In the early 1960s, psychologist Stanley Milgram ran a series of experiments at Yale University that became one of the most referenced findings in all of social science. In the best-known version, 65 percent of ordinary participants delivered what they believed were potentially lethal electric shocks to a stranger, simply because a researcher in a lab coat told them to continue.1Philosophia Scientiae. Stanley Milgram’s Obedience Studies: A Critical Review of the Most Influential Explanatory Accounts The experiments were designed to probe how far people would go when an authority figure gave harmful orders, and the answer shocked Milgram himself. But six decades of scrutiny have complicated the picture in ways that the textbook summary rarely captures.
What Actually Happened in the Lab
Milgram recruited men from the New Haven, Connecticut area through newspaper ads promising a small payment for participating in a “study of memory.” When a volunteer arrived at the Yale lab, he met another apparent volunteer who was actually an actor working with the research team. Through a rigged drawing, the real participant was always assigned the role of “teacher,” while the actor became the “learner.” The learner was strapped into a chair in an adjacent room, and the teacher sat at a shock generator with switches labeled from 15 volts up to 450 volts, with verbal labels ranging from “slight shock” to “danger: severe shock.”
The teacher’s job was to test the learner on word pairs and deliver an increasingly powerful shock for every wrong answer. The shocks were fake, but the teacher did not know that. As the voltage climbed, the learner (following a script) grunted, protested, screamed, complained of a heart condition, demanded to be released, and eventually fell silent. When the teacher hesitated or asked to stop, the experimenter used a series of prods, escalating from “Please continue” to “You have no other choice, you must go on.”
Milgram did not run one experiment. He ran more than twenty variations, adjusting the setup each time. He moved the learner into the same room as the teacher. He removed the experimenter from the room. He used female participants. He had peers pressure the teacher to stop. He changed the location from Yale to a nondescript office building. Each variation produced a different obedience rate, sometimes dramatically so.
Eight Factors That Shaped Whether People Obeyed
A large-scale analysis that pooled data from Milgram’s full set of experimental conditions identified eight factors that predicted whether a participant would go all the way to the maximum shock. These were not abstract personality traits but features of the situation itself.2PubMed Central. Meta-Milgram: an empirical synthesis of the obedience experiments
- Experimenter directiveness: The more forcefully and consistently the experimenter pushed, the more participants complied.
- Experimenter legitimacy: Obedience dropped when the experiment moved away from Yale’s prestigious setting to a run-down office, eroding the authority figure’s credibility.
- Experimenter consistency: If two experimenters gave contradictory instructions, participants almost universally stopped.
- Group pressure to disobey: When two confederate “teachers” refused to continue, the real participant was far more likely to stop as well.
- Proximity of the learner: The closer the learner was to the teacher, and the more direct the interaction between them, the harder it was for the teacher to keep going.
- Distance from the experimenter: When the authority figure left the room and gave instructions by phone, obedience plummeted.
What made these findings powerful was the sheer range. In some configurations, nearly everyone obeyed. In others, almost no one did. The implication was that obedience was not simply a character flaw of weak individuals but something shaped by context, proximity, and the structure of the situation around them.
Why People Kept Going
Milgram’s own explanation centered on what he called the “agentic shift.” In his account, people who enter a hierarchical situation undergo a psychological transition: they stop seeing themselves as autonomous agents and begin seeing themselves as instruments of the authority figure. Once that shift happens, the person feels responsible for carrying out instructions competently but no longer feels responsible for the consequences of those instructions.3PubMed Central. Obedience induces agentic shifts by increasing the perceived time between own action and results Recent experimental work has found a measurable correlate of this shift: when people act under orders, they perceive a longer time gap between pressing a button and the outcome, as if the connection between their action and its result has loosened in their mind.
Not everyone finds the agentic shift explanation satisfying. An alternative view, developed from archival research into Milgram’s own files at Yale, argues that many participants were not blindly following orders at all. Instead, they were actively persuaded that the experiment served an important scientific purpose. Letters written by participants after the study show many expressing pride in having contributed to science. The researchers who developed this “engaged followership” model argue that the experimenter’s real power was not coercive command but the ability to frame the task as virtuous, making participants feel they were doing something meaningful rather than harmful.4PubMed. ‘Happy to have been of service’: the Yale archive as a window into the engaged followership of participants in Milgram’s ‘obedience’ experiments
This distinction matters beyond the lab. If obedience is primarily about surrendering personal agency to a powerful figure, then the remedy is to break the hierarchical spell. But if people comply because they genuinely believe in the mission, then the problem runs deeper: you need to change what people think is worth doing, not just remind them that they have a choice.5Annual Review of Law and Social Science. 50 Years of “Obedience to Authority”: From Blind Conformity to Engaged Followership
The Surprisingly Important Role of “Please Continue”
One of the most persistent critiques of the Milgram experiments concerns how tightly scripted they actually were. When a participant hesitated, the experimenter was supposed to use four prods in sequence, the last being “You have no other choice, you must go on.” Rhetorical analysis of Milgram’s session recordings revealed that participants regularly drew the experimenter into back-and-forth negotiations about whether to continue, leading to substantial departures from the planned script.6PubMed. Milgram’s obedience experiments: a rhetorical analysis The fourth prod, the one that sounds most like a direct order, was strikingly ineffective: when participants heard “you have no other choice,” they tended to stop.
A more recent study of the session data found something that reframes the famous results. The mildest intervention, “Please continue,” appeared in virtually all sessions, both obedient and disobedient. The stronger, more scripted prods were actually used in fewer than half of obedient sessions. The researchers calculated that if Milgram had defined disobedience as any session where a prod stronger than “Please continue” was needed, the obedience rates would still have been historically striking, but the picture of iron-fisted authority crumbles somewhat.7European Journal of Social Psychology. ‘Please Continue’: Implicit Communication and the Experimenter’s Interventions in Stanley Milgram’s Obedience to Authority Series In other words, what kept people going was often a polite nudge, not a command. That finding aligns more with the engaged followership model than with the image of helpless subjects steamrolled by authority.
Did Participants Even Believe It Was Real?
A persistent question about the experiments is whether participants truly believed they were delivering painful shocks. If some were skeptical, then what looked like obedience was really just going along with an act they saw through. Milgram himself acknowledged that a fraction of participants doubted the setup, but he argued the majority were convinced.
Archival material from Milgram’s own lab paints a more nuanced picture. An unpublished analysis by Milgram’s assistant, Taketo Murata, examined the relationship between participants’ beliefs and their behavior across 23 variations of the experiment. In 18 of those 23 conditions, participants who fully believed the shocks were real delivered lower shock levels on average than those who had doubts.8Social Psychology Quarterly. Credibility and Incredulity in Milgram’s Obedience Experiments: A Reanalysis of an Unpublished Test The pattern is counterintuitive but important: skepticism inflated obedience. If you thought the whole thing was staged, it was easy to keep flipping switches. If you genuinely thought someone was being hurt, you were more likely to stop.
This does not invalidate the experiments, but it does complicate the headline number. The 65 percent figure includes both genuinely distressed people who kept obeying and relatively calm people who suspected the shocks were fake. The real rate of people knowingly inflicting serious pain on an innocent person was likely lower, though how much lower is hard to pin down with precision.
Do the Results Hold Up in Modern Replications?
Because of ethical restrictions, no full replication of the Milgram paradigm has been conducted in recent decades. But partial replications have come close. In 2009, psychologist Jerry Burger devised a “150-volt solution,” stopping the experiment at 150 volts (the point where the learner first screams and demands to be let out) rather than running all the way to 450. This allowed institutional review boards to approve the study while still capturing the critical decision point. Burger found obedience rates comparable to Milgram’s original results, even though participants were screened in advance for psychological vulnerability.9PubMed. Reflections on “Replicating Milgram” A condition in which a confederate refused to continue did not significantly reduce the real participant’s obedience, a finding that surprised many observers.
Virtual reality has opened another window. In VR replications, participants know they are interacting with a computer-generated avatar, not a real person. Nobody is actually being harmed, and participants are told this explicitly. Yet the studies consistently find that participants show subjective distress, physiological stress responses, and behavioral hesitation as if the situation were real.10PLOS ONE. A Virtual Reprise of the Stanley Milgram Obedience Experiments Participants who could see and hear the virtual learner struggled more with continuing. Even in a follow-up VR study, researchers found that participants expressed genuine concern for the virtual learner’s suffering and reported being stressed by the experience.11PubMed Central. Participant concerns for the Learner in a Virtual Reality replication of the Milgram obedience study These VR studies cannot directly measure obedience to harm, since participants know no harm occurs, but they demonstrate that the psychological architecture Milgram tapped into is not a historical artifact. The situation’s pull is real enough to override explicit knowledge that the scenario is simulated.
Gender, Personality, and Who Obeys
A common assumption is that certain types of people are more susceptible to authority. Gender is usually the first variable people ask about. In Milgram’s original Experiment 8, which used an all-female group, obedience rates did not differ significantly from the male participants, though the women reported experiencing higher stress. Modern research has confirmed this pattern and extended it: the gender of the authority figure does not seem to matter much either. In a lab experiment, 88 percent of participants complied with a female experimenter compared to 90 percent with a male experimenter, a negligible difference. An online study with nearly 800 participants produced the same result.12Social Psychology. Authority Knows No Gender – Gender Effects in Exerting Obedience in Milgram’s Experiment
Personality does show some predictive power, though not in the direction most people expect. A study using a Milgram-style paradigm found that participants scoring higher on conscientiousness and agreeableness were more willing to administer intense shocks. These are traits generally considered positive in everyday life: being dutiful, cooperative, and organized. But in the context of an authority-driven situation, those same tendencies become liabilities, making a person more inclined to follow through on uncomfortable orders. Political orientation and social activism also predicted behavior, with more politically conservative and less socially active participants tending toward greater obedience.13PubMed. Personality predicts obedience in a Milgram paradigm
These findings are worth putting in perspective. Personality variables accounted for a meaningful share of the variation in behavior, but they pale compared to the situational factors identified in the meta-analysis of Milgram’s variations. How close the learner was, how the experimenter communicated, and whether peers were present mattered more than any trait the participant walked in with.
What Happens in the Brain When You Follow Orders
Neuroscience has begun to map what Milgram described as the agentic shift in measurable brain activity. In fMRI studies using coercion paradigms (where participants either freely choose to shock another person or are ordered to do so), following orders consistently reduces the brain’s response to the victim’s pain. Activity drops in regions associated with empathy and vicarious pain, including the anterior cingulate cortex, the insula, and the temporoparietal junction. Participants also showed reduced activation in neural signatures associated with guilt, and they reported feeling less responsible for the outcome.14PubMed. Obeying orders reduces vicarious brain activation towards victims’ pain
A separate study found that the medial frontal gyrus plays a role in maintaining a sense of personal agency under coercion. Participants whose brain activity in this region stayed relatively high during coerced actions also maintained a stronger subjective sense that they were the ones causing the outcome. Those whose medial frontal activity dropped showed the most pronounced agentic shift, perceiving a greater temporal gap between their action and its consequence.15PubMed Central. The obedient mind and the volitional brain: A neural basis for preserved sense of agency and sense of responsibility under coercion Further work comparing civilians and military personnel found that following orders reduced the sense of agency across both groups, with a network of frontal and parietal regions tracking the difference between free and coerced moral decisions.16PubMed. Neural correlates of the sense of agency in free and coerced moral decision-making among civilians and military personnel
The upshot of these studies is that “just following orders” is not merely a convenient excuse. When people act under orders, their brains literally process the victim’s pain differently, feel less guilt, and experience a weakened connection between action and consequence. The machinery of moral concern does not shut off entirely, but it dims. Whether that dimming reflects an automatic psychological process or something closer to motivated reasoning remains debated.
The Ethical Fallout
Milgram’s experiments are as famous for their ethical controversies as for their findings. Participants were deceived about the purpose of the study, led to believe they were harming another person, and subjected to intense psychological stress. Some left the lab visibly shaken. Milgram defended himself by pointing to post-experiment debriefings and a follow-up survey showing that the vast majority of participants said they were glad to have taken part. But critics have argued that debriefing someone after inducing them to believe they nearly killed a stranger does not undo the experience.17Research Ethics. What can Milgram and Zimbardo teach ethics committees and qualitative researchers about minimizing harm?
The studies played a direct role in the development of modern research ethics standards. Institutional review boards, which now review proposed studies for potential harm to participants before they can proceed, grew partly out of reactions to experiments like Milgram’s. No ethics board today would approve the original procedure. Even Burger’s partial replication in 2009 required extensive screening, stopping at 150 volts, and immediate debriefing to win institutional approval. The experiments exist in a strange ethical limbo: they could never be repeated, but they remain among the most cited and taught studies in psychology, their findings treated as foundational knowledge about human nature.
How the Findings Have Traveled Across Cultures
Replications and adaptations of the Milgram paradigm have been conducted in countries across Europe, Africa, Asia, and Australia. A cross-cultural review of these studies found that the basic pattern of obedience to authority holds across diverse cultural settings, though exact rates vary depending on how closely a given replication matched Milgram’s original method. Gender differences in obedience have been similarly small or nonexistent across cultures, consistent with what Milgram found and what later lab studies confirmed.
One area where cross-cultural work has been less conclusive is the mechanism behind obedience. Whether the agentic shift operates the same way in collectivist cultures as in individualist ones remains an open question. In societies where deference to authority is more explicitly valued, it can be difficult to disentangle culturally normative respect from the psychological surrender of personal agency that Milgram described. The behavioral outcome looks similar, but the internal experience may differ.
Obedience in Real Institutions
One of the earliest and most striking extensions of Milgram’s work took place in a hospital. In a 1966 study, researchers called nurses on hospital wards, identified themselves as doctors the nurses had never met, and instructed them to administer a dose of medication that was clearly above the safe maximum listed on the bottle. The vast majority of nurses moved to comply before a researcher intercepted them. The study illustrated that the dynamics Milgram captured in a lab were not confined to an artificial setting. In hierarchical institutions with established chains of command, obedience could translate directly into dangerous action.
Modern healthcare has tried to address this through protocols like crew resource management, borrowed from aviation, which explicitly empower junior team members to challenge unsafe orders. Military training in many countries now includes instruction on the obligation to refuse illegal orders. These institutional reforms represent a practical legacy of Milgram’s work, even as scholars continue debating exactly what his experiments proved and what they merely suggested. The research did not just describe a problem; it changed how organizations think about the relationship between hierarchy and harm.