the social utility of a shift from one level to another is the same for all
individuals. On the one hand, this procedure would weigh identically
those changes involving the same number of discriminations that indi-
viduals
felt differently about, some having stronger feelings than others;
while on the other hand, it would count more heavily the changes experi-
enced by those individuals who appear to make more discriminations.
Surely it is unsatisfactory to discount the strength of attitudes, and espe-
cially to reward so highly the capacity for noting distinctions which may
vary systematically with temperament and training.
45
Indeed, the whole
procedure seems arbitrary. It has the merit, however, of illustrating the
way in which the principle of utility is likely to contain implicit ethical
assumptions in the method chosen for establishing
the required measure
of utility. The concept of happiness and well-being is not sufficiently
determinate, and even to define a suitable cardinal measure we may have
to look at the moral theory in which it will be used.
Analogous difficulties arise with the Neumann-Morgenstern defini-
tion.
46
It can be shown that if an individual’s choices between risky
prospects satisfy certain postulates, then there exist utility numbers corre-
sponding to the alternatives in such a way that his decisions can be
interpreted as maximizing expected utility. He chooses as if he were
guided by the mathematical expectation of these utility numbers; and
these assignments of utility are unique up to a positive linear transforma-
tion. Of course, it is not maintained that the individual himself uses an
assignment of utilities in making his decisions. These numbers do not
guide
his choices, nor do they provide a first-person procedure of delib-
eration. Rather, given that a person’s preferences among prospects ful-
fill certain conditions, the observing mathematician can, theoretically at
least, compute numbers that describe these preferences as maximizing
expected utility in the sense defined. So far nothing follows about the
actual course of reflection, or the criteria, if any, that the individual relies
upon; nor is anything implied about what features of the alternatives the
utility numbers correspond to or represent.
Now assuming that we can set up a cardinal utility for each person,
how is the interpersonal measure to be established?
A familiar proposal is
the zero-one rule: assign the value zero to the individual’s worst possible
situation and value one to his best situation. Offhand this seems fair,
45. For these difficulties, see Sen, ibid., pp. 94f; and W. S. Vickrey, “Utility, Strategy, and Social
Decision Rules,”
Quarterly Journal of Economics,
vol. 74 (1960), pp. 519–522.
46. For an account of this, see Baumol,
Economic Theory and Operations Analysis,
pp. 512–528;
and
Luce and Raiffa,
Games and Decisions,
pp. 12–38.
283
49. Comparison with Mixed Conceptions
perhaps expressing in another way the idea that each is to count for one
and no more than one. Yet there are other proposals with comparable
symmetry, for example, that which assigns the value zero to the worst
alternative and the value one to the sum of the utilities from all alterna-
tives.
47
Both of these rules seem equally just, since the first postulates
equal
maximum utility for everyone, the latter equal average utility; but
they may lead to different social decisions. Furthermore, these proposals
postulate in effect that all individuals have similar capacities for satisfac-
tion, and this seems like an unusual price to pay merely to define an
interpersonal measure. These rules clearly determine the concept of well-
being in a special way, for the ordinary notion would appear to allow for
variations in the sense that a different interpretation of the concept would
be equally if not more compatible with common sense. Thus for example
the zero-one rule implies that,
other things equal, greater social utility
results from educating people to have simple desires and to be easily
satisfied; and that such persons will generally have the stronger claims.
They are pleased with less and so presumably can be brought closer to
their highest utility. If one cannot accept these consequences but still
wishes to hold the utilitarian view, some other interpersonal measure
must be found.
Further, we should observe that while the Neumann-Morgenstern pos-
tulates assume that individuals do not enjoy the experience of risk, the
actual process of gambling, the resulting measure is nevertheless influ-
enced by attitudes toward uncertainty as defined by the overall prob-
ability distribution.
48
Thus if this definition
of utility is used in social
decisions, men’s feelings about taking chances will affect the criterion of
well-being that is to be maximized. Once again we see that the conven-
tions defining interpersonal comparisons have unexpected moral conse-
quences. As before the measure of utility is influenced by contingencies
that are arbitrary from a moral point of view. The situation is very differ-
ent from that of justice as fairness as shown by its Kantian interpretation,
the embedding of ideals in its principles, and its reliance upon primary
goods for the necessary interpersonal comparisons.
It would appear, then, that the vagueness of
the utilitarian principle is
not likely to be satisfactorily removed simply by a more precise measure
of utility. To the contrary, once the conventions required for interpersonal
comparisons are examined, we see that there are various methods for
47. See Sen,
Do'stlaringiz bilan baham: 
