Preference utilitarianism
Most modern contributions to utilitarianism retain the preference model of utility, such that utility is simply the satisfaction of preferences. The “strong preference” model is the one that is most commonly used in social sciences: individual preferences are revealed in the actual choices of individuals.
Specifically, utilities are the numerical representation of preferences as revealed by choices. If I choose wine when I could drink tea, my preference is for wine over tea. In terms of utility, the numerical value of the utility that I associate with wine is therefore greater than the numerical value of the utility that I associate with tea. For example, I can capture either the choice situation or the preference by the following utility numbers: U(Wine) = 10 and U(Tea) = 0 or, equivalently, 4 and -2.67 respectively. In this context, ordinal measures of utility are all that counts, but cardinal utility is meaningless - such that summing and comparison are impossible. Arrow and Hahn, Debreu, Hicks and Samuelson, among many others, had been convinced of the primacy of this model for economic science since the forties. The revealed preference model is the basis for the interpretation of the utilities found in our microeconomic textbooks. Nevertheless, the same model of utility is likely to be used in other chapters of these textbooks, those which tackle issues of welfare economics, and even applied public economics. In this context, social welfare is hence supposed to depend on individual utilities alone, which only represent individual choices. The use of this model in welfare economics is heavily criticized for several reasons (for example, Hausman 2012). For one thing, it ignores the fact that individuals may act against their own interests: for instance, a knowledgeable person who continues to smoke, or a penniless young man who prefers to work for a low wage rather than concentrate on his studies. More generally, we cannot distinguish actions by interest, lack of information, whether they are based on false beliefs or lack of will, constrained by social norms, and so on. As a consequence, it is difficult to interpret the concept of utility corresponding to the strong preference model as the well-being of the person, and even more as relevant information to compute social welfare.While the strong model of preferences -that is, revealed preference theory - is mostly retained in economics, most philosophers prefer to retain a weak model of preferences. They claim that utility stands for specific kinds of preferences: they should be rational, informed, and based on true beliefs (for example, Hare 1981; Griffin 1986; Sumner 1996). The introduction of these conditions has two consequences. First, this model answers most of the criticisms that plagued utilitarianism both in its hedonist version and in the strong preference model. Second, it implies that, for utilitarianism, the relevant preferences are not actual preferences but some kind of ethical preferences: actual preferences should be “laundered” (Goodin 1986) - notice that the issue of who should be entitled to perform the actual task of laundering preferences when utilitarian policies are implemented remains open so far.
Further, if you suppose we know the rational and informed preferences, and if you can transform each individual utility by any positive monotonic function f(.), then you can design some ordinalist utilitarianism where:
It has been shown that the Rawlsian ranking is a special case of all utilitarian ordinalist functions, where the concavity of f is extreme (Arrow 1973).