Ask basketball fans to estimate the probability that each of eight NBA teams will win the playoffs. They estimate each team individually, focusing on how that team could win. The eight estimates sum to 240%. The confirmatory imagination generated a plausible victory scenario for each team — and generated no scenarios in which they lose.

The Framework

Confirmation bias is System 1's default search mode: when given a hypothesis, it searches for evidence that confirms the hypothesis and ignores or underweights evidence that contradicts it. Kahneman presents it in Chapter 7 as the "strongest" version of WYSIATI — not only does System 1 build stories from available evidence, it actively seeks evidence that fits the story it's already building.

The basketball fan experiment (Chapter 30) demonstrates the mechanism in action: when each team is made the focal event, confirmatory imagination generates scenarios of victory. The contradicting evidence (the other seven teams competing for the same championship) fades from consideration. The result: each team's probability is overestimated, and the eight estimates are wildly inconsistent (summing to 240% instead of 100%). The bias disappears when the question is restructured to force comparison: "Will the winner be from the Eastern or Western conference?" — with a specified alternative, estimates sum to 100%.

Where It Comes From

Confirmation bias has been studied since Peter Wason's 1960s card selection task. Kahneman treats it in Chapter 7 as a fundamental operating characteristic of System 1 — not a flaw to be corrected but an architectural feature that enables rapid story construction. Without confirmation bias, System 1 couldn't build coherent stories from fragments of evidence in milliseconds. The cost is that the stories are systematically biased toward whatever hypothesis happened to come to mind first.

> "System 1 does not keep track of alternatives that it rejects, or of the fact that there were alternatives. Conscious doubt is not in the repertoire of System 1." — Thinking, Fast and Slow, Ch 7

Cross-Library Connections

Voss's "That's right" technique in Never Split the Difference exploits the confirmation bias constructively: by reflecting the counterpart's worldview so accurately that they say "That's right," Voss confirms their story rather than attacking it. The confirmation feels genuinely persuasive because System 1 rewards confirmation with trust.

Fisher's brainstorming protocol in Getting to Yes (separate inventing from deciding) is a confirmation-bias corrective: during the inventing phase, no ideas are evaluated, preventing the confirmation bias from killing options that don't fit the initial narrative. Evaluation comes later, in a separate phase.

Cialdini's commitment principle in Influence leverages confirmation bias: once someone commits to a position (even a small one), their System 1 actively seeks confirming evidence for that position and dismisses contradicting evidence — deepening the commitment automatically.

The Implementation Playbook

Decision-Making: Before committing to any significant decision, deliberately search for disconfirming evidence. Ask: "What would convince me I'm wrong?" If you can't find any disconfirming evidence, you probably aren't looking hard enough — confirmation bias is ensuring that only confirming evidence reaches your attention. Assign a devil's advocate or conduct a premortem.

Research and Analysis: When analyzing data, formulate the opposite hypothesis first and search for evidence supporting it. If your hypothesis is "this product will succeed," search for evidence that it will fail before searching for evidence of success. The order matters: whichever hypothesis you test first gets the benefit of confirmatory imagination.

Feedback and Learning: Seek feedback from people who disagree with you rather than people who share your views. Confirmation bias means that your social circle reinforces your existing beliefs. The most valuable information comes from outside the echo chamber — but it's the last place System 1 wants to look.

Journalism and Media Literacy: Every news story is a confirmation exercise — the journalist selected facts that confirmed the narrative and omitted facts that contradicted it. Read coverage from sources with different editorial perspectives to break the single-narrative confirmation trap.

Key Takeaway

Confirmation bias isn't a flaw — it's the search algorithm that makes rapid cognition possible. System 1 builds coherent stories by finding confirming evidence quickly, and the stories are often approximately correct. The cost is paid when the stakes are high and the truth doesn't fit the first story that came to mind. The only defense: deliberately, effortfully, counterintuitively search for the evidence that would prove you wrong. It will never feel natural. It will always be valuable.

Continue Exploring

[[WYSIATI]] — The master principle: confirmation bias ensures "all there is" confirms the existing story

[[Halo Effect]] — Confirmation bias applied to people: the first impression selects the confirming evidence

[[Premortem (Klein)]] — The structured exercise that forces the search for disconfirming scenarios

📚 From Thinking, Fast and Slow by Daniel Kahneman — Get the book