What the work claims
That the simulated prison situation—not individual personality—caused extreme, abusive behaviour. That roles, rules, symbols, group identity, and situational validation can elicit actions ordinarily repulsive to individuals. That situational forces exert far greater influence on behaviour than most people appreciate.
How it was done
Twenty-four psychologically stable male college students were randomly assigned as prisoners or guards in a mock prison in the basement of Stanford’s psychology building. The experiment began with real Palo Alto police conducting mock arrests. Prisoners were confined to 6-by-9-foot cells with cots; solitary confinement occurred in a small unlit closet. It was funded by the US Office of Naval Research to investigate guard–prisoner conflict and antisocial behaviour.
What holds up
The rapid role adaptation, escalating psychological abuse over five days, Zimbardo’s own immersion as superintendent, and the intervention by Christina Maslach—all are directly attested. The experiment ended on day six—not due to external review, but because Zimbardo recognised his compromised position after her challenge.
What does not
It does not establish that personality is irrelevant to behaviour. It does not show that situational forces always override individual agency. It does not demonstrate how prisons operate in reality—only how one ad hoc simulation unfolded under specific, uncontrolled conditions with no follow-up or independent oversight.
Why it matters beyond the lab
It exposed how quickly institutional roles can override moral self-constraint—even among educated, screened volunteers—and how authority structures invite unchecked control. Its legacy is not predictive power, but diagnostic: a warning about design, not a blueprint for human conduct.
Is it worth your time
Yes—if you need to understand how situational power operates in institutions, and how easily ethical boundaries collapse when roles are uncritically adopted. Not if you seek generalisable data on human nature: the sample was tiny, unblinded, unreplicated, and lacked controls for demand characteristics or experimenter influence.