Where does rational choice collide?

Economists treat rational choice as an assumption inside a model: an agent picks the most preferred option from those a budget or constraint allows, and preferences are taken to be complete and transitive. Samuelson's 1938 revealed preference recasts this without any appeal to unobservable utility, showing that if observed choices obey a consistency condition, the agent can be represented as if maximizing a utility function. Becker's 1976 The Economic Approach to Human Behavior carries the same assumption beyond markets into crime, marriage, and discrimination, treating stable preferences and maximizing behavior as the method itself. The claim is usually as-if rather than literal: the theory does not say people compute utilities, only that their choices are consistent enough to be described that way. In this field rational choice is a modeling convention, judged by the predictions it yields rather than by any account of how people actually deliberate.[1][2]

Unit

A consistent choice from a constrained set of options

Recovery means

Not applicable

Measured by

Tested by the consistency of observed choices against the axioms of revealed preference, and by predictive fit

In sociology the same idea is used as a mechanism to explain how individual action adds up to social order. Coleman's 1990 Foundations of Social Theory begins from purposive actors who control resources and events that others value, and builds exchange, trust, norms, and authority out of their choices, with the hard problem being the transition from the micro level of action to the macro level of the system. Coleman's 1988 account of social capital treats relationships and obligations as a resource an actor can draw on, one produced largely as a by-product of activity pursued for other reasons. Here rational choice is not a claim that people are calculating, but a starting point that lets social structure be derived rather than assumed. The interest is in emergence and order, so the actor is deliberately kept simple in order to see what the structure of relations does.[3][4]

Unit

An actor's choice over resources within a structure of relations

Recovery means

Not applicable

Measured by

Assessed by how far micro-level action reproduces observed macro-level patterns, and by network and exchange data

Political scientists use rational choice as a positive theory of voting and collective decisions. Downs's 1957 An Economic Theory of Democracy models parties as vote-maximizers and voters as choosing the party closest to them in a policy space, which yields predictions about convergence toward the median voter and about rational ignorance. The framework also produces the paradox of voting: if a single vote almost never decides the outcome, the expected instrumental benefit of voting is tiny, so a purely instrumental citizen should abstain, yet turnout stays high. Riker and Ordeshook's 1968 calculus of voting answers this by adding a term for the private satisfactions of the act itself, such as a sense of duty, so that voting can be rational even when it is not decisive. In this field rational choice is a theory whose sharpest results are the puzzles it raises about collective action and turnout.[5][6]

Unit

A voter's or party's choice in an electoral or collective decision

Recovery means

Not applicable

Measured by

Evaluated against turnout, party positioning, and electoral outcomes

A fourth strand offers no new definition but asks whether the shared assumption survives testing. Green and Shapiro's 1994 Pathologies of Rational Choice Theory reviews the empirical record of rational choice work in political science and argues that many applications were method-driven, specified arbitrarily after the fact, and rarely put to fair tests against rival explanations. Their target is not the internal logic of the models but the claim that the models had earned their standing from evidence. This overlaps with the older contrast drawn by bounded rationality, which holds that real actors satisfice under limits of information and computation rather than maximize. Taken together these lines mark the edge of the construct: it can be defended as a useful as-if convention, but the further it is read as a literal account of how people decide, the more the evidence is disputed.[7]

Unit

A judgment about whether the maximizing assumption holds empirically

Recovery means

Not applicable

Measured by

Weighed by out-of-sample prediction and by fair tests against alternative theories