Confirmation bias can influence the evidence you notice, how you evaluate conflicting information and how readily you reconsider a belief. Explore classic experiments, selective exposure, motivated reasoning and practical ways to make more reliable decisions.
In This Article
What Is Confirmation Bias?
Imagine buying a new phone after weeks of comparing options. Once you have made the purchase, you begin noticing reviews that praise its camera, battery or design. A review mentioning a serious weakness seems less persuasive, even though you would have considered that same criticism important before spending your money. The product has not changed, but your relationship to the information may have.
Confirmation bias describes a tendency to look for, interpret or evaluate information in ways that favor beliefs and expectations you already hold. In his influential 1998 review, psychologist Raymond Nickerson examined several forms of this tendency across reasoning and judgment.
The important distinction is between finding evidence that genuinely supports an explanation and treating support as sufficient merely because you prefer that explanation. A belief can be accurate even when someone evaluates it poorly. Equally, a confident person can reach an incorrect conclusion despite encountering several apparently supportive facts.
Confirmation bias is not a diagnosis, and the title’s reference to the brain should not be interpreted as a claim that a single brain structure produces it. The research discussed here concerns psychological processes and observed behavior. Our related guide to cognitive biases that affect mental health explores how other tendencies in attention, interpretation and memory can influence everyday functioning. This article concentrates on how we test beliefs and decide which information deserves our confidence.
The Classic Experiment That Revealed A Problem With Testing Beliefs
In a 1960 experiment, Peter Wason gave participants a sequence of three numbers and asked them to discover an underlying rule. They could propose additional sequences and receive feedback about whether those sequences satisfied the rule. Only six of the 29 participants identified the intended answer without first offering an incorrect one.
You can explore the reasoning involved using a similar puzzle. Suppose the numbers 2, 4 and 6 satisfy a hidden rule. You might guess that the rule requires three consecutive even numbers. Testing 8, 10 and 12 would produce a positive result if your guess were correct, but it could also produce a positive result under many alternative rules.
A more informative test could involve numbers that do not satisfy your initial guess. If 1, 3 and 7 are accepted, the rule cannot require consecutive even numbers. If 9, 7 and 5 are rejected, that provides another useful distinction. In Wason’s task, the actual rule concerned increasing numbers rather than a particular interval or parity.
The experiment illustrates how a series of correct observations can support a mistaken conclusion when competing explanations predict the same result. It does not establish that everyone reasons this way in all situations. The artificial task allowed investigators to examine a particular problem with hypothesis testing under controlled conditions.
In daily life, the corresponding question is not merely whether an observation supports your current explanation. It is whether the observation distinguishes that explanation from plausible alternatives.
When Looking For Positive Evidence Is Actually Useful
It would be a mistake to conclude from Wason’s experiment that people should always seek evidence against what they believe. Sometimes a test that examines where an outcome is expected to occur provides exactly the information needed.
In their 1987 analysis, Joshua Klayman and Young-Won Ha distinguished a positive test strategy from a universally harmful preference for confirmation. They argued that examining cases expected to show a property can be an effective approach in some settings and misleading in others.
Imagine that a printer stops working whenever a particular application is open. You suspect that the application contributes to the problem. Reproducing the problem under the suspected conditions is a sensible first test. You should also investigate whether the printer operates normally with the application closed and whether the issue occurs with other applications. Relevant comparisons help distinguish among explanations.
The purpose of testing is not to contradict yourself for its own sake. It is to obtain information that would change your confidence in competing accounts. A carefully chosen positive test may be informative, whereas searching through hundreds of irrelevant counterexamples may waste time.
Think about the consequences of being wrong and the resources available to investigate. A minor preference does not require the same examination as a decision involving health, financial commitments or the safety of other people. Match the effort devoted to checking a belief to the importance of the decision.
Our article on anchoring bias discusses another way initial information can influence judgment. The two concepts overlap in some circumstances but describe distinguishable phenomena.
Selective Exposure: How The Information You Encounter Changes
Before you evaluate a claim, you have to encounter it. Information selection can therefore influence what appears to be a convincing body of evidence. A search query framed around a conclusion may produce a substantially different collection of results from one framed around an open question.
In a 2009 meta-analysis, William Hart and colleagues investigated choices between information supporting and challenging existing attitudes, beliefs or behavior. They identified a moderate average preference for supportive information, with an effect size of 0.36. The extent of this preference varied with circumstances, including the relevance of the belief to a person’s values and the usefulness of the information.
For example, someone who suspects that working remotely always reduces productivity might search exclusively for stories about employees working less effectively at home. Another person committed to remote work might collect only reports describing its advantages. Neither search necessarily reveals how the arrangements compare across different jobs or how outcomes depend on working conditions.
A more informative search would specify the population, outcome and comparison of interest. It might examine productivity among workers doing similar tasks under different arrangements, while considering whether the available studies used comparable methods.
You do not need to devote equal attention to every claim. The quality of evidence, independence of sources and relevance to your question are important. A well-supported account does not deserve less confidence simply because another account exists.
Review the way you frame a search when the answer matters. Ask whether someone using the opposite assumption would encounter substantially different information and whether a more neutral question would help you evaluate both accounts.
Why Beliefs That Matter Personally Can Be Difficult To Reconsider
Some conclusions have consequences beyond the question of whether they are accurate. Beliefs may influence a person’s relationships, professional reputation or understanding of earlier decisions. Learning that a familiar explanation is mistaken can raise uncomfortable practical questions.
In her 1990 review of motivated reasoning, Ziva Kunda explored how the desire for accuracy and the desire to reach particular conclusions can influence the reasoning strategies people use. The framework describes ways a preferred conclusion may affect what explanations or information receive attention, subject to the individual’s ability to find a justification that seems plausible.
Imagine a manager who selected a new supplier despite concerns raised by a colleague. When delivery failures occur, the manager may quickly identify reasons why those failures appear temporary. They might evaluate similar problems involving a different supplier more critically.
The example does not reveal the manager’s motives. The delivery problems might genuinely be temporary, and the manager might have relevant information unavailable to the colleague. What deserves investigation is whether the same standards are being applied to comparable evidence.
A practical safeguard is to agree on performance criteria in advance. If the team records what reliability, quality and cost it expects before a supplier is selected, later decisions can refer to those standards. The manager can revise the arrangement without needing to defend every element of the original choice.
This approach also reduces the tendency to confuse reconsidering a decision with making a judgment about the person who made it. New information can justify a change even when an earlier decision was reasonable at the time.
Our related article on cognitive distortions describes unhelpful thinking patterns that can arise in ordinary and clinical contexts. These are related concepts, not interchangeable labels.
How Confirmation Bias Can Influence Relationships
In close relationships, behavior is often interpreted against a history of shared experiences. That history matters, but an established expectation can also shape the meaning assigned to new events.
Suppose you believe that a friend does not value your time. When they arrive late, the event appears consistent with your expectation. When they arrive early or make a considerable effort to help, you may treat those occasions as exceptions. Over time, your account of the friendship may emphasize a selected portion of the actual experiences.
The reverse can also happen. Someone convinced that a person is trustworthy may repeatedly disregard evidence of unreliable or harmful behavior. A bias favoring a reassuring belief can be as consequential as a bias favoring an unfavorable one.
A useful review involves separating observations from inferences. How often have arrangements been canceled? Were explanations offered? What agreements were made about timing? Does the other person respond constructively when a concern is raised? The answers may support the original concern or identify a need to revise it.
Do not approach credible evidence of coercion, abuse or other unsafe conduct as merely a thinking error. Take practical safety concerns seriously and seek appropriate assistance. Reviewing ordinary interpretations is not a requirement to remain in a harmful relationship.
If you repeatedly find yourself revisiting ambiguous interactions without gaining useful information, our article on rumination and worry explores related patterns. Persistent distress may benefit from individualized psychological assessment rather than increasingly extensive self-analysis.
Consider The Opposite Without Pretending All Explanations Are Equal
A useful corrective technique is to identify a credible alternative explanation and ask what evidence would support it. This differs from trying to force yourself to believe the opposite of what you currently think.
In two experiments published in 1984, Charles Lord, Mark Lepper and Elizabeth Preston investigated instructions that encouraged participants to consider opposing possibilities in exercises involving social judgment. The strategy performed better than general requests to avoid bias in the particular experimental tasks.
For a practical example, imagine that your team has concluded that customers are abandoning an order because the checkout form is too long. An alternative explanation is that unexpected delivery fees appear late in the process. Reviewing at which stage orders are abandoned could help distinguish the possibilities. The result might suggest that both explanations matter or that neither adequately accounts for the pattern.
The technique works best when the alternative is plausible and relevant. Inventing an implausible explanation does not make a comparison stronger. Neither does repeatedly challenging a well-supported fact until you become uncertain about everything.
If you are evaluating a claim involving technical or medical expertise, consult qualified sources rather than relying exclusively on your own ability to generate explanations. A credible objection has to be assessed in relation to the quality of evidence available.
The objective is to identify information your current explanation might overlook, while keeping an appropriate standard for what would justify changing your mind.
A Practical Review Before An Important Decision
Choose an upcoming decision for which being wrong would have meaningful consequences. Begin by stating the question without building a preferred answer into it. For example, ask which of two service providers best meets your requirements rather than searching for reasons why the provider you already favor is the best.
Next, record the criteria you intend to apply to both options. These might include relevant experience, cost, reliability, contractual terms and the ease of resolving problems. Distinguish between conditions that are essential and those that are merely desirable.
Identify what information could reasonably change your decision. If evidence of repeated missed deadlines would make one provider unsuitable, apply that standard consistently. Where an independent review is practical, ask another person to examine the same criteria without first telling them which option you prefer.
Record the decision and the most important reasons for it. If a problem arises later, revisit the original criteria rather than inventing a new standard specifically to protect the previous choice. Some information will be unavailable until after a decision is made, so a record helps distinguish an unreasonable choice from an unpredictable outcome.
This process is intended for decisions that justify the effort. Applying detailed written reviews to every routine choice may become burdensome. Select the safeguards appropriate to the decision, and observe established professional procedures whenever they are required.
These steps are an illustrative application of research on judgment, not a validated stand-alone assessment tool or a guarantee of unbiased reasoning.
Can You Train Yourself To Reduce Confirmation Bias?
Research suggests that reasoning strategies can improve with appropriate instruction and practice. In a 2015 study by Carey Morewedge and colleagues, participants received interventions involving an instructional video or an educational game. The experiments assessed several judgment biases, including confirmation bias, and found improvements on research measures that persisted at a two-month follow-up.
This work is encouraging, but it does not demonstrate that every participant becomes unbiased or that any online training product will achieve equivalent results. Performance on an assessment is also not the same as accuracy in every consequential real-world decision.
An individual can practice by examining a limited set of decisions where feedback is available. For instance, a learner might make a prediction about how a particular study method will affect their performance, record the conditions under which they use it and examine the subsequent result. Clear outcomes make it easier to notice when a prediction is not supported.
Institutions can use additional safeguards. Independent assessments, consistent decision criteria, reproducible tests and procedures for raising concerns may reduce the extent to which important judgments depend on a single person’s expectations.
Confirmation bias is best understood as a reason to improve how beliefs are evaluated, not as a label for dismissing everyone who disagrees with you. You may have good reasons for your current conclusion, and another person may have access to evidence you have not yet considered. The goal is to make it possible for relevant information to influence decisions, whether that information supports the original belief or requires it to change.
Frequently Asked Questions
What Is Confirmation Bias?
It is a tendency to favor information that supports an existing expectation, belief or hypothesis when searching for or evaluating evidence.
What Is An Everyday Example Of Confirmation Bias?
A shopper may favor favorable reviews of a product after purchasing it while applying stricter scrutiny to unfavorable reviews.
Is Confirmation Bias Always Harmful?
Not every search for positive evidence is biased. A positive test can be informative when it helps distinguish plausible explanations.
How Is Confirmation Bias Different From Motivated Reasoning?
Confirmation bias concerns the preferential treatment of supporting evidence. Motivated reasoning concerns the broader influence of goals and desired conclusions on reasoning.
How Can I Reduce Confirmation Bias?
Use consistent criteria, examine plausible alternatives, seek information that could change your judgment and obtain independent review for important decisions.
Can Confirmation Bias Be Eliminated Completely?
Research suggests that training and decision procedures can reduce certain biases, but no technique guarantees complete impartiality.
Research And Further Reading
Nickerson (1998), Confirmation Bias: A Ubiquitous Phenomenon In Many Guises.
Wason (1960), On The Failure To Eliminate Hypotheses In A Conceptual Task.
Klayman And Ha (1987), Confirmation, Disconfirmation, And Information In Hypothesis Testing.
