Confirmation Bias Explained: Why You Only See What You Already Believe

You decide a colleague dislikes you. Their short reply becomes evidence. Their friendly greeting becomes politeness. Their delayed message becomes more evidence. Before long, almost anything they do can fit the same story. The belief feels increasingly solid, even though you have quietly made it difficult for any observation to count against it.

Confirmation bias describes tendencies to seek, interpret or remember information in ways that favour what we already believe. It does not mean every belief is wrong, or that people consciously falsify evidence. Often the process feels like ordinary careful thinking. We ask sensible questions inside a frame we have never tested, then mistake the accumulation of compatible answers for a serious challenge to that frame.

Wason's numbers and the missing challenge

In Peter Wason's classic 1960 rule-discovery study, participants were given the number sequence 2, 4, 6 and asked to identify the rule it followed. They could propose other triples and receive feedback about whether those triples satisfied the experimenter's rule. The actual rule was broader than many participants initially guessed: numbers increasing in order.

Someone who suspects "even numbers increasing by two" might try 8, 10, 12 and receive a positive answer. That is compatible with the guess, but also with many other rules. Trying 1, 2, 3 would be more informative because success would contradict the proposed restriction to even numbers. A useful test separates competing explanations rather than merely generating another example they both predict.

Many participants had difficulty abandoning narrower hypotheses, illustrating the limits of seeking positive instances. This is the 2-4-6 rule-discovery task, not Wason's separate card-selection task. The distinction matters because familiar psychology stories are easily compressed into a single inaccurate anecdote. The original study concerned a particular reasoning task, not a numerical estimate of how biased everyone is in daily life.

Why confirmation feels like being right

Supportive information is often easy to process because it fits an existing explanation. Contradictory information requires extra work: perhaps the source is wrong, perhaps the circumstances differ, or perhaps the explanation needs revising. Any of these could be true. Bias appears when we apply demanding scrutiny only to the information we dislike and accept friendly evidence with much less inspection.

Beliefs can also become attached to identity, relationships or past decisions. Changing an opinion may feel like admitting that your group was mistaken or that an expensive purchase was unwise. This does not require conscious dishonesty. It creates an emotional reason to notice some possibilities sooner than others. The useful question is whether your standards of evidence remain consistent across the conclusions you prefer and reject.

News feeds can narrow the questions

A feed learns from what attracts your attention, while you choose whom to follow and which stories to open. Those selections can reinforce each other. Soon a familiar interpretation appears in many places, giving the impression of independent confirmation even when the posts trace back to the same report or community. Repetition changes visibility without necessarily adding evidence.

This overlaps with the frequency illusion, where something newly noticed seems suddenly everywhere. But the mechanisms are not identical. Confirmation bias concerns support for a belief; frequency illusion concerns perceived prevalence. Before treating a feed as a survey of reality, ask who is missing, whether the sources are independent, and what information would look different under a competing explanation.

Shopping research can become a defence brief

Suppose you want a particular laptop. You search for its best features, read enthusiastic reviews and explain away reports of poor battery life because those owners probably use it differently. The problem is not that the laptop must be bad. It is that your research has become a case for buying it rather than a comparison of whether it meets your needs.

Write the decision criteria first: battery life under your workload, repair options, keyboard comfort and total cost. Then examine drawbacks and alternatives using the same standards. A criticism that is irrelevant to your needs can reasonably carry little weight. But dismiss it because you already want the product, and the conclusion has started choosing the evidence instead of the other way around.

Arguments reward winning more than checking

During a disagreement, we often listen for the mistake that will let us reply. We collect weak examples of the opposing view while keeping its strongest evidence out of focus. Then we feel confident because we have defeated a version that few thoughtful opponents would actually defend. The conversation may produce sharper rhetoric without producing a better account of the issue.

Steelmanning means stating the strongest reasonable version of another person's argument, including its evidence and assumptions. Ask whether they recognise your summary before criticising it. This does not oblige you to agree, and it is inappropriate to invent credibility for harmful falsehoods. Its purpose is to make disagreement answer a real argument rather than an easy target assembled for victory.

Consider the opposite, concretely

An instruction to "be open-minded" is too vague to guide a search. Instead, ask: if my explanation were wrong, what else could explain these observations? If the colleague does not dislike you, a heavy workload might explain short replies. If a policy did not cause an outcome, an existing trend might explain part of it. Identify predictions that could distinguish the accounts.

Then decide in advance what would shift your confidence. A single anecdote may be inadequate for a large causal claim, while a direct clarification may settle a misunderstanding about a message. You need not lower all beliefs to equal uncertainty. Confidence should track the quality and relevance of evidence, with a genuine route to revision when better information arrives.

Seek disconfirmation without chasing every claim

Choose credible sources with different perspectives, especially people who can explain their methods and acknowledge limitations. Read the original report where possible instead of relying on a headline describing it. Ask whether an objection addresses the actual claim. Endless exposure to incompatible opinions is not automatically useful: unsupported assertions can add noise without offering a serious test.

The Dunning-Kruger effect concerns calibration of competence, not a licence to dismiss anyone who disagrees as ignorant. Similarly, negativity bias concerns the weight of negative information, not proof that criticism is mistaken. Naming a bias is helpful when it improves your own checks. It becomes another confirmation tool when every label is reserved for opponents.

Curiosity needs permission to correct you

MindSnap is our app. Its 2-minute daily lessons and five flagship story collections plus unlimited topics can introduce unfamiliar ideas beyond a familiar feed. That wider input is an invitation to curiosity, not a guarantee of unbiased thinking. A lesson becomes more useful when you ask what surprised you, what remains uncertain and where a further source could deepen or challenge the account.

Try ending a research session with two sentences: the strongest evidence for your conclusion, and the strongest unresolved objection. If you cannot write the second, you may have found an unusually clear case, or you may not have looked seriously enough. The aim is not permanent indecision. It is the ability to hold a view while leaving a real door open for evidence to change it.