The concept of utility has a long and complex history, and may be traced back to Aristotle. In modern (neoclassical) economics, utility is generally taken to be a number that represents an individual’s preferences. Despite this definition (or perhaps because ofit), the notion of utility remains one of the most debated economic concepts of our time; after centuries of discussion, there is still no consensus among economists about the precise definition of the term “utility”.
In contrast to this wealth of debate regarding utility, the economic literature contains very little discussion about the meaning and significance of the corresponding and equally important concept of preference. In particular, the literature mentions nothing about the origin and development of this concept. How then can we say that we have defined utility if we do not know what is meant by preference? In Part I of this thesis it is shown that the neoclassical heavily-circumscribed conception of preference, as it pertains to utility theory, is merely an abstract representation of the Utilitarian notion of pleasure. This is accomplished through an examination of the works of the principal contributors to the historical development of the utility concept, namely Jevons, Edgeworth, Fisher and Pareto, and the post-1900 contributions of economists such as Wicksteed, Hicks, Wold, Debreu, Samuelson and Chipman.
It is demonstrated that the Utilitarian notion of pleasure implies a form of “moralistic hierarchy”. This conception of pleasure, as described by Bentham and JS Mill (which the Utilitarians had adapted from the philosophical conception of pleasure propagated by the ancient Greeks and the Scholastics), was adopted by Jevons and Edgeworth as the starting point of their respective theories of utility measurement. Both limited the Utilitarian definition in order to produce a model that could “measure” pleasure. Successive generations of economists then progressively reduced the pleasure concept from a broad psycho-philosophical notion to the bare construct of preference that we see today. The contemporary preference concept is largely due to Wold, who formally defined preference by deriving conditions that a preference relation must satisfy in order that a continuous utility function may exist, and Debreu, who provided the definitive distinction between the concepts of preference and utility by showing that there can be preference without utility. In Part II of this thesis, we present a series of essays on concepts central to preference and utility theory, including the “measurability” of utility, and the meaning of “indifference”, “ordinal utility” and “cardinal utility”; we also look at how the concepts of preference and utility are applied in other disciplines outside the realm of economics. We find that, for the early utility theorists, the measurability of utility was not an issue; the theoretical possibility of measurability was accepted as given. However, their respective interpretations of “measurability” differ. We examine why most economists unquestioningly assume that utility is a real number, and show that this view is due to Jevons, and that it has come about because of the particular structure of Jevons’ utility functions.
We examine the competing views of preference and indifference, and find that although these concepts have appeared in the literature since the time of Edgeworth, they largely remained undefined, their meanings taken as given, until the late 1930s; during the mid-twentieth century the meanings of these concepts began to be questioned.
We investigate the origin and evolution of the concepts of “ordinal utility” and “cardinal utility”, and find that the economic interpretations of the concepts of “ordinality” and “cardinality” bear no resemblance to the mathematical definitions of these terms, and further we conclude that attempting to categorise the early theories of utility as either ordinal or cardinal in nature does little to aid our understanding of these theories. Weshow that Edgeworth was the first to distinguish between “ordinal” and “cardinal” utility, although he made use of an entirely different nomenclature, differentiating instead between the terms “quantitative” and “numerical”.