The Milgram Experiment: Obedience, Shock, and What It Really Showed

In the summer of 1961, at Yale University, Stanley Milgram ran an experiment that seemed to reveal something terrible about human nature: give ordinary people a lab coat, an authority figure, and an instruction, and a large majority will administer what they believe to be near-lethal electric shocks to a stranger. Sixty-five percent of Milgram's participants went to the maximum voltage of 450 volts. The experiment became a cultural monument. It also has more cracks in it than the textbooks admit.

The Question Milgram Was Asking

Milgram designed the study in the immediate aftermath of the Adolf Eichmann trial. Eichmann, the Nazi bureaucrat who organized the logistics of the Holocaust, had claimed in Jerusalem in 1961 that he was simply following orders. Hannah Arendt, covering the trial for The New Yorker, coined the phrase "the banality of evil" — the idea that monstrous outcomes could emerge from mundane compliance rather than sadistic intent.

Milgram wanted to test this empirically. Could ordinary Americans — recruited from New Haven, not selected for any particular ideology — be induced to harm a stranger simply because an authority figure told them to?

Destructive obedience — compliance with authority even when compliance causes clear harm to others — was the behavior Milgram sought to measure and explain.

Source: Milgram, S. (1963). Behavioral study of obedience. Journal of Abnormal and Social Psychology, 67(4), 371–378.

Source: Arendt, H. (1963). Eichmann in Jerusalem: A Report on the Banality of Evil. Viking Press.

The Setup

The Shock Generator

At the center of the experiment was a large, impressive-looking piece of equipment: a shock generator with 30 switches, labeled in 15-volt increments from 15 volts to 450 volts. Groups of four switches were labeled with verbal descriptors: "Slight Shock," "Moderate Shock," "Strong Shock," "Very Strong Shock," "Intense Shock," "Extreme Intensity Shock," "Danger: Severe Shock," and, at the top two switches, simply "XXX."

The generator was fake. No shock was ever delivered. But participants did not know this.

The Participants and the Confederate

Milgram advertised for male participants between 20 and 50 for a study on "memory and learning," paying $4.50 for one hour (roughly $44 in 2026 dollars). Participants were drawn from a cross-section of New Haven: factory workers, laborers, salespeople, engineers, teachers. They arrived at Yale's Linsly-Chittenden Hall.

On arrival, each participant met another "participant" — a mild-mannered, 47-year-old man named Mr. Wallace, described as an accountant. Wallace was actually a confederate, an actor following a precise script. A rigged drawing always assigned the real participant to the role of "Teacher" and Wallace to the role of "Learner."

The Learner was strapped into a chair in an adjacent room, electrodes attached to his wrist. The Teacher sat before the shock generator with an experimenter in a gray lab coat seated nearby. The experimenter's role was to issue scripted verbal prods whenever the Teacher hesitated.

Source: Milgram, S. (1974). Obedience to Authority: An Experimental View. Harper & Row.

The Protocol

The task was simple: the Teacher read a list of word pairs ("blue, box; nice, day; wild, duck") to the Learner via intercom. Then the Teacher read the first word of each pair, and the Learner pressed one of four buttons to indicate which of four options was the correct match. If the Learner answered incorrectly, the Teacher delivered a shock, increasing the voltage by 15 volts with each error.

As voltages increased, the Learner (following his script) delivered pre-recorded responses through the wall:

• At 75 volts: a grunt.
• At 120 volts: "Hey, this really hurts."
• At 150 volts: "Experimenter! That's all. Get me out of here. I told you I had heart trouble. My heart is starting to bother me. Get me out of here, please."
• At 300 volts: "I absolutely refuse to answer any more. Get me out of here. You can't hold me here. Get me out. Get me out of here."
• Above 330 volts: silence.

Silence, in Milgram's instructions, was to be treated as a wrong answer. Teachers were told to continue administering shocks even when the Learner had gone silent.

When participants expressed reluctance or tried to stop, the experimenter delivered one of four scripted prods in order:

1. "Please continue" (or "Please go on").
2. "The experiment requires that you continue."
3. "It is absolutely essential that you continue."
4. "You have no other choice, you must go on."

If the participant refused to continue after all four prods, the experiment ended. If not, it continued to 450 volts.

Source: Simply Psychology. Milgram Shock Experiment | Summary | Results | Ethics. [simplypsychology.org]

The Results

In the baseline condition — real participant as Teacher, Mr. Wallace as Learner, separated by a wall, Yale University setting — 65% of participants (26 out of 40) administered the full 450 volts. Every single participant went to at least 300 volts.

This stunned Milgram. Before running the experiment, he had asked Yale psychology students, other psychologists, and psychiatrists to predict the outcome. The most common prediction: fewer than 1–3% would go to the maximum.

Milgram ran 18 variations between 1961 and 1962, with over 600 participants. Obedience rates varied dramatically depending on conditions:

Condition	Obedience Rate
Experimenter in same room, Learner in adjacent room (baseline)	65%
Experimenter by telephone (not in same room)	20.5%
Learner in same room, visible to Teacher	40%
Teacher must physically hold Learner's hand to plate	30%
Study conducted in run-down office building (not Yale)	47.5%
Two confederate experimenters give conflicting orders	0%
Two confederate Teachers both refuse to continue	10%

The pattern was clear: obedience dropped when the authority figure was less present or less legitimate, when the victim was more visible, and when social support for defiance was available.

Source: Milgram, S. (1963). Behavioral study of obedience. Journal of Abnormal and Social Psychology, 67(4), 371–378.

Source: Wikipedia. Milgram experiment. [en.wikipedia.org]

Why People Obeyed: Milgram's Explanations

Milgram proposed two interlocking mechanisms.

The first was the agentic state — a psychological mode in which a person stops seeing themselves as a moral agent and begins to see themselves as an instrument of another's will. In the agentic state, responsibility for outcomes feels as though it has been transferred upward, to the authority who issued the instruction. This explains why so many participants continued while expressing verbal distress: they were behaving as though the experimenter, not they themselves, was responsible for the Learner's suffering.

The second was binding factors — elements of the situation that made it difficult to leave: the commitment already made by showing up, the incremental nature of the task (each shock was only 15 volts more than the last — a ratchet mechanism), the social awkwardness of defying someone in authority, and the legitimacy conferred by the Yale name and the lab coat.

Think of it like the "boiling frog" metaphor applied to moral decision-making. No single step felt catastrophic; the 330-volt switch was only 15 volts more than the 315-volt switch the participant had just accepted. By the time the stakes were obvious, participants were already deeply committed.

Source: Milgram, S. (1974). Obedience to Authority: An Experimental View. Harper & Row.

The 1963 Publication and Its Reception

Milgram's first paper appeared in the Journal of Abnormal and Social Psychology in 1963, before the variations were complete. The paper was a sensation and a scandal simultaneously. The American Psychological Association held up Milgram's membership application for a year while investigating the ethics of the study — participants had been deceived, had experienced genuine acute distress, and had been inadequately debriefed.

The ethical critique was serious. An estimated 600 participants left Yale believing they had actually shocked and potentially harmed a human being. Gina Perry, whose 2013 investigation examined Milgram's archives, estimated that many were not properly informed during debriefing that the shocks were never real. Some went home genuinely disturbed.

Source: Perry, G. (2013). Behind the Shock Machine: The Untold Story of the Notorious Milgram Psychology Experiments. Scribe Publications.

The Controversy: Cracks in the Foundation

Perry's Archival Investigation

Gina Perry spent years interviewing participants and examining Milgram's unpublished archives at Yale. Her 2013 book Behind the Shock Machine raised several significant concerns.

First, selective reporting. Milgram conducted more than 18 variations, but he reported only those that fit his narrative. Perry identified a "Relationship Condition" conducted in 1962 but never published: when the Teacher and Learner were relatives or close acquaintances, and the Learner protested, approximately 85% of participants refused to continue. That 85% disobedience rate — nearly a direct inversion of the famous headline finding — was apparently set aside.

Second, Perry concluded that "only half of the people who undertook the experiment fully believed it was real and of those, 66% disobeyed." If many participants suspected the shocks weren't genuine, the baseline obedience rate is measuring something different from genuine willingness to harm strangers — perhaps willingness to play along with a theater exercise.

Third, debriefing was inadequate in many sessions. Milgram's notes indicate that experimenters varied substantially in how they handled the end of each session.

Source: Perry, G. (2013). Behind the Shock Machine. Scribe Publications. [ir.ung.edu — peer review]

Source: Los Angeles Review of Books (n.d.). Psych, Lies, and Audiotape: The Tarnished Legacy of the Milgram Shock Experiments. [lareviewofbooks.org]

The Belief Problem

The question of whether participants "really believed" is not merely academic. The entire inferential chain — from "person administers shocks" to "ordinary people will harm innocents under authority" — depends on participants genuinely believing harm was occurring. If a substantial fraction suspected they were in a theater exercise, the study does not demonstrate what it claims to demonstrate.

Milgram himself acknowledged in his notes that some participants seemed "not fully convinced," but the published papers did not systematically analyze this variable.

Replication: Burger (2009)

In 2009, psychologist Jerry Burger at Santa Clara University conducted the first major replication of the Milgram paradigm in the United States since the 1970s. Ethical constraints had previously made full replication impossible — contemporary IRB standards prohibited inducing the level of distress Milgram's original participants experienced.

Burger's solution: stop at 150 volts, the point at which the Learner first explicitly demands to be released. This was the moral crossroads — the moment when continuing was most clearly wrong.

The result: 70% of participants (combining males and females) continued past 150 volts when the experimenter asked them to — nearly identical to the 82.5% Milgram had found at the same decision point in 1961. The gender difference was not statistically significant. Burger's conclusion: obedience rates at that critical juncture have not meaningfully changed in 45 years.

Burger also ran a condition where participants watched a confederate Teacher refuse to continue at 150 volts before they took the role. Even with that model of defiance, obedience remained high — 63% continued past 150 volts.

Source: Burger, J. M. (2009). Replicating Milgram: Would people still obey today? American Psychologist, 64(1), 1–11. [apa.org]

The Timeline

Date	Event
April 1961	Eichmann trial begins in Jerusalem; Milgram develops the experiment concept
July 1961	First sessions begin at Yale's Linsly-Chittenden Hall
1961–1962	18 variations conducted; ~636 participants total
1963	First paper published in Journal of Abnormal and Social Psychology
1963–1964	APA holds up Milgram's membership, citing ethics
1974	Obedience to Authority published — the full account
1984	Milgram dies of a heart attack, age 51
2004	Jerry Springer Game Show The Game of Death (French) — televised Milgram-style replication
2009	Burger partial replication published; 70% comply at 150 volts
2013	Perry publishes Behind the Shock Machine
2015	Film Experimenter (Michael Almereyda) released, featuring Peter Sarsgaard as Milgram

Ethical and Methodological Takeaways

The Milgram experiments remain among the most methodologically complex studies in psychology's history — simultaneously landmark and problematic.

The ethical problems are clear. The methodological problems are also real but do not erase the core finding: under controlled conditions, many people will continue to follow instructions even when those instructions cause visible distress to another person, especially when the authority is nearby and legitimate. Burger's replication, which passes contemporary ethical standards, supports this much.

What remains contested is the inferential leap from laboratory to life. Can obedience in a Yale lab, toward a researcher in a lab coat, for a study on "learning," predict behavior in war, genocide, or institutional cruelty? Milgram believed it could. Critics argue the analogy is too simple — that the Holocaust was produced by ideology, dehumanization, social structure, and escalating commitment over years, not the 45 minutes in a New Haven basement.

The Stanford Prison Experiment faced an analogous challenge: does the behavior of a small group of undergraduates in a basement generalize to institutions? Both studies ask the right question but generate evidence that cannot carry the interpretive weight placed on it.

Source: ResearchGate. Coverage of recent criticisms of Milgram's obedience experiments in introductory social psychology textbooks. [researchgate.net]

Where the Science Stands Today

Contemporary social psychology has moved toward a more nuanced account of obedience. The binary of "obey vs. defy" misses how people actually navigate authority: they slow down, they express reluctance, they seek social support, they look for exit routes. Milgram's own variation data showed this — social context was the most powerful predictor of behavior, far more than individual psychology.

The most robust modern finding is also the most uncomfortable: obedience is not a personality trait. It is a situational outcome. You cannot reliably predict in advance who will comply and who will refuse, based on personality tests, demographic characteristics, or stated values. This conclusion — arguably the genuine scientific contribution of Milgram's work — does not require the original study to be methodologically sound. It has been replicated and extended in better-controlled paradigms since.

Source: ResearchGate. Replicating Milgram (Burger 2009). [researchgate.net]

The 24 Variations Milgram Actually Ran: What the Headline Number Hides

Sixty-five percent is the number that made it into every textbook. It is also the number from exactly one of the 24 condition variants Milgram ran between 1961 and 1962. The other 23 conditions, which produced obedience rates ranging from 0% to 92.5%, were published in Milgram's 1974 book Obedience to Authority — a work that received serious academic attention but never generated a headline to match the original 65%.

Source: Milgram, S. (1974). Obedience to Authority: An Experimental View. Harper & Row.

The variation data is worth examining directly, because the spread of results tells a more precise and more interesting story than the baseline finding alone. When two confederate "teachers" both refused to continue, real participants followed their lead in 90% of cases — obedience collapsed to 10%. When a confederate peer actively pushed the teacher to keep going, obedience climbed to 92.5%. These are not minor perturbations around a stable trait. They are swings of 80 percentage points produced by changing who else was in the room.

The proximity conditions are equally striking. When the Learner was placed in the same room as the Teacher, visible rather than walled off, obedience dropped to 40%. When the Teacher had to physically hold the Learner's hand against a metal shock plate, it fell further to 21% — the closest the study came to placing the harm in direct physical contact. At the other extreme, when the experimenter gave instructions by telephone rather than sitting in the room, obedience dropped to 20.5%. Authority's power was directly proportional to its physical presence.

The gender condition produced a result that consistently surprises people: women obeyed at 65% — statistically identical to men. Milgram ran this variation partly because many observers assumed women would be less susceptible to authority pressure. They were not. Female participants reported more verbal distress during the procedure, but their behavior at the shock generator was the same.

One condition that never found its way into standard textbook accounts: when the experimenter was replaced by a regular man in street clothes rather than a researcher in a lab coat, obedience fell to roughly 20%. The lab coat and the institutional setting — Yale University — were doing substantial work that had nothing to do with the content of the instruction. The experiment was, in significant part, a study of institutional legitimacy.

Source: Reicher, S. D., Haslam, S. A., & Miller, A. G. (2014). What makes a person a perpetrator? The intellectual, moral, and methodological arguments for revisiting Milgram's research on the influence of authority. Journal of Social Issues, 70(3), 393–408.

The selective amplification of the 65% figure reflects a broader problem in how psychology research travels from the lab to the culture. A single dramatic number attached to a clean moral story propagates easily. The full variation dataset — which shows that the same people who obeyed under one condition defied under another — requires more careful attention to absorb. It is also the more important finding: obedience is not a stable trait that some people have and others lack. It is an output of a specific configuration of authority, visibility, social support, and institutional framing. Change the configuration and you change the behavior. The headline number hides this, and the 23 other conditions reveal it.

FAQs

Q: Were the electric shocks in the Milgram experiment real? A: No. The shock generator was a prop, and Mr. Wallace was an actor. No participant was ever harmed physically. However, participants were not told this during the experiment, and many experienced genuine acute distress believing they were harming someone.

Q: What was the significance of 450 volts specifically? A: 450 volts is well above the threshold for cardiac arrest in a human being. The "XXX" label on the final switches was chosen to signal unambiguous lethality. The fact that 65% of participants in the baseline condition reached that level was the headline finding of Milgram's 1963 paper.

Q: Did Milgram's results prove that the Holocaust could have been carried out by ordinary people? A: Milgram believed they provided evidence for this, and it was his stated motivation. Most historians and psychologists now think the analogy is too simple. Holocaust perpetration involved years of ideological indoctrination, dehumanization of victims, peer pressure, career incentives, and escalating commitment — not the 45-minute dynamic of a Yale psychology study.

Q: Were women tested? Did they obey at the same rate? A: Milgram ran a variation with female participants. They showed an obedience rate statistically similar to men at 65%, though they expressed more verbal distress during the procedure. Burger's 2009 replication also found no significant gender difference.

Q: Why isn't this experiment replicated more often? A: Contemporary institutional review boards (IRBs) will not approve studies that induce the level of psychological distress Milgram's baseline condition created. Burger's 2009 study was approved because it stopped at 150 volts and included more rigorous real-time screening for distress. The full protocol cannot be replicated under current ethical standards.

Q: What does the Milgram experiment tell us that actually holds up? A: The clearest finding is that situational factors — proximity of authority, legitimacy of the institution, presence of social support for defiance, and physical proximity to the victim — powerfully influence compliance with harmful instructions. Individual character is a poor predictor. This conclusion survives the methodological critiques and has been supported by subsequent research.

---

Read next:

• The Stanford Prison Experiment
• The Double-Slit Experiment
• What Is the Bystander Effect?