Support Theory

Definition

Rather than attaching probabilities to objective events, people assign probabilities to mental representations (called hypotheses or descriptions of events). The probability of a description depends on the support provided by evidence, which differs between explicit and implicit descriptions of the same event. An implicit description is collapsed ("a car crash"), while an explicit description unpacks possibilities ("a crash due to road construction, driver fatigue, or brake failure"). Unpacking increases support by bringing neglected possibilities to mind or amplifying the weight of specific components.

In the Book

Tversky and Koehler show that when an event is described implicitly, it receives lower probability judgment than when the same event is explicitly decomposed. For example, the probability of "a car accident" increases when it is unpacked into specific causes. A survey asking respondents to judge "a deadly accident in the next year" yielded different responses than "a deadly accident in the next year by drunk driving, reckless driving, mechanical failure, or weather." The unpacked version receives higher aggregate probability judgment than the implicit version, even though both logically describe the same event.

This effect also explains the "partition dependence" observed in frequency judgments and probability estimates. When college students list "ways a person could die," they provide different probability estimates than when explicitly asked to estimate causes (natural, accident, homicide, suicide). The revision of a hypothesis through unpacking systematically increases both the support of the explicit components and the overall probability. Support theory further predicts that complementary events (a hypothesis and its negation) will sum to unity, which is typically observed in data.

Why It Matters

Support theory formalizes how representation shapes probability judgment, showing that two logically identical events can receive different probability assessments based on how they are described or decomposed. This has practical importance: in risk communication, how a disease or accident is broken down into causes affects perceived likelihood. It also explains why people's probability judgments often violate standard probability axioms—not because they can't do logic, but because they're implicitly following a different system based on evidence strength rather than event structure. The theory bridges extensionality (matching logical structure) with psychological representation (how people mentally encode possibilities).