Confirmation bias is the tendency to search for, interpret, and recall information in ways that confirm one's existing beliefs while giving disproportionately less attention to evidence that contradicts them. The Wason selection task demonstrated that even in abstract logical puzzles, people preferentially test cases that would confirm a hypothesis rather than cases that could falsify it. This bias operates at every stage of reasoning: in what questions we ask, which sources we consult, how we interpret ambiguous data, and what we remember afterward. It is arguably the single most pervasive obstacle to objective reasoning.
Try the Wason selection task yourself before learning the answer. Then practice 'steel-manning' — deliberately constructing the strongest version of a position you disagree with. Keep a reasoning journal where you note cases of seeking out only confirming evidence.
You already know that cognitive biases are systematic patterns where mental shortcuts — heuristics — produce predictable errors. Confirmation bias is the most pervasive of these failures: the tendency to gather, interpret, and remember information in ways that support what we already believe. It is not a single error but a cluster of related tendencies that distort reasoning at every stage, from initial hypothesis to final conclusion.
The clearest demonstration is the Wason selection task. You're shown four cards labeled A, D, 4, and 7. The rule is: "If a card has a vowel on one side, it has an even number on the other." Which cards must you flip to test the rule? Most people choose A and 4 — picking the card that would confirm the rule (A has a vowel; check if even) and the card that might confirm it (4 might have a vowel). But the correct answer is A and 7. You need A (a vowel — verify the even-number side) and 7 (an odd number — verify there's no vowel, since a vowel on the back of 7 would falsify the rule). Card 4 is irrelevant: whether it has a vowel or consonant cannot falsify the rule. The pattern is diagnostic: people seek verification when they should seek falsification.
Confirmation bias operates at three distinguishable stages. Biased search affects what information you seek out — people ask questions whose "yes" answer supports their hypothesis, read media that confirms their views, and notice evidence that fits. Biased interpretation affects how you process ambiguous information — the same study result reads as "suggestive" when it supports your position and "inconclusive" when it opposes it. Biased memory affects what you retain — confirming evidence is remembered more vividly and accurately. Together, these create a self-reinforcing cycle: beliefs become increasingly resistant to revision not because evidence against them is absent, but because the believer systematically fails to find, weigh, and recall it.
The unsettling finding is that confirmation bias survives expertise. Scientists, doctors, and judges exhibit it, often without awareness. The most effective countermeasures are not introspection but structural practices: pre-registration of hypotheses before data collection (removing the freedom to redefine what you were testing), actively generating and testing alternative explanations before settling on a conclusion, and assigning an explicit devil's advocate role. The goal is to institutionalize falsification — to make the systematic search for disconfirming evidence a habit rather than a reluctant concession that only happens when someone forces it.
Topics in reflective domains aren't scored by quiz answers. Read, reflect, and mark when you've thought it through.