An immanent transcendental Foucault, Kant and critical philosophy
Every philosophy conceals a philosophy; every opinion is also a hiding place, every word also a mask.
Nietzsche, Beyond Good and EvilThe relation of Foucaultʼs work to philosophy remains an unsettled issue. Indeed, Foucault sometimes preferred to present himself as ʻthe masked philosopherʼ. Much like Nietzscheʼs ʻhermitʼ, Foucault wrote books to conceal what lies within, a deeper cave behind every cave, ʻa stranger more comprehensive world beyond every surface, an abyss behind every ground, beneath every “foundation”ʼ.  However, a number of readers of Foucault have noticed that he constantly returned in his published work and interviews to an encounter with Kantian philosophy and the concept of the ʻtranscendentalʼ. Although these readers – including Gilles Deleuze, Jürgen Habermas, Beatrice Han, Gary Gutting and others – represent a broad range of interpretations of his work, the idea of the transcendental plays a key role in these readings providing the grounds for the legitimation, critique or disqualiﬁcation of Foucaultʼs thought and its relation to philosophy. What is the status of the transcendental in Foucaultʼs work and what is Foucaultʼs relation to transcendental philosophy? Is the transcendental just another mask that is temporarily utilized and then abandoned in Foucaultʼs thought when it became clear that forging a new relation between the transcendental and empirical would eventually lead to insurmountable logical and theoretical difﬁculties? Or, rather, is there perhaps an attempt on Foucaultʼs part to ʻrestore the forgotten dimension of the transcendentalʼ,  developing a conception that goes ʻall the way downʼ, so to speak, an immanent conception of the transcendental consistent with a thought without ground?
Drawing from some of the readings examined here I want to argue for this latter view. Foucaultʼs philosophy can be understood in terms of the development of his own conception of an immanent transcendental out of resources provided, in part, from Kantʼs own work. Foucaultʼs work could thus be seen as a ʻradical transformation of Kantianism, a re-invention of the critique which Kant betrayed at the same time as he conceived it, a resumption of the critical project on a new basis and with new conceptsʼ.  Rather than a set of inconsistent, contradictory or viciously circular relations between the transcendental and the empirical, as some of Foucaultʼs best-known readers have claimed, I argue that in Foucaultʼs reinvention and transformation of Kantianism he develops an immanent conception of their relation as contingent and differential, a circle in which transcendental elements are immanently ʻcaught up in the very things they connectʼ  without being reduced to the same or to a simple repetition. By contrasting these differing accounts of the transcendental I will attempt to renew the question of what is at stake in Foucaultʼs critical project more than twenty years on, as well as raise important questions about the contemporary value of the transcendental, the Kantian legacy and the nature of philosophy itself.
In the ﬁrst section I introduce the concept of the transcendental in Kant and post-Kantianism, indicating brieﬂy how this has been taken up in contemporary philosophy. In the main section I explicate and contrast interpretations of the transcendental in some of the best-known recent readings of Foucaultʼs work. In the ﬁnal section I raise some questions regarding these interpretations, the continuing value of transcendental philosophy and the nature of philosophy itself, and I conclude by laying out the grounds for Foucaultʼs conception of the immanent transcendental.
Kant and the post-Kantian transcendental
In modern philosophy it is of course with Kant that the concept of transcendental analysis undergoes a momentous transformation. In contrast to the medieval ʻtranscendentalsʼ Kant tied the very meaning of the ʻcritical projectʼ to an analysis of transcendental conditions. Kant attempted to ground the possibility of knowledge by deﬁning the necessary and universal conditions of experience as a priori and hence irreducible to the empirical. The famous answer given in the First Critique is that the necessary and universal conditions of experience rest on an a priori analysis of subjectivity. The Kantian analysis demonstrates that the transcendental subject synthesizes or ʻschematizesʼ the sensible forms of intuition with the categories of the understanding to produce the conditions for the possibility of objects of experience and experience itself. In Kantʼs well-known formulation ʻthe conditions of the possibility of experience in general are likewise the conditions of the possibility of the objects of experienceʼ.  Experience, then, for Kant must have certain structurally necessary conditions and these conditions are double. There are two ʻrootsʼ or sources of knowledge and these are the a priori or transcendental conditions of space and time and the categories. The ﬁrst critique is an analysis of how these sources could come together in a new kind of assertion: a synthetic a priori judgement. It is in answering the famous question of how synthetic a priori judgements are possible that Kant believed that he had secured an empirical realism on the basis of a transcendental idealism. For Kant this means two things: ﬁrst, external objects can be perceived and exist independently of us: they are empirically real. Second, these real objects in space and time are relative to the a priori forms of experience, the set of necessary conditions that must obtain if experience and the object world are to have the character that they do for us. Our experience of the world as empirically real is possible because the limit-conditions of experience are transcendently ideal. Knowledge that transcends the bounds of these limitconditions is impossible. Thus, the Kantian transformation of the transcendental concerns not the objects of knowledge themselves but our mode of knowing them, or, as Kant puts it, transcendental philosophy is an investigation of ʻour mode of cognition of objects insofar as this is possible a prioriʼ.  If every aspect of Kantʼs account in the ﬁrst critique is subsequently challenged it is arguably this insight regarding the genesis of modes of knowing a priori – upon how knowledge is legitimated, made possible or produced by either ʻimmanentʼ or ʻtranscendentʼ conditions that precede it a priori – that is retained and taken in new directions by post-Kantianism.
Although modern philosophy is deeply indebted to this Kantian legacy it is clear that the major postKantian traditions have been unable to accept Kantʼs solution to the critical question of legitimating how knowledge is produced, and so modern philosophies, in the wake of Kant, either abandon the transcendental project altogether, radically limit its range and scope, or try to develop their own modiﬁed and reconﬁgured notions of the transcendental. Within the ʻanalyticʼ tradition, for example, philosophers have on the whole remained suspicious of the transcendental with its attendant issues of idealism and veriﬁcationism, and so transcendental philosophy has tended to become narrowly focused upon epistemological debates over the nature and structure of ʻtranscendental argumentsʼ and whether they can be ʻnaturalizedʼ or at least do sufﬁcient work to defeat the sceptic. This kind of work is best exempliﬁed by Stroud and Strawson and most recently by Cassam, Sacks and others.  By contrast, in the recent ʻcontinentalʼ tradition philosophers have continued to appeal to modiﬁed versions of the transcendental despite its negative association with metaphysics, with the speculative, the ahistorical, the universal, the subjective or the foundational. In the early part of the twentieth century it is Husserlʼs commitment to a phenomenologically-inspired and reformulated transcendental idealism that remained enormously productive yet problematic for a whole generation of German and French thinkers beginning with Heideggerʼs own complex transformation of Husserlʼs transcendentalism and culminating, towards the end of the century, with Derridaʼs ʻquasitranscendentalsʼ, Irigarayʼs ʻsensible transcendentalʼ, Deleuzeʼs ʻtranscendental empiricismʼ and, of course, Foucaultʼs own ʻhistorical a prioriʼ.
Whether Foucaultʼs work can be understood from within the post-Kantian tradition as an effort to engage the concept of the transcendental is a question that a number of Foucaultʼs best-known readers have explored. The issues that divide those readers revolve around the extent of Foucaultʼs indebtedness to this tradition of transcendental thinking within European philosophy, whether his conception of transcendental thought either escapes or transforms the problems it was meant to deal with, or, indeed, whether it is inconsistent or aporetic and, therefore, insufﬁcient as a strategy for dealing with the problems he addresses. Gary Gutting, for example, one of the best-known interpreters of Foucault, argues that Foucaultʼs relation to transcendental philosophy, and post-Kantianism generally, is something of a non-issue since Foucaultʼs thought has little or no connection to these traditions. Gutting would have us abandon reference to transcendental talk in Foucault. Foucaultʼs work is not the work of a transcendental philosopher for Gutting since a transcendental project is deﬁned by a commitment to establishing necessary conditions for the possibility of knowledge. These necessary conditions refer to the constituting power of the transcendental subject that functions as ground for experience. Given that Foucaultʼs work for Gutting undercuts this commitment by refusing to refer necessary conditions to a transcendental subject and by showing how necessary limits are historically contingent limits in disguise then, for Gutting, Foucault cannot be a transcendental philosopher. As Gutting says, Foucaultʼs project in Kantʼs terminology is critical (examining assumptions regarding the scope and limits of our knowledge), but it is not, like Kantʼs own project, transcendental. It does not, that is, claim to discover necessary conditions for knowing that determine categories in terms of which we must experience and think about the world and ourselves. 
Thus, for Gutting, Foucault utilizes historical rather than strictly a priori philosophical methods for his critical project since Foucaultʼs works are ʻprimarily works of historyʼ and his main concern is with ʻforging a new approach to historical analysisʼ.  For Gutting The Order of Things is Foucaultʼs most philosophical book but if we construe its claims, for example in chapter 9, as a critique of individual philosophers it ends up as a discussion of the history of ideas and it ceases to work as an archaeological investigation of unconscious structures; on the other hand, if we view Foucaultʼs discussion here as genuinely archaeological then it is the modern episteme governed by the concept of ʻmanʼ that is shown to be incoherent: ʻin neither instance has Foucault made an effective case for or against a standard philosophical positionʼ.  Thus Gutting makes the claim that ʻeven in his most apparently philosophical moments, Foucault is not a participant in the debates of modern post-Kantian philosophyʼ. 
Transformation of the transcendental
That a ʻnew approach to historical analysisʼ could be conducted alongside and in conjunction with a post-Kantian philosophical project is, since Schelling and Hegel, at least plausible, rather than mutually exclusive as Gutting seems to imply. One could argue that Foucault is not simply doing either history or philosophy, neither simply history of ideas nor history of philosophy, but working out a new philosophical relation to history. Moreover, that a new style of historical analysis could be conducted in association with a reconﬁgured form of transcendental philosophy is precisely the kind of project that Foucault himself had already signalled an interest in, as early as his D.E.A., with the title ʻthe constitution of a historical transcendental in Hegelʼ.  References to a transformation of Kantian thought span the entire range of Foucaultʼs work, beginning with Foucaultʼs introduction to his translation into French of Kantʼs Anthropology from a Pragmatic Point of View right through to the very last works. The Kantian a priori is reworked in several key texts, including The Birth of the Clinic, where Foucault introduced the concept of the ʻconcrete a prioriʼ in order to examine the ʻhistorical and critical understanding of the old experienceʼ  of disease, to late references to ʻgames of truthʼ as the historical a priori of experience. In The Archaeology of Knowledge Foucault explicitly distinguishes what he called a ʻformal a prioriʼ from his own ʻhistorical a prioriʼ since he claims he was not interested in discovering the Kantian conditions of validity for judgements but rather in determining the concrete conditions of reality for ʻstatementsʼ. ʻThe formal a priori and the historical a priori neither belong to the same level nor share the same nature: if they intersect, it is because they occupy two different dimensions.ʼ  Thus, the conditions that Foucault is interested in operate in another ʻdimensionʼ from the Kantian formal a priori but are no less ʻphilosophicalʼ. In this other ʻarchaeologicalʼ dimension the conditions are not ahistorical and universal rules that determine in advance what could be given or said but are rather the historically changing rules of what is actually given and said. And these rules are themselves a ʻtransformable groupʼ since they do not sit above events like an ʻatemporal structureʼ in some ʻunmoveable heavenʼ but are ʻcaught up in the very things they connectʼ.  It is perhaps the nature of this being ʻcaught upʼ that Foucaultʼs work constantly strives to understand and explicate, and it is also, as we will see, one of the major points of contention in interpreting Foucaultʼs transcendentalism.
In the archaeological period, rather than rediscovering in Kantian fashion what might legitimate an assertion, Foucault claims that his historical a priori reveals the principles according to which statements survive, are transformed or disappear, and these principles are caught up in the very things they connect. Although at this stage the nature of these ʻprinciplesʼ that govern discourse are not entirely clear (as Foucault was later to claim, what else could he have been talking about here but ʻpowerʼ?) what does seem clear is that Foucault often and explicitly developed his methodological approaches by transposing the Kantian formal a priori into another dimension, a dimension that does not change historically – as Foucault says, the historical a priori is not simply a formal a priori endowed with a history – but a dimension that changes with history. Foucaultʼs approach to this space is not dependent upon an a priori analysis of subjectivity or upon universal and necessary conditions, but upon a relative and variable historical a priori of knowledges. Thus, the idea that Foucaultʼs archaeologies and genealogies are not transcendental in the strictly Kantian sense but perhaps still best viewed as contributing to a form of historicized transcendental philosophical discourse appears to be recognized by Foucault himself.
Apart from Gutting, whose conceptions of the transcendental and of philosophy appear perhaps too narrow to capture what Foucault is up to here, this novel relation between history and philosophy is recognized by other readers of Foucaultʼs work, even if, in the end, some of these readers think this relation collapses under the pressure of its own internal contradictions. For example, Jürgen Habermasʼs well-known criticisms of Foucault are based upon accepting at least that the critique of conventional historiography (with its residual anthropologism and humanism) found in Foucault emerges out of what Habermas calls a ʻtranscendental historicismʼ. For Habermas Foucaultʼs historicism is set within a ʻweakʼ or looser sense of transcendental rules that are formed, displaced and regrouped as formations shift and are reshaped by nothing other than the ever-renewed technologies of power. In Habermasʼs account Foucault utilizes this power principle to replace the Kantian transcendental subject:
what the synthetic power of transcendental consciousness was hitherto supposed to accomplish for the one and general universe of the objects of possible experience – this synthesis – is now degraded into the subjectless will of a power effective in the contingent and disordered to-and-fro of discursive formations. 
Foucauldian power, in Habermasʼs reading, is joined together with the transcendentalist meaning of synthesis to produce a ʻpurely structuralistic activityʼ,  a ʻKantianism without the subjectʼ as Ricoeur put it. But this structuralist activity cannot be simply ʻpureʼ in Foucault, as Habermas points out, since the transcendental generativity of power is combined and connected with the emergence of events. The event is articulated with and immersed in structures of power, or structures are caught up in the events they connect. 
It is, however, this (con)fusion of the transcendental with the empirical, of connectives with things connected, that acts as the source of Habermasʼs and othersʼ criticisms of Foucault. For Habermas Foucaultʼs concept of power operates in an ʻirritating double roleʼ:19 on the one hand, Foucaultʼs work functions in an empirical role where we are given descriptive and ʻneutralʼ analyses of the technologies of power, yet, on the other hand, Foucault wants his work to operate in a critical transcendental role where analyses of the technologies of power explain how discourses about man are possible at all. Power in Foucault is ʻcontaminatedʼ for Habermas because Foucault forces it to play these ʻparadoxicalʼ yet incompatible twin roles. As Habermas puts it: ʻIn his basic concept of power Foucault has forced together the idealist idea of transcendental synthesis with the presuppositions of an empiricist ontology.ʼ  By attempting to historicize and temporalize the a priori, Foucault thus ʻundertakes a fusion of opposed meaningsʼ that performatively constitutes power as a conception both of ʻtranscendental generativity and of empirical self-assertionʼ  simultaneously. For Habermas this cannot be a ʻway out of the philosophy of the subjectʼ  because the concept of power that provides the resources for both empirical and transcendental roles is drawn from the philosophy of the subject itself. On Habermasʼs reading Foucaultʼs theory of power is trapped within a vicious circle of its own devising – the well-known ʻperformative contradictionʼ – complete with all the aporias of the philosophy of the subject that, according to Habermas, Foucault thought he had left behind. These aporias are deﬁned and speciﬁed in Foucaultʼs case by what Habermas famously called ʻpresentismʼ, ʻrelativismʼ and ʻcrypto-normativismʼ. 
Where Gutting argues that the transcendental project entails a commitment to a Kantian subject and so Foucault could not be doing transcendental philosophy, Habermas also appears to assume that Foucaultʼs transcendental project ultimately depends on a Kantian conception of the subject as man, without really providing additional argument for it, and that turning from this subject to a historicized transcendental form of power will not escape the conceptual constraints of the modern subjective tradition. However, the idea that Foucaultʼs conception of the transcendental requires a Kantian understanding of the a priori subject is not ʻnecessaryʼ if the transcendental is rethought as no longer dependent upon uncovering apodictic certainty, absolute foundations for knowledge or the discovery of what is true independently of experience. We have already suggested that Foucault was perhaps searching for a non-anthropological and historicized dimension of the transcendental (a ʻcompletionʼ of Kant that Foucault often designates as the ʻNietzschean experi-enceʼ) without presupposing the Kantian subject as ʻmanʼ or as condition of possibility. I would suggest that in carrying out this project Foucault can be interpreted as offering a ʻtranscendentalʼ account of Kantʼs own transcendental subject, a ʻcritique of critiqueʼ, or what the post-Kantian tradition has sometimes called ʻmetacritiqueʼ, and he ﬁnds some of the resources to do so in Kant.  A Foucauldian metacritique will involve the creation and analysis of a new critical space opened by the Kantian reﬂection, yet inverting the (Kantian) direction of the critical process and the signiﬁcance of experience: instead of anticipating the possibility of all knowledge by prescribing in advance its own laws (Kant), Foucault begins with already constituted forms of knowledge in order to deﬁne retrospectively that which rendered them possible. Experience is a given whose conditions must be archaeologically or genealogically traced to their historical a priori. Thus Foucaultʼs metacritique offers a kind of turning around of the famous Kantian ʻturnʼ; beginning with the subjects and objects of real experience and genealogically uncovering their historical conditions rather than beginning with the subject and deducing the universal, ahistorical conditions of possibility for experience. If Foucaultʼs archaeologies/genealogies are his explorations into this new immanent transcendental dimension, then what remains to be elucidated is the way in which Foucault carries out this transformation of Kant by explicating the relation between his nonanthropological form of the transcendental and the empirical forms conditioned by it – how the conditions are ʻcaught up in the very things they connectʼ – and whether and how this conception avoids the aporetic doubling that Foucault diagnosed in Kant and modern thought.
The interest of Beatrice Hanʼs reading of Foucault in her Foucaultʼs Critical Project25 lies in her searching examination of these issues. Indeed, Han argues that the central and unifying theme of Foucaultʼs work is the effort to develop a new historicized interpretation of the transcendental modifying Kantʼs project by attempting to detach it from his ʻanthropologicalʼ solution while retaining the form of the ʻcritical questionʼ. The ʻslalomʼ of Foucaultʼs constantly changing methodological frameworks are explained as various attempts to ﬁnd a working version of this historical transcendental that is coherent and consistent in its rejection of ʻman and his doublesʼ. The Foucauldian a priori, Han says, ʻis given in history, … transforms itself with it, and … nevertheless lies beyond it in deﬁning the conditions of possibility, themselves variable, from which the knowledge of an epoch can and must form itselfʼ.  Han tries to show how this deﬁnition is variously transposed, reworked and refocused as Foucaultʼs thought responds to the internal demands and aporia of holding the historical and the transcendental together and apart. Like Habermas, Han argues that Foucault is ultimately unsuccessful in these attempts and falls back into a form of the ʻdoublingʼ or reductive oscillation between the transcendental and the historical that Foucault himself had uncovered as one of the characteristic problems of modern thought.
In view of Hanʼs thesis, perhaps the most illuminating contrast will be with Gilles Deleuzeʼs reading in his little book Foucault.  Deleuzeʼs book on his friend is important here because it explicitly describes Foucaultʼs work as a unique sort of neo-Kantianism that attempts to seek out the historical a priori conditions of experience, of what makes something visible or readable, sayable or seeable, and so on. Thus, a ʻstatementʼ for Deleuze is precisely not to be confused with propositions or phrases since it is the condition of propositions or phrases. Equally, the ʻvisibleʼ in Foucault, according to Deleuze, is not what can be seen but the condition of what can be seen. However, what really sets Deleuzeʼs analysis apart is that he thinks that Foucaultʼs ʻneo-Kantianismʼ is precisely the problem of the ʻmutual presuppositionʼ, reciprocal determination and ʻcoadaptationʼ of these two forms, of seeing and saying, light and language – without one being reducible to or simply collapsing into the other. Deleuze articulates a ʻtransposition of the transcendental themeʼ and a logic of thought in Foucault that directly challenges the theses of Gutting, Habermas and Han. In other words, Deleuze offers an account of the a priori in Foucault that engages the problems diagnosed by these readers yet displaces the logic of reduction and the issues of internal consistency and methodology that they identify.
Han and Deleuze thus agree that Foucault develops a modiﬁed transcendental project, but they disagree over the extent, nature and coherence of that modiﬁcation. Against Gutting, both Han and Deleuze agree that Foucaultʼs emphasis on conditions enables him to construct a completely new philosophical relation to history: for Han a ʻmiddle pathʼ between idealism and materialism; for Deleuze a philosophical experiment with the real beyond history. The primary question that divides them here is precisely the nature of the a priori conditions and their relation with the conditioned. For Han the stakes of Foucaultʼs philosophical project lie precisely in whether he can hold the critical space between the transcendental and the historical open without one collapsing into the other. Initially, it is on archaeology, ʻsister and rival of phenomenologyʼ  Han says, that Foucault will confer the task of ﬁnding this non-anthropological version of the historical transcendental. Thus I will focus here only on parts of Hanʼs analyses of archaeology and contrast them brieﬂy with Deleuze.
Han ﬁnds differing archaeological phases of the historical a priori in Foucault corresponding to the early archaeological texts. In The Birth of the Clinic Han concludes that Foucault ends up with a ʻconfused phenomenologyʼ  generated by attempting to integrate the implicit presuppositions of a Merleau-Ponty style of phenomenology into Foucaultʼs nascent archaeological framework. For Han, Foucault may have been inﬂuenced by Merleau-Ponty because of his constant references to concepts like the ʻgazeʼ, ʻperceptionʼ, the ʻvisibleʼ, and so on, and the effort to identify historical variations of the structures of perception in a given domain. However, the fact that Foucault does not have a theory of ʻoneʼs own bodyʼ (le corps propre) and refers the relation between seeing and saying to an a priori space anterior to perception rules out this possibility. Thus, The Birth of the Clinic for Han ʻﬁnds itself without any real theoretical supportʼ,  caught in the confusion between the transcendental and the empirical.
In Deleuzeʼs reading, by contrast, Foucaultʼs repeatedly stated efforts to move beyond any phenomenological framework are developed by uncovering an a priori visibility that cannot be reduced to the acts of a seeing subject or the data of a visual meaning for a body–subject. Thus, in The Birth of the Clinic Foucault says the a priori is ʻanteriorʼ to all perceptions, ʻgoverning them from afarʼ.  There are moments for Deleuze in The Birth of the Clinic when Foucault invokes a ʻmedical gazeʼ that appears to rely on a unitary subject whose position never changes with respect to objects. However, Deleuze argues that the ʻarchaeology of the gazeʼ refers predominantly to an anonymous ʻvirtualʼ space or ʻabsolute visibilityʼ outside of the gaze, which is not deﬁned by sight but rather by ʻmultisensorial complexesʼ  that occupy a depth space. This precedes perceptual contents and makes them possible. Equally, Deleuze ﬁnds a separate sphere of the articulable in The Birth of the Clinic that conditions what is said such that the ʻnon-relationʼ between the articulable and the visible is clear even if the priority of the articulable is not sufﬁciently emphasized. Foucaultʼs denunciation of the subtitle of The Birth of the Clinic amounts, for Deleuze, then, not only to an assertion of the primacy of the articulable over the visible but also to a block on any phenomenological recuperation of the ʻgazeʼ, regardless of whether that philosophy is committed to a phenomenology of the subject or the body. Indeed for Deleuze, although Foucault may have found inspiration in the late Merleau-Ponty, his major archaeological achievement consists in this: ʻthe conversion of phenomenology into epistemologyʼ  where ʻknowledgeʼ (savoir) is understood as the ʻnon-relationʼ between seeing and speaking, an irreducible disjunction and doubling where each form has its own objects and subjects. (For example, there is no ʻsingleʼ object madness that a consciousness could direct itself towards. Madness is seen in different ways just as it is articulated in different ways from one period to the next and even in different stages of a period.) In the next phase of the a priori, Han argues, The Order of Things abandons all reference to perception and, in a way that bears similarities with Deleuze, reconstructs the a priori as a relation between the separate forms of language and being, each with their own autonomous and independent ontological modes of existence, that more or less ʻcorrespondʼ to produce differing historical relations between words and things. This ʻhidden metaphysicsʼ  is ʻironicalʼ for Han not because Foucault is interested in what precedes and makes ʻwords and thingsʼ possible, as it was for Deleuze, but because The Archaeology of Knowledge will disavow the existence of these separate ontological regions in favour of a ʻnominalismʼ that ﬁnds ʻobjectsʼ constituted in discourse rather than an independent zone of things. Where Deleuze sees a conversion of phenomenology into epistemology in Foucault, Han sees a conversion of ontological or metaphysical realism into a nominalist or discursive idealism. Thus, from The Archaeology of Knowledge on, Han posits another phase in the archaeological endeavour to ﬁnd the conditions for the possibility of knowledge as well as a decisive break in Foucaultʼs thought.
For Han, in this ﬁnal archaeological phase the status of the rules of the a priori in The Archaeology of Knowledge, and the speciﬁc kind of determination they exert, remain problematic. After a discussion of the prescriptive and descriptive nature of the rules, Han, following Dreyfus and Rabinow, argues that for Foucault the rules must be descriptive (they canʼt be prescriptive because Foucault is committed to the ʻneutralityʼ of archaeology and he denies that his a priori operates causally), and yet, as Han points out, Foucault argues that the a priori ʻmakes possible and governsʼ. So Han concludes that ʻFoucault ends up in the difﬁcult position of claiming for the historical a priori an efﬁcacy which is excluded by archaeologyʼs very theoretical premises, hence the strange notion of regularities which regulate themselvesʼ.  For Han this repeats the empirico-transcendental confusion evidenced in the earlier archaeological versions of the a priori and risks falling into the ʻanthropological sleepʼ that Foucault had himself warned us of.
However, for Deleuze this is to treat statements as if they were formed by rules operating on a transcendent level whose status is constant in relation to a homogenous system. For Deleuze, on the contrary, the rules of the discursive formation are immanent and found on the same level as the discursive, but that level is shifting and in continuous variation, operating neither laterally nor vertically but transversally. Thus ʻstatements of a discursive formation move from description to observation, calculation, institution and prescription, and use several systems or languages in the process.ʼ  In effect, the regularity of statements is self-regulating for Deleuze since the formation of statements is governed by rules of inherent change or variation that are neither exclusively formal nor purely extrinsic – that is, determined by social practices. For Deleuze, Foucaultʼs conditions are concerned with real and not possible experience; they are immanent to the ʻobjectʼ and the historical and are therefore essentially ʻrareʼ or limited according to the formation in question. According to Deleuze ʻthe conditions are never more general than the conditioned element and gain their value from their particular historical status. The conditions therefore are not “apodictic” but “problematic”.… What in fact they present is the way in which the problem appears in a particular historical formation.ʼ  Thus Deleuze deﬁnes the rules or conditions of the discursive formation as functioning according to the logic of the ʻmultiplicityʼ, and, if the Archaeology does signal a decisive break in Foucault, for Deleuze it is because ʻit represents the most decisive step yet taken in the theory-practice of multiplicitiesʼ.  Thus, Foucaultʼs work is an experiment with multiplicities, ʻa pragmatics of the multipleʼ.  In addition, the Archaeology for Deleuze registers the completion of the detachment from phenomenology ʻalways alreadyʼ under way in the previous books; an emphasis on the primacy of the articulable (now the ʻstatementʼ or the ʻdiscursiveʼ) over the visible (now designated as the ʻnon-discursiveʼ) which was not sufﬁciently ﬂagged in the preceding books. Rather than a move away from any ʻhidden metaphysicsʼ the Archaeology signals the emergence of something like a ʻﬂat ontologyʼ (Foucault will later say ʻhistorical ontologyʼ) that develops out of and transforms the ʻdepthʼ or ʻverticalʼ analysis of ʻBeingʼ found in The Order of Things. Thus, the ʻbeing of languageʼ of The Order of Things becomes ʻthere is languageʼ in The Archaeology of Knowledge, but in either case, for Deleuze, ʻone speaksʼ in an anonymous murmur. Thus The Archaeology of Knowledge is both an experiment with the autonomy and self-regulation of the articulable as an independent, a priori, anonymous multiplicity (which is not to be confused with words, phrases or propositions) and an exploration of the metaphysical topology of its surfaces – ʻthe positivity of the dictumʼ  – that are neither visible nor hidden. Behind the curtain there was nothing to see, but all the more important each time to describe the complex folds of the curtain. For Deleuze, in fact, this is Foucaultʼs most important historical principle. Deleuze has Foucault experimentally drawing out a sort of immanent ʻtransversalʼ space that cuts across traditional unities, groupings, disciplines, and so on, a self-organizing transcendental space that has ʻno need whatsoever of unity to form a systemʼ.  Learning to reach what Deleuze calls the ʻextractive conditionsʼ of this space required both a critical development of what was there all along and the creative construction of new concepts. So, rather than a strict discourse on method, Foucaultʼs The Archaeology of Knowledge becomes the creative ʻpoem of his previous worksʼ. 
For Habermas and Han, Foucaultʼs answer to the critical question swings from the transcendental to the historical and back again in an unstable reproduction of the same and a repetition of the ʻdoublesʼ that mirrors the ʻanalytic of ﬁnitudeʼ. In contrast Deleuze sees in Foucaultʼs transposition of the transcendental a vital grappling with the history and becoming of thought, a creative continuity that develops more like a volcanic chain, moving seismically from one crisis to another, revealing a deeper consistency and coherence that cannot be easily measured from the surface. Although the stakes of Foucaultʼs critical project as each of our authors understands it could be seen as at points compatible, they fundamentally disagree over how one evaluates the philosophical apparatus that underpins it. Han, for example, holds Foucaultʼs work on the historical a priori in The Archaeology of Knowledge accountable to a strict logic of ʻexclusive disjunctionʼ where the rules, which must be either descriptive or prescriptive, inhabit a transcendent space clearly delimiting them from the empiricities they govern. Habermas ﬁnds similar logical problems in the tensions between transcendental and empirical elements in Foucaultʼs genealogical work. Deleuze, however, ﬁnds in Foucault an immanent transcendental that operates as an ʻinclusive disjunctionʼ accounting for itself and its ʻobjectʼ according to the logic of the multiplicity. Here statements and their spaces of dispersion merge at the level of the rules of their formation, tracing out lines of inherent variation. Habermas, Han and Deleuze agree that in Foucault the relation between condition and conditioned is (or ought to be) radically disjunctive, heterogeneous and differential, but they disagree over the nature of the disjunction or difference involved (exclusive or inclusive, external or internal difference) and whether this is coherently maintained.
What ultimately divides these readings is, in Deleuzeʼs idiom, the logic or ʻimage of thoughtʼ upon which their respective interpretations are premissed. At stake here, then, are not only differing understandings of the Kantian legacy and interpretations of the function and nature of the transcendental, but the commitments one has to the very image of philosophical and critical thought, of how one should proceed in thought, of how we should be ʻdoing philosophyʼ. For Habermas, for example, although critique would require the rejection of the Kantian role of philosophy as ʻjudgeʼ, critique cannot abandon judgement itself. A tribunal or court is still necessary and would proceed in accordance with the priority of the concepts of communicative action, consensus and the principles of procedural rationality. For Deleuze, by contrast, the ʻjudgementsʼ of philosophy can only be afﬁrmed through ʻcriteriaʼ governed by creativity. On this view, philosophy begins in the ʻmiddleʼ, when we are provoked and compelled to think – to take the ʻwitches rideʼ – leaving behind any external ʻfoundationʼ or ʻgroundʼ. And when we are forced to think philosophically we do not reﬂect, contemplate or communicate. Rather, philosophical thought proceeds through the creation of concepts that respond to problems that change. In the creation of the concept – if it is a good concept – one ʻcounter-actualizesʼ the problem, changing the ʻspace of possibilitiesʼ through which we think about the problem and live with it. For Habermas and Han, however, philosophical discourse requires a foundation. To be sure, they do not mean a metaphysical ground or ﬁrst principle but a foundation deﬁned, at least in part, in terms of criteria derived from rationalized principles of consistency and coherence. For Deleuze, however, Habermasʼs and Hanʼs understanding of these terms still amounts to the application of external criteria, of principles of transcendence or juridical concepts to a body of work whose logic of development doesnʼt conform to their conception of a consistent or stable rational system. For Deleuze, if there is a foundation in Foucault it is immanent or self-founding, driven by a rhythm of thought far from equilibrium. There is no direct ground beneath our feet but only an indirect conditioning by the transcendental condition of the historically given. Thinking in the absence of transcendent foundations is, as Foucault says, a ʻperilous actʼ. For Deleuze, Foucaultʼs work amounts not only to an encounter with the ʻhistory of thoughtʼ; it also involves a perilous experiment with its becoming in which thought thinks its own history, but in order to free itself from what it thinks and be able ﬁnally to think otherwise. For Deleuze the application of external criteria from outside the work will not understand the nature of this conditioning and the experimental becoming at its heart – will not do justice to the dangers and passions of this work of thought as a critical and creative project of thinking-otherwise. How, then, might we formulate a conception of the transcendental consistent with this approach to philosophy?
AN immanent transcendental
In order to address this question we can characterize the different responses to the role of Kant and the thought of the transcendental in Foucault in several ways, but the differences can, for the sake of brevity, be reduced to one primary issue: whether and how Foucaultʼs replacement for the Kantian a priori subject – a non-anthropological transcendental – that is itself historically ʻformed and modiﬁedʼ in experience escapes his own critique of the ʻempiricotranscendental reduplicationʼ which purports to show ʻhow what is given in experience and what renders experience possible correspond to one another in an endless oscillationʼ.  The criticsʼ objections to what they take to be Foucaultʼs transcendental approach here are close to what we will call the ʻstandard objectionʼ to transcendental argumentation per se. The standard objection to transcendental arguments generally is that they are question-begging and circular, that their conclusions are presupposed in their premisses, or that their conclusions are simply premisses in disguise.  The accusation of ʻcontradictionʼ or circular reasoning is a claim that both Habermas and Han level at Foucault, underpinning their critical interpretations of his conception of the transcendental and supporting their claim that Foucaultʼs own conception of the transcendental suffers from aporias and doublings similar to those he himself identiﬁed within anthropological thought. Although a circular argument is not formally invalid, since if the premisses are true then the conclusions will be also, it will not be of use as a ʻproofʼ because if the conclusion merely repeats the truth contained in the premises then nothing will have been ʻprovenʼ true. Although neither Habermas nor Han relies on the idea that a transcendental argument must offer ʻproofsʼ, the issues that separate them from Deleuze depend upon how one understands the coherence of Foucaultʼs methodologies, the ʻinternal consistencyʼ of his argumentation, the need for a ʻfoundationʼ or ʻgroundʼ and its application to the relation between the transcendental and the empirical. Although neither Habermas nor Han is opposed to a modiﬁed transcendental, both appear to develop their criticisms in parallel with what they take to be Foucaultʼs own critique of the ʻempirico-transcendental duplicationʼ and the standard objection to transcendental arguments. What I want to question here, then, is the relevance and validity of the claimed parallel between the premisses of Foucaultʼs own critiques, for example his critique of anthropology in The Order of Things, and the criticisms levelled at him by Habermas and Han. Foucaultʼs objections to the ʻempirico-transcendental duplicationʼ cannot be viewed as exclusively premised upon or derived from notions of logical contradiction, coherence and consistency, nor should Foucaultʼs methods be seen as aiming at breaking free from or avoiding the process of doubling.  Rather, I want to suggest that Foucaultʼs objections to the analytic of ﬁnitude are indeed based on the form of circling, doubling or duplication and the type of content this doubling presupposes and permits, but not doubling, folding or duplication as such. The doubles that constitute the analytic of ﬁnitude are all determined ʻanthropologicallyʼ by the form of identity in their concepts or representations so that their content is a repetition of the same. This ʻunveiling of the Sameʼ or the identical involves a dialectical relation or doubling where the content that is repeated remains within the form of (identity of) the concept incorporating otherness, difference or distance. In a crucial passage in The Order of Things Foucault says:
From one end of experience to the other, ﬁnitude answers itself; it is the identity and the difference of the positivities, and of their foundation, within the ﬁgure of the Same. It is apparent how modern reﬂection, as soon as the ﬁrst shoot of the analytic appears, … moves towards a certain thought of the Same – in which Difference is the same thing as Identity. 
Foucaultʼs objections here are to that form of the circle or doubling – captured in the play of identity and difference within the positivities and their foundation – governed by the principle of the ʻSameʼ that makes all difference correspond to identity. In its anthropological conﬁguration the transcendental syntheses performed by the ʻI thinkʼ are doubled by the empirical syntheses of a living, speaking, labouring individual since Manʼs ﬁnitude ʻanswers itselfʼ by referring his positive forms to the background and foundation of his own ﬁnitude as both a subject that knows and an object of knowledge. Thus, in Foucaultʼs critique of ʻmodern reﬂectionʼ, the principle that governs synthesis and the reproduction of the doubles in the synthesis is an external principle or condition of the Same where the transcendental repeats the empirical. I want to suggest that we can ﬁnd a more afﬁrmative, differential principle of synthesis in Foucault, a non-identical sense of circularity, doubling or folding positively recognized by Deleuze. However, the nature of this doubling, the way in which the ʻtranscendentalʼ elements in Foucault are afﬁrmed and internally ʻcaught up in what they connectʼ without being governed by the ﬁgure of the Same, is the very aporetic condition of what Foucaultʼs thought is all about – indeed the very basis of the ʻcritical freedomʼ implicit in his approach to philosophy – at least from the perspective of Deleuzeʼs reading.
The differences on this key point between various interpretations of Foucaultʼs transcendental can now be recast in terms of how we characterize the unity and separation in the relation between transcendental and empirical elements – what I referred to earlier in Deleuzean terms as a distinction between an exclusive and inclusive disjunction. Foucaultʼs transcendental does not seek a foundation for experience in something outside it or in some presuppositionless beginning but in a structure immanent with yet irreducible to the experience it generates. Foucaultʼs texts move within the circle of the ʻalready saidʼ, the circulation of ʻdiscoursesʼ, ʻpowersʼ and ʻknowledgesʼ through which experience is constituted. What is at issue is precisely the structure of a ʻregionʼ of experience (e.g. ʻmadnessʼ, ʻpunishmentʼ, etc.) with all of its interconnected elements and the shaping patterns or ʻsystems of thoughtʼ that condition it. Thus the ʻunityʼ of experience has the character of what we referred to earlier in Deleuzean terms as a ʻmultiplicityʼ that organizes the separate components together in a complex ʻinternalʼ articulation of a differentiated structure. In Foucaultʼs transcendental there is not one set of elements that ʻgroundsʼ another set of elements in a ʻgroundedʼ whole. Rather, the transcendental structure (the historical a priori) is not independent of the elements but fully dependent upon them just as the elements depend upon the relations between themselves and the structure. The structure and the elements are held together in such a way as to form an essentially indeterminate and open-ended multiplicity where the relations themselves (relations of force, knowledge, power, self, resistance, etc.) determine the distribution of elements, places, functions, and so on, in the ʻexperienceʼ being produced. Thus the elements double, fold or encircle the structure in a differential relation just as the structure differentiates itself from its own genetic elements. This whole relational structure of the immanent transcendental is put into play in Foucaultʼs texts through a dynamic differential temporality, historical processes that stratify and those that lead to our becoming in the ʻpresentʼ. Foucaultʼs archaeological and genealogical descriptions attempt to lay out the structure of the experience concerned (madness, illness, punishment, sexuality, etc.) in the light of this differential temporality, where what lies within the circle of experience is exposed and opened to the forces of the outside. If this conception of the double as differential is worked out ʻepistemologicallyʼ in the earlier texts and ʻstrategicallyʼ in the texts on power, it is in the ﬁnal texts when Foucault focuses the immanent transcendental on the interior experience of ʻsubjectivationʼ, the relation to oneself, that the concepts of the fold, the double and a differential circling are brought to the fore and take on a ʻcompletely new appearanceʼ while retaining their ontological importance.  In the late Foucault, experience, in its exposure to the forces of the outside, is now to be analysed in terms of the way these forces fold back upon themselves and affect themselves as the affect of self upon self, enabling the creation of ʻnew forms of subjectivityʼ. As Deleuze puts it:
the theme that has always haunted Foucault is that of the double. But the double is never the projection of the interior; on the contrary, it is an interiorization of the outside. It is not a redoubling of the One but a redoubling of the Other. It is not a reproduction of the same but a repetition of the different. 
Replacing the vicious circle of anthropological thought with its repetitions of the Same this conception of the folds and doublings of the immanent transcendental in Foucault becomes valid as a ʻdiagnosticʼ principle for us:
understood in this way, the diagnostic does not establish the facts of our identity by means of the interplay of distinctions. It establishes that we are difference, that our reason is the difference of forms of discourse, our history is the difference of times, that our selves are the difference of masks. 
The fold or double as difference becomes the nonanthropological or diagnostic principle of the internal genesis and ʻunityʼ of experience in Foucault, the ʻgroundless groundʼ of the immanent transcendental.
The immanent transcendental in Foucault develops out of resources provided by Kant.  Foucaultʼs transformation of these resources began very early on through a reading of the Anthropology, underwent further development in a number of works with his novel conceptions of the a priori, and continued through into the ﬁnal texts with his work on ʻproblematizationʼ, ʻgames of truthʼ and ʻsubjectivationʼ. I have argued that one condition for doing justice to Foucaultʼs critical project would involve carefully situating his thought within the rich, multiple and often conﬂicting trajectories and traditions of modern philosophy for which Kant and his critical project of transcendental philosophy stands as the principal ʻﬁgureheadʼ. Within these traditions the exploration of the transcendental pushes up against questions relating to the very limits, scope and nature of philosophy itself. I have suggested that a certain ʻdifferentialʼ and temporal conception of circling or doubling is necessarily bound up for Foucault not just with an immanent understanding of the transcendental and transcendental inquiry but with philosophical inquiry itself. The immanent, differential and historical nature of the transcendental in Foucault will distinguish his philosophical and critical approach from formal/logical procedures (structuralism, ʻanalyticʼ, etc.), phenomenology and hermeneutic inquiry. Indeed, for Foucault the opening of such a space would not only mark the ʻreturn of the beginning of philosophyʼ; it would be ʻnothing less and nothing more, than the unfolding of a space in which it is once more possible to thinkʼ. 
1. ^ F. Nietzsche, Beyond Good and Evil, trans. R.J. Hollingdale, Penguin, London, 1973, §289, p. 216.
2. ^ M. Foucault, The Order of Things (OT), Tavistock Publications, London, 1970, p. 321.
3. ^ G. Deleuze, Nietzsche and Philosophy, trans. H. Tomlinson, Athlone, London, 1983, p. 52.
4. ^ M. Foucault, The Archaeology of Knowledge (AK), trans.
A. Sheridan Smith, Routledge, London, 1972, p. 127.
5. ^ I. Kant, Critique of Pure Reason, trans. P. Guyer and A.Wood, Cambridge University Press, Cambridge, 1998, A158/B197.
6. ^ Ibid., A11–12/B25.
7. ^ There is a large body of literature – both ʻanalyticʼ and ʻcontinentalʼ – on this topic. A couple of representative texts are Robert Stern, ed., Transcendental Arguments: Problems and Prospects, Clarendon Press, Oxford, 1999; Hans-Johann Glock, ed., Strawson and Kant, Clarendon Press, Oxford, 2003.
8. ^ G. Gutting, Foucault: A Very Short Introduction, Oxford University Press, Oxford, 2005, p. 59.
9. ^ G. Gutting, ʻReview of Foucaultʼs Critical Project, B.
Hanʼ, Notre Dame Philosophical Reviews, 1 May 2003, at http://ndpr.nd.edu/review.cfm?id=1262.
10. ^ Gutting, Foucault, p. 66.
11. ^ Ibid., p. 66. This is a striking claim that Gutting does not appear to make in his earlier Michel Foucaultʼs Archaeology of Scientiﬁc Reason, Cambridge University Press, Cambridge, 1989.
12. ^ ʻLa Constitution dʼun transcendental historique dans La Phénomenologie de LʼEsprit de Hegelʼ, cited in D. Eribon, Michel Foucault, Flammarion, Paris, 1991, p. 47.
13. ^ Foucault, The Birth of the Clinic (BC), trans. A.M.
Sheridan Smith, Routledge, London, 1973, p. xv.
14. ^ Foucault, AK, p. 128.
15. ^ Ibid., p. 127.
16. ^ J. Habermas, The Philosophical Discourse of Modernity (hereafter PDM), trans. F.G. Lawrence, Polity Press,
Cambridge, 1987, pp. 253–4.
17. ^ Ibid., p. 256.
18. ^ This is why Foucault objected to being called a ʻstructuralistʼ, since structuralism tried to ʻevacuate the concept of the eventʼ, as Foucault put it in an interview in 1977.
See The Foucault Reader, ed. P. Rabinow, Penguin, London, 1986, p. 56.
19. ^ Habermas, PDM, p. 273.
20. ^ Ibid., p. 274
21. ^ Ibid., p. 256.
22. ^ Ibid., p. 256.
23. ^ Ibid., p. 276.
24. ^ I did not have space to include a discussion of Andrew Cutrofelloʼs excellent book Discipline and Critique, SUNY Press, New York, 1994, which attempts to work out a nonjuridical or metacritical version of Kantian ethics in a Foucauldian framework. Also see Amy Allenʼs ﬁne paper, ʻFoucault and Enlightenment: A Critical Reappraisalʼ, Constellations, vol. 10, no. 2, 2003.
25. ^ B. Han, Foucaultʼs Critical Project: Between the Transcendental and the Historical (hereafter FCP), Stanford University Press, Stanford, 2002.
26. ^ Ibid., p. 4.
27. ^ G. Deleuze, Foucault, Athlone, London, 1988.
28. ^ Han, FCP, p. 6.
29. ^ Ibid., p. 65.
30. ^ Ibid., p. 50.
31. ^ Foucault, BC, p. 5.
32. ^ Deleuze, Foucault, p. 59.
33. ^ Ibid., p. 109.
34. ^ Han, FCP, p. 65.
35. ^ Han, ʻReply to Gary Guttingʼ, pp. 5–6; available at: http://privatewww.essex.ac.uk/~beatrice/Gutting%20_answer_%202003–05.pdf.
36. ^ Deleuze, Foucault, p. 5.
37. ^ Ibid., p. 114.
38. ^ Ibid., p. 14.
39. ^ Ibid., p. 84.
40. ^ Ibid., p. 15.
41. ^ Deleuze, Difference and Repetition, trans. P. Patton,
Athlone, London, 1994, p. 182.
42. ^ Deleuze, Foucault, p. 18.
43. ^ Foucault, OT, p. 336.
44. ^ This is of course a claim originally laid at the door of Kant and critical philosophy. For a now classic version of this argument in the literature, see Stephan Korner, ʻThe Impossibility of Transcendental Deductionsʼ, The Monist 51, 1967, pp. 317–31.
45. ^ The original version of this argument is found in Dreyfus and Rabinowʼs Michel Foucault: Beyond Structuralism and Hermeneutics, Harvester Press, Brighton, 1986 (see Part 1, ch. 4) and is repeated by both Habermas and Han.
46. ^ Foucault, OT, p. 315.
47. ^ Deleuze, Foucault, p. 111.
48. ^ Ibid., pp. 97–8.
49. ^ Foucault, AK, p. 131.
50. ^ Foucault arguably found a trace of this nonanthropological conception of the transcendental in Kantʼs Anthropology, which he suggests repeats the a priori of the critique ʻin a temporal dimensionʼ. Time is no longer a non-empirical condition of synthesis as it is in the ﬁrst critique. Rather, time gnaws at the synthetic activity itself. One could argue that Foucault here uncovers a line of development, not carried forward by Kant, that he will himself take up and transform. See Foucault, Introduction à Lʼanthropologie de Kant, thèse complémentaire; available at www.generation-online.org/p/fpfoucault8.htm.
51. ^ Foucault, OT, p. 342.