Give someone a rule-discovery puzzle: sequences of three numbers follow a rule. Their job is to figure out the rule by generating sequences and getting feedback on whether each fits. Most people get the answer wrong — not because the puzzle is hard, but because they never try to prove themselves wrong.
They propose "2, 4, 6" (fits). They try "4, 8, 12" (fits). They try "10, 20, 30" (fits). They declare the rule is "even numbers that double." The actual rule? Any three numbers in ascending order. They could have discovered this in two tests — but their brain only generated sequences that confirmed the hypothesis it already had.
This is confirmation bias in its clearest form.
What confirmation bias is
Confirmation bias is the systematic tendency to favor, search for, and interpret information in ways that support existing beliefs — while discounting, ignoring, or explaining away evidence that contradicts them. It operates across three distinct stages:
Search bias: You seek out information that's likely to agree with what you already think. If you believe a supplement works, you search for studies supporting it. You don't spontaneously seek the disconfirming evidence.
Interpretation bias: Ambiguous information gets interpreted in the direction of your prior belief. The same study results look compelling to a believer and flawed to a skeptic. The data is identical; the reading is not.
Memory bias: You're more likely to remember evidence that confirmed your view than evidence that challenged it. Over time, your mental record of "what the evidence says" becomes increasingly skewed toward confirmation.
The original study
Wason, P.C. (1960) — "On the failure to eliminate hypotheses in a conceptual task." Quarterly Journal of Experimental Psychology, 12(3), 129–140. This is the 2-4-6 task described above. Wason found that participants almost exclusively tested sequences that confirmed their hypotheses rather than trying to falsify them. Of his original subjects, very few discovered the correct rule. This paper launched decades of research into motivated reasoning and hypothesis testing.
Three real-world examples
Investment decisions. An investor who believes a stock will rise reads every piece of news through that lens. Positive analyst notes are absorbed at face value. Negative news is dismissed as noise or short-term. The position grows even as contrary evidence accumulates. Many significant investment losses trace back not to bad initial judgments but to confirmation bias in maintaining them.
Performance reviews. A manager who forms an early impression of an employee — strong performer, weak performer — subsequently interprets the employee's work in light of that belief. The same outcome is attributed to talent (for the favored employee) or luck (for the less favored). The employee's actual performance becomes nearly invisible behind the manager's prior expectation.
Online news consumption. Recommendation algorithms aren't entirely to blame for filter bubbles. People actively seek sources that confirm their worldview. Studies show that even when exposed to opposing viewpoints, people spend less time reading them and rate them as less credible. The selective exposure happens before the algorithm even intervenes.
Why it exists and why it's so hard to correct
Confirmation bias isn't a flaw in the traditional sense — it's a feature of how cognition manages limited resources. Evaluating every piece of information from first principles is exhausting. Using prior beliefs as a filter is efficient. The problem is that efficiency here comes at the cost of accuracy.
The deeper issue is motivational. Beliefs have social and psychological stakes. Being wrong feels threatening. Seeking disconfirming evidence means risking being wrong. The bias protects self-esteem and group identity at the cost of epistemics.
Research consistently shows that knowing about confirmation bias doesn't eliminate it. You can improve at recognizing it, especially in domains where you have no emotional investment. But where beliefs are tied to identity — politics, religion, deeply held worldviews — confirmation bias persists robustly even in people who understand the mechanism perfectly well.
The most practical corrective: actively consider what evidence would change your mind, then go look for it. Not "what would disprove this belief" as an abstract exercise, but as an actual search. The 2-4-6 task shows that the gap between knowing you should seek disconfirmation and actually doing it is wider than most people expect.