THE MILGRAM OBEDIENCE EXPERIMENT

When authority asks us to cross a moral line

The experiment placed personal conscience in direct conflict with the demands of legitimate-seeming authority.

A man sits before a machine with thirty switches. In the next room, another person is supposedly taking part in a memory test. Each mistake must be followed by a stronger electric shock. The man at the controls hesitates. He protests. The researcher behind him calmly insists that the experiment must continue.

No shocks are actually being delivered. The “learner” is an actor, the machine is a convincing prop, and the true subject of the experiment is the person being told to press the switches. How far will he go when scientific authority asks him to continue?

Stanley Milgram’s obedience studies became some of the most famous-and most troubling-experiments in psychology. Their importance does not lie in a simple claim that people are naturally cruel. Most participants were visibly distressed, many challenged the researcher, and a substantial minority refused. The deeper question is how a structured situation can make harmful-seeming action feel incremental, authorised and difficult to stop.

Obedience is not the same as conformity

The distinction matters. In Solomon Asch’s experiments, people faced pressure to agree with a unanimous group even though no one explicitly ordered them to do so. Milgram studied obedience: an authority figure directly instructed one person to act against another person’s wishes.

The participant therefore faced two competing demands. The researcher invoked duty to the experiment, while the learner’s protests called for the procedure to stop. Continuing could be described as obedience to authority; refusing was an assertion of personal responsibility.

The experiment participants thought they had joined

Milgram recruited 40 men between 20 and 50 years of age from the New Haven area. They represented a range of occupations and educational backgrounds. Each was paid for attending and was told that the study concerned the effects of punishment on learning.

After meeting another apparent volunteer, the men drew lots to become either the “teacher” or the “learner.” The draw was arranged so that the genuine participant always became the teacher. The learner, who worked with the research team, was placed in an adjacent room and connected to what appeared to be an electrode. The participant received a small sample shock, strengthening the plausibility of the equipment.

Participants believed they had entered a study of memory and punishment; the learner was actually a trained confederate and received no shocks.

A machine designed to make escalation feel orderly

The teacher sat before a simulated shock generator containing 30 switches, rising in 15-volt steps from 15 to 450 volts. The panel’s labels progressed from mild descriptions to warnings of severe danger. The teacher read word pairs, tested the learner’s memory, and was instructed to increase the shock after every incorrect response.

The learner’s reactions followed a prepared script. As the apparent voltage rose, he protested, referred to a heart condition, demanded release, struck the wall, and eventually stopped answering. The teacher could hear these responses but could not see the learner in the baseline condition.

When a participant resisted, the experimenter used a sequence of standard verbal prompts, beginning with a request to continue and becoming progressively firmer. If the participant still refused after the final prompt, the session ended. The researcher did not physically force anyone to proceed.

The result that shocked psychology

Before the study, Milgram asked groups of people-including psychiatrists-to predict what participants would do. They expected almost everyone to stop well before the end and believed that only a tiny pathological minority would reach the maximum setting.

The actual result was stark. In the reported baseline condition, 26 of 40 participants-65%-continued to the final 450-volt switch. Every participant reached at least 300 volts before stopping or continuing. Yet the number alone conceals the conflict inside the room. Participants sweated, trembled, laughed nervously, argued, paused and repeatedly sought reassurance. Obedience was often anguished rather than calm.

Fourteen participants refused before the maximum. Their behaviour is not a statistical footnote. It demonstrates that the situation exerted powerful pressure without eliminating choice. Milgram created a laboratory conflict, not a machine that automatically converted people into passive instruments.

Why did so many continue?

No single explanation is sufficient, but several features worked together.

• Legitimate authority: the study took place at Yale, and the researcher appeared to represent competent scientific expertise.

• Gradual commitment: the procedure began with a minor action and increased in small, regular steps. Stopping later meant rejecting a course of action already accepted many times.

• Division of roles: the experimenter designed the procedure, the teacher operated the machine and the learner answered questions. Responsibility could feel distributed across the system.

• Ambiguous responsibility: the researcher’s assurances allowed participants to treat continuation as part of an authorised scientific task rather than a wholly personal decision.

• Conflicting norms: keeping a promise, respecting expertise and completing a study competed with the obligation not to harm another person.

Milgram later described an “agentic state” in which people see themselves as carrying out another person’s wishes and shift responsibility upward. This remains influential, but it is not the final word. A later account known as identification-based followership proposes that participants continued partly because they identified with the scientific mission and actively cooperated with what they believed was worthwhile research. On this view, obedience is not mindless submission; it can involve commitment to a valued cause.

The variations reveal the real lesson

Milgram conducted numerous variations, and obedience changed markedly when the social architecture changed. When the learner was physically closer, refusal became more common. When the experimenter issued instructions remotely, obedience fell. When peers rebelled, participants found it much easier to stop. When another person performed the direct action, responsibility became more distant and obedience increased.

These patterns argue against a fixed “obedient personality” as the complete explanation. Behaviour depended on proximity, institutional legitimacy, peer conduct, responsibility and the relationship between the participant, the researcher and the learner. The famous 65% belongs to one carefully constructed condition; it is not a universal constant.

Disobedience became more likely when the authority’s legitimacy weakened, the learner became more immediate, or peers modelled refusal.

What the experiment does not prove

Milgram’s study does not show that 65% of people will obey any order, in any culture, under any authority. It does not establish that participants enjoyed the task, believed every detail equally, or surrendered all moral judgement. Nor can a brief laboratory encounter explain genocide, military violence or institutional abuse on its own. Those phenomena involve ideology, social identity, coercion, bureaucracy, history and power on a far larger scale.

Archival research has also complicated the classic account. Participants varied in how strongly they believed that the learner was receiving painful shocks, and greater scepticism was associated with higher apparent obedience in many experimental variations. This does not make the studies meaningless, but it limits the inference that pressing a simulated switch is equivalent to knowingly injuring a real person.

Could the experiment be conducted today?

Not in its original form. The study relied on extensive deception, exposed some participants to intense emotional strain and created uncertainty about whether they could genuinely withdraw once the experimenter insisted on continuation. Diana Baumrind’s early ethical critique argued that scientific value did not remove the researcher’s duty to protect participants’ welfare and dignity.

Contemporary research would require independent ethical review, careful screening, explicit safeguards, clear stopping rules, monitoring for distress and a prompt, thorough debriefing. Jerry Burger’s 2009 partial replication incorporated extensive protections and stopped at 150 volts-the point at which the learner first strongly objected-rather than continuing to the end of the original apparatus. Even with these limits, situational pressure to continue remained observable.

A proper debriefing reveals the deception, confirms that no harm occurred, addresses distress and restores the participant’s informed understanding.

Why Milgram still matters

The enduring lesson is not that authority always wins. It is that ethical failure can emerge from ordinary systems in which responsibility is fragmented, escalation is gradual and questioning the procedure feels socially costly. The person giving the instruction may never act; the person performing it may feel that someone else owns the decision.

This matters in hospitals, universities, businesses, public institutions and military organisations. Professional status and hierarchy can support coordination, but they can also suppress challenge. Safe systems therefore make dissent legitimate before a crisis develops.

●      Clarify who remains personally accountable for a decision.

●      Invite concerns before the most senior person states a view.

●      Create protected routes for speaking up and reporting unsafe practice.

●      Treat discomfort as information, not merely as resistance to be overcome.

●      Pause when a procedure advances only because each next step seems slightly easier than reversing course.

The takeaway

Milgram’s participants were neither robots nor simple villains. They were people caught between conscience, commitment, uncertainty and legitimate-seeming authority. Many continued; many protested; some refused. Their conflict is precisely why the experiment remains unsettling.

The most useful question is not, “Would I have pressed the final switch?” From a safe distance, nearly everyone imagines they would refuse. A better question is: “What structures would help me recognise the moral line earlier-and make it possible for me and others to stop?”

References and further reading

Milgram, S. (1963). Behavioral study of obedience. Journal of Abnormal and Social Psychology, 67(4), 371–378. https://doi.org/10.1037/h0040525

Milgram, S. (1965). Some conditions of obedience and disobedience to authority. Human Relations, 18(1), 57–76. https://doi.org/10.1177/001872676501800105

Baumrind, D. (1964). Some thoughts on ethics of research: After reading Milgram’s “Behavioral Study of Obedience.” American Psychologist, 19(6), 421–423. https://doi.org/10.1037/h0040128

Burger, J. M. (2009). Replicating Milgram: Would people still obey today? American Psychologist, 64(1), 1–11. https://doi.org/10.1037/a0010932

Reicher, S. D., Haslam, S. A., & Smith, J. R. (2012). Working toward the experimenter: Reconceptualizing obedience within the Milgram paradigm as identification-based followership. Perspectives on Psychological Science, 7(4), 315–324. https://doi.org/10.1177/1745691612448482

Perry, G., Brannigan, A., Wanner, R. A., & Stam, H. J. (2020). Credibility and incredulity in Milgram’s obedience experiments: A reanalysis of an unpublished test. Social Psychology Quarterly, 83(1), 88–106. https://doi.org/10.1177/0190272519861952