Bayesian Prior Neglect

Background information that should genuinely shape a judgment can get ignored entirely. Attention lands on the new evidence, and the prior facts that should have anchored the decision get left out.

This bias underweights or ignores the prior probability when updating a belief in light of new evidence. The resulting estimate relies disproportionately on the new likelihood information, reducing how coherent the update actually is.

A doctor learns a patient tested positive for a rare disease with a 99% accurate test. Ignoring that only 1 in 10,000 people have the disease, the doctor concludes the patient almost certainly has it — when in reality, most positives are false alarms given how tiny the base rate actually is.

An intelligence analyst receives a signals intercept indicating a 90% likelihood of an imminent attack at a specific location. Without explicitly folding in the historical base rate for credible warnings at that kind of location — say, 2% — the analyst's estimate implicitly collapses to near 90%, far exceeding the properly calculated figure of around 15%. That gap leads to misallocated readiness resources and an operational overreaction, traceable directly to letting the new signal dominate over the prior under time pressure.

New reports draw far more attention than past rates do, so the prior simply gets ignored. That shift in attention is exactly what tilts the judgment toward the recent evidence alone.

The prior gets downweighted relative to the new likelihood signal at the point of belief update. Limited capacity to integrate both, combined with a bias toward whatever's most salient, enforces that imbalance.

Reminding people of the base rate before showing them new evidence is the direct fix. Stepping back to combine the old facts with the new information keeps the judgment properly calibrated.

Presenting explicit base-rate information alongside a structured update rule recalibrates the prior's weight directly. Decision aids that force an explicit representation of the prior, compared against the new likelihood, keep the two properly combined.

Overreliance on anecdotal evidence; Underestimation of base rates; Erroneous posterior confidence

An adversarial actor can exploit Bayesian prior neglect by flooding a target audience with vivid, salient new evidence while suppressing or obscuring base-rate information, causing the audience to form posterior beliefs that are unanchored from realistic priors. In legal, medical, or intelligence contexts, a manipulator can strategically present a compelling individual case narrative that crowds out unfavorable statistical baselines, distorting risk or guilt assessments. This technique is especially potent in high-stakes rapid-decision environments where decision-makers lack time or tools to explicitly retrieve and weight prior probabilities.

Decision-makers should adopt structured belief-updating protocols that require explicit documentation of prior probabilities before new evidence is introduced, preventing salience-driven displacement of the prior node. Use of calibrated decision aids—such as Bayesian calculators or natural frequency formats—forces prior representation and comparative likelihood assessment into the deliberation pipeline. Institutional training in base-rate retrieval and periodic audits of posterior estimates against known priors can systematically reduce prior-weight reduction in recurring judgment tasks.