Cognitive Biases: 20 Common Thinking Errors Explained

InteractiveSpot these biases in everyday scenarios. Open Bias or Fallacy? →

A cognitive bias is not a random slip. It is an error with a shape: it shows up in the same direction, in the same kinds of situations, across very different people. That predictability is the core insight of Amos Tversky and Daniel Kahneman's 1974 paper in Science, which argued that people rely on a few mental shortcuts (they named representativeness, availability, and anchoring) that are "quite useful" most of the time but lead to "severe and systematic errors" in others (Tversky & Kahneman, 1974).

This page collects twenty of the biases you are most likely to meet in daily life, explains the research behind the big ones, and is honest about which famous findings have wobbled under replication. If you want the argument-level cousins of these errors, see our guide to logical fallacies: a bias is a flaw in how you judge; a fallacy is a flaw in how an argument is built.

The 20 biases at a glance#

Bias	What it is	Everyday example	Countermeasure
Confirmation bias	Seeking and weighting evidence that fits what you already believe	Only reading reviews that agree with the phone you already want	Ask "what would change my mind?" before you look
Anchoring	Estimates drift toward the first number you see	A "was $400, now $199" tag makes $199 feel cheap	Form your own estimate before seeing anyone else's
Availability heuristic	Judging frequency by how easily examples come to mind	Fearing plane crashes more than car trips after news coverage	Look up the actual rate
Base-rate neglect	Ignoring how common something is in favour of a vivid description	Assuming a quiet, bookish stranger is a librarian rather than a salesperson	Start with "how common is this in general?"
Dunning–Kruger effect	Low performers overestimating their performance (contested; see below)	Feeling sure you aced a test you failed	Get external, objective feedback
Sunk cost fallacy	Continuing because of what you have already spent	Sitting through a bad film because the ticket was expensive	Decide as if starting fresh today
Survivorship bias	Drawing conclusions only from the cases that made it through a filter	"Dropouts become billionaires," based on famous dropouts	Ask where the failures went
Hindsight bias	"I knew it all along" after the outcome is known	Calling a market crash obvious, afterwards	Write predictions down in advance
Framing effect	Same facts, different choice depending on wording	"90% survival" sounds better than "10% mortality"	Restate the choice the other way round
Loss aversion	Losses loom larger than equal gains	Refusing a fair coin-flip bet	Judge the whole portfolio, not each loss
Status quo bias	Preferring the current state of affairs by default	Never switching a pricey phone plan	Schedule a yearly "would I choose this today?" review
Planning fallacy	Underestimating how long your own tasks will take	The weekend project that takes a month	Use how long similar past projects really took
Fundamental attribution error	Explaining others' behaviour by character, not circumstances	"That driver is a jerk" (they may be rushing to a hospital)	Ask what situation would make you act the same way
Halo effect	One good trait colours judgments of unrelated traits	Assuming an attractive candidate is also competent	Score criteria separately, before an overall rating
Bias blind spot	Seeing bias in others more readily than in yourself	"I'm immune to advertising"	Use processes, not self-assessment
Law of small numbers	Treating small samples as representative	Judging a restaurant on one visit	Ask how many observations you really have
Optimism bias	Expecting good outcomes for yourself more than is likely	Skipping insurance because "it won't happen to me"	Plan for the median case, not the best case
IKEA effect	Overvaluing things you helped build	Loving your own wobbly bookshelf	Get a neutral person to evaluate it
Choice overload	Too many options leading to worse or no decisions (mixed evidence)	Leaving a store without buying because there were 40 versions	Narrow to three before comparing
Bandwagon effect	Adopting beliefs because many others hold them	Buying a stock because "everyone" is	Ask what you would think if nobody else did

Confirmation bias#

Raymond Nickerson's widely cited review called confirmation bias "a ubiquitous phenomenon in many guises": people seek evidence that supports a hypothesis they already hold, interpret ambiguous evidence in its favour, and give less weight to evidence against it (Nickerson, 1998). It is especially dangerous because it feels like research. You did look at evidence; you just looked in the places most likely to agree with you.

Wrong: "I searched and found ten articles saying this supplement works."

Right: "I searched for the best evidence that it doesn't work, and here's why it didn't convince me."

Anchoring#

In the classic demonstration, participants watched a wheel of fortune produce a number (different groups received 10 and 65) and were then asked whether the percentage of African countries in the United Nations was higher or lower than that number, and then for their own estimate. The group that saw 10 gave a median estimate of 25; the group that saw 65 gave 45, even though the number obviously had nothing to do with the question (Tversky & Kahneman, 1974). Anchoring is one of the sturdiest effects in the literature: it was among the effects tested in the large Many Labs replication, where only flag priming and currency priming clearly failed (Klein et al., 2014).

Availability#

We judge how likely something is by how easily examples come to mind (Tversky & Kahneman, 1973). That usually works, because common things are easier to recall. It fails when memorability and frequency come apart: dramatic events, recent news, and personal anecdotes are easy to recall without being common. Scammers exploit the same machinery in reverse: an urgent text about a "suspicious charge" makes one frightening scenario vivid enough to crowd out the far more likely explanation, which is that the message itself is the scam. allthingsN has a practical walkthrough on how to spot a scam text before you tap.

Base-rate neglect#

When given a personality sketch, people tend to judge which profession someone has by how well the description matches a stereotype, largely ignoring how many people are in each profession to begin with (Kahneman & Tversky, 1973). The same error drives misreadings of medical tests: if a condition is rare, even an accurate test can produce more false positives than true positives. Always ask: before I heard this detail, how likely was it?

The Dunning–Kruger effect, honestly#

Justin Kruger and David Dunning's 1999 paper reported that participants who scored lowest on tests of humour, grammar and logical reasoning markedly overestimated how well they had done, and argued that the skills needed to perform well are the same skills needed to recognise good performance (Kruger & Dunning, 1999).

The finding has a real statistical problem. If everyone's self-estimates cluster toward "a bit above average" and actual scores are bounded between 0 and 100, the bottom group will appear to overestimate and the top group to underestimate, even with no special metacognitive deficit. A 2022 paper in Frontiers in Psychology argues the effect can be derived largely as such an artifact (Magnus & Peresetsky, 2022), and McGill's Office for Science and Society summarises the critiques under the blunt headline that the effect is "probably not real" (McGill OSS).

Sunk cost#

Hal Arkes and Catherine Blumer showed that people keep investing in something because of what they have already put in. In one scenario, many participants chose a more expensive ski trip they expected to enjoy less over a cheaper one they expected to enjoy more, after learning they had accidentally booked both. In a field study, theatre customers randomly given discounted season tickets attended fewer plays in the first half of the season than those who paid full price (Arkes & Blumer, 1985). Money already spent is gone either way; only future costs and benefits should drive the decision. The sunk cost fallacy and status quo bias together explain a lot of forgotten monthly charges. If that sounds familiar, here is a guide to finding and cancelling forgotten subscriptions, and more on the money side in common money mistakes.

Survivorship bias and Abraham Wald#

During the Second World War, the statistician Abraham Wald worked on how to armour aircraft. The naive approach is to reinforce the places where returning planes show the most damage. Wald's insight, reconstructed by Marc Mangel and Francisco Samaniego, was that the planes being examined were the ones that survived: damage in the places where returning planes were clean was likely what brought the others down (Mangel & Samaniego, 1984). The same filter distorts business advice drawn only from successful companies, investment strategies judged only on funds that still exist, and "they don't make them like they used to" judgments based on the old things that happened to last.

Hindsight bias#

Baruch Fischhoff found that once people learn an outcome, they rate it as having been more predictable than people without that knowledge do, and they often fail to notice their judgment has shifted (Fischhoff, 1975). Hindsight bias makes post-mortems unfair ("how did nobody see this coming?") and stops us learning, because a surprise that feels obvious in retrospect teaches nothing. It is a big reason why engineering incident reviews, like those behind the famous software bugs on this site, focus on what was knowable at the time.

Framing, loss aversion and the rest#

Describing the same options in terms of lives saved or lives lost reliably flips people's preferences between a sure thing and a gamble (Tversky & Kahneman, 1981), and prospect theory formalised the observation that losses weigh more heavily than equivalent gains (Kahneman & Tversky, 1979). Other well-documented patterns in the table include the planning fallacy (Buehler et al., 1994), the halo effect (Nisbett & Wilson, 1977), the fundamental attribution error (Ross, 1977), the bias blind spot (Pronin et al., 2002), the "law of small numbers" (Tversky & Kahneman, 1971), optimism bias (Sharot, 2011) and the IKEA effect (Norton et al., 2012).

Replication caveats: not every famous bias survived#

Psychology went through a reckoning in the 2010s. When the Open Science Collaboration repeated 100 published studies, 97% of the originals had reported significant results but only 36% of the replications did (Open Science Collaboration, 2015). Some casualties are relevant here:

Ego depletion (the idea that willpower is a resource that runs down): a preregistered test across 23 labs and more than 2,000 participants found an effect close to zero (Hagger et al., 2016).
Behavioural priming: a replication of the famous claim that reading words about old age makes people walk more slowly found no effect when experimenters were blind to condition (Doyen et al., 2012). Flag and currency priming also failed in Many Labs (Klein et al., 2014).
Choice overload: the well-known jam-tasting study found that shoppers offered fewer varieties were more likely to buy (Iyengar & Lepper, 2000), but a meta-analysis of 50 experiments found a mean effect of virtually zero, with large variation between studies (Scheibehenne et al., 2010).

The heuristics-and-biases core (anchoring, framing, sunk cost) has generally held up better than the more surprising social-priming claims. Treat any single striking study, including ones on this page, as a hypothesis rather than a law.

A debiasing checklist#

Write your prediction and confidence down before the outcome is known.
Look up the base rate before you look at the specific case.
Make your own estimate before you see anyone else's number.
Run a pre-mortem: imagine the plan failed, then list why.
Ask someone to argue the opposite side, and listen.
For ongoing commitments, ask "would I sign up for this today?"

Frequently asked questions

What is a cognitive bias?

A cognitive bias is a systematic, predictable deviation from rational judgment. Tversky and Kahneman's 1974 paper in Science argued that mental shortcuts (heuristics) are usually efficient but produce characteristic errors, which is why biases show up reliably across people rather than as random noise.

Is the Dunning-Kruger effect real?

The 1999 paper found that low scorers overestimated their performance. Later work, including a 2022 paper in Frontiers in Psychology, argues much of that pattern can arise as a statistical artifact of regression and score boundaries. The safest reading: people are generally poor at judging their own skill, but the popular 'stupid people are the most confident' cartoon overstates the evidence.

Did all the famous bias studies replicate?

No. A 2015 effort to replicate 100 psychology studies got statistically significant results in about 36% of cases, and a 23-lab test of ego depletion found an effect close to zero. Anchoring, framing and sunk-cost effects did replicate in the Many Labs project, while flag and currency priming did not.

Can you get rid of cognitive biases?

Not entirely, and knowing about them is not enough: people readily spot bias in others but not themselves (the 'bias blind spot'). Process fixes work better than willpower: checklists, pre-mortems, writing predictions down, considering base rates first, and asking someone to argue the other side.

Sources

Every factual claim on this page links to one of these. Checked October 2026. See how we source.

The 20 biases at a glance#

Confirmation bias#

Anchoring#

Availability#

Base-rate neglect#

The Dunning–Kruger effect, honestly#

Sunk cost#

Survivorship bias and Abraham Wald#

Hindsight bias#

Framing, loss aversion and the rest#

Replication caveats: not every famous bias survived#

A debiasing checklist#

Frequently asked questions

Sources

Keep reading