244
PARTISAN REVIEW
ception of duty to see the advantage of finding language that allows
us to distinguish between the person doing his
duty
(by his lights) and
doing what he
prefers.
This distinction makes no sense if one's con–
ception of duty is just part of one's preferences . By the same token,
such a formulation has no place for the Aristotelian notion of
akrasia
or weakness of will: you cannot fall short of your ideals by yielding to
your preferences if the former are part of the latter. And finally - the
issue most relevant to Hirschman - the idea of
changing
your pref–
erences to conform to your ideals, what we might call moral aspira–
tion, makes no sense on the neoclassical assumption of unchanging
preferences.
Sen has suggested that these problems can be overcome by
describing a moral ideal not as a preference among states of affairs,
like other preferences, but rather a preference
among
preferences: an
expression of what it is you would prefer to want. Such a preference,
naturally called a "second-order preference" or "metapreference,"
draws an appropriate distinction of levels between ideals and pref–
erences, and clearly creates a useful framework for analyzing
akrasia
and preference change.
This example illustrates a number of features that characterize
Sen's work more generally. The first job of serious ethical analysis,
formal or informal, is to get an adequate description of the moral
phenomena in question . Given the intricacies of real moral judg–
ment and deliberation, this is likely to require some considerable
complexity in the basic formulation of the problem - in particular a
richer conception of the person than neoclassical theory will allow .
Formal theorizing helps principally by providing an analytical ap–
paratus capable of expressing these complexities with precision.
Such a stress on the inherent complexity of moral reasoning im–
plies a characteristic attitude toward the project of moral theory .
Systematic moral thinking should cease being so preoccupied with
deciding which oversimplified system of moral thinking is right and
aim instead at developing the analytical and conceptual capacity to
describe ways of thinking about morality that are rich enough to
have at least a chance of being helpful. Any adequate moral theory
will have to have room for both Kantian and utilitarian aspects. And
it will also have to allow for the great importance we attach to secur–
ing individual rights- as Robert Nozick insists- while avoiding the
blindness to consequences in defining those rights, that Nozick's
theory displays, blindness that can lead to the sorts of disturbing
