In 1977, Amartya Sen called the agents of economic theory "rational fools." Their single preference ordering explains everything — and therefore explains nothing. His life's work tears it apart.
scroll to begin
Standard economics assumes each person has a single preference ordering — a complete, transitive ranking of everything they might choose. This ordering simultaneously explains what they choose, represents what they prefer, and measures what's good for them.
Elegant. Parsimonious. And, Sen argues, deeply misleading.
Samuelson's revealed preference theory (1938) made this operational. If a person picks A when B is available, A is "revealed preferred" to B. No psychology needed — just watch what they do.
From enough observations, we can reconstruct the complete ranking. Behavior is all economics needs.
Three claims, compressed into one assumption: choice reveals preference. Preference indicates welfare. Rationality equals consistency.
This chain is the foundation of welfare economics. Sen's life's work is a systematic assault on each link.
You're offered an apple or nothing. You take the apple. Then the host brings out a fruit basket — the last one, which you know they wanted. From the larger menu, you take nothing.
Same apple, different meaning. Your behavior violates contraction consistency — the axiom underlying revealed preference. Yet you are perfectly rational.
A person refuses to cross a picket line, even though shopping would benefit them and they feel no personal distress. They choose from commitment — a judgment about what is right.
Their choice does not maximize their welfare. The gap is not irrationality. It is moral agency.
Sen argues we must distinguish: what you do, what you judge best, and what's good for you.
A person may prefer to act on a moral commitment that reduces their welfare. Their choice reflects their preference, not their self-interest. One ordering cannot carry all three meanings.
Prude and Lewd disagree about who should read Lady Chatterley's Lover. Each person's liberty says they control their own reading. The Pareto principle says if both prefer X to Y, so should society.
These two reasonable principles generate a cycle — no social ordering exists. The content of preferences matters, but the formal apparatus cannot see it.
People under chronic deprivation adjust their desires downward. The underfed laborer may not report hunger. The woman denied education may not want schooling.
If welfare = preference satisfaction, these people appear satisfied. Their preferences have been deformed by the very conditions we should be evaluating.
Sen's capability approach evaluates well-being by what people are actually able to do and be — their real freedoms, not their reported satisfaction.
It does not discard preferences. It situates them in a richer framework that can detect deprivation even when preferences have adapted to conceal it.
In 1977, Sen coined the phrase to describe the agents of economic theory — beings whose every action springs from a single, self-interested preference ordering. Real people are richer: they sympathize, they commit, they reason about their reasons.
An economics that treats them as utility-maximizing automata is not merely incomplete. It is actively harmful — licensing policies that ignore the difference between chosen deprivation and enforced deprivation, between principled sacrifice and helpless acceptance.
Every ethical framework is defined by the information it uses — and the information it excludes. Revealed preference has the narrowest basis of all.
Sen's key methodological insight: the information a framework ignores is as important as the information it uses.
The deepest lesson of Sen's work is that preferences are not just rankings — they have content, context, and history. A preference about your own body is different from a preference about your neighbor's reading habits, even if both are binary comparisons over social states.
Ignoring this distinction doesn't simplify the analysis. It corrupts it.
Set two people's preference orderings over three social states and assign each a personal domain. Can liberty and the Pareto principle coexist?