You are here: Home / Synopses / Michel Foucault: ‘Réponse au Cercle d’épistémologie’

This project is funded by an Arts and Humanities Research Council (AHRC) research grant and is supported by the Centre for Research in Modern European Philosophy (CRMEP) and Kingston University's Faculty of Arts and Social Sciences.

AHRC logo

CRMEP logo

Synopsis of Michel Foucault, ‘Réponse au Cercle d’épistémologie’

[‘Response to the Circle of Epistemology’]

CpA 9.2:9–40

As its title suggests, Foucault wrote or at least framed this piece in response to a series of questions posed by the Cercle d’Épistémologie, as part of an issue of the Cahiers on the ‘Genealogy of the Sciences’ (CpA 9.1). In 1966-67, as Foucault’s biographer David Macey notes, Althusser and several of his students were enthusiastic about Foucault’s early ‘archaeological’ works on madness and medicine, and both ‘Etienne Balibar and Althusser believed that [Foucault’s] Les Mots et les choses [1966, translated into English as The Order of Things] would help to provide a general theory of ideology. It was increasingly possible to see Foucault and Althusser as part of the same theoretical project’; hence the growing interest in Foucault’s work among the editors, at the Ecole Normale, of the Cahiers Marxistes-Léninistes and then the Cahiers pour l’Analyse. It was in the ‘highly intellectualised, increasingly politicised’ milieu of the Cahiers, Macey continues, that ‘Foucault found an audience for his Archéologie du savoir. The philosophy of the concept once associated with Cavaillès and Canguilhem was beginning to find a new incarnation.’1 The contents of the piece were quickly reworked and published in modified form as a section of his 1969 L’Archéologie du savoir [The Archaeology of Knowledge].

Foucault’s first section is entitled ‘History and Discontinuity’2 and its opening remarks note a ‘curious intersection’ between two tendencies in contemporary historiography.3 On the one hand, many historians have been turning to the study of long periods of time, and attempting to formulate systems of large-scale periodisation. For them, the discontinuities of experience take place on a backdrop of continuity and large epochal ‘units’.4 On the other hand, the attention of another group of historians has been displaced from the vast units forming an ‘epoch’ or ‘century’ towards the study of phenomena of rupture. Gaston Bachelard, Georges Canguilhem and Martial Gueroult have each in their own way charted the ruptures and interruptions that occur across the history of the sciences. However, in the course of this intersection of tendencies, what has happened is that, rather than one tendency opting for continuity, and the other for discontinuity, the very notion of discontinuity has ‘changed in status’ (CpA 9.2:11; trans. 299). In classical historiography, discontinuity was always presented as ‘the stigma of temporal dispersion which it was the historian’s duty to suppress from history’ in order to reveal ‘the continuity of concatenations’ [enchaînements]. But it has now undergone a mutation and become ‘one of the fundamental elements of historical analysis’. Thus what the contemporary historian now ‘undertakes to discover is the limits of a process, the point of change of a curve, the reversal of a regulatory movement, the bounds of an oscillation, the threshold of a function, the emergence of a mechanism, the moment a circular causality is upset’. Discontinuity is no longer an obstacle, but a practice, involving the ‘regulated use of discontinuity for the analysis of temporal series’ (11/300).

History and historiography have often served as a ‘privileged shelter for consciousness’, reserving a place for the restoration of a greater subject. But today psychoanalysis, linguistics and ethnology have all ‘dispossessed the subject of the laws of its desire, the forms of its truth and speech, the rules of its action, and the systems of its mythical discourses’. We are witnessing the disappearance of ‘that form of history which was secretly … transferred to the synthetic activity of the subject’. History can no longer be declared to be ‘living and continuous’ (12/301).

In the second section, ‘The Field of Discursive Events’, Foucault explains that the definition and application of the concept of discontinuity in the history of ideas and the analysis of discourse demands the renunciation of certain notions. Anthropological categories and ‘unregulated syntheses’ must be abandoned. First, there is negative task of breaking free of the framework of continuity that is supported by teleological notions such as ‘tradition’, ‘influence’, and ‘development’ [développement]. Disciplinary categories (for instance, the division between ‘literature’ and ‘politics’) also can no longer be treated as normative, but rather must be minimally approached as ‘facts of discourse’ (14/303). Other ‘unities’ must be questioned, such as that of the ‘book’, ‘work’ and ‘author’ (the ‘immense swarm of verbal traces that an individual leaves around him at the moment of death’ cannot be contained by a proper name). Two ‘postulates’ must also be renounced. The first is the assumption that ‘it is never possible to find the irruption of a genuine event in the order of discourse; that beyond every apparent beginning, there is always a secret origin […], an indefinitely distant point, never present in any history. The point itself could only be its own emptiness; all beginnings from that point, could only be recommencements or occultations (strictly speaking both, in one and the same gesture)’ (16/305). Affirming this postulate leads to the historical analysis of discourse subordinating itself to a repetition of a primordial origin. This kind of view is linked to the second postulate, that ‘every manifest discourse rests on an “already said”’, that is at the same time a ‘never said’ or ‘non-said’. These two postulates ‘function to guarantee the infinite continuity of discourse and its secret presence to itself in the actions of an absence that is always one stage farther back’ and must be renounced. Instead, ‘each moment of discourse must be welcomed in its irruption as an event; in the punctuation where it appears’, and in its ‘temporal dispersion’.

Eliminating these ‘unregulated syntheses of discourse’ liberates the immense domain of ‘the set of all effective statements (spoken or written) in their dispersion as events and in the immediacy that is proper to each’. Prior to classification according to genre, author, etc, discursive material is ‘to be handled in its initial neutrality [as] a population of events in the space of discourse in general’. This requires the ‘project of a pure description of the facts of discourse’ (16/306), which is different from a linguistic analysis. A language is a finite system for the generation of possible statements, with an infinite number of performances. Discourse, by contrast, ‘is the always finite and temporally limited set of those linguistic sequences that have been formulated’ (17/307; trans. modified). Where linguistics inquires as to the rules by which a statement is construction, the description of discourse asks: how is it that this statement appeared, rather than some other one in its place? The aim should be to ‘grasp the statement in the narrowness and singularity of its event; to determine the condition of its existence, to fix its limits as accurately as possible’, and to show what it excludes.

Even if its consequences remain unapparent, ‘a statement is always an event that neither language nor meaning can completely exhaust’. It is enunciated in acts of writing or speech, but, as an ‘enunciative event’, maintains a residual existence in memory and in reproduced texts. Foucault suggests that these enunciative events are also related to ‘events that are not discursive in nature, but may be of a technical, practical, economic, social, political or other variety’ (18/308).

Foucault then defines ‘the series of rules which determine in a culture the appearance and disappearance of statements’ as an ‘archive’ [archive] (18/309). What Foucault calls ‘archaeology’ thus involves the exploration of the archive of statements in this specific, rarefied sense. Taking a retrospective look back at his previous books, History of Madness (1961), The Birth of the Clinic (1963) and The Order of Things (1966), Foucault suggests that this is what he has been doing all along. His task has always been to find a vantage point by locating the set of statements by virtue of which categories are constituted. In the History of Madness, for instance, the statement, everywhere present but never directly said as such is ‘confine the mad’. More recently, in The Order of Things, he has attempted to excavate the statements behind the rise of the ‘sciences of man’, or human sciences.

But then Foucault notes the reflexive predicament in which he now finds himself:

I should indeed acknowledge that such a project of description as I am trying at present to circumscribe is itself caught up in that region that I am trying, by way of an initial approach, to analyse, and that it runs the risk of being dissociated under the effect of analysis. I am investigating that strange and quite problematic configuration of human sciences to which my own discourse is tied. I am analysing the space in which I speak. I am laying myself open to undoing and recomposing that space which indicates to me the first indices of my discourse [… Therefore] everything I say could well have the effect of displacing the place from which I am saying it […]. So I must now acknowledge that I can myself no longer speak from where I showed [these discourses] to be speaking without saying it, but instead only from that difference, that infinitesimal discontinuity which my discourse has already left in its wake (21/311).

These problems of reflexivity will be the focus of the ‘Nouvelles Questions’ put by the Cercle d’Épistémologie in the article that follows (CpA 9.3).

In the third section, ‘Discursive Formations and Positivities’, Foucault presents a new theory of discourse. His guiding question is: how can statements form a set [ensemble]? How can a set of statements be individuated? Foucault presents four initial hypotheses, outlining how he came to abandon them for a more complex position.

His first initial hypothesis was that certain statements could form a set by referring to one and the same object. Statements about madness, for instance, may come from legal or medical domains, but, by this hypothesis, they would still concern the same object. However, this cannot work. The object is constituted by what is said in the statements. What is said about madness may be said because of the function it plays in a particular discourse. Individuating a set of statements, says Foucault, must instead mean describing their ‘system of distribution’. This is their only ‘referential’; there is no ‘object’ outside discourse.

The second hypothesis Foucault initially entertained about the individuation of statements concerned the type of enunciation used. Perhaps nineteenth century medical discourse constituted a certain ‘style’ or constant form of enunciation? Medicine, for instance, seemed to be in the process of ‘formalising itself as a series of descriptive statements’. But this early hypothesis too had to be abandoned: ‘I had to admit that clinical medicine was just as much a set of political prescriptions, economic decisions, institutional settlements and educational models as it was a set of descriptions’ (23/314). A set of statements can be individuated by referring to their system of distribution: ‘the rule of formation of these statements in their heterogeneity, in the very impossibility of their integration into a single syntactic chain, is what I shall term enunciative divergence [l’écart énonciatif]’ (24-5/315).

A third possible criterion for the individuation of statements might involve an appeal to permanent and internally consistent concepts. For instance, the concept of judgment would be defined as ‘the general and normative form of every sentence’, while the concepts of subject, verb and attribute could be seen to have undergone development. But ‘one has to admit that new concepts appear’, some of which may indeed be conceptually derivable, ‘but others of which are heterogeneous, and some even incompatible with them’. Beauzée’s notion of natural syntactic order, for instance, introduced at the beginning of the eighteenth century, cannot be derived from Port-Royal logic even if it is consistent with it. Instead, the ‘common system’ beneath the various concepts of Classical grammar must be constituted by a ‘set of rules of formation of concepts’.

Foucault suggests that any set of such rules can be divided into four ‘formative schemata of concepts’ (26-7/317-18):

  1. Attribution: the set of rules that relates statements to each other.
  2. Articulation: the set of rules that permit a ‘description of the relations between the different signifying elements of the sentence, and the different elements of what is represented by these signs’.
  3. Designation: the set of rules that govern the emergence of concepts charged with expressive value.
  4. Derivation: the set of rules that allow for the formation of bodies of discourse.

These four formative schemata suffice to generate a ‘theoretical network’ [réseau théorique] for the theory of discourse. What gives the discourse individual and independent existence is ‘the system of points of choice which it offers from a field of given objects, from a determinate enunciative scale’ (29/320). Behind every option taken there is a ‘field of strategic possibilities’. Foucault claims that these four criteria cover all the aspects of discourse: rules of formations of (1) objects, (2) syntactic types, (3) semantic elements and (4) operational eventualities. Where one referential, one divergence in enunciation, one theoretical network and one field of strategic possibilities are isolated, then we arrive at a discursive formation. It is this fully ‘positive’ four-level system that underlies discourse.

In the fourth section, on ‘Knowledge [Savoir]’, Foucault claims his method is both empirical, in that it is articulated through precise enquiries into historical material, and critical, ‘since it concerned the place from which I was posing the question, the region which situated it’. Making an implicit reference to his own identification of a ‘transcendental-empirical doublet’ in Kantianism and phenomenological thought5 Foucault asks himself whether his own works end up also relying despite themselves on ideologically given premises and merely deducing their conditions.

In order to handle this question, he proceeds to distinguish his conception of a theory of discourse from epistemology. In the background is the conception of epistemology proposed by Bachelard and taken up by Althusser and the contributors to the Cahiers pour l’Analyse. Foucault contends that Bachelardian theories of epistemic change need support from a theory of discursive change. Clinical medicine, for instance, ‘is an enunciative set [that is] both theoretical and practical, description and institutional, analytical and prescriptive, made up of inferences as well as decisions, of assertions as well as degrees’ (32/324). These discursive formations are ‘neither current sciences in gestation’, nor ‘sciences’ no longer recognised as such. ‘They are unities of a different kind and on a different level from what is called today (or was once called) a “science”’. In order to characterize them, the distinction between scientific and non-scientific is not pertinent: they are ‘epistemologically neutral’. The set that is manifested ‘in the unity of a discursive formation is called a knowledge [savoir]’. This knowledge is not the sum of scientific knowledges [connaissances]. The latter are true or false, but this distinction does not apply to knowledge as savoir, which is grounded in a positive system that generates the formation of statements. ‘One cannot ask that their description be equivalent to a history of knowledges, a genesis of rationality, or the epistemology of science’ (33/325). Formal criteria may govern the question of a scientificity of science, but these criteria ‘can never account for its factual existence, that is, its historical appearance, the events, episodes, obstacles […] etc. that stamp its actual destiny’ (33/325). The conditions of the appearance of a science are determined in the element of savoir.

There are thus two kinds of ‘conditions of possibility’ of a science. First, ‘the formal and semantic rules required for a statement to belong to a science’. These conditions of scientificity are ‘internal to scientific discourse in general and cannot be defined other than through it’. But on the other hand, there is the possibility of a science in its historical existence. This is conditioned by the discursive formations, which have their own kind of consistency. ‘In a word, knowledge is not science in the successive displacement of its internal structures; it is the field of its actual history’.

In the final section, ‘Several Remarks’, Foucault comments on two types of ‘extrapolation’ or one-sided abstraction in accounting for the history of a discourse. On the one hand, in ‘epistemological extrapolation’, a science is ‘given the responsibility of explaining its own historicity’; on the other hand there are ‘genetic extrapolations’ which make the ‘genetic fallacy’ of reducing systems to their historical determinations. Both abstractions involve a type of miscognition (35/327; trans. modified).

To show what he means by epistemological extrapolation he takes the example of the history of mathematics. This is indeed a science of such a type that ‘every episode of its historical development can be taken up again in the interior of its deductive system; its history can, in fact, be described as a movement of lateral extension, then of repetition and generalization at a higher level, such that each moment appears either as a special region or a definite degree of formalisation’ (36/327-28). Foucault mentions as an example the generalisation of Cartesian algebra by Joseph-Louis Lagrange, Niels Henrik Abel, and Evariste Galois. But an excessive attention to formalisation misses the level of discursive formations at work beneath internal epistemological breaks and ruptures. Implicitly addressing the epistemology of mathematics inaugurated by Cavaillès, Foucault says that mathematics should not serve as a privileged model but rather as a ‘limit-case’. ‘All sciences (even ones as highly formalised as mathematics) presuppose a space of historicity that does not coincide with the interaction of its forms’ (36/328). Although one should not go so far as to encourage a ‘doxological illusion’, based on the actuality of opinions (rather than analysis of the choices and ‘strategic possibilities of conceptual games’), it is equally necessary to avoid the ‘formalising illusion’:

One can always establish the semantics and syntax of a scientific discourse. But it is necessary to protect oneself from what one would call the formalising illusion [illusion formatrice]: that is, from imagining that these laws of construction are at the same time and with full title the conditions of existence […]. The formalist illusion elides knowledge (the theoretical network and enunciative repartition) as the site and law of formation of concepts and propositions (37-38/330; trans. modified).

With regard to the problem of ‘genetic extrapolation’, Foucault takes psychology as an example. ‘If there has only be one psychology since Gustav Fechner, if there has only been one sociology since Auguste Comte […] it is not insofar as it possible to assign a single epistemological structure […] to so many diverse discourses; it is insofar as sociology and psychology have at each moment located their discourse in a historical field they themselves had traversed in the critical mode of confirmation or invalidation’ (36/328). The misrecognition involved in the ‘genetic extrapolation’ is that it never reaches the level of formalisation. For these reasons the ‘intermediate sciences’ - such as biology, physiology, political economy, linguistics and philology - ought rather to provide the models for the integration of science and savoir.

Finally, Foucault warns of the ‘illusion of experience’ (38/330): there are no regions or domains of things that ‘present themselves spontaneously to an activity of idealization’. Science does not work on an immediate, concrete and lived experience. ‘The referent itself’ does not ‘contain the law of the scientific object’. Nevertheless, that does not mean one should go to the other extreme and ‘imagine that science is established by an act of rupture or decision’. Rather, each scientific discourse has its own ‘relations of reference and separation’ to its own history.

The archaeology of savoir moves us beyond the ‘great historico-transcendental thematic that ran through the nineteenth century’ (39/332). Savoir ‘is free of any constituent activity, disengaged from any reference to an origin or to a historico-transcendental teleology, detached from any reliance upon a foundational subjectivity’. Thus Foucault overturns the exclusion characteristic of the nineteenth century, which expelled the ‘anonymous discontinuity of knowledge’ from discourse, casting it out into the ‘unthinkable’ (40/332-33).

References to this text in other articles in the Cahiers pour l’Analyse:

  • Cercle d’épistémologie, ‘Nouvelles questions’. CpA 9.3.

English translation:

  • ‘On the Archaeology of the Sciences: Response to the Epistemology Circle’, translated in Theoretical Practice 3-4 (Autumn 1971). Revised translation in Foucault, The Essential Works, vol. 1: Aesthetics, Method, and Epistemology, ed. James D. Faubion. London: Penguin, 1998.


  • Foucault, Michel. Histoire de la folie à l’âge classique [1961]. Paris: Gallimard, 1972. History of Madness, trans. Jonathan Murphy and Jean Khalfa. London: Routledge, 2006.
  • ---. Naissance de la clinique. Paris: PUF, 1963. The Birth of the Clinic, trans. Alan Sheridan. London: Tavistock Publications, 1973.
  • ---. Les Mots et les choses. Paris: Gallimard, 1966. The Order of Things, trans. Alan Sheridan. New York: Random House, 1970.
  • ---. ‘Les Mots et les choses’. Interview with Raymond Bellour. Les Lettres françaises, 31 March 1966. ‘The Order of Things’, trans. John Johnston. In Foucault Live: Collected Interviews, 1961-1984. New York: Semiotext(e), 1996.
  • ---. ‘Le Discours de l’histoire’. Interview with Raymond Bellour, Les Lettres françaises, 15 June 1967. ‘The Discourse of History’, trans. John Johnston in Foucault Live.
  • ---. ‘Réponse à une question’, Esprit 371 (May 1968). ‘History, Discourse and Discontinuity’, trans. Anthony Nazzaro. Salmagundi 20 (Summer-Fall 1972), reprinted in Foucault Live.
  • ---. L’Archéologie du savoir. Paris: Gallimard, 1969. The Archaeology of Knowledge, trans. Alan Sheridan. New York: Pantheon, 1982.
  • ---. ‘L’Archéologie du savoir’. Interview with Jean-Jacques Brochier. Quinzaine Littéraire, April-May 1969. ‘The Archaeology of Knowledge’, trans. John Johnston. In Foucault Live.
  • ---. L’Ordre du discours. Paris: Gallimard, 1971. ‘The Discourse on Language’. In Foucault, The Archaeology of Knowledge.
  • ---. Dits et écrits, 1954-1984, I, eds. Daniel Defert and François Ewald. Paris: Gallimard, 1994.
  • Gutting, Gary. Michel Foucault’s Archaeology of Scientific Reason: Science and the History of Reason. Cambridge: Cambridge University Press, 1989.
  • Macey, David. The Lives of Michel Foucault. London: Vintage, 1994.


1. David Macey, The Lives of Michel Foucault, 197-198. Daniel Defert’s chronological outline of Foucault’s activities suggests, however, that Foucault had already finished a first draft of L’Archéologie du savoir in Spring 1967 (‘Chronologie’, in Dits et écrits, I), so it remains uncertain how far Foucault’s text was written as a direct response to the questions asked by the Cercle d’Épistémologie.

2. The subtitle is omitted from the text in Dits et écrits and from the English translation of Foucault, The Essential Works, vol. 1.

3. The ‘curious intersection’ may also refer to the intersection of Foucault’s own research with that of the Cercle d’Épistémologie. The phrase is cut from the opening paragraph of the Archaeology of Knowledge.

4. Foucault does not cite names here, but in a contemporary interview on The Archaeology of Knowledge, he specifies ‘the historians of the Annales school, Marc Bloch, Lucien Febvre, Ferdinand Braudel, have tried to enlarge the periodisations that historians usually make’ (Foucault, ‘The Archaeology of Knowledge’, Foucault Live, 58.

5. Foucault holds the ‘transcendental-empirical doublet’ – whereby prejudices about empirical objects inform conceptions of the transcendental conditions of those objects in a viciously circular manner – to be a hallmark of the post-Kantian and phenomenological ‘analytic of finitude’, as well as of the contemporary notion of the ‘human’ sciences (The Order of Things, 318-21).