How BBC recreated a highly controversial 1974 psychology prison experiment to test human obedience
The idea of recreating one of psychology’s most controversial experiments for television should have been unworkable from the outset. When the BBC announced in 2002 that it would run a controlled prison simulation as a documentary series, it was immediately compared to the 1971 Stanford Prison Experiment, a study so widely criticised for its ethics that it is now taught as a warning as much as a finding. The programme, titled The Experiment, set out to revisit the same question, how ordinary people behave when given power over others, but under conditions designed to avoid the failures of the original.
What the Stanford experiment set out to prove, and what went wrong
In August 1971, psychologist Philip Zimbardo and his colleagues built a mock prison in the basement of Stanford University. Twenty-four male students, selected from a larger pool of volunteers screened for physical and psychological health, were randomly assigned roles as “guards” or “prisoners.” They were paid $15 a day and told the study would last up to two weeks.The objective was grounded in a broader line of research on obedience and authority, building on work such as Stanley Milgram’s famous obedience experiment , where participants obeyed authority by administering supposed electric shocks to others. Zimbardo wanted to test whether behaviour could be shaped by situation alone, whether psychologically stable individuals would adopt the behaviours expected of their assigned roles inside a prison-like system.The simulation was built around a carefully designed structure, where prisoners were held in small cells, identified by numbers rather than names, and subjected to routines that simulated loss of autonomy. Guards worked in shifts and were given broad authority to maintain order, though they were instructed not to use physical violence. Cameras and microphones recorded interactions throughout.Within days, the situation deteriorated. Shocking evidence emerged that guards were becoming increasingly aggressive and dehumanising toward prisoners. Participants showed signs of acute stress, anxiety, emotional breakdowns and withdrawal, and five prisoners had to be released early. Zimbardo himself, who had taken on the role of prison superintendent, became absorbed in the simulation and overlooked the abusive behavior of the jail guards until graduate student Christina Maslach voiced objections to the conditions in the simulated prison and the morality of continuing the experiment.
The experiment, intended to run for 14 days, was stopped after six. It later became one of the most cited studies in psychology, often used to support the idea that people conform to roles and that situations can override individual personality. At the same time, it has been criticised on multiple grounds: lack of ethical safeguards, inadequate informed consent, the psychological harm caused to participants, and questions about whether guards were implicitly encouraged to behave harshly. By modern standards, it would not be approved under established research ethics frameworks.
Why the BBC attempted it again
Three decades later, psychologists Alex Haslam and Steve Reicher worked with the BBC to design a new study that would revisit the same core question under stricter scientific and ethical conditions. Their aim was not simply to replicate Zimbardo’s findings, but to test them. Fifteen male participants were selected and placed in a purpose-built prison environment inside a television studio in Elstree. As in the original, they were randomly assigned roles as guards or prisoners. The study was scheduled to run for eight days and was continuously filmed for broadcast.The experiment introduced tighter safeguards to avoid the failures of the original, operating under independent ethical oversight, allowing participants to withdraw at any time, ensuring continuous psychological monitoring, and keeping researchers from taking on direct authority roles within the system. The objective was more specific than in 1971. Haslam and Reicher wanted to examine how inequality is maintained or challenged, whether people accept hierarchical roles or resist them, and under what conditions authority becomes stable or collapses.
What actually happened inside the BBC study
The outcome did not follow the trajectory of the Stanford experiment. From the outset, the guards struggled to form a cohesive identity. They were reluctant to assert authority and appeared uncomfortable enforcing discipline. Without a shared sense of purpose or group cohesion, their position weakened. The prisoners, by contrast, began to develop a stronger collective identity. Over time, they coordinated their actions, questioned the legitimacy of the guards’ authority, and resisted the imposed hierarchy. This shift became more pronounced as the study progressed.
The BBC prison experiment showed prisoners forming alliances, refusing instructions, staging a prison break, and later attempting a self-governing commune.
By the sixth day, the structure had effectively broken down, culminating in a prison break staged by participants that rendered the guard–prisoner regime unworkable. In its place, they attempted to form a self-governing commune based on shared decision-making, but it quickly collapsed amid internal tensions, particularly among those who had led the earlier resistance. A smaller group then proposed creating a new regime, with themselves as guards, this time intending to impose a stricter and more authoritarian structure.At that point, researchers intervened and ended the study early, as the emerging dynamics suggested a shift toward a more extreme system that could pose risks to participants’ well-being.
What the BBC study found, and why it matters
The BBC study stood in contrast to the Stanford experiment, finding no evidence that individuals naturally conform to roles of authority or submission. Power did not automatically produce tyranny. Instead, behaviour depended on group dynamics, particularly whether individuals identified with their role and whether they could form a cohesive group around it. This aligns with the psychological concept of deindividuation, where a person’s sense of individual identity becomes submerged within a group, making them more susceptible to collective behaviour seen in settings such as protests or crowd movements, where ordinary individuals can sometimes act in more extreme or uncharacteristic ways. The guards’ failure was not a refusal of authority in principle, but a lack of shared identity. Without cohesion, their authority remained fragile. The prisoners’ ability to challenge the system emerged from the opposite condition: a growing sense of collective identity that allowed them to act together. These findings led Haslam and Reicher to argue that tyranny is not an inevitable outcome of power. It depends on social conditions, particularly whether a dominant group can organise itself and whether those subject to it accept or resist that structure.
The study was later published in academic journals and is often cited as a direct challenge to the conclusions drawn from the Stanford experiment. It shifted the focus from individual conformity to group processes, suggesting that leadership, identification and collective behaviour are central to understanding how systems of power operate.
Two experiments, two conclusions
Placed side by side, the two studies describe different mechanisms. The 1971 experiment suggested that roles and situations can drive individuals toward extreme behaviour, even in the absence of prior tendencies. The 2002 study argued that roles alone are not enough, that power depends on whether people believe in it, organise around it, and accept its legitimacy.Both studies have important limitations and cannot fully replicate real-world institutions. A key issue in each is lack of ecological validity: the artificial settings, whether a simulated prison or a controlled behavioural environment, do not capture the complexity, pressures and unpredictability of real prison life or authority systems. As a result, while they offer insight into behaviour under structured conditions, their findings are constrained by the environments in which they were produced.
