Confirmation Bias
The tendency to seek, interpret, and remember evidence in ways that favor existing beliefs, and how researchers test for it.
Confirmation bias is the tendency to search for, interpret, favor, and recall information in ways that confirm one's preexisting beliefs or hypotheses. It is not a single mechanism but a family of effects: biased search for evidence, biased interpretation of ambiguous evidence, and biased memory for what was encountered. It operates even when people intend to be objective, which is why it has been called the best-known and most frequently replicated bias in the psychology of reasoning.
The classic demonstration is Peter Wason's 1960 2-4-6 task. Participants were told that the triple 2-4-6 conforms to a rule and asked to discover the rule by proposing their own triples. Most participants proposed triples consistent with their current hypothesis — ascending even numbers — and rarely proposed triples that could refute it, even when the true rule was simply "any ascending numbers." They tested what they believed rather than what would discriminate. In a complementary 1979 study, Charles Lord, Lee Ross, and Mark Lepper showed participants mixed evidence about capital punishment; supporters and opponents both rated the evidence supporting their own position as more convincing and reported their beliefs becoming more extreme — a pattern called biased assimilation.
Researchers distinguish motivated reasoning, in which directional goals distort processing, from more neutral cognitive constraints such as limited attention and the costs of testing many hypotheses. Under some accounts, confirmation bias reflects a search for evidence that is sufficient for action rather than an attempt to be wrong as rarely as possible. This distinction matters because the two explanations imply different remedies.
Consequences are documented across domains: belief persistence after disconfirmation, political and scientific polarization, overconfidence in medical and legal judgment, and the formation of echo chambers where one-sided evidence dominates. In science, confirmation bias is a recognized hazard addressed by institutional practices — preregistration of hypotheses, blind analysis, adversarial collaboration, and the convention that a result must survive an attempt at refutation. Bayes' theorem provides the normative benchmark: beliefs should be updated by evidence in proportion to its diagnosticity, and the research on confirmation bias largely documents how far intuitive updating falls short of that standard. Debates continue about the magnitude and boundary conditions of the effect, and some early findings are under replication scrutiny — but the core phenomenon is not in dispute.
Tags
cognitive biases psychology reasoning