A landmark of social psychology research was “The Milgram Experiment,” but a new look at the audio tapes and other evidence collected during that experiment suggests that we may have been interpreting it incorrectly. Here is the Wikipedia summary of the experiment, showing how it is typically portrayed:

Yale University psychologist Stanley Milgram… intended to measure the willingness of study participants to obey an authority figure who instructed them to perform acts conflicting with their personal conscience. Participants were led to believe that they were assisting in a fictitious experiment, in which they had to administer electric shocks to a "learner". These fake electric shocks gradually increased to levels that would have been fatal had they been real.

The experiments unexpectedly found that a very high proportion of subjects would fully obey the instructions, with every participant going up to 300 volts, and 65% going up to the full 450 volts.

I don’t know about that “unexpectedly” part. I think the researchers suspected, in the wake of e.g. the Holocaust, that people were generally willing to obey awful instructions in ways that they failed to account for. Their experiment was designed to answer not whether but how much.

Interpreting the results

The results have since been interpreted as a kind of cynicism or caution about human nature, and about people’s tendencies to let their consciences be silenced by the trappings of authority.

But such takes may have been too optimistic.

Milgram interviewed his subjects after the experiment and found that those who stopped giving shocks felt that they were responsible for what they were doing, while those who continued giving shocks felt that the experimenter (the one giving the instructions to the subject) was responsible. Milgram theorized that his subjects, in the presence of an authority figure, stepped into a corresponding role: the “agentic state.” Once you are in that state, you stop considering yourself responsible for what you are doing and for the effects of what you are doing, and judge your actions only on whether you are doing it according to how the authority wants it done.

Arne Johan Vetlesen, in Evil and Human Agency (2005), pointed out that there is another possible interpretation: Milgram’s subjects may have had genuine sadistic impulses. In subjecting their victims to pain, they were not being somehow coerced by their situation to do things they would ordinarily not want to do, but that they were being allowed by their situation to do things they were ordinarily inhibited from doing.

He quoted Ernest Becker, who took a second look at Freud’s take on mob violence:^[1]

…[M]an brings his motives in with him when he identifies with power figures. He is suggestible and submissive because he is waiting for the magical helper. He gives in to the magic transformation of the group because he wants relief of conflict and guilt. He follows the leader’s initiatory act because he needs priority magic so that he can delight in holy aggression. He moves in to kill the sacrificial scapegoat with the wave of the crowd, not because he is carried along by the wave, but because he likes the psychological barter of another life for his own: “You die, not me.” The motives and the needs are in men and not in situations or surroundings.

And Hannah Arendt, whose examination of the Adolf Eichmann trial was going on at around the same time as the early Milgram experiments, warned that the excuse of “obedience” (as used by the compliant Milgram subjects to explain their actions after-the-fact, and secondarily by Milgram himself in his theory) was not an explanation but a “fallacy”:

Only a child obeys. An adult actually supports the laws or the authority that claims obedience.^[2]

A new review of the evidence

Now David Kaposi and David Sumeghy have gone back through the audio tapes and other documentation preserved from the original Milgram experiments.^[3]

What they found was that the “obedient” subjects were not in fact very compliant at all. Indeed none of them actually followed the experimental procedures they had been instructed to comply with.

Only a few of the subjects complied with the experimental procedures they were given in full, and all of them were among those who eventually refused to continue with the experiment.

(1) no entirely “fully obedient” participant fully obeyed the procedures Milgram’s experimenter instructed them to do; (2) violations of the procedures occurred on average 48.4% of the time in “fully obedient” sessions; and (3) violations occurred significantly more frequently in “fully obedient” than in the obedient phase of “disobedient” sessions.

Tellingly, these procedural violations were not efforts to avoid giving shocks, but actually increased the likelihood that an opportunity to give another shock would arise:

The most frequent violation in obedient sessions involved reading the memory test questions over the simulated screams of the learner. Doing this effectively guaranteed that the learner would fail the test and receive another shock. By talking over the protests, the obedient subjects abandoned the [ostensible] goal of testing memory and simply facilitated continuous shocks.^[4]

The implication is that when Milgram interviewed the “obedient” subjects after the experiment was over, these subjects represented themselves as having merely obeyed because this was an excuse for their behavior that had been dangled before them temptingly during the experiment, and they anticipated that this excuse would be accepted. Milgram, by being willing to accept this excuse at face-value, in effect validated it and cooperated with the subjects in whitewashing their surrender to sadistic temptation.

1. ^{^}

   See also H.L. Mencken (Damn: A book of Calumny) making a similar point about the supposedly hypnotic influence of the mob:

   The numskull runs amuck in a crowd, not because he has been inoculated with new rascality by the mysterious crowd influence, but because his habitual rascality now has its only chance to function safely. In other words, the numskull is vicious, but a poltroon. He refrains from all attempts at lynching a cappella, not because it takes suggestion to make him desire to lynch, but because it takes the protection of a crowd to make him brave enough to try it.

   ⋮

   In other words, the particular swinishness of a crowd is permanently resident in the majority of its members — in all those members, that is, who are naturally ignorant and vicious — perhaps 95 per cent. All studies of mob psychology are defective in that they underestimate this viciousness. They are poisoned by the prevailing delusion that the lower orders of men are angels. This is nonsense. The lower orders of men are incurable rascals, either individually or collectively. Decency, self-restraint, the sense of justice, courage — these virtues belong only to a small minority of men. This minority never runs amuck. Its most distinguishing character, in truth, is its resistance to all running amuck. The third-rate man, though he may wear the false whiskers of a first-rate man, may always be detected by his inability to keep his head in the face of an appeal to his emotions. A whoop strips off his disguise.

   
   

2. ^{^}

   Moral Responsibility under Totalitarian Dictatorships

3. ^{^}

   David Kaposi & David Sumeghy “From legitimate to illegitimate violence: Violations of the experimenter's instructions in Stanley Milgram’s ‘obedience to authority’ studies” Political Psychology (2026)

4. ^{^}

   Karina Petrova “Audio tapes reveal mass rule-breaking in Milgram’s obedience experiments” PsyPost 28 March 2026