sciencebriefs
13:00in productionCh. 1 · A trial in Jerusalem, an experiment in New Haven/ 13:00 · ceiling 15 min
Neuroscience

Milgram experiment

In 1961, months after Adolf Eichmann's trial opened, Stanley Milgram found 65 per cent of volunteers at Yale would deliver what they believed was a 450-volt shock to a stranger simply because told to continue — far more than psychiatrists had predicted.

Stanley Milgram began his obedience experiments at Yale in 1961, prompted by the trial of Adolf Eichmann and the question of how ordinary people could take part in atrocities under orders. Volunteers, believing they were administering shocks in 15-volt increments up to 450 volts to a learner in another room, kept going far more often than experts had predicted: 65 per cent reached the maximum, against a psychiatric estimate before the study of under 4 per cent doing so. Obedience varied sharply with the setting, from 21 per cent when instructions came by telephone to 65 per cent in the standard version, and the study drew immediate ethical criticism over the distress it caused participants who believed they were really hurting someone, criticism that has shaped research ethics rules ever since.

Chapters & takeaways6
  1. 0:08
    A trial in Jerusalem, an experiment in New Haven

    Milgram began the study in 1961, three months into Adolf Eichmann's trial, to ask how ordinary people follow orders that harm others.

  2. 2:10
    Two slips, both rigged

    Participants believed a draw decided who would be teacher and who would be learner, but both slips read teacher, guaranteeing the volunteer administered the shocks.

  3. 4:20
    A prediction that missed by a mile

    Psychiatrists surveyed beforehand expected well under four in a hundred people to reach the maximum shock; sixty-five in a hundred actually did.

  4. 6:30
    Obedience that depended on distance

    Compliance fell from 65 per cent in the standard setup to 21 per cent when the experimenter gave instructions only by telephone.

  5. 8:40
    A study that changed how studies are run

    Diana Baumrind's 1964 critique argued Milgram owed his distressed participants more protection than informed consent alone provided, feeding into today's research ethics rules.

  6. 10:50
    Same result, tighter rules

    Jerry Burger's 2009 partial replication, stopped early to meet modern ethics standards, found obedience rates close to Milgram's originals.

Worth your time?

Yes. Study the whole thing.

4.5/ 5
What works
  • the specific numbers, the predicted 3.73 per cent versus the actual 65 per cent, make the result concrete rather than a vague claim about human nature
  • the variations across settings, especially the drop to 21 per cent for telephone instructions, show the effect responding to real conditions rather than being fixed
  • Burger's 2009 replication under stricter ethical limits gives the original finding a genuine, modern check rather than leaving it as a historical curiosity
What does not
  • later critic Gina Perry argued in 2012 that many participants did not fully believe the setup was real, complicating what the obedience rate actually measures
  • current researchers including Haslam and Reicher argue engaged support for the experiment's scientific goals, not blind obedience, may explain much of the behaviour
Study it if
  • anyone who has cited this experiment secondhand and never seen the actual percentages
  • readers interested in how a single controversial study reshaped the ethical rules governing all human research
  • anyone curious whether a 1961 finding about obedience still holds up under modern, more constrained replication
Skip it if
  • readers wanting a single, uncontested explanation for why participants kept going, since researchers now dispute whether it was really about obedience to authority
  • anyone looking for reassurance that people are inherently good or inherently compliant; the variation across conditions argues against either simple story
The written brief4 min read

A trial in Jerusalem, an experiment in New Haven

The claim under test was stark: that ordinary people, given no threat and no direct coercion, would inflict serious harm on a stranger simply because a person in a position of authority told them to keep going. Stanley Milgram set up the study at Yale in 1961, three months after Adolf Eichmann’s trial began in Jerusalem, explicitly motivated by the question of how ordinary Germans could have participated in the Holocaust under orders. Volunteers were told they were taking part in a study of learning and punishment, and were assigned, through a rigged draw in which both slips actually read teacher, to administer what they believed were real electric shocks to another participant, in fact an actor, whenever that person answered a word-pairing question incorrectly.

Two slips, both rigged

The shocks were presented as rising in 15-volt increments from 15 up to 450 volts, with the actor in the learner’s role delivering pre-recorded protests, banging on the wall, and eventually falling silent as the supposed voltage climbed, while the actual experimenter, present in the room with the volunteer, instructed them to continue whenever they hesitated. No real shocks were ever delivered. Before running the study, Milgram surveyed psychiatrists about how far they expected participants to go, and they predicted that under four in a hundred, precisely 3.73 per cent, would continue past 300 volts. In the actual experiment, every participant in the original condition reached at least 300 volts, and 65 per cent went all the way to the maximum 450 volts, a result so far outside expert prediction that it became the study’s defining number.

A prediction that missed by a mile

What has held up, across dozens of later replications catalogued in a meta-analysis by Thomas Blass, is that high rates of obedience in this basic setup are a real, repeatable phenomenon rather than a one-off artefact of Milgram’s particular sample, with obedience across studies ranging from 28 to 91 per cent, averaging 61 per cent in the United States and 66 per cent outside it. The result also proved sensitive to circumstance in ways that have replicated consistently: obedience fell to 47.5 per cent when the study moved to a modest office setting rather than Yale itself, to 40 per cent when the volunteer and the learner were in the same room, and to 21 per cent when the experimenter delivered instructions only by telephone rather than in person. Jerry Burger’s 2009 partial replication, which stopped at 150 volts to satisfy modern ethical requirements, found obedience rates close to Milgram’s own from decades earlier.

Obedience that depended on distance

What has not held up without challenge is a simple reading of the result as proof that people blindly obey authority. Gina Perry argued in 2012 that only about half of participants fully believed the shock apparatus was genuine, and that among those who did believe it, 66 per cent actually refused to continue, a very different picture from the headline 65 per cent obedience figure if belief in the setup is factored in. More recent researchers, including Alexander Haslam and Stephen Reicher, have proposed that many participants continued not out of blind obedience but out of what they call engaged followership, a active identification with the experimenter’s scientific goals rather than passive submission to an order. The ethical criticism, led by Diana Baumrind in 1964, focused less on whether the result was real and more on whether Milgram’s informed consent adequately protected participants who showed visible signs of distress, sweating and trembling, during the study itself.

A study that changed how studies are run

The stakes of the finding extend well beyond the psychology laboratory: obedience research of this kind has been invoked to help explain events including the My Lai massacre and to inform how police interrogation and military command structures are trained to guard against unquestioning compliance. The ethical fallout mattered just as much practically, since the criticism the study attracted, including an initial delay in Milgram’s American Psychological Association membership over concerns about his methods, fed directly into the development of institutional review boards and informed consent standards that now govern human research broadly. Milgram himself reported that 84 per cent of former participants said afterward they were glad or very glad to have taken part, a figure defenders cite against claims of lasting harm, though critics regard it as insufficient justification on its own for the distress caused during the study itself.

Same result, tighter rules

Yes, and reading past the popular summary is worth the time specifically because the details complicate the story people usually repeat. The bare statistic, that most people will shock a stranger if told to, is only the entry point; the more interesting material is in how sharply that rate moved with distance, setting and physical presence, and in the genuine, ongoing argument among researchers about whether obedience to authority is even the right description of what participants were doing. It is also a rare case where the ethical controversy is as consequential as the scientific finding itself, since the rules that now protect participants in psychological research trace directly back to the objections this particular study provoked.

Same field · Neuroscience4 of 45
Up next in Science

Oil drop experiment

· 9:54

Charge is not smooth — it clicks, like coins in a jar.

9:54