● Cognitive bias
From Latin auctoritas, "power, influence, legitimacy"; named as a persuasion principle by psychologist Robert Cialdini, 1984; foundational research by psychologist Stanley Milgram, 1961-63.
The tendency to give a claim or instruction more weight simply because it comes from someone who looks, sounds, or is labeled as an authority, even when that label has nothing to do with whether they're actually right.
Also called: obedience to authority, the authority heuristic. Not the same as trusting real expertise (see Section 01).
The tendency to assign greater accuracy, credibility, or legitimacy to a statement or instruction because of the perceived authority of its source (a title, uniform, institutional affiliation, or confident manner), independent of the actual soundness of the claim itself.
Plain version
Your brain uses "who said it" as a shortcut for "how much should I believe or obey this," instead of actually checking the claim on its own merits. The shortcut fires off the symbols of authority (a badge, a lab coat, a job title, an official-sounding voice) almost as readily as it fires off real, relevant expertise.
Aliases
Obedience to authority and the authority heuristic are both used for this pattern in the research literature. Psychologist Robert Cialdini's later, more consumer-facing framing calls it simply the authority principle of persuasion.
Not the same as trusting real expertise
This is the distinction worth getting right: trusting a cardiologist about your heart, or a structural engineer about a bridge, isn't a bias. It's the correct use of relevant, verified expertise, and society mostly runs on exactly this kind of reasonable deference. Authority bias is what happens when the deference gets decoupled from actual relevant expertise: when the lab coat, the title, or the confident tone does the persuading even though the person wearing it has no real qualification to speak on the claim at hand, or when a genuine expert is followed uncritically outside the narrow area where their expertise actually applies.
Scope
Hard to catch, though not quite as silent as the hardest biases on the site. Deferring to authority sometimes comes with a felt signal (Milgram's participants trembled, sweated, and laughed nervously while still continuing), but that discomfort usually reads as "this task is unpleasant," not "I am being biased," so it rarely triggers self-correction. The cover story is also unusually strong ("they're the expert, not me"), and going along with legitimate-seeming authority hands you real relief: you don't have to own the decision.
Why your brain does this
You cannot personally verify almost anything you're told: you didn't run the clinical trial, audit the bridge design, or witness the historical event yourself. Deferring to people who plausibly know more than you is, most of the time, a completely reasonable way to function in a world too complex for any one person to independently check. Historically, the people society marked as authorities (elders, trained specialists, officials) usually did know more, so the shortcut paid off often enough to stick around. This isn't a broken brain. It's a brain conserving effort on a problem it can't fully solve alone.
Where it misfires
The shortcut breaks down when the markers of authority get separated from the substance behind them: when a title, uniform, or confident tone is real but irrelevant to the claim being made, or when the marker itself is faked. In those cases the brain still fires the "defer" response, because it's reacting to the symbol of authority, not auditing whether the authority is actually warranted here.
Two ways researchers frame this (both worth knowing)
Gerd Gigerenzer's "fast and frugal heuristics" research treats authority deference as ecologically rational: in an information-rich, time-poor world, using a credible-seeming source as a stand-in for verifying every claim yourself is often the best available strategy, not a design flaw. On this view, the bias is a generally sound heuristic misapplied outside the conditions where it works well.
The Kahneman/Tversky heuristics-and-biases tradition frames it differently: authority deference is a case of attribute substitution, where a hard question ("is this claim actually correct?") gets swapped for an easier one ("does this source seem legitimate?") without the person noticing the swap happened. The site doesn't take a side between these framings; both are live in the research, and both agree on the practical lesson: this is a shortcut built for a world of limited time and information, not evidence of gullibility or low intelligence.
How it activates, step by step
1. You encounter a claim or instruction accompanied by markers of authority: a title, a uniform, an institutional seal, a confident and technical tone.
2. Those markers get used as a fast proxy for "this is probably correct" or "I should probably comply," standing in for actually evaluating the claim.
3. Independent scrutiny of the claim's actual merits gets reduced or skipped, especially if evaluating it yourself would take real effort or expertise you don't have.
4. You adopt the belief or follow the instruction, sometimes with visible discomfort, but usually without a clear internal flag that a shortcut, rather than a real judgment, just did the work.
Authority bias is one of the most directly targeted vulnerabilities on this site; several propaganda techniques exist almost entirely to trigger it.
The most direct hit: citing a person, institution, or credential as if that alone settles the claim, regardless of whether the source is actually qualified on this specific question. Your authority bias does the rest of the work without any further argument being needed.
Borrows a recognizable, often unrelated authority figure or celebrity to vouch for a claim or product. The credibility doesn't come from the endorser knowing anything about the subject; it comes from your authority bias transferring their general status onto this specific, often unconnected claim.
Pairs a claim or actor with the prestige of a respected symbol, institution, or seal of legitimacy (a flag, a uniform, an official-looking logo), so the borrowed authority of the symbol rubs off on the message, without the message having earned it independently.
Elevates a single leader's authority to a level where it's treated as effectively beyond question. Once that happens, authority bias stops working claim-by-claim and starts working wholesale: followers extend the same reflexive deference to nearly everything the leader says, insulating the leader's claims from the scrutiny any individual one would otherwise get.
Outside politics
Advertising leans on this constantly and mostly non-maliciously: an actor in a white coat, or the line "nine out of ten dentists recommend," borrows the authority of a medical or expert consensus that may or may not be real or relevant. In its predatory form, phone and text scams impersonate police officers, tax authorities, or bank fraud departments specifically because an official-sounding voice and enough procedural jargon can get a target to hand over money or information they'd never give a stranger who simply asked.
What increases it
Visible symbols of authority: psychologist Leonard Bickman's 1974 field experiment found pedestrians in Brooklyn were far more likely to obey a stranger's request when he wore a security-guard uniform (89% complied with a request to give a bystander a dime for a parking meter) than when he wore a milkman's uniform (57%) or ordinary clothes (33%). The uniform alone did most of the persuading.
Domain unfamiliarity: the less you personally know about a subject, the harder it is to independently evaluate a claim, so you lean more heavily on who's making it.
Distance from personal responsibility: Milgram's own variations found obedience dropped sharply as the "teacher" was placed physically closer to the person being harmed, and rose when responsibility for the outcome could be shifted onto the authority giving the order.
What does NOT predict it
This one matters: Milgram's original sample spanned a wide range of occupations and education levels with no meaningful protective effect from more schooling, and psychiatrist Charles Hofling's 1966 hospital study found trained, licensed nurses (professionals with real domain expertise) still complied with an unfamiliar "doctor's" order 21 out of 22 times, despite knowing it broke hospital protocol. Expertise in general, or in an unrelated area, does not reliably protect against authority bias in the moment it's actually tested.
State-dependent factors
Time pressure, unfamiliarity with the situation, and being asked to act quickly all increase reliance on the authority shortcut, since there's less opportunity to independently reason through the claim. Milgram's transcripts also show that a graduated, step-by-step escalation (never one single, obviously extreme request) made continued compliance easier at every stage than a single big ask would have been.
The pattern across all four: authority bias doesn't require a villain giving orders, or a gullible person receiving them. It shows up in trained nurses, ordinary volunteers, wartime bureaucrats, and everyday phone calls alike, because the mechanism only cares about the presence of authority markers, not about who's standing in front of them.
First named/documented
The foundational research is Stanley Milgram's 1963 paper "Behavioral study of obedience" (Journal of Abnormal and Social Psychology, 67(4), 371-378), though the pattern was later formalized as one of six persuasion principles in psychologist Robert Cialdini's popular-audience book Influence (1984), which coined the more consumer-facing "authority principle" framing used in marketing and compliance research.
The landmark study
Milgram recruited 40 male volunteers at Yale University for a supposed study of "punishment and learning." Each was assigned the role of "teacher" and instructed by a lab-coated experimenter to shock a "learner" (an actor, in a separate room) with increasing voltage for each wrong answer, up to a labeled 450 volts marked "Danger: severe shock." All 40 continued to at least 300 volts; 65% continued to the maximum, even as the "learner" pounded on the wall and then fell silent. Milgram had surveyed psychiatrists beforehand, who predicted well under 1% would go that far.
Since then
Milgram ran at least 18 documented variations, manipulating factors like the experimenter's physical proximity, the setting's institutional prestige, and the presence of peer "teachers" who refused. Obedience rates shifted substantially across these conditions, showing the effect is heavily situational rather than fixed. A 2017 partial replication in Poland by psychologists Dariusz Doliński and Tomasz Grzyb, adapted for modern ethical standards, found comparably high, in fact slightly higher, obedience rates decades later, suggesting the underlying pattern has not simply faded with time or cultural change. Psychologist Leonard Bickman's 1974 uniform study (see Section 04) extended the finding from live verbal commands to visual authority symbols alone.
Cross-tradition note
France offers a genuinely distinct real-world extension of this research, not just a translation of it. In 2010, French public broadcaster France 2 aired Le Jeu de la Mort ("The Game of Death"), a documentary in which volunteers were recruited for a fake game-show pilot and told, by a television host and a studio audience, to deliver increasingly severe shocks to a "contestant." Roughly 80% complied to the maximum level, matching or exceeding Milgram's original rate, but with the authority figure replaced entirely by a media host and the social pressure of a live studio audience rather than a scientist in a lab coat. Social psychologists involved in the production, including Jean-Léon Beauvois, argued this showed television itself functioning as a modern authority figure. The project has also drawn academic criticism for not fully replicating Milgram's original conditions (notably the added presence of a studio audience) and for using the same media techniques it set out to critique, worth knowing if this example comes up, rather than treating it as a clean scientific replication. See Section 02 for Gerd Gigerenzer's ecological-rationality framing, the relevant German research tradition for this page's adaptive-origin section.
Score: 2 out of 5 (see the self-check card after Section 01 for the three sub-scores). Hard to catch, though not quite as silent as the hardest biases on the site.
The in-the-moment question
Before complying or believing: "If I stripped away the title, the uniform, or the confident tone, would this instruction or claim still make sense on its own?" If the answer only holds up because of who said it, that's the signal, not any feeling of certainty, which is usually still present either way.
Why this is hard to catch
Milgram's own participants often showed real discomfort (trembling, sweating, asking to stop), yet the large majority kept going anyway. That's the core problem with this bias: the internal signal, when it exists at all, gets interpreted as "this is unpleasant" rather than "I am deferring to something I haven't actually verified." The cover story is also unusually strong ("they're the expert, I'm not qualified to question this"), and complying provides real relief from having to own a hard decision yourself, which is exactly why even visible distress often isn't enough to produce self-correction in the moment.
Debiasing research note
The evidence here is thinner than for some other biases on this site. Milgram-style obedience has proven resistant to simple awareness-raising: knowing about the experiment doesn't reliably make people less obedient in a fresh situation. What does show real effect in the research is structural: verification requirements and distributed responsibility, which don't rely on any individual successfully catching the bias in themselves in the moment.
Commonly co-occurs with
Often mistaken for
Looks similar from the outside, both end in "going along with it," but the mechanism is different. Bandwagon effect is horizontal: deferring to what a lot of your peers already believe or do. Authority bias is vertical: deferring to a single person or institution perceived to hold legitimate power, expertise, or rank over you, regardless of what the majority thinks.
Compounding effect
Authority bias frequently feeds the halo effect: once someone is accepted as an authority in one domain, their perceived credibility often bleeds into unrelated domains where they have no real expertise at all. Left unchecked over time, this compounding is exactly the mechanism behind cult of personality (Section 03): a leader's authority, reinforced repeatedly and never independently checked, gradually expands until it's treated as blanket credibility across everything they say.
Primary sources
Accessible reading