Confirmation Bias
People notice, seek out, and remember evidence that supports what they already believe more readily than evidence that contradicts it, which makes existing beliefs unusually good at defending themselves.
What Is It?
Francis Bacon described the underlying tendency in 1620, noting that people notice confirming instances more readily than disconfirming ones. Modern experimental grounding for the specific mechanism of confirmatory hypothesis testing is strongly associated with Peter Wason’s studies in the 1960s, and the broader concept has since been developed across a wider body of judgment and reasoning research: given a hypothesis, people disproportionately look for information that would confirm it rather than information that would test or disconfirm it, and can interpret ambiguous evidence in ways that preserve the existing belief. The bias operates at every stage, what evidence people seek out, how they interpret ambiguous evidence once they have it, and what they remember afterward. None of this requires bad faith. People aren’t usually aware they’re doing it, the belief they already hold quietly shapes what counts as a fair test of itself, and the whole process can feel like careful, evenhanded reasoning from the inside.
Why Does It Matter?
A leader who’s already decided a strategy is working reads ambiguous quarterly numbers as confirmation and a leader who’s already decided it’s failing reads the identical numbers as confirmation of the opposite, using the same data. A team that likes a product idea interprets neutral user feedback as validating, and a team that’s skeptical of it interprets the same feedback as damning. This isn’t a story about dishonesty, it’s what happens by default when evidence is genuinely ambiguous and a prior belief is available to resolve the ambiguity in whatever direction the belief already points.
The organizational risk compounds when a decision has already been publicly committed to. Once a leader has stated a position, confirmation bias can reinforce Escalation of Commitment: one mechanism filters the evidence toward what already supports the decision, while the other raises the social and psychological cost of reversing course, and together they’re part of why bad decisions can survive so much contrary evidence before anyone acts on it.
The effect can also operate at organizational scale, beyond any one person’s reasoning. A leader who believes customers want a particular feature doesn’t just interpret ambiguous customer feedback that way, they can authorize research that asks about that feature, allocate analysts to study it, and fund experiments around it. The belief doesn’t only shape how existing evidence gets read, it can shape what evidence gets produced in the first place, which means the organization’s information environment itself can end up organized around confirming what leadership already believes.
What Changes Once You See It?
You start actively asking what evidence would change your mind, before you go looking for evidence, since defining disconfirmation in advance makes it much harder to quietly redefine what counts as a fair test after you already know what you’re hoping to find.
You start treating a search for evidence that turned up only supporting examples as a sign to check whether the search itself was capable of surfacing disconfirmation, rather than a sign you were right.
You also get more deliberate about asking someone to construct the strongest falsifying case, searching explicitly for failed predictions, or consulting people whose incentives and prior views make them likely to notice different evidence, rather than relying on a token “argue the other side” exercise that can end up as performative as it is useful.
Common Misunderstandings
- It isn’t a claim that people are lying or arguing in bad faith when they favor confirming evidence. The bias operates on genuinely honest reasoning, which is part of what makes it hard to self-correct.
- It isn’t the same as motivated reasoning in general, which covers a broader set of ways people’s desires shape their conclusions. Confirmation bias is specifically about the asymmetric search for and treatment of evidence relative to an existing belief.
- It doesn’t mean every instance of finding supporting evidence is biased. Sometimes a belief really is well-supported and the evidence genuinely points that way, the bias is in the asymmetric search process, not in every case of a hypothesis turning out to be correct.
- It isn’t fixed by simply deciding to “keep an open mind.” Better protection comes from structuring the inquiry itself so disconfirming evidence has a real chance to appear, actively seeking it out, assigning dissenting roles, rather than relying on willpower or good intentions alone.
Diagnostic Question
What would we expect to see if this belief were wrong, and have we actually gone looking for it?
Explore Further
Field Notes
- None yet.
Related Field Guide
Origin
The general pattern was described by Francis Bacon in Novum Organum (1620); the term “confirmation bias” and much of its modern experimental grounding trace to Peter Wason’s studies of hypothesis testing in the 1960s.