The Stanford Prison Experiment: How Normal Students Became Sadistic Guards in 6 Days

The ad in the New Haven newspaper seemed innocent enough. "Volunteers wanted for a study of memory and learning at Yale University. Payment: $4." None of the people who would respond to this ad could have imagined that they would be participating in one of the most shocking experiments in the history of psychology — an experiment that would reveal a truth so disturbing, it would forever change the way we see ourselves.

It was August 1961. Three months earlier, in Jerusalem, the trial of Adolf Eichmann had begun — the bureaucrat who organized the transport of millions of Jews to the death camps. His defense was simple and chilling: “I was just following orders.” A young psychologist at Yale, Stanley Milgram, decided to test whether this excuse held any truth — whether ordinary, everyday people could indeed torture a stranger, simply because someone in authority asked them to.

The answer would be far worse than anyone expected.

Stanley Milgram was born in 1933 in the Bronx, New York, to a family of Jewish immigrants from Eastern Europe. He grew up in the shadow of the Holocaust — many of his relatives had been living in Europe during the war. The question that haunted him was deeply personal: how could millions of people participate in such massive atrocity? Were the Germans somehow “different,” or could it happen anywhere?

After his studies at Harvard, where he worked with Solomon Asch — renowned for his conformity experiments —, Milgram took a position at Yale. There he designed an experiment that he originally intended to conduct first on Americans as a “control group” and then on Germans, expecting the latter to show far greater obedience. He never needed to go to Germany. The results from the Americans were shocking enough.

Forty psychiatrists from a medical school were also asked. They predicted that at 300 volts, when the “victim” would refuse to answer, only 3.73% would continue. And they believed that “just over one-tenth of one percent” would reach the maximum shock. Reality would prove them terrifyingly wrong.

Three people took part in each session. The “experimenter” — a stern man in a white lab coat, representing authority. The “teacher” — the actual participant, an ordinary citizen who believed he was assisting in a memory study. And the “learner” — an actor, though the participant did not know he was an actor.

The teacher and the learner drew lots for their roles. But the draw was rigged — both slips of paper read “teacher.” The actor always “drew” the role of the learner. The learner was then strapped into a chair in an adjacent room, connected to electrodes — fake ones, but the teacher didn't know that.

Before the session began, the teacher received a sample electric shock of 45 volts — this one was real — so he could “feel” what the learner would supposedly experience. It was a painful jolt of electricity. Enough for the teacher to believe that the learner was truly in pain.

The shock generator had 30 switches, ranging from 15 to 450 volts, in 15-volt increments. Above each group of switches were labels: “Slight Shock,” “Moderate Shock,” “Strong Shock,” “Very Strong Shock,” “Intense Shock,” “Extreme Intensity Shock,” “Danger: Severe Shock,” and finally, two red switches marked simply “XXX.”

The procedure was simple. The teacher read pairs of words to the learner. Then he asked for the correct match. Every wrong answer meant a shock. And each successive shock was 15 volts stronger than the last. At 75 volts, the learner groaned. At 120, he cried out in pain. At 150, he begged them to stop. At 300, he refused to answer and pounded on the wall.

At 330 volts — silence. Absolute, frozen silence. For the teacher, this could only mean one thing: something very bad had happened in the next room.

When the teacher wanted to stop — and almost all of them did at some point — the experimenter used four successive "prods":

The results were devastating. Every single participant — every one, without exception — continued up to 300 volts. The point at which the learner had already screamed in pain, begged for release, and refused to participate. And 65% — 26 out of 40 — went all the way. To 450 volts. To the last, red “XXX” switch — and pressed it three times in a row, as the protocol required.

This doesn't mean it was easy for them. The participants sweated, trembled, stuttered, bit their lips, groaned, and dug their fingernails into their skin. Fourteen of the 40 exhibited nervous laughter or seizures. Some offered to return the $4 payment. Some begged the experimenter to stop. All of them asked “is this all right?” at least once.

But in the end, 65% obeyed. Not a single one of those who refused to go to 450 volts demanded that the experiment be stopped entirely — no one went to the next room to check whether the learner was alive. They stopped, but they took no action. Obedience was not just about “do it” — it was also about “don't do anything beyond what you're told.”

Milgram did not stop at one experiment. In his book “Obedience to Authority” (1974), he describes 19 variations, each designed to answer a different question. What happens if the conditions change? If the learner is closer? If the experimenter is further away? If others refuse?

When the experimenter gave instructions by telephone, obedience dropped to 21%. Some participants even deceived the experimenter, pretending to administer shocks while actually pressing the lowest switches. The physical presence of authority, then, was decisive.

When the learner was in the same room as the teacher — instead of a separate space — obedience dropped to 40%. When the teacher had to physically hold the learner's hand down on the metal shock plate, only 30% continued to the end. Proximity to the victim dramatically reduced obedience. It was easier to inflict pain on someone you couldn't see.

Location also mattered. When the experiment was moved from Yale's impressive facilities to a drab office in Bridgeport, Connecticut — with no university affiliation — obedience dropped from 65% to 47.5%. The prestige of the institution served as a shield for one's conscience.

But the most striking variation involved social pressure. When two additional “teachers” — also actors — refused to continue, only 4 out of 40 participants went all the way. Conversely, when a second “teacher” obeyed fully, 37 out of 40 followed suit. It wasn't just authority that determined behavior — it was also what those around you were doing.

Another experimental variation used exclusively female participants. Obedience did not differ significantly, although the women reported higher levels of stress. Authority's ability to compel obedience knew no gender.

The publication of the results ignited a firestorm. On June 10, 1964, psychologist Diana Baumrind published a brief but devastating article in the American Psychologist: “Some Thoughts on Ethics of Research: After Reading Milgram.” She argued that even if Milgram had obtained the participants' consent, he was morally responsible for their psychological trauma. The people who sweated, trembled, and had fits of laughter were not all right — and the experimenter should not have let them continue.

Baumrind's critique led to a radical overhaul of ethical standards in psychological research. Today, an experiment like Milgram's would be impossible to approve by any ethics committee. The psychological pain experienced by participants — the discovery that they were capable of torturing someone on command — was a wound that some carried for years.

Milgram vigorously defended his experiment. In a survey he conducted among former participants, 84% said they were “glad” or “very glad” to have taken part. Only 1.3% said they regretted it. Former participants wrote him letters of gratitude, offered their help, and asked to join his team.

There were, however, skeptics. Researcher Gina Perry, in her book “Behind the Shock Machine” (2012), argued that Milgram had manipulated the results and that many participants didn't truly believe the shocks were real. According to her analysis, only half believed it was real — and of those, 66% refused to obey. But the truth is that dozens of replications of the experiment in many countries — from Australia to Italy and South Africa — yielded similar results.

Milgram proposed two theories to explain the results. The first was conformity theory, based on Solomon Asch's experiments: a person lacking experience or decision-making ability in a crisis defers to the group and its hierarchy. The second was the theory of the "agentic state": a person ceases to see himself as responsible for his actions and becomes merely an instrument for carrying out orders.

This “agentic state” was not merely theory. It was observed in real time. Participants laughed nervously, sweated, looked to the experimenter for approval — but did not assume personal responsibility. The question “are you responsible if something goes wrong?” — and the answer “yes, I take responsibility” — was enough for many to continue.

Philosopher Hannah Arendt, observing Eichmann's trial, coined a phrase that would become immortal: “the banality of evil.” Eichmann was no monster. He was a bureaucrat. He followed rules. He paid his employees properly. He efficiently organized train schedules. The fact that those trains ended up at gas chambers was, for him, a detail outside his jurisdiction.

Milgram claimed that “a common psychological process is centrally involved in both” — in his laboratory and in Germany. But critics, such as James Waller, noted essential differences: the participants believed there would be no permanent harm, did not know the victim, harbored no hatred, and the experiment lasted one hour — whereas the Holocaust lasted years, with full knowledge and intent.

Milgram's experiment was not unique. It inspired — or perhaps revealed — a series of experiments into the dark side of human nature. In 1971, Philip Zimbardo, a former high school classmate of Milgram's, conducted the Stanford Prison Experiment: students were randomly assigned as “guards” or “prisoners” in a mock prison basement. Within six days, the “guards” had become so sadistic that the experiment was terminated early.

The same had been demonstrated, in a quieter way, by the Hofling experiment (1966) in a real hospital. An unknown “doctor” phoned nurses and asked them to administer double the maximum permitted dose of a (fictitious) drug to a patient. 95% of the nurses prepared to carry out the order, despite safety protocols. Authority had won again.

And on real battlefields, the mechanics of obedience were laid bare at Mỹ Lai, a small community in South Vietnam. On March 16, 1968, American soldiers under the orders of Lieutenant William Calley killed between 347 and 504 unarmed civilians — elderly men, women, children. Many soldiers later reported that they were “following orders.” Only one, Warrant Officer Hugh Thompson, refused — turning his weapons on his own comrades to protect civilians.

Even more chilling was a 1972 experiment by Charles Sheridan and Richard King, who wanted to test whether Milgram's participants had realized the shocks were fake. They used a real victim — a small puppy — that actually received electric shocks (harmless, but real). Seven out of 13 men obeyed to the end. All 13 women obeyed to the end. Many were crying as they pressed the switch.

In 2006, psychologist Jerry Burger managed to obtain approval for a partial replication of the experiment, stopping at 150 volts instead of 450, for ethical reasons. The results were “nearly identical” to Milgram's from 1961. People had not changed in half a century.

The most shocking replication came from France. In 2010, in the documentary “Le Jeu de la Mort” (The Game of Death), researchers redesigned the experiment as a pilot television game show. Volunteers were paid 40 euros and told they wouldn't win anything — this was just a test run. Only 16 out of 80 “contestants” stopped before administering the maximum punishment. The remaining 80% went all the way. Television — the modern authority — proved even more persuasive than the white lab coat.

Thomas Blass, in the largest meta-analysis of replications of the experiment, found that the rate of full obedience ranged from 28% to 91%, depending on conditions. There was no significant trend of change over time. The average in the U.S. was 61%, while in non-U.S. studies it was 66%.

Stanley Milgram died in 1984, at the age of 51, from a heart attack — his fifth. He remained controversial even at the end, but his experiment endures as perhaps the most important psychological study of the 20th century. He didn't tell us anything new — but he proved, numerically, what we didn't want to admit.

The actor who played the “learner” in dozens of sessions was named James McDonough — though Milgram referred to him functionally as “Mr. Wallace” in his publications. He sat strapped to the chair, listening to the teachers increase the voltage, button after button. He knew the shocks were fake. But he experienced something real: the moment when ordinary, polite, educated people decided to “kill” him — because someone in a lab coat told them it was all right.

What frightened him most was not the electricity. It was their eyes. The process by which a pair of eyes full of concern gradually transformed into a pair of eyes that simply followed orders. The moment when a person stopped thinking and simply began to execute. That moment was not hidden behind crematoria or battlefields. It was found in a Yale courtyard, on a Tuesday afternoon, in the eyes of a neighbor, a postman, a schoolteacher.

Perhaps this is the most terrifying thing Milgram taught us: that the line between “them” and “us” is far, far thinner than we think.