Confirmation Bias: Why You Only See What Proves You Right

Imagine you have decided that a new neighbor is unfriendly. She did not wave back on Monday. On Wednesday she walked past with her phone out, and you thought, "Of course." On Friday she smiled and held the door, and you decided she was probably just being polite because someone was watching. Three events, one conclusion, and the conclusion never had to change.

That pattern has a name, confirmation bias. It is the habit of searching for, noticing and remembering what supports what you already believe, and giving less room to what does not. It is not stupidity and it is not lying. It tends to happen quietly, to most people.

This guide covers where the idea came from, what the main experiments show, and what the pattern can look like in relationships, at work and online. It also covers how it can keep people in painful situations and in unfounded suspicion, and it ends with habits and reflection questions. Everything here is offered for reflection. It describes a common tendency and does not diagnose you or anyone else.

What Is Confirmation Bias?

Psychologist Raymond Nickerson, in a 1998 review in Review of General Psychology, wrote that the term "connotes the seeking or interpreting of evidence in ways that are partial to existing beliefs, expectations, or a hypothesis in hand" (Nickerson, 1998). That covers three habits worth pulling apart.

• Searching. You ask questions, pick sources and choose conversations that are likely to return a yes. If you suspect a friend is drifting, "You seem busy lately, is everything okay?" invites an answer that fits.
• Noticing and interpreting. Ambiguous events get read in the direction you already lean. A short reply is coldness if you expect coldness and a busy afternoon if you expect warmth.
• Remembering. The moments that fit the story are the ones that stick, and the ones that do not fit fade or get filed as exceptions. Nickerson's review describes research by Mark Snyder and colleagues suggesting that what people see or remember can match what they are looking for more closely than the facts justify.

Nickerson notes that the term usually refers to "unwitting selectivity in the acquisition and use of evidence," and he separates it from the deliberate case-building of a lawyer or debater, who knows the job is to argue one side. Here, nobody feels they are arguing. It feels like looking at the facts.

Having beliefs is not the problem. The trouble starts when the way you test a belief is arranged so that the belief cannot lose, because a belief that cannot lose is not really being checked.

Where Did the Idea Come From?

The observation is old. Lord, Ross and Lepper open their 1979 paper with a passage from Francis Bacon, written in 1620, saying that once the human mind has adopted an opinion, it draws all things else to support and agree with it, and that it neglects or sets aside the instances on the other side (Lord, Ross and Lepper, 1979).

The classic experiment dates to 1960. Psychologist Peter Wason published a task in the Quarterly Journal of Experimental Psychology in which people had to work out the rule behind a set of three numbers. As Nickerson's review describes it, the experimenter gave a triplet such as 2-4-6, the participant guessed a rule, and then the participant tested the guess by proposing new triplets and hearing whether each one fit the hidden rule.

People typically guessed something like "successive even numbers" and tested it by offering more even numbers, such as 8-10-12. Every one got a yes. But if the real rule was broader, such as "any numbers in increasing order," those tests could never reveal the mistake. To find it, a person had to try a triplet the guess said should fail, such as 7-3-1.

Nickerson makes a careful point about this. The feedback participants took as strong confirmation was not strongly confirming in a logical sense. In his words, the "confirmation" the participant receives "is, to some degree, illusory" (Nickerson, 1998). The review also notes that other researchers questioned how to interpret the task, so it is best treated as an illustration, not a final measurement.

A test that can only say yes feels like evidence, but it tells you very little.

Does It Get Stronger When We Care About the Answer?

Nobody has feelings about triplets of numbers. What happens when people care about the answer?

Biased assimilation. Lord, Ross and Lepper (1979) recruited 48 undergraduates, 24 who favored capital punishment and 24 who opposed it. Each read about two invented studies on whether it deters crime, one seeming to support deterrence and one seeming to undermine it. Both groups rated the study matching their existing view as more convincing, and the net effect was increased polarization: people who read the same mixed evidence ended up further apart (Lord, Ross and Lepper, 1979).

Motivated reasoning. In a 1990 paper in Psychological Bulletin, Ziva Kunda proposed that wanting a particular conclusion shapes how we access, build and judge beliefs. She summarized the evidence as showing that people are more likely to arrive at conclusions they want, but that they are limited by their ability to construct seemingly reasonable justifications for those conclusions (Kunda, 1990). We do not believe just anything. We look for a case that would sound plausible to a fair-minded observer, and then tend to find it. Kunda also wrote that the motivation to be accurate pushes people toward the beliefs and strategies they consider most appropriate.

Choosing what to read. Charles Taber and Milton Lodge (2006) ran two experiments in which students judged arguments about affirmative action and gun control. Arguments matching a student's prior attitude were rated stronger, opposing ones were counterargued, and when students could choose which sources to read, they tended to pick confirming ones. The biases were strongest among those with the strongest prior attitudes and the highest political knowledge (Taber and Lodge, 2006).

A caution is fair. These studies used student samples and contested political topics in a lab, and they describe average tendencies, not what any one person will do. Still, the picture is consistent: the stronger the stake, the more the mind acts like a lawyer and the less like a judge.

What Does It Look Like in Relationships?

Relationships combine high stakes with ambiguous evidence, which suits this bias well. It can show up in several ways.

Once a label is set, every act seems to prove it. Imagine you have concluded that your partner is selfish. He picks the restaurant, and that is selfish. He cooks dinner, and that is him making up for it. He forgets the milk, and that confirms it. The kind acts are not unseen so much as explained, while the unkind ones are counted at full value. This is close to what we describe in our guide to the horn effect, where one negative impression colors everything that follows. The same trap closes around a mind-reading accusation, where every denial gets filed as more proof.

The reverse is just as common. Imagine you have decided that someone is a good person at heart. They cancel plans at the last minute, so they must be stressed. They go cold for days after an argument, so they just need space. They check your phone, which shows how much they care. Each warning sign gets its own gentle explanation, and the conclusion survives untouched. No single explanation is unreasonable. The pattern across them may be worth a look.

First impressions give the bias a head start. What you decide early often becomes the reference point for what follows, a pull we describe in our guide to anchoring bias. Once it is in place, searching, noticing and remembering lean toward it.

There is also a quieter loop. Nickerson's review describes work by Snyder, Tanke and Berscheid (1977) in which expectations helped draw out the behavior expected. Men who had been told a woman was attractive behaved differently toward her on the phone, and their behavior evoked more desirable responses from the "attractive" partners. Expectations can become self-fulfilling. In a relationship, treating someone as cold may make them cooler, and the coolness then looks like proof (Nickerson, 1998).

How Can It Keep Someone in a Painful Situation?

This bias can keep a person in a situation that hurts. It does not require them to miss the painful parts. It works on how those parts are filed.

Imagine someone in a relationship or a job that regularly leaves them drained. The bad days get filed as exceptions: they were under pressure, anyone would snap. The good days get filed as the real thing: this is who they really are. Each choice is believable alone, and together they hold up a hopeful story that more evidence may never disturb.

Searching helps keep it going. They may look up "how to make a relationship work" instead of "signs a relationship is not working," or talk only to the friend who always says things will improve.

A second force is often at work. When what we do, such as staying, and what we see, such as being hurt, do not fit together, the mind tends to ease the discomfort by adjusting the story. We cover that tension in our guide to cognitive dissonance. Confirmation bias supplies the material, making reassuring evidence easier to find and the other kind easier to set aside.

None of this means a hopeful read is wrong. Sometimes the good days are the real pattern and the bad ones are the exception. The point is that the question "is this getting better, and is it good for me?" is hard to answer fairly when only one kind of evidence is being gathered. If you ever feel unsafe, please contact a local helpline or someone you trust.

How Can It Keep Someone in Suspicion?

The same bias can run in the other direction, toward suspicion. Here the belief is that something is wrong, and the evidence is gathered to match.

Imagine someone suspects their partner is hiding something. A late reply is proof. A fast reply is covering. An offer to show their phone is a performance. A calm denial is rehearsed, and an angry one is guilt. Whatever happens, the belief wins, the same trap as the 2-4-6 task.

It also tends to feed itself. Constant questioning and checking can make the other person defensive or distant, and that reaction then looks like confirmation, the self-fulfilling loop described above.

Suspicion is not always wrong. Sometimes it is picking up on something real, and a pattern of lies or secrecy deserves to be taken seriously. A useful test is a simple one: what would have reassured me? If there is a specific thing that would have settled the question, it can be asked about directly. If the honest answer is "nothing," then the belief is not being tested, and the person holding it may be carrying more distress than any evidence can fix.

Carrying a conclusion you cannot test can be exhausting, and it can be hard to step back from alone.

What Does It Look Like at Work?

At work, the bias often hides behind professional language. Imagine a manager who has decided one team member is a weak performer. That person's mistake goes in the notes. A colleague with a strong reputation makes the same mistake, and it is a blip. Over a year, one file is thick and the other thin, and both look like objective records.

It can run inward too. Someone who has decided they are not good at their job may count every critical comment and discount every compliment as politeness.

Teams show it too. Once a group has committed to a plan, the data shared in meetings tends to support the plan. Nobody needs bad intent. The person preparing the slides simply went looking for support, and found it. It often overlaps with the fundamental attribution error, where a colleague's slip is explained by who they are instead of what they were dealing with, and the explanation then becomes one more thing that the next slip confirms.

One counter is to ask before a decision, "What would we expect to see if this plan were wrong?" and to assign someone to look for exactly that.

What Does It Look Like Online?

Online, searching is effortless. You can type almost any belief into a search box and find someone agreeing with it, and the phrasing can steer the results: "Is this supplement dangerous?" and "Is this supplement safe?" often return different pages. Many platforms also tailor what you see to what you click.

Taber and Lodge's finding that students chose confirming sources when free to pick is a small version of this (Taber and Lodge, 2006).

There is a more personal version as well. A person reads a list of traits or symptoms, recognizes themselves in it, and feels a click of certainty. Nickerson's review mentions a study by Bertram Forer (1949) in which people readily accepted a personality sketch as a description of themselves, with little inclination to ask whether it would fit others as well (Nickerson, 1998). That applies to any quiz, including ours. A result can be a useful prompt for reflection, but it is not proof, especially when you were hoping for a particular answer.

What Habits Help You Catch It?

The bias is unlikely to disappear, but these habits may help you catch it more often.

1. Ask what would prove you wrong, and say it before you look. Write one sentence: "I would change my mind if I saw ___." Then look for that on purpose. If you cannot fill in the blank, that says something about how firmly the belief is held.
2. Try the triplet that should fail. In Wason's task, finding the flaw in a guess meant testing a case the guess said would not work. In daily life, that means asking a question where a "no" is possible, or searching for the strongest case against your view and reading it properly instead of skimming for flaws.
3. Ask someone neutral. Choose someone with no stake in the answer and no habit of taking your side. Describe what happened as plain events, leave out your conclusion, and ask what they make of it.
4. Make a two-column list. For a week or two, write down what points toward your belief on one side and what points against it on the other, including small things. Then look at the balance. A lopsided list can mean the belief is right, or that you have only collected one kind of evidence.
5. Slow down when you are certain and upset. When you feel strongly, the pull to build a convincing justification, the process Kunda describes, may be at its strongest. A day between the event and the conclusion costs little.

These habits do not ask you to doubt everything. They add the missing step of letting the belief be tested, so that a yes means something because a no was possible.

What Questions Are Worth Asking Yourself?

These are prompts, not a test. There are no right answers.

• What is one belief about a person or a situation that I hold strongly right now?
• What would I expect to see if that belief were wrong, and have I looked for it?
• Which sources, friends or searches do I turn to because they are likely to agree with me?
• Is there a recent event I explained away quickly? What was the other possible reading?
• If a friend described my situation to me, what would I notice that I may be missing?
• What is one small, safe way I could test my belief this week?

What Else Do People Ask?

What is confirmation bias?

It is the tendency to seek, notice and remember evidence that fits what you already believe, and to give less weight to evidence that does not. Researchers usually mean an unwitting habit, not deliberate dishonesty, and it is a common tendency, not a diagnosis.

Is confirmation bias the same as lying to yourself?

Not quite. In his 1998 review, Raymond Nickerson described it as usually unwitting, meaning a person is not aware of being selective. Lying to yourself implies knowing the truth, while confirmation bias often feels like seeing clearly.

Does being smart or well informed protect you from it?

Not reliably. In two experiments on political arguments, Charles Taber and Milton Lodge found biased evaluation was strongest among people with the strongest prior attitudes and the highest political knowledge. Participants were university students, so this does not cover every situation.

How can I tell if I am doing it?

Possible signs are relief when something agrees with you, irritation when it does not, asking only questions that can be answered yes, and being unable to say what would change your mind. None proves the bias is present, but each deserves a second look.

How can I reduce it?

You may not remove it completely, but you can catch it more often. Decide in advance what would count against your view, look for it on purpose, ask someone with no stake, and write down what points each way.

Sources: Nickerson, R. S. (1998), "Confirmation bias," Review of General Psychology, 2(2), 175 to 220, doi 10.1037/1089-2680.2.2.175. Lord, C. G., Ross, L., and Lepper, M. R. (1979), "Biased assimilation and attitude polarization," Journal of Personality and Social Psychology, 37(11), 2098 to 2109, doi 10.1037/0022-3514.37.11.2098. Kunda, Z. (1990), "The case for motivated reasoning," Psychological Bulletin, 108(3), 480 to 498, doi 10.1037/0033-2909.108.3.480. Taber, C. S., and Lodge, M. (2006), "Motivated skepticism," American Journal of Political Science, 50(3), 755 to 769, doi 10.1111/j.1540-5907.2006.00214.x. Wason, P. C. (1960), "On the failure to eliminate hypotheses," Quarterly Journal of Experimental Psychology, 12, 129 to 140, as described by Nickerson (1998). Snyder, Tanke, and Berscheid (1977) and Forer (1949) are as described by Nickerson (1998).

For self-reflection and educational purposes only. Nothing here is a diagnosis or a substitute for professional mental health support. Only a qualified clinician can assess or diagnose any condition. If you feel unsafe, please contact a local domestic abuse or crisis helpline.