FutureIQ KnowledgeBase

Confirmation Bias

mechanism

Confirmation bias lets prior belief judge the evidence meant to test it: attention preserves supporting hits, interpretation excuses conflicting misses, and memory turns the resulting imbalance into apparent proof.

An astrologer can offer thirteen descriptions, get only four right, and still seem uncannily accurate. The four matches produce the jolt of recognition; the other eight quietly disappear from your verdict.

E1

The belief grades its own test

A prior belief first directs attention toward compatible evidence. It then supplies asymmetric explanations: a favorable result counts at face value, while an unfavorable one gets attributed to a bad day, bad method, or special circumstance. Memory completes the loop by retaining the satisfying hits more readily than the awkward misses.

The evidence therefore never receives a neutral hearing. The same aptitude score can mean “bad day” when it threatens your view of an IIT candidate and “cheating” when it conflicts with your view of a local-college candidate. Repeated over time, this biased accounting makes a weak belief feel repeatedly verified.

E1 E2

Where it shows up

Politics filters the dataset

On climate change, an existing political position determines which findings feel trustworthy and which can be ignored. The apparent weight of evidence changes because the admissible evidence has changed.

E3

Agreement becomes methodological rigor

In the death-penalty study, people praised methods that produced their preferred conclusion and attacked methods that did not. The conclusion was supposedly downstream of the method, yet it reached backward and graded the method.

E4

Expectation reaches into taste

Even tasting is not insulated: the brain uses surrounding cues to form an expectation, then interprets the sensory input as confirmation. What feels like direct perception may already be an evidence-check shaped by a prior.

E6

Online certainty without learning

When an information environment keeps returning congenial material, looking at more content need not mean encountering more evidence. It can strengthen confidence while preventing correction—an overlap with the Illusion of Information Adequacy.

E5

A prior is not automatically a prejudice

Existing beliefs are necessary for interpreting noisy evidence, and agreement with a prior does not make evidence false. The mechanism becomes distorting when standards change with the conclusion—when hits need no explanation, misses always do, or methodological objections appear only after an unwelcome result.

E2 E4

Score the misses before reading the hits

Tomorrow, test one belief with a two-column ledger: write in advance what would count for it and what would count against it, then record every relevant result under those fixed rules. If you find yourself adding a rescue explanation only to the “against” column, require the same kind of explanation for the supporting evidence—or downgrade the belief.

Episodes that teach this

Related concepts

Expertise Selection BiasPost-hoc RationalizationCognitive biases arise from System 1 overconfidenceBayesian PriorsObserver Bias and Blinded StudiesUnconscious BiasBayesian InferenceSelection BiasIdentity-Protective CognitionSurvivorship BiasSource Credibility as EvidenceCherry Picking