Cognitive Biases · CB-28
People routinely judge a specific, detailed, vivid combination of events as more probable than either event alone — even though a combination can never be more probable than its parts.
The error of judging the probability of two events occurring together (a conjunction) as higher than the probability of either event occurring alone — a logical impossibility, since the conjunction of two events can never be more likely than either individual event, but a mistake people make reliably when the combined description feels more coherent or representative.
Demonstrated in Amos Tversky and Daniel Kahneman's famous 1983 'Linda problem': given a description of a socially progressive philosophy graduate, most participants rated 'Linda is a bank teller and active in the feminist movement' as more probable than 'Linda is a bank teller' alone — a straightforward logical impossibility.
The Mechanism
A specific, coherent story beats its own logically necessary superset
The conjunction (bank teller AND feminist) cannot logically be more probable than either part alone — every feminist bank teller is also simply a bank teller, so the set of 'bank tellers' is always at least as large. Tversky and Kahneman's participants nonetheless rated the more specific, narrative-coherent conjunction as more likely, an error that persisted even among statistically trained participants.
01 · IT'S DRIVEN BY REPRESENTATIVENESS, NOT LOGIC
A vivid story that 'fits' the character description feels more probable than the math allows
The specific conjunction matches the personality profile given (Linda sounds like the kind of person who'd be a feminist) far better than the bland, generic category alone — this representativeness, or narrative fit, overwhelms the straightforward logical constraint that a conjunction cannot exceed its components.
02 · IT PERSISTS EVEN AMONG PEOPLE WITH FORMAL STATISTICAL TRAINING
Knowing the rule doesn't reliably prevent violating it in the moment
Tversky and Kahneman found the effect held up even among graduate students with statistical training, when the problem was presented in a narrative, character-based form — formal knowledge of probability rules doesn't automatically override the intuitive pull of a coherent, representative story.
03 · IT EXPLAINS WHY DETAILED, SPECIFIC SCENARIOS OFTEN FEEL MORE 'REALISTIC' THAN THEY ARE
Adding vivid, plausible-sounding detail to a forecast makes it feel more likely, not less
A detailed, specific forecast ('a recession caused by a housing correction combined with a banking crisis') can feel more probable than a vaguer, more general forecast ('a recession'), even though every added specific detail necessarily narrows the range of scenarios it covers and therefore can only lower or maintain, never raise, its true probability.
Where It Fails / Inversion
Where it fails / inversion
The fallacy specifically concerns probability judgments, not other legitimate reasons a specific scenario might be worth attention (it may be more actionable, more informative for planning, or more consequential if true) — a detailed scenario can be more useful to plan around even while being, correctly, judged as less probable than its more general superset.
How To Use It
Worked example · evaluating a detailed, vivid business risk scenario
A specific, narratively compelling risk scenario ('a key supplier fails due to a currency crisis triggered by a specific geopolitical event') often feels more concerning and more probable than the broader, blander category it belongs to ('a key supplier fails for any reason') — but the broader category is, by strict logic, always at least as likely, and risk planning should account for the full broader category, not just the most vivid specific instance of it.
How to use it
When a detailed, specific scenario feels highly probable because it 'hangs together' as a coherent story, explicitly check whether you're judging the conjunction of several conditions rather than the broader category it falls within — the broader category is always at least as likely, and is usually the more useful thing to actually plan around.
See Also