Support theory
Probability judgments attach to descriptions of events, not the events themselves.
What it means
A descriptive theory of subjective probability holding that people assign likelihood to explicit descriptions of events rather than to the events as such, so two logically identical events can receive different probabilities depending on how they are described. Each description recruits a degree of 'support' — the strength of evidence and scenarios it brings to mind — and a more detailed, unpacked description typically musters more support than a terse one. This single assumption predicts a family of findings at once: subadditivity (parts sum to more than the whole), the unpacking effect, and systematic violations of the standard probability calculus. The theory reframes many 'biases' as the lawful output of how the mind evaluates hypotheses, and it matters for forecasting and risk communication because the framing and granularity of a question, not just its content, drive the numbers people give.
Examples
Clinicians estimate a higher chance of a patient dying when the prognosis is broken into specific causes than when asked about death from 'any cause.'
Travellers will pay more for cover against death 'from terrorism or mechanical failure' than for cover against death from any cause on the same flight — the unpacked list summons vivid scenarios.
Ask a team the odds the launch slips and you hear 'maybe twenty percent.' Ask separately about hiring, vendors and bugs, and the three parts add up well past half.
First described in Tversky & Koehler (1994); Rottenstreich & Tversky (1997).