Authority bias
We over-defer to experts, titles, and symbols of authority.
What it means
Authority bias is the tendency to attribute greater accuracy to, and comply more readily with, the views of a figure perceived as an authority, sometimes triggered by mere symbols such as titles, uniforms, or badges, independent of the argument's content. The mechanism is a heuristic shortcut: authorities often do know more, so the mind treats authority cues as a fast proxy for correctness. The danger is that the cue detaches from genuine competence: people move 48 percent of the way toward advice framed as high quality, against 32 percent for the same advice framed as low quality. It is conventionally tied to Milgram's obedience research, though that reading is contested — in an online analogue of the paradigm the prompt most resembling a direct order worked worst — and to the halo effect, where one authoritative quality colors unrelated judgments. It matters in advertising, medicine, finance, and organizations, where the right response is to weigh the argument and the track record, not the symbols of status.
Why it happens
Deference to authority is cheap cognition. Working out whether a claim is true is expensive; checking whether the speaker has a title is nearly free, and across most of life the title correlates with being right, so the shortcut pays. What turns it into a bias is that the shortcut does not merely tilt judgment, it can replace it. In one neuroimaging study, participants chose between certain payments and lotteries while an economist's recommendation was shown or withheld. Without advice, activity in valuation regions tracked the lottery's value. With advice present, that tracking disappeared, and participants followed the expert's deliberately suboptimal rule on 73 percent of trials, against 64 percent when no advice was shown at all. The authors called it offloading: the expert's presence switched the private calculation off rather than informing it. That reading rests on a null result in 24 scanned participants, though, and a correlation falling below threshold is weaker evidence than a computation switching off.
What the evidence shows
The headline evidence is durable but narrower than it looks. Burger's partial replication of Milgram found 70 percent of participants still continuing at the point where Milgram had recorded 82.5 percent, a difference that was not statistically significant, and replications in Poland have since produced obedience near 90 percent. The phenomenon travels, even if the original dataset is shakier than its fame suggests: archival work documents that Milgram's experimenter improvised coercion well beyond the four scripted prods, that obedience varied enormously across two dozen largely unreported variants, and that many participants doubted the shocks were real. A cleaner estimate comes from advice-taking research, which measures how far people move toward an advisor's number. A meta-analysis of 346 effect sizes from more than 17,000 participants put the average at 0.39, and perceived advisor quality was the only unique predictor: people moved 48 percent of the way toward advice framed as high quality, against 32 percent for low quality. Credentials measurably reprice a judgment, even when the underlying argument is identical.
Where the authority reading breaks down
The Milgram interpretation that gives this bias its name is contested. When Haslam and colleagues ran an online analogue, in which participants were asked to describe ever-more-likeable groups with ever-nastier adjectives, and varied which of Milgram's four prods pushed them onward, the one most like a naked order, the insistence that you have no other choice and must go on, was least effective and most likely to make them quit. What predicted continuation was the appeal to the scientific project, which points to identification with a cause rather than submission to a badge. Writing rude adjectives on a website is a long way from believing you are electrocuting a stranger, so that inference travels only as far as the analogue does. The prod finding itself survives in the real paradigm: an analysis of Milgram's own archived recordings found participants typically stopped rather than continued after the fourth prod, and a coding of Burger's replication found none carried on after hearing it. Complicating both accounts, a 2025 replication found obedience unmoved by whether the study ran in a university or a rented flat, and unmoved by whether it was billed as science or as a marketing test: 50 of 58 participants went to the last switch regardless (p = .607). With fewer than 60 participants it could not detect a modest effect, but it is not evidence for the science account either.
Using it in practice
The practical lever is not rank but knowledge and the standing to use it. When Rank and Jacobson repeated the Hofling study, in which 21 of 22 nurses were willing to give an unfamiliar drug on an unknown doctor's telephoned order, three things changed at once: the drug was one the nurses knew well, they were allowed to consult colleagues first, and, cutting against the result, the order now came from a real staff physician they knew rather than Hofling's total stranger. Sixteen of eighteen refused. Authority went up; compliance collapsed. The design cannot tell us which change did the work, but knowledge and the standing to use it are the plausible candidates. That is the design brief: give the junior party the facts and a legitimate route to raise them. Note too that people underweight advice on average, so the failure is rarely blanket credulity. It is that status cues quietly decide which advice gets checked at all.
Examples
An actor in a doctor's coat sells more medicine; people defer to a confident 'expert' outside their field.
A fund manager in an expensive suit with a corner office attracts money on the strength of the trappings, long before anyone checks whether the returns beat a plain index tracker.
A junior engineer spots the bug but says nothing because the principal engineer wrote the code — the same objection would have been raised in a second to a peer.
A radiologist's preliminary read is signed off by a junior who has independently spotted the nodule the senior missed, but downgrades their own confidence because of who wrote the first read; the same image, re-rated on the strength of a name.
A council adopts a traffic scheme because a famous consultancy's partner presented it in person; nobody asks to see the underlying model until the junctions are already built and jammed.
First described in Cialdini; Milgram's obedience research.
Key references
- Grzyb, T., Dolinski, D., Sudol-Malisz, M., Kulik, G., & Mielczarek, L. (2025). Doubting the power of prestige: Obedience to authority beyond institutional and research justifications. Scientific Reports, 15, 24911. doi.org/10.1038/s41598-025-10331-4
- Bailey, P. E., Leon, T., Ebner, N. C., Moustafa, A. A., & Weidemann, G. (2023). A meta-analysis of the weight of advice in decision-making. Current Psychology, 42(28), 24516-24541. doi.org/10.1007/s12144-022-03573-2
- Haslam, S. A., Reicher, S. D., & Birney, M. E. (2014). Nothing by mere authority: Evidence that in an experimental analogue of the Milgram paradigm participants are motivated not by orders but by appeals to science. Journal of Social Issues, 70(3), 473-488. doi.org/10.1111/josi.12072
- Gibson, S. (2013). Milgram's obedience experiments: A rhetorical analysis. British Journal of Social Psychology, 52(2), 290-309. doi.org/10.1111/j.2044-8309.2011.02070.x
- Burger, J. M., Girgis, Z. M., & Manning, C. C. (2011). In their own words: Explaining obedience to authority through an examination of participants' comments. Social Psychological and Personality Science, 2(5), 460-466. doi.org/10.1177/1948550610397632
- Burger, J. M. (2009). Replicating Milgram: Would people still obey today? American Psychologist, 64(1), 1-11. doi.org/10.1037/a0010932
- Dolinski, D., Grzyb, T., Folwarczny, M., Grzybala, P., Krzyszycha, K., Martynowska, K., & Trojanowski, J. (2017). Would you deliver an electric shock in 2015? Obedience in the experimental paradigm developed by Stanley Milgram in the 50 years following the original studies. Social Psychological and Personality Science, 8(8), 927-933. doi.org/10.1177/1948550617693060
- Perry, G. (2012). Behind the Shock Machine: The Untold Story of the Notorious Milgram Psychology Experiments. Melbourne: Scribe Publications. archive.org/details/behindshockmachi0000perr_b8r1
- Engelmann, J. B., Capra, C. M., Noussair, C., & Berns, G. S. (2009). Expert financial advice neurobiologically 'offloads' financial decision-making under risk. PLoS ONE, 4(3), e4957. doi.org/10.1371/journal.pone.0004957
- Hofling, C. K., Brotzman, E., Dalrymple, S., Graves, N., & Pierce, C. M. (1966). An experimental study in nurse-physician relationships. Journal of Nervous and Mental Disease, 143(2), 171-180. pubmed.ncbi.nlm.nih.gov/5957275/
- Rank, S. G., & Jacobson, C. K. (1977). Hospital nurses' compliance with medication overdose orders: A failure to replicate. Journal of Health and Social Behavior, 18(2), 188-193. pubmed.ncbi.nlm.nih.gov/881549/