Social Psychology
Milgram Obedience Experiments
Ordinary people delivering "lethal" shocks under authority — 1961-1963 Yale studies
Stanley Milgram (Yale, 1961-1963) tested how far ordinary Americans would obey authority. Participants believing they were in a "learning" study delivered shocks of increasing voltage to a confederate "learner" who screamed and pleaded. An experimenter in a lab coat instructed them to continue. About 65% of subjects went to the maximum 450 volts, labeled XXX. Milgram had predicted under 1% would do so. The findings — published as Behavioral Study of Obedience (1963) and Obedience to Authority (1974) — emerged in the shadow of the Eichmann trial and reframed the Holocaust question from German national character to universal human susceptibility. Replications, reanalyses (Gibson 2013), and ethical critiques continue. The exact obedience rate varies (30-65%) by condition, but the basic finding — distressed compliance under perceived legitimate authority — is robust.
- ResearcherStanley Milgram, Yale (1961-1963)
- Headline result~65% delivered maximum 450V shocks
- Predicted rate<1% (psychiatrists' estimate)
- Maximum voltage label450V "XXX"
- Ethical legacyHelped establish IRB review (1974 NRA, Belmont 1979)
- Famous variantProximity, peer rebellion, two-experimenter conditions
Interactive visualization
Press play, or step through manually. The visualization is yours to drive — try it before reading on.
Watch the 60-second explainer
A condensed visual walkthrough — narrated, captioned, under a minute.
Why Milgram matters
- Institutional design. Hierarchies must build in checks against authority-driven harm.
- Medical ethics. Hospital chain-of-command failures (wrong-site surgery) match the Milgram pattern.
- Military training. Lawful disobedience, refusal to follow illegal orders, taught as core duty.
- Compliance cultures. Whistleblower protections counter agentic deference.
- Research ethics. The studies shaped IRB systems still in use.
- Cult dynamics. Gradual escalation and authority concentration parallel Milgram conditions.
- Holocaust education. Reframes "monsters" as ordinary actors in monstrous systems.
Common misconceptions
- Real shocks were delivered. They weren't; the learner was a confederate.
- 65% applies universally. Rates ranged from 0% to 93% across conditions.
- Obedience equals blindness. Recent work suggests engaged identification, not zombie compliance.
- Only sadists comply. Most participants showed visible distress while continuing.
- Couldn't happen today. Burger 2009 and replications show comparable rates.
- Milgram's data was clean. Recent archival work (Perry) revealed undisclosed methodological issues; the qualitative truth survives, the statistics need caveats.
Frequently asked questions
What was the procedure?
A naive "teacher" was paired with a confederate "learner" (always assigned the latter role via rigged drawing). The teacher read word pairs; for each error, they delivered a shock from a console rising in 15V increments from 15V to 450V, labeled "Slight Shock" through "Danger: Severe Shock" and "XXX." No actual shocks occurred; the learner produced scripted screams, demands to stop, and eventually silence. The experimenter, in a gray lab coat, gave four scripted prods if the teacher hesitated.
What did Milgram find?
In the baseline study (Experiment 5), 26 of 40 participants (65%) administered the maximum 450V despite the learner's pleading and apparent collapse. All went to at least 300V. Many showed acute distress — sweating, trembling, nervous laughter — but continued under instruction. Milgram had surveyed 14 Yale psychiatry seniors who predicted only the most pathological 1-2% would proceed to maximum.
Why did people obey?
Milgram proposed an "agentic state" — under legitimate authority, people shift from autonomous actor to agent of another's will, deflecting responsibility upward. Other contributors. (1) Gradual escalation — the foot-in-the-door problem. (2) Unfamiliar situation, no clear script for refusal. (3) Authority of Yale and science. (4) Physical distance from victim. (5) Time pressure preventing reflection. Conditions reducing any of these reduced obedience.
What were the variations?
Milgram ran 24 conditions. Obedience dropped sharply when. (1) The learner was in the same room (40%). (2) Teacher had to physically place the learner's hand on a shock plate (30%). (3) Two experimenters disagreed (≈0%). (4) Peers refused to continue (10%). (5) The experiment moved from Yale to a generic Bridgeport office (47%). It rose when authority was reinforced or victim made abstract. Variation matters — the 65% figure is a single point on a wide range.
Was it ethical?
Highly contested. Diana Baumrind (1964) argued the deception caused real psychological harm. Milgram countered with follow-up surveys showing most participants reported the experience as valuable. The studies influenced US federal regulations on human-subjects research (National Research Act 1974, Belmont Report 1979) and current IRB review would not approve them. Recent reanalyses (Gibson, Perry) report some participants suffered lasting distress not captured in published debriefs.
Have replications confirmed it?
Yes, with caveats. Burger (2009) ran an ethics-modified version stopping at 150V — 70% of participants kept going past that threshold, matching Milgram's data. Doliński et al. (2017) Polish replication showed similar rates. Cross-cultural studies generally find obedience to legitimate authority, though absolute rates vary. The phenomenon is real; debate centers on interpretation, especially whether it's "blind obedience" or "engaged followership" with shared identity.
How does it explain atrocities?
Milgram framed his work against the Eichmann trial — Hannah Arendt's "banality of evil." Implication: the Holocaust required not sadists but ordinary people obedient to bureaucratic authority. Recent scholarship (Haslam, Reicher) argues participants often actively identified with the experimenter's mission rather than obeying mindlessly — "engaged followership." Both readings highlight that systems can recruit ordinary moral agents into harmful action, with implications for institutional design.