Cognitive Biases: 20 Common Examples and How to Avoid Them

Ingrid has already decided to buy the house. She loved the kitchen the moment she walked in, and since then every piece of information has somehow confirmed it. The inspector's note about the roof? "Every old house has something." The long commute? "I'll listen to podcasts." The price? It seemed reasonable next to the first house she saw, which was absurdly overpriced.

Ingrid is not careless. She is smart and good with money. She is also human, which means her brain runs shortcuts she cannot see. Those shortcuts have names: confirmation bias, anchoring, and a few others we will meet below.

This guide is a practical list of cognitive biases: what they are, how they differ from logical fallacies, 20 common examples with a debiasing tip for each, a quick-reference table, and the techniques that hold up best in real decisions.

What is a cognitive bias?

A cognitive bias is a systematic pattern in how people judge and decide, where the error is predictable rather than random. It is not a lack of intelligence. It is a side effect of the mental shortcuts (psychologists call them heuristics) that let us make thousands of decisions a day without grinding to a halt.

The modern study of biases took off in the 1970s with the psychologists Amos Tversky and Daniel Kahneman. Their 1974 paper "Judgment under Uncertainty: Heuristics and Biases" described shortcuts like anchoring and availability, and their later prospect theory explained why losses loom larger than gains. Kahneman received the Nobel Memorial Prize in Economic Sciences in 2002 for this work; Tversky had died in 1996, and the prize is not awarded posthumously.

System 1 and System 2, in plain words

Kahneman's book Thinking, Fast and Slow popularized a simple picture of where biases come from, using terms borrowed from the psychologists Keith Stanovich and Richard West:

• System 1 is fast, automatic and effortless. It reads a face, finishes "bread and ...", and tells you a deal feels right before you know why.
• System 2 is slow, deliberate and effortful. It checks a contract clause and weighs two job offers line by line.

Most of the time System 1 is excellent. The trouble is that it answers quickly and confidently even when the question is hard, and System 2 is lazy about checking. Kahneman described the two systems as a useful metaphor, not two separate parts of the brain. Treat them that way.

Cognitive bias vs logical fallacy

People mix these up constantly. A cognitive bias is a pattern in how a mind processes information, and it usually operates unnoticed. A logical fallacy is a flaw in the structure of an argument, something you can point to on the page. You catch biases with process and habits; you catch fallacies by inspecting the argument.

The two are related, because biases often produce fallacies. The bandwagon effect is the tendency to follow the crowd; the bandwagon fallacy is the argument that something is true because it is popular. For the argument side of the story, our guide to logical fallacies covers 25 of them with examples. This article is about the mind that produces the arguments.

How we see evidence

These biases shape what information we notice, how much weight we give it, and how we read it.

1. Confirmation bias

Definition: Searching for, interpreting and remembering information in a way that supports what you already believe.

Example: Mateo thinks his new manager dislikes him. When she skips his update in a meeting, that proves it. When she praises his report, she is "just being polite."

Debias: Before you look for evidence, write down what would change your mind, then go looking for that first.

2. Anchoring bias

Definition: Relying too heavily on the first number you encounter when making an estimate.

Example: Ingrid's first house was listed far above its value, so every house after it looked cheap, including one that was actually overpriced. Salary talks work the same way: the first number mentioned tends to pull the final one toward it.

Debias: Build your own estimate before you see anyone else's number.

3. Availability heuristic

Definition: Judging how likely something is by how easily examples come to mind.

Example: After a week of news about a plane incident, Zara drives eight hours instead of flying, even though driving is generally the riskier way to travel. Vivid, recent events feel more frequent than they are.

Debias: Ask "how often does this actually happen?" and look for a real rate rather than a memorable story.

4. Framing effect

Definition: Reacting differently to the same information depending on how it is presented.

Example: A treatment with a "90% survival rate" sounds better than one with a "10% mortality rate," though they are identical. In Tversky and Kahneman's classic "disease problem," people's choices flipped depending on whether outcomes were described as lives saved or lives lost.

Debias: Restate the choice as a gain, then as a loss, and see if your preference survives both.

5. Survivorship bias

Definition: Drawing conclusions from the cases that made it through a filter while ignoring those that did not.

Example: Brendan reads interviews with college dropouts who built huge companies and concludes dropping out is smart. He never hears from the much larger group who dropped out and built nothing, because nobody interviews them.

Debias: Ask "who is missing from this picture?"

How we judge ourselves

These biases distort how we see our own knowledge, skill and track record.

6. Overconfidence bias

Definition: Being more certain of your judgments than your accuracy justifies.

Example: Lucia is "sure" a project will take two weeks and "certain" she remembers the contract deadline. She is wrong on both, and her certainty stopped her from checking.

Debias: Put a number on your confidence ("70% sure") and track how often you are right at each level.

7. Dunning-Kruger effect

Definition: The finding, from a 1999 paper by Justin Kruger and David Dunning, that people with low skill in an area tend to overestimate their performance, partly because judging the task takes the same skills as doing it.

Example: After two weekends of trading videos, Hassan feels ready to actively manage his savings. He does not yet know what he does not know.

Honest note: This is one of the most debated findings on the list. Some researchers argue much of the classic pattern can be produced by statistical artifacts, such as regression to the mean plus a general tendency to rate ourselves above average. The popular "Mount Stupid" chart is an internet meme, not a figure from the paper. The modest idea (beginners often misjudge their skill) is still a useful caution, not a law of nature or an insult to throw at people.

Debias: Get feedback from someone more skilled before you raise the stakes.

8. Hindsight bias

Definition: The "I knew it all along" feeling: once you know the outcome, it seems more predictable than it was.

Example: After a startup fails, everyone on the team remembers having doubts. Few wrote them down at the time.

Debias: Keep a decision journal (more below). Your written prediction is the only reliable record of what you knew.

9. Self-serving bias

Definition: Crediting successes to your skill and failures to bad luck or other people.

Example: When Renata's campaign beats its targets, it was her strategy. When the next one flops, it was the algorithm and the budget.

Debias: For every outcome, win or lose, list one factor you controlled and one you did not.

10. Bias blind spot

Definition: Seeing bias clearly in others while believing you are less affected by it.

Example: Wendell reads this article and thinks of five colleagues who should read it too. He does not think of himself.

Debias: Assume you are as biased as anyone, and build processes and invite people who will catch it for you.

How we decide under uncertainty

These biases show up when we choose between options with unknown outcomes, especially with money and time.

11. Sunk cost fallacy

Definition: Continuing to invest in something because of what you have already put in, rather than what you will get out of it from here.

Example: Mateo has spent 18 months and a chunk of savings on a side business that is not growing. Quitting feels like wasting it all. But the money and months are gone either way; the only question is whether the next 18 months are worth it.

Debias: Ask: "If I were starting fresh today, would I choose this?"

12. Loss aversion

Definition: Losses tend to feel more painful than equivalent gains feel good, so people take extra risks to avoid them. It is central to prospect theory.

Example: Zara holds a stock that has fallen for two years because selling would make the loss "real," while quickly selling winners to lock in gains.

Honest note: Loss aversion is well established as a general pattern, but researchers still debate how large it is and how much it depends on context. Be wary of any single precise number.

Debias: Judge each holding as if you had cash instead: "Would I buy this today at this price?"

13. Planning fallacy

Definition: Underestimating how long a task will take and what it will cost, even when similar tasks ran over before. Kahneman and Tversky named it in 1979.

Example: Brendan plans his kitchen renovation for six weeks. His last two home projects took about twice as long as planned, but this one feels different.

Debias: Start from how long similar projects actually took, not from your plan.

14. Base rate neglect

Definition: Ignoring how common something is in general and focusing on vivid details of the case in front of you.

Example: Lucia meets a quiet, bookish stranger and guesses librarian rather than salesperson, ignoring that there are far more salespeople than librarians.

Debias: Start every estimate with the base rate, then adjust for the details.

15. Status quo bias

Definition: Preferring things to stay the same, because change feels risky and staying put feels like no decision at all.

Example: Hassan has kept the same expensive phone plan and underperforming retirement fund for years. Switching would take an afternoon. Not switching costs him every month.

Debias: Treat "keep things as they are" as an active choice and give it the same scrutiny as the alternatives.

Social biases

These biases come from the fact that we think in groups and care what others think.

16. Bandwagon effect

Definition: Adopting a belief or behavior because many others have.

Example: Renata's company rushes to add a feature because competitors did, without checking whether its own customers asked for it.

Debias: Ask "would I do this if nobody else were?"

17. Halo effect

Definition: Letting one positive trait (charm, looks, a famous employer) color your judgment of unrelated traits.

Example: A polished candidate interviews well, and Wendell rates his technical answers higher than they deserved. A nervous candidate with better answers scores lower.

Debias: Score each quality separately, before forming an overall impression. That is why structured interviews exist.

18. Groupthink

Definition: A group's desire for harmony suppresses dissent and leads to a poor decision. The psychologist Irving Janis popularized the term in the 1970s.

Example: The senior manager clearly loves the launch date. Two people have doubts, see nodding heads, and stay quiet. Everyone leaves believing the team agreed.

Debias: Collect written views before discussion, have the most senior person speak last, and assign someone to argue the other side. Our guide to the devil's advocate covers how to do that without it becoming theater.

19. In-group bias

Definition: Favoring people in your own group (team, town, political side) and judging outsiders more harshly.

Example: When someone in Zara's department misses a deadline, it is an understandable crunch. When someone in another department does, they are disorganized.

Debias: Swap the labels and ask how you would judge the same behavior from your own side.

20. Fundamental attribution error

Definition: Explaining others' behavior by their character and your own by circumstances.

Example: A driver cuts Mateo off: "what a jerk." When Mateo cuts someone off, it is because he was late for school pickup and the lane ended suddenly.

Debias: Before judging, list two situational explanations for what the person did.

Quick-reference: 20 cognitive biases at a glance

A note on the wider research: psychology's replication crisis hit some famous findings hard. Ego depletion, the idea that willpower runs out after use, did not show up in a large multi-lab replication effort. Treat any single striking study, including those in pop psychology books, with healthy caution.

Debiasing techniques that actually help

Here is the uncomfortable part: simply knowing about cognitive biases does not reliably protect you from them. What helps more is changing the process around a decision.

The pre-mortem

The psychologist Gary Klein proposed the pre-mortem: before committing to a plan, imagine it is a year later and the plan has failed. Everyone writes down, independently, why it failed. Then you discuss. "What could go wrong?" invites polite hedging. "It went wrong. Why?" invites specifics, and gives people permission to voice doubts that groupthink would bury.

Consider the opposite

When you notice a strong belief, ask: "What are the reasons I might be wrong?" Research from the 1980s found that instructing people to consider the opposite reduced biased evaluation of evidence better than simply telling them to be fair. It is the core of steelmanning: building the strongest version of the view you reject.

The outside view and base rates

Kahneman and Dan Lovallo contrasted the inside view (the details of your plan) with the outside view (how similar projects usually turn out). Find a reference class, such as "how long do kitchen renovations like mine usually take?", start there, then adjust for what is truly different about your case. It is the best single fix for the planning fallacy.

Red team

A red team attacks a plan as a determined opponent would. A small version works anywhere: ask one colleague to spend an hour trying to break your proposal. Because attacking is their assigned role, they pay no social price for disagreeing.

Decision journal

Before an important decision, write down what you chose, why, what you expect, and how confident you are. Revisit it months later. A journal fights hindsight bias, overconfidence and self-serving bias, and it separates good decisions from good outcomes: a sound choice can still turn out badly, and a sloppy one can get lucky.

Cooling-off period

Emotion and urgency make System 1 louder. For big purchases, angry emails and offers that "expire tonight," build in a pause, such as 48 hours for anything over a set amount. High-pressure sales tactics are designed to prevent exactly that pause, which tells you how well it works.

For a full decision process built on these ideas, see how to make better decisions with AI.

Cognitive biases at work and in money decisions

Biases do the most damage where stakes are high and feedback is slow.

At work

• Hiring: The halo effect and in-group bias favor candidates who resemble the interviewer. Use the same questions and score each area separately.
• Projects: The planning fallacy and sunk costs combine: the project runs late, and because so much has been spent, nobody wants to stop. Set kill criteria up front: "If we have not hit X by date Y, we stop or re-plan."
• Meetings: Anchoring and groupthink mean the first confident opinion often becomes the decision. Collect written opinions first.
• Strategy: Survivorship bias makes "copy the winners" look smarter than it is. Before a new product, validate the startup idea before building with evidence that could prove you wrong.

A skeptical finance voice is a quick outside view, which is the role James, the CFO character, plays in a debate.

With money

• Investing: Loss aversion keeps losers open, overconfidence encourages frequent trading, and the bandwagon effect pulls people into whatever is rising.
• Big purchases: Anchoring on list prices and framing like "only $19 a month" hide the total cost. Always calculate the full amount.
• Subscriptions: Status quo bias keeps you paying for things you no longer use. Review recurring charges on a schedule.

Here is a classic decision where several biases pile up at once:

How AI chatbots can amplify confirmation bias

There is a new source of confirmation bias many people have not noticed: the chatbot on their phone.

AI assistants are trained partly on human feedback, and people tend to rate agreeable answers highly. The result is a documented tendency researchers call sycophancy: models leaning toward agreeing with the user, softening criticism, or abandoning a correct answer when pushed. Our article on why AI agrees with you explains how this happens.

Combine that with confirmation bias and you get a loop. Ingrid asks, "Is it smart to buy a house with a long commute if I love the kitchen?" An agreeable assistant reinforces the lean, and she feels her decision was "checked," when it was really echoed.

A few habits help: ask neutrally ("What are the strongest reasons not to buy this house?"), ask for the opposite case first, do not reveal your preference until you have heard the arguments, and get more than one voice. One assistant agreeing with you is weak evidence. Several arguing with each other is harder to fake.

Using an AI debate as a debiasing tool

That last point is the idea behind Debatly. Instead of one assistant answering you, several AI models and characters debate your question over rounds, arguing different sides, and an AI judge weighs the arguments and gives a verdict with reasoning. It is a structured way to consider the opposite and red team your own decisions.

• Against confirmation bias: Put your preferred option forward and let Petra, the devil's advocate, attack it.
• Against overconfidence and the planning fallacy: Add Viktor, the skeptic, to question your assumptions and your numbers.
• Against sunk costs and status quo bias: Frame the question as a fresh choice ("Starting today, should I continue or stop?") and let the debaters argue both options.
• Against one confident answer: Read the verdict from Judge Alex and, more importantly, the reasoning. If the judge credits an argument you dismissed, that is the one to think about.

New to the format? What an AI debate room is explains how a session runs. The AI decision maker is a good start for personal choices, and the debate room lets you set up your own question and pick the participants.

Lucia's routine before a big decision: write her preferred option and her confidence as a percentage, run a debate with a devil's advocate, write down the single strongest argument against her choice, then wait 48 hours before acting. Sometimes she changes her mind. More often she keeps her choice but adds a safeguard she would have missed.

A good one to try, because sunk costs make it hard to see clearly:

An honest limit

AI debaters have their own biases. Models learn from human writing, so they can reflect common assumptions, favor popular views, sound confident when wrong, and miss context only you know. A judge's verdict is a reasoned opinion, not a fact. Use a debate to surface arguments you had not considered, then check the important claims yourself and talk to people who know your situation.

Exercises to catch your own biases

1. The bias audit. Pick a recent decision, go through the quick-reference table, and mark every bias that could have played a role. Pick the likeliest one and ask what you would do differently.
2. Calibration practice. For a week, attach a confidence percentage to small predictions. At the end, check: are your 80% calls right about 80% of the time?
3. Consider the opposite, in writing. Take a belief you hold strongly and write three reasons it might be wrong.
4. Find the base rate. Before your next estimate, find a reference class and a rough rate before looking at your specific case.
5. The fresh-start test. List three things you continue mainly because you started them. Would you choose each of them today?

Start with one decision this week

Knowing about biases is not enough. Change the process: run a pre-mortem, consider the opposite, start from base rates, invite a red team, write decisions down, and give yourself time to cool off.

Pick one real decision this week, write down your preferred option and your confidence, and look for the best case against it.

If you want that case made as forcefully as possible, run it as a debate on Debatly and see how the judge weighs it. Give it a try, starting with the decision you feel most sure about. Plans are on the pricing page.