Mind Bending · Cults & Group Psychology · Obedience · Authority · 1961

The Milgram Experiment

In 1961, ordinary Americans administered what they believed were lethal electric shocks to an innocent stranger — simply because a man in a white coat told them to continue. 65% went all the way to 450 volts. The most disturbing experiment in the history of psychology. And the most important.

Researcher

Stanley Milgram — Yale University

Year

1961–1962 · Published 1963

Key finding

65% of participants administered maximum shock

Context

Designed in response to the Eichmann trial

The Eichmann question. Stanley Milgram designed his experiment in direct response to the 1961 trial of Adolf Eichmann — the Nazi official responsible for organising the transportation of millions of Jews to extermination camps. Eichmann's defence was that he had merely followed orders — that he was an ordinary bureaucrat doing his job. Hannah Arendt's coverage of the trial produced the phrase "the banality of evil." Milgram's experiment was designed to answer a specific question: was the Holocaust the product of uniquely German psychology, or could ordinary Americans be induced to harm innocent strangers simply by being told to by an authority figure?

Section 01

The Experiment

The setup appeared simple. A participant arrived at Yale University's psychology laboratory, having responded to a newspaper advertisement offering $4.50 for one hour's participation in a study of memory and learning. They were introduced to another participant — actually a confederate, an actor — and assigned the role of "teacher" through an apparently random draw that was secretly rigged. The confederate was assigned the role of "learner."

The learner was taken to an adjacent room and strapped to a chair with electrodes attached to their wrists. The teacher was seated in front of a shock generator — a convincing piece of equipment with 30 switches ranging from 15 volts to 450 volts, labelled in 15-volt increments from "Slight Shock" through "Danger: Severe Shock" to, at the highest level, simply "XXX." The teacher's task was to read word pairs to the learner, test their recall, and administer an electric shock for each wrong answer — increasing the voltage by 15 volts with each error.

The learner — the actor — gave wrong answers according to a predetermined script and responded to the increasing shocks with escalating protests: grunts at 75 volts, complaints at 120 volts, demands to be released at 150 volts, refusal to answer at 300 volts, and silence thereafter. When participants hesitated or refused to continue, the experimenter — a stern man in a grey lab coat — delivered one of four standardised prods: "Please continue," "The experiment requires that you continue," "It is absolutely essential that you continue," and "You have no other choice, you must go on."

No actual shocks were delivered. The learner was unharmed. But the participants did not know this. They believed — and their physiological responses confirmed — that they were causing genuine pain to a real person. And 65% of them administered shocks all the way to the maximum 450 volts.

Section 02

The Results

Before conducting the experiment, Milgram surveyed psychology students, middle-class adults, and psychiatrists about what percentage of participants they expected to go all the way to 450 volts. The predictions clustered around 1–3% — a small minority of sadists or psychopaths, the assumption being that normal people would refuse long before reaching lethal levels. The actual result — 65% — was not merely higher than predicted. It was categorically different from what anyone had expected.

The participants were not monsters. Milgram's recordings show them sweating, trembling, laughing nervously, pleading with the experimenter to stop the study. Many showed visible signs of extreme distress. They were not enjoying what they were doing. They continued anyway. The disjunction between their emotional state and their behaviour is the experiment's most disturbing finding — and its most important one. These were not people who wanted to harm strangers. They were people who, when placed in a specific situational context with a specific authority structure, harmed strangers despite not wanting to.

The Number

65% — Not a Minority

Two thirds of participants administered what they believed were potentially lethal shocks to a stranger. Not a small minority of disturbed individuals — two thirds of an ordinary American sample. The result has been replicated across cultures, genders, and decades with remarkably consistent findings. A 2008 replication by Jerry Burger found 70% compliance to 150 volts (the point at which the learner first demanded to be released). The finding is not an artefact of 1960s American psychology. It is a feature of human psychology under conditions of authority.

The Distress

Knowing and Doing Anyway

The participants knew what they were doing was wrong. They expressed distress, asked the experimenter to stop, and showed physiological signs of extreme stress. They continued anyway. This is the finding that most directly challenges the assumption that moral knowledge produces moral behaviour. The participants had the moral knowledge — they knew harming an innocent person was wrong. They had the emotional response — they were distressed by it. They continued because the situational authority structure was stronger than their individual moral agency.

The Surprise

Nobody Predicted This

The gap between predicted behaviour (1–3% compliance) and actual behaviour (65% compliance) is itself significant. It means that our intuitive model of how we and others would behave in this situation is wildly inaccurate. We believe we would refuse. The evidence suggests most of us would not. This discrepancy between self-prediction and actual behaviour is not a limitation of Milgram's participants — it is a feature of how human beings model their own psychology. We systematically overestimate the power of our moral convictions relative to situational pressure.

Section 03

Why They Obeyed

Milgram's analysis of why two thirds of his participants administered maximum shocks identified several interlocking factors — none of which alone would have produced the result, but which together created a situational pressure powerful enough to override individual moral judgement in the majority of cases.

Institutional Legitimacy

Yale · White coat · Scientific authority

The experiment took place at Yale University — one of America's most prestigious institutions. The experimenter wore a grey lab coat. The study was presented as scientific research with legitimate academic purposes. The institutional context conferred authority on the experimenter's instructions that the same instructions would not have carried from a stranger on the street. When Milgram replicated the experiment in a run-down commercial building rather than Yale, compliance dropped from 65% to 47.5%. Institutional legitimacy is itself a mechanism of obedience.

Incremental Escalation

The foot-in-the-door · 15 volts at a time

The voltage increased gradually — 15 volts at a time, from a starting point of 15 volts that caused only minor discomfort. By the time participants reached the dangerous levels, they had already committed to the procedure through a long series of smaller steps. Each step seemed only marginally more serious than the previous one. Refusing at any point required acknowledging that all previous steps had been wrong — a psychologically costly acknowledgement that most people avoided by continuing. The incremental structure is the same technique used in cult recruitment, radicalization, and the gradual escalation of institutional violence.

Diffusion of Responsibility

The experimenter is responsible — not me

When participants expressed concern, the experimenter's prods explicitly assigned responsibility: "I'm responsible for what happens here." This diffusion of responsibility — the transfer of moral accountability from the actor to the authority — is a central mechanism of institutional harm. Milgram's soldiers, bureaucrats, and corporate employees who participate in harmful actions while believing that responsibility lies with those who gave the orders are not moral aberrations. They are operating exactly as Milgram's experiment predicts.

Section 04

Variations — What Changes Obedience

Milgram ran 21 variations of his experiment — systematically changing one element at a time to identify which factors increased or decreased compliance. These variations are as important as the original finding: they reveal the specific situational levers that control obedient behaviour and, therefore, the conditions that can interrupt it.

Variation 01

Proximity — Compliance Drops with Visibility

When the learner was in the same room as the teacher — visible and audible — compliance dropped to 40%. When the teacher had to physically hold the learner's hand onto a shock plate, compliance dropped to 30%. The closer the victim, the lower the compliance. Physical distance from the consequences of one's actions increases willingness to continue them. This is the mechanism that makes bureaucratic violence possible: the person who signs the order never sees the person who suffers its execution.

Variation 02

Peer Rebellion — The Most Effective Interruption

When two confederates posing as fellow teachers refused to continue at 150 volts, compliance dropped to 10%. The presence of peers who modelled resistance was the single most effective factor in reducing obedience — more effective than moving the victim closer, more effective than making the authority figure less prestigious. Peer dissent breaks the assumption that compliance is the universal norm. It gives individual resisters permission to do what they already want to do. This is why authoritarian systems suppress dissent so vigorously: not because one dissenter is dangerous but because one dissenter enables others.

Variation 03

Authority at Distance — Compliance Drops

When the experimenter gave instructions by telephone rather than in person, compliance dropped to 20%. Many participants in this condition said they would continue but secretly administered lower voltages than instructed. Physical presence of authority — the direct gaze, the immediate social reality of the authority figure — is a significant factor in compliance. Digital and remote authority structures are, on this evidence, less coercive than in-person ones. The Zoom meeting is, in this sense, slightly less authoritarian than the boardroom.

Section 05

The Agentic State

Milgram's theoretical explanation for his findings centred on what he called the "agentic state" — a psychological condition in which an individual ceases to act as an autonomous agent and instead becomes an instrument of an authority's will. In the agentic state, the individual's sense of responsibility for their actions shifts from themselves to the authority — they experience themselves as executing another's will rather than exercising their own.

The transition into the agentic state is not conscious — it happens through the gradual absorption into an institutional role with its associated expectations, obligations, and authority structures. Once in the agentic state, the individual's moral discomfort does not disappear — as Milgram's participants' visible distress demonstrated — but it is reinterpreted as personal weakness rather than legitimate moral signal. The "good soldier," the "loyal employee," the "professional" executes instructions despite personal discomfort because the discomfort is understood as a private feeling that should not interfere with the performance of duty.

The Agentic State

Instrument vs Agent

The shift from autonomous agent to instrument of authority is the psychological pivot point of the Milgram experiment — and of every institutional atrocity. The soldier who follows orders, the bureaucrat who processes the paperwork, the engineer who builds the weapon: each has entered the agentic state. They have not lost their moral knowledge or their emotional responses. They have reframed their role — from person who acts to person who executes — in a way that transfers moral responsibility to the authority and removes it from themselves.

The Connection

Milgram & the Holocaust

Milgram explicitly connected his findings to the Holocaust — and specifically to Eichmann's defence. His conclusion: the Holocaust was not the product of uniquely German pathology. It was the product of universal human psychology placed in specific institutional conditions. The camp guards, the train operators, the bureaucrats who managed the deportations were not monsters who wanted to kill Jews. They were ordinary people in the agentic state, executing the instructions of a legitimate authority structure, with responsibility diffused across thousands of participants none of whom felt primarily responsible for the whole.

The Relevance

Every Institution Is a Milgram Apparatus

Every hierarchical institution — corporation, military, government, religious organisation, university — creates the conditions for agentic state psychology: authority figures, legitimate institutional purposes, incremental escalation of requests, diffusion of responsibility, and peer conformity pressure. The Milgram experiment does not describe an unusual situation. It describes the normal operating conditions of institutional life. The question it poses — at what point does the individual reassert moral agency against institutional pressure? — is asked daily by millions of people in ordinary workplaces.

Section 06

The Experiment Today

The Milgram experiment has been replicated, criticised, defended, and extended in the sixty years since its original publication. Its ethical conduct — participants were subjected to extreme stress without adequate informed consent — makes direct replication impossible under modern research ethics standards. But partial replications, using lower shock levels and stopping before maximum compliance, consistently find compliance rates in the same range as Milgram's original findings.

The experiment's relevance to contemporary life has, if anything, increased since 1961. Digital authority structures — the algorithm, the platform, the employer monitoring system, the social credit score — are new forms of the authority apparatus that Milgram's experiment illuminates. The person who posts inflammatory content because the algorithm rewards it, the employee who silences ethical concerns because the corporate culture penalises them, the citizen who complies with surveillance because the state authorises it — all are navigating versions of the situation Milgram created in a Yale laboratory in 1961.

The most important legacy of the Milgram experiment is not the depressing finding about human compliance. It is the hopeful finding about its disruption: one person who says "I won't continue" reduces compliance by 55 percentage points. The power of individual moral agency, when actually exercised, is enormous — precisely because it is so rarely exercised. The experiment is not a counsel of despair about human nature. It is a precise description of the conditions under which human moral agency fails — which is also, necessarily, a map of the conditions under which it can be protected.

"The disappearance of a sense of responsibility is the most far-reaching consequence of submission to authority."

Stanley Milgram — Obedience to Authority, 1974