Bentham and Mill held that well-being is pleasure and the absence of pain. Many later utilitarians kept the structure of the theory but changed its account of the good. Instead of asking how much pleasure an outcome contains, they asked how far it gives people what they want.
Preference utilitarianism
Preference utilitarianism holds that well-being consists in having one's preferences satisfied. The best outcome is the one that satisfies the most preferences, weighted by their strength. This view has several attractions. It respects people's own aims instead of imposing a single picture of the good life. A person who prefers a hard, meaningful life to an easy, pleasant one gets what she wants. Preferences can also be studied through choices, while pleasure is hidden inside each mind.
R. M. Hare reached preference utilitarianism through his theory of moral language. To make a moral judgment, he argued, is to prescribe something for everyone in relevantly similar situations, including yourself in the other person's place. If I imagine myself in each affected person's position with that person's preferences, I will end up giving everyone's preferences equal weight. The Australian philosopher Peter Singer adopted a preference view in his early work, but Singer later moved from preference utilitarianism to hedonism, in a 2014 book written with Katarzyna de Lazari-Radek.
Which preferences?
Not every preference seems worth satisfying. Actual preferences can rest on false beliefs, as when someone wants a drink that is in fact poison. So most preference utilitarians count informed preferences: what a person would want if she knew the relevant facts and thought clearly. This helps, but it raises its own puzzles. How much would she need to know? And are informed preferences still hers if she does not actually have them?
The American philosopher Ronald Dworkin raised a sharper problem in 1977. He distinguished personal preferences, about one's own life, from external preferences, about what others should have or how they should live. If a majority prefers that a minority be denied jobs, counting that external preference lets prejudice weigh against its victims, so the minority loses twice: once in its own preferences and once in others' preferences about it. Dworkin argued that a fair calculation must leave external preferences out, which in practice means protecting people with rights.
Harsanyi's impartial chooser
The Hungarian-American economist John Harsanyi gave utilitarianism a new kind of defense in the 1950s. Imagine choosing the rules of a society without knowing who you will be in it. Harsanyi argued that an impartial chooser with an equal chance of being anyone would pick the arrangement with the highest average welfare. That is just what a rational person does when facing risks to herself. Utilitarianism, on this view, is what self-interest would choose behind a screen of impartiality. John Rawls later used a similar device but argued that a rational chooser would instead protect the worst position, and the two men debated the point for years.
Welfare economics
Utilitarian ideas shaped economics. Early welfare economists spoke freely of maximizing total satisfaction. Then the British economist Lionel Robbins argued in 1932 that comparisons of satisfaction between different people cannot be tested scientifically. We cannot measure whether a rich man's loss of one pound hurts him less than a poor man's gain helps her.
Economists responded by building tests that avoided such comparisons. A Pareto improvement is a change that makes at least one person better off and no one worse off, named after the Italian economist Vilfredo Pareto. But almost every real policy has some losers. So in 1939 Nicholas Kaldor and John Hicks proposed a looser standard: a change passes if the winners gain enough that they could compensate the losers, even if no payment is made. This compensation test underlies much modern cost-benefit analysis. Critics note that it can approve projects that make poor people worse off, since compensation that is never paid comforts no one.
Worked example: the bypass road
A region plans a bypass that will save forty thousand drivers about ten minutes a day, but will run beside a village of three hundred people, whose homes will lose value and whose air will grow worse. Is it a Pareto improvement? No, since the villagers lose. Does it pass the Kaldor and Hicks test? Probably: the drivers' total time savings may be worth far more than the villagers' losses, so the winners could in principle pay the losers. A preference utilitarian would want more than this. She would ask about the strength of the villagers' informed preferences about their homes, not only their market value. Dworkin would ask whether any preferences being counted are external, such as hostility to the villagers themselves. And many would conclude that the project should go ahead only if compensation is actually paid, turning a paper test into a fairer outcome.
