[Book Notes] Michel Foucault: The Panopticon and the Habits of Discipline
Published:
This post supports English / 中文 switching via the site language toggle in the top navigation.
Introduction: Who designed the Panopticon?
Jeremy Bentham proposed the Panopticon prison design. Michel Foucault made it central to an influential analysis of disciplinary power. The book to read is Foucault’s Discipline and Punish: The Birth of the Prison, originally published as Surveiller et punir: Naissance de la prison by Gallimard in 1975. The discussion appears especially in Part Three, Discipline, in the chapter Panopticism. Alan Sheridan’s English translation first appeared in 1977. The French National Library catalogue and University of North Texas catalogue establish the editions and chapter placement.
The question that draws me to this part of the book is what happens when a person cannot tell whether anyone is watching. The uncertainty can enter an ordinary decision: when to pause, how to sit, which action will look acceptable. Authority may then operate through a person’s anticipation of being assessed, even between actual inspections.
That mechanism is unsettling enough. Yet stopping at surveillance would leave much of the book unread. Observation works alongside routines, classifications, records, and corrective exercises. Reading the Panopticon through those other practices changes the question from how much an institution sees to what it trains people to become.
1. A building, a mechanism, and a historical argument
Bentham’s scheme places an inspection lodge at the centre of a circular building, with the people under inspection arranged around it. Lighting and partitions make their activities observable while concealing the inspector’s presence. His Panopticon; or, The Inspection-House presents the arrangement as a practical project for institutional reform. UCL’s account adds a useful qualification to the attribution: the central inspection principle came from Bentham’s brother Samuel; Jeremy developed the prison proposal. He also envisaged uses in factories, hospitals, and schools. The scope beyond prisons was already there. UCL’s historical account helps keep that history distinct from Foucault’s later reading.
The book places the modern prison within a wider growth of practices for training and assessing people. My question then becomes how these practices acquire authority across everyday institutions, and how their ordinary usefulness can make that authority difficult to notice.
For Foucault, the architecture makes a relation of power unusually easy to examine. One party can inspect another without being equally exposed to inspection. The building gives that relation a physical form, but its operation can be discussed independently of a particular floor plan.
I find it useful to keep three levels separate while reading:
| Level | What it refers to | What would be an overreach |
|---|---|---|
| Bentham’s Panopticon | An architectural and institutional proposal for inspection | Attributing the design to Foucault |
| Foucault’s panopticism | A generalizable arrangement in which possible observation helps regulate conduct | Calling every instance of visibility panoptic |
| Disciplinary power | A wider set of practices that organize, train, compare, and correct individuals | Reducing the whole argument to cameras or prison buildings |
The distinction also prevents a mistake about Foucault’s purpose. An analysis of how an arrangement exercises power does not amount to an endorsement of that arrangement.
2. Visible power, unverifiable observation
In Foucault’s reading, the subject knows that observation is possible but cannot establish whether it is happening now. The tower is visible; the observer’s activity is unverifiable. An authorized reprint of Panopticism, hosted by Colby College, sets out this mechanism and its intended persistence between inspections.
Consider the difference between an announced inspection and an inspection that might occur at any time. In the first case, a person can prepare for the scheduled inspection. In the second, there is no clearly protected interval. If assessment carries consequences, adjusting one’s conduct continuously can become a reasonable response to the uncertainty.
This is how I understand the common phrase internalized discipline here: the person begins to carry out some of the supervisory work. An order need not be repeated each time. Nor must the person believe that the authority is wise or fair. Someone can resent an arrangement and still learn to anticipate its demands.
The distinction matters because compliance is easy to mistake for agreement. A quiet classroom or a uniformly busy office does not reveal what its occupants think. It reveals conduct under particular conditions. To infer wholehearted acceptance from that conduct would give the observer more knowledge than observation provides.
The model also needs a credible institution behind the possibility of inspection. An empty tower is no proof that authority can function forever without people, maintenance, records, or sanctions. Self-regulation remains tied to arrangements that make being noticed consequential.
3. A record needs a standard before it can judge
Suppose an institution can see every action. It still needs to decide which actions deserve praise, correction, or investigation. Surveillance alone supplies no answer.
In The Means of Correct Training, the chapter preceding Panopticism, Foucault connects hierarchical observation, normalizing judgment, and examination. Normalizing judgment compares conduct with a standard, including small departures that may break no law. Examination joins observation to that judgment and records an individual as a case. The Stanford Encyclopedia of Philosophy’s discussion of the prison explains this relation between knowledge and control.
A classroom example helps me distinguish the operations. Recording that a student took twenty minutes establishes a duration. Calling that student slow adds a comparison and a judgment about what the task should require. Prescribing extra exercises turns the judgment into an intervention. A file can then preserve the label after the original exercise has been forgotten. This is an illustrative example, not a case reported by Foucault.
The troubling step is often small: a limited observation acquires authority over a larger description of the person. How much of the student’s understanding did that timed exercise actually test? Did the task reward care or speed? The record can be accurate about the minutes and still support a poor judgment about the learner.
This is why I would read the discussion of examination alongside the tower. A system can collect more evidence while leaving its criteria largely unquestioned. Making its observations accurate will not, by itself, settle whether those criteria deserve to govern someone’s opportunities.
4. Discipline trains bodies as well as habits of thought
Foucault’s account also attends to the organization of bodies: assigned places, schedules, repeated movements, and exercises. His term docile bodies names the connection between increased usefulness and increased governability. The publisher’s guide to the book situates these techniques within its larger history of punishment and institutional power.
I read this as a warning against judging a training arrangement only by whether it produces skill. A tightly organized exercise may make people much better at a task. It can also reserve decisions about pace, acceptable movement, and the purpose of the task for someone else. Competence and discretion need separate attention.
The distinction is especially relevant to a technical education. Following a laboratory procedure can teach accuracy and make dangerous work possible. But a learner also needs occasions to ask why that procedure exists and when its assumptions fail. Those questions require a different relationship to the rule than simply performing it correctly.
There is no useful criticism in declaring all practice oppressive. The harder question is how a person gains enough command of a practice to question its use. That question keeps the value of training in view while refusing to treat obedient execution as a complete account of learning.
5. Why the prison does not contain the argument
Foucault examines the circulation of disciplinary techniques across institutions with different purposes. A school, hospital, workshop, and prison can all organize space, observe individuals, retain records, and use assessments to prescribe further activity. His claim concerns these relationships and practices; it does not require every building to have a central tower.
Bentham’s own text makes the comparison instructive. He discusses how inspection could serve several institutional purposes. Foucault asks what such arrangements reveal about the exercise of power. Reading the public-domain Inspection-House beside Foucault makes the difference between the reformer’s proposal and the critic’s question easier to see.
The analogy should still leave room for institutional differences. A patient asking a clinician to review a record, a child compelled to attend school, and a person confined in prison occupy very different positions. The ability to leave, challenge a decision, seek another opinion, or refuse an assessment changes what observation means.
For me, that is the practical value of the comparison. It directs attention to concrete arrangements that might otherwise look like neutral administration. It becomes less useful when similarity in one technique is allowed to erase differences in purpose, coercion, or accountability.
6. Where the model can mislead the reader
The tower is memorable enough to become an answer before an inquiry has begun. Once everything looks like a Panopticon, finding one more example becomes too easy.
Foucault presents an abstract model whose ideal operation sets aside resistance and friction. Actual institutions require evidence of their own. People can misunderstand rules, ignore them, negotiate exceptions, coordinate with others, or learn how to appear compliant. A design’s ambition tells us something about its operators; it does not establish what the design achieves.
The wider historical argument also deserves scrutiny. A shift toward disciplinary techniques does not establish that bodily violence or sovereign authority disappeared. Nor does a genealogy focused on particular European institutions supply a complete history of every society. I would use the book to formulate questions about a setting before using it to explain that setting.
There is a more personal temptation in this reading: to feel that recognizing power puts the reader outside it. Naming a score as a disciplinary device does not make the score stop affecting a career. It may, however, help specify what should be challenged: the measure, the interpretation, the consequences, or the absence of a way to appeal.
That is a smaller promise than total escape, but it gives criticism something definite to work on.
7. Workplaces, schools, and AI: a contemporary interpretation
The examples in this section are my present-day applications of the reading. They are hypothetical arrangements, not cases described by Foucault or claims that he anticipated today’s AI systems.
Imagine a workplace that records computer activity and lets managers inspect an employee’s score at any time. An employee who knows this may reorganize the day around visible activity. Time spent thinking through a difficult problem becomes awkward to defend if the measure recognizes only frequent interaction. The resemblance to the Panopticon lies in the combination of possible inspection, unequal access to the record, and consequences for the person assessed.
Now change one condition. Suppose employees use the same records to document impossible workloads, and management must respond to their findings. Observation has acquired another direction. This would not settle every objection to monitoring, but it would change the relation being analyzed. The device’s existence alone cannot tell us who gains the ability to question whom.
School metrics raise a related difficulty. A teacher may need evidence that a student is struggling. Yet a system that publicly rewards rapid completion could teach students to avoid tasks that slow them down. I would want to know whether the record helps a teacher understand the difficulty or simply turns difficulty into a lower rank. The same number can inform very different educational decisions.
AI adds a further distinction. Consider a system that infers attention from video and sends a score to a supervisor. Its claim to measure attention needs evaluation before any discussion of panopticism: an inference can be wrong. Even an accurate measurement would leave questions about who chose the standard and why a departure should have consequences. Technical validity and institutional justification are separate problems.
Some algorithmic decisions fit the tower poorly. A person may never know that a hidden classification affected which opportunities were shown to them. In that case, the system can affect their circumstances without their anticipating an observer or adjusting their conduct. Explaining this solely through anticipated surveillance would obscure the very mechanism that needs investigation. The analogy is strongest when it explains how a particular arrangement changes behavior; it should be revised when that mechanism is absent.
8. Who can question the assessment?
My strongest reservation about using this book in discussions of technology is how easily the image of total surveillance can replace an examination of ordinary decisions. A disputed attendance mark, an unexplained productivity score, or a record that cannot be corrected may tell us more about an institution than the number of cameras it owns.
I would therefore begin with the path from an observation to a consequence. What was recorded? How was it interpreted? Who can challenge that interpretation, and what happens while the challenge is considered? These are my practical questions for the present, rather than a policy programme supplied by Discipline and Punish.
Some observation deserves to continue. Records may help detect a dangerous practice or show that someone needs support. My judgment would depend on whether those benefits can be demonstrated, whether a less intrusive arrangement could provide them, and whether the people assessed can question both errors and the standard itself. A system that welcomes correction is materially different from one that demands permanent correction only from its subjects.
The question I would carry back from this book to a school, a research group, or a workplace is therefore quite specific: when an assessment changes someone’s life, what power does that person have to question it?
Further Reading
- Foucault, M. (1975). Surveiller et punir: Naissance de la prison. Gallimard. Original-edition record at the Bibliothèque nationale de France.
- Foucault, M. (1977; Vintage edition, 1995). Discipline and Punish: The Birth of the Prison. Translated by Alan Sheridan. Read Part Three, especially The Means of Correct Training and Panopticism. Edition and contents at the University of North Texas Libraries; publisher’s book page and guide.
- Foucault, M. (2008). Panopticism from Discipline & Punish: The Birth of the Prison. Race/Ethnicity: Multidisciplinary Global Contexts, 2(1), 1–12. An authorized reprint hosted by Colby College; a useful accessible entry into the argument.
- Bentham, J. Panopticon; or, The Inspection-House. In The Works of Jeremy Bentham, volume 4, edited by John Bowring. The public-domain primary text allows comparison with Foucault’s interpretation.
- UCL: The Panopticon. Historical context for Jeremy and Samuel Bentham and the inspection scheme.
- Stanford Encyclopedia of Philosophy: Michel Foucault. See History of the Prison for normalization, examination, and disciplinary power.