Put good people into a bad role, and they turn bad. That is the lesson generations have drawn from the Stanford Prison Experiment — next to Milgram, the most famous study in social psychology. There is just one problem: the lesson does not survive scrutiny. Today the Stanford experiment is less a proof than a case study in how a scientific myth is made.
The Setup
In the summer of 1971, the psychologist Philip Zimbardo built a mock prison, the "Stanford County Jail", in the basement of the Stanford psychology building. 24 male students, previously screened as psychologically healthy, were recruited for $15 a day and assigned by coin toss to be "guards" or "prisoners". The "prisoners" were "arrested" at home by real police, deloused, put into numbered smocks; the "guards" were given uniforms, mirrored sunglasses and batons. Two weeks were planned.
It was called off after six days. The guards, so the story goes, quickly became abusive and at times cruel; the prisoners apathetic and broken. "Prisoner" No. 8612, Douglas Korpi, suffered an apparent breakdown after about a day and a half. Zimbardo's conclusion made him world-famous: not evil characters but the power of the situation and the role had turned normal young men into tormentors. From this idea came his book The Lucifer Effect (2007), and Zimbardo drew the line all the way to the torture images of Abu Ghraib.
The Collapse of the Legend
As catchy as the story is, the evidence beneath it has proven brittle. The heaviest blow came from the French researcher Thibault Le Texier, who worked through the Stanford archive and interviewed 15 participants. His paper Debunking the Stanford Prison Experiment (American Psychologist, 2019) shows:
- The guards were coached, not observed. In the orientation the team explicitly told the guards the desired outcome — they wanted to create the psychological conditions that lead to "mob behaviour" and violence. The brutality was therefore not the spontaneous result of a role but a requested performance. Zimbardo's assistant David Jaffe, as "warden", actively pushed guards to be tougher.
- The famous breakdown was staged. Decades later Douglas Korpi told the author Ben Blum (The Lifespan of a Lie, 2018) that he had faked the episode to get out of the study — he wanted to study for an upcoming exam, and the guards had refused him his books. His verdict on his own scene: "Anybody who is a clinician would know that I was faking."
- The cruellest guard was acting too. Dave Eshelman, whom the prisoners called "John Wayne", later said he had consciously played his sadistic character — partly inspired by a film — to give the experimenter what he evidently wanted to see.
- Methodologically it was not an experiment. There was no control group and no real hypothesis test; Zimbardo himself ran the scene as "superintendent" and shaped it. The results were first spread through the media, not through the usual peer-reviewed publication — an approach that put the stage before the method.
Little is left of the "proof": what Stanford showed was not that roles automatically brutalise people, but how far people will go to do what an authority visibly expects of them — when they identify with its purpose. That is exactly the finding the honest reading of Milgram's obedience experiment yields.
The Counter-Test: the BBC Prison Study
If roles really do lead automatically to tyranny, it should be repeatable. In 2002 Stephen Reicher and Alex Haslam did exactly that — as a scientifically monitored study, broadcast by the BBC. The result was the opposite of Stanford:
- The "guards" did not identify with their role as a matter of course and were reluctant to exercise authority.
- There was no "natural" drift into cruelty. For a time the participants even formed an egalitarian system.
- What mattered was not the role but the shared identity: whether a group experienced itself as a capable "we" determined whether it submitted or resisted.
Reicher and Haslam framed it as "rethinking the psychology of tyranny": people do not blindly follow a role; whether oppression arises depends on identification and leadership — not on a uniform. The same logic of shared identity is known from conformity research.
Zimbardo's Rebuttal
In fairness: Zimbardo has rejected the criticism. He argues that an orientation about expectations does not invalidate the observation, that Korpi's later denial is retrospective reinterpretation, and that the experiment never claimed to be a strictly controlled laboratory study but a demonstration. One can grant that — only a demonstration does not carry the strong claim that was drawn from Stanford for decades. That is exactly the difference between an impressive piece of theatre and a robust finding.
What Really Remains
The Stanford Prison Experiment is no proof that a torturer sleeps in everyone, waiting only for the role to wake him. But it is not simply nothing either. The honest balance matches the whole cluster around this question:
- It is not the role as such that brutalises, but identification with an authority and its purpose — amplified by mirrored glasses, uniforms, anonymity.
- People are not determined. Whether someone goes along or refuses depends on situation, leadership and shared identity — and is therefore changeable.
- One's own judgement does not switch off because a role overwrites it, but because one voluntarily adopts someone else's framing without examining it.
Read correctly, then, Stanford fits the same series as Milgram, Hannah Arendt's thoughtlessness and Dietrich Bonhoeffer's "stupidity": evil on a large scale needs no monsters and no magic roles, only people who surrender their own examining.
And the uncomfortable point remains, the one that cuts both ways: for decades Stanford was a comfortable, canonical "proof" — in every textbook, every documentary. That it took fifty years to debunk shows how tenaciously a well-told finding survives. The same care this site demands toward uncomfortable data (near-death, mediumship) applies just as much to the comfortable ones. Examine first, then believe — especially when everyone already thinks they know.
Sources
- Haney, C., Banks, C. & Zimbardo, P. (1973): Interpersonal Dynamics in a Simulated Prison. International Journal of Criminology and Penology — the original account.
- Zimbardo, P. (2007): The Lucifer Effect. — Zimbardo's situationist interpretation and the Abu Ghraib link.
- Le Texier, T. (2019): Debunking the Stanford Prison Experiment. American Psychologist 74(6), 823–839 — archival work, coaching of the guards, demand characteristics.
- Blum, B. (2018): The Lifespan of a Lie. Medium — Douglas Korpi's account of the faked breakdown.
- Reicher, S. & Haslam, S. A. (2006): Rethinking the psychology of tyranny: The BBC prison study. British Journal of Social Psychology 45 — no automatic role conformity.
- Zimbardo, P.: Response to Recent Criticisms (prisonexp.org) — the rebuttal.