Foucault - Discourse, Discipline and the Production of Subjects

Michel Foucault (1926–1984) is the most cited author in the social sciences and humanities, and probably the most misrepresented. He rejected almost every label applied to him — structuralist, post-structuralist, postmodernist, historian, philosopher — and his work changed direction at least twice.

What holds it together is a single reversal, and it is the thing worth taking from this lesson: power does not principally forbid. It produces.

Two ways to make a person work.

The first. A foreman with a whip stands over a gang of labourers. He watches; when they slow, he strikes. Power here is visible, external, intermittent and negative. It stops things. It costs money — the foreman must be paid — and it works only while he is looking.

The second. A modern workplace. There is no foreman. There is a timesheet, a performance review, a set of objectives agreed at the start of the year, a peer feedback process, a professional development plan, and a set of metrics visible on a dashboard that the worker can consult at any time.

Nobody watches, and the worker watches herself. She knows what the metrics say. She knows what "underperforming" looks like and she does not want to be that. She has internalised a standard, measures herself against it continuously, and adjusts.

And here is the part Foucault insists on. She has not merely been constrained. She has been made into a particular kind of person: someone with objectives, a development plan, a self she works on, and a set of categories — high potential, needs improvement — in which she understands who she is.

The whip produced compliance. The review produces a subject.

Foucault's claim is that this second kind of power is the distinctively modern kind, that it is far more effective than the first, and that it is what we mean when we talk about the individual — because the individual, in the modern sense, is one of its products.

The problem Foucault poses is a rejection of how everyone in Part 5 so far has thought about power.

The standard model — call it the juridical or sovereign model — treats power as: Possessed by someone (a class, a state, an elite); Exercised over someone else; Negative in operation — it prohibits, censors, represses, excludes; Centralised in identifiable institutions; And external to those it acts on, who exist prior to it with their own interests and identities.

Every theory in this Part assumes it. Marx's ruling class, Weber's authority, Dahrendorf's imperative coordination, Lukes's three faces, Gramsci's hegemony — all of them have someone holding power and someone subject to it.

Foucault's argument: this model describes a form of power that has been historically superseded, and it makes the distinctively modern form invisible.

"We must cease once and for all to describe the effects of power in negative terms: it 'excludes', it 'represses', it 'censors', it 'abstracts', it 'masks', it 'conceals'. In fact power produces; it produces reality; it produces domains of objects and rituals of truth. The individual and the knowledge that may be gained of him belong to this production."

And his methodological consequence: do not ask who has power. Ask how power operates — through what techniques, what procedures, what forms of knowledge, and with what effects on the people it constitutes.

Power/knowledge

The hyphenated term (pouvoir/savoir) is not a claim that knowledge is a weapon of the powerful. It is a claim that the two are mutually constitutive.

"There is no power relation without the correlative constitution of a field of knowledge, nor any knowledge that does not presuppose and constitute at the same time power relations."

The mechanism: to govern a population you must know it — count it, classify it, measure it, record it. And the act of knowing produces the objects it studies: the criminal, the homosexual, the delinquent, the schizophrenic, the unemployed, the at-risk child.

These are not natural kinds waiting to be discovered. They are categories produced by particular knowledges, at particular times, for particular administrative purposes — and once produced, people are sorted into them, treated accordingly, and come to understand themselves in their terms.

Foucault's term for the systems that do this is discourse.

Discourse, in Foucault's sense, is not conversation or language. It is a regulated system of statements that produces the objects it speaks about, together with the rules governing what can be said, by whom, from what position, and what counts as a serious claim.

A discourse determines: what can be an object of knowledge; who is qualified to speak about it; what procedures establish truth; and what statements are even possible — not forbidden, but unsayable, because they would not register as claims at all.

His example in The Order of Things: a proposition in eighteenth-century natural history and one in nineteenth-century biology may look similar and belong to entirely different systems, in which "life," "organism" and "species" mean different things and are established by different procedures.

The consequence: truth is not outside power. "Truth isn't outside power... Each society has its regime of truth, its 'general politics' of truth: that is, the types of discourse which it accepts and makes function as true."

Note carefully what this does and does not claim. It does not claim that all statements are equally valid, or that facts do not exist. It claims that what counts as a serious candidate for truth — what can be asserted, tested, believed or disputed — is historically produced. Whether the earth goes round the sun is not up for grabs; what makes a question askable, and by whom, is.

Discipline and the panopticon

Discipline and Punish (1975) is his most read book and the source of the concepts most used in sociology.

It opens with a contrast that is the book's whole argument in two scenes.

1757: the public execution of Damiens the regicide, described over several pages in appalling detail — torn with pincers, molten lead poured into the wounds, and drawn apart by four horses that could not manage it, before a crowd.

Eighty years later: a timetable for young prisoners in Paris. Rise at five forty-five. Prayer. Work. Meals at fixed times. Instruction. Silence. Every hour accounted for.

Foucault's question: what happened in between?

The standard answer: humanitarian reform. Society became more civilised and less cruel.

Foucault's answer: the target changed and the technique changed. Punishment moved from the body to the soul; from spectacle to surveillance; from the exemplary infliction of pain on a few to the continuous correction of the many. Not less power — a different economy of it, and a far more efficient one.

Disciplinary power operates through identifiable techniques, and this is the part most useful to sociologists:

The distribution of bodies in space — enclosure, partitioning (each individual in their place), functional sites, and rank. The classroom with assigned desks; the hospital ward; the factory floor; the open-plan office.

The control of activity — the timetable, the temporal elaboration of the act, the correlation of body and gesture, and the exhaustive use of time. Taylorism (4.1.4) as a disciplinary technology.

The organisation of genesis — dividing training into stages, with exercises of graduated difficulty and examinations at each threshold.

The composition of forces — assembling individuals into an efficient machine, as in the drilled infantry unit.

Hierarchical observation, normalising judgement and the examination are the three instruments, and the second is the one people miss.

Hierarchical observation — architecture and organisation arranged so that seeing is built in.

Normalising judgement — a small penal mechanism operating in every disciplinary institution, punishing not offences against law but departures from the norm: lateness, inattention, incorrect posture, insufficient zeal. It compares, differentiates, hierarchises, homogenises and excludes — and above all it measures everyone against an average and treats deviation as correctable.

The examination — the ceremony combining both, in which the individual is made visible, measured, documented and turned into a case: an object of knowledge with a file.

And Foucault's observation about the file is one of his sharpest. For most of history, being written about was a privilege of the powerful — chronicles, monuments, genealogies. The disciplinary institution reverses this: it is now the ordinary and the deviant who accumulate records. "The turning of real lives into writing... functions as a procedure of objectification and subjection."

Anyone who has a medical record, a school record, a credit file, an employment history and a browsing history is the subject of this reversal.

Jeremy Bentham's 1791 architectural design: a circular building of cells around a central inspection tower, backlit so that each inmate is visible and the inspector is not.

The effect Bentham intended, and Foucault's point:

"The major effect of the Panopticon: to induce in the inmate a state of conscious and permanent visibility that assures the automatic functioning of power... the inmates should be caught up in a power situation of which they are themselves the bearers."

Three features make this the emblem of modern power:

Surveillance becomes unverifiable. The inmate cannot know whether they are being watched, so must behave as though they always are.

Power becomes automatic. It no longer requires the exercise of force. "He who is subjected to a field of visibility, and who knows it, assumes responsibility for the constraints of power... he becomes the principle of his own subjection."

And it is economical. One inspector, or none. The tower can be empty and the effect persists.

Foucault's generalisation: panopticism is not a building but a diagram of power — "a figure of political technology that may and must be detached from any specific use" — applicable to prisons, schools, hospitals, factories, barracks and asylums. He calls modern society a disciplinary society, in which these techniques have spread from institutions into the general administration of populations.

Now the criticisms, which are substantial and should be stated.

Historically, the panopticon was barely built. Bentham failed to get one constructed in Britain; a handful of radial prisons approximate it; and most nineteenth-century prisons did not follow the design. Foucault uses it as a diagram, not a description — and readers routinely take it as history.

Empirically, surveillance frequently fails to produce docility. Prison ethnography, workplace studies and school research consistently document resistance, evasion, gaming and the informal orders that surveillance generates rather than suppresses (5.3.1's negotiated order). People watched behave differently from what the model predicts, and they know they are being watched, which changes the game.

And the digital application needs care. Contemporary surveillance is frequently not panoptic: it is distributed rather than central, dataveillance rather than observation, and it often operates by prediction and sorting rather than by inducing self-discipline. Deleuze's short essay on "societies of control" (1990) argues that discipline's enclosed institutions have given way to continuous modulation — no walls, no completion, just perpetual adjustment. Zuboff's "surveillance capitalism" frames the mechanism as behavioural prediction sold as a product. These are extensions of and departures from Foucault, and conflating them loses what is distinctive about each.

Biopower and governmentality

The later work, and the part most useful for political sociology.

Biopower — the exercise of power over life and over populations, as against sovereign power over death.

Foucault's formulation of the shift: the sovereign's ancient right was "to take life or let live." Modern biopower is "to make live and to let die."

Its two poles: Anatomo-politics of the human body — discipline, as above: the individual body as machine, optimised and integrated into efficient systems. Biopolitics of the population — the species body: birth rates, mortality, health, longevity, fertility, migration, epidemics. The population as an object of knowledge and management, through statistics, public health, insurance, urban planning and welfare.

And this is where sociology's own tools appear as instruments of power. The census, the rate, the survey, the actuarial table (4.2.4) are all biopolitical technologies, and Foucault's account puts the discipline itself inside the picture it draws.

Governmentality — his term for the "conduct of conduct": the arts, rationalities and techniques of governing, understood broadly, from the government of a state to the government of a household to the government of oneself.

The crucial extension: liberal government does not govern less. It governs through freedom — by producing subjects who are autonomous, responsible, choosing and self-governing, and by arranging conditions so that their free choices produce the desired aggregate. Freedom is not the limit of this power; it is its instrument.

Go deeper

Sexuality, and the repressive hypothesis

The History of Sexuality, Volume 1 (1976) contains his most counterintuitive argument.

The received story: the Victorians repressed sexuality; the twentieth century liberated it; and speaking about sex is therefore an act of liberation.

Foucault calls this the repressive hypothesis and rejects it.

His evidence: the period supposedly characterised by silence produced an explosion of discourse about sex — in medicine, psychiatry, criminology, demography, pedagogy and above all in the confessional and its secular descendants. Sex was not silenced; it was made into the object of an enormous machinery for producing speech about it.

And the mechanism he identifies is the confession, transposed from religion to medicine and psychiatry: the obligation to speak the truth about oneself to an authority who interprets it, and the belief that this speech liberates.

His famous claim about homosexuality illustrates the production thesis: "The sodomite had been a temporary aberration; the homosexual was now a species." Before roughly the 1870s, sodomy was a category of forbidden acts anyone might commit. Nineteenth-century psychiatry and law produced a category of person — with a history, a childhood, a physiology and a case file.

Note what this claims and does not. It does not claim same-sex desire is modern. It claims that the identity — the idea that a person is a homosexual, that this is a kind of being — is a historical production of a particular knowledge.

And the political sting: liberation movements that demand recognition of the identity are operating on terrain produced by the power they oppose. "Where there is power, there is resistance" — and the resistance is shaped by what it resists.

(Historians of sexuality have complicated this. Some evidence of earlier subcultures and self-understandings — the London molly houses of the early eighteenth century, for instance — suggests the discontinuity is less sharp than Foucault claimed. The general point about the production of identity categories has held up better than the specific dating.)

The criticisms

Where is agency? If subjects are produced by discourse, who resists and on what basis? Foucault insists resistance is coextensive with power, and critics reply that this is asserted rather than explained. Charles Taylor and others argue that critique requires a standpoint, and that Foucault's framework denies him one.

The Habermas objection (5.4.2), and it is the sharpest. If all discourse is power-saturated and there are no criteria of validity independent of a regime of truth, why is Foucault offering arguments rather than simply exercising power? Habermas called this a "cryptonormativism": Foucault's work is plainly indignant about prisons, asylums and psychiatry, and the indignation requires normative grounds his theory forbids. Foucault died in 1984 before a direct exchange could occur, and the confrontation has been conducted by proxies ever since.

The history is unreliable in places. Madness and Civilization has been challenged extensively by historians of psychiatry — on the scale of the "great confinement," on the dating, and on the characterisation of pre-modern attitudes to madness. H. C. Erik Midelfort's critique is the standard reference. Foucault's response, in effect, was that he was writing philosophical history, which is a defence some historians accept and others regard as an evasion.

Feminist critiques are mixed and productive. Some — Sandra Bartky, Susan Bordo — found the disciplinary framework enormously useful for analysing how femininity is produced through practices on the body. Others — Nancy Hartsock — argued that a theory dissolving the subject arrives just as previously silenced groups are claiming subjecthood, which is unfortunate timing at best.

And the political ambivalence. Foucault's analysis of how welfare, medicine and social work operate as disciplinary systems was taken up by critics of the welfare state from several directions, including some he would not have chosen. His own late interest in neoliberalism — the 1979 Collège de France lectures — has generated a substantial and contested literature about whether he was analysing it or partly attracted to it.

What sociology took

Discourse analysis as a method, in several forms (7.4).

The production of categories. The entire literature on how kinds of person are made — Ian Hacking's "making up people" and his concept of the looping effect, in which classified people change in response to being classified, thereby altering the classification. This is Foucault made precise, and it is one of the best ideas in the philosophy of social science.

Surveillance studies as a field, with the qualifications above.

Governmentality studies — Nikolas Rose, Peter Miller and others, on advanced liberal government, responsibilisation, audit and the government of the self. This has been enormously productive in the sociology of health, welfare, education and risk.

The sociology of the body and of medicine — medicalisation, the clinical gaze, and the constitution of the patient.

And a permanent methodological question: what would happen if we asked, of any category we use, when was this produced, by whom, in what discourse, for what administrative purpose, and what happened to the people sorted into it?

"Foucault said power is everywhere, so nothing can be done." He said power is a relation rather than a possession, and that "where there is power, there is resistance." The quietism charge is a real criticism, not his position.

"Power/knowledge means knowledge is just power." It means the two are mutually constitutive: governing requires knowing, and knowing produces its objects. It is not a claim that evidence is irrelevant.

"Foucault denied that truth exists." He argued that each society has a regime of truth determining what counts as a serious candidate for truth. That is a claim about the conditions of assertion, not a denial of facts.

"The panopticon describes modern society." It is a diagram, barely built historically, and contemporary surveillance is frequently distributed and predictive rather than panoptic.

"Discipline replaced sovereign power." Foucault describes a shift in the dominant modality; sovereign power did not disappear, as the persistence of imprisonment, deportation and war makes obvious.

"He said homosexuality was invented in the nineteenth century." He said the identity category — being a kind of person — was produced then. Historians have complicated even that.

Why it matters

It reverses the question about power, from who has it to how does it operate and what does it produce. That reversal makes visible a range of phenomena — professional expertise, classification, record-keeping, self-improvement, audit, therapeutic culture — that the possession model cannot see at all.

It makes categories historical. Every category of person a society uses was produced somewhere, by someone, in a discourse, for a purpose. Asking when and why is one of the most productive moves available, and Hacking's looping effect shows that the categories then change the people in them.

It puts sociology inside its own picture. The census, the survey, the rate and the case file are biopolitical instruments, and a discipline that produced them cannot analyse governmentality as though it were standing outside it. That is uncomfortable and correct.

And it poses the hardest question in this Part. If subjects are produced by the discourses that describe them, what is left of the person who resists — and what grounds does the critic stand on? Foucault did not answer this, Habermas pressed it, and the standing tension between them is the most important unresolved argument in late twentieth-century social theory.

Foucault's reversal: against the juridical model of power as possessed, negative, centralised and external, he argues that "power produces; it produces reality; it produces domains of objects and rituals of truth." The methodological consequence: ask how, not who.

Power/knowledge — mutually constitutive: governing requires knowing, and knowing produces its objects. Discourse is a regulated system of statements that produces what it speaks about and governs what can be said, by whom, and what counts as a serious claim. Each society has a regime of truth.

Discipline and Punish: the shift from the execution of Damiens to the prison timetable is not humanitarian progress but a change of target (body to soul) and technique (spectacle to surveillance). Discipline operates through the distribution of bodies, control of activity, graduated training and composition of forces — with hierarchical observation, normalising judgement (punishing departures from the norm, not offences against law) and the examination, which turns the individual into a case with a file. And the historical reversal: being written about was once a privilege; now it is the ordinary and the deviant who accumulate records.

The panopticon is a diagram — surveillance made unverifiable, power made automatic, the inmate becoming "the principle of his own subjection." It was barely built, surveillance frequently produces resistance rather than docility, and digital surveillance is often distributed and predictive rather than panoptic (Deleuze's societies of control; Zuboff).

Biopower: from "take life or let live" to "make live and let die" — anatomo-politics of the body and biopolitics of the population, through statistics, public health and insurance. Governmentality: the conduct of conduct, in which liberal government governs through freedom rather than against it.

The repressive hypothesis rejected: the Victorian period produced an explosion of discourse about sex, not silence; and "the sodomite had been a temporary aberration; the homosexual was now a species."

The criticisms: the missing account of agency; Habermas's cryptonormativism charge — indignation requires grounds the theory forbids; unreliable history in Madness and Civilization; and mixed feminist assessments.

Discourse — a regulated system of statements producing its objects and governing what can be said.

Power/knowledge — the mutual constitution of relations of power and fields of knowledge.

Regime of truth — the types of discourse a society accepts and makes function as true.

Disciplinary power — power operating through surveillance, normalisation and examination on individual bodies.

Normalising judgement — the penal mechanism punishing departures from a norm rather than offences against law.

The examination — the ceremony that makes the individual visible, documented and a case.

Panopticism — the diagram of unverifiable surveillance producing self-regulation.

Docile bodies — bodies rendered useful and manageable through disciplinary training.

Biopower — power over life and populations: "to make live and to let die."

Biopolitics — the management of populations through statistics, health and welfare.

Governmentality — the conduct of conduct; the rationalities and techniques of governing, including of the self.

Repressive hypothesis — the rejected view that modern sexuality was silenced and then liberated.

Looping effect — Hacking's term for classified people changing in response to their classification.

Societies of control — Deleuze's successor concept: continuous modulation replacing enclosed discipline.

One — audit a disciplinary institution. Take a school, office, hospital or gym. Identify the distribution of bodies in space, the control of time, the examinations, and what counts as a departure from the norm. All four will be present.

Two — date a category. Take any category of person — the teenager, the commuter, the consumer, the burnout, the at-risk pupil, the gifted child. Find out when it appeared and in what literature. Every one has a birthday.

Three — find a looping effect. Identify a diagnostic or administrative category and ask how people so classified have changed in response — and whether the category then changed too.

Four — test the panopticon claim. Take a monitored setting you know and ask whether surveillance produced docility or produced evasion, gaming and an informal order. The answer is usually the second, which is the finding.

Five — read the opening thirty pages. Discipline and Punish begins with Damiens and ends with a timetable. Read both and formulate your own answer to what happened in between before reading his.