II. THE MORAL QUESTION
When the political question is raised, and the state’s normative claims are brought before the court of morality, this court’s jurisdiction is for the most part taken for granted. That the political question arises between two contrasting poles—of prudence and morality—reminds us, however, of the challenge that the self-interested individual poses not just to the state and its law but to morality as well. For this individual, keen to advance her interests and satisfy her desires, morality presumes to stand in the way. Why would the individual care about its demands? How are we to understand morality’s grip in possible derogation of our own interests and desires? In Kant’s well-known formulation, how is morality possible? Call this the moral question. Clearly, our answer to the political question must be linked to our answer to the moral question.
Adding the moral question to the political question, while compounding difficulties, also provides a clue. Both questions must respond to the same challenge, posed by the self-regarding individual. Given the similarity between the two questions and the common challenge they face, strategies for coping with the moral question may be employed in coping with the political question too. One response to the moral question, of which Kant’s own moral theory is a prime example, resorts to abstraction. Since morality purports to speak in a single voice to or on behalf of individuals whose interests and desires potentially conflict, it presumably requires a unitary standpoint, occupied by every human being. Abstraction paves the way. By abstracting from actual, concrete individuals, their interests and desires, we efface differences and construct a single platform on which they all stand. In Kant’s case this feat is accomplished by means of the noumenal self, characterized exclusively by the possession of a rational will, and by the uplifting image of a Kingdom of Ends, a forum in which abstractly conceived noumenal selves spell out the practical implications of their shared humanity.2
It is instructive to note that the most influential recent engagement with the political question, that of John Rawls, purports to follow Kant in this regard. Since Rawls considers justice to be the primary virtue of political institutions, his response to the political question takes the form of a procedure for constructing a society’s constitution, laws, and institutions that embody sound principles of justice. Rawls explicitly models his procedure on Kant’s approach to the moral question.3 The participants in the original position, a forum analogous to the Kingdom of Ends, are abstracted from actual human beings by means of the veil of ignorance, and so reach principles of justice in their shared capacity as citizens, oblivious to distinguishing characteristics and conflicting ends that keep them apart.
On a closer look, however, Rawls’s use of abstraction turns out to be at once too timorous and excessive in ways that help reveal some of the broader issues involved. To appreciate the first weakness, we need to compare Rawls’s theory to Kant’s. Despite their surface similarity, the approaches are fundamentally different, exposing a crucial ambiguity in the notion of abstraction and its relationship to the self. In employing abstraction, Kant is making a metaphysical claim. His moral theory is grounded in a bifurcated metaphysics that distinguishes between the world of appearances—that is, the world as it appears to creatures with the particular perceptual and cognitive capacities that human beings happen to possess—and the world as it exists apart from humans’ perception of it, the world of things-i n-themselves. People belong to both domains. As phenomenal selves they belong to the world of appearances, in which psychological inclinations participate in the same system of perceptual and cognitive capacities by means of which all of human reality is constructed. Qua noumenal selves, however, they are things-in-themselves, to which ex hypothesi they have no experiential access. We can, however, use our philosophical imagination to project on this blank screen the aspects of our moral condition that the phenomenal self cannot by itself accommodate. Specifically, we can view moral reasons as applying to us as noumenal selves and motivating us in this capacity.4
Post-Kantian philosophy, however, is generally averse to this bifurcated metaphysics, and at any rate Rawls abjures it. Cut off from such metaphysical moorings, Rawls’s abstraction differs fundamentally from Kant’s. Unlike the Kingdom of Ends and its noumenal inhabitants, the original position is a hypothetical meeting of imaginary representatives, whose characteristics purport to be nothing more than theoretical stipulation. The original position and its abstract inhabitants accordingly play a much more attenuated role in answering the political question than the Kingdom of Ends and the noumenal self play in answering the moral. The normative force of the principles of justice and of the laws and institutions they generate comes from outside the theoretical devices Rawls employs. He appeals from the start to people who are assumed to possess a sense of justice; the original position serves only as a heuristic device designed to instruct them about what justice, to which they are independently committed, requires.5 But appealing in this way to a sense of justice is unsatisfactory. If we are puzzled about the source of our alleged commitment even toward a just state, positing a sense of justice that accommodates such a commitment from the start is too ad hoc, and does little to solve the puzzle.
Rawls’s abstraction is also excessive for the task he undertakes. Depriving the participants in the original position of all individuating characteristics is designed to replicate Kant’s subject of morality, the noumenal self. But what would stop such an abstract self from assuming a universal perspective? Why would its interest in justice and the scope of the principles it adopts be confined to domestic institutions and apply only to citizens of a single state? This indeed is the gist of the critique that communitarians launch against Rawls. On the communitarian view, only a “situated” self, thickly constituted by communal norms and practices, can sustain the burdens of communal life and exhibit the other-regarding concerns that justice mandates.6
But this communitarian critique of Rawls’s position, and the alternative it presents to liberalism’s abstract strain, raise difficulties of their own. First, by privileging the community and its norms, the communitarian position militates against a universal morality, and weighs instead in favor of moral relativism that many, including some communitarians, find unappealing. Second, when the communitarian trains her critique on Kantian abstraction, she tends to downplay the individualist challenge to which the political question must also respond. After all, the communitarian’s situated self isn’t quite the concrete, prudential self either. The integration of the individual into the community denoted by the “situated” conception of self risks displacing not only the universal standpoint of morality but also the unique standpoint of the individual and its normative significance. I consider these next. 
impulse, instinct, or whim. Much as you’re inclined to escape an occur- rent pain, prudence might require that you endure it, say, for medical reasons. Removing one’s hand from the fire is explained by the fact that the fire hurts. But when you refrain from doing so on account of prospects of greater future pain, we need an altogether different account, since unlike occurrent pain, future pain does not hurt. Why would you resist present desires or assume burdens on behalf of a future self?7 Call this the prudential question.
This question too can be posed in the idiom of autonomy. I have earlier mentioned the distinction between personal and moral autonomy, and suggested that political autonomy represents a distinctive category intermediate between the two. But what does personal autonomy amount to, and what does it have in common with moral autonomy? A possible answer invokes Kant’s distinction between psychological inclinations and rationality. Just as moral autonomy is a matter of subjecting psychological promptings to the discipline and oversight of a universal standpoint that encompasses humanity as a whole, personal autonomy requires subjecting those same promptings to similar control from a standpoint representing one’s life as a whole. Juxtaposing the alleviation of one’s own occurrent pain to that of someone else’s conflates two different issues: the contrast between the self-regarding and the other-regarding with the contrast between inclination and rationality. To exhibit personal autonomy requires that one submit one’s psychological inclinations, even when self-regarding, to a regime of prudence that resembles in this respect the regime that governs other-regarding concerns as well. The addict, for example, has cravings for narcotic drugs, and yet to be autonomous he must comply with prudential considerations that mandate that these cravings be resisted. How are we to understand this regime and the autonomy it enables?
Not only are these serious puzzles, but they resemble the ones raised by law and morality. When considering the political question and the moral question, we saw how abstraction can provide the requisite unitary perspective, universal in one case, communal in the other. Abstraction from what? The natural answer presumes a concrete individual, whose properties are fully determinate and given. But reflection on the problem of prudence discloses that no such individual exists. A temporally unified individual must be constructed in light of some template, idea, or plan. Here too unity must be imposed on an endless experiential manifold and an equally unruly menu of potential responses and acts. And here, too, abstraction is the route to the unity we seek. As we have just seen, prudence can be every bit as demanding and cumbersome as morality and law. The difference is that when, as in our example, prudence requires that I sustain some occurrent pain, it points to my self-interest (say, in a medical procedure) and so it speaks on behalf of my own future self. But as we have also noted, this future self is an abstraction, and to give its claims priority over the present suffering self I must espouse a position of spatiotemporal neutrality between the two; I must subsume the occurrent experience and the immediate urge to withdraw from the pain within the same abstraction that includes the future promise of health or other enjoyment or relief. In short, no less than the other branches of ethics— morality and law—prudence too depends on abstraction; it requires, if you like, a veil of ignorance of its own.
We can draw two lessons from these remarks. One concerns the crucial role that abstraction plays even at the level of the individual and indeed in constituting one. The second is that the abstraction in the case of prudence is not the same abstraction employed in the case of either morality or law. Prudence would seem to require that we introduce yet another abstract conception of the self, a conception that would now be in competition with both the universal abstraction of the noumenal self and the communal abstraction of the situated self. 
in moral and prudential terms, since these two domains pull in opposite directions, and are themselves under a similar shadow of doubt. An answer to the political question must be part of a more general account that encompasses morality and prudence as well. We have also considered attempts to account for each of the three normative domains we’ve distinguished in terms of a corresponding abstract conception of self. But some formidable difficulties arise, of which three are particularly salient.
The first concerns the relationship among these conceptions. To align each of the various normative standpoints (universal, communal, individual) and their correlative normative orientations (morality, law, prudence) with a suitable conception of self is to replay the tension among the normative domains as a conflict among conceptions of self, and so does not bring us any closer to a unified account. The second difficulty concerns each of the accounts that an appeal to abstraction is expected to provide for its respective domain. It is natural to speak in this connection about an abstract conception of self. But this locution conjures up a certain imagery in which the abstraction is a mere representation of something else. This imagery comes into play, since the notion of an abstract conception of self is naturally construed in light of the broader idea of “an abstract conception of X,” in which X is implicitly taken to be some concrete material object, such as an elephant or a chair. And obviously, only representations of such objects can be more or less abstract, not the objects themselves. A drawing or description of an elephant may render it in various degrees of resolution and detail, and so be more or less abstract. But it makes no sense to talk about a more or less abstract version of Jumbo itself. Now if abstraction relates to human beings as to elephants, then human beings cannot be any more abstract than Jumbo can. On this view, and as our discussion of Rawls illustrates, abstraction can yield only hypothetical representations of human beings, thereby creating a gap between the unitary normative standpoint that abstraction is expected to create, and the concrete individuals that are supposed to occupy this standpoint. Unless we suppose in each case that people are already disposed from the start to act prudentially, or legally, or morally, in what way does an abstract representation help account for the normative grip that each of these domains is supposed to exert? Why should any of us care for one or another representation of ourselves? Finally, the third challenge that a unified account of ethics faces results from the exclusivity or comprehensiveness that each of the domains appears to claim. Morality, law, and prudence, each claims authority over human life as a whole, at least in the sense of being in charge of defining over which issues each has a final say. This suggests an apparently inescapable conflict that a unified account would be hard put to resolve.
The key to the response I propose requires that we reconceive the relationship of abstraction to self. Instead of a competition among variously abstract representations of self, we need to think of a single conception of self as abstract. On this conception, abstraction pertains to the actual self, rather than being a property of its representations, so that different levels of abstraction can be all internal to the self. This conception of self is implicit in the meaning-oriented, constructivist approach to the self we’ve discussed in previous chapters, and in the traditions of thought on which we have already drawn: it is the view of the self as an ordered configuration of meanings, for which the literary and dramaturgical imageries provide some familiar templates. Such literary and dramaturgical analogies alter our understanding of the way abstraction relates to human beings. In the case of physical objects there is a clear distinction between the object and its representation, for example, Jumbo and a drawing or description of it. In the case of literary objects, this distinction is effaced. The point can be made succinctly in terms of the two different uses of the verb “tell,” intransitive and transitive. In telling me about a physical object, an elephant or a car, you provide a description of the object or an account of it. The description or the account is external to the object: in describing the car, you don’t give me the car or any part of it. But when it comes to literary objects, tell can be used transitively. To tell a story or a joke is not to describe but to transcribe it; it is to convey to the listener the very story or joke that is the subject matter of the telling. Telling a story, transcribing it, can be performed at various levels of abstraction or detail. Suppose that I ask you to tell me the story of Macbeth, and you oblige with a synopsis. This may be fully responsive to my request. Whether the level of abstraction of your narration is adequate will depend on such contextual considerations as the degree of my curiosity or whether I am in a rush, and such considerations may call for greater abstraction as much as for more detail. Just as interpretation can add detail to a story without changing it, a synopsis gives us a shortened version of it. Different renditions of a story that vary in level of abstraction are equally versions of the story itself.
In addition to such “vertical” differences among versions in level of abstraction, versions can also diverge “horizontally,” when they differ in some of their detail, for example, the story of Faust as rendered by Marlowe, Goethe, Lessing, Heine, and Mamet. In what sense are they all, despite their differences, versions of a single story? Here too the answer lies in abstraction. Since increased abstraction effaces differences among the versions, at a higher level of abstraction the different versions merge into a single story, whereas at lower levels of abstraction (or higher levels of resolution) the differences among the versions appear. By the same token, when we think of the self as abstract, the content or meaning constitutive of the self can also range over various levels of abstraction. Distinguishing characteristics that appear at lower levels of abstraction are effaced at higher levels, and so interpersonal commonalities and unities appear.
This point can also be made in terms of the dramaturgical imagery. Many of the roles constitutive of our identities are nested: a cardiologist and a dermatologist are both physicians. In what sense do they occupy different roles and in what sense one and the same? As in the case of the different versions of the same story, roles too can diverge at lower levels of abstraction and converge at a higher level. Now as we further ascend the ladder of abstraction, we reach the idea of a person understood in terms of the convergent abstract content of all human lives. When, moving in the opposite direction, we descend the ladder of abstraction, and increase resolution, individuals come into view. As we have seen earlier, to be an individual also involves abstraction, though the level of resolution is much greater (or, conversely, the level of abstraction lower) than that pertaining to being a person. Individuals enact or realize at a high level of specificity, and therefore in vastly ramified and divergent ways, a singular meaning or content that pertains to all persons as such.
Person and individual thus label the two polar extremes on a spectrum of abstraction over which the self ranges. This spectrum contains innumerable intermediate levels, such as those occupied by the role of cardiologist and physician just mentioned. But here we need draw a further distinction. Both person and individual are comprehensive terms, in that at their respective levels of abstraction they each pertain to a human being as a whole, whereas cardiologist and physician are partial, pertaining to some aspects of their bearer’s identity but not to others. In addition to such terms referring to partial roles, however, there is logical room for a comprehensive term that applies to a human being as a whole, but at an intermediate level of abstraction. Citizen is such a term.8 To be French, for example, is to be constituted by a concatenation of meanings that at a suitable level of abstraction defines a common identity of being French. These three terms—i ndividual, citizen, person—accordingly designate the same human being conceived at different levels of abstraction: individual alludes to a cluster of meanings unique to her, citizen to meanings she shares with the other members of a political community, and person to the more abstract content shared by every human being as such. 
than standing for a disjunction or an opposition. Even so, they can each be loosely associated with a different value or goal. Applying to people at the highest level of abstraction, morality upholds dignity, the value all persons have qua human beings. Law spells out the more specific requirements of justice among the members of a political community. Prudence, operating at an even greater level of specificity, at which each individual’s particular experiences come into view and take pride of place, is oriented toward the individual’s happiness. Acting in one’s capacities as an individual, a citizen, and a person, one acts, respectively, prudently, legally, and morally, and so one pursues happiness, realizes justice, and respects dignity.
This picture suggests straightforward answers to the questions regarding the practical domain we have raised. First, to see morality, law, and prudence as operating at various levels of abstraction explains how each of them can apply to one’s life as a whole, without being in necessary conflict with the others. Since they are each other’s abstractions, or in reverse order, each other’s elaborations, each of these normative systems can claim exclusive dominion over the self’s corresponding level of abstraction, consistent with recognizing the others’ exclusivity at other levels. This picture also relieves the pressure to divide all interests, reasons, attitudes, and the like into self-regarding and other-regarding. This binary division is replaced by a continuum of increasing abstraction and correspondingly greater convergence of content, a continuum of which the unique individual and humanity as a whole are the two extreme poles. Political reasons (attitudes, etc.) pertain to intermediate levels of abstraction, which create smaller clusters of partial convergences of content, and hence more limited pockets of solidarity than the entire human race. Finally, the tripartite division of autonomy into personal, political, and moral also finds its place. Autonomy at all three levels involves subjecting impulse to norm. The norm must be internal, though, rather than externally imposed. But to be internal it need not be, indeed it cannot be, invented by the agent or pulled out of thin air. Rather, a norm is internal insofar as it fits, at a suitable level of abstraction, within the structure of meanings that defines the agent as an individual, a citizen, or a person, or, to put the same point differently, insofar as the agent identifies with it, or endorses it, as an element within the overall structure of meanings she enacts. Within this picture, the subject of the self-government exercised by the state, and hence of political autonomy, is not an aggregate of individuals, nor is it an impersonal collective entity, but rather each citizen, abstractly conceived.
-  THE PRUDENTIAL QUESTION In contemplating both the moral and the political question, the self-regarding individual provides the natural, taken-for-granted point ofdeparture, posing a seemingly obvious challenge with which moralityand law must contend. The claims of morality and of law are commonlyperceived as demands made on the individual, and so her responding tothem is deemed in need of explanation in a way that her pursuing herown interests is not. Removing your hand from a burning stove is easilyexplained in terms that don’t seem to apply to your pulling someone else’shand from harm’s way. Nothing corresponding to the heavy machineryof morality or law that comes into play in the latter case seems to be involved in the former. Your own sharp pain does all the motivating as wellas explanatory work. But even this simple example reveals a difficulty in the notion of theself- interested individual and in the normative challenge it is taken topresent. To act in a self-interested manner is not the same as to act on
-  THE ABSTRACT SELF We have started our discussion by attending to the political question: whyshould we obey the law or recognize the authority of the state is a familiarand persistent challenge. As it turns out, however, an adequate answerto the political question must do more than assess the claims of the state
-  ANSWERING THE QUESTIONS We can now combine this conception of self as abstract with the tripartitedivision of ethics discussed earlier. On the resulting picture, the threesubdivisions of ethics—prudence, law, and morality—relate to us in thesame kind of way: morality defines in part what a person is, thereby helping constitute the common identity of all human beings; law defines inpart what a citizen is, thereby helping constitute the common identityof, say, the Brazilians or the French; and prudence defines in part whateach individual is, thereby helping constitute each individual’s uniqueidentity. Since the three branches of ethics correspond to different levelsof abstraction of the self, they represent points on a continuum rather