The Psychology of Obedience: Why Good People Follow Evil Orders

About this psychology article

Ordinary, decent people delivered what they believed were increasingly painful electric shocks to a stranger, simply because a man in a lab coat told them to continue. Here's what psychologists act...

You're told the shocks are necessary for the experiment. The man in the lab coat says the results are essential. The person behind the wall is screaming, then pounding on the wall, then silent. And you keep pressing the switch, because someone with authority told you it was fine.

Most people, asked directly, are certain they wouldn't do this. They'd refuse. They'd walk out. They'd never knowingly hurt another person just because someone in a position of authority told them to continue. This certainty is almost universal, and it is, according to one of the most replicated and unsettling findings in the history of psychology, almost always wrong.

In 1963, psychologist Stanley Milgram set out to answer a question that had been haunting the postwar world since the Nuremberg trials: how do ordinary people end up participating in atrocities? Not sadists, not the mentally disturbed, but regular citizens, employees, neighbors. His answer, built from one of the most famous and most misunderstood experiments ever conducted, wasn't about a few uniquely evil individuals. It was about a specific set of conditions that can push the overwhelming majority of people toward causing harm they would otherwise consider unthinkable, simply because someone else was willing to take responsibility for it.

What Obedience to Authority Actually Means

Obedience, in the psychological sense Milgram studied, describes compliance with a direct instruction from a perceived legitimate authority, even when that instruction conflicts with a person's own moral judgment or stated values. This is a specific and narrower concept than simply following rules or cooperating with others. It describes the particular moment when someone's internal sense of right and wrong comes into direct conflict with an external command, and the command wins.

What makes this psychologically distinct from cowardice or moral failure is the mechanism underneath it. Milgram's research suggested that obedience isn't primarily driven by a personal decision to do something harmful. It's driven by a shift in how a person experiences their own agency, moving from feeling like the author of their actions to feeling like an instrument carrying out someone else's decision. That shift, more than any deficiency in character, is what the research kept pointing back to.

The Milgram Experiment: What Actually Happened Inside That Room

As covered in full detail in The Milgram Obedience Study , participants were told they were taking part in a study on learning and memory. They were assigned the role of "teacher," instructed to administer increasingly severe electric shocks to a "learner" in another room every time that learner answered a question incorrectly. The shock generator was labeled in ascending increments, from a mild 15 volts up to a clearly marked 450 volts, with the highest levels ominously labeled "XXX." Unbeknownst to the participants, the learner was an actor, and no real shocks were ever administered. But the participants didn't know that. As far as they understood, they were causing real, escalating pain to a real person in the next room.

When the learner began audibly protesting, then screaming, then pounding on the wall, then falling completely silent, an experimenter in a lab coat standing nearby delivered a scripted series of increasingly firm prods: "Please continue." "The experiment requires that you continue." "It is absolutely essential that you continue." "You have no other choice, you must go on."

Before running the study, Milgram surveyed psychiatrists, asking them to predict what percentage of participants would go all the way to the maximum 450-volt shock. The consensus estimate was around one in a thousand, reserved for a small population of sadistic outliers. The actual result: roughly 65% of participants delivered the maximum shock, continuing to the very end of the generator's range, simply because a man in a lab coat told them the experiment required it.

Why This Isn't About Being a "Bad Person"

This is the finding that made Milgram's work so uncomfortable, and so enduringly relevant. The participants weren't cruel. Many of them showed visible, extreme distress throughout the process, sweating, stuttering, laughing nervously, asking repeatedly who would take responsibility if something happened to the learner. They weren't enjoying it. Most of them clearly wanted to stop. And most of them kept going anyway, because the responsibility for continuing had been, at least in their own minds, transferred to someone else.

Milgram called this the agentic state: a psychological shift in which a person stops experiencing themselves as the origin of their actions and instead experiences themselves as an agent carrying out someone else's will. In this state, the usual sense of personal moral responsibility doesn't disappear entirely, but it becomes muted, redirected toward the authority figure who issued the instruction. The participant isn't thinking "I am hurting this person." They're thinking "the experimenter is running this experiment, and I am doing what's required of me within it."

This connects directly to the mechanism explored in The Stanford Prison Experiment , where ordinary participants assigned the role of "guard" escalated into genuinely abusive behavior largely because the situational structure, not any preexisting trait, gave them both permission and a role to inhabit. In both cases, the disturbing lesson wasn't about identifying which individuals were secretly capable of cruelty. It was about how thoroughly situational structure can override individual conscience, in people who show no unusual predisposition toward harm in any other context of their lives.

The Specific Conditions That Make Obedience More Likely

Milgram didn't stop at the original setup. He ran more than twenty variations of the experiment, deliberately manipulating specific conditions to isolate exactly what increased or decreased obedience, and the pattern that emerged was remarkably consistent and specific.

Physical and psychological distance mattered enormously. When the learner was in the same room as the participant, obedience dropped sharply compared to when the learner was in a separate room and could only be heard, not seen. When participants had to physically place the learner's hand on a shock plate themselves, obedience dropped further still. Distance from the consequences of an action, it turned out, made those consequences dramatically easier to enact.

Proximity to the authority figure mattered just as much, in the opposite direction. When the experimenter gave instructions by phone rather than standing in the room, obedience dropped substantially, with several participants claiming to continue while actually stopping once out of the experimenter's direct view. And in one of the most striking variations, when participants watched two other "teachers," actually confederates, refuse to continue partway through, obedience collapsed to roughly 10%. Seeing even a single other person defy the authority figure was enough to give most participants the psychological permission to do the same.

This last finding deserves particular attention, because it offers something closer to genuine hope than most of Milgram's other results. Obedience wasn't a fixed trait people either had or didn't have. It was a probability, heavily shaped by the specific social conditions surrounding the moment of decision, conditions including whether anyone else in the room was willing to say no first.

Obedience Is Not the Same as Conformity

This distinction is worth stating clearly, because the two get treated as interchangeable far more often than they should be. As explored in Conformity Explained , conformity describes adjusting your behavior or stated beliefs to match a group of peers, driven primarily by a desire to fit in or a genuine uncertainty about what's correct. Obedience describes something structurally different: compliance with a direct instruction from a perceived legitimate authority, driven by a hierarchical relationship rather than peer pressure.

The two mechanisms can compound each other, and often do in real institutional settings, but they operate through different psychological pathways. Solomon Asch's classic conformity experiments showed people doubting their own perception to match a group of equals. Milgram's obedience experiments showed people overriding their own conscience to comply with a hierarchical command. Someone can be highly resistant to peer conformity while remaining just as vulnerable to authority-driven obedience, or the reverse, which is part of why understanding the specific mechanism at play in a given situation actually matters.

Beyond the Lab: Why This Research Still Matters

Milgram's work was directly motivated by an attempt to understand how large numbers of ordinary German citizens participated in the machinery of the Holocaust, a question philosopher Hannah Arendt had grappled with just two years earlier while covering the trial of Adolf Eichmann, coining the phrase "the banality of evil" to describe what she saw: not a monster, but a bureaucrat who insisted he was simply following orders. It's worth being precise here about where the evidence actually sits: Arendt's concept was a philosophical and journalistic observation, not an experimentally derived finding, while Milgram's research provided empirical, replicated laboratory evidence for a related but distinct mechanism. The two concepts inform each other without being identical, and treating them as interchangeable overstates what either one, on its own, actually demonstrates.

The pattern Milgram identified has since been observed, with appropriate caution about oversimplifying complex real-world events, in a range of documented institutional failures: military personnel following unlawful orders, corporate employees participating in fraud they privately found troubling, cult members following escalating directives from a leader, and healthcare workers, as some later studies specifically examined, deferring to a physician's clearly incorrect instruction rather than risk an uncomfortable confrontation. It's worth noting that Milgram's original findings have also faced legitimate scholarly scrutiny; historian Gina Perry's 2013 investigation into the original experimental records found more variability in how strictly the script was followed and how participants actually responded than the popularized version of the study often suggests. This doesn't invalidate the core finding, which has been replicated in modified forms as recently as Jerry Burger's 2009 partial replication, but it's a useful reminder to hold even famous, foundational research with appropriate nuance rather than treating it as a single, clean, unassailable number.

The Belief That Keeps Obedience Running

Underneath most destructive obedience sits a quiet, rarely examined belief: if someone with legitimate authority says this is necessary, my personal discomfort with it probably isn't a reliable guide. This belief has an intuitive logic in many everyday contexts, since deferring to genuine expertise, a surgeon's judgment, a pilot's instructions during an emergency, is often exactly the right call, and questioning every instruction from every authority figure would make functioning within any institution nearly impossible.

The danger isn't the belief itself. It's the absence of any internal checkpoint distinguishing legitimate expertise from illegitimate demands. Once that checkpoint goes quiet, the agentic state can take over almost automatically, redirecting responsibility upward and allowing behavior to escalate incrementally, one small step at a time, in a pattern closely related to the mechanism explored in The Foot-in-the-Door Experiment : each shock level felt only slightly worse than the one before it, which made stopping at any single point feel arbitrary, even as the cumulative distance traveled from the starting point became enormous.

Can Awareness of This Pattern Actually Protect You?

To some meaningful degree, yes, though it's worth being honest about the limits of what awareness alone accomplishes. Simply knowing about the Milgram experiment doesn't automatically make someone immune to its underlying mechanism; the participants in later replications, some of whom had likely heard of the study, still showed substantial obedience rates. But specific, concrete strategies do show real protective value, based on exactly the variables Milgram's own variations identified.

Deliberately closing the physical or psychological distance to the person who would be affected by a harmful instruction reliably reduces compliance, since abstraction is part of what makes obedience easier. Seeking out or simply noticing dissent from even one other person in the room dramatically increases the odds of resisting an improper instruction, which is why institutional cultures that actively protect the ability to voice disagreement tend to see far less destructive compliance than cultures that quietly punish it. And deliberately pausing to ask a direct, specific question, "who is actually accountable if this goes wrong, and is that acceptable to me personally," reintroduces the exact sense of personal agency the agentic state works to dissolve.

What This Pattern Is Not

Being capable of obedience to a harmful instruction isn't evidence of a weak character or a latent capacity for cruelty. Milgram's research is unsettling precisely because it implicates ordinary psychological machinery, present in the vast majority of people, rather than some rare defect present in only a few. Recognizing this isn't an excuse for harmful compliance when it happens. It's a more accurate, and ultimately more useful, starting point than assuming harmful obedience only happens to people fundamentally different from yourself. The mechanism is common. The specific conditions that activate it are identifiable. And the same research that revealed how easily obedience escalates also revealed exactly what interrupts it.

See Your Own Response to Pressure More Clearly

Reading about a landmark experiment is different from recognizing how authority, pressure, and diffused responsibility actually show up in your own decisions, at work, in relationships, or anywhere a request starts to feel like it isn't optional.

→ Take the Manipulation Detection Assessment

Free. No account required.

Curious how you specifically respond to pressure and boundary-testing in your own relationships and environment?

→ Explore the Boundary & Pressure Pattern Assessment in the PsychOS Assessment Lab

Frequently Asked Questions

What was the main finding of the Milgram obedience experiment?

Approximately 65% of participants administered what they believed were the maximum, potentially lethal electric shocks to another person, simply because an authority figure instructed them to continue, despite showing visible distress and a clear desire to stop.

Why do ordinary people obey harmful instructions?

Milgram's research points to a psychological shift called the agentic state, in which a person stops experiencing themselves as personally responsible for an action and instead experiences themselves as an instrument carrying out someone else's decision.

What factors make obedience more or less likely?

Physical and psychological distance from the person being harmed increases obedience, while proximity to the authority figure issuing instructions also increases it. Witnessing even one other person refuse to comply dramatically decreases obedience.

Is obedience the same thing as conformity?

No. Conformity describes matching the behavior of peers, typically to fit in or resolve uncertainty. Obedience describes complying with a direct instruction from a perceived legitimate authority, a distinct hierarchical mechanism studied separately by researchers like Solomon Asch and Stanley Milgram.

Has the Milgram experiment been criticized or questioned?

Yes. Later historical investigation, including Gina Perry's 2013 review of original experimental records, found more variability in how the study was actually conducted than commonly assumed. The core finding has still been replicated in modified, more ethically constrained forms as recently as 2009.

Want to see how PsychOS handles psychology research, sources, limitations, and interpretation? Read Editorial Standards