific theory. Realists not only reject the basic assumptions behind the instrumentalist interpretation of theory but sometimes claim that their rendition of scientific explanation is not only a description of scientific practice but itself a kind of empirical hypothesis which is supported by the practical success, and progress, of science in explaining, predicting, and controlling natural events. Although realism involves a rejection of traditional empiricism, its basic philosophical aspirations have been largely the same, that is, to demonstrate how scientific knowledge is possible and authoritative. Philosophical realism is opposed by versions of anti-realism which, while also unsympathetic to traditional empiricism, are sceptical of the metaphysical premises of realism as well as of the claim that the truth of scientific theories, and the doctrine of realism itself, can be demonstrated by the history and practice of science. Anti-realists, such as Dummett argue that much of the conduct of natural science is based on anti-realist premises. Anti-realists do not revert to the idea that theories are merely tools for apprehending observable facts, and, like realists, they maintain that theories are often actually about unobservable entities. They claim, however, that the value of theories cannot be based on an abstract criterion such as correspondence with the ‘world’ and instead argue, like the ‘constructive empiricism’ of Bas van Fraassen,Footnote ¹¹ that good theories are not necessarily literally true. Both realists, and anti-realists such as Larry Laudan,Footnote ¹² are, on the whole, however, like realists, worried about arguments such as those of Kuhn,Footnote ¹³ which they claim threaten to undermine the objectivity of science by questioning the philosophical commensurability of scientific theories.

These objections to work such as that of Kuhn are, however, something of a red herring. It is not the objectivity of science that is at issue but rather a particular philosophical idea of objectivity and the capacity of philosophy to specify criteria of commensurability. Although Kuhn is sometimes categorised, and criticised, as an idealist because of his argument that factual statements are theoretically constituted, his position, as well as that of individuals such as Wilfrid Sellars (who coined the phrase the ‘myth of the given’ in his critique of the central tenet of positivism), Richard Rorty, and Nelson Goodman, might be best described as theoretical realism.Footnote ¹⁴ They all agree that what we mean by the ‘world’ and ‘reality’ is basically a function of operative scientific theories and that it makes little sense to speak of an unrepresented ‘real’ world to which scientific concepts must ultimately correspond. There has been a kind of mythology that has been attached to Kuhn's work which ascribes to him an inversion of the positivist theory/fact dichotomy and the view that that facts are ‘theory-laden’, but this phrase, once used by Norwood Russell Hanson,Footnote ¹⁵ but not signifying what critics often impute, misses Kuhn's primary point, that is, that theories in science are ontological claims specifying what ‘really’ exists and the manner of its behaviour and that philosophy has no basis for positing a more profound ontology or account of reality and criteria of scientific truth. The term ‘theory-laden’ is misleading, because it assumes a binary relationship between theory and fact that was at the core of positivism. For Kuhn, theories are conceptually constitutive of facts.

Most self-ascribed forms of realism, however, remain tied to the basic problematic of representational philosophy and the correspondence theory of truth, which pivot on the epistemological problem, that emerged with Locke and modern empiricism, of how thought and language make contact with an external world. Philosophers as diverse in many ways as Wittgenstein and Donald Davidson have challenged the basic premises of representational philosophy.Footnote ¹⁶ Kuhn's contribution to the transformation of the philosophy of science did much to propagate the core assumptions of anti-representationalism which Rorty summed up in his critique of philosophy's claim to have the capacity to specify the basis on which mind could mirror nature. And now even Putnam who had done so much to bolster and popularise the realist position has, despite some lingering differences with individuals such as Rorty, rejected such a bifurcation of language and the world or at least challenged the idea that philosophy can supply some general answer to the problem. Theories are not viewed by these individuals, as some critics claim, as, so to speak, creating the world, but rather theories are construed as claims about what constitutes the world and which are often incommensurable with respect to both past theories and competing reality or world-defining claims such as those of religion and common sense. For these individuals, there is, and cannot be, any general philosophical answer to questions such as why one theory displaces another. Each transformation carries with it its own story of conflict, persuasion, and resolution. If, for example, we detailed the dramatic shift in the middle of the twentieth century that brought plate tectonics and continental drift into prominence in geology, it would be close to meaningless to say, from a philosophical standpoint, that the underlying reason was that the theory corresponded more closely to reality. The term ‘reality’ simply began to refer to a different concept. All of this constitutes a fundamental challenge to philosophical realism.

Although L P/E is now philosophically obsolescent, there has, at least since the mid-1990s, been increasing difficulty in assessing exactly where the current weight of opinion resides in the philosophy of science and where the lines are to be drawn as far as parsing the arguments within and between versions of realism and anti-realism. Even some realists suggest that the only consensus in the philosophy of science is with respect to the assumption that logical positivism is dead and that there is a need to sort out the basic contemporary contending positions.Footnote ¹⁷ Arthur Fine has suggested that both realism and anti-realism are basically defunct and that the best approach is to adopt what he has termed the ‘natural ontological attitude’,Footnote ¹⁸ which amounts to the claim that we should accept scientific claims as true, much in the same sense that we accept common sense assertions. Realism, anti-realism, and L P/E are, however, all representationalist philosophies, and the initial burden in all three cases is to demonstrate that the very questions that they pose about truth and reality are, outside a particular scientific context, philosophically and practically meaningful. Despite its remaining influence in social science, the philosophical decline of L P/E has left social science with considerable perplexity about its scientific identity, and it has left critics of a ‘scientific’ approach to inquiry with an anomalous target. This situation has, in some quarters, prompted a search for a new meta-theory and rhetoric of inquiry – a new scientific ideology. Such a project, however, has emanated less from mainstream political science, in which ideas, or at least slogans, characteristic of L P/E remain dominant, than from a new wave of critics who, for various reasons, feel the pangs of scientific insecurity. This may be somewhat ironic in view of the fact that even though the philosophy of science, from its prototypical origins to the present, has been devoted to validating the activity of science, it has seldom been advanced by philosophers as a foundation for scientific practice.

The normative language and tone of L P/E may have belied the claim of individuals such as Hempel that their goal was neither to judge nor to instruct scientific practice, but, today, most philosophers of science, maybe with the exception of some of those who follow Popper, do not present their work as a model for informing practice. To the extent that recent invocations of realism point to the deficiencies of the positivist meta-theory and its liabilities when applied to social inquiry, they are helpful. The claim, however, that the alternative is the introduction of another meta-theory is where the position most significantly goes awry. But while positivism was largely unreflectively absorbed into the discourse of social science, contemporary advocates of realism have almost programmatically turned to the philosophy of science as a basis of social scientific inquiry. Self-ascribed ‘realists’ in social science have also been concerned with countering what they believe are some of the inhibiting relativistic implications of certain arguments in post-positivist philosophy such as hermeneutics and post-modernism, even though, in their deference toward philosophical authority, they are also wary of not, in some way, incorporating these views. In this respect, as well as with regard to their general belief in the need for a scientific meta-theory, contemporary advocates of realism largely repeat the concerns that, two generations earlier, attracted social science to positivism.

Part of the criticism of positivism in social science emanated from the work of individuals such as Peter Winch,Footnote ¹⁹ who challenged claims about the unity of science and argued for the logical autonomy of both social phenomena and social scientific inquiry. While these arguments were welcomed by social scientific critics of positivism, they also, like the work of Kuhn, called into question, at least implicitly, the basis of claims to epistemic privilege on the part of meta-practices such as philosophy and social science. Although Kuhn did not focus explicitly on the issue of the relationship between philosophy and the practice of science, what prompted so many strong reactions against his work was the fact that he had, in effect, transferred the issue of specifying the criteria and locus of scientific truth and reality to science. This relegated philosophy and meta-theoretical claims in general, to what many considered to be a diminished role. The ‘relativism’ that is attributed to positions such as that of Kuhn and Winch is not, however, something that is a threat to the integrity of the practices that are the object of inquiry but rather a threat to the claims of cognitive authority characteristic of much of philosophy and social theory.Footnote ²⁰ As a position in this contest between, what is sometimes referred to as, ‘rationalism’ and ‘relativism’,Footnote ²¹ philosophical realism has not only gained a following in philosophy but has been invoked in a number of fields including social science, ethics, literature, and legal theory. The arguments of the philosopher John Searle are typical of these concerns.Footnote ²² Although Searle, maybe more than anyone else, has advanced a detailed account of what he designates as the unique character of ‘mind-dependent’ or ‘constructed’ social facts, he has insisted that such facts are erected on the foundation of ‘mind-independent’ or ‘brute’ facts in terms of which they can be both explained and judged on the basis of a correspondence theory of truth. Searle, like someone such as Alan Bloom,Footnote ²³ worries about the decline of modern culture in the face of relativistic tendencies. This same tone has carried over into the language of recent social scientific proponents who often espouse a ‘critical’ version of realism.

‘Critical realism’, which is often closely associated with ‘scientific realism’, was a term originally coined in the early twentieth century by the American philosopher Roy Wood Sellars whose naturalistic image of science was formulated in opposition to nineteenth century idealism. The phrase is still used to refer to a general contemporary philosophical argument,Footnote ²⁴ but in social theory, ‘critical realism’, in its various incarnations, involves the search for a meta-theory devoted to sustaining the idea of explaining and evaluating social phenomena in terms of the existence of unobservable structures, generative mechanisms, and underlying causal relations. In some instances, this approach builds on the Marxist tradition of structural explanation which, as in the case of Lenin, invoked a realist philosophy of science to support the theory of dialectical materialism as well as entailed claims about exposing the realities behind false-consciousness. Although critical realism often seeks support from the literature of scientific realism, much of contemporary critical realism, as well as other social scientific appeals to realism, are philosophically, to say the least, eclectic. What is involved is often less a one-dimensional defence of some version of philosophical or scientific realism than an attempt to reconcile various and sometimes contending post-positivist epistemologies, account for what is sometimes viewed as the problem of sorting out the relationship between agency and structure, and yet retaining a critical perspective with practical implications. The meta-theorist Christopher Norris, who has been one of the most prolific and accurate secondary summarisers and proponents of realism, is also a fierce critic of post-modernism and deconstructionism.Footnote ²⁵ He has been adamant in defending hard-core philosophical realism against a wide range of intellectual currents which he believes have threatened the idea of reality as a transcendental assumption and, consequently, the intellectual authority of social theory. Although Norris's claims once depended heavily upon the work of Putnam, he now chides Putnam for what he takes to be philosophical backsliding, and he continues to press the case for traditional scientific realism and apply it to everything from literary criticism to social science.

In the philosophy of social science, one prominent and influential version of critical realism, which has developed into what has come close to an institutional academic movement, is closely associated with the work of Roy Bhaskar.Footnote ²⁶ Stressing the need to take account of both the epistemological and ontological dimensions of social inquiry, Bhaskar has insisted on the independence and reality of the objects of science and their knowability, but he has also emphasised the contingent and socially situated nature of knowledge. This project was linked with and informed by a critical and normative goal. Bhaskar has drawn upon a wide range of meta-theoretical claims ranging from the philosophy of science to hermeneutics in order to formulate a programme of critical inquiry devoted to a general project of ‘human emancipation’. Although Bhaskar acknowledges the contribution of arguments such as that of Kuhn, which stress the inseparability of scientific concepts and the ‘world’, he criticises what he claims are its subjectivist and idealist implications and the inability, or more accurately refusal, of this approach to advance a general explanation for transformations in science. The somewhat unstated or understated assumption in this case is that philosophy has the capacity to achieve such an explanation and to provide ontological and epistemological foundations for science. Bhaskar defends a form of ‘metaphysical realism’ as a transcendental deduction that would sustain particular claims to knowledge and provide a basis for critical social inquiry. While for individuals such as Bhaskar and Norris, realism serves as a distinctly left-leaning political ideology, the same realist arguments are also typically invoked by many on the right. Realism is an equal opportunity metaphysic, but what joins somewhat strange ideological bedfellows is the concern with establishing the cognitive, and practical, authority of meta-practical judgment in fields such as philosophy and social science.

Liaisons dangereuses

The application of scientific realism to the study of International Relations is, in part, devoted to a critique of positivist inspired instrumentalist accounts of theory which, as in most of political science, had become prevalent in the field and its various research strategies, but these realist arguments are also, somewhat ambivalently, concerned with what is often generically labelled ‘constructivism’. The latter term, however, whether embraced or pejoratively deployed, is, like ‘realism’ and ‘relativism’, less the name of a definite philosophical position than a categorical designation. It encompasses a variety of claims to the effect that reality is less a given datum than a function of interpretive frames deployed by both social actors and social scientists, but, as used in a number of fields, including International Relations, it is a very porous concept. Constructivism emerged in the field of International Relations, in various forms,Footnote ²⁷ as a counter to political realist and neo-realist theories regarding power and interest in the interactions among states (which should not be confused with realism as a philosophy of science).

The propagation of philosophical realism in the field of International Relations manifests some persistent problems in the relationship between political inquiry and the philosophy of science. The philosophy of science, as already noted, has not been conceived by most philosophers as a foundation of scientific practice, and much of post-positivist philosophy, such as the work of Kuhn and Feyerabend, explicitly rejected such an assumption. In addition, the field is a highly contested one from which it is not defensible, without confronting indigenous arguments, to extract bits and pieces which serve a particular prior agenda. And there is the fundamental logical problem of how philosophy relates to its subject matter. For example, few would suggest that the philosophy of religion is, or could be, a guide to religious practice or that the philosophy of mind is the basis of intelligent thought. What is maybe most striking about the use of scientific realism in International Relations is that it reflects the fact that there is typically an intellectual lag between what is happening in philosophy and the echo that reverberates in the discourse of social science. And there is a further lag between political theory and the study of International Relations. At the very point, for example, at which realism might be construed as reaching its peak in social theory, its popularity is arguably undergoing a decline in philosophy, and while the popularity of realism in political theory is probably on the wane, it is widely embraced in the field of International Relations. Many of the references, however, are often to a somewhat dated literature, such as in the case of reliance on the early work of Putnam. The fact that Putnam has radically altered his view of realism does not necessarily entail that his past conception was incorrect, but it does demand confronting the validity of his current account. The recent interest in the philosophy of science on the part of scholars in International Relations has also somewhat unreflectively reprised debates that, during the 1960s and 1970s, had characterised arguments about behaviouralism in political science. Contemporary advocates of realism often seem dimly aware of the literature in which empiricism, deductivism and instrumentalism were hotly contested issues.Footnote ²⁸ There are continuing difficulties revolving around the extent to which social scientists adequately interpret the content and purpose of the literature in the philosophy of science and sufficiently discriminate among various positions. Most advocacy of realism is not conducted on the basis of immanent philosophical argument but rather consists primarily of invocations of the work various philosophers with whom most readers in the field are unfamiliar as well as unequipped to access and assess. Finally, there is the persistent underlying issue of the application of the philosophy of natural science to social scientific inquiry.

The arguments for and against the application of scientific realism to the study of International Relations have, at times, been edifyingly and expansively discussed,Footnote ²⁹ but this has largely been in terms of the relative merits of different meta-theories. The general issue of the relationship between philosophy and social science has not fully emerged, and it has not been directly confronted. It is not possible, and it is not the intention of this article, to do full justice, either descriptively or critically, to the arguments of the individuals discussed below, and no attempt is made either to assess the substantive contributions the authors have made to the study of International Relations or to discuss how their arguments have changed or evolved. The concern is instead to extract an image of the basic kind of project that is represented in this material and to point to some specific common problems that characterise this literature.

Although there is reason to suggest that the current conversation began with the attempt of Steve Smith and Martin Hollis to reconcile ‘explanation’ and ‘understanding’ in the social scientific investigation of International Relations,Footnote ³⁰ there would seem to be a general consensus, and considerable evidence, that the realist vogue in International Relations sprung largely from the work of Alexander Wendt, who many also associate with constructivism. Some of the basic problems with this vogue are evident in Wendt's early work.Footnote ³¹ Wendt's principal purpose was to attack the hold of positivism as a meta-theory in the study of International Relations, and particularly the instrumentalist account of theory which dominated the social sciences. He believed, however, that this required advancing an alternative meta-theory – or combination of meta-theories. His argument was less a defence of some distinct version of realism in philosophy than a general composite or collage of philosophical claims designed to serve various purposes in redirecting the study of International Relations. Realism, constructivism and positivism were all explicitly partially incorporated, ending in a hybrid that he dubbed ‘modernist constructivism’. Wendt was quite correct in maintaining that social scientists are today less secure than natural scientists and ‘have turned to philosophers for methodological guidance’, and this was precisely the path that Wendt followed. One might indeed argue that some philosophical renditions of science are better than others and that we cannot adequately combat the influence of positivism in social science without confronting it on its own philosophical grounds, but it is quite another thing to assume that the way to better science is through philosophy or, particularly, through an integration and application of diverse epistemologies.

Wendt claimed that realism entails putting ‘ontology before epistemology’, and although this claim, which has become something of a mantra in the literature, is defensible in the sense that the postulation of what is real must be prior to the question of how we gain knowledge of it, realism is primarily an epistemology. What might be termed scientific ontologies, that is, those embodied in scientific theories, are claims about the constitution of reality – not simply the philosophical claim that there is a reality independent of theories. Realism, like positivism, assumes that there is a mind/world dichotomy or that theories are one thing and that reality is another. The fundamental issue posed is the epistemological one of how theory and reality make contact. In other words, it treats theories as if they were high-level empirical hypotheses. In philosophical debates, this claim, even among realists, is highly contested. While such a perspective might make sense within the practice of science, it does not make much sense as a theory about science or a theory of theories. Wendt's statements that ‘theory reflects reality’ and that ‘meaning is also regulated by a mind-independent, non-linguistic world’ are quintessential epistemological claims, and these statements require a great deal unpacking in order to identify exactly what is being claimed. Similarly, despite Wendt's emphasis on the distinction between what is ‘unobservable’ and ‘directly observable’, these terms not only have different meanings in different philosophical accounts but have no general scientific application apart from specific theoretical contexts.Footnote ³²

By the point that Wendt's book was published, much of the defence of realism in the philosophy of science had given way to views that were often nearly indistinguishable from anti-realism. Most prominent in this respect, as already noted, is the case of Putnam on whom Wendt relied for much of what he had to say about matters such as ‘natural kinds’, which putatively secured the idea of a reality independent of theory. The incarnation of realism on which Wendt tended to depend was advanced by Putnam in the 1970s. Putnam began his philosophical career as a materialist and hard-core metaphysical realist and representationalist, but, in the course of a generation, he gravitated to what he termed ‘internal realism’, which rejected any ‘god's eye’ view of reality and suggested that ‘truth’ meant little more than idealised rational acceptability in the context of a particular theory and which closely resembled anti-realism as propounded by individuals such as Dummett. Finally, by the 1990s, Putnam had embraced what he today refers to as ‘natural’, ‘direct’, and ‘pragmatic’ realism. This position, which he largely identifies with both American pragmatism and Wittgenstein, entails a rejection of the assumption that it is philosophically meaningful to speak about an ‘interface’ between science and the ‘world’. He follows Wittgenstein's later philosophy in claiming that the relationship between thought and reality is to be found in the grammar of our language, that is, that representing nature takes place through the application of language in particular contexts such as science and everyday life.Philosophy can provide no general answer to how this is, or should be, achieved. He insists that this position is not the same as that of Rorty, who has recommended jettisoning traditional metaphysics and epistemology and adopting pragmatic criteria of truth which would involve what, in varying circumstances, people are willing strongly to defend. The basic differences between Rorty and Putnam, however, involve minimal family quarrels. Putnam essentially joins Wittgenstein, Davidson, Kuhn, and others in rejecting the assumptions and central problematic of representational philosophy and in concluding that ‘there is a way to justify our sense that knowledge claims are responsible to reality without recoiling into metaphysical fantasy’.Footnote ³³ So to be an authentic disciple of philosophical realism, one must keep moving quickly on the philosophical treadmill.

More specifically, with respect to the study of International Relations, what, for example, does it mean to say, as Wendt did, that ‘the state and state systems are real structures whose nature can be approximated through science’? The term ‘state’ is a word that, in the past two centuries, has been attached to quite different concepts. If Wendt was talking about the concept to which this word, as a term of art in political science, commonly refers, it was to a category of institutions and actions. It designates a stipulated class of things bound together by family resemblances and certain historical connections. So surely it makes sense to speak of this class as a class of real things as opposed, for example, to a merely taxonomic concept or to the manner in which a theorist such as David Easton might posit a political system as an analytical construct. Wendt's claim that states have a ‘nature’ which is ‘unobservable’ is less than clear unless it means that classes or species are, as such, not observable in the manner that members of the species are observable or unless all that it means is that, like board games, we can ascribe to them certain commonalities which might not immediately meet the eye. During the nineteenth century, however, it made sense, given the dominant approach to the study of politics, to claim that the state had a ‘nature’ which was unobservable and that all particular states were instances of an essential nature. The concept to which the term ‘state’ referred was that of an invisible spiritual community or public, which by the early twentieth century was deemed as unreal as phlogiston. Accordingly, in the nineteenth century, it was meaningful to speak of the ‘national interest’ and the ‘public interest’ as substantive attributes of states, while today there seems to be considerably less adherence to such a concept. Wendt claimed that in political science today the word ‘state’ refers typically to an instrumental construct, but it seems more accurate to suggest that states are viewed primarily as a species of somewhat loosely conceived political phenomena. The actual theoretical or ontological issue involves what kinds of things states are instances or manifestations of, that is, what are social facts such as institutions, actions, and conventional objects in general. This is the theoretical issue, which does not have a meta-theoretical solution.

The basic motive behind Wendt's project was a significant one: to turn social scientific theory, and the theory of International Relations, in the direction of fundamental claims about the nature of social reality and to move away from the typical empiricist construal of theories as conceptual frameworks for organising independent observable data. The problem, however, was with attempting to enlist philosophical realism as the vehicle for accomplishing this task as well as with attempting to conceive of agency and structure as ontological entities. The use of concepts of agency and structure involves a crucial and controversial but often inadequately explicated claim in the social scientific realists’ repertoire. These concepts bear no demonstrable resemblance to the unobservable objects and reality claims that are typical of natural science. They are basically categories or taxonomic terms which are in many ways the ghosts of old philosophical theories such as voluntarism and determinism, the remnants of post-Marxism social theory, and the residue of nineteenth century arguments about individualism and holism. The even more recent genealogy of these discussions involves the mid-twentieth century arguments of individuals such as Anthony Giddens. In many respects, Giddens’ work is a paradigm case, as well as source, of many of the problems and contradictions manifest in the attempt of scholars in International Relations to adopt realism. Giddens was among the first to argue for ontology over epistemology and to attempt to posit structure and agency as ontological entities, and he was a leader in the movement to solve the problems of conflicting epistemologies, such as naturalism and interpretivism or action and causality, by analytically synthesising them and deploying the product as both a framework and method of inquiry. Little is gained, however, by resurrecting this now exhausted discussion and claiming that agency and structure can be construed as theoretical entities comparable to those of natural science.

Wendt attempted to define his position by contrasting it with a characterisation of certain post-modernists who, he claimed, believe that ‘cats and dogs’ do not exist except as discursive inventions. This was far too simplistic for a description of any seriously developed philosophical argument, but it is not philosophical realism that makes it possible to conceive of states as real entities anymore than it is realism that makes it possible for natural scientists to claim that molecules exist. This is a function of scientific theory, not philosophy. Although it is tempting, and common, to say, as Wendt did, that natural objects are ‘mind-independent’, while ‘social kinds’ are the product of ideas and beliefs and therefore ‘mind-dependent’, this kind of distinction can be ambiguous if it is meant to imply more than the fact that, at any time, we are likely to draw a line between phenomena we designate as natural and those we take to be conventional. That line, however, has hardly been constant. It was not, for example, until the early part of the twentieth century that scientists accepted atoms as natural. The distinction is misleading if it is interpreted as meaning that natural objects are in some way given in a preconceptual manner that can be specified apart from the theories of science or that of whatever discourse we take to be authoritative. It would be trivialising to say that nature is only a product of the scientist's mind, and no one in contemporary philosophy has actually made such an idealist claim. Wendt took pains to argue that despite their ‘mind-dependence’, there are ways of approaching social kinds (in terms of causal analysis, focusing on structures, etc.) that make it possible to treat them functionally as if they were natural kinds, but this claim was, and is, also typical of positivism in social science.

For those unfamiliar with the contemporary conversations in the philosophy of science, Wendt's work did not offer clear guidance. The average reader in International Relations would have had little in the way of resources for evaluating the claims and citations regarding philosophical realism. Just as Wendt built his version of realism by mixing various meta-theories, he constructed an opponent composed of various elements of post-modernism, constructivism, and anti-realism. When he did mention a specific author such as Kuhn, he sometimes significantly misstated the claim. For example, despite Wendt's statement, Kuhn never said that ‘paradigms create “different worlds”’ but only that the theoretical components of such paradigms were logically incommensurable. What Wendt referred to as the ‘ultimate argument for realism’, that is, the capacity of science to ‘manipulate the world’, was, as already indicated, not only highly controversial in philosophy but so deeply embedded in certain forms of realism that it can hardly be called upon as a justification of the position. In the case of International Relations, it was unclear how this test could be applied to its theories, and to say that there is ‘some disarray in the realist camp about how to deal with this problem’ was a distinct understatement.Footnote ³⁴

Wendt's book was followed by Heikki Patomäki's contribution to the Bhaskar inspired ‘studies in critical realism’. One of the problems with this work was, once again, that while it subscribed to the title of ‘philosophical realism’, no distinct philosophical argument was advanced. Patomäki's image of realism was composed from the prior ecumenical epistemological compilations of Habermas, Giddens, Paul Ricoeur, and Bhaskar – with the addition of elements of Wittgenstein, Winch, Kuhn, Jacques Derrida, Michel Foucault – and many of the major philosophers of Western civilisation. Apart from Giddens and Bhaskar, most received short shrift, because the premise was that everything had led up to Bhaskar's ‘seminal’ work and his ‘theory of emancipation’ which transcended ‘both positivism and post-positivism’. Unlike Wendt, however, Patomäki wished to add a more distinct critical dimension and stance to the theory of International Relations.

Patomäki provided a lucid overview of the evolution of critical realism in the philosophy of science, and from this literature, he claimed he had extracted three basic ‘transcendental arguments’: ‘ontological realism’ or the view that the world exists independent of our knowledge of it; ‘epistemological relativism’ or the assumption that beliefs and knowledge claims are ‘socially produced’; and ‘judgemental rationalism’ or the claim that in spite of ‘interpretive pluralism’, we can make models of the world and determine their truth. After accessing such truth, we would, he claimed, be in a position to evaluate critically the ‘lay meanings’ which ‘create social practices and relations’. Patomäki identified his basic intellectual task as reconciling and synthesising the work of Giddens and Bhaskar, both of whom, he claimed, went beyond individualistic social theory by adding hermeneutics and the critical perspective of the Frankfurt school. The arguments of Giddens and Bhaskar were already highly syncretic, but, Patomäki claimed, while Giddens focused more on interpretation and agency, Bhaskar emphasised causal structures and relations. This, he argued, would yield a ‘social ontology’ based on ‘a conceptual analysis’ of action, structure, power, and system. And by taking account of the ‘double hermeneutic’ required by social inquiry and adding ‘iconic modeling’, it would be possible to achieve ‘emancipatory explanation’. All of this, however, required ‘a plurality of initial hypotheses drawn from different theories’; ‘hermeneutic mediation between theoretical concepts’; and the simultaneous embrace of a ‘reductive approach’.Footnote ³⁵ Patomäki's image was, indeed, the epitome of meta-theoretical fantasy, and it manifested many of the same problems as the literature on which he predicated his argument.

There is certainly nothing, in principle, wrong with eclecticism, pluralism, or interdisciplinary borrowing as a general value. The problem emerges when there is no attention to basic philosophical incompatibilities among the sources on which one draws as well as with the assumption that one can simply, as if making a quilt, construct a position out of the elements of diverse, and often contending, philosophies. If one attempts to do so, the burden is on the person who is assuming or arguing for compatibility. In the case of these authors, the argument for joining things such as realism and hermeneutics is not only tenuous, but even the assumption that there is compatibility among various versions of philosophical realism requires much more support than what is typically offered in this literature.

Although Colin Wight wished to make a stricter commitment to realism than Wendt, he presented his approach as ‘broadly consistent’ with that of Wendt, and it was in many respects closely related to the position of Patomäki. It was certainly consistent with Wendt in that, despite all the criticisms of positions such as positivism and idealism, there was no direct examination of, or confrontation, with the argument of any specific individual. To a large extent, Wight focused on an indictment of positivism and its conception of theory, but positivism appeared more as a general image or mood and its manifestations in social science, and this image was largely constructed from references to secondary literature rather than from the specific claim of any particular positivist philosopher or group of philosophers. ‘Scientific realism’, although sometimes identified with specific arguments in philosophy, was expanded far beyond anything most philosophers would accept, as it became entangled with a particular image of ‘critical realism’. What was presented was once more a variegated meta-theory derived largely from the work of Bhaskar and the invocation of the claims of various, and often not obviously compatible, philosophers. As in the case of Wendt, the assumption was that the practices of science rest on and can be described in terms of philosophical metaphysics which also can be the basis of scientific reformation. While this would be a very dubious claim with respect to natural science, there is some basis for arguing that much of social science, from its origins, has emanated more from philosophical claims than anything resembling theory in natural science. But if this was the root of the problem, it was not solved by substituting one meta-theory for another.

Like Patomäki, Wight embraced transcendental commitments to: ‘a reality independent of […] mind[s]’; the assumption that ‘all beliefs are socially produced’; and a faith that it is ‘possible, in principle, to choose between competing theories’.Footnote ³⁶ But once again, on their face, these claims were not very contentious and could be equally attributed to positivism as well as scientific realism. Taken individually and in the abstract, it would be hard to imagine anyone disagreeing. Even more than in the case of Wendt, what was presented was less a distinct argument confronting another distinct argument than a series of categorical assertions about realism and its opponents which were supported by often unexplicated references to a wide range individual authors (Bhaskar, Mario Bunge, Ian Hacking, Peter Berger, Richard Boyd, Martin Hollis, Kuhn, Hume, Paul Feyerabend, Imre Lakatos, Michel Foucault, Pierre Bourdieu, etc.). Redescribing Bhaskar's meta-theoretical mélange as an authority in support of Wight's particular account of realism deflected a confrontation with the issues involved. Anyone unfamiliar with the work of some of the individuals cited by Wight would have had little grasp of what they actually said or the extent to which they agreed and disagreed with one another.

As in the case of Wendt, it was very difficult to discern exactly what was meant by Wight's claims about what is and what is not ‘mind-dependent’ and ‘socially produced’. The bottom line in Wight's argument about realism was largely a variation of the claim, characteristic of a wide range of theorists such as Alasdair MacIntyre and Steven Lukes, that although social phenomena are the manifestation of human agents and their ideas, which must be taken into account, there are deeper underlying structural realities which can be studied much in the way that we study natural objects and which can explain why people hold the ideas that they do and why they behave accordingly – and often in a both substantively and logically irrational manner. At this level of generality, it would be difficult to respond concretely, and the particulars buried in this position, regarding, for example, what is observable and not observable, were not elaborated. As Wight noted, he ‘had not been reluctant to draw upon ideas from quite different sources’ even though some might see it as ‘an unacceptable electicism’. The problem with such eclecticism, however, was, again, not simply that of attempting to meld often quite different philosophical positions but the tendency simply to cite and subscribe to these positions as authorities, without a full elaboration and critical analysis of the arguments.

One of Wight's central concerns was with establishing the centrality of agency and structure, but this was somewhat regressive in that it manifested many of the problems already inherent in the work of Giddens and already prominent in the work of Wendt and Patomäki. Giddens, for example, despite his rejection of positivism, actually assumed an instrumentalist conception of theory which was the very antithesis of realism in the philosophy of science. He claimed that ‘conceptual schemes that order and inform processes of inquiry into social life are in large part what “theory” is and what it is for’, and he included in this category functionalism and other such typical approaches in social science which he claimed were tacit ontologies. His ‘theory of structuration’ was explicitly presented as an alternate framework for the ‘understanding of human agency and of social institutions’ but one that would also serve a critical function.Footnote ³⁷ Thus it had all the characteristics of the very instrumentalist conception of theory that realism in the philosophy of science was focused on criticising. As in the case of Wendt, many of the criticisms that Wight made regarding contemporary practices and approaches in the study of International Relations were relevant and persuasive, but these criticisms were not necessarily entailments of his meta-theoretical prolegomenon. For someone so committed to rejecting instrumentalism and conventionalism as philosophical accounts of theory, Wight's ‘manifesto’ for realism still clung to the positivist idea of observations as ‘theory-laden’, and, like Wendt, he still cited Putnam's now long renounced claim that without the assumptions of scientific realism, science would seem to be a ‘miracle’.Footnote ³⁸

Beyond realism and anti-realism

The recent fascination with realism in International Relations has not been without its critics. Fred Chernoff, for example, has strongly advocated a form of anti-naturalism or conventionalism as a meta-theory, and with greater sensitivity to the actual content of the philosophical literature than many of the advocates of realism. Although Chernoff, like the realists, joined in the critique of positivism, a commitment to methodological pluralism, and the embrace of causal explanation, he claimed that a meta-theory based on a ‘version of conventionalism’, inspired in large measure by the work of the philosopher Pierre Duhem, offers the best hope and provides the basis of a predictive capacity which would be of practical aid to policymakers. Conventionalism, Chernoff argued, particularly conceived in terms of ‘quasi-Duhemianism’, does not deny the ‘reality of the world beyond sense perception’, but it assumes that ‘physical theory is inescapably conventional’ and, unlike realism, does not insist that all ‘scientific theories refer to real entities’.Footnote ³⁹ After describing and discussing various positions in the philosophy of science, examining challenges to a ‘scientific approach’ such as those emanating from ‘critical theory, poststructuralism, and interpretive constructivism’, and concluding that ‘conventionalism’ constituted ‘a worthwhile alternative metatheory’, Chernoff nevertheless maintained that meta-theory is the foundation of sound scientific practice. He stated, for example, that ‘the answer to the question, how do we determine the best theory does not come from IR or social science but from philosophy’, and he continued to assume that precepts in the philosophy of science reflect scientific practice and ‘have allowed scientists to develop better theories over time’ and ‘have led the natural sciences […] to such great successes’.Footnote ⁴⁰ There is, however, little evidence, or argument to the effect, that the progress of natural science has been a result of subscribing to the precepts of the philosophy of science.

One of the basic problems in all this literature has been a neglect of the issue of what kind of enterprise has been represented in the history of the philosophy of science and in its contemporary form. More specifically, there has been a lack of attention to the relationship of this field to scientific practices and particularly to social science. Little is gained in social science by importing and becoming mortgaged to philosophical debates, such as that between realism and anti-realism. What is important is less making a choice between realism and conventionalism as theories of theory than recognising that the philosophy of science is not the key to successful science. Still, it may be asked, what, if anything, can philosophy do forsocial science? In answering this question, it is important to be clear about what contemporary realism and much of anti-realism, in philosophy are actually all about. They are in part a reaction against the increasingly dominant anti-representationalist critique of metaphysics and epistemology. Metaphysics and epistemology, in this sense, are not generic images of the ways things are and the nature of knowledge, which could be construed as assumptions in the background of all forms of science, but rather a specific kind of philosophical endeavour which is devoted to statements about reality and knowledge which transcend, but illuminate, the substantive claims of any practice of inquiry. Anti-representationalism not only calls into question this form of philosophy but implies a much more modest role which would make one cautious about seeking from philosophy the answer to the proper practice of science.

There are different strands of anti-representationalism, but they come together in the rejection of the traditional metaphysical and epistemological project. It is here that International Relations and other forms of social inquiry might fruitfully turn in search of philosophical enlightenment – an enlightenment that would encourage the autonomy of science rather than an attempt to base social science on, often transient, philosophical reconstructions. As noted earlier, one strand of anti-representationalism is apparent in the work of Wittgenstein who was arguably the most influential philosopher of the twentieth century, and another strand is manifest in the work of Davidson who, in the last half century, was one of the most important figures among professional analytical philosophers. One respect in which Wittgenstein and Davidson differed, and in terms of which some important philosophical disputes emerge, is the manner in which they treated the relationship between language and the ‘world’, but what their work shares is an insistence on the impossibility of separating the two. What forms of anti-representationalism have in common, and what Rorty appropriated from both Wittgenstein and Davidson as well as from Kuhn, is a critique of traditional philosophy's attempt to answer the question of how the language of science can ‘mirror nature’. As Wittgenstein, and William James, noted, realism and idealism are primarily different ‘battle cries’ uttered in defence of scientific knowledge, but the philosophical postulation of a world that exists outside that articulated by natural science is, as Rorty and Goodman both avowed, ‘a world well-lost’. It is not that science either mirrors or constructs the world – the ontological autonomy of both the world and language are assumed – but rather that it is only in the application of language in human practices that the ‘world’ appears. To recover that world philosophically is like seeking God outside the formulations of religion. The history of the ‘world’ and the history of ‘god’ are respectively the history of science and religion. In the case of scientific practices, the space between language and the world is only opened up within the discourse of science where there are specific theoretical criteria regarding truth-values and such matters as what constitutes natural kinds, taxonomic distinctions, etc. The philosophies of realism and anti-realism, however, seek to state the conditions of knowledge in a manner that both transcends and underwrites specific practices of knowledge, but it has no more (or less) significance for the conduct of scientific inquiry than, for example, the philosophy of language has for speaking grammatically. The philosophy of science should be devoted, as advocated by Kuhn and others, to attempting to make sense of the practice of science, much like social science attempts to make sense of social practices, but philosophy is no more the basis of science than social science is the basis of society.

Just as anti-representationalism is a threat to the aspirations of much of traditional philosophy, it is doubly a threat to the critical and practical aspirations and pretensions of social inquiry, which are still manifest in arguments such as those of Bhaskar. It is a threat, first of all, because the social sciences, which arose from reform movements and moral philosophy, have nearly always based their claims to practical authority and the capacity to speak truth to power on their putative possession of epistemic privilege. Anti-representationalism called into question the basis of this claim to authority, but it also called into question the particular images of science and the rhetoric of inquiry which social scientists had appropriated from philosophy. As much as realism involves a rejection of the epistemology of positivism, the turn to realism in social science is another example of the kind of appeal to philosophy that led to the saturation of the discourse of social science with the tropes of positivism.

Rejecting the allure of representationalism in philosophy entails, first of all, recognition of the need of social science to confront substantive theoretical issues regarding the nature of social phenomena. What is somewhat ironic about the popularity of realism is that despite all the emphasis on ontology and on the priority of ontology vis-à-vis epistemology, little attention has been given to these theoretical or ontological questions. Second, the field of philosophy is not something of one piece. Much of contemporary philosophy, in areas such as the philosophies of action, language, and mind, is not merely meta-theoretical and involves fundamental issues regarding the nature of social reality. Third, the problem of the relationship between social science and its object of inquiry is not simply cognitive but practical. If the general philosophical issue of how mind makes contact with nature raises a bogus philosophical question in the case of natural science, it is, oddly enough, an authentic question in the case of the relationship between social science and its object of inquiry.

In the recent Perestroika controversy in political science, with all the posturing about the need, or lack of need, for methodological pluralism and the adoption of both qualitative and quantitative approaches, what was lost was recognition of the fact that all social science, whatever its methods, is interpretive, because it speaks about, and sometimes to, another conceptual domain. The very concept of interpretation implies the discursive preconstitution of the object of inquiry. The now frequent claim, sometimes associated with post-modernism, that interpretation, including social scientific discourse, ‘goes all the way down’ misses the fact that social phenomena are conceptually and behaviourally prefigured prior to the reconstructions of social science and thus have a reality status or ‘nature’ quite independent of the claims of social science. What is real as an object of social inquiry is in fact what is conventional. Social objects are not in the first instance conceptualised by the scientist but rather, whatever the methodology, conceptually reconstructed. Consequently, the idea of the object of inquiry existing apart from the ‘mind’ of the scientist actually makes sense in the case of social science. After natural science has told us about the constitution of the world, it would make little sense to say that we must check and see if the natural scientist (or the theologian, for that matter) has got it right, while, on the other hand, it would always be reasonable to ask, from a third-order perspective, if the reconstructions of social science adequately convey the phenomena that are their object of inquiry. Just as there may be what might be called a ‘lateral’ cognitive conflict between science and religion, there may be ‘vertical’ conflicts between social science and its subject matter. But no more than between science and religion is there the possibility of a neutral, apodictic, philosophical resolution.Footnote ⁴¹

What this means for social science is first and foremost the need to confront theoretical issues regarding the nature of social phenomena, which precedes categories such as agency and structure. There can be no theory of politics, and certainly no theory of International Relations, in the sense in which we typically tend to use the word ‘theory’ when referring to a class of claims in natural science. Politics is a historically and socially particular configuration of human conventions, International Relations is a species of politics, and agents and structures are social scientific classifications of elements of that species. It is precisely this confrontation between the preconstituted constructions of first-order practices such as politics and the reconstructions of second-order practices such as social science that underlies much of the interest in realism. It is in part a manifestation of the wish of social science to trump its object of inquiry, with respect to the reality issue, by its claims to causal knowledge and a superior ontology – just as some of the impetus behind realism in the philosophy of science reflects philosophy's wish to be a kind of master-science. The degree to which social science can justify a claim to accurately and authoritatively describe and pass judgment on its subject matter is a highly debatable issue, but it cannot be settled epistemologically. Realism in social science would entail recognising the reality of social phenomena – and sometimes, what we might all agree, is the all too real nature of such phenomena and the problematic of representing that reality.

From the beginnings of social science, the goal has been to bring to social scientific inquiry a form and method that could be applied to the subject matter in order to make it cognitively accessible, as well as to make normative judgments about it, and the dilemma has always been that of reconciling that form with the preconstituted forms of social life. The answers have come from across a spectrum ranging from an insistence on a single form to an insistence on a plurality of forms, and sometimes, as in the case of Max Weber, to a search for multiple but changing ideal typifications. The basic answer of behavioural political science was the application of diverse and hopefully complementary conceptual frameworks such as systems analysis and decision theory. This gave way, in the enthusiasm for rational choice analysis, to a return to the hope for a unitary framework, which in turn gave way to a balance between rational choice and institutionalism and then finally, by the twenty-first century, once again to the ethic of allowing a thousand approaches to bloom. But in all this, the assumption was that because there is no ‘blank-mind’ approach, and therefore no social science without the imposition of form, we must settle on some kind of form. This assumption is once again reflected in the turn to realism and its commitment to categories such as agency and structure, but it is this fundamental assumption that must be critically interrogated. A meaningful epistemology of social science and relevant methods and forms of inquiry can only emerge as the entailment of a substantive theory of social phenomena, but as much as the turn to realism involves a wish to push the slogan of ‘ontology first’, what it is really doing is once again searching for an a priori meta-theoretical form to guide inquiry.

Although this is not the place to elaborate in any detail on Wittgenstein's, or Winch's and Kuhn's, claims about the relationship between language and the world,Footnote ⁴² it is important to rescue them from academic folklore, so prevalent in adaptations of realism in social science, which often attributes to these individuals some philosophical theory such as idealism, relativism, and constructivism. These ascriptions are the product of a continuing attachment to the binary terms of representational philosophy. There are a number of places in Wittgenstein's work where he says that language is not directly accountable to reality, that essence is a function of grammar, and that the relationship between language and what it represents is ‘arbitrary’. But there are also a number of places where he also makes it clear that if the facts of nature were different, our concepts would surely be different. His principal point, however, was that our language is not a reflection of the natural world, and, as already stressed, the relationship between language and the world is only manifest in the practical application of language in various human practices. What he rejected were philosophical theories such as realism and idealism that sought a transcendental answer.

At the heart of philosophical realism lies what John Dewey called the ‘quest for certainty’ and what Wittgenstein similarly criticised as the search for the indubitable, but, as Wittgenstein famously stated, this urge of philosophers ‘satisfies a longing for the transcendent, because insofar as people think they can see the “limit of human understanding”, they believe of course that they can see beyond these’.Footnote ⁴³ As Cora Diamond has pointed out, philosophical realism, like traditional empiricism, adopts a position that is like ‘elementary realism’ or simply the common sense notion of the difference between subject and object, which informs most practices.Footnote ⁴⁴ What is unrealistic is the attempt to transform elementary realism into a meta-theoretical proposition. A truly ‘realistic spirit’ in philosophy would be one that recognised the futility of that fantasy, and a realistic spirit in political inquiry would be one that recognised that scientific practices are based on scientific theories rather than on diverse and contentious theories about science.