[The collection from which this excerpt is taken was originally published in 1964 under the title Essais critiques.]

[In Folie et déraison: Histoire de la folie à l'âge classique, Foucault] has not written the history of madness, as he says, in a style of positivity: from the start he has refused to consider madness as a nosographic reality which has always existed and to which the scientific approach has merely varied from century to century. Indeed Foucault never defines madness; madness is not the object of knowledge, whose history must be rediscovered; one might say instead that madness is nothing but this knowledge itself: madness is not a disease, it is a variable and perhaps heterogeneous meaning, according to the period; Foucault never treats madness except as a functional reality: for him it is the pure function of a couple formed by reason and unreason, observer and observed. And the observer (the man of reason) has no objective privilege over the observed (the madman). It would thus be futile to try to find the modern names for dementia under its old names.

Here is a first shock to our intellectual habits; Foucault's method partakes at once of an extreme scientific discretion and of an extreme distance with regard to "science"; for on the one hand, nothing happens in the book which is not nominally given by documents of the period;… and on the other hand, the historian here studies an object whose objective character he deliberately puts in parentheses; not only does he describe collective representations (still rarely done in history), but he even claims that without being mendacious these representations somehow exhaust their object; we cannot reach madness outside the notions of men of reason (which does not mean, moreover, that these notions are illusory); it is therefore neither on the side of (scientific) reality nor on the side of the (mythic) image that we shall find the historical reality of madness: it is on the level of the interconstituent dialogue of reason and unreason, though we must keep in mind that this dialogue is faked: it involves a great silence, that of the mad: for the mad possess no metalanguage in which to speak of reason. In short, Michel Foucault refuses to constitute madness either as a medical object or as a collective hallucination; his method is neither positivist nor mythological; he does not even shift, strictly speaking, the reality of madness from its nosographic content to the pure representation men have made of it; he keeps identifying the reality of madness with a reality at once extensive and homogeneous with madness: the couple formed by reason and unreason. Now this shift has important consequences, both historically and epistemologically.

The history of madness as a medical phenomenon had to be nosographic: a simple chapter in the general—and triumphant—history of medicine. The history of reason/unreason, on the other hand, is a complete history which brings into play all the data of a specific historical society; paradoxically, this "immaterial" history immediately satisfies our modern insistence on a total history, which materialistic historians or ideologists appeal to without always managing to honor it. For the constitutive observation of madness by men of reason is very quickly seen to be a simple element of their praxis: the fate of the mad is closely linked to the society's needs with regard to labor, to the economy as a whole; this link is not necessarily causal, in the crude sense of the word: simultaneous with these needs appear representations which establish them in nature, and among these representations, which for a long time were moral ones, there is the image of madness; the history of madness always follows a history of the ideas of labor, of poverty, of idleness, and of unproductivity. Michel Foucault has taken great care to describe simultaneously the images of madness and the economic conditions within the same society; this is doubtless in the best materialist tradition; but where this tradition is—happily—transcended is in the fact that madness is never offered as an effect: men produce in the one impulse both solutions and signs; economic accidents (unemployment, for example, and its various remedies) immediately take their place in a structure of significations, a structure which may well pre-exist them; we cannot say that the needs create values, that unemployment creates the image of labor-as-punishment: rather the two meet as the true units of a vast system of signifying relations: this is what Foucault's analyses of classical society unceasingly suggest: the link which unites the foundation of the Hôpital Général to the economic crisis of Europe (beginning of the seventeenth century), or on the contrary the link which unites the disappearance of confinement to the more modern sentiment that massive internment cannot solve the new problems of unemployment (end of the eighteenth century)—these links are essentially signifying links.

This is why the history described by Michel Foucault is a structural history…. Without ever breaking the thread of a diachronic narrative, Foucault reveals, for each period, what we should elsewhere call sense units, whose combination defines this period and whose translation traces the very movement of history; animality, knowledge, vice, idleness, sexuality, blasphemy, libertinage—these historical components of the demential image thus form signifying complexes, according to a kind of historical syntax which varies from epoch to epoch…. A more formalistic mind might have exploited more intensely the discovery of these sense units: in the notion of structure to which he appeals, Foucault emphasizes the notion of functional totality more than that of component units; but this is a matter of discourse; the meaning of the procedure is the same, whether we attempt a history of madness (as Foucault has done) or a syntax of madness (as we can imagine it): the question is still one of making forms and contents vary simultaneously. (pp. 164-67)

[What] we can infer from Foucault's analyses (and this is the second way in which his history is structural) is that madness (always conceived, of course, as a pure function of reason) corresponds to a permanent, one might say to a transhistorical form; this form cannot be identified with the marks or signs of madness (in the scientific sense of the term), i.e., with the infinitely various signifiers of what is signified (itself diverse) which each society has invested in unreason, dementia, madness, or alienation; it is a question, rather, of a form of forms, i.e., of a specific structure; this form of forms, this structure, is suggested on each page of Foucault's book: it is a complementarity which opposes and unites, on the level of society as a whole, the excluded and the included…. Naturally, we must repeat, each term of the function is fulfilled differently according to period, place, society; exclusion (or as we sometimes say today: deviation) has different contents (meanings), here madness, there shamanism, elsewhere criminality, homosexuality, etc. But a serious paradox begins here: in our societies, at least, the relation of exclusion is determined, and in a sense objectified, by only one of the two humanities participating in it; thus it is the excluded humanity which is named (mad, insane, alienated, criminal, libertine, etc.), it is the act of exclusion, by its very nomination, which in a positive sense accounts for both excluded and "included" ("exile" of the mad in the Middle Ages, confinement in the classical period, internment in the modern age). Thus it is on the level of this general form that madness can be structured (not defined); and if this form is present in any society (but never outside of a society), the only discipline which could account for madness (as for all forms of exclusion) would be anthropology (in the "cultural" and no longer "natural" sense the word increasingly acquires for us). In this perspective, Foucault might have found it advantageous to give some ethnographic references, to suggest the example of societies "without madmen" (but not without "excluded groups"); but also, no doubt, he would regard this additional distance, this serene purview of all humanity as a kind of reassuring alibi, a distraction from what is newest about his project: its bewilderment, its vertigo.

For this book, as we realize when we read it, is different from a book of history, even if such history were audaciously conceived, even if such a book were written, as is the case, by a philosopher. What is it, then? Something like a cathartic question asked of knowledge, of all knowledge, and not only of that knowledge which speaks about madness. Knowledge is no longer that calm, proud, reassuring, reconciling act which Balzac opposed to the will which burns and to the power which destroys; in the couple constituted by reason and madness, by included and excluded, knowledge is a taking of sides; the very act which apprehends madness no longer as an object but as the other face which reason rejects, thereby proceeding to the extreme verge of intelligence, this act too is an act of darkness: casting a brilliant light on the couple constituted by madness and reason, knowledge thereby illuminates its own solitude and its own particularity: manifesting the very history of the division, it cannot escape it.

This misgiving … is inherent in Foucault's very project; once madness is no longer defined substantially ("a disease") or functionally ("antisocial conduct"), but structurally on the level of society as a whole as the discourse of reason about unreason, an implacable dialectic is set up; its origin is an obvious paradox: for a long time men have accepted the idea of reason's historical relativity; the history of philosophy is invented, written, taught, it belongs, one may say, to a hygiene of our societies; but there has never been a corresponding history of unreason; in this couple, outside of which neither term can be constituted, one of the partners is historical, participates in the values of civilization, escapes the fatality of being, conquers the freedom of doing; the other partner is excluded from history, fastened to an essence, either supernatural, or moral, or medical; doubtless a fraction—tiny, more-over—of culture acknowledges madness as a respectable or even inspired object, at least through certain of its mediators, Hölderlin, Nietzsche, Van Gogh; but this observation is very recent and, above all, it admits of no exchange: in short it is a liberal observation, an observation of good will, unfortunately powerless to dissipate bad faith. (pp. 167-69)

Foucault shows very well that the Middle Ages were actually much more open to madness than our modernity, for madness then, for from being objectified in the form of a disease, was defined as a great passage toward the supernatural, in short as a communication (this is the theme of The Ship of Fools); and it is the very progressivism of the modern period which here seems to exert the densest form of bad faith….

The history of madness could be "true" only if it were naïve, i.e., written by a madman; but then it could not be written in terms of history, so that we are left with the incoercible bad faith of knowledge. This is an inevitability which greatly exceeds the simple relations of madness and unreason; indeed, it affects all "thought," or to be more exact, all recourse to a metalanguage, whatever it might be: each time men speak about the world, they enter into a relation of exclusion, even when they speak in order to denounce it: a metalanguage is always terrorist. This is an endless dialectic, which can seem sophistical only to minds possessed of a reason substantial as a nature or a right; the others will experience it dramatically, or generously, or stoically; in any case, they know that bewilderment, that vertigo of discourse on which Michel Foucault has just cast so much light, a vertigo which appears not only upon contact with madness, but indeed each time that man, taking his distances, observes the world as different, which is to say, each time he writes. (p. 170)