What it is
Confirmation bias is the tendency to search for, interpret, favor, and recall information in a way that confirms beliefs already held. People selectively gather evidence, selectively attend to what they find, and selectively remember in ways that are skewed toward their prior views. The bias operates at all stages of the information processing chain: what we look for, what we notice, how we interpret ambiguity, and what we retain.
This is the most well-documented bias in the entire psychology of judgment, first described by Peter Cathcart Wason in the 1960s and replicated across hundreds of studies in different domains and cultures.
The classic example
Wason's selection task: subjects are shown four cards — say, showing E, K, 4, 7. They are told the rule "if a card has a vowel on one side, it has an even number on the other." They must choose which cards to flip to test whether the rule is true.
The correct answer is to flip E (to check for an odd number on the back) and 7 (to check for a vowel on the back). These are the cards that could falsify the rule. Most subjects flip E and 4 — seeking confirmation rather than falsification.
Falsification is the correct logical approach. To test a hypothesis, you must look for evidence that would prove it wrong. But people instinctively seek confirming evidence and ignore or avoid disconfirming evidence.
In investing
A fund manager forms a thesis that a company is undervalued. From that point forward, confirming evidence — strong earnings, favorable industry trends, management credibility — is weighted heavily and interpreted positively. Disconfirming evidence — margin compression, customer churn, competitor moves — is dismissed as temporary or already reflected in the price.
This pattern is well-documented in analyst research. Analysts who initiate coverage with a "buy" rating interpret subsequent ambiguous news more favorably than those who initiated with a "sell." The initial stance creates an anchor that filters subsequent interpretation.
The phenomenon is particularly destructive in concentrated portfolios. The more capital committed to a position, the stronger the motive to interpret new information in its favor — and the more catastrophic the consequence when reality intrudes.
In research and analysis
The replication crisis in social psychology is partly a result of confirmation bias in research design. Researchers who expect to find an effect design studies that are more likely to find it, run analyses until they find a significant result, and interpret ambiguous data favorably. The published literature reflects researchers' prior beliefs as much as reality.
In due diligence for acquisitions, confirmation bias is especially expensive. Once a deal team has decided they want a company, their due diligence process tends to confirm rather than test the investment thesis. Red flags get rationalized away. The deal closes; the thesis turns out to be wrong.
The right way to think about it
The pre-mortem is the most practical antidote. Before committing to a decision, imagine that it has been made and has failed spectacularly — then work backward to explain why. This forces active generation of disconfirming scenarios rather than passive filtering of them.
Red team exercises serve the same purpose at the institutional level: explicitly designating someone to argue against the prevailing view, with the responsibility to find the strongest possible counterarguments.
In analysis, the discipline of seeking out the most credible bear case — and steelmanning it before evaluating the bull case — creates at least some resistance to confirmation bias. The bearish analysis you produce yourself, honestly, is worth more than ten bearish analyses from others you read selectively.
One thing most people get wrong
Confirmation bias is strongest precisely where stakes are highest and beliefs are most personal. In casual domains — music preferences, restaurant choices — people are relatively open to contradicting evidence. But for beliefs tied to identity (political, religious, professional), confirmation bias operates at maximum intensity.
This is why investment theses that become public or social — shared with colleagues, argued in investor letters, presented in thesis competitions — become most resistant to updating. The public commitment creates an identity stake that confirmation bias defends aggressively. The corrective is to distinguish between having a view and being committed to that view. Updating should be welcomed; it signals you are reasoning from evidence rather than defending a position.