Ignoring the prior probability of an event in favour of vivid, specific, case-level detail. The analyst reasons about the individual case as if it existed outside any population — "this person is behaving suspiciously, therefore this is an attack" — without asking how often that behaviour occurs in the general population.
In intelligence analysis
- A triggering event is assessed from its own detail only; the denominator (how many such events occur for benign reasons) is never estimated, so a weak indicator is treated as a strong one.
- It compounds Vividness Bias: the memorable case is both over-weighted and stripped of its base rate.
- Rare capabilities (catastrophic terrorism, novel malware capable of destroying physical plant, state-level attribution) are systematically over-estimated because their base rate is nearly invisible, while the consequences dominate attention.
- It drives alert fatigue and false-positive volume in detection work: indicators are selected for sensitivity without a base-rate check on how much benign traffic they match.
Countermeasure
- For every indicator, state the benign explanation rate alongside the malicious one, even approximately.
- Use explicit likelihood language (Estimative Language) that separates confidence in the indicator from confidence in the inference drawn from it.
Related
- Cognitive Bias — the taxonomy this bias sits within
- Vividness Bias — the concrete case crowds out the prior
- Representativeness Heuristic — the mechanism that produces base-rate neglect
- Insensitivity to Sample Size — the same failure applied to the evidence itself