Marxism and philosophy: a critique of critical realism - Richard Gunn

What follows is an attempt to reopen an old question, that of the nature and conceptual status of the categories of Marxist thought. The two 'classic texts' of Marxism which raise this question are Georg Lukács's History and Class Consciousness and Karl Korsch's Marxism and Philosophy, both published in 1923 . In the Stalinist 1930s and 40s, with some notable exceptions such as the writings of the Frankfurt School, the question disappeared from view. It came into prominence again in the debates over methodology in the 1960s and 70s but since then has once more entered eclipse as part of the general 'decline of the left'. Hence my aim is to renew, at least by implication, certain of the 1970s debates. This seems a matter of political urgency since a left unclear about the fundamental concepts of its own thinking is a left all-too-ready to concede the points it should be defending and to define its agenda by whatever markers it finds planted on the enemy's terrain. Examples of this latter tendency are the infatuation with new technology in the latest redraft of the Communist Party's programme and the theme of active citizenship (cf. Plant 1988) in the Labour Party's policy review. To a large extent, the left's response to the right's ascendancy has been to prioritise directly practical-political issues over methodological-conceptual ones, but this is surely short-sighted not least because (as argued below) first-order 'empirical' points and second-order 'methodological' points must form a unity in the development of Marxist thought.

The form my article takes is that of an argument around some formulations advanced by Roy Bhaskar in a paper for the 1988 Socialist Conference at Chesterfield, published in Interlink 8 (Bhaskar et al. 1988). Bhaskar advocates a philosophy he terms 'Critical Realism' . The programme of Critical Realism, as outlined in his Interlink article, is one of elucidating the 'enduring structures and generative mechanisms underlying and producing observable phenomena and events'; in Marxist terms, his model is a particular understanding of the relation between 'appearances' (which may be misleading, mystifying and so forth) and social 'essence' or 'reality' which Marx propounds in Capital (cf. Geras 1972) . Critical Realism is critical because, rightly, it refuses to take appearances at their face value; it is realist because it believes that the 'structures' and 'mechanisms' which it regards as generating appearances exist not merely as theoretical constructs but objectively, and in practice, as well. The founding text of Critical Realism is Bhaskar's A Realist Theory of Science, published in 1975. There, the problematic is not so much one of Marxist or political issues but of disputes within the philosophy of science. One of my contentions will be that these philosophy-of-science roots entwine all too closely with the programme of Critical Realism in its present, political, guise.

This said, I shall be less concerned to debate Critical Realist doctrines in detail than to sketch in contrast to Critical Realism an alternative understanding of the conceptual status of Marxist thought. I have selected Critical Realist formulations as my point of departure because — unlike those of the aftermath of 'structuralist' Marxism (for a critique see Bonefeld, 1987b) and Rational Choice Marxism, two other currently influential methodological schools — they throw the question of Marxism and philosophy into relief. The question of Marxism's relation to philosophy is inescapable for an exploration of Marxism's conceptual status if only because 'philosophy' is the discipline, par excellence, which has reserved to itself the right of debating issues of a categorial and conceptual kind. Bhaskar opens his Interlink contribution by saying that 'the left needs to take philosophy seriously', and for the reason just given he is right to do so. It is on this broader question of Marxism and philosophy rather than on Critical Realism as such that my presentation will concentrate, although I shall offer comment on some specifically Critical Realist doctrines in due course.

The basis of my disagreement with Bhaskar can be stated very simply. When Bhaskar says that 'the left needs to take philosophy seriously' he seems to mean that the left for its own part needs a philosophy, his own philosophy of Critical Realism being offered as filling this alleged conceptual gap. My contention, as against Bhaskar, will be that the (Marxist) left has no need for a philosophy: there is no conceptual gap within Marxism which 'philosophy' might fill . In saying this, however, I am far from implying that Marxism amounts to a positivism or scientism uninterested in categorial questions; rather, I am concerned — as were many strands in the debates of the 1970s — to underscore the 'Hegelian' dimension within Marx . Marx saw Hegel as the paradigmatic 'philosopher' but, I would urge, he was never more Hegelian than when the critique of philosophy is present as a figure of his thought.

The question of whether Marxism needs a philosophy at all is logically prior to the question of whether it needs a specifically Critical Realist philosophy. And it is the question of Marxism-and-philosophy (Korsch's 1923 title) which highlights the crucial issues so far as Marxism's conceptual status is concerned. This explains why, in the present paper, a discussion of the tenets of Critical Realism per se takes second place. Likewise it explains why a (brief) account of the nature of 'philosophy' is the place where the argument of my paper should begin.

Philosophy

Which questions are 'philosophical' ones? Oversimplifying, we can say: philosophy does not ask 'Is X true?' but, rather, 'What is truth?'. More precisely, it asks after the validity of the categories (the set of terms or conceptual framework) in virtue of which X counts as true: the kind of truth we arrive at depends on the conceptual framework we employ. In other words philosophical questions are ones of a metatheoretical' (a second- or higher-order) as distinct from a theoretical' (a first-order or empirical) kind. Philosophy, as a discipline or discourse, separates metatheory from theory and reserves to itself the former as its own, specifically philosophical, domain.

Philosophy has good reason for projecting this separation. If first-order theory were to undertake the justification of its own categories then theory couched in some set of terms would have to validate these same terms; pulling itself up by its own bootstraps it would presuppose what it was supposed to show; and vicious circularity would be the result.1 The specification of a realm of 'metatheory' as distinct from 'theory' breaks out of this viciously circular trap. And hence philosophy legitimates itself. This said, however, the theory/metatheory separation on which philosophy turns has fateful consequences. If it is second-order theory that is to validate the categories of first-order (empirical) theory then, presumably, we need a third-order theory to evaluate the categories of second-order theory . . . and so on, without hope of halt. Vicious circularity is avoided, but at the cost of infinite regress: from the devil to the deep sea. If this is so then the programme of philosophy is flawed at source, and the flaw is precisely the notion of a discrete and separable region of metatheory (labelled 'philosophy') which at first seemed to promise such logically cogent results.

How might the dilemma 'vicious circularity or infinite regress?' be overcome? Perhaps it is tempting to think that it might be overcome by ignoring philosophical (or metatheoretical) questions, and indeed the Marx of The German Ideology (Marx 1975, vol. 5 p. 236) advises us to 'leave philosophy aside' and to concentrate on the actual or real world. But matters are not so simple, since any theorisation, however empirical, employs categories. Unless it undertakes to be answerable for them, relativism — the view that any conceptual perspective is as good as any other2 — is the automatic result. The questions of category-validation which philosophy construes as its own are thus unavoidable, and to this extent Bhaskar is right.

I suggest that the above-mentioned dilemma can be overcome only by developing a notion of theorising which is theoretical and metatheoretical (or first-order and second-order) at the same time. Such theorising must be theoretical qua metatheoretical and vice versa. Theorising of this kind would overcome vicious circularity since it is already metatheory. And it would overcome infinite regress since it does not separate theory from metatheory: it does not, in the manner of philosophy, treat the latter as a discrete and autonomous conceptual domain. Not only does theorising of the kind here indicated place its own first-order truth-claims at issue by posing to itself the question 'Which categories are valid?' ; additionally, and conversely, it places the validity of its categories at issue before its first-order truth-claims themselves. Neither level of its theorisation is reducible to the other, but both are reciprocally informing: as it were, they are allowed to interact. Such a mode of theorising addresses 'philosophical' questions but, with good reason (infinite regress having to be avoided) it does not adduce a philosophy to do so. It contains no gap which philosophy might fill. And so, to the extent that this notion of theorising can be clarified and supported, Bhaskar is wrong.

I further suggest that Marxist theorising is theorising of just this theoretical-metatheoretical kind. Marxist theorising answers to the just-mentioned requirements, despite the inadequacies of Marx's occasional comments (e.g. my above citation) on this score. It is this notion of theorising which gives substance to Marx's critique of philosophy in his polemics of the 1840s, and which informs the critique of political economy (a critique which, Engels and dialectical materialism notwithstanding in no way depends on philosophy) developed by Marx in the course of his later life. We need to pay attention to Marx's critique of philosophy. We need also to pay attention to the notion of a theory which is also metatheory on which this rejection turns. This notion of a theory-metatheory unity gives us a first insight into what I should like to call Marx's theoretical 'totalisation'. A totalisation is a dynamic unity-in-difference to which nothing can be added, and from which nothing can be subtracted, without destroying the totalisation as a whole. For the 'totality' is wholly present in each of its 'moments' or parts. To add philosophy to the totalisation of Marxist theorising may therefore be to undermine it; and such I suggest is the case. Further: it is the configurations of this totalisation which, in outline, my article aims to bring to light.

Philosophy: a very short history

If philosophy disrupts Marxism, some contextualisation of Marx's notion of theorising is in order. It was not always the case that philosophy construed itself as pure metatheory. Roughly speaking, it only did so from Kant onwards. Ancient Greek philosophy understood itself as asking both first-order and higher-order questions; it could do so because it saw itself as interrogating a world, or cosmos, which it viewed as already meaningful in itself. In the order of things, the categories needful to speak truthfully of this order were believed to be already inscribed. Plato's 'Theory of Forms' is the best-known instance of philosophising in this cosmological mode.

The demythologisation of modern 'enlightenment' did away with the ancient notion of a cosmos (an in-itself meaningful world order) once and for all. Philosophy responded to this demythologisation by abandoning cosmology, which entailed a theory-metatheory unity, and by taking refuge in a solely metatheoretical redoubt. Thenceforth questions of the validity and status of categories were construed as sheerly metatheoretical questions, with all the dangers of infinite regress signalled above.

To be sure the move to the notion of philosophy as metatheory was both hesitant and complex. Seventeenth-century 'metaphysics' still carried cosmological overtones. Only in the twentieth century, the tendency of which is to assimilate philosophy to 'methodology', does the equation between philosophy and metatheory become complete: Russell's theory of logical 'types' and the 'philosophy of science' which unfolds from it are the most signal instances of the transition concerned. It is Russell who enunciates the case for a theory/metatheory separation in the clearest terms (cf . Davie, 1986; Gunn, 1987c). The consequences of this separation are (a) the positivism of a first-order theory which disallows reflection on categories and (b) the tedium of a philosophising which, as purely metatheoretical, treats engagement with worldly issues as an infection of a non-philosophical kind. Anyone who has studied either the social sciences or philosophy knows what this positivism and this tedium mean.

It is Marx (and, I would suggest, Hegel before him) who attempts to renew the notion of a theory-metatheory unity without moving on to cosmological ground. This Hegelian-Marxist renewal has two steps. The first (especially clear in Hegel: cf. Hegel, 1874 para. 41; 1977 pp. 47-8) is to resist the notion of metatheory as a distinct conceptual realm which philosophy might appropriate as its own. The second step (especially clear in Marx although already prefigured in Hegel's Phenomenology of Spirit: cf. Hegel, 1977 pp. 490-1 ; Gunn, 1988a) is to link theorisation, not with a cosmos, but with practice. For Marx theory is an aspect, or moment or dimension, of practice. Conversely, for Marx, practice includes theory: for example capitalism as a mode of production includes the ideologies of 'freedom, equality, property and Bentham [or atomised self-interest]' (Marx, 1976 p . 284) as intrinsic to the reproduction of capitalist relations of production through time . And it is Marx who, in 1844 (cf. Marx, 1975, vol. 3 p. 332), says that uncritical positivism complements — instead of contradicting — uncritical idealism, i.e., the idealism of philosophy. Philosophy as sheerly second-order metatheory trails the positivism of sheerly first-order theorising in its wake, and vice versa. Marx condemns the dichotomous severance of theory from metatheory inherent in Russell's programme more than half a century before this programme was announced.

It follows that Marx also, by implication, condemns the notion of 'methodology'. (I shall return to this.) A Marxist philosophy and/or methodology — in fact any notion of a Marxist metatheory understood as separate from first-order theory — is a contradiction in terms. For example, to read Marx's Capital either as sheerly first-order and empirical (the reading attempted by bourgeois sociology) or as sheerly second-order and philosophical (the reading attempted in e.g. Althusser, 1970) is to miss its challenge. Capital is both first- and second-order. It is both because it is neither on its own; it is neither because it is both.

Practical reflexivity

I have proposed that it is by understanding theorisation as linked to — as forming a unity with practice that Marx can unify theory and metatheory, thereby dispensing with the need for a 'philosophy': but how exactly is this so? The answer to this question lies in what I shall term 'practical reflexivity'.

Theory is reflexive when it reflects on the question of the validity of, or justification for, its own categories. Theory is practically reflexive when in doing so it reflects upon and understands itself as inhering in a practical (a social) world. Marx — and after him Lukács, Horkheimer, Adorno and Gramsci, to name only a few — elaborates theory in this practically reflexive way. (For references and a more extended discussion see Gunn, 1987a.)

To see how practically reflexive theorising unifies theory and metatheory we should note (a) that practically reflexive theorising includes itself within its own object, viz. social practice, and (b) that it thinks about the validity of its own categories precisely while reflecting on its presence (its situatedness) within the society it attempts to understand. Its theorisation of its object, of its presence within its object and of the validity of its categories (as categories appropriate to the theorisation of precisely that object) are not three separate conceptual moves but a single totalisation. Each of the three dimensions of this totalisation already effects the other two. For practically reflexive theorising, thinking about its object already raises the question of its presence in that object; and thinking about itself already raises the question of the object — the totality of social practice — in which it inheres. Further, its reflection on society is already reflection on its own truth-criteria and vice versa; this has to be so if, in accordance with Marx's thesis of a theory-practice unity, theorisation is to understand itself as practically rooted without remainder or, in other words, as practical through and through . Not just the first-order but also the second-order (or metatheoretical) dimension of theorising inheres in practice; this is what Marx tells us when (Marx, 1975, vol. 5 p. 5) he urges that it is 'in human practice and the [reflexive] comprehension of this practice' that theoretical (read: so-called philosophical) 'mysteries' stand to be resolved.

Thus it is that practically reflexive theorising overcomes the theory/metatheory (the philosophy/non-philosophy) separation. It does so because it meets the condition of being first- and second-order at once: simultaneously and on one and the same conceptual movement, it advances on each of these two fronts. Put otherwise: the same body of theory plays now a 'theoretical' and now a 'metatheoretical' role. Each of these two dimensions of practically reflexive theory informs and interrogates the other. The first-order 'object' of practically reflexive theory has something to say about the categories which might validly cognize it, and the categories themselves have something to say about how the 'object' is to be constituted and defined. Neither dimension is reducible to the other and, within their totalisation, neither has exclusive rights.3 Vicious circularity would result if all the rights were given to first-order discourse about theory's object; infinite regress would result if all the rights were given to second-order discourse about theory's categories; but neither is here the outcome because, according to the idea of practical reflexivity, only a single discourse — or, better, a single totalisation or interaction as between theoretical levels — is entailed .

If theory were merely reflexive and not practically reflexive these totalising conclusions would not follow. For it is only for practically reflexive theorising that the questions of the character of its subject- matter and of the validity of its categories appear within the same level of theorising (this level being theoretical or metatheoretical as the case may be). Reflexive but non-practically-reflexive theorising finds itself pinned on the horns of the dilemma: vicious circularity or infinite regress? So: not merely can practically reflexive theorising avoid the dilemma; additionally, only practically reflexive theorising can do so. And here, I suggest, an important conceptual strength of Marxist theorising lies.

The importance of this point appears to me to be such as to deserve emphasis. All that I have said follows from the idea of theory's understanding itself as, without remainder, a moment in and of the practice (the society) whose character and contradictions it aims to make clear. For such theory, to drive deeply into first-order elucidation is to generate metatheoretical insights, and to drive deeply into metatheory is to carry forward first-order elucidation itself. To raise metatheoretical questions is at the same time to raise social questions, not least because every social crisis is an epistemological crisis as well. Two consequences follow. One is that the question 'Should Marxists devote themselves to methodological or empirical research?' becomes meaningless: to be a Marxist is to do both (because sheerly neither) at the same time. The second consequence is that, within practically reflexive theorising, there is no space which a 'philosophy' — a metatheory disconnected from theory — might appropriate as its own . In other words, as averred earlier: the gap within leftist theorising which Bhaskar offers to fill with the philosophy of Critical Realism does not exist.

Critique of Critical Realism (1)

Before pursuing the account of the status of Marxist categories on to which the notion of practical reflexivity opens, I shall ask: to what extent does the above represent a criticism of Bhaskar's views? Something resembling practical reflexivity makes its appearance in Bhaskar's Interlink article, since we there read that 'social theory and social reality are causally interdependent' in the sense that 'social theory is practically conditioned by, and potentially has practical consequences in, society'. Does this imply practical reflexivity in the full sense of my preceding section?

The answer to this question has to be No. The 'causal' interrelatedness which Critical Realism envisages as between theory and practice is at best an external interrelatedness: it has to be, since a causal explanation is tautologous unless the term in the causal relation which explains and the term which is to be explained are reciprocally independent and distinct. This is so even if the terms concerned are held to be 'causally interdependent', i .e ., to explain one another. Some definitions are helpful here. An external relation (and all causal relations have to be external relations) is one which leaves unaffected the terms between which the relation obtains. An internal relation is one which affects and constitutes the related terms. Marx for his part sees theory and practice as being linked not just in an external — a causal — but in an internal way. Practice constitutes theory and vice versa. Only in virtue of this internal relatedness do they form a totalisation. Introducing causalism into this totalisation undermines it. Worked out to the end, it is only in the form of a determinist conception of political action that a causalist approach to Marxist discourse can unfold. Either causalism or internal-relatedness: there is no way to combine them. Space forbids, here, an exploration of determinism's consequences. But from Engels through Plekhanov and Lenin we can learn from history — if history teaches us anything — where a political practice premised on determinist theory leads.

Primarily, the claim just raised to the effect that theory and practice are internally related is a first-order one: it relates to the 'theory' or ideology and the 'practice' or social action which goes forward in social worlds. But, secondarily, it is also a second-order claim inasmuch as without remainder theory grows from practical roots. It is the second-order (or metatheoretical) implications of the point which are relevant just now. Once theory and practice are construed as externally (e.g. causally) related it becomes impossible to say, as was said earlier in the course of defining practical reflexivity, that theory which reflects on its own practical situatedness and which reflects on its categorial validity does both together and at the same time. The two dimensions fall apart. They do so because it is possible for theorisation to acknowledge that is has practical and social preconditions while also holding off from this social-situatedness the question of its categorial validation as a topic to be addressed in a purely and practice-independent metatheoretical realm. It was on the basis of this separation as between the causal explanation of a theory's occurrence and its categorial validation that bourgeois 'sociology of knowledge' was born. To keep the theoretical and metatheoretical dimensions of theorising together we need not an external but an internal relation of theory to practice. And since Marxism understands theory and practice as, thus, internally linked we can say: the totalisation it effects within theory is a totalisation as between theory and practice as well. It is a totalisation which opens on to practice and exposes itself to the challenges (ones of struggle and unpredictability) which practice presents (cf. Bonefeld, 1987a).

We can also say: Bhaskar advertises a philosophy, in the above sense of a pure and infinitely regressive metatheory, in the same conceptual movement as he understands the theory-practice relation in an external and (reciprocally) causal way. Philosophy closes itself off against practice — it forms an allegedly practice-independent 'realm of its own' (cf. Marx, 1975, vol. 5 p. 447) — and so stands at the opposite pole to practically reflexive theorising, which places itself at issue within the social practice it reports (Marx, 1975, vol. 5 p. 3 : Thesis II).

But is this criticism of Critical Realism fair? Is Critical Realism a 'philosophy' in a solely metatheoretical sense? Perhaps Bhaskar might endorse all of my anti-philosophical points and yet say that they do not apply to him. Perhaps he philosophises in what, given my above definitions, amounts to a non-philosophical way.

Certainly, from his Interlink contribution, it is difficult to gather what Bhaskar understands the conceptual status of Critical Realism to be. The most we learn is that it 'provides a set of perspectives on society (and nature) and on how to understand them'. To make sense of this, we should once more take historical stock.

In my 'very short history of philosophy' I reported methodology as being heir to the seventeenth-century metaphysics which first projected philosophy as a metatheory disconnected from all else. Metaphysics still carried in its wake cosmological traces. Russell's critique of metaphysics, which in the twentieth century transposed philosophy into the key of philosophy-of-science 'methodology', attempted to purge these traces once and for all. But the notion of methodology appears unable to effect the definitive purgation since, after all, for a methodology to be an empirically rich one it must in some fashion stand in a relation of correspondence or analogy — a mimetic relation, to use the terminology of Ancient Greek philosophy — with the world whose character it purports to make clear. A purely unworldly methodology could achieve nothing, except perhaps death, and we know that science has advertised for us all manner of worldly and practically successful goods. For this reason, the notion of philosophy as pure metatheory has involved an attack even on the notion of 'method', Karl Popper's onslaught against a 'logic of scientific discovery' being the first move in this post-methodological game. The second move is towards a 'post-empiricist' philosophy of science (cf. Kuhn, 1962) which condemns the residual elements of methodologism, and thereby of cosmology, in Popper himself . From the assuredness of metaphysics we move through the middle ground of methodology to the idea of a set of 'perspectives' (Bhaskar) or of 'paradigms' (Kuhn). The point of this short rendition of philosophy's twentieth century history is to suggest that it is precisely by attenuating itself that the notion of philosophy as a metatheory severed from theory holds itself in play. The informality of the idea of 'a set of perspectives' fails to equate with philosophical innocence. On the contrary, such informality can amount to philosophy-as-metatheory bending over backwards in order to maintain its domain.

In fact, as indicated earlier, it is in post-Russellian philosophy of science that Critical Realism's roots lie. And philosophy of science (whether or not one devoted to setting forth a scientific 'methodology') is, we can now say, philosophy's paradigmatic twentieth-century guise. Philosophy as philosophy-of-science reproduces the idea of philosophy as turning on a theory/metatheory divide.

If all this is so then my criticism of Critical Realism is indeed a fair one. It is a fair criticism because neither a first-order theory cut off from its categorial self-reflection nor a second-order theory which abnegates its own (practical) worldliness can open itself on to the challenges of unpredictability, happenstance and struggle which any social world contains. The severance of the two modes of theory means that each becomes conformist. The critique of conformism belongs with theory which is practically reflexive (cf. Horkheimer, n.d. p. 229), since only a theory which thematises its own presence in society can make a question out of the way in which society proffers, as ideologies, categories whose seeming 'obviousness' suggests that they have only to be understood to be endorsed. This kind of false or socially-constructed obviousness can only be challenged by theorising which reflects on its own social involvement, i.e., its own presence in the world of practice where ideologies are inscribed. Not even the most diluted and informal notion of philosophy as metatheory — for example the notion of philosophy as supplying 'a set of perspectives' — can achieve the same, critical and interrogative, result. Left theory has indeed to take the questions raised by philosophy seriously, in order to be answerable for its categories and to understand its own way forward, but at its peril (at the peril of abnegating its critical standpoint) it takes them seriously in a philosophical way.

Immanent critique

To say, as I have said, that practically reflexive theorising overcomes the theory/metatheory separation is one thing. To say how it does so is another. Notice that, in order to be consistent with the above line of argument, we have to be able to say how it does so without construing practical reflexivity as a 'methodology' for its part. My suggestion is that the key to this difficult question lies in the circumstance that the modus vivendi of practically reflexive theorising is that of immanent critique. Again, some definitions help to clear the way.

Critique calls into question the truth-claims and/or the categories of whatever view it interrogates: it refuses to succumb to obviousness which may be false, or to take claims at their face value. Out of its interrogation arises opposition, depending on the results which critical evaluations produce.

External critique effects its interrogation by measuring a view's claims and/or categories against some pre-given conceptual yardstick, i.e., against some truth-criterion presupposed as valid in advance. By contrast immanent critique operates by interrogating or challenging a view from within. It plays off a view against the view itself, for example by asking whether the conclusions drawn from the view's premises are consistent with one another and with the premises themselves. It does not impose its own truth-criteria on the view criticised, as does external critique; rather, it placed its own truth-criteria at issue within the critical engagement and develops itself (instead of merely reconfirming its own validity) in and through the process of critique it undertakes. The classic texts of immanent critique are Hegel's Phenomenology of Spirit and the critique of political economy unfolded in the Grundrisse and Capital of Marx .

In the present connection, the relevance of immanent critique is that it does not presuppose the validity of any metatheory (any categorial basis from which critical pronouncements are delivered) but, in the critical process, places at issue the categories in terms of which it initiates the critical play. Put otherwise: immanent critique proceeds in the manner of 'good' conversation (Gunn, 1988c).
Immanent critique converses with its critical targets, in contrast to external critique which holds no brief for answerability in any conversational (or 'dialogical') sense. The programme of immanent critique turns on the Hegelian notion of intersubjectivity or 'recognition' (Hegel, 1977, p. 112; Gunn, 1988a). External critique is for its part 'monological', and in the last instance throws intersubjectivity to the winds.

In other words: the notion of conversation helps us to understand something of what immanent critique involves. At the opposite extreme from 'good' conversation there is conversation which is 'disappointing' or boring . Disappointing conversation restricts itself either to sheerly first-order points (e .g . 'Did X really murder Y?') or to sheerly external points (e.g. 'What counts as murder?'); its mode is either empirical or philosophical but never both at the same time. On the other hand, good conversation allows first-order and metatheoretical points to interact, to inform one another and to unfold together: it is not two disconnected totalisations but one. As it were, the 'vertical' line of the distinction between its theoretical and metatheoretical dimensions and the 'horizontal' line of the distinction between conversational partners become a single line, without either of the conversational partners being allocated a sheerly theoretical or metatheoretical role and without the conversation itself becoming either sheerly metatheoretical or first-order as the case may be.

'Good' conversation is good rather than 'disappointing' — it does not merely chew over factual disputes or retreat into a play of disembodied concepts — because it, and it alone, allows conversational partners to challenge one another and to learn from one another in a fashion which brings all things about each partner into play. To discuss with someone whether or not they think it empirically true that it was X who murdered Y (and to discuss this alone) is to leave their conception of what constitutes 'murder' out of account; to discuss with someone what they think constitutes 'murder' (and to discuss this alone) rapidly becomes, in the worst sense, academic unless we can also ask: 'Do you think for example that Stalin murdered Bukharin when he brought him to trial in 1938?'. We recognize our conversational partner — to employ this Hegelian expression once again — only when both theoretical and metatheoretical considerations are open to being conversationally addressed. And, as the example of Bukharin's execution signals, such recognition presupposes that theoretical and metatheoretical considerations are not held separate. We have to be able to say things like 'You define murder in such-and- such a way? But then think of Bukharin!' and 'Yes of course Stalin was the cause of Bukharin's death; but does it count as murder to do what Stalin did?'. In the flow (the totalisation) of 'good' conversation first- and second-order points inform one another and interact, and this interaction is part and parcel of the interaction — the play of recognition — between those amongst whom conversation occurs. What I have called 'good' conversation is, in fact, the most familiar instance of a theorisation which refuses to compartmentalise itself in a theory-as-distinct-from-metatheory sense. To be sure, as a 'local' shift within such a conversation, now-metatheoretical and now-theoretical questions may occupy centre-stage. But these shifts are indeed local ones. What conditions them is the appeal from the one level to the other, i.e., the flow or totalisation of the recognitive conversation itself.

As with conversation, so with immanent critique. Such critique plays off against one another all of the dimensions of the view it interrogates, including the theoretical and metatheoretical dimensions; it does so by placing all of itself, including the adequacy of its own theory to its own metatheory and vice versa, at stake in the critical (the conversational) battle which it joins. There is more than a relation of analogy between 'good' conversation and immanent critique. The space of immanent critique just is the space of good conversation which also just is the space of the theory-metatheory totalisation. And so not three terms but one. The severance of theory from metatheory would in itself be sufficent to hedge the bets of immanent critique in such a way as to allow it to hold either its first-order truth-claims or its second-order categories and truth-criteria back from the play of critique itself. Thereby immanent critique would devolve back towards external critique. And so the notion of immanent critique has to be dialogical: one can play off the theoretical and metatheoretical dimensions of one's opponent's or partner's theorisation against one another only by throwing the relation of these two dimensions of one's own discourse into play as well. It is this dialogical aspect which transforms severance into totalisation. Conversely, 'good' conversation has to be immanently critical: it has to address the theory/metatheory articulation advertised in a conversation, not externally (i.e dogmatically and monologically, in the light of a pre-given metatheory) but from within.

Marxism has always become dogmatic when, as for example in Engels' Dialectics of Nature or Lenin's Materialism and Empirio-criticism, it has come forward in the guise of a philosophy. (I suggest that, all its differences from Engels-Lenin dialectical materialism notwithstanding, it is in this tradition of specifically philosophical Marxism that Critical Realism belongs.) In monological theorising, philosophy (construed as pure metatheory) and external critique find their common basis. And the positivism of a first-order theorising disconnected from its categorial self-reflection is their common heir. Engels' determinism and Lenin's Taylorism are the best-known instances of this heritage. Another such inheritance is 'structuralist' Marxism, which in the form of Jessop (see Jessop, 1988) completes my polemical circle by announcing itself to have Critical Realist philosophical roots. (The problem of 'Marxism and philosophy' was always a difficulty for structuralist Marxism: see note 3, above.) The first-order, or political, implications of Critical Realism remain to date unclear. Bhaskar (1988) talks of a 'transformational' and 'relational' approach to social relations, and appears to understand the politics of this in a New Social Movements sense. All this sounds welcoming . But, as is clear from the Communist Party's latest programmatic redraft (cf. Gunn, 1988b), a pluralist celebration of new social movements can all too easily make common cause with a technological determinism which might have enthused Engels himself. What guards against positivism, and its long historical flirtation with determinism, is practical reflexivity (understood as immanent critique). If this is so, then the philosophical form of Bhaskar's re-opening of Marxist methodological questions supplies grounds for political disquiet rather than delight.

On the score of dogmatism, finally, it can be noted that Marx himself theorises non-dogmatically because he theorises non-philosophically. He theorises at once theoretically and metatheoretically because he theorises in the mode of immanent critique. His work on the critique of political economy from 1844 onwards records the process of a long, painful and severe conversation with his opponents, some of whom — for example the 'classical' economists and the 'left' Ricardians — become in the course of the conversation his partial friends. He throws against them all the charges that practical reflexivity entitles, but yet relates to them dialogically because he thematises the practical-rootedness of his own discourse for its part. He places himself on their terrain (the terrain of the practice of capitalism) in the same movement as he draws them on to his. Thus everything is at issue and nothing guaranteed in advance. Solely for this reason, without dogmatism the urgencies of class struggle can be seen as traversing not just Marxist practice but (cf. Cleaver, 1979; Negri, 1984) Marxist theorising itself.

Immanent critique and practical reflexivity

It remains to show why the approach of immanent critique is a practically reflexive one and vice versa.

Practically reflexive theorising is immanently critical because it asks after the validity of its own categories in the course of understanding itself as practically situated. Inasmuch as it does these two things at once it has to place its own categories (and the truth-claims they entitle) at issue in the face of the theorising enunciated by its opponents, who share the same social and practical world. That is to say, it cannot confine itself to reporting the practical situatedness of its opponents — as state-sycophants or bourgeois apologists or whatever — in a purely 'third-person' fashion, as a bourgeois sociology of knowledge might do. It has to engage 'in the first person' with the views which, 'in the third person', it finds its own social world to contain: again, not two separate totalisations, but one, in which the first-person and the third-person moments interrogate and inform each other. Its debate within itself is also a debate with its opponents and vice versa. This means that practically reflexive critique has to be immanent critique (it has to be dialogically structured) since a critique which was merely external and third-person would omit the moment of 'in-the-course-of' of self-risk. Only immanent critique achieves first-person openness. Indeed the recognitive depth of 'good' conversation — conversation able to cross the theory/metatheory boundary because it is able to bring all things about each partner into play as conversational topics — and the practical depth of practical reflexivity are one and the same: to ask about one's partner's definition of 'murder', to continue the above example, is also to ask about the practical situatedness of this partner (as 'bourgeois' or 'Stalinist' or neither, for example) in virtue of which some definition of 'murder' seems to carry obvious force. Conversely, to raise the topic of one's partner's practical situatedness is at the same time to signal one's own practical situatedness as a possible topic and so to place at issue the validity of the categories within which one has raised the topic on one's own behalf.

The same point — practical reflexivity implies immanent criticism — can be arrived at by reflecting that it is an internal relation between theory and practice which is advocated by Marx. The notion of an external or causal theory-practice relation (as in base/superstructure versions of Marxism: e .g . Cohen, 1978) entails that one approaches the theory which is to be practically situated in a purely third-person fashion, i.e., without risk. The notion of internal relatedness, by contrast, takes out no such categorial insurance policies: this notion implies that it is in the course of thinking about practical situatedness that one thinks about validity. As already reported, the phrase 'in the course of carries with it the idea of openness in the first person to the challenges which one's social world (including the theoretical dimension of that world) presents.

Just as practical reflexivity implies immanent criticism so, conversely, immanent criticism implies theorising of a practically reflexive kind. This is clear from what has just been said: rigorous — recognitive — interaction holds open the possibility of raising dialogically the question of 'where one's opponent is coming from', i .e ., of who and what (socially speaking) they are. And since this question is raised dialogically it is a question whose force one appreciates as applied to oneself. Practical reflexivity and immanent critique form a single conceptual figure. Whenever they are separated, external critique and (its complement) a sociology of knowledge become the order of the day.

The category which allows us to think the practical reflexivity/immanent criticism interrrelation is that of theory's audience. (For a discussion of this category in relation to Hegel see Gunn, 1988a.)
The notion of an audience is a practical category since, quite independently of any given theory's pathos or eloquence, an audience for the theory may or may not exist. If practically reflexive theorising is immanently critical and thereby dialogical, it must raise the question of its own validity in the course of projecting the notion of an audience whose response to it (or recognition of it) might validate it. And if immanently critical theory is practically reflective theory it has to pose the question of its audience in a directly social and political way. Only in the first instance does the audience for immanent critique consist of those against whom the critique is targetted; the larger audience for such theorising consists of all those who, joining the conversation, are in a position to address the question of the validity of the critique concerned.

This leads us to the paradox of a critical theory whose own practical reflexivity tells it that the audience for its points and categories is non-existent. This was the paradox of 'Frankfurt School' critical theory in the 1930s and 40s (cf. Horkheimer n.d. pp. 220- 1). The further reflection of it is Ernst Bloch's notion of a not-yet- existing audience able to recognize itself in the 'militant optimism' of a revolutionary hope-principle (Bloch, 1986) and Sartre's notion of a 'virtual', as opposed to an empirically actual, public towards which radical literature might be addressed (Sartre, 1950). Lukács's distinction between 'imputed' and 'empirical' class consciousness (Lukács, 1971 p. 51) amounts to an attempt to dissolve the paradox by placing it in the context of an evolutionary and developmental historical scheme.

But the paradox is not to be dissolved so easily. In an estranged world, where is the audience whose freedom allows it to evaluate the truth of a theorisation premised upon, and aiming towards, estrangement's critique? Practically reflexive/immanently critical theory thus turns on paradox wherever its own reflexivity compels it to acknowledge alienation as the order of the practical day.

This paradox is not, however, a self-destructive or self-consuming one. It would be self-consuming if the question of an audience were addressed solely in first-order and third-person terms: does or does not the audience obtain? But, if first-order points are already in themselves metatheoretical points, then theorisation can claim for itself the role of defining its own (possible) audience as it proceeds. And, if third-person points are already in themselves first- person points, then theorisation can set out to conjure the audience to whom it makes its appeal. The condition of both of these moves is that the audience signalled is not one which is sheerly notional, ahistorical and ideal (as is, for example, the 'ideal speech situation' as described in at any rate Habermas's later works: e.g. Habermas, 1986 p. 90). The theory which invokes the possibility of a not-yet-existing audience must be able to establish this possibility as one rooted in the contradictions of an existing social world. To this extent first-order and third-person points are methodologically indispensable: utopias must be historically and practically concrete utopias, as Bloch said. Once more: a totalisation and not a disconnection. In an estranged world critical theory must be counterfactual theory (cf. Horkheimer and Habermas) but, pace the later Habermas, the counterfactual claims raised by such a theory must be capable of redemption and validation via the conversation of an audience which at least in principle is able to make its appearance in practical, i.e. social and political, terms. For, as in Kant, the notion of a counterfactuality severed from history and practice only renews the fateful theory/metatheory separation once more.

Put differently, in terms which are 'practical' rather than 'theoretical': estrangement, or alienation, is never a self-complete or seamless whole. Rather it is the movement of a contradiction. The 'real movement' of this contradiction is 'communism' (Marx, 1975, vol . 5 p . 49). 'Communism is neither a teleology of the capitalist system nor its catastrophe. It is . . . thus a concept that we can only formulate within the form of the transition' (Negri, 1984 p . 165).

A note on methodology

I said earlier that, on pain of self-contradiction, the modus vivendi of practically reflexive theorising must admit of being specified without its being specified as a methodology; for it is on the theory/metatheory severance that the idea of 'a methodology', no less than the idea of 'a philosophy', turns. I have specified the modus vivendi of practical reflexivity as that of immanent critique. Does immanent critique amount to a methodology for its part?

The answer to this question is 'Yes' if, and only if, immanent critique fails to understand itself in a dialogical way. For then it can only count as a method which is to be externally applied. Transposing itself into the mode of external critique, it undermines itself. Methodology and external critique legitimise one another; immanent critique is critique in non-methodological guise. Arguably for this reason some of the best Hegelian scholars (Kojeve, 1969, p. 176; Dove, 1970) urge that it is precisely rejection of the idea of methodology which underpins the rigour of Hegel's thought. As it were, for Hegel, the openness-to-all-comers of good or 'recognitive' conversation makes possible a more severe testing of truth-claims than does any 'method' deployed — as all methods must be — in a more-or-less a priori sense.

Michael Rosen, in a fascinating discussion of the internal logic of immanent critique (Rosen, 1982 Ch. 2), makes clear the difficulties which the notion of methodologically based critique involves. These difficulties turn on what Rosen calls the post festum paradox', viz., the paradox of being able to evaluate the results of critique only by depending on these same results' validity; this amounts to the vicious circularity of evaluating a method by its outcome while at the same time evaluating the outcome by the method. External critique may seem to avoid this vicious circularity, but only because (and to the extent that) it sheerly presupposes its own standards and criteria as beyond reproach. Immanent critique may seem to entail this vicious circularity, but in fact does so only to the extent that it embraces 'method' (and thereby contradicts itself) since, of course, once a method is pressed into service the question of whether the method is prior to its results or vice versa comes into play.
Rosen is right in identifying the vicious circularity of methodological critique, but wrong in thinking that this viciousness is a problem for immanent critique per se as distinct from immanent critique in a methodologically debased form. For there is no question of what I have just called 'priority' for immanent critique properly understood, and it is only when the question of priority ('Which comes first?') gains purchase that scenarios of vicious circularity unfold. For immanent critique, it is not a matter of priority but of interaction (or totalisation): as was stressed earlier in connection with practical reflexivity, both the categories and the object — as it were, the 'method' and the 'outcome' — of such theorising have something to say about one another; neither has exclusive rights. And as we have also seen, this holds true for all 'good' conversation. Rosen's critique of immanent critique has therefore the following use-value: it serves to highlight what it is in immanent critique that allows it to pass through the meshes of his condemnatory net.

Thus: immanent critique escapes the aporias of methodology by placing itself at issue within conversation. Methodology (a department, and perhaps the fulfilment, of philosophy) destroys conversation, unless it is forced down into the crucible — the totalisation — of conversation itself. In a sense, the only methodology of good conversation is that it has no methodology: all claims can be raised. Invocation of method is at most a moment in conversation, and most often a suspect one: it is not methodology which defines the parameters of rigorous conversation but vice versa. This openness on the part of good conversation — no categorial holds are barred — corresponds to the Hegelian and Marxian perception of the individual, or conversational participant, as social and intersubjective through-and-through (Hegel, 1977, pp. 110-1; Marx, 1975, vol. 5 p. 4 Thesis VI). Theorisation and practice; theory and metatheory; first- and third-person theorising; practical reflexivity and immanent critique; immanent critique and conversation; all these form moments of a single totalisation in which the notion of methodology, like that of philosophy, has no place. And in case the notion of 'conversation' should seem too polite to capture the Marxist notion of class-hatred (Negri), this is to be noted: nothing is less polite than rigorous conversation pursued to its end. For Hegel, such conversation amounted to a 'drama of suspicion' (Gunn, 1988a); for Marx, famously, immanent critique plunges over into satire and vitriol. No- holds-barred conversation precludes none of this. Least of all does it mean that one's placing of oneself at issue entails deriving one's agenda and categories from those planted on enemy terrain. All that it means is that the dialogical condition of immanence is sustained and, thereby, the dogmatisms of philosophy and methodology are avoided. No-one says that one has to like the opponents whom, literally for the sake of argument, one agrees to respect. And no-one can say in advance where (into what issues of life-and-death struggle) good conversation may lead.

Abstraction

I should like to carry my account of the conceptual status of Marxist theorising one step further. The themes of practical reflexivity, immanent critique, etc., are followed through by Marx into his understanding of abstraction. All theorising employs abstractions, Marxist theorising included. But Marx's abstractions, inhering as they do in a particular theoretical self-understanding (which I have attempted to explicate), take a quite specific form.

For the purposes of clarifying this form we can say, schematically: abstractions are of two kinds. There are empiricist and there are determinate (sometimes also called substantive) abstractions (cf. for what follows Bonefeld, 1987c; Gunn 1987b) . Of these two kinds, empiricist abstraction is the most familiar. Marx signals this notion, and offers it provisional endorsement, when in his 1857 Introduction to the Grundrisse he says that 'Production in general is an abstraction, but a rational abstraction insofar as it . . . fixes the common element [i.e. the common element in all modes of production] and thus saves us repetition' (Marx, 1973, p. 85). The idea of production in general' is that it abstracts from concrete social circumstances, in the sense of leaving what makes them different from one another out of account. This notion of abstraction-from is the defining mark of abstraction in its empiricist sense.

The notion of determinate abstraction is a less familiar one, and it is to this notion that the 1857 Introduction devotes the greater space. The example of determinate abstraction which Marx offers is that of labour'. Labour is, to be sure, an empiricist abstraction in the sense that all possible modes of production entail work: Capital Vol. I reports the production of use-values as the 'eternal, nature-imposed condition' of any conceivable form of social life (Marx, 1976, p. 194). But labour is not just an empiricist abstraction. It is also a determinate abstraction because 'this very simple category . . . makes an historic appearance in its full intensity only in the most developed [i.e. in modern capitalist) conditions of society. By no means does it wade its way through all economic relations' (Marx, 1973, p. 103). Labour as an empiricist abstraction is the genus under which historically specific forms of labour fall. Labour as a determinate abstraction is an historical (a capitalist) species of labour while remaining no less abstract than the genus for its part. Labour as a (determinate) abstraction makes its historical appearance only when commodity-production has been generalised. Capitalism itself, as a mode of social practice, effects the abstraction from concrete kinds of labour; and so the abstraction has not merely a theoretical but a social and practical status, as anyone who sells the use of their labour-power discovers fast enough. Moreover, if abstract labour had merely a generic and not a socially specific existence, Marx's contention (1976, Ch . 1, section 2) that a contradiction obtains as between abstract and concrete labour would be unintelligible: for between a genus and its species no contradiction can occur. It is in the nature of a genus/species hierarchy that it becomes increasingly abstract the closer to its apex we move .

It is the notion of this hierarchy that Marx challenges. Moreover, it is on the distinction between empiricist and determinate abstraction that his critique of political economy turns. Political economy, he urges, elides determinate with empiricist abstraction thereby understanding the abstract - or value-producing - labour of capitalism as the condition of all societies of whatever kind. Thereby all societies are construed as capitalist, or at least commodity- producing, societies: the hegemony of value becomes the order of any imaginable social day.

Without entering into detailed textual discussion I should like to add, in parenthesis, a qualification to what has just been said. A good deal turns on how the category of 'labour' in the above example is understood. Labour as such is already an abstraction half-way towards empiricism. I suggest that Marx understands labour as, like capital, a social relation. In short it is the capital-labour relation. The determinate abstraction Marx calls 'labour is class struggle. 'Work which is liberated is liberation from work' (Negri, 1984, p . 165). The political consequences of this qualification cannot be debated here.

What can and must be debated is the significance of the idea of determinate abstraction in Marx's post-1857 writings. Labour, for Marx, is never labour as such. I said that Marx provisionally endorses the notion of empiricist abstraction. The force of this 'provisionally' is that Marx tells us that 'there is . . . no general production' (Marx, 1973, p. 86). Thereby, he invites us to construe all abstraction, and all the categories of his own theorisation — 'labour and 'production' included — as determinate abstractions. Class, class-struggle and value have to be understood in this way. In the most logically radical sense, in other words, he places his own theory socially at issue (at issue within struggle). Not merely do his categories have practical roots and practical effects: they have practical existence as well. If my earlier discussion of practical reflexivity and immanent critique is to be meaningful, therefore, its relation to the idea of determinate abstraction must be made clear.

The relation concerned can be stated very simply. Practical reflexivity opens on to the notion of determinate abstraction (it makes it possible); and the notion of determinate abstraction presupposes that theorisation understand itself in a practically reflexive way.

This is so because, in the first place, theorising which operates in terms of determinate abstraction has to be theoretical and metatheoretical at the same time. The categories in terms of which such theorising raises its truth-claims (e.g. labour, value, class) are the categories which that same theorising understands to be constitutive in practice of its social world. The categories which metatheoretically 'control' discourse are also the categories which at a first-order level discourse 'finds'. This articulation becomes nonsensical unless a theory-metatheory unity is possible, and it is this possibility which I have sought to establish above.

And in the second place theorising which operates in terms of determinate abstraction has to be third-person and first-person at once. 'Labour', 'value', etc . have a third-person existence as part of a determinate social world, a world which (Marx, 1973 pp. 100-1) goes on existing whether it is theorised or not. But on the other hand categories, including practically existing ones, exist in the first person; they can at least in principle be engaged with and understood. This requisite third-person/first-person unity was part of what, earlier, we saw the self-consistency of practically reflexive theorising to involve.

The conditions of determinate abstraction and of practical reflexivity are accordingly one and the same. Only practical reflexivity (see above) effects the synthesis of theory with metatheory which determinate abstraction requires; determinate abstraction for its part is the conceptual and political horizon towards which practically reflexive critique clears the way. And this tells us something about the trajectory of Marx's own thought. Marx thematizes practical reflexivity in (especially) his polemical writings of the 1840s and determinate abstraction in (especially) his critique of political economy in its post-1857 phase. The implication of the argument I have just proffered is that he thereby deepens rather than abandons his earlier insights. Pace Althusser for example (Althusser, 1969, pp. 33- 8, 227-31), there is no 'break' in the development of his work. On the contrary a single totalisation ever-more-richly unfolds, viz., a totalisation open on to practice. Marx pursues the implications of this totalisation — one which I have attempted schematically to reconstruct — not with breaks and jumps (Althusser) or with lacunae and gaps (Bhaskar) but self-consistently and from end to end.

Critique of Critical Realism

As argued above, Critical Realism understands the theory-practice interrelation in a causal sense: the internal interrelatedness of a totalisation falls out with its horizon. And in virtue of this the question of categories' social and political entailments floats free from the question of their implications for truth. The 'in-the-course-of movement of practically reflexive theorising becomes unattainable. Bhaskar misses the notion of categories' practical existence, on which the notion of determinate abstraction turns. He misses it because he construes critique as philosophical critique. Determinate abstraction and Critical Realism stand at opposite conceptual poles.

This has a specific political consequence, to which I shall draw attention only briefly. Determinate abstraction's understanding of abstractions as socially existing allows it to mount an ideology-critique which is directly, and at the same time, social critique. To criticise ideas just is to criticise practical relations; and conversely. This is how classic Marxism has always understood the matter. But Critical Realism, lacking the moment of determinate abstraction, severs ideology-critique from the critique of society itself. The relation between the two becomes an external one, because their respective objects are understood as related in an external (a causal) way. Thereby, whether wittingly or not, Bhaskar lends the credence of his position to the view — perhaps more fashionable a decade ago than today — that the left has to fight an all-but-interminable 'war of position' (cf . Gramsci, 1971, pp . 235-43) against the ideologies of the right. The fight is indeed interminable as long as it goes forward merely ideologically since the line dividing the critique of ideas from the critique of social conditions is a line drawn on rightist terrain. Already for Marx in 1844, by contrast, the 'weapon of criticism' and 'criticism by wapons' (Marx, 1975, vol. 3, p. 182) were internally linked. Determinate abstraction is his subsequent conceptual and political deepening of just this interconnection. It was towards this that his 1840s critique of Young Hegelian philosphical critique cleared the way. A purely ideological critique is politically impotent since it presupposes the validity of the separation of 'ideas' from 'actuality' which (to the status quo's advantage) allows practice to ignore the challenges theory presents. Critical critique becomes severed from political critique. Within the confines of the present article it is possible to state these points only generally, but nonetheless they serve to endorse my main contention: the philosophical separation of theory from metatheory, although apparently a technical and academic matter, is fatal for the political project of the left.

Critique of Critical Realism (3)

So far, I have not commented directly on what Bhaskar's Interlink article presents as being the central themes of a Critical Realist approach. My reasons for this indirection were made clear in my introductory remarks. But now some pivotal doctrines of Critical Realism can be placed, polemically, on centre-stage.

Bhaskar tells us that Critical Realism is concerned to 'identify the structures at work which generate . . . events and discourses'; we are invited to recognize that 'there are enduring structures and generative mechanisms underlying and producing observable phenomena and events' (Bhaskar et al. 1988; for a further development of the same points see Bhaskar, 1975, esp. Ch. 1: the terminology is more or less unchanged4 ). The difficulty with these seemingly innocuous formalae is that they involve us in tautology. On the one hand there are observable phenomena, on the other generative mechanisms. The criterion for truthfully identifying a generative mechanism is that it successfully explains (or at least renders intelligible) the phenomena. But the criterion for truthfully identifying the phenomena can only be appeal to some generative mechanism. The first of these points may be readily granted but the second is more contentious; and so let me explain.

There is no such thing — it is banal to state this — as a 'brute' (a category-neutral) fact. Facts, and thereby descriptions and identifications of phenomena, count as such only within some categorial framework. In the natural sciences, it may seem that the identification of facts is unproblematic (either the litmus paper turns red or it doesn't) but in the social sciences there is a further question of ideological mystification to take into account. For in the first place mystifying categories — e.g. 'abstract labour' — may, as determinate abstractions, be socially real. (It is not necessarily just us who are mystified about society; society may be mystifying for its part.) And so, in the second place, what we count to be a fact will depend on the theorisation — in Bhaskar's terms the specification of 'generative mechanisms' — which we endorse. There is no such thing as a social litmus-test (although even in natural science the seemingly straightforward matter of litmus-testing gives rise to categorial problems of its own: Kuhn, 1962). And it is in this way, I take it, that Critical Realism is critical: it refuses to accept appearances at their face value as though they were givens of a 'brute' kind. It is interested in generative mechanisms or structures which 'are not spontaneously apparent in the observable pattern of events' (Bhaskar et al. 1988), and so distinguishes truthful from misleading appearances in the light of the generative mechanisms which it finds. But then appearances become the criterion of generative mechanisms (of reality) while generative mechanisms (reality again) become the criterion of phenomena or appearances. We can evaluate claims about reality only in the light of claims about appearance and vice versa. In formal terms: tautology is the outcome. Less formally: we are sent from pillar to post.

These problems affect not just a Critical Realist account of social science but a Critical Realist account of natural science as well. On a Critical Realist approach, natural science escapes tautology (cf . Hegel, 1977, pp . 94-5) if and only if the categories in terms of which observable 'facts' are identified can be specified independently of the categories in terms of which the generative mechanisms serving to explain these facts are specified for their part. This may seem a simple enough requirement, and yet in the history of science it has never been met. To Alexandre Koyre (1978), and after him to Husserl (1970, pp. 48-9), we owe the demonstration that the rise of modern natural science relied on an idealised 'mathematization' of space and time which was not itself something empirically discoverable but which, on the contrary, was a categorial presupposition of scientific explanations turning on the notion of predictability (in the sense of all causes, everywhere and everywhen, having the same effects). Space and time become homogenised at the start of the modern period — they become abstracted from the qualitative space and time of the everyday lifeworld — by natural science on the one hand and by the exchange-relation (cf. Adorno, 1973) on the other. To the same effect, the early-modern philosophers of science — John Locke, for example — treat only those experiential qualities which can be quantified (e.g. distance but not scent) as fully real. In short science has always precast, in its own categories, the 'observable phenomena and events' which it explains. Certainly this is not an objection to natural science as such inasmuch as, in numerous cases, a tautologous statement can be informative and to the point. But it is an objection to a philosophy of science since, from such a theorisation, we might hope at least to discover how and why it is that science's tautologies are of an informative kind. Critical Realism repeats science's tautology, perhaps faithfully, but without casting light upon it. The tautology which should be its subject-matter becomes the tautology of its discourse itself.

Like the dilemma 'vicious circularity or infinite regress?', tautology is the vice of discourse from which the notion of totalisation is absent. A totalising discourse addressing epistemological problems of natural science would attempt to show how mathematization and observation interact as moments of the same project, for better or for worse. My criticism of Critical Realism's central doctrines relates to my criticism of it as a 'philosophy', this latter being the argument on which the present paper has turned.

Appearance and reality

Is it an accident that Critical Realism becomes trammelled in the problems of tautology? I suggest that it is not. Once more, as just signalled, the problem stems from a disregard of the totalisation on which Marx (together with Hegel) set store.

Wherever appearance and reality are severed from, and dualistically counterposed against, one another the danger of tautology looms. We experience 'only appearances' ; reality is 'something else'. If this is so then only appearances can serve as a guide to reality; and only reality can serve as the yardstick against which — unless we are willing to say that no appearances can be mystifying or misleading — appearance is to be judged. Vicious circularity/infinite regress; theory/metatheory. In just the same way: tautology is the nemesis which an appearance/reality severance has to confront. This reflection enables us to report one further moment in the theoretical totalisation which Marx, following upon Hegel, attempts.

Historically speaking, the appearance/reality severance goes hand in hand with the rise of natural science and that of the exchange-relation. The Ancient world did not see the relation of appearance to reality in the same terms. The pre-Socratic philosopher Anaxagoras characterised appearances, or phenomena, as 'a glimpse of the obscure' (cit. Kirk & Raven, 1960, p. 394). He meant that what is revealed in and through appearances (viz. reality or 'being') was irreducible to them while at the same time being internally linked with them: as it were, it is no accident that reality appears (even misleadingly) thus-and-thus. To be sure, Anaxagoras is replying to Parmenides, who counterposes the 'way of seeming' against the 'way of being' (Kirk & Raven, 1960 Ch. 10) in a thoroughly dualistic sense. To be sure, too, Plato is more faithful on this score to Parmenides than to Anaxagoras; but even Plato, in his Republic (509-11 in the standard citation) agrees to construe appearance and reality as two segments of one and the same divided line. That is, he allows their internal relation (the unity of the line) to be primary. It is in the modern — the capitalist and natural-scientific — world that the idea of a pure external relation of appearance to reality comes to the fore. And again, as in the case of the theory/metatheory severance, it is Hegel and Marx who attempt to re-establish unity and internal-relatedness albeit on a non-cosmological ground. Plato could presuppose a unity of appearance and reality because he presupposed a cosmos to whose meaningful order the philosopher might assimilate himself (Republic 500c-e): there had to be an assured ladder leading from appearance to truth. Marx and Hegel presume no such cosmos but, instead (and construing even natural appearances as ones which are socially constructed), address the appearance-reality relation on a solely practical terrain.

Hegel declares that 'essence [in the context of the present argument: reality) must appear: 'Essence is not something beyond or behind appearance' while appearance, for its part, is not a mere illusion of 'show' (Hegel, 1874, para . 131; cf. Lukacs, 1979 on 'determinations of reflection'). Marx too, although he is famous for having penetrated through appearance to underlying reality, thinks of appearance as the mode in which the capital-labour relation exists (e.g. the reference to the world-market at Marx, 1966, p. 110; cf. Negri, 1984 on 'social capital'). Thus: exchange-value is the mode of existence of value, and money the mode of existence of exchange-value. Social structures are themselves the mode of existence of action and struggle: structures are struggles existing in the mode of being denied (cf. Gunn, 1987b; and as background Hegel, 1977, pp. 263-4). The ossification of structure is the appearance which overlies the reality of struggle, as the opening sentence of the Communist Manifesto reports, but once again there is just a single totalisation (whose key is struggle) and not two. As it were: social structures are non-existent, but they are non-existent in the mode of being real. The movement from the first to the third volumes of Marx's Capital is — all this being so — not a move from reality to unreality, or from production to ideology, but an increasingly concrete picture of what the mode of existence of capital involves. The same point can be made by saying that the famous 'abstractions' of Volume One are abstractions of a determinate kind.

In other words Marx (and Hegel) argue for just the internal relatedness of appearance and reality from whose absence Critical Realism's trammelment with tautology stems. Tautology arose because two allegedly separate things were supposed to make sense of one another, within a causal-explanatory frame. An internal as distinct from an external relation between appearance and reality overcomes tautology by understanding these two things, precisely in their difference, as one. Nothing is explained by anything else or, put differently, there are no 'generative mechanisms'. Instead there is determinate abstraction: the existence of unity in difference and of the abstract in the concrete. The notion of determinate abstraction is the totalisation of appearance and essence. Theory forecloses on practice in the same movement as, minus determinate abstraction but plus philosophy, it understands itself as the most surface level of appearance itself. Every 'in-the-course-of' becomes disrupted. No doubt despite Bhaskar's intentions, the aged base/superstructure metaphor becomes the order of the discursive day.

*If not, why not? Certainly Marx says things which tell against my argument: for example he advises us to 'leave philosophy aside' and, in the opening pages of Capital, he sets out a putative deduction of a labour theory of value which is not only fallacious in its own terms (e.g. Bohm-Bawerk, 1975, pp. 68-80) but which has the consequence of presenting 'labour' as an abstraction not of a determinate but an empiricist kind. The immeasurable richness of what Marx does is, however, something else again. It is on these riches that I have attempted to concentrate. In the 1980s it is fashionable to say that they are exhausted: that we live in a post-Marxist world. The more friendly version of this thesis is to proffer to Marx all kinds of aid. Bhaskar supplies the oxygen-mask of philosophy. Others advertise the crutches of Rational Choice.5 Still others assimilate Marxism to a bourgeois sociology of inert structures. The answer to the question 'If not, why not?' is twofold. The first part of it turns on emphasising that an external or 'causalist' disconnection of theory from practice, as in the base/superstructure metaphor or a thousand other versions of sociological and philosophical determinism, has not the least in common with the discourse of totalisation we are still able to recover from Marx's works. Bhaskar et al. (1988) offer to overcome the dichotomy within Marxism of 'fundamentalism versus revisionism', but this is perhaps their most suspect procurement since fundamentalist readings of Marx had, at least, the virtue of being readings of a careful kind.

And the second part of my answer is still more straightforward. Our own weakness may not be Marx's. To insist that a healthy ambulant walk with crutches (in Marx's case the crutches of philosophy) is to require that he flatter an audience, whose applause he might well have regarded as anathema, by falling over his own theoretical feet.

* Critical Realism's offer of a philosophical reformulation of Marxism seems obviously helpful, but in fact, as I have shown, is not.

References

Adorno, T.W. (1973) Negative Dialectics. Routledge and Kegan Paul.
Althusser, L. (1969) For Marx. Allen Lane.
Althusser, L. (1970) Reading Capital. New Left Books.
Bhaskar, R. (1975) A Realist Theory of Science. Books .
Bhaskar, R. et al . (1988) 'Philosophical Underlabouring' Interlink 8.
Bloch, E. (1986) The Principle of Hope. Blackwell.
Bohm-Bawerk, E. (1975) Karl Marx and the Close of His System. Merlin Press.
Bonefeld, W. (1987a) 'Open Marxism' Common Sense 1.
Bonefeld, W. (1987b) 'Reformulation of State Theory' Capital and Class 33.
Bonefeld, W. (1987c) 'Marxism and the Concept of Mediation' Common Sense 2.
Bubner, R. (1982) 'Habermas's Concept of Critical Theory' in Thompson J.B. & Held D. (eds) Habermas: Critical Debates. Macmillan.
Cleaver, H. (1979) Reading 'Capital' Politically. University of Texas Press.
Cohen, G.A. (1978) Karl Marx's Theory of History: A Defence. Oxford University Press.
Davie, G.E. (1986) The Crisis of the Democratic Intellect. Polygon Books.
Dove, K .R. (1970) 'Hegel's Phenomenological Method' Review of Metaphysics. Vol. XXIII.
Elster, J. (1985) Making Sense of Marx. Cambridge University Press.
Elster, J . (1986) 'Further Thoughts on Marxism, Functionalism and Game Theory' in Roemer, J. (ed.) Analytical Marxism. Cambridge University Press .
Geras, N. (1972) 'Marx and the Critique of Political Economy' in Blackburn, R. (ed.) Ideology in Social Science. Fontana/Collins.
Gramsci, A. (1971) Selections from the Prison Notebooks. Lawrence and Wishart.
Gunn, R. (1987a) 'Practical Reflexivity in Marx'. Common Sense 1.
Gunn, R. (1987b) 'Marxism and Mediation' Common Sense 2.
Gunn, R. (1987c) 'George Davie: Common Sense, Hegelianism and Critique' Cencrastus 27.
Gunn, R. (1988a) 'Recognition in Hegel's Phenomenology of Spirit' Common Sense 4.
Gunn, R . (1988b) 'Facing Up to the Communist Party' Common Sense 6 (forthcoming).
Gunn, R. (1988c) 'In Defence of a Consensus Theory of Truth' (unpublished MS).
Habermas, J. (1986) Autonomy and Solidarity. Verso Books.
Hegel, G.W.F. (1874) The Logic of Hegel. Clarendon Press.
Hegel, G.W.F. (1977) Phenomenology of Spirit. Oxford University Press.
Holloway, J. (1988) 'An Introduction to Capital: or, How I Fell in Love with a Ballerina' Common Sense 5.
Horkheimer, M. (n.d.) Critical Theory: Selected Essays. Seabury Press.
Husserl, E. (1970) The Crisis of European Sciences. Northwestern University Press.
Jessop, B. (1988) Paper to conference on Regulation Theory, Barcelona.
Kirk, G.S. & Raven, J.E. (1960) The Presocratic Philosophers. Cambridge University Press.
Kojeve, A. (1969) Introduction to the Reading of Hegel. Basic Books.
Korsch, K. (1970) Marxism and Philosophy. New Left Books.
Koyre, A. (1978) Galileo Studies. Harvester Press/Humanities Press.
Kuhn, T.S. (1962) The Structure of Scientific Revolutions. University of Chicago Press.
Lukacs, G. (1971) History and Class Consciousness. Merlin Press.
Lukacs, G. (1979) Ontology of Social Being: (1) Hegel. Merlin Press.
Marx, K. (1973) Grundrisse. Penguin Books.
Marx, K. & Engels, F. (1975) Collected Works. Lawrence and Wishart.
Marx, K. (1976) Capital Vol. I Penguin Books.
Marx, K. (1966) Capital Vol. III Progress Publishers.
Negri, A. (1984) Marx beyond Marx: Lessons on the Grundrisse. Bergin and Garvey.
Plant, R. (1988) Citizenship, Rights and Socialism. Fabian Society Pamphlet 531.
Rosen, M. (1982) Hegel's Dialectic and Its Criticism. Cambridge University Press.
Sartre, J.-P. (1950) What Is Literature?. Methuen.

• 1 E.g. Althusser. 1970 p. 59, where we are told that 'theoretical practice [first-order 'science'] is . . . its own criterion' only to learn that this thesis is restricted to sciences 'once they are truly constituted and developed'. Is it science itself which decides that it is truly constituted?
• 2More precisely: relativism consists in the inference that since viewpoints differ categorially therefore compelling argument across the boundaries of category-systems is impossible. For an argument which accepts the premise but resists the inference from it, see Gunn, 1988c.
• 3'Philosophy' has equivocated on this issue. In its most hubristic moments it has reserved all rights to metatheory. In its more modest guise — the guise of science's 'underlabourer' — it has confined itself to clearing what science declares to be rubbish from science's path . Althusser's shifting formulations on 'philosophy' moved, over a number of years, from the 'hubristic' to the 'modest' view. My contention is that this shiftiness is intrinsic to philosophy. At each of its stages, including its most self-effacing ones, the same dilemmas are reproduced.
• 4One terminological change should however be noted. What Bhaskar now calls 'critical' realism he used to call 'transcendental' realism (Bhasker, 1975 p . 25). In other words, as in just about all philosophy of science, the inspiration is Kant. For a sharp comment on the relation of Kantian critique to Hegelian-Marxist critique see Bubner, 1982.
• 5Although a critique of Rational Choice Marxism falls outside the bounds of the present paper, it is worth pointing out that it is on the theory/metatheory separation that the case for Rational Choice depends. Jon Elster rejects the view that 'methodological individualism . . . must go together with an assumption of universal selfishness' (Elster, 1986 p. 209; cf. Elster, 1985 Ch. 1) not least in order to distance his individualism from that condemned outright by Marx (1976, p. 284). Despite this he says that there is a methodological presumption in favour of attributing selfish motives to individuals inasmuch as-allegedly-altruism is parasitic on selfishness rather than vice versa. The arguments for and against this kind of lunacy are decidedly avoidable. I wish only to stress that Elster's case for distinguishing real selfishness from methodological selfishness, and thereby the entirety of his case for a Rational Choice Marx-reconstruction, self-evidently turns on the idea of a theory/metatheory separation once more. We are supposed to be atomised and self-interested agents only as a matter of methodology, apparently, and not as a matter of first-order and social fact.