Learn it → Apply it → Save as Skill
A request from a perceived authority gets followed far more readily than the same request from a peer, and that readiness tracks the symbols of authority, a title, a uniform, a confident tone, a prestigious setting, more closely than it tracks whether the person is actually right. Authority is the mental model for noticing that pull, using real expertise honestly when you have it, and checking a confident instruction against the evidence before you comply with it.
The Dr. Wallace’s Favorite Mental Models series is designed to turn useful ways of thinking into reusable tools. Each model gives you a fast visual overview, deeper interactive learning, practical application exercises and an AI-ready Markdown skill you can save for future use.
Instead of merely reading about mental models, you can gradually build your own AI skills library: a collection of portable reasoning skills that can help ChatGPT, Claude, Grok and other AI assistants apply the mental models you value.
When an instruction comes from someone wearing the costume of authority, a title, a uniform, a lab coat, a prestigious institution's name, people tend to comply with it, often without separately checking whether the instruction is actually correct. The pull isn't really about verified expertise. It's about recognizable signals of authority, and those signals can be genuine, borrowed, or entirely fabricated.
Illustrative, not exact: bar heights reflect the published full-obedience rates from Stanley Milgram's 1963 Yale study and his own later variations, not a universal multiplier. Strip away physical presence, institutional prestige or peer deference, and the identical instruction loses most of its pull.
Authority is most useful before you comply with an instruction, defer to a credential, or lend your own expertise to someone else's claim, in healthcare and safety decisions, workplace instructions, sales and marketing claims, and everyday moments where a confident voice is nudging you toward agreement.
Authority doesn't ask you to distrust every expert. It asks you to notice when a title or a confident tone is carrying the argument, check whether that authority is actually relevant to this specific decision, weigh the instruction's content separately from who delivered it, and decide deliberately rather than by reflex.
Notice when a request leans on a title, uniform, credential or confident tone rather than on its own stated reasoning.
Check whether the authority is actually relevant expertise for this specific decision, not just borrowed prestige from a different domain.
Weigh the content of the instruction on its own merits, independent of who delivered it or how confidently they said it.
Comply when the authority's relevant expertise and the evidence agree; pause and verify through a second channel when they don't.
Open a use case to see what the model changes.
When an order or instruction from a physician or senior clinician doesn't match what a reference, a chart or a second check would show, verifying through that second channel before acting is what separates deference from diligence. Hospitals that built mandatory callback verification into procedure did so precisely because individual judgment alone, under pressure from a title, proved unreliable.
"Recommended by experts," a white coat in an ad, or a vague professional title attached to a claim is designed to transfer the credibility of the role onto the product, whether or not that expertise is actually relevant to the claim being made. Checking what the credential actually certifies, and whether it is relevant here, is the model in action.
Hierarchy makes organizations function, but it also makes people hesitate to flag an error coming from above them. Separating "this person outranks me" from "this person is right about this specific thing" is what lets a legitimate concern get raised instead of silently absorbed.
A steep "authority gradient," where junior crew or staff feel unable to challenge a senior figure's decision, has been identified as a contributing factor in documented aviation incidents. Crew Resource Management training was developed specifically to flatten that gradient and make challenging a senior figure's call a normal, expected part of the job.
Stating your actual, relevant credentials and reasoning, rather than relying on title alone, builds authority that holds up under scrutiny. Borrowing prestige from an unrelated credential may work briefly, but it collapses the moment someone checks whether it's actually relevant.
Noticing that a confident tone, a title or a uniform is moving you toward agreement faster than the actual argument would justify is itself a use of the model: naming the pull is often enough to decide deliberately whether the authority here is actually relevant.
This is an illustrative example designed to teach the method.
Expand for where the idea came from and how solid the evidence actually is.
The psychologist Robert Cialdini formalized authority as one of the core principles of persuasion in his 1984 book "Influence: The Psychology of Persuasion," drawing together decades of social-psychology research on obedience and credibility into a single, named pattern: people comply more readily with requests that carry recognizable markers of authority, titles, uniforms, and other trappings, often with little separate scrutiny of whether that authority is actually relevant.
Stanley Milgram's obedience experiments, conducted at Yale University in 1961 and 1962 and published in 1963, found that 65% of participants, acting as "teachers," obeyed an experimenter's instructions all the way to the maximum 450-volt shock level, despite the "learner's" apparent distress. Before the study, psychiatrists Milgram surveyed had predicted that only about 0.1% of participants would go that far, making the gap between expert prediction and actual obedience one of the most cited findings in social psychology.
Genuinely strong and replicated outside the lab. In a 1966 study, the psychiatrist Charles Hofling had researchers posing as an unfamiliar physician call hospital nurses and instruct them, by phone, to administer a drug dose that was twice the maximum stated on the label. 21 of 22 nurses, 95%, began to comply before being stopped, even though hospital policy barred unauthorized phone orders of that kind. A separate group of 33 nurses, asked hypothetically how they would respond to such an order, had 31 say they would refuse it, a sharp gap between stated intentions and actual behavior under real authority pressure.
No, and that's a central, documented part of the principle. Cialdini's own research catalogued how mere symbols of authority, a title, a uniform, an expensive car, a diploma on a wall, trigger a similar deference reflex to substantive, relevant expertise, even when the symbol has nothing to do with the claim being made. That's what makes the principle usable honestly, stating real, relevant credentials plainly, and usable manipulatively, borrowing an unrelated credential's prestige to make an unrelated claim land.
An authority who sounds certain isn't automatically right, and a title that's real in one domain carries no special weight in an unrelated one, a celebrity endorsing a medical claim, a senior executive overruling a specialist outside their own field.
Hofling's study found 95% of nurses began complying with an unauthorized order rather than checking it first, even though checking was the lower-risk option; deferring instead of verifying is riskiest exactly when the cost of being wrong is highest.
A steep authority gradient, where a junior team member feels unable to flag a senior figure's error, has been identified as a contributing factor in documented aviation incidents, which is why Crew Resource Management training exists to make raising a concern a normal, expected part of the job rather than an act of defiance.
Citing a real credential that has nothing to do with the claim at hand, a doctor's title used to sell an unrelated product, transfers unearned credibility and erodes trust once the mismatch is noticed.
In Milgram's own variations, full obedience fell from 65% to 20.5% once the experimenter gave instructions by phone instead of standing in the room, showing that the pull of authority leans heavily on proximity and presence, not just on the words of the instruction.
People complying with an instruction says something about the pull of authority; it says nothing on its own about whether the instruction was correct. The two get conflated constantly, and keeping them separate is most of what this model asks of you.
the decision is low-stakes, easily reversible, and the authority's expertise is genuinely relevant and well-established for exactly this kind of decision. Re-verifying a routine, low-risk instruction from a qualified source mostly just slows things down.
a way to dismiss legitimate input from someone with less formal status but more relevant, direct knowledge, a frontline worker, a junior specialist, a patient describing their own symptoms. Authority bias runs in both directions: deferring too readily to titles, and discounting people without one.
the cost of being wrong is high, medicine, aviation, finance, safety-critical operations. Individual judgment alone, even well-intentioned, has repeatedly proven an unreliable check against a confident authority figure in exactly these high-stakes settings.
This browser tool structures your thinking. It does not send your information anywhere or perform AI analysis.
Save as Skill converts the model into portable Markdown instructions that teach an AI when and how to apply the authority principle honestly, what separates a relevant credential from a borrowed one, what mistakes to avoid and how to structure its analysis.
Save the skills you find useful and organize them into your own reusable library. The Markdown is intentionally model-neutral, so it can be used wherever an AI system accepts uploaded files, project knowledge, custom instructions or reusable text context. Platform capabilities differ, so this is a portable skill file rather than a claim of one-click installation.
Often, yes, and that's precisely why the heuristic exists: deferring to a genuine, relevant expert is usually a reasonable shortcut when you can't independently verify every claim yourself. The problem isn't that the shortcut exists; it's that it responds to symbols, titles, uniforms, confident tone, that can be present with or without the underlying expertise they're supposed to signal.
Before running the study, Milgram asked psychiatrists to predict how far participants would go; they expected only a small, unusual minority, around 0.1%, to reach the maximum shock level, assuming obedience that extreme would require a rare, sadistic disposition. The actual 65% full-obedience rate suggested something closer to a situational pull that most ordinary people are susceptible to under the right conditions, not a trait limited to a few.
It describes how a steep hierarchy can discourage a subordinate, a co-pilot, a junior nurse, from voicing a safety concern to someone senior, even when they've noticed a real problem. Crew Resource Management training was developed in aviation partly in response to this pattern, specifically to flatten that gradient and normalize challenging a senior figure's decision when safety is at stake.
Training helps, but hospitals and airlines that have reduced authority-driven errors generally paired training with structural changes, mandatory callback verification, standardized checklists, blameless incident reporting, rather than relying on individuals to override the pull of authority through willpower alone in the moment it actually matters.