PART 1—
BEGINNINGS
I think you're right, Lysis, to say that if we were looking at things the right way, we wouldn't be so far off course. Let's not go in that direction any longer.
Socrates—from Plato's Lysis
This book is about human agency, practical reason, and morality, in this order. The order is important, because the results of inquiry are often dictated by where it begins. I start then with a brief explanation of why I proceed in the way that I do and where I think it leads. I distinguish my approach from two others, one that begins with morality and proceeds to agency and practical reason and another that begins, as I do, with agency. The latter view has gained much currency among contemporary moral philosophers with the revived interest in the ethics of Immanuel Kant. I refer to a school of thought called Kantian internalism. Among the proponents of this version of Kantianism are Barbara Herman, Marcia Baron, Alan Donagan, Thomas Hill, Jr., Onora O'Neill, Henry Allison, David Cummiskey, and Christine Korsgaard.[1] The project of this school is to provide a rational defense of an agent-centered conception of morality that can be
[1] . See Barbara Herman, The Practice of Moral Judgment (Cambridge, Mass.: Harvard University Press, 1993); Marcia Baron, Kantian Ethics Almost Without Apology (Ithaca: Cornell University Press, 1995); Alan Donagan, The Theory of Morality (Chicago: University of Chicago Press, 1977); Thomas Hill, Dignity and Practical Reason in Kant's Moral Theory (Ithaca: Cornell University Press, 1992); Onora O'Neill, Constructions of Reason: Explorations of Kant's Practical Philosophy (Cambridge: Cambridge University Press, 1989), and Towards Justice and Virtue: A Constructive Account of Practical Reason (Cambridge: Cambridge University Press, 1996); Henry E. Allison, Kant's Theory of Freedom (Cambridge: Cambridge University Press, 1990); David Cummiskey, Kantian Consequentialism (New York: Oxford University Press, 1996); and Christine M. Korsgaard, Creating the Kingdom of Ends (Cambridge: Cambridge University Press, 1996), and The Sources of Normativity (Cambridge: Cambridge University Press, 1996).
constructed from Kant's ethics. My primary goal here is to construct an agent-centered conception that is largely Aristotelian in its structure and to argue that it is superior to the alternative of Kantian internalism.
I.
According to a long and venerable philosophical tradition dating back to the Enlightenment, morality is a normative perspective that has three essential features. It is impartial in its evaluations and demands; it universally applies to all human beings who can properly be called agents; and it is one that is rational for all rational human agents to take against alternative perspectives. A major reason that the order that begins with human agency is important is that if we proceed in this way we find that all three Enlightenment claims regarding morality are false, or so I will argue. An even more important reason is that we can discover a better alternative.
The reverse order of the one suggested here is the approach historically associated with one interpretation of the Enlightenment tradition. Because the Enlightenment tradition, on this interpretation, begins with an analysis of the morality of actions and obligations and proceeds to the morality of agents and character, it is act-centered rather than agent-centered. A major puzzle of this approach is that it quickly generates the problem of why we should be moral: Somehow we are first to understand what morality requires of us and then to worry over whether those requirements are rational. But if we are concerned with rationality, why begin with morality at all? Why not begin with an account of practical reason and see if anything resembling morality shows up within the account? The danger of the traditional methodology is that distortions of practical reason will be imposed to save the conception of morality.
This explains, then, why I think that practical reason comes before morality. But why think that a prior conception of agency is necessary for a conception of practical reason? I defend the claim that practical rationality is character-relative in the sense that what is practically rational for an agent turns on the land of person the agent is. In this regard, we sometimes describe others as courageous, temperate, or long-suffering. When we do, we employ aretaic concepts involving self-control. On other occasions, we describe others as respectful, sympathetic, loving, or devoted. When we do, we employ concepts involving virtues that I call virtues of caring. It is in terms of the latter, the virtues of caring, that I claim practical reason is character-relative. That there is no standpoint outside character that is the
foundation for practical reason I take to be a central Aristotelian point. Thus I begin not with an account of morality in terms of moral agency but with an account of practical reason in terms of human character, in terms of the integrity of human agency. I then provide a conception of morality within that conception of practical reason.
Finally, the claim that practical reason is character-relative (a point to which I will return at the end of this introduction) does not in itself deny the universality of morality, for character might be shared among human beings in a way that establishes a substantial universal form of practical reason and therefore of morality. Very reluctantly, I have come to the conclusion that there is no such form. Nevertheless, there is a conception of morality, I will argue, that is practically rational for a surprisingly large portion of humanity. Things are relative but not as relative as some relativists would have us believe. Constructing this conception of morality is by far the main task of this book.
I argue for a conception of human agency that begins with the concept of human integrity. In the process, I argue that criteria for practical rationality are character-relative. I then argue that normative beliefs, including beliefs about obligation, have rational foundation in terms of their providing solutions to what I call integration problems. Moreover, it is the notion of integrity that provides us with a test for adjudicating philosophical disputes about the nature of rational agency. This test I call the integration test. It will turn out that the dispute between the view presented here and the view of Kantian internalism turns on a dispute about the nature of rational agency. If such a dispute can be rationally adjudicated and one of the views is correct, there must be some test that selects for the better theory. It is the integration test that plays this adjudicating role. Later, we will see what this test is.
A reason for suspecting that integrity is the clue to providing an adequate conception of practical reason involves a central issue of this book: the issue of regulative norms and their relationships within practical reason. Less technically and more generally, the idea is something like the following. People care about a variety of things in a variety of ways. They love their friends, their families, and their communities; they respect, sympathize with, and esteem strangers; they are dedicated to their work and enjoy their play, and they are committed to causes, practices, and principles. All these things factor into practical reason being what it is for at least most human beings. As factors within practical reason, these various concerns take on the status of norms that regulate our thinking about how to live and
what to do. Some of these norms are "partial" because they involve personal connectedness in a sense that others do not. Loving a child is a good example of a partial norm. Other norms lack this personal element and are best understood as "impartial." Respect and sympathy for strangers are usually good examples. One of the central questions of this book is, How are we are to understand these norms and their interrelationships such that they are the norms of a person as a practical reasoner? A very general answer is that they must be integrated as a roughly coherent set of norms. To be a set of norms that reflect personhood and practical reason they must achieve a certain kind of integrity as a whole. As such, they must constitute a psychology .
Another, more specific answer, of course, is to be found among the various Enlightenment conceptions of practical reason and morality, the most developed of which is the Kantian branch of that tradition. In chapter 2, I argue that to achieve the necessary kind of wholeness or integrity and to account for the right psychology, the relationships among these regulative norms cannot be hierarchical in the way required by the Enlightenment tradition. More precisely, I argue that this is true on the assumptions of our considered moral judgments.
Against some of the most recent attempts to defend an Enlightenment view, I argue that human agents must integrate the kinds of concerns they have into a manageably coherent life in order to be agents at all and that there is no way of achieving such integration through a dominant norm that regulates other norms without being regulated in any way by them. That is, the regulative relationships among norms must be symmetrical rather than asymmetrical in order to achieve integration of the sort necessary for at least most humans. If this is true, I argue, then neither impartial respect nor impartial sympathy can do the work required of them by Kantians and utilitarians. Moreover, I argue that no norm can be asymmetrically dominant over other norms within practical reason, no matter whether it is partial or impartial. The reader might think of this as a hypothesis and the remainder of the book as a test of it. On this view, the role any norm plays in an agent's life will be recognized by how it regulates and is regulated by other norms in an agent's psychology in a way that preserves the integrity of that psychology. It is from this perspective that I develop a specific answer to our question that is an alternative to the Enlightenment tradition.
The methodology I find distorting, then, is one that begins with a conception of moral agency according to which an agent is a moral agent only
to the extent to which he or she deliberates on normative criteria that are in principle entirely independent of considerations of the agent's own good or particular personal commitments. Call this an agent-neutral theory, since it employs an asymmetrical regulative norm that is neutral regarding the more personal concerns of an agent's life. To be sure, any agent-neutral theory may either allow or require an agent under appropriate circumstances to pursue his or her own good, but this occurs in only two ways. One involves moral coincidence: The permission to pursue one's own good is the coincidental result of applying moral criteria in a completely agent-neutral way; as moral luck would have it, happiness coincides with duty. The other involves an instrumental necessity to the ends of an agent-neutral conception of morality: The pursuit of one's own good is instrumentally necessary to an overall agent-neutral good; duties to oneself flow from the necessity for moral preparedness in one's duties to others.
Increasingly, many contemporary philosophers share the view that human agents are surely not the kinds of agents required by agent-neutral theories of morality. They also share the view that it is a good thing they are not.[2] Some even have the view that human agents are not moral agents. Lacking in this literature, however, is any developed conception of practical reason and morality that represents an alternative to the Enlightenment tradition. What is needed is an alternative moral psychology that will supply the account of regulative norms that the Enlightenment tradition has failed to provide. Moreover, that account must accommodate what is valid in recent Kantian defenses against contemporary philosophical attacks.
While I share the view that it is a good thing we are not the kinds of agents required by agent-neutral conceptions of morality, I do not reject the moral status of our agency. What allows me to do this is a reversal of methodology, which is Aristotelian in spirit. I begin by asking what human agents are like. Is there an essential connection between the agency of any
[2] . See Michael Stocker, "The Schizophrenia of Modern Ethical Theories," and Susan Wolf, "Moral Saints," both in The Virtues: Contemporary Essays on Moral Character , ed. Robert B. Kruschwitz and Robert C. Roberts (Belmont, Calif.: Wadsworth, 1986), 36–45 and 137–52, respectively; Michael Slote, Goods and Virtues (Oxford: Clarendon Press, 1983); Lawrence Blum, Friendship, Altruism, and Morality (London: Routledge and Kegan Paul, 1980); John Cotringham, "Ethics and Impartiality," Philosophical Studies 43 (1983): 83–99, and "Partiality, Favouritism and Morality," Philosophical Quarterly 36 (July 1984): 357–73; Julia Annas, "Personal Love and Kantian Ethics in Effi Briest," Philosophy and Literature 8 (April 1984): 15–31; Bernard Williams, Moral Luck (Cambridge: Cambridge University Press, 1981).
particular human and the good of that human? Only after coming to a somewhat affirmative and informative answer to this question do I proceed. I then ask if there is any importance in retaining a moral /nonmoral distinction regarding human agents. I believe that there is.
If I am right about this, then all morality must be agent-centered in one very important sense: There can be no acceptable theory of moral obligation that is entirely agent-neutral. Moral obligation cannot be entirely independent of what is good for the agent who has the obligations of morality. Nor can it be completely without connection to the agent's particular personal commitments.
The connection for which I argue is this: To be an ongoing agent is not to be the agent of a principle that comes into play willy-nilly as dictated by events external to the agent's reasons for living; rather, it is to be an agent whose actions have their coherence and meaning within the agent's way of life with its own constitutive good. Being an ongoing agent, therefore, essentially involves finding life worthwhile and having reasons for living. It also involves having reasons for living one way rather than another. No way of life, then, is morally obligatory that is entirely independent of a human agent's own good. If this is true, then no individual act that is destructive to the most fundamental goods of that life can be morally obligatory for an agent.
My methodology, therefore, commits me first to come to terms with the connection between the concept of human agency and the concept of a particular human agent's own good. Only then does it allow me to proceed to questions of obligation. Moreover, it commits me to giving an account of an agent's sense of obligation in terms of how an agent's reasons for action emerge in the context of those things in virtue of which the agent finds life meaningful.
II.
If we are to think of morality as a function of practical reason and understand practical reason in terms of the integrative functions of human consciousness, we need a general account of an integrative function. For this we need a general account of integrity. I call this general account the thin conception of integrity.
The thin conception of integrity focuses on the person to whom we are willing to attribute at least four features. Here I will simply state these features, leaving developed discussion and argument for them for the remain-
der of part 1 beginning especially in chapter 3. The first of these features is a sense of self that is sufficiently unified to allow us to say that the person has at least one and no more than one basic "self." The second is a level of self-knowledge inconsistent with a life of pervasive self-deception (or insanity) regarding that person's basic sense of self. The third is the strength of character to meet significant challenges to that which makes this sense of self possible. And the fourth is a sense of self as intrinsically important to some degree as a separate and numerically distinct person. A human lacks integrity in this sense, then, by lacking sufficient unity of self to be at least one and no more than one person, by being unable to live with a level of self-knowledge that reveals who and what the person basically is, by lacking the capacity to meet challenges natural to a human environment, and/or by being unable to assign any intrinsic importance to the fact that he or she is a separate and numerically distinct person.
I say much more about the concept of integrity and motivate its elements in chapter 3, but it seems important for purposes of clarity to comment briefly here on the first two elements. To say that a person of integrity has at least one and no more than one basic "self" is not to endorse any strongly Cartesian views on the unity of consciousness. It may very well be true that the best explanation of certain forms of irrationality is in the notion of "partitioned" systems of consciousness.[3] Self-deception is a possible example. But if partitioning is the best explanation for such phenomena, certain conditions must hold. Foremost is that each of these systems must achieve a certain level of integration or unity within its own domain to constitute a personality system. This system then must be segregated in significant ways from other personality systems that somehow functionally relate to the same human organism. Each such system must have an identity of its own. If these identifiable personality systems are sufficiently conflicting and separate, no overall unity of personality is possible for the human to which they functionally relate. If this is true, then the human lacks the substantial unity of personality required for personal integrity. Similar comments apply if the personality system that represents a particular human's "true self" is partitioned from explicit consciousness. Such a person lacks integrity despite having a very unified personality on any particular occasion.
[3] . See Donald Davidson, "Deception and Division," and David Pears, "The Goals and Strategies of Self-Deception," both in The Multiple Self , ed. Jon Elster (Cambridge: Cambridge University Press, 1985), 79–92 and 59–78, respectively; and David Pears, Motivated Irrationality (Oxford: Clarendon Press, 1984).
It is important, then, to keep in mind two points regarding the first two elements of integrity. First, there must be some overall substantial unity or integration of personality for a human to be a person with integrity. This is in contrast to being a human with several partitioned personality systems with their own self-contained integration. Second, the human who suffers from serious cases of fragmented personality is not a person to whom integrity is attributable, and the same is true of the human whose sense of unity of personality is partitioned from conscious recognition.
From these observations, we can now say in general what an integrative function is. The general conception of an integrative function as it applies to practical reason is that it is a function of one's psychology that makes the basic elements of integrity in the thin sense possible. Later, it will become clear that one major difference between the view advocated here and Kantian internalism is the account of what makes such a function in the general sense possible.
Any specific, as opposed to general, conception of an integrative function would require a substantive conception of integrity. If there are different substantive conceptions of integrity that realize the basic elements of the thin conception and they yield integrative functions significantly different in their configurations, practical relativism is the result.[4] The conception of morality defended here is a function of one specific conception of integrity. I call it the thick conception of integrity.
The thick conception of integrity includes the features of the thin conception but adds others as well. More accurately, the thick conception is only one version of how the thin conception might have substance rather than mere form. This is to say that there might be many different thick conceptions that realize the thin conception other than the one I consider; hence the possibility of relativism.
The agent I have in mind is one who has self-respect and self-esteem, as well as impartial respect, sympathy, and esteem for others. These attitudes regarding others, then, express the relevant impartial norms of the agent of integrity in the thick sense. But in addition to these concerns, this agent also has a number of other concerns that express his or her partial norms.
[4] . I must hedge here. I am now only beginning to take seriously Isaiah Berlin's claim that moral pluralism is neither relativistic nor universalistic. So while I will continue to use the term "relativism," it should be understood here simply as the denial of universalism, leaving open the possibility that there are at least two ways of denying universalism, by appeal to relativism proper or to moral pluralism. For Berlin's discussion, see his book, The Crooked Timber of Humanity (Princeton: Princeton University Press, 1990), esp. 1–20.
Some of these involve personal love for others, including being a loving parent, a loving friend, and a loving neighbor. Finally, this agent leads a life containing a great deal of intrinsically meaningful activity, much of which is aesthetic in value and is dedicated to excellence, that is, to doing what he or she does well, where these concerns are among his or her partial norms. So described, it is important to note that there is nothing about this agent that Kantians should find objectionable. In fact, Kant is plausibly interpreted as claiming that we have duties to cultivate personal capacities in regard to all the features of the thick conception of integrity. The issue is whether he and his advocates can account for this in an acceptable way. More generally, the question is, How can an agent who has these concerns, partial and impartial, integrate them in a way that makes the basic elements of integrity in the thin sense possible? Most of this book is given to answering this question.
I argue for a specific conception of the integrative function of practical reason as it applies to the agent of integrity in the thick sense and consider alternative conceptions. In this regard, it is important to distinguish three major competitors, one of which is associated with Aristotle and two of which are associated primarily with Kant.
The first I call simply the Aristotelian conception of the integrative function as it applies to the thick conception of integrity, with the understanding that the conception is Aristotelian in spirit and not in historical detail. The conception defended here is clearly revisionary in regard to its Aristotelian origins, as is much of Kantian internalism in regard to its Kantian origins. It reflects an inclusive-ends view of eudaimonia (the good for a human) and includes the following two claims:
a. The rational grounds for practical judgments are those multiple goods in terms of which the agent finds life meaningful from his or her own point of view. (This does not mean that practical reason aims at eudaimonia, but at the goods that make up the life of eudaimonia for the particular agent.)[5]
[5] . Eudaimonism, like Kantianism, is very easy to caricature. It is also subject to different contending forms. If we think of hedonistic eudaimonism as the paradigm, we are bound both to be misled and to overlook the most promising forms. Most important, we are likely to misunderstand the best eudaimonistic account of practical reason. Korsgaard has attempted to show that Aristotle and Kant have very similar views on the role of reason and reflectiveness in practical reason. The eudaimonistic account offered here is meant to be plausibly Aristotelian in both form and content and different in significant ways from Kant. For Korsgaard's account, see Christine M. Korsgaard, "From Duty and for the Sake of the Noble," in Aristotletotle, Kant, and the Stoics: Rethinking Happiness and Duty , ed. Stephen Engstrom and Jennifer Whiting (New York: Cambridge University Press, 1996), 203–36.
b. None of the regulative norms of such a conception are asymmetrical in their regulative functions.
Several comments on (a) are in order. First, although scholars debate whether Aristotle endorsed an inclusive-ends conception of eudaimonia or a dominant-end conception (which emphasizes the highest value of contemplation),[6] I do not pursue this issue here: first because this is not a scholarly book on Aristotle and second because the implications of my argument clearly rule out the dominant-end conception.[7] Hereafter, then, when I refer to the Aristotelian conception I am referring to the inclusive-ends view.
Second, some clarification is necessary of how considerations of eudaimonia enter into the deliberations of a practically rational agent. On what is to my mind a rather crass view, Aristotle's agent simply asks what is conducive to his or her happiness, and means-ends reasoning ensues. Interpreted narrowly, a passage from book 1 of the Nicomachean Ethics might seem to support this. There Aristotle says:
[6] . W. F. R. Hardie was the first to make the distinction between the inclusive ends and the dominant end interpretations of Aristotle, in W. F. R. Hardie, "The Final Good in Aristotle's Ethics," in Aristotle: A Collection of Critical Essays , ed. J. M. E. Moravcsik (Garden City: N.Y.: Anchor Books, 1967), 297–322. Leading defenders of the dominant end view are David Keyt, "Intellectualism in Aristotle," in Essays in Ancient Greek Philosophy, ed. John P. Anton and Anthony Preus (Albany: State University of New York Press, 1983), 364–87; and Richard Kraut, Aristotle on the Human Good (Princeton: Princeton University Press, 1989). Advocates of the inclusive ends interpretation are J. L. Ackrill, "Aristotle on Eudaimonia," in Essays on Aristotle's Ethics , ed. Amelie Oksenberg Rorty (Berkeley: University of California Press, 1980), 15–34; Sarah Broadie, Ethics with Aristotle (London: Oxford University Press, 1991); and Anthony Kenny, The Aristotelian Ethics (London: Oxford University Press, 1978) and Aristotle on the Perfect Life (London: Oxford University Press, 1992).
[7] . The Aristotelian conception interpreted as involving a dominant end would include the following two claims:
a. The grounds for practical judgments are those multiple goods in terms of which the agent finds life meaningful from his or her point of view, and among these goods is one that functions as a dominant end. (Again, this does not mean that practical reason aims at eudaimonia, but at the goods that make up the life of eudaimonia for the particular agent.)
b. Though the norms grounded in subdominant ends might be symmetrical in their regulative functions vis-à-vis each other, the norms grounded in the dominant end are asymmetrical in their regulative functions vis-à-vis the norms of the subdominant ends.
For our present purposes, we may draw the conclusion from the preceding argument that happiness is one of the goods that are worthy of honor and are final. This again seems to be due to the fact that it is a starting point or fundamental principle, since for its sake all of us do everything else. (NE 1101b:351102a:4)[8]
However, I think if we consider book 1 more carefully, we need not get the crass view.
One of the primary functions of book 1, I believe, is to get clear on the sense of good relevant to ethical inquiry. When we find clarity on this issue, we get a much more plausible view of how eudaimonia enters into an agent's practical deliberations, and this is important to how I am interpreting the Aristotelian scheme.
The first thing Aristotle wants to establish is that the sense of good relevant to the study of ethics is one that must be relevant to practical reason. Consider in this regard the following passage that begins at Nicomachean Ethics , 1096b:53. He says:
Perhaps one may think that the recognition of an absolute good will be advantageous for the purpose of attaining and realizing in action the goods which can be attained and realized. By treating the absolute good as a pattern, [they might argue,] we shall gain a better knowledge of what things are good for us, and once we know that, we can achieve them. This argument has, no doubt, some plausibility; however, it does not tally with the procedure for the sciences. For while all the sciences aim at some good and seek to fulfill it, they leave the knowledge of the absolute good out of consideration. Yet if this knowledge were such a great help, it would make no sense that all the craftsmen are ignorant of it and do not even attempt to seek it. One might also wonder what benefit a weaver or a carpenter might derive in the practice of his own art from a knowledge of the absolute Good, or in what way a physician who has contemplated the Form of the Good will become more of a physician or a general more of a general. For actually, a physician does not even examine health in this fashion, he examines the health of man, or perhaps better, the health of a particular man, for he practices his medicine on particular cases.
The central point here is that even if there is a good in the absolute sense in which Plato asserted, its relevance to ethics is questionable in the same
[8] . Aristotle, Nicomachean Ethics , trans. Martin Ostwald (Indianapolis: Library of Liberal Arts, Bobbs-Merrill, 1962). Hereafter, all quotes from the Nicomachean Ethics are from the Ostwald translation.
way that its relevance is questionable to the weaver, the carpenter, the general, and the physician. Why? Because the crafts and ethics are concerned with how to live and act, rather than with what is true in the Platonic sense. Thus the relevant sense of good must be one that guides or is capable of guiding practical reason. This is to say, in Barbara Herman's terms, that the sense of good relevant to the study of ethics is the one that applies to things that appear within our deliberative field, that present themselves to us as things to be pursued, cherished, nurtured, maintained, respected, loved, and so on.1[9] Being clear on this, however, only tells us something about what we are looking for in the relevant sense of good. We need to know much more.
The comments on the physician are especially important. Aristotle seems to express some ambivalence about whether the sense of good should be relativized to the individual, as in the physician's case, or should be taken as good for "man" in the sense of the good for humanity. Actually, I do not believe that it is either of these, for there is another possibility. Consider the difference between the good for humanity, the good for this particular human, and the good for this particular kind of human, for example, the ideal Athenian. Taking the latter as a guide to interpretation, we can read Aristotle as rejecting as too broad (and thin) the conception of humanity as the proper subject of inquiry and as too narrow the study of some particular human. The question is how to specify the subject matter in a way that is neither too broad and thin for substance nor too narrow and particular for purposes of generalization.[10]
I take it that the ideal Athenian, for us, fails on the latter grounds. Still, we can take the reference to the ideal Athenian in another way. We can take it to mean that any study of ethics that yields any substantive results will always be relative to a way of life and to the character of those for whom that way of life is in some sense a natural expression of who they are at the core. In this sense, the person with integrity on the thick conception is the subject of the current study, rather than the ideal Athenian. And by taking the thick conception as the subject, we both leave behind some objectionable features of Aristotle's view and remedy some deficiencies. The views
[9] . Herman, The Practice of Moral Judgment , 152, 166, 168, 172, 179, 180, 181, 182–83, 191, 193–94, 196–202.
[10] . Aristotle shows some sensitivity to this concern at NE 1097B-6. There he says, "We do not mean a man who lives in isolation, but a man who also lives with parents, children, a wife, and friends and fellow citizens generally, since man is by nature a social and political being. But some limit must be set to these relationships, for if they are extended to include ancestors, descendants, and friends of friends, they will go on to infinity."
on slavery and gender are left behind, and added are the concerns of respect and sympathy for those who are not in any important sense closely connected to us. The Aristotelianism I defend here, therefore, is clearly a revised version. Unlike neo-Kantians, I am more inclined to revise Aristotle to accommodate impartial norms than I am to think that Kant can be understood or revised in a way to accommodate partial norms. The project of Kantian internalism is not, as I understand it, to be perfectly true to Kant's own project but to construct from some understanding of the categorical imperative a conception of the personal life that is rich and robust but also appropriately demanding in terms of impartial respect for self and others as rational agents.[11]
Similarly, the current project of Aristotelian internalism is not to be perfectly true to Aristotle. Aristotle was wrong about all sorts of things, as was Kant. So I will not be defending Aristotle as a disciple of some sort, devoted to showing that the master had things right. Rather, I will be constructing a conception of what it is to be a person of integrity in the thick sense that runs in the opposite direction of the Kantian analysis. Rather than fit the analysis of partial norms within the context of the categorical imperative with its perfect and imperfect duties, I argue that it is best to fit the concerns for those who are not closely related to us within a conception of our own good in the sense that their well-being is central to the meaningfulness of our own lives when we are persons of integrity in the thick sense. Much more will be said as I proceed.
The relevant sense of good, then, is one that can guide practical reason and is relative to a character of a certain sort and to the way of life expressive of that character. But we need to know more about how this sense of good enters into the deliberations of the agent in the form of eudaimonia. Here we must see how the criteria of finality and self-sufficiency function and in what sense the Aristotelian agent (as constructed here) aims at eudaimonia. I think the best way of understanding Aristotle is that practical reason aims at eudaimonia only in the sense that practical reason is guided by a sense of good that raises the issues of finality and self-sufficiency in its evaluation of how to act and to live. About finality, Aristotle says:
What is never chosen as a means to something else we call more final than that which is chosen both as an end in itself and as a means to something else. What is always chosen as an end in itself and never
[11] . The most developed account of this sort is by Nancy Sherman in her book, Making a Necessity of Virtue: Aristotle and Kant on Virtue (Cambridge: Cambridge University Press, 1997).
as a means to something else is called final in an unqualified sense. (NE 1097a:30-35)
And about self-sufficiency, he says:
For the present we define as "self-sufficient" that which taken by itself makes life something desirable and deficient in nothing. (NE 1097b-15)
I construct Aristotle's scheme to mean that one aims at eudaimonia in the sense that one evaluates actions and their place in life in terms of the criteria of finality and self-sufficiency and that this is not to aim at some mental state (or any other state) called "happiness."[12] On this construction, finality and self-sufficiency are criteria to be employed in practical reasoning itself; they are not simply criteria employed in philosophical debate about the ultimate goal of life. Whether Aristotle actually meant the criteria to be employed in this way is a matter of unimportance to my project.
What does all this come to in regard to the agent of integrity in the thick sense? In terms of finality, it means that the agent experiences life in a way that many things appear within his or her deliberative field as good. That is, they are goods that are relevant to practical reason, things to be pursued, cherished, nurtured, maintained, respected, and loved, and they are, as such, valued as ends. However, they are final only as they appear as ordered within a life as a whole, for it is only from the viewpoint of life as a whole that the issue of finality can arise. In this regard, to aim at eudaimonia is nothing more than to attempt to see the various goods of life as ordered in a way that the choice of the life in which they appear is chosen for itself and for no further end. To the extent to which a life with its goods is ordered in a way that meets the criterion of finality it is practically rational. Why? Because it is to that extent guided by the relevant sense of good. The other criterion is self-sufficiency, which operates a bit differently than finality. Whereas finality requires that the goods be ordered in a way that makes them the proper object of final choice, self-sufficiency requires that all the goods get into the ordering, if at all possible. Nothing of importance can avoidably be left out. In this regard, to aim at eudaimonia is nothing
[12] . Aristotle seems to mean one thing by self-sufficiency in book 1 of the Nicomachean Ethics and another thing in book 10. In book 10 , he emphasizes the notion of independence rather than the notion of completeness when he provides the divinity argument for the highest value of contemplation. The idea is that the gods, who spend their time in contemplation, are invulnerable to supporting conditions for their way of life in a way that other creatures are not. Since appeals to divinity will play no positive role in my argument or the Aristotelianism I construct, I will not employ this notion of self-sufficiency.
more than the reflective concern that every good thing or as many kinds of the most important good things as possible get a place in life. Again, it is not to aim at some state called "happiness." To aim at eudaimonia, then, is simply to employ as reflective criteria the criteria of finality and selfsufficiency in the evaluation of how to live and to act, given that we value some things as ends.
As I employ these concepts, however, I want to make four clarifications: one having to do with the concept of eudaimonism itself; one, with the criterion of finality; one, with the concept of evaluating from the perspective of a life as a whole; and one, with self-sufficiency.
Beginning with eudaimonism, I want to distinguish between what I call subjective eudaimonism, on the one hand, and objective eudaimonism, on the other. The distinction between the two can best be made out in terms of Aristotle's definition of eudaimonia, that is, activity of the soul in accordance with virtue, and how he understands the status of external goods in the life well lived. According to what I call the subjectivist interpretation, external goods are not a part of one's well-being but are the equipment one needs for living the life of virtue. Perhaps the best expression of this view is found in Richard Kraut's book, Aristotle on the Human Good . His interpretation of Aristotle is that external goods are not a part of one's wellbeing but are the equipment for acting virtuously. In some cases, this makes sense. Wealth is the mere equipment whereby the generous person can act magnanimously. But what about the people who benefit from such generosity? Are they merely equipment needed for virtuous activity? Kraut's answer is that they are not, but insofar as they are valued for themselves, they do not reflect on the well-being of the generous person. Rather, concern for them reflects an altruistic attitude toward others. A similar analysis is given of the virtues of friendship and justice. On this view, that things go badly for other people is not something that makes life go badly for the agent, except insofar as things going badly for others lessens the occasions for one's virtuous activities. Thus to say that eudaimonia is activity of the soul in accordance with virtue is to assert that there is an identity relationship between eudaimonia and some exercise of the self. Since external goods are not parts of the self or its activities, they are not components of eudaimonia, the life well lived.
According to the objectivist construction of eudaimonia that I want to defend, some external goods, most notably friends, family, fellow citizens, and others, are intrinsic constituents of eudaimonia itself for creatures like us. On this understanding, the definition of eudaimonia is best understood as activity of a psychology (soul) in accordance with virtue and the goods
to which that virtue is attached. The virtue of justice attaches us to persons worthy of respect; the virtue of sympathy, to persons in need; and the virtues of friendship and parental love, to our friends and children. Our character, on this Aristotelian construction, attaches us to items in the world and to the intrinsic well-being of those items. If this is true, then when things go badly for those items, things go badly for us. When asked how his life is going, the virtuous person's answer will often be couched in terms solely related to the well-being of others. Contrast in this regard these responses: Things are going badly because I have a headache, I have nothing virtuous to do, my child is ill, my friend's home was destroyed in a fire, workers are being cheated out of their pay, people are starving in Africa. Notice that many of the responses on the list do not mention the self but others, but they are mentioned in the context of why life is not what it could be. Also contrast these positive responses to the same question: Things are going really well because I feel great, my child just had her first recital, my friends and I played bridge last night, the workers strike has been settled in a way that is fair to all sides, and the drought has ended in Africa. These responses, both negative and positive, are intrinsic to what it is to thrive as a social being. Now, when eudaimonia is understood in this way, the self does not dominate the understanding of how well life is going in the way that it does on Kraut's understanding of Aristotle. And it should be noted that eudaimonia on this view involves vulnerability because of the way in which one's character attaches one's well-being to items in the world that are external goods. When Aristotle says that we do not deliberate about external goods, part of what that means on this interpretation is that social beings bring certain values to the task of practical reason and with these values in place, they ask, How am I to live in a way that best accommodates these goods?[13]
The criteria of finality and self-sufficiency are brought to the task of answering this question understood on the objectivist reading of eudaimonia. Regarding finality, it is important to keep in mind that it can come in degrees. Among alternative ways of life open to an agent, none of them might satisfy the criterion perfectly. In such cases, we can speak of the way of life that is most final, the one that is most chosen for itself as an end and not as a means to anything else. This clarification accommodates the fact that
[13] . Aristotle is usually taken to be a perfectionist, but the Aristotelianism I construct here is pluralistic in that there are goods other than perfection. For a contrasting, perfectionist construction of Aristotle, see Thomas Hurka, Perfectionism (New York: Oxford University Press, 1993).
none of us has among his or her alternatives the perfect way of life. Also, I will understand the criterion of finality as having two aspects: the first applies to a life as a whole, that it be chosen for itself and for no further end; the second applies to the goods within the life that is chosen for itself. This second aspect of finality requires that if something appears within one's deliberative field as good as an end, then it is rational, if at all possible, to order one's priorities within a way of life to accommodate that good as the kind of end it is taken to be within that deliberative field. The second aspect of the finality criterion is not met, then, if one's priorities do not properly recognize a good for the kind of good it is within one's deliberative field. This can occur in a variety of ways: by misconstruing the value of a good as not good at all, by misconstruing the value of a good that is an end as a means only, or by misconstruing the kind of end a good is. This aspect of the finality requirement is such that a failure to meet it means a failure to meet the self-sufficiency requirement as well. The difference between the second aspect of finality and self-sufficiency is that in the former case the value of some good is misconstrued; whereas in the latter a good is missing, either because its value has been misconstrued or because it simply is not there. To take a friend for granted is one thing; not to have a friend is another. But in neither case is there a life that is chosen for itself and self-sufficient.
Regarding the evaluation from the perspective of a life as a whole, it needs to be recognized that we do not have the kind of access to our lives that allows us in any literal way to evaluate from this perspective. We have only a relatively vague notion of what our lives will turn out to be as a whole. This is why we should understand this perspective as simply requiring us to place our choices, at least the most fundamental ones, within the context of how our priorities fit within a way of life, and we do have some idea of what this is. The contrast is a decision model the rationality of which is defined independently of a concern for how things fit within a way of life at all.
Finally, there are two concepts of self-sufficiency that should not be confused. The first, which is the one employed here, is the idea that all the goods for a meaningful life are included. The second is that the kind of life that includes all the goods in the first sense is free of the contingencies of moral luck. Both the Stoics and Aristotle thought it rational to pursue a life that was self-sufficient in the first sense, but they differed on whether the life of eudaimonia was self-sufficient in the latter sense.[14] Because he thought ex-
[14] . There is a dispute about this among Aristotelians. Kraut takes Aristotle to put a great deal of emphasis on invulnerability on his intellectualist interpretationand takes himself to be disagreeing with Martha Nussbaum. See Kraut, Aristotle on the Human Good , and Martha Nussbaum, The Fragility of Goodness (Cambridge: Cambridge University Press, 1986). For a philosophical defense of the notion of vulnerability as it applies to human dignity, see my Dignity and Vulnerability: Strength and Quality of Character (Berkeley: University of California Press, 1997).
ternals were intrinsic goods of the best life, Aristotle thought that we need a modicum of good luck for good living. Because they identified the best life with untroubledness (ataraxia), the Stoics rejected the value of externals and identified the best life with the life of virtue and hence immune to the forces of luck. For the Stoics, the goods of virtue are all under our control, immune to luck, and are all the goods there are in a life well lived. As will become clear, I side with Aristotle. It is only in understanding the fact that many of us are creatures with a character full of caring about externals that practical reason is made what it is for us.
To sum up, then: The Aristotelian conception first involves the appearance of things that are good (in the relevant sense) within an agent's deliberative field. The agent then considers different sets of priorities as candidates for accommodating those goods. Each set of priorities is considered as the basis for the imaginative projection of a way of life and how the things thought of as good by the agent appear within that way of life. Finally, the criteria of finality and self-sufficiency are employed to evaluate which set of priorities is rational. To the extent to which a set of priorities meets the criteria of finality and self-sufficiency it solves the integration problems of the agent. All this assumes that it is the appearance of things that are meaningful within the agent's deliberative field that generates the need for an integrative function in the first place. It makes no assumption about happiness as the goal of one's deliberations.
In contrast with the Aristotelian conception of the integrative function of practical reason are two impartial conceptions. The first I call externalist impartialism, which includes the following two claims:
a. There is a dominant impartial norm that is asymmetrically regulative of any other norms within consciousness, though there are other subdominant norms that are symmetrical in their regulative functions vis-à-vis each other.
b. The rational grounds for the dominant, asymmetrical norm are independent of those goods that make life meaningful from an agent's own point of view.
On this conception of impartialism, the demands of practical reason are pure in the sense that they are in no way dependent on the psychological
attachments of the agent. Shelly Kagan endorses a view of this sort.[15] But also there is a plausible interpretation of Kant that takes this line, for there are passages from Kant, as I note later, that suggest we have the duty not to commit suicide, even if we do not find life meaningful at all. On any view of practical reason, whether Kantian or non-Kantian, that takes this form, there is a radical distinction between the demands of practical reason and a particular agent's own good. The Kantian version of externalist impartialism I call traditional Kantianism.
The second conception of impartialism I call internalist impartialism, and it includes the following two claims:
a. There is a dominant impartial norm that is asymmetrically regulative of any other norms within consciousness, though there are other subdominant norms that are symmetrical in their regulative functions vis-à-vis each other.
b. The rational grounds for the agent's norms, including the dominant, asymmetrical norm, are the goods that make life meaningful from an agent's own point of view.
In contrast to traditional Kantianism, Barbara Herman, Marcia Baron, Christine Korsgaard, Henry Allison, Nancy Sherman, and David Cummiskey (all of whom I discuss later) defend versions of internalist impartialism. It is important to note that internalist Kantianism is a conception of practical reason that bridges some of the gap between the Aristotelian conception and traditional Kantianism. On this view, impartial respect for persons and their rational nature is still a dominant impartial norm, but it functions within consciousness as a dominant good apart from which no rational agent would find life meaningful from his or her own point of view with the basic elements of integrity in the thin sense intact. Later, I point out advantages of internalist Kantianism over the traditional variety. But it is the central thesis of the book that the thick conception of integrity requires the Aristotelian (inclusive-ends) conception of an integrative function of consciousness. The negative (as opposed to the positive) thesis of the book can be stated as follows: Against traditional Kantianism, I argue that all norms have their foundation in the goods that make life meaningful from the agent's own point of view. The argument for this occurs primarily in part 2, where I show that the goods of respect, as Kantian internalists insist, must be given an internalist account. If traditional, externalist Kantianism cannot account for the role of respect in our lives, it stands no chance of ac-
[15] . See Shelly Kagan, The Limits of Morality (Oxford: Clarendon Press, 1989).
counting for other goods. The remainder of the book is given to an argument against internalist Kantianism, where I argue that no conception of practical reason that employs an asymmetrical regulative norm can solve the integration problems of the agent of integrity in the thick sense.
III.
The basic contrast of this book, then, is between two conceptions of practical reason. As best I can, I want here to provide a brief sketch of the contrast. Herman has argued that one requirement of an adequate conception of morality is that the rightness of an action must be the nonaccidental result of its motive. This she (rightly) takes to be one of Kant's most fundamental points. If we combine this requirement with the claim that morality is based on practical reason, then the rightness of an action is both (i) the nonaccidental result of its motive and (ii) practically rational, all things considered. All Kantians, I think, would agree. Moreover, on the interpretation given here, the Aristotelian would as well. The difference between the Kantian and the Aristotelian conceptions of practical reason is in how they account for these two requirements and in what Korsgaard has called reflective endorsement.[16]
Consider first the Kantian account. Here some comments by Herman are helpful. She makes an illuminating distinction between the end or object of an action and the motive for an action. The end or object of an action, she says, "is that state of affairs the agent intends his action to bring about."[17] About motives, she says, "The motive of an action, what moves the agent to act for a certain object, is the way he takes the object of his action to be good, and hence reason-giving."[18] Now consider how this distinction might shed light on how the rightness of an action could be the nonaccidental result of its motive and rational, all things considered.
Imagine a case in which three people—A, B, and C—all do the right thing from different motives—A from narrow self-interest, B from natural sympathy, and C from the Kantian sense of duty. Let us assume that the right thing to do in the context is to render aid to someone in distress. A takes the object of his action to be good as a means of promoting his own narrowly self-interested goals, and it is in this sense that he sees the object as giving him a reason for action. Perhaps the person in distress is someone
[16] . See Korsgaard, The Sources of Normativity , 49–89.
[17] . Herman, The Practice of Moral Judgment , 25.
[18] . Ibid.
likely to benefit him in some way. Though his action is right, its rightness is not the nonaccidental result of his motive. B takes the object of his action to be good simply because it relieves the distress of someone in need, and it is in this sense that it gives him a reason for action. There is no further motive of self-interest. However, the rightness of B's action is not the nonaccidental result of its motive. This is because B takes the fact that his action relieves the distress of someone in need as a sufficient condition for its goodness. But surely this is not a sufficient justification. For in some circumstances relieving the distress of someone might be wrong because of other considerations. If this is true, then B's action, while right, is not the result of full rationality, that is, rational, all things considered. It is this failure that shows that B's action is not the nonaccidental result of its motive. C, on the other hand, takes the object of his action as good because he believes that relieving distress in the circumstances is the right thing to do. This is because he believes that doing so is required by the categorical imperative, that is, rational, all things considered from an impartial point of view. Unlike A and B, the rightness of C's action, Herman claims, is both the nonaccidental result of his motive and rational, all things considered.
No doubt, the Kantian view is a very powerful one, one not to be taken lightly or dismissed with caricature. How does the Aristotelian view, as I construe it here, differ? The major differences are these. First, full rationality, on the Aristotelian view, is not achieved by an impartial decision procedure, and second, there are no norms, on the Aristotelian view, that are asymmetrical in their regulative functions. On the Kantian view, the dominant norm of practical reason is the concern that one's deliberations take a certain form, namely, the impartial employment of the CI procedure (categorical imperative decision procedure). It is the fact that deliberations take this form that guarantees that the rightness of an act is the nonaccidental result of its motive. But on the Aristotelian view there is no such procedure. Rather, norms are the various ways an agent has of caring about himself or herself and others and other things in the natural and social environment. Hence, on this view, norms are at once both psychological and ethical. How, then, does the Aristotelian view guarantee that the rightness of action is the nonaccidental result of its motive? Here the concept of integrity plays a crucial role. The most general answer is that the character of an agent of integrity is such that any form of caring is influenced and balanced by the need to make a place for the other forms of caring indicative of the agent's character. Different things appear as good within the agent's deliberative field because the agent cares about a variety of things in a variety of ways. Practical reason, then, is not the capacity to employ an impartial decision pro-
cedure. Rather, it is a complex capacity that assists a psychology in its movement toward the equilibrium of integrity. This involves not only means-end reasoning but other things as well. Among them is the capacity for meriological analysis (the ability to relate parts to wholes) and the exercise of imagination. In order to gain an intuitive understanding of this, consider the agent of integrity in the thick sense.
The agent of integrity in the thick sense is one who cares in a variety of ways, including those of self-interest, natural sympathy, and impartial respect for others. It is the fact that such an agent is both sympathetic and respectful that he is not narrowly self-interested. Hence, A (above) reasons badly because his sense of self-interest is not regulated by other concerns. The Aristotelian self-interested agent could not see rendering aid as good simply because it serves some narrow self-interest. Why? Because his character is such that he cares about things other than himself, and his practical reason is guided by a sense of good that employs the criterion of selfsufficiency. Remember that self-sufficiency is the concern that all goods that appear within an agent's deliberative field be included within a way of life. Moreover, this concern is at once both psychological and ethical. Thus, his concerns, guided by the criterion of self-sufficiency, do not allow him to reason egoistically, because the objects of his actions are not seen by him to be good in a way that allows for such practical reasoning. Not only is it false that actions that render aid have only instrumental value for him; they are intrinsically valued as parts of the life most worth living. This consideration reflects the criterion of finality. The Aristotelian agent's practical reasoning, then, is one in which he is moved by the imaginative projection of a life of a certain sort. Why is it that he is moved by such a thought? Because of his character, the kind of person he is, one who is caring in a variety of ways and whose evaluations are guided by a certain sense of good. Thus he does not act for the goal of eudaimonia in the sense of pursuing happiness but for those goods envisioned within a way of life in which they are most meaningfully secured and balanced. It is his character that makes the vision both intrinsically alluring and rational. It is not the thought that these goods are means to the life most worth living.
Now consider the Aristotelian conception of a naturally sympathetic agent. The Aristotelian agent would not see the case of relieving distress as good simply because it serves some narrow self-interest. Nor would the Aristotelian sympathetic agent see rendering aid as good simply because it relieves distress. He would be concerned that it also be consistent with all the other things with which he is concerned. That is, he would see it as good
because it relieves distress and it is consistent with the respect, love, and other concerns he has. Again, this concern reflects practical reason guided by a certain conception of good. But the concern would not be a concern that the agent's deliberations take a certain form, as on the Kantian view.[19] Rather, it would be entirely because of the substance of what the agent is concerned about, guided by the criteria of finality and self-sufficiency. In both the case of self-interest and the case of natural sympathy, the rightness of the Aristotelian's action would be the nonaccidental result of its motive. This is because the Aristotelian's normative conception of self-interest and sympathy are regulated by the other norms of the agent's psychology. By this I mean that what such an agent sees as in his or her interest or as sympathetic is shaped by other considerations. Thus, on the Aristotelian view as understood here, it is substance, not form, that regulates substance. Moreover, the norms of this psychology are such that they are all symmetrical in their regulative functions—impartial respect notwithstanding.
The contrast between the Kantian and Aristotelian views should now stand out in greater relief in the way that they account for both the requirement that an act be the nonaccidental result of its motive and the requirement that a right act is rational, all things considered. On the Kantian view, agents reason from an impartial perspective and the CI procedure, and the capacity for such reasoning governs their natural inclinations, which might otherwise lead them to do the wrong thing. This perspective employs an asymmetrical regulative norm and informs agents of how their characters are to be formed. On the Aristotelian view, the reasoning of an agent is shaped by the fact that an agent has a certain core character, one that cares in a variety of ways about himself or herself and other things in the natural and social environment and is guided by a certain sense of good. This is what it means to assert that practical reason is character-relative. It is the imaginative projection of a life shaped by a set of priorities that ultimately determines how the agent sees the object of his or her action as good and reason-giving, and it is the fit between vision and character that ultimately constitutes full rationality. If we apply these thoughts to the agent of in-
[19] . Later I will consider a distinction between procedural constructions of Kant and nonprocedural, or substantive, constructions. I take Sherman to suggest a nonprocedural understanding of the categorical imperative, whereas Korsgaard endorses a procedural view. Allen Wood also defends a procedural understanding in "The Final Form of Kant's Practical Philosophy," Kant's "Metaphysics of Morals," Spindel Conference 1997, Southern Journal of Philosophy 36, supplement (1998): 1–20.
tegrity in the thick sense, we will get, or so I argue, a better account of the rightness of actions. Nothing about such projection, however, requires an impartial decision procedure, and I take it as a requirement of the Aristotelian view to show how this is possible.
I also argue that the Aristotelian view gives us a better account of reflective endorsement than does the Kantian account recently defended by Korsgaard in her Tanner Lectures. There she says:
"Reason" means reflective success. So if I decide that my desire is a reason to act, I must decide that on reflection I endorse that desire. And here we run into the problem. For how do I decide that? Is the claim that I look at the desire, and see that it is intrinsically normative, or that its object is? . . . Does the desire or its object inherit its normativity from something else? Then we must ask what makes that other thing normative, what makes it the source of a reason. And now of course the usual regress threatens. What brings such a course of reflection to a successful end?[20]
The Kantian solution to the regress problem entails a view of what full practical rationality is in terms of a conception of what, on its view, is the widest possible reflective endorsement. Complete reflective endorsement is reached on this view by the employment of the CI procedure. This means that the widest possible reflective endorsement is from the impartial point of view of respect for self and others. On the Aristotelian view, the widest reflective endorsement is from the point of view of a life as a whole and how that life accommodates all the goods found by the agent to be intrinsically important (finality and self-sufficiency). This means that one cannot accept the results of the CI procedure until one can see how those results factor into a life as a whole, a way of life. It also means that should those results fail to integrate within a way of life the goods that appear as ends within the agent's own deliberative field, then acting from the point of view of the CI procedure is neither fully reflective nor fully rational. I say more about this notion of reflective endorsement in the following chapters.
IV.
Part 1 is concerned with the thin conception of integrity and the general conception of practical reason appropriate to it. Chapter 1 is concerned with methodology, especially as it bears on the dispute between the Aristotelian conception of practical reason and Kantian internalism. There I consider
[20] . Korsgaard, The Sources of Normativity , 97.
views by Henry Allison, Barbara Herman, Christine Korsgaard, and John Rawls. My goal is to show that contrary to Allison and Korsgaard, the deepest division between the Aristotelian alternative and Kantian internalism is not that of differing conceptions of rational agency (though these differences are certainly there). Rather, the deepest differences are methodological. I argue that how we understand the internalism requirement regarding practical reason turns on our beliefs about rational inquiry. This in turn leads to a different general conception of rational agency. Chapter 2 provides a sustained argument that an adequate conception of practical reason for agents like us, or at least most of us, must consist of norms that are symmetrical in their regulative functions if it is to capture our considered moral judgments. Chapters 3 and 4 provide the general conceptions of integrity and practical reason that accommodate these functions. Parts 2 through 4, the bulk of the book, are given to a substantive account of integrity (the thick conception) and practical reason that I claim applies to a large portion of humanity. Part 2 gives a preliminary analysis of the role of impartial respect within the psychology of the agent of integrity in the thick sense and is designed to put traditional Kantianism aside and to prepare the way for the debate between Aristotelian and Kantian internalism. Part 3 addresses various partial norms involving different forms of personal love. The issue at each point is how to understand the regulative functions among the agent's norms in a way that solves the integration problems of an agent who is not only impartially respectful of others but also personally loving in a variety of ways. Finally, part 4 addresses further integration problems involving intrinsically valued activities and our concern with excellence. There I consider the role of the aesthetic dimensions of life in practical reason. In each of the parts devoted to the various goods, the analysis is intended to show two things: first, the ways in which these goods function as the grounds for both practical reason and the meaning of the agent's life from his or her point of view; and second, that the integration of these goods is achieved without norms that are asymmetrical in their regulative functions. This is to say that the analysis in parts 3 and 4 is intended to show that a thorough implementation of the integration test will reveal that the Aristotelian conception of practical reason is true regarding a large portion of humanity and that Kantian internalism is false.
I turn now to a discussion of the internalism requirement and a defense of the methodological significance of the integration test for rational inquiry concerning the nature of practical reason.
1—
The Internalism Requirement and the Integration Test
My concern in this chapter is with the terms of debate with Kantian internalism. Consequently, I will say little in argument against the externalist view. As I construe them here, Kantian internalism and Aristotelianism have several things in common. They both seek to provide a foundation for morality in practical reason, and they are both forms of internalism. As forms of internalism they seek to provide a foundation for morality within those goods that make life meaningful from a rational agents own point of view. They are united, then, in their opposition to externalism, the view that an agent can have reasons for action that are not founded in the goods that give life meaning. A significant difference between them, however, is in their conceptions of regulative functions. The Kantian view has it that impartial respect has an asymmetrical regulative function within practical reason, and the Aristotelian view has it that the norms of practical reason are all symmetrical in their regulative functions.
Another crucial difference is in their conceptions of rational agency. Henry Allison and Christine Korsgaard, both Kantians, have argued that the most fundamental difference between contemporary thinkers such as Bernard Williams, who belong to an Aristotelian/Humean tradition, and themselves is that they have rival conceptions of rational agency. I argue in this chapter that although there are fundamental differences about rational agency, we should attempt to find as much common ground as possible between the two traditions, including common methodological commitments that will allow us to adjudicate disputes about rival conceptions of rational agency. A great deal turns on what one considers a rational method of inquiry in this regard. I show that there are differences on this issue that affect how one sees internalism as a requirement for any adequate concep-
tion of practical reason and that differences about rational agency are often underwritten by these differences of method.
On the Kantian view, as understood by Allison and Korsgaard and by what I call the Kantian metaphysical school, rational inquiry about human agency proceeds a priori, but on the Aristotelian approach, it proceeds empirically. In turn, this difference in method leads to a different understanding of the internalism requirement. I argue that Korsgaard's claim that almost any moral theory passes the internalism requirement is a function of methodological commitments that we should reject. However, if we accept an empirical approach, the internalism requirement will lead to the integration test as the appropriate method of adjudicating disputes about human agency, and it is questionable that any current theory meets the internalism requirement on such a reading. The central thrust of the argument is that any conception of practical reason must assign a functional role to practical reason within an overall psychology and that these functional claims are best construed in a way that render them empirically falsifiable. This might seem to rule out a priori the possibility of Kantian internalism, but it does not. It does not because there is a difference among Kantian internalists about methodology. In addition to the metaphysical school of Kantian methodology, there is an internalist school of thought called Kantian constructivism. Members of the constructivist school want to distance themselves from Kant's methodology insofar as it is tied to his metaphysics. I argue that Kantian constructivists must accept an understanding of the internalism requirement that, when combined with their understanding of the function of practical reason, requires the integration test for adjudicating disputes about rational agency. This method, however, leaves open whether it is some constructivist version of Kantian internalism or Aristotelian internalism that passes the integration test. Thus the issue over the internalism requirement and the integration test turns not so much on different conceptions of rational agency as on different conceptions of rational inquiry about that topic.
1.
In Kant's Theory of Freedom , Henry Allison attempts to defend Kant against an attack by Bernard Williams, whose views have influenced the conception of practical reason defended here. One line of Allison's defense is to argue that Williams starts from a different conception of rational agency than Kant and that from this conception his criticisms logically fol-
low Allison then asserts that Kant's view "may be mistaken, perhaps even deeply mistaken . . ; but Williams has done nothing to show that it is."[1] I take this as a valid request for an argument and to imply the assumption that there is some test that can rationally adjudicate the dispute between these rival conceptions of rational agency.
This assumption naturally raises the issue of what the test might be, and it might be thought that one's conception of the test will turn on how one understands the internalism requirement. In a way this is true, but I will argue that how one understands the internalism requirement turns on one's methodologcal commitments.
A fundamental commitment of the Aristotelian/Humean tradition is that we should understand the capacities for practical reason as psychological capacities. Whatever else they are, they are features of our psychologies. Why have such a commitment, other than its inherent plausibility? One very plausible understanding of this tradition and its reason for thinking in this way is that it is only by thinking in this way that practical reason can be adequately studied. The thought is that the only way to gain any objectivity to claims about practical reason is to construe what practical reason is in such a way that disputes about it are rationally resolvable. This means that there is a fundamental tension between this tradition and divine command theories of morality and with intuitionist views that require special moral facts. The tradition also looks skeptically on any methodology that attempts to settle claims about rational agency without appeal to the empirical facts regarding human psychology. And it is this, rather than some dogmatic commitment to a conception of rational agency, that leads to skepticism about Kant's a priori conception of practical reason.
At another level, there is skepticism from this tradition about theories of practical reason that speak of rational agency per se rather than of human agency. How one is to study rational agency per se is not at all clear. One possibility is to gather all the known species of animals that we are inclined to call rational and conceptualize a notion of practical reason in light of the empirical facts of their psychologies. From the currently known facts, that would probably leave us with a study of human psychology. Most surely, that study would not include reference to psychologies that could not be empirically investigated, especially the psychology of God. This is why for Aristotle, considerations of neither beasts nor gods were of much influence
[1] . Henry E. Allison, Kant's Theory of Freedom (Cambridge: Cambridge University Press, 1990), 196.
in his conception of rational agency.[2] Of course, we would need a conceptual scheme, a philosophical psychology, that would allow us to map those facts onto a conception of practical reason and rational agency. The point of this tradition, however, is that the arguments for the categories of the philosophical psychology must be such that they show that those categories are both empirically accommodating and rather tightly connected to the study of humans. By "empirically accommodating," I mean that an argument for the categories of a philosophical psychology must be carried out in terms of how those categories allow us to comprehend the facts of our experience. It is this, I suggest, that motivates Hume's insistence on the necessity of desire for rational agency, and this is a methodological motivation.
My major points here about skepticism are three. First, the tradition of skepticism about pure practical reason is founded on a methodological skepticism that insists that claims about the essentials of human agency be in principle empirically testable. Second, the categories of any adequate philosophical psychology must be empirically accommodating. Whether we insist that motives include desires is one thing, but whether claims about motives include claims about our psychologies that are subject to empirical testing is another. And third, the kind of test that adjudicates between competing conceptions of rational agency must be carried out in regard to humans, rather than gods, beasts, or imaginary creatures far removed from human experience. Thus, skepticism about the Kantian claim that reason alone can motivate is skepticism about two things: (i) whether that issue has been conceptualized in a way that empirically accommodates the facts of our psychology and (ii) whether inquiry about that issue takes the right point of departure, namely, the study of ourselves.
This method leads quite naturally to the issue of what function practical reason is to serve within a psychology, which in turn leads to the thought that the appropriate test for a conception of practical reason can be carried out in terms of its functional claims. In what follows, I will consider in this
[2] . See Politics , 1253a. There is some room for debate on this point. For one view, see Martha Nussbaum, "Aristotle on Human Nature and the Foundations of Ethics," in World, Mind, and Ethics: Essays on the Ethical Philosophy of Bernard Williams , ed. J. E. J. Altham and Ross Harrison (Cambridge: Cambridge University Press, 1995), 86–131. For a somewhat different view of the role of god in Aristotle's ethics, see Christine M. Korsgaard, "From Duty and for the Sake of the Noble," in Aristotle, Kant, and the Stoics: Rethinking Happiness and Duty , ed. Stephen Engstrom and Jennifer Whiting (New York: Cambridge University Press, 1996), 203–36.
regard methodological implications of views of three prominent Kantians: Barbara Herman, Christine Korsgaard, and John Rawls.
2.
Consider first Herman's understanding of Kant and the nature of practical reason.[3] A brief analysis of the structure of her version of Kantian internalism as developed in The Practice of Moral Judgment will help in articulating the aforementioned integration test and its importance.
The key difference between Herman's understanding of Kant and the traditional understanding is that she thinks that the concept of value is more basic to Kantian theory than the concept of right action. Kant, on her view, gives priority to the "good" over the "right" within his conception of morality. It is this fact, when properly understood, that allows us to see that Kantian morality is not an alien force threatening to disrupt our lives and intrude on our integrity in the way that many claim it does on the traditional interpretation of Kant that begins with obligation.
The theory of value, however, does not function the way it does in a consequentialist theory. It does not serve as a means of ranking states of affairs from best to worst on impartial criteria. What, then, is its function within the theory as a whole? Traditionally, the other way in which value has played a central role in a conception of morality is as a measure of human well-being, of human flourishing, of human good. I have in mind the kind of agent-centered morality found in Aristotle, where the most fundamental issue is not what one ought to do but what is the kind of life most worth living. In this vein, Herman interprets Kant's treatment of the unqualified value of the good will as the most fundamental element in his conception of the well-being of any rational being as such. This being the case, respect for human beings as agents of pure practical reason is a regulative norm that shapes and limits the other projects within an agent's life. So construed, respect for others is not a factor external to the meaningfulness of an agent's life from his or her own point of view but the most fundamental value of a meaningful life for a rational being. Thus when our respect for others puts limits on what we can do for our loved ones, this respect is not an alien force external to what gives our lives meaning but something central to the meaning we assign to love and its place in life. Therefore, to
[3] . Allison's views are greatly influenced by Herman. See Allison, Kant's Theory of Freedom , 278 n. 49.
live a good life, a life of flourishing, for a rational being involves regulating the other values in one's life by respect for ourselves and others as agents of pure practical reason.
To be an agent of pure practical reason is both to deliberate in a certain way about the components of one's life and to have one's will determined by this type of deliberation rather than by the affective and conative capacities of our empirical nature. Our sentiments, interests, wants, and desires stand on a different level than the willings of pure practical reason. Indeed, our natural affective and conative capacities provide only incentives to action, never motives; only pure practical reason, which is a purely cognitive capacity, can provide reasons by what it does with the incentives of inclination. It is the decision procedure involved in the various formulations of the categorical imperative that constitutes the pattern of deliberation indicative of a person of good will. Following Herman, we can call this the CI procedure.
The subtleties of Herman's interpretation of the CI procedure are rich and worthy of detailed examination, an examination that is beyond anything I can attempt here. What I want to focus on is the general function of the CI procedure within the kind of moral theory she construes Kant's to be. If the CI procedure is to do the work she thinks it is designed to do, it will provide a solution to any of the integration problems of any rational being. It will organize all the elements that factor into the meaning of a person's life from his or her own point of view in a way that is consistent with respect for persons as agents of pure practical reason. Without this, it cannot accommodate the integrity of human beings and the complexity of that which gives their lives meaning. As applied to this complexity, the CI procedure governs the deliberations of an agent, since that agent is respectful of self and others. Moreover, the patterns of deliberation captured in the CI procedure are the rational standard for the integration of the other elements of a person's life: sympathy for others, personal love, the value of intrinsically meaningful activities, and so on. That is, it is the standard if the person is rational. This, I believe, is the guiding principle in how to interpret the demand of the procedure and how it works. And on this view, Kantianism is not as far from Aristotelianism in terms of the structure of moral theory as has been thought, even though, on Herman's view, deliberations about one's own good are regulated by an impartial decision procedure.[4] On
[4] . For further development of her views in regard to Aristotle, see Barbara Herman, "Making Room for Character," in Aristotle, Kant, and the Stoics , ed. StephenEngstrom and Jennifer Whiting (New York: Cambridge University Press, 1996), 33–62.
this view, practical reason is an integrative function of consciousness that relates to other features of consciousness somewhat as output function to input, where the output function is a purely cognitive faculty and our wants, needs, desires, and sentiments are inputs.[5]
Herman's view suggests that what shows that the Kantian conception of rational agency is correct is that it, and only it, will pass the integration test. That is, it is the only conception of rational agency that includes a conception of an integrative function that will actually integrate the various concerns of rational human agents in a way that preserves integrity. On this view, practical reason, interpreted as the capacity for employing the CI procedure, plays a certain functional role within the psychology of any rational agent: It regulates all the concerns embedded in the psychology of a rational being in a way that preserves integrity.
Now this seems to provide an excellent test. So construed, this conception of the function of practical reason has the very desirable feature that substantive claims about its content are falsifiable. Alternatives to the capacity for employing the CI procedure can be constructed and tested against each other in terms of their integrative success as applied to the psychology of actual human beings. Among these alternatives is the Aristotelian conception defended here. If the Aristotelian alternative makes more sense of the integrative capacities of actual human beings than the CI procedure on its most favorable Kantian interpretation, then Herman is wrong about the CI procedure. If, however, the best Kantian interpretation of the CI procedure achieves superior integrative success than all other candidates, she is right, and those of us who have opposed Kant can now embrace him without the need for faith or his metaphysics.
There are some deep questions, however, that must be raised about the CI procedure on this view. Herman wants to interpret the procedure in a way that steers a course between an interpretation of Kant that places moral demands on a foundation outside the factors that go into the meaningfulness of one's life from one's own point of view and a clearly Aristotelian view that does not involve an impartial decision procedure. The issue, then, is whether she can construct the procedure in a way that does not collapse either into the traditional conception of the procedure or into a form of
[5] . For the Kantian, of course, it is important that this output function is a feature of agency that can be combined with freedom of the will.
Aristotelianism that is not, for her, appropriately impartial. It is here that the concept of pure practical reason comes into potential conflict with the requirement that the procedure solve the integration problems of human agents.
I cannot see that it is a purely conceptual issue whether a pattern of deliberation will provide a solution to such problems for humans. This is surely in part an empirical issue, which illustrates the necessity of the integration test. Whether a certain way of attempting to integrate the kinds of concerns an agent has will in fact integrate them is in part an issue to be settled by empirical observation. Later, in chapter 3, we will see what behavioral responses will count as evidence that integration has failed to occur. The point here is that if Herman simply defines the procedure as whatever pattern will pass the integration test, she runs the risk of collapsing Kantianism into a more robust Aristotelianism than she desires. This would be true if impartiality and respect do not have the kind of role in the procedure one would expect on a Kantian view. On the other hand, if she defines the procedure independent of its ability to pass such a test and then it fails the test for large numbers of apparently rational humans, she has to say that they are not rational beings, which would reveal that the test is superfluous. And she would have to say this regarding any form of practical consciousness in which impartial respect is regulated by some other norm, partial or impartial. However, denying the status of rationality to such a form of practical reason reverts to the traditional, externalist view of Kant, for it imposes a conception of the meaningfulness of life on an agent from a source independent of that agent's own point of view. As long as she retains the notion of pure practical reason and the a priori cognitive psychology that goes along with it, I do not believe that she can avoid the dilemma. Moreover, I believe this applies to any version of Kantian internalism.
3.
My reasons for thinking that this dilemma applies to any Kantian internalism with strong a priori commitments arise in response to Korsgaard's influential arguments regarding internalism.[6] I show that Korsgaard must ei-
[6] . See Christine M. Korsgaard, "Skepticism about Practical Reason," Journal of Philosophy 83, no. 1 (January 1986): 5–25.
ther revise her understanding of the internalism requirement in a way that entails the integration test as a means of adjudicating disputes about rational agency or run the risk, as Herman does, of becoming an externalist. I will again show that the commitment to the integration test need not prejudice a priori the case for or against either Kantian or Aristotelian internalism. This, I believe, makes a good case for considering the integration test as a part of a methodology that provides common ground between Kantians and Aristotelians for adjudicating disputes about rational agency.
Kantian internalism includes the claim that reason alone can motivate, and according to Korsgaard, to be skeptical about this is to be skeptical about pure practical reason. The reason Kantians, whether internalist or externalist, feel compelled to claim that reason alone can motivate is that they see no other way of ensuring that morality is universally binding on all rational beings. Their reasoning is that as long as it is possible that a kind of motive is one that a rational being might not have, then it is not the kind of motive that can bind all rational beings. Moreover, there is, on their view, no necessary feature of rational agency attached to desires and sentiments. Their point is not only that there are no particular desires and sentiments that all rational agents have; it is also that not all rational agents even have desires and sentiments. It is not difficult, then, to see what drives the Kantian to a conception of practical reason in which reason alone can motivate. Nor, I take it, is it difficult to see why someone might be skeptical about this claim. After all, much of Kant's writings in moral philosophy are intended to remove such skepticism.
Korsgaard's argument is a defense of only one particular kind of basis for skepticism about pure practical reason or the view that reason alone can motivate. Specifically, her argument is a defense against the view that skepticism about pure practical reason follows from internalism itself.
The target of Korsgaard's argument is her understanding of a skeptical argument by David Hume and Bernard Williams. She distinguishes between content skepticism, which questions that reason alone can provide us with a mechanism for deciding what to do, and motivational skepticism, which questions that reason alone can function as a motive. She then claims that the latter has no independent force, which is to say that if it could be shown that reason alone can provide a decision mechanism, the claim that it could not motivate would itself have no philosophical motivation. Thus any rational skepticism can only take the form of content skepticism. The following is my formulation of the argument that her argument is designed to attack.
1. The internalism requirement: Practical claims, if they are really to present us with reasons for action, must be capable of motivating rational persons.[7]
2. For any rational person, P, P has a motive to do x only if P has a desire to do x or x is a means to satisfying some of P's desires.
3. Therefore, for any practical claim, R, that P do x, R gives P a reason to do x only if P has a desire to do x or x is a means to satisfying some of P's desires.
Korsgaard wants to assert that since the second premise is necessary to get to the conclusion, which constitutes the denial that reason alone can motivate, it is not the internalism requirement itself that leads to skepticism of pure practical reason. From this, she wants to assert that Williams's insistence on internalism itself does no work against Kant. It is only the additional claim involved in the second premise—that desire is necessary for motivation—that we get the skeptical conclusion, and this premise needs an independent argument.
My purpose here is not to defend Williams but to consider Korsgaard's argument that internalism itself does not entail skepticism about the issue of the motivational force of pure practical reason.[8] 1 believe that she is right that the internalism requirement itself does not preclude the truth of the claim that reason alone can motivate. But when the internalism requirement is understood from the perspective of a methodological commitment to the integration test, it requires that any functional claims about the role of practical reason within an overall psychology are falsifiable in terms of that test. To deny this, is to endorse a methodology that is not tightly connected to the study of ourselves. Hence how one understands the implications of the internalism requirement turns on one's methodological commitments. Thus both the claim that reason alone can motivate and its denial must be understood in a way that they can be put to the integration test. Moreover, I will argue that we should be committed to the integration test
[7] . This is Korsgaard's exact wording. See "Skepticism about Practical Reason," 11.
[8] . Williams's view is more complex than represented here. He has never meant by a desire a mere animal urge, as Allison puts it. And Williams has recently said that Kantian internalism is a limiting case of internalism. See Bernard Williams, "Replies," in World, Mind, and Ethics: Essay & on The Ethical Philosophy of Bernard Williams , ed. J. E. J. Altham and Ross Harrison (Cambridge: Cambridge University Press, 1995), 186–94.
as the appropriate method of adjudicating between rival conceptions of rational agency.
To see this, we have to recast the argument in a manner that reframes the issue to allow for the possibility of rationally resolving the question of whether reason alone can motivate.
Consider the following argument.
1. The internalism requirement: Practical claims, if they are really to present us with reasons for action, must be capable of motivating rational persons.
2. For any rational person, P, P has a motive to do x only if the consideration of x by P has some psychological pull with P for doing x.
3. Therefore, for any practical claim, R, that P do x, R gives P a reason to do x only if the consideration of x by P has some psychological pull with P for doing x.
Now it might seem that this is just a superficial attempt to evade the issue of desire versus reason as motivators, that "psychological pull" is just a synonym for "desire." But whether this is true depends on the conditions placed on assigning a truth value to the claim that a certain consideration has psychological pull with a rational agent. If Korsgaard is right that motivational skepticism turns on content skepticism, we should interpret the second premise in light of alternative conceptions of the content of practical reason. I propose that we understand the claim that reason can motivate with the claim that the CI procedure can have psychological pull with a human being. This defines practical reason in terms of a content, assuming that the CI procedure generates content, and a function, namely, an integrative function. But what about the second premise? We must understand this premise in a way that its meaning leaves open the possibility that the CI procedure can have psychological pull. This is to say that one way of understanding the meaning of (2) is in terms of what might make (2) true, and we understand this in terms of a certain content practical reason might have, that provided by the CI procedure, and the function it is to serve.
Of course, we must also understand the second premise in a way that leaves open the possibility that it is false that reason can motivate. How shall we understand, then, the denial of the thesis that reason alone motivates? Instead of formulating the most general logical denial, I propose we contrast the thesis with one among several ways in which that thesis might be de-
nied. Since our concern is with a dispute between Kantian and Aristotelian versions of internalism, we can say that the Kantian thesis is false if the regulative norms of practical reason are all symmetrical in their regulative functions. We can say this because the CI procedure requires that impartial respect is a norm within practical reason that functions asymmetrically in relationship to all other norms. This will be required of any content it generates. On the other hand, the Aristotelian alternative will be true just in case the norms of practical reason are all symmetrical in their regulative functions and the goods that give rise to those norms are those one would expect on an inclusive reading of an Aristotelian view, which is reflected in the thick conception of integrity.
Both alternatives, that reason alone motivates and that it does not, are defined in terms of content and function. Thus, in this sense, the issue of motivational skepticism turns on content skepticism, just as Korsgaard requires. But to be skeptical of content in regard to the CI procedure is to be skeptical that the content of the CI procedure can carry the burden of the integrative function of consciousness with regard to human beings. This is to be skeptical that the CI procedure will pass the integration test, that the CI procedure can provide an integrative function that preserves the elements of integrity for admirable human beings. Thus content skepticism is based on methodological skepticism.
How is the integration test to be administered? Kantian internalism is committed to the thesis that the CI procedure can motivate, which means that the CI procedure has a content and a function. To test this claim we need first some idea of what that content is, and then we need to put that content to a test to see if it can serve its designed function. That test is the integration test: Can the CI procedure serve the role of the integrative function of human consciousness in a way that preserves the elements of integrity for human beings of the sort reflected in the thick conception of integrity? I shall argue in parts 2 through 4 that it cannot.
I will construe the Kantian internalist claim that reason can motivate as the claim that the CI procedure cannot only have psychological pull for human beings but considerable pull. If the Kantian asserts that the CI procedure has no psychological pull for a rational being but nonetheless provides a reason for action, I will understand that to be a denial of internalism and an endorsement of externalism. Otherwise, it is difficult to see what would count as externalism. The considerable pull that the CI procedure must have is that it must play a certain role within the psychology of a rational being. Its pull must be the central integrative force of the psychology of a rational
human being. This does not mean that the agent cannot act against that force. It does mean, however, that if the agent does act against that force there are repercussions for the structure of the agent's psychology. I will say more about this when I come to chapter 3, the thin conception of integrity, and categorical interests. But it is important here to recognize that these observations underscore the threat of the previous dilemma facing Kantian internalism: If the integration test is rejected, externalism is waiting in the wings. It is waiting in the wings because the rejection of the integration test imposes a conception of practical reason on humans from a source external to their psychology, which includes their motivational set.
It is not, however, only the CI procedure that will be put to the integration test but the Aristotelian conception of practical reason as well. According to the Aristotelian conception, there is a clear path from the goods that make life meaningful for most human agents to the thick conception of integrity in which a network of practical norms are all symmetrical in their regulative functions. Thus applying the integration test will involve a detailed elaboration of the content of the Aristotelian model as it applies to the agent of integrity in the thick sense.
My argument, therefore, in no way depends on the claim that internalism itself settles the issue of whether reason can motivate. Moreover, it accepts the claim that motivational skepticism turns on content skepticism. The issue of content skepticism, however, turns on an empirical issue of whether the CI procedure on its best interpretation can play a certain role in human psychology: Can it integrate the various concerns a human agent has in a way that preserves the elements of integrity? If the relevant concepts can be made clear enough, the rest is a matter of empirical testing. I want to formulate the Kantian and Aristotelian theses so that they can be tested and to argue that the evidence tells against the Kantian conception and for the Aristotelian conception of practical reason for a rather large portion of humanity.
4.
Here it is important to consider Kantian alternatives. One possibility is for the Kantian to reject the methodology that leads to the integration test. I have no doubt that Kant himself would have taken this course and that it must be the course of the metaphysical school of Kantian methodology. According to the metaphysical school, as I understand it, we can derive a conception of rational agency from the concept of what it is to act for a reason.
This will include considerations of universalizability, consistency, impartiality, and the like. The point is that the rational method of formulating a concept of rational agency is an a priori method that relies in no way for its essentials on the facts of human psychology. Indeed, according to this methodology, we can learn psychology from moral theory. Korsgaard herself distinguishes between two types of internalism: "one that takes the psychological facts as given and supposes that we must somehow derive ethics from them in order to achieve an internalist theory, and one that supposes that metaphysical investigations—investigations into what it is to be a rational person—will have psychological conclusions."[9] On her view, Kant's thought is an example of the second, and I take it that hers is as well. When she says, then, that "the internalism requirement is correct, but there is probably no moral theory that excludes it,"[10] she is presupposing internalism that is tied to a certain methodological interpretation, and it is just that interpretation that is a stake in the dispute between the two traditions.
Thus the denial of the relevance of the integration test might rest on the Kantian doctrine that the rationality of pure practical reason is established a priori, the so-called fact of reason argument presented in The Critique of Practical Reason .[11] For metaphysical Kantians, this will mean that we can know that reason alone can motivate because we can know a priori that rational beings have certain obligations. This doctrine implies that the procedure determines a priori what our obligations are independent of whether we can as a matter of fact perform the actions they require. And since ought implies can , we can perform these actions. The syllogism is something like the following:
1. Necessarily, P ought to do x.
2. Necessarily, if P ought to do x, P can do x.
3. Therefore, P can do x.
It is the fact that the first premise is necessarily true on the Kantian view that prevents its falsification by showing that the consequent of (2) is false. This argument can then be employed to show that reason motivates by add-
[9] . Korsgaard, "Skepticism about Practical Reason," 23 n. 17.
[10] Ibid., 23.
[11] . See Immanuel Kant, Critique of Practical Reason , trans. Lewis White Beck (Indianapolis: Bobbs-Merrill, 1956), 5:30–31; Marcia Baron, Kantian Ethics Almost Without Apology (Ithaca: Cornell University Press, 1995), 44–45; Allison, Kant's Theory of Freedom , 230–49.
ing the premise that necessarily P can do x only if P has a motive to do x. Thus we can learn psychology from moral theory.[12]
It is difficult to see how to accept this argument as it stands without endorsing conclusions that I am sure neither Kant nor Korsgaard would accept. For example, the CI procedure is supposed to put some rather severe restrictions on breaking one's promises, perhaps not as absolute as Kant himself thought, but nonetheless severe. But surely the fact that keeping my promise would require violating the laws of logic or mathematics provides a good reason both for thinking that I cannot keep my promise and for thinking that I do not have the obligation to do so. If I promised you yesterday that I would today square the circle, surely I do not have the obligation to square the circle just because that is something I cannot do. The CI procedure on its best interpretation, then, will assume that we cannot get to the truths of logic and mathematics through moral theory. Similarly, if keeping my promise would require violating the laws of physics, a similar conclusion would follow. If I promised you yesterday that I would today jump over the moon, surely I do not have the obligation to jump over the moon just because that is something I cannot do. The CI procedure on its best interpretation will assume that moral theory cannot get us to the truths of physics. Moreover, that interpretation of the procedure will construe the truths of physics in such a way that they are accessible to us only through a methodology that is a posteriori in nature. Thus the best interpretation of the procedure in this regard imposes limits on the metaphysics, and this is the central point. Our metaphysical understanding of agency is structured not only by what we know about the formal truths of logic and mathematics but also by what we know about the laws of nature. Good metaphysics and good metaphysical methodology recognize this.
But, then, this is to admit that some considerations of fact a posteriori can falsify the "claims of reason," on a fully a priori understanding of the claims of reason. Now, however, comes the slippery slope. What is to stop some of these facts from being the facts of our psychology? Once the Kantian has construed the best interpretation of the CI procedure to accommodate the a posteriori truths of physics, it is difficult to see what principled methodological reason there can be for exempting the truths of psychology from their independent status. And, of course, if the truths of psychology are independent of moral theory and to be discovered by an independent methodology and if internalism is correct, then the correct version of internalism is not the one Korsgaard has in mind. It is one that leads to a de-
[12] . Korsgaard, "Skepticism about Practical Reason," 23–25.
nial of the independent status of moral theory. Whatever role metaphysics plays in our formulation of a concept of rational agency, it is not one that is a priori in the way that Korsgaard and Kant require. The best conception of metaphysics is one according to which metaphysics begins with and is restricted by a large body of empirical knowledge. Good metaphysics, then, is never entirely a priori, perhaps not even largely so. The best conception of rational agency will recognize this, as will the best interpretation of the CI procedure, neither of which will yield the result that the facts of psychology are accessible through an independent moral theory. And, according to the methodology that requires the integration test, it is irrational to accept a methodology that requires us to ignore a vast body of knowledge we already have in the formulation of our metaphysics. This is no less true of psychology than it is of logic, mathematics, physics, or biology. That Kant would reject this was in part a function of his belief that psychology is not a science, but our increasing understanding of both cognitive and clinical psychology will not allow us this belief.
I cannot, then, see how it can be rational to insist that the facts of psychology do not have the kind of standing that can invalidate a normative claim or a conception of practical reason. To do so would be to endorse an irrational methodology that requires that we ignore knowledge that we might already have or that we might come to have by a reliable method. To advance a claim about a person having the capacity for a certain motive is to advance a claim about that person's psychology (which, on the most plausible view, includes a claim about the capacities of a certain neurology). It is to advance a claim that a certain kind of consideration has psychological pull. To advance a conception of practical reason, then, is to advance a claim about the functional capacities of a certain psychology. This is an issue concerning a claim the truth of which can be determined independent of moral theory by means of the integration test and the best methods of social science.[13] To conceptualize the issue of rational agency in such a way
[13] . Finally, there is a possible coherence problem with such a denial for an internalist, especially for Korsgaard. According to her, intuitionism is a version of externalism, and "intuitionists do not believe in practical reason" ("Skepticism about Practical Reason," 10). Korsgaard is right to insist that there is no essential connection between the truth of a moral claim and its practical rationality according to intuitionism. This is directly due to the fact of intuitionism's externalism. Yet Korsgaard seems to deny that we should understand what it is to have a motive in terms of what Williams calls an agents subjective motivational set. I am not sure what motivational set is to replace it, but if she is suggesting that we should understand having a motive independent of the fact that a certain consideration has psychologicalpull, then it is not clear that she is even talking about motives. Indeed, it seems that she is an externalist after all. Ironically, however, this means that she has given up on founding morality on practical reason, which is a requirement of her own theory.
that it is not subject to such testing seems motivated less by a concern for inquiry than by the desire to protect a favored conception against any possible incriminating evidence. And, in this regard, it is difficult not to notice that Kant's metaphysics maps very neatly onto his Christian heritage. For it is crucial to his argument that it is possible to be a rational agent and not have desires and sentiments because God is such a being. How one is to evaluate claims about God's psychology, however, is beyond empirical testing. In fact, how one is to evaluate such claims is mysterious by any means, and this is a worry about a conception of agency that arises from a worry about method.[14]
It is difficult to know what to say if deep disagreement over methodology persists past these observations. I can point to the kind of illumination philosophy might be capable of when structured within a larger body of knowledge that is achieved through the reliable methods of science. If a philosopher does not accept the requirement that philosophical knowledge must be not only consistent with scientific knowledge but also continuous with it, I simply do not know what else to say. That metaphysical Kantians do not accept the continuity requirement seems evident in the fact that they conceptualize the issue of rational agency in a way that the conditions for the possibility of rational agency might be discontinuous with the facts of human psychology. With this in mind, I can point to the dangers of conceptual schemes that have allowed themselves to reject the continuity requirement. The history of theology is full of examples. At any rate, I should have said enough by now to make clear the reasons for my methodological commitments, even for those who will not accept them. Perhaps this is all that can be done.
[14] . Kant's distinction between an impurely rational being and a purely rational being corresponds to a distinction between human beings, who have desires and inclinations, and God, who does not. Since we need a distinction between purely rational and impurely rational agents, how is a conception of agency that is tightly connected to the study of human beings to accommodate the distinction? Rather easily, I think. A purely rational being is not one who does not have desires, sentiments, and inclinations but one who deliberates in the light of the relevant facts, is fully competent in the formal requirements of logical reasoning, and does not suffer from psychological encumbrance. Whether such an agent employs the CI procedure is an open question. And it is the integration test that provides the means of answering it.
5.
Unlike the metaphysical school, however, Kantian constructivists can accept much, perhaps all, of what has been said so far regarding methodology. I will argue in what follows that if Kantian constructivists want both to base morality on practical reason and to do so in a way that is independent of metaphysics, then they must accept the integration test rather than what they call reflective equilibrium as the ultimate arbiter of rival conceptual schemes. Moreover, I will argue that the integration test does not rule out Kantian internalism a priori.
In his Dewey Lectures, John Rawls claimed that there is a kind of moral theory that has yet been unnoticed, which contrasts markedly with the Kantian metaphysical tradition.[15] This kind of theory Rawls calls a constructivist theory. As I understand the structure of such a theory, any instance of it has three features: a decision model spelled out in terms of an original position, a conception of reflective equilibrium that provides an explanation of our moral sentiments and values, and a conception of the justification for moral principles.
The construction of a decision model employs Rawls's now-famous notion of an original position. One aim of the decision model is to describe the original position in a way that will allow anyone in that position to make a rational choice regarding alternative principles to govern behavior, policy, or the development of character. Of course, the decision model can be set up in a variety of ways, depending on how the original position is described. John Harsanyi, for example, has long ago constructed an original position that leads to the choice of average utilitarianism.[16] No doubt different models could be constructed to yield a wide variety of other possibilities. How, then, is one to choose between these possibilities, and what does the choice reveal?
It is the notion of reflective equilibrium that allows the choice between the alternative ways of setting up the original position. Reflective equilibrium is gained when the original position has been set up in a way that not only allows a choice of principles that are rational from that position but also where the results of the choice model yield maximal coherence be-
[15] . John Rawls, Political Liberalism (New York: Columbia University Press, 1993), 89–130.
[16] John Harsanyi, "Morality and the Theory of Rational Behavior," in Utilitarianism and Beyond , ed. Amartya Sen and Bernard Williams (Cambridge: Cambridge University Press, 1982), 39–62.
tween one's judgments regarding particular cases and general moral principles.[17] Thus the argument for one conception of the original position over another is that one yields a greater coherence within our moral experience. And the claim is that the reflective equilibrium resulting from the model succeeds in providing us with a certain kind of explanation. What is explained is something about our moral psychology, namely, what our deepest moral sentiments and values are. So construed, a constructivist theory is, according to Rawls, "a theory of moral sentiments (to recall an eighteenth century title) setting out the principles governing our moral powers."[18] On this view, the rationality of the choice of one set of principles over another by those in the original position is explained by the fact of the sentiments of a certain psychology, the ones built into the assumptions of the original position. That the characterization of the original position yields a result that achieves reflective equilibrium reveals that these assumptions are features of our psychology, not the psychology of those in the original position.
Rawls's own attempt is to construct a model that is "Kantian" in content and that applies to issues of distributive justice regarding the basic institutions of society. The claim is that if we set up a decision model in a certain way—a way that provides a clear rationale for basic principles of justice—and if this way of setting up the model achieves the desired reflective equilibrium, then we have a good argument that the model captures our notion of moral rationality. Were we to set up the model in a different way that yielded different principles but nevertheless achieved the desired reflective equilibrium, this would provide an argument for another conception of moral rationality. Rawls's claim about distributive justice is that when we set up the decision procedure with Kantian assumptions about rational agency we are able to test different substantive conceptions of distributive justice in a way that allows us to get reflective equilibrium. The reasonable assumption is that though there is a kind of predictability that if you start with a procedure built on Kantian assumptions you will get substantive Kantian results, there is no advance reason for thinking that this will result in reflective equilibrium. Hence the achievement of equilibrium adds something to the argument: It constitutes evidence that Kantian assumptions are behind our considered judgments.
The methodological point is that we should understand moral rational-
[17] . John Rawls, A Theory of Justice (Cambridge, Mass.: Harvard University Press, 1971), 20 ff., 48–51, 120, 432, 434, 579.
[18] . Ibid., 51.
ity as proceeding from our deepest moral sentiments and in terms of that decision model that allows us to achieve reflective equilibrium regarding our considered moral judgments. The task of moral theory is to provide such a model, and Rawls claims that, in this sense, moral theory is an independent philosophical discipline, one that is independent of metaphysics, as well as theory of meaning, epistemology, and the philosophy of mind.[19] Whether moral theory is independent of psychology and social science is another matter, one that we will come to later.
The third feature of a constructivist model is a conception of what justifies moral principles. Not only does reflective equilibrium explain our moral sentiments by revealing how they structure our judgments; the principles yielded by the decision model that achieves reflective equilibrium are justified by that very fact. The reason they are justified is that they are the rational result of our deepest moral sentiments and values. As Rawls says, "I have not proceeded then as if first principles . . . have special features that permit them a peculiar place in justifying moral doctrine. They are central elements and devices of theory, but justification rests upon the entire conception and how it fits in with and organizes our considered judgments in reflective equilibrium."[20] Kantian constructivism, then, is the claim that if we set up a decision model on Kantian normative (but not metaphysical) assumptions, we will get reflective equilibrium in a way that we will not on any non-Kantian model. By so doing, we gain both an explanation of what our deepest sentiments are and a justification for the resultant moral principles.
This conception of justification reveals a methodological commitment regarding the study of rational agency that is decidedly different from that of the metaphysical school. At the most general level, constructivist method is committed to the view that what morally justifies a set of principles is that
[19] . I cannot see that this is a good argument for the independence of moral theory from psychology, if what is meant is that moral obligations have a nonpsychological foundation. On this view, the method of reflective equilibrium is a method of psychological inquiry, one that is designed to reveal features of our psychology. Perhaps the claim is that methodologically constructivism is not among the disciplines of psychology as a social science. Fair enough. But this is a fairly weak claim about the independence of moral theory from psychology, one that metaphysical Kantians would not allow as strong enough. Constructivism does not yield a priori truths about our moral sentiments. It takes the data of what we report as our considered moral judgments and makes explicit the values behind such judgments. This is the study of the norms embedded in a psychology. See John Rawls, "The Independence of Moral Theory," in Proceedings of the American Philosophical Association (New York: American Philosophical Association, 1974), 5–22.
[20] . Rawls, A Theory of Justice , 579.
it expresses the deepest moral sentiments and values of a certain psychology, namely, ours. Moreover, it is committed to an a posteriori method of discovering what those values are. To be sure, there is the philosophical construction of the decision model and the original position. But the test for this model is whether it achieves reflective equilibrium. And it is a contingent fact whether any particular model achieves this result, a fact that rests on the contingencies of our psychology. What we must understand through reflective equilibrium is what moral rationality comes to for humanity or some subsection thereof. There is no a priori argument that only the CI procedure will result in the desired reflective equilibrium. Understood in this way, constructivists are committed to the empirical tradition: Moral rationality serves the moral values most deeply embedded in our psychology. Hume and Aristotle would most heartily agree. For constructivists, what those values are can be discovered through constructivist methodology, which is at once a kind of moral theory and a kind of psychological study. It is not a branch of metaphysics, and it proceeds independently of metaphysics.
This methodological difference between constructivist and metaphysical Kantians requires a difference in a general conception of rational agency. For constructivists, the fact that is explained by reflective equilibrium is that the principles yielded by the decision model rationally express our deepest moral sentiments. Moreover, it is this very fact that rationally justifies these principles. No metaphysical deduction of the CI procedure is appealed to, nor could it be and reflective equilibrium be the justification for the principles. For the metaphysical Kantians, on the other hand, reflective equilibrium could never play this justificatory role. The reason that the metaphysical school persists is that they see that Kant was right that the only kind of moral theory that will yield the rational necessity of moral principles that bind all rational agents is one that employs a metaphysical deduction of the moral law. If this is true, then the most reflective equilibrium can do is elucidate our deepest moral values; it can do nothing to justify them. The justification must come through pure practical reason. For this reason, one cannot be both a constructivist and a metaphysical Kantian, where constructivism includes the claim that reflective equilibrium justifies the acceptance of moral principles.[21] Thus on the Kantian constructivist
[21] . It is possible to endorse something called constructivism and delete the third feature, the claim that reflective equilibrium justifies the selection of moral principles. The cost of the deletion, however, is that the theory loses any claim to being normative.
view, moral rationality comes to acting on our deepest values, and the evidence that we have so acted is in principle provided by a moral theory that achieves maximal coherence in explaining what our deepest values are. And methodologically, even if the CI procedure were to result in reflective equilibrium, there would be no necessity in this fact. It would simply be a function of a contingent fact about our psychology. Hence the constructivist methodology is committed to a conception of rational agency that can be empirically studied.
There is, however, a problem. The problem is that it follows that acting on our deepest moral sentiments and values is practically rational only if we assume that the conception of moral rationality constructed in the original position is practically rational. But reflective equilibrium of our moral judgments could never assure this. Only a reflective equilibrium of all our considered practical judgments could yield this result. What is needed, then, is some further test that supports the claim that our deepest moral sentiments and values, on some interpretation of what they are, are in fact fundamental to our overall psychology. Since the method of reflective equilibrium reveals that the foundation for the rationality of our moral judgments is to be found within our psychology, namely, within our deepest moral sentiments, consistency requires that the foundation for the rationality of our practical judgments is to be found in the deepest values of our psychology. This follows from the psychological basis for the explanation of rationality appealed to vis-à-vis the notion of reflective equilibrium. And it should be noted that this model for rationality is squarely within the Aristotelian/Humean tradition. Thus to assert that the Kantian constructivist view of the CI procedure is rational is to assert that it and it alone can serve a certain psychological function, the one discovered in the discussion of Herman. Therefore, if we are to accept constructivism as a method that bases morality on practical reason, as Kantian internalism does, then either the notion of reflective equilibrium will have to include the integration test or the integration test should be understood as the arbiter of whether our deepest moral sentiments are rational. Without the integration test, constructivism will not have yielded the claim that our deepest moral sentiments and values have a rational foundation in practical reason. And thus constructivism will have reverted either to externalism or to the metaphysical school, in which case the notion of reflective equilibrium is justificatorially inert.
If a constructivist model is to be independent of metaphysics and it is to be defensible as practically rational, it will have to show that our deepest moral sentiments provide a basis for practical reason that can sustain
the elements of integrity in the thin sense over time. I do not see how the method of reflective equilibrium can achieve this, as long as the equilibrium, however "wide," includes only our moral sentiments and judgments.[22] For it seems that from the perspective of practical reason, it is always an open question whether independently specifiable moral concerns are among the deepest concerns of our psychology. That they are (on some interpretation of what "moral" concerns are) could be revealed by the integration test. For if it is true that moral concerns on some interpretation are among the deepest concerns of our psychology, then the frustration of those concerns rising from the attempt to live apart from them or contrary to them will result in the inability to sustain the elements of integrity in the thin sense.
It is instructive in this regard to think of how equilibrium might be achieved and lost. At one point in the not too distant past, reflective equilibrium regarding the morality of gender and race relations would have yielded much different results than today. What explains this? A glib answer is that we have different expectations of ourselves and others today because we now live in a different culture. A more plausible answer is that we tried to live with certain expectations of ourselves and others and found that we could not sustain those values over time and maintain the elements of integrity. The burden of self-deception imposed by those expectations and the stress incurred by the attempt to find meaning in them simply would not allow them to retain a central place in our psychology. And this is why they are not and were never practically rational. This is not to say that reflective equilibrium is not an important device of inquiry. But it is to say that what explains reflective equilibrium is the belief that we have found the values that are at the core of our psychology and that they can sustain integrity over time. Thus if it is the belief that the integration test reveals that any particular reflective equilibrium is trustworthy, then it is the integration test that is the ultimate arbiter of rival conceptual schemes. I see no reason why constructivists could not accept this point as a necessary stage of constructivist inquiry.
If, however, constructivists reject the necessity of this stage of inquiry, they are doing one of two things: either they are covertly endorsing Kant-
[22] . For discussion of a conception of wide reflective equilibrium different from that of Rawls, see Norman Daniels, "Wide Reflective Equilibrium and Theory Acceptance in Ethics," Journal of Philosophy 76 (1979): 256–82, and "Reflective Equilibrium and Archimedean Points," Canadian Journal of Philosophy 10, no. 1 (March 1980): 83–103.
ian metaphysics, or they are suspending judgment about whether Kantian morality is practically rational. If they are doing the latter, it is hard to see in what sense they are defending a conception of morality founded on practical reason.
The best interpretation of constructivism, then, is one that accepts the integration test and is willing to put the CI procedure and the conception of practical reason associated with it to this test. In chapter 3 we will see more of what the integration test involves in terms of human behavior. But here the important thing is that there is nothing a priori that rules out the possibility that Kantian internalism, understood as the claim that the CI procedure must serve a functional role within our psychology, might pass the test. In what follows, then, I will proceed with the methodological commitment of applying the integration test to both the CI procedure and the Aristotelian conception of practical reason.
The argument against Kant, however, must be against the best interpretation of the CI procedure and its results. This task is made more difficult by the fact that there are both consequentialist and deontological interpretations of the procedure, as well as various deontological interpretations. These are disputes over the content of the CI procedure. Rather than define and address these issues here, I will address them as they become relevant to the myriad contexts in which the CI procedure, on any interpretation, must exercise its function. I bring up this issue now for two reasons. The first is to reassure the reader that I will make a good faith attempt to evaluate the most plausible and developed views of the CI procedure rather than implausible caricatures. The second is to insist on two features that must accommodate any employment of the integration test. The test must be administered in a way that is both context-sensitive and contextually thorough. To be context-sensitive it must be applied to contexts in terms of all and only the relevant features of the context, and to be contextually thorough it must be applied to a substantial range of contexts to which the conception of rational agency might plausibly apply. My argument will be that the deliberative strategies available to Kantians through the CI procedure are not rational across the range of contexts an agent must face in living a life. This means that any contextually thorough implementation of the integration test will reveal that the CI procedure fails in significant contexts and that the Aristotelian alternative does not. It also means that implementing the test requires very careful and detailed work. Showing that a conception of rational agency is plausible in some contexts is not sufficient for rational acceptance of that conception. This is why the analysis in parts 2 through 4 is so detailed. As in science, there is no shortcut to testing a
theory, and, as in art, glossed blemishes eventually penetrate the veneer designed to conceal them. Philosophy involves hard, detailed work.
6.
The result, then, is that I accept Allison's requirement for an argument for the conception of rational agency employed here and explicated in general in the remainder of part 1. The test is the one implied in Herman's arguments for the CI procedure. If Kantians believe that this is not the appropriate test, they must give us philosophical reasons for rejecting the integration test as the appropriate test and then provide the appropriate test with an argument for it. Without this, the claim that I have not given an argument for rejecting the Kantian conception of rational agency will seem mysterious.
2—
Impartiality, Regulative Norms, and Practical Reason
This chapter aims at establishing a point about our considered moral judgments, namely, that they reflect a conceptual scheme in which the relationships between regulative norms are all symmetrical in their regulative functions. None of these norms function asymmetrically. If this is true, then our considered moral judgments reflect a central feature of the Aristotelian conception of practical reason rather than either of the Kantian conceptions. Establishing this will provide prima facie evidence that no Kantian constructivist model can achieve reflective equilibrium. Of course, we might discover after philosophical reflection that our considered moral judgments lack a rational foundation. Nevertheless, if it can be shown that the Aristotelian scheme has a home in our judgments, this constitutes an important first step in an argument for that scheme. It will remain to be shown, then, that there is a rational foundation for this scheme. This can be provided if the scheme can be developed sufficiently and, once developed, can be shown to pass the integration test.
Before I begin with the argument, however, I wish to engage in a little diplomacy, which is sometimes called for in philosophy when argument breaks down due to mistrust. Both sides of a dispute feel that their arguments have been treated unfairly; so they withdraw and accuse. This is somewhat true of the current debate between Kantians and non-Kantians concerning the role of impartiality in ethics. Kantians often complain, sometimes bitterly, that objections to Kantian ethics are based on caricature and that if more careful attention were paid to the role of regulative norms in practical reason the objections would lose their force. Non-Kantians express frustration about the cursory treatment by Kantians of alternative positions, accusing them either of providing only a gesture in the direction
of a response to serious alternatives or of an unwillingness to specify criteria that might show that they are wrong. The evidence suggests, I believe, that there are good grounds for complaint on both sides: the tendency to see everything that is said against one's views as caricature can be as pernicious a form of dismissal as the tendency to caricature itself.
The only way to combat both tendencies is with hard philosophical analysis. What I intend to do here is to construct an inquiry regarding the role of regulative norms in practical reason in a way that studiously avoids caricature and argue that recent Kantian defenses fail to consider a very important conceptual alternative to their view. What they have said thus far in the debate does not address the alternative I will present. If I succeed, the only appropriate response is a philosophical argument that shows where I fail. On the principle, then, that a careful argument requires a careful reply, I accept the Kantian challenge to avoid caricature in understanding the Kantian project, but I lay down the challenge of a non-Kantian view that requires an equally serious and detailed reply rather than a gesture in the direction of a response. One final diplomatic point. Kantian ethics has for some time now enjoyed a favored status, especially in the United States. This is due in no small way to the contributions of some very talented thinkers. But if wide consensus among philosophers establishes the presumption that the view held should be taken seriously and not oversimplified by its opponents, then the dangers of consensus require that the prestige of opinions plays no argumentative role in inquiry.
So much, then, for diplomacy; now for argument.
1.
Recent defenses of Kantian impartiality have employed the concept of regulative norms.[1] My purpose here is to clarify this concept, to show how it defuses some criticisms of Kant, and finally to argue that a clear understanding of regulative norms undermines Kantian claims about the role of
[1] . "Principles" is preferred among Kantians instead of "norms," but there are philosophical reasons for the latter. It might turn out that the regulatory work thought to be done by principles is actually done by other kinds of norms. "Norms," then, allows us to employ a philosophically neutral vocabulary. See especially, BarTbara Herman, The Practice of Moral Judgment (Cambridge, Mass.: Harvard University Press, 1993), 222–23. Others might have other sources of discomfort with the term, but I have been unable to find another term that is as neutral between philosophical conceptions.
impartiality in practical reason. I argue that impartial norms play a very important role in the practical reason of an admirable agent but not the one attributed to them by Kant and his defenders. Moreover, the position I defend is not one that has yet been considered on either side of the debate regarding the role of impartiality in ethics. So it is important not to read what follows as a defense of some standard objection to Kantian impartiality.
I define a regulative norm in terms of what I call its regulative effect on other norms. Two definitions will be given: one in the language of practical reason and one in the language of morality. For the sake of simplicity, the definitions assume contexts in which only two norms are relevant.
For any two norms, A and B, A is regulated by B just in case there is some practical context in which (i) acting on A is irrational because it violates B and (ii) were it not for B it would be rational to act on A.
The second condition should be understood as "(ii) were it not for B it would be either (iia) not irrational to act on A or (iib) irrational not to act on A." That (i) and (iia) are sufficient for regulative effect is reflected in the fact that in a context in which were it not irrational to act on A condition (i) could not obtain. Since (iib) entails (iia), (i) and (iib) also reflect regulative effect, but (iib) is not a necessary condition. A third feature of regulative effect is that when A is regulated by B in the previous sense, not acting on A in such contexts is not contrary to A. For example, where, say, acting for the sake of a friend is regulated by the concern for respect for others, not acting for the sake of a friend is not contrary to friendship. (I will say more about this third feature shortly.)
The same concept can be formulated in moral terms.
For any two norms, A and B, A is regulated by B just in case there is some practical context in which (i) acting on A is wrong because it violates B and (ii) were it not for B it would be permissible to act on A.
The second condition should be understood as "were it not for B it would be either (iia) not wrong to act on A or (iib) wrong not to act on A." (i) and (iia) are sufficient for regulative effect, since any context in which acting on A is not wrong is a context in which condition (i) does not obtain. Of course, if not acting on A is wrong in a context, it is not wrong to act on A in that context, though acting on A might not be wrong without its being obligatory. Finally, the third feature of regulative effect expressed in moral terms is that when A is regulated by B in the previous sense, not acting on A in such contexts is not wrong in terms of A itself. For example, where acting for the sake of a friend is regulated by the concern for respect for others,
not acting for the sake of a friend is not an act of betrayal or disloyalty. This third feature of regulative effect contrasts with a way of talking about competing duties overriding each other. In contexts of the sort just mentioned, the language of regulative effect allows us to avoid saying that the duty of respect for others overrides the duty of friendship. This latter way of talking implies that in the context friendship requires acting in one way and respect for others in another. But the language of regulative effect for such contexts would not have us say that friendship requires us to do one thing and respect requires that we do another but that since favoring the friend would be disrespectful of others, friendship does not require what it otherwise would permit. This turns out to be a very powerful conceptual feature of regulative effect.[2]
I will assume without argument that any acceptable conceptual scheme that employs regulative norms in moral terms inherits the burden of mapping that conceptual scheme onto a defensible conception of practical reason. In other words, the language of morals must be translated into the language of practical reason, where it is shown that the obligations of morality are practically rational. This does not rule out uniquely moral obligations, but it does require that if there are such obligations, they must be practically rational for agents like us.
2.
To understand how this conception of regulative norms aids in a defense of Kant, it is first necessary to understand the kind of objection to which it is a response. The kind of objection I have in mind applies to the impartiality requirement as a necessary feature of any rational morality. Those who defend such a requirement have come to be known as "impartialists" and those who deny such a requirement as "partialists." Both terms are distorting, I will argue, for they are misleading as characterizations of either morality or practical reason. Among the leading partialists is Michael Stocker, and one of his examples has become a touchstone in the debate. I quote his example from his now-famous article, "The Schizophrenia of Modern Ethical Theories." He says:
Suppose you are in the hospital, recovering from a long illness. You are very bored and restless and at loose ends when Smith comes in once again. You are now convinced more than ever that he is a fine fellow and a real friend—taking so much time to cheer you up, traveling
[2] . I would like to thank A. M. MacLeod for comments on this point.
all the way across town, and so on. You are so effusive with your praise and thanks that he protests that he always tries to do what he thinks is his duty, what he thinks will be best. You at first think he is engaging in a polite form of self-deprecation, relieving the moral burden. But the more you two speak, the more clear it becomes that he was telling the literal truth: that it is not essentially because of you that he came to see you, not because you are friends, but because he thought it his duty, perhaps as a fellow Christian or Communist or whatever, or simply because he knows of no one more in need of cheering up and no one easier to cheer up.
Surely there is something lacking here—and lacking in moral merit or value. The lack can be sheeted home to two related points: . . . the wrong sort of thing is said to be the proper motive; and, in this case at least, the wrong sort of thing is . . . essentially external.[3]
According to these observations, loving, partial acts must be motivated directly by love, and when apparently loving acts are done from nonloving, impartial motives, they cease to be loving acts. This is surely correct. But Stocker seems to suggest that if an act is done from a sense of obligation it cannot be done from love, and it is here that some impartialists think he has misunderstood the role of regulative norms in practical reason.
3.
The point against Stocker has been made most explicitly by Marcia Baron whose view is very similar to Herman's.[4] They agree that there is something objectionable about the moral psychology of the person in Stocker's example, but they disagree with him that what is objectionable is due to any inherent problem with a fully impartialist conception of morality. We can see this when we understand how a Kantian conception of morality employs regulative norms. I will make what I take to be their point in my own terms.
[3] . See Michael Stocker, "The Schizophrenia of Modern Ethical Theories," in The Virtues: Contemporary Essays on Moral Character , ed. Robert C. Kruschwitz and Robert C. Roberts (Belmont, Calif.: Wadsworth, 1986), 41, 42.
[4] . See Marcia Baron, "The Alleged Repugnance of Acting from Duty," Journal of Philosophy 81 (1984): 197–220. See also her Kantian Ethics Almost Without Apology (Ithaca: Cornell University Press, 1995) for a developed discussion of the role of imperfect duties and their regulatory role in Kant's moral theory. For Herman's views, see The Practice of Moral Judgment . Add to this Allison's Kant's Theory of Freedom (Cambridge: Cambridge University Press, 1990) and the result is a trilogy of books that presents a powerful picture of Kantian morality that relies heavily on the notion of regulative norms.
A norm can function either regulatively or determinately. It functions determinately when it is act-specific in its action-guiding role: when (with ample background conditions) it is sufficient for determining the act from among those available to the agent that is practically rational for the agent to perform in the circumstances. On the other hand, a norm functions merely regulatively in a context when though it is sufficient (given ample background conditions) for determining that some actions are practically irrational for an agent in the circumstances, it is not act-specific in its actionguiding role: it is not sufficient to determine which specific act is rational. Note that this way of making the distinction does not ascribe the functional features of norms to norms simpliciter but to their function in a context. So described, the distinction involves the concept of rational justification.
Also important, however, is the concept of motivation. In contexts in which a norm functions determinately as a justifier, it is also the determinate motive, as affection is to a hug.[5] Similarly, where a norm functions regulatively, it is only the regulative motive, as the desire for good health is to the avoidance of fatty foods.[6] What impartialists like Baron and Herman want to claim is that for any moral act there must be either a determinate or a regulative justifying norm that is completely impartial. Moreover, this does not prevent there being contexts in which the determinate norm is partial in both its justificatory and motivating aspects. Such contexts are those in which no impartial norm functions as a determinate justifier. Thus friends are free to visit each other in the hospital out of love where the love is determinate both as justifier and motivator and where such visits do not contradict the practical requirements of any relevant impartial norm. Most important, being motivated by a concern not to violate such impartial requirements does not intrude on our psychology in the way that Stocker imagines.
If Baron and Herman are right, and I think they are, we should be able to see in our own conceptual scheme the features of the regulative effect of impartial norms on partial norms, like the various forms of personal love. If we stay with the hospital situation and friendship, we should be able to see the following:
Friendship is regulated by Impartial Respect just because there is some practical context in which (i) acting on Friendship is irrational (wrong)
[5] . Influenced by Herman, Baron calls this a primary motive. See "The Alleged Repugnance of Acting from Duty," 207.
[6] . Again following Herman, Baron calls this a secondary motive. "The Alleged Repugnance of Acting from Duty," 207.
because it violates Impartial Respect and (ii) were it not for Impartial Respect it would be rational (permissible) to act on Friendship.
Now consider two contexts, one in which an impartial norm is regulative and another in which it is determinate, both of which illustrate that we do in fact accept the truth of the above claim of the regulative effect of impartial respect on friendship.
In the first context, Smith's visiting you in the hospital does not involve his doing so at the cost of his failing to show proper impartial respect for others. He visits you as a loving friend who is also respectful of others. Here respect for others functions regulatively both as justifier and as motive, but friendship functions determinately in both regards. The regulative motive is revealed by some counterfactual statement to the effect that Smith would not have felt free to visit you had he thought it contrary to respect for others. Where the counterfactual condition is satisfied, Smith's sentiments of personal affection are free to express themselves directly rather than "externally," as Stocker would have it. The test, then, for whether some determinate norm is regulated by some other norm is whether there is some counterfactual condition at work in our approval of the determinate norm. The test is that when a norm is regulatively influenced by another norm yet is determinate in the context there is some counterfactual statement to the effect that there is some other context in which conditions (i) and (ii) hold regarding some other norm's regulative effect on that norm.
We need only imagine a second context to see that we do indeed recognize such a counterfactual in the background of our approval of Smith's visit in the first context. In the second context, Smith believes that his visiting you in the hospital would come at too great a cost to others and would not be consistent with impartial respect for them. Perhaps you are only slightly ill and Smith is scheduled to testify in court concerning the innocence of someone falsely accused. Here one would expect any fully admirable agent to have his or her course of action determined by the impartial consideration of respect for others. That is, one would expect that friendship should be regulated in such a way that it is respect that is the determinate justifier as well as motive in practical conflicts of this sort. Thus, on this view of friendship and impartial respect, impartial respect does not allow Smith to visit you in the second context, and friendship does not require it. It is this last feature—what I have previously referred to as the third feature of regulative effect—that shows that we accept a regulated conception of friendship as a partial norm. Surely, anyone alienated by such a conception of friendship suffers from bad character.
The point of both examples is that determinate justifiers and motives match up in just the way one would expect for a practical reasoner with good character. As far as this goes, Baron and Herman are right that Stocker's example does not pay due attention to the role of regulative norms and how they can function on an impartialist theory. On this analysis, however, the claim that morality is impartial comes to the claim that for any fully admirable moral agent there are no partial norms that are not subject to the regulating functions of impartial norms. In what remains, I will argue that such a claim is not sufficient reason to endorse fully impartialist conceptions of morality, and it is here that I will present an alternative not yet considered in the debate over impartiality in ethics. I will limit my comments to Kantianism.
4.
Kantians are right to insist that our view of partial norms and sentiments, like friendship and parental love, involves their being "filtered" through other considerations. Consider a conception of friendship not filtered by the regulative effects of other norms, partial or impartial. On this view, friendship requires one always to give priority to friends no matter what other considerations are at stake. No matter whether it involves love of family, sympathy for those in need, or respect for the dignity of others, friends always come first. On this view, any failure to give priority to friends in cases of conflict is an act of betrayal, for when the counterfactual test is applied to this conception of friendship, it reveals that there is no possible world in which others take priority over one's friends. The counterfactual test shows, then, that friendship on this conception is asymmetrical in its regulative effect on all other norms within this conception of practical reason. Other kinds of norms on other conceptions of practical reason can have this same property. Abraham's willingness to sacrifice Isaac seems to give religious faith this kind of status in Kierkegaard's understanding of the teleological suspension of the ethical.[7] Also, many moral theories afford some norms this status. What should be clear at this point is that attributing such a property to partial norms is counterintuitive, to say the least.
For now, however, it is important to see that some regulative norms can and do have another property, namely, the property of symmetrical regula-
[7] . Søren Kierkegaard, Fear and Trembling , trans. Walter Lowrie (New York: Anchor Books, 1954).
tive effect or function.[8] Consider in this regard friendship and parental love. For most of us, conceptions of these norms ascribe to them the property of having symmetrical regulative effect on each other. That is, not only do we believe that
Friendship is regulated by Parental Love just because there is some practical context in which (i) acting on Friendship is irrational (wrong) because it violates Parental Love and (ii) were it not for Parental Love it would be rational (permissible) to act on Friendship
we also believe that
Parental Love is regulated by Friendship just because there is some practical context in which (i) acting on Parental Love is irrational (wrong) because it violates Friendship and (ii) were it not for Friendship it would be rational (permissible) to act on Parental Love.
Within a conception of practical reason where the only regulative influences on these norms come from each other, considerations of family sometime override considerations of friendship but there are no other exceptions for the priority given to friends. On this view, some cases of favoring family over friends are not acts of betrayal and should not be so viewed by the friends. Think here of what you would expect of a friend if you had to cancel an appointment because your child was ill. A conception of friendship that makes exceptions in this way only for family is a conception of friendship involving a minimal regulating influence, which is revealed by the scope of the counterfactual conditions regarding other norms. A maximally regulated conception of friendship would be one the normative dimensions of which were influenced by the entire range of considerations a fully admirable agent would have, including impartial sympathy and respect for others. So it is implausible that any acceptable conception of friendship or any other partial sentiment will reflect the concept of asymmetrical regulative effect on other norms. Moreover, it will reflect the regulative effect of other norms on it, including the regulative effect of im-
[8] . The concept of symmetrical regulative function or symmetrical regulative effect is similar to the concept of bidirectionality found in the work of Henry Richardson. I discovered and read Richardson's excellent book, Practical Reasoning about Final Ends , only after the manuscript of Agent-centered Morality was sent to the copy editor. For his discussion of bidirectionality, see Henry S. Richardson, Practical Reasoning about Final Ends (Cambridge: Cambridge University Press, 1994), 141, 143, 176–77, 182.
partial respect. This can be captured in a rather complex statement of the counterfactual conditions included as a test of the extent to which the partial norm is regulated by other norms. Partial norms, then, are regulated both by other partial norms and by impartial norms, and at least some of the regulative functions between partial norms are symmetrical.
If my conception of friendship includes within it a reflection of a regulating influence of parental love, then I will not think of some cases of my giving priority to my children over my friends as cases of betrayal of my friends; and I will expect my friends to see this as such. Similarly, if my conception of parental love includes within it a reflection of a regulating influence of friendship, then I will not think of some cases of my giving priority to my friends over my children as cases of betrayal of my children; and I will expect my children to see this as such. It is only through conceptions of sentiments or norms that display such symmetry of regulating influence that we can integrate various kinds of partial sentiments and norms into our lives and avoid devastating alienation. Think what it would be like for friends and children to lack an intuitive understanding of the symmetry of norms alluded to here.
5.
Now consider how impartial norms might be regulatively influenced by other norms. First, impartial norms can reflect the regulative influence of other impartial norms. Consider the Kantian point about sympathy, for example. According to Kant, the following is true of the regulative function of impartial respect in regard to impartial sympathy.
Impartial Sympathy is regulated by Impartial Respect just because there is some practical context in which (i) acting on Impartial Sympathy is irrational (wrong) because it violates Impartial Respect and (ii) were it not for Impartial Respect it would be rational (permissible) to act on Impartial Sympathy.
We do not think, as Kant rightly insists, of sympathy unregulated by other considerations in its conception as being mature sympathy.[9] Suppose I have a conception of sympathy such that it acts as justifier and motive for me regardless of other considerations. I will then think of anyone not giving pri-
[9] . Immanuel Kant, Groundwork of the Metaphysic of Morals , trans. H. J. Paton (New York: Harper Torchbooks, 1964), 66.
ority to considerations of sympathy regardless of other considerations as lacking in sympathy. But we do not think this. We do not think of a person who sometimes places respect for autonomy over the concern to alleviate distress as necessarily being unsympathetic. What this shows is that our concept of sympathy does not require that we always give priority to those with whom we are sympathetic on pain of being guilty of being unsympathetic, just as our concept of friendship does not require that we always give priority to friends on pain of being disloyal or guilty of betrayal.
Here, however, begins the problem with Kantianism. Respect is a norm that is a regulating influence on all other norms and sentiments, even including other impartial sentiments. This seems right. But what seems deeply wrong is that respect is not regulated by other norms, as is the case on a Kantian view. Two points are crucial here to seeing that, according to Kant, impartial respect is asymmetrical in its regulative features. First, to respect self and others on the Kantian view just is to be willing to deliberate in a certain way regarding how one should act. It is to submit one's deliberations regarding self and others to the categorical imperative decision procedure. Second, the CI procedure is an all things considered form of rationality according to Kant; therefore, there cannot be anything more that can regulate such rationality or such respect. Moreover, it would be odd to say from a Kantian perspective that there is a symmetrical regulative relationship between sympathy and impartial respect. This would mean that a natural inclination has a regulating influence over rationality.
Now consider, in contrast to Kant, how the regulative relationship between impartial sympathy and impartial respect might be symmetrical. Let us admit with Kant that the regulative influence of impartial respect is found in our conception of impartial sympathy. Thus the previous claim regarding the regulative effect of impartial respect on impartial sympathy is true. But where symmetry rather than asymmetry exists between these two impartial norms, the following will also be true.
Impartial Respect is regulated by Impartial Sympathy just because there is some practical context in which (i) acting on Impartial Respect is irrational (wrong) because it violates Impartial Sympathy and (ii) were it not for Impartial Sympathy it would be rational (permissible) to act on Impartial Respect.
Why would anyone think that this is true, especially as applied to our considered moral judgments? A striking case, I believe, involves our feelings toward animals. According to Kant, respect is an attitude appropri-
ate only toward rational agents, but sympathy can be had toward so-called lower animals (animals that lack some human cognitive capacities) that are not rational agents. Since moral concern, on Kant's view, is motivated directly only by impartial respect and concern for agents per se , this means that we have no direct moral concern for animals. (I will come to the point of indirection shortly). But consider a person who is both respectful of agents and sympathetic toward lower animals. He leaves his home one day to meet you for a luncheon appointment he promised to keep. If he misses the appointment, you will be a bit disappointed but nowhere near devastated. Between home and the restaurant he comes upon a man cruelly beating a dog. It is clear that the beating is a case of cruelty rather than something somehow necessary for a greater good. Moved by his sympathy for the dog, he stops and intervenes to prevent any further cruelty or harm to the animal. As a consequence, he is unable to make the appointment and only afterward is he able to call and explain the circumstances that prevented his meeting you.
The first thing to note is that these actions, given the circumstances, were surely both the right thing to do and perfectly rational for someone who both respects people and has sympathy for animals.[10] The second thing to note is that the actions on behalf of the dog affected the autonomy of two people—the dog beater and yourself—and were done for the sake of an animal that has no Kantian autonomy. Moreover, it is clear, I believe, that, contrary to Kant, we admire such actions because they show direct concern for animals.[11] Some advocates for animal rights prefer the language of "respect" for lower animals. With Kant, I believe that this is a mistake. Respect is an attitude appropriate toward rational agents, and dogs are not rational agents. They are nonetheless sentient creatures that can be cruelly treated and harmed. For this reason, we can sympathize with them. This is just what our exemplar did. Our exemplar, then, displays a conception of practical reason in which impartial respect for the autonomy of rational agents is regulated by something other than respect for the autonomy of rational agents. Moreover, I do not think any of us would say that the actions
[10] . The details might need filling in a bit to make this obvious, but no doubt it can be done.
[11] . We should be careful to distinguish direct concern from unreflective concern. When one has direct concern for another, one's concern is not the function of some other concern. This does not mean that the concern cannot be reflective and take into account other concerns.
of our exemplar were disrespectful of either you or the dog beater. Still, it remains true that were it not for his sympathy for the dog it would have been disrespectful to you for him simply not to have shown up for the appointment. Yet, because of the dog, it was not disrespectful. Conditions (i) and (ii) are clearly satisfied, as they were in the previous claim regarding the effect of respect on sympathy, which illustrates the symmetrical regulative functions of impartial respect and impartial sympathy.
Kantians, as far as I can see, have only one line of defense here, and it is an embarrassing one. They can argue that neither condition (i) nor (ii) is satisfied because it would have been wrong on the Kantian view itself to keep the appointment under the circumstances. The only way they can argue for this, however, is by appeal to the indirect connection between lack of concern for animals and concern for respecting rational agents. What one is morally concerned about, on this view, is not the welfare of the animal but either (a) the effect on one's own capacity for respecting rational agents if one allows oneself to be indifferent to the plight of animals or (b) the fact that the animal somehow belongs to a rational agent and hence harming it would be disrespectful to the owner. This seems to run directly into Stocker's problem in another context: there is a clear mismatch between reason and motive for action. Not preventing the cruelty would have been a wrong done to the dog , even at the expense of your slight disappointment and the sadistic interests of the dog beater. The Stocker problem is that Kantians cannot get the dog properly into the picture, as long as they treat impartial respect as asymmetrically regulative within practical reason.
If the Kantian does not allow that this is a counterexample to the asymmetry of impartial respect as a regulative norm, it is difficult to see why. If we should accept Kant's appeals to ordinary moral experience to establish that sympathy is regulated by respect in our understanding of a respectful agent of good character, then why should the above example not establish the same point regarding the regulative effect of sympathy on respect? Of course, if it is a counterexample, it implies that we are not pure practical reasoners. That we are not and that we do not aspire to be is evident in these very sentiments. One thing is certain: The argument from this example relies in no way on a caricature of Kant. It is appropriate, then, to expect a reasoned reply. And in the context of working out a theory of regulative norms, a familiar objection to Kant takes on added significance; it cannot be shrugged off as a minor flaw in his theory. More important, I will argue that the analysis of sympathy for lower animals generalizes in a very important sense (which is to be made clear) to the analysis of other norms.
6.
Indeed, the consideration of our sympathy for lower animals should open the way for a wholesale challenge to a conception of practical reason in which impartial respect for the autonomy of rational beings as such is asymmetrical in its regulative effect on our other concerns. In this section, I will argue that within our concern for persons, where lower animals are not an issue, impartial respect and impartial sympathy are symmetrical in their regulative functions, which is contrary to either the Kantian or the utilitarian view of these norms. I will focus on Kant's account of a duty to render aid, or what is sometimes called the Duty of Beneficence, and argue that the appeal of the Kantian analysis is not what Kantians take it to be. To illustrate the point, I will consider the constructivist model and argue that a distinction crucial to the Kantian project is not preserved.
The crucial distinction is between norms that have been "filtered," on the one hand, and norms that are the result of the "special application" of the CI procedure to the circumstances of human beings, on the other. These are not the same thing, and collapsing the distinction leads to serious errors about practical reason. Moreover, when we understand the distinction, we can see that the Kantian view must argue for the special application interpretation of the CI procedure over the filtering view, and I will argue that what actually occurs in Kantian universalizability tests, when they are successful, is filtering, not a special application of a decision procedure.
A stark way of seeing the filtering view is to consider how one can adjust two desires or sentiments to make room for both. Reflection on one's desires for both intrinsically meaningful work and the intimacy of a family life reveals various possibilities, one of which is a smaller family and a less demanding career. Becoming aware of this possibility appears as the solution to your problem when it adjusts each desire to the other within your psychology by being projectable into the future as your life. If the adjustment does not take place on reflection, filtering in this way has not yet solved the problem. The function of reflection in this case is simply to bring to consciousness different ways of living in which the intimacy of family life and a meaningful career can be conjoined or disjoined so that the psychological adjustment of desire can have its effect. And, of course, sometimes filtering does not adjust desires to each other; it sometimes removes a desire. For example, reflection on the marketing conditions involving young calves might extinguish one's desire for veal. Here one's tastes have been filtered through one's sentiments. Finally, filtering can sometimes fail
to resolve conflicts simply because desires and sentiments reach a point of incommensurability: choosing which of your children to surrender to the Nazis might simply render you paralyzed regarding the future.
Filtering cannot be what is happening on the special application view, and it is the latter that is required by Kant. The worry that generates the special application view is ultimately traced to the notion that moral imperatives must be categorical and necessary rather than hypothetical and contingent. To secure categorical necessity, practical reason must be pure, which means, among other things, that the content of one's judgment is rational in virtue of its form alone, namely, the form of reasoning of a rational being as such. Thus the Kantian thought is that form is applied to content, but it is the form that is crucial; content is only the special occasion for judgment. Otherwise, contingency enters to vitiate the purity of practical reason and to undermine the categorical demands of morality. It does this when sentiment and desire—contingent features of a psychology—are not merely the occasion for judgment but the ultimate determinants of action and rational choice. Not only are particular desires and sentiments contingencies of rational agency on Kant's view, but the mere having of desires and sentiments is also a contingency of rational agency. For Kant, God, who is a rational agent, does not have desires and sentiments; yet God can employ the CI procedure. In some of its applications, the CI procedure adjusts to the special circumstances of human beings who are rational agents and who do have desires and sentiments. This is not because God desires that human beings do well or because God has sympathy or some other sentiment for human beings. God is completely unsentimental. He is, however, rational. He can apply the CI procedure to the special circumstances of human beings and is allegedly motivated without desire or sentiment to do so. This means that the CI procedure must be such that it can yield a rational choice independent of the kind of filtering alluded to earlier; hence the special application interpretation of the CI procedure cannot involve filtering for God. Nor can it for us, for filtering cannot ensure the categorical necessity sought for in the universalizability tests. Whatever, then, is distinctively moral about moral reflection is purely intellectual and must be so if the requirements of morality are to be categorical. It is crucial here to keep in mind that to make sense of the special application view, Kant must take recourse to the notion of God as a rational being who does not have desires and sentiments.
With this distinction between the filtering view and the special application view of moral reflection in mind, consider Kant's analysis of the duty to render aid, or the Duty of Beneficence.
Kant himself believes that our duties to others involve more than mere noninterference. He believes that it is rational for a respectful person to be concerned about more than mere noninterference with the autonomy of others and that it is sometimes irrational for a respectful person to be concerned only with noninterference. He endorses a version of the Duty of Beneficence, that we should help others where doing so would come at little cost to ourselves or others and would benefit others significantly.[12] But why is one conception of respect that sometimes requires more than mere noninterference rational and the other conception that never requires more than noninterference irrational? The reason, I believe, for our thinking this is parallel to the case of animals. In the case of animals, sympathy for them regulates our conception of respect for persons because we care, in different ways, for both lower animals and for human agents. Were we to accept respect for persons as an asymmetrical regulative norm, our conceptual scheme would be different from what it in fact is. In the case of human agents, our sympathy regarding the limits of human self-sufficiency regulates our concern for noninterference, the overall effect of which is a regulated conception of respect for human autonomy. Again, the regulative effect is brought about by the fact that we care in different ways for human beings. Were we to accept concern for noninterference as an asymmetrical regulative norm, our conceptual scheme would be different from what it in fact is. Thus our concept of respect for autonomy includes sympathy for humans as limited in autonomy, which, to my mind, clearly reveals that our conception of respect for persons has undergone its own filtering process.
But it is not enough simply to assert this. To do so would be simply to assume the truth of the very thing at issue, namely, that our sense of respect for persons is filtered through our sympathy for persons. I must give some principled reasons for thinking that the animal case generalizes in an important sense to the case of persons.
In this regard, consider an example very similar to Kant's own example in The Groundwork for the Metaphysics of Morals . Imagine a person walking along the shore of a lake, when suddenly she hears someone crying for help. Immediately it is clear both that if she does not help the person will drown and that the only cost to her if she does help will be the minor inconvenience of getting her clothes dirty. There is no risk to her at all, and there are no other pressing considerations. Now most of us surely believe that the person not willing to help in such circumstances lacks something of
[12] . Kant, Groundwork , 90.
moral importance. Yet let us assume that though she would not be willing to help, she would nonetheless take great risks to avoid interfering with the rights of people to express their political views. What is lacking here?
One might say that she has respect for people but no sympathy for them. This cannot be the Kantian response, because Kant believes that the categorical imperative, and thus impartial respect for the drowning victim, requires the actions of the Duty of Beneficence in such a case. She cannot, then, have Kantian impartial respect for persons in such a case and feel free to stand on the excuse that she is not interfering, though she is not rendering aid either. Another possibility is that she has neither respect nor sympathy for people. But if this were true, how could we explain that she is quite willing to take on significant risks not to interfere with free political speech? What we should say here, I believe, is that she has a distorted conception of impartial respect; moreover, I believe that Kant would say this as well. If this is correct, then she is acting irrationally due to a distortion. The question now becomes, What undistorted form of practical consciousness is there that would make acting in accordance with the Duty of Beneficence practically rational for her?
We have already given one reason, namely, that she has a conception of impartial respect that is regulated by a conception of impartial sympathy. The effect of this is that her respect for human autonomy is filtered through her sympathy for the limits of self-sufficiency, a very straightforward, utterly familiar sentiment. Kantians, however, must give a different account. They must distinguish between "filtering," on the one hand, and "taking account of the special circumstances of human agents," on the other, and argue for the latter to secure the purity of practical reason and the categorical status of moral demands.
Kantians can take this line, of course, only on the condition that there is a universalizability test that yields the right results and that the test does not involve the filtering of impartial respect. Here they might appeal to Kantian constructivism, or to the constructivist model of impartial respect.
The constructivist model can be understood as like the one found in John Rawls's A Theory of Justice .[13] Though Rawls himself did not intend the model to be applied to morality in general, it is instructive to consider how it might be applied in the current context. On the proposed application, we
[13] . See John Rawls, A Theory of Justice (Cambridge, Mass.: Harvard University Press, 1971), esp. pt. 3, 118–94; Stephen L. Darwall, Impartial Reason (Ithaca: Cornell University Press, 1983), 220, 230–31, 240–49. Darwall believes that the Kantianinterpretation of Rawls provides the basis for a general conception of practical reason.
can imagine self-interested practical reasoners operating with limited knowledge behind a veil of ignorance faced with a choice between the Duty of Beneficence and a simple Duty of Noninterference. The reasoners in the original position have general knowledge but no specific knowledge regarding their special circumstances. The general knowledge includes all the general facts of human psychology, especially those related to the limits of human self-sufficiency, but no specific knowledge of their own psychological characteristics. Finally, the reasoners are risk averse, making it rational for them to adopt strategies that avoid worst possible outcomes. Would it be rational for them to choose the Duty of Beneficence over the simple Duty of Noninterference? Given the choice situation and the assumptions regarding the reasoners, I believe that it would.
Kantians will think that this proves that there is a purely rational derivation of the Duty of Beneficence from the concept of rational agency per se and the special circumstances of human agents. Thus one need not appeal to sympathy as a basis for the practical rationale for the Duty of Beneficence. Since the principle is established merely by appeal to the results of an intellectual procedure—an interpretation of the CI procedure—there is no need to ascribe regulative effect to impartial sympathy on our conception of impartial respect, and the asymmetrical regulative function of impartial respect is preserved for our conception of practical reason. Kant is then vindicated, and pure practical reason prevails.[14]
There is, I believe, a decisive response to this. It is that whether one comes out with the rationality of the Duty of Beneficence or the Duty of Noninterference turns on how the decision procedure is set up in the first place. By making the reasoners rationally self-interested, by excluding special knowledge, and by assuming certain attitudes toward risk, we assured the result we wanted. Yet in doing so, we made the reasoners deliberate in ways that assured sympathy, just as we made them deliberate in ways that assured respect. Notice that God, who has neither inclinations nor interests, is not a party to the contract. We are only able to get the desired result by appealing to a conception of agency that includes as an essential part of agency itself the having of interests. When we control for special self-knowledge by implementing the veil of ignorance, we transform self-
[14] . I believe something like this is the import of the view suggested by Stephen Darwall regarding sympathy in Impartial Reason ; see esp. 173 ff.
interest into generalized sympathy. The assumption of aversive attitudes toward risk ensures the separateness of persons and impartial respect. The conception of agency we have instantiated in the model, then, is not the reasoning of a pure practical reasoner, a God who does not have desires and sentiments, but that of a sympathetically respectful and a respectfully sympathetic agent, a being whose psychological apparatus includes both sympathy and respect as sentiments essential to its capacities of practical reason.[15] But this is filtering; it is not the application of a purely intellectual universalizability test to the special circumstances of human agents. That it is filtering is made clear by two facts: we could not have set up the model to get any results had we not made the persons in the original position interested behind the veil of ignorance, and we could have gotten different results by setting up the model with other constraints. The first point shows that we do not have a conception of rational agency without desires and sentiments, and the second point shows that only some forms of rational agency yield the desired results. These two facts give us principled, sufficient reasons for thinking of the regulative functions between sympathy and respect for persons as being analogous to the relationship between sympathy for lower animals and respect for persons. For the only way we could set up a constructivist model to cover the case of lower animals would be to make the persons in the original position care about lower animals, and this would clearly be filtering, not the application of a purely intellectual procedure to the special circumstances of lower animals. It is in this very important sense that the analysis of the case of lower animals generalizes to the case of sympathy and respect for persons.
Could we have staged a decision context in which the reasoners were concerned with noninterference but not with rendering aid? Perhaps we could have by employing some notion of risk. The reason we do not set up the procedure in that way is because we want our agents to be sympathetic in their understanding of the limits of self-sufficiency. That we can set up the procedure to secure this result only reinforces the fact of the regulative
[15] . Kantians might assert that God is both sympathetic and loving as well as respectful but that sympathy and love are purely cognitive capacities in God's case; they carry no affective or conative functions. They are simply epistemic capacities that allow God to employ the CI procedure in its application to the special circumstances of human agents. But surely this is implausible? We have no awareness of what it would be like to love or sympathize with another person where affective and conative capacities are not in play. It is hard, then, to see this kind of response as anything other than ad hoc.
influence of impartial sympathy on our conception of impartial respect. It does not confirm the Kantian notion of impartial respect as an asymmetrical regulative norm, and this should be reinforced in retrospect by our thoughts about sympathy for lower animals. Lower animals do not have even instrumental rationality, or at least many of them do not, let alone the capacity for the autonomy of pure practical reason. How are they to be parties to the contractual model envisioned by Kantian constructivism? Yet we are firm in our beliefs about the importance of lower animals, which is a clear instance of our impartial respect being filtered through our sympathy for animals.
Kantians, of course, worry that if we make a place for sympathy like this in our conception of practical reason, then we will lose not only the concept of the purity of practical reason but also the guarantee that morality is universal.[16] Those who lack sympathy will lack this conception of impartial respect.
So they will, and there is no way to weasel out of this. In the end, one has to choose between the concept of guaranteed universality[17] and some other concepts. Among these are a conceptual scheme that is true to our actual values—think here again of our sympathy for lower animals—and a plausible philosophy of mind. Kant's views on human volition and action and his general philosophy of mind are controversial even among Kantians.[18] On the other hand, that sympathy and respect, understood as sentiments, have a mutual influence on each other as psychological norms within human behavior is plausible on its face. We understand what it is like to be agents and to a degree self-sufficient, and we understand what it is like to be vulnerable and limited in our capacities. That we care about both is enough to make plausible the symmetry of these two impartial norms, without appealing to the concept of God, to a questionable metaphysics, or to a hypo-
[16] . For the general point about contingencies of human agency and universality, see Onora O'Neill, Constructions of Reason: Explorations of Kant's Practical Philosophy (Cambridge: Cambridge University Press, 1989), 52–55. For more specific discussion of sympathy and universality, see Darwall, Impartial Reason , 162–63,173–74.
[17] . For Kant, universality must be secured by necessity rather than by empirical fact: intersubjectively shared norms cannot provide practical laws, no matter what their content or the scope of their sharedness among humans.
[18] . John Rawls, Political Liberalism (New York: Columbia University Press, 1993), 89–130; Thomas Hill, Dignity and Practical Reason in Kant's Moral Theory (Ithaca: Cornell University Press, 1992), 226–50.
thetical decision procedure. This is not to deny that some hypothetical decision procedure might help us clarify what is rational for us to do, given that it captures the symmetrical relationship between sympathy and respect. What it cannot do is establish a purely intellectual reason for this rationality, for it will be rational for us to accept the artificial procedure only if we are in fact respectfully sympathetic and sympathetically respectful in the way that the procedure assures.
The crucial point of this section, then, is that the Kantian must construe the constructivist model to establish that there is a purely intellectual reason for the Duty of Beneficence because the choice of duties or principles is merely an application of the categorical imperative to the special circumstances of human agents. That this is not the case is reinforced by our discussion of sympathy for lower animals, because the CI procedure, even in its constructivist form, cannot be construed to include lower animals as parties to the social contract and hence as a special application to their circumstances. Just as our sympathy for lower animals sometimes regulates our concern for the autonomy of human beings, our sympathy for human beings sometimes regulates our respect for the autonomy of other human beings. And I can see no reason that a similar analysis does not apply to the test for imperfect duties sketched by Kant in the Groundwork , out of which the constructivist model is built.[19]
Kant's universalizability test for imperfect duties, which is the test that would apply to the kind of context we are considering, is this: as a rational being who has respect for self and others one must be able to will that any rational being live in a system of nature in which one's maxim is a natural law. The imaginative thought project here cannot be carried out by someone who is not sympathetic in his or her respectful response to the plight of others. At least, this must be true if the duty to render aid is to be the rational result. To secure the result we want regarding the plight of lower animals, we would have to extend the imperfect duties test to cover nonrational, lower animals in our concern for the environment in which creatures have to live. We have, then, the same reasons for thinking of the imperfect duties test as filtering as we do for the constructivist model. Hence the argument regarding the constructivist model applies without dissimilarity to the imperfect duties test. Whatever plausibility these tests have is due to the fact that they filter our sentiments, including impartial respect; it is not due to the purely intellectual application of a procedure to the spe-
[19] . Kant, Groundwork , 89–92.
cial circumstances of human agents. To a large extent, this explains why there is sometimes disagreement about what the results of the procedure are for particular cases among people of similar intelligence but different sentiments. It also explains why the purely formal aspects of the procedure provide little in the way of guidance to someone who has no sentiments, which is to say that such a procedure would provide Kant's God with little direction. We should be very suspicious of theories of practical reason that rely on appeals to God, and as we see here, when we eliminate God from the picture we see that filtering is what is secured by the tests.
7.
Thus far we have seen that impartial norms regulate partial norms and that impartial norms regulate each other in our conception of practical reason. Regarding the latter, we have seen that impartial norms are symmetrical in their regulative functions, both as they apply to relationships between persons and lower animals and as they apply to relationships among persons themselves. I turn now to the issue of whether our conception of practical reason reflects symmetrical regulative functions between partial norms and impartial norms regarding relationships among persons. If it does, then our conception of practical reason does not give a place to impartial norms that are asymmetrical in the way required by both Kantianism and other modern moral theories.
I will consider impartial respect and the partial norm of parental love and argue that they are symmetrical in their regulative functions. Once again, I will argue that there is something very important about the analysis of sympathy for lower animals and respect for persons that generalizes to the analysis of parental love and impartial respect. I leave it to the reader to extend the analysis to other partial norms. My goal in this section, then, is to clarify the problem facing the Kantian view. The goal of the next section is to provide an argument against a Kantian solution.
That impartial respect regulates parental love is not difficult to establish. There are many contexts in which if we favor our own children over others we will be acting disrespectfully. Moreover, in those very same contexts, our favoring others over our own children is not betrayal, though it would be were it not for our respect for others. My serving jury duty might come at some cost to my child, though my exacting that same cost for other reasons might constitute betrayal. In some contexts, then, refusing to serve jury duty is disrespectful and is not required by parental love, though it otherwise would be.
The more controversial issue is whether impartial respect exhibits the regulative effect of parental love. I believe that it does, and if I am right, we accept the truth of the following claim.
Impartial Respect is regulated by Parental Love just because there is some practical context in which (i) acting on Impartial Respect is irrational (wrong) because it violates Parental Love and (ii) were it not for Parental Love it would be rational (permissible) to act on Impartial Respect.
Consider the following case. I can either send my daughter to college where she can get the best education consistent with her abilities or use that same amount of money to send several other youths to college. An important difference is that my child has been accepted to several of the best private and public schools in the country. The problem for the other students is that they cannot afford to go to the more expensive superior schools, though these students have also been accepted. There are other schools to which these other students might go, however, that are quite good and less expensive, but the students and their families cannot afford them. Morever, these other students are as deserving or maybe even slightly more deserving than my daughter and there are more of them than there are scholarships and loans to support. There are even other very deserving students, a greater number of them, who were unable to get into the very top first-and second-rung schools but who have been admitted to some very good to decent schools, none of which they can afford. Now I can either send my daughter to the best school I can afford that is consistent with her talents or I can support a greater number of students at the second-rung schools or even a greater number at the third-rung but still decent schools. Now assume that the talents of my daughter are such that there would be a significant difference in the results of her education at the very best schools than at the second-best schools and that this is also true of some of the other students. Assume, also, that everything else is equal. That this is a realistic case in terms of costs can be confirmed by checking the widely different costs of higher education, depending on the institution involved. The example is selected because many of us in higher education have talented children who put us in the position of making just such a choice.[20]
What would it be rational for me to do if I am both a respectful person
[20] . I thank James Klagge for suggesting this kind of example.
and a loving parent? Would my sending the other students to college be a betrayal of my daughter's love? Would my sending my child to the best college consistent with her abilities be disrespectful to the other children? Surely the answer to the second question is yes and to the third question no, both of which provide an answer to the first question. A conceptual scheme that cannot allow for the fact that an impartially respectful but loving parent is significantly more committed (but not to an unlimited degree) to his or her own children than to others condemns itself to irrelevance to loving parents. I take it, then, that this example is firm in its result regarding what would be the rational and right thing to do, given the circumstances.
Now we must see if this example presents any problems for the Kantian view of impartial respect as a regulative norm that is asymmetrical in its regulative effect on the partial norm of parental love. A clear rationale for the priorities is available on a conception of practical reason within which impartial respect and parental love are symmetrical in their regulative functions vis-à-vis each other. The jury duty example illustrates the regulative influence of impartial respect on parental love, and the education example illustrates the regulative influence of parental love on impartial respect. Conditions (i) and (ii) of the counterfactual condition are met in both directions, which is the mark of symmetrical regulative function.
Kantians, however, might, in an attempt to preserve our values, argue for the rationality and rightness of sending one's own child to college in such circumstances. This would seem to defuse the example. And indeed it would if the argument to this effect were to meet certain conditions. To do so, the argument must first establish precisely how impartial respect expressed in the CI procedure as applied to the context would require sending one's own child to school and not using one's money for the other children. This is not a case of it being merely permissible to favor one's own. The Kantian position seems plausible as long as the focus is on contexts involving merely permissible actions. We are thinking here of parental obligations, not mere permissions. Moreover, the argument must be constructed in a way that does not employ a conception of respect that has already been filtered through considerations of parental love. Just as Kantians can rightly insist that a conception of parental love that has already been filtered though considerations of impartial respect cannot pose a threat to the moral status of impartial respect, non-Kantians can insist that a conception of impartial respect that has already been filtered through considerations of parental love cannot pose a threat to the moral status of parental love. How, then, can Kantians ensure against such filtering? There are only
a few possibilities, which represent different conceptual models for understanding impartial respect as an asymmetrical regulative norm.
The Simple Noninterference Model
One strategy mentioned before involves a minimalist view of impartial respect. It depends on an understanding of respect that requires only noninterference and nothing in the way of positive aid, where failing to promote is not counted as interference. On this view, as long as my sending my child to college did not interfere with the other children and their education, I would not be showing disrespect for them by not coming to their aid. This being true in the present case, I would be free to send my own daughter to college.
There are several problems with such a view, two of which are most relevant here. First, it is clearly not Kant's view, which should be evident from previous discussion. Only a caricature could saddle Kant with this model. Second, it fails, as Kant saw, because it is not, in some respects, demanding enough. Kant's worry was that it does not demand enough in terms of positive duties to render aid, and he was surely right about that. Therefore, it is enough to reject this view that it is not demanding enough in terms of impartiality alone.[21] Moreover, there are few Kantians or other impartialists who want to defend anything like this model. I include it only to indicate the counterintuitiveness of minimalist moralities.
The Strict Egalitarian Model
A more complicated model is more demanding and has an initially intuitive appeal. On this view, impartial respect would have us treat everyone on a strictly equal basis, which, though strictly egalitarian, would not guarantee sameness of outcome. Anticipating differential outcomes, I might rationally favor one person over another while treating them equally from the perspective of impartial respect. The idea is something like the following. Persons are respected equally when the relative importance of their interests from their own points of view are given equal weight in practical deliberation from the moral point of view. Suppose we distinguish interests. lexically and roughly as first order, second order, and third order interests.
[21] . There is another sense in which such a morality would be too demanding. If we could never justify interfering with the autonomy of others, we simply could not get on with much of anything. On any plausible view of noninterference, we can interfere even if it is only to minimize noninterference. It is notoriously difficult, however, to specify a rationale for mere noninterference that does not lead to positive duties.
The idea is that first order interests are those that are the most important from a particular person's point of view (their ground projects, what life is most about for them); second order interests are the second most important from that person's point of view (very important but not absolutely central interests); and third order interests are the least important from the personal point of view (comparatively minor interests). The idea, as suggested by Korsgaard's Tanner Lectures, is that a person's autonomy is gauged best in terms of the concerns that factor most centrally in his or her identity.[22] For present purposes, further fine tuning of these distinctions regarding kinds of interests is unimportant, though doing so will be crucial to elaborating the theory of practical reason presented in following chapters. All we need to note here is the general conception of equality on such a view. It is this: To treat two persons equally and with respect is to give equal weight to their interests as long as the interests are of the same category on the lexical scale. Otherwise, disputes are resolved by giving first priority to first order interests over second order interests and to second order interests over third order interests. This allows strict equality of consideration but differential outcome, a plausible view in many deliberative contexts.
Moreover, it is consistent, in many contexts, with being both a loving parent and an impartially respectful person. For example, suppose I can satisfy some minor interest of my daughter but only at the cost of some vital, first order interest of some innocent third party. I am faced, say, with taking her to the circus or rescuing another child from a burning building. How can it be impartially respectful to give priority to my child's interest in the circus over the vital interests of the other child, and on what conception of parental love is this required? The point is that no plausible conception of impartial respect would allow such inequality, and no plausible conception of parental love would require it. What we see in this conception of respect that we did not in the previous conception is a more plausibly demanding impartial norm.
The problem is that though this conception of impartial respect has rational results in many cases, it does not in others. Most notably, it has irrational results for the loving and respectful parent in our education example. Strict equality would require a loving parent to send the other children to school instead of his or her own. This is a direct result of this conception of
[22] . See Christine M. Korsgaard, The Sources of Normativity (Cambridge: Cambridge University Press, 1996), 17, 18, 102–3, 73. Korsgaard does not endorse anything like Strict Egalitarianism. Such a view is closer to David Cummiskey's view in Kantian Consequentialism (New York: Oxford University Press, 1996).
strict equality, even construed as not requiring sameness of outcome: to send one's own child on these assumptions would afford a disproportionate weight to the fact that it is the interests of one's own child that are at stake. Once we make this conception of impartial respect an asymmetrical regulative norm, there seems little place for the most commonplace priorities of partial sentiments. That this is not our conception of impartial respect is evident in the sentiments of people who love their children and yet do a good deal for other people. Although their priorities do not reflect such strict egalitarianism, we admire them for loving their children in the way they do. We cannot think, then, that our education example satisfies conditions (i) and (ii) of our claim and accept the strict egalitarian conception of impartial respect.
Of course, these observations refute Kantian ethics only if this is the only sense in which impartial respect could be asymmetrical in its regulative function vis-à-vis other norms. To conclude this at this point, however, would be premature. What, then, are the other possibilities?
The Indirect Model
Someone might object that our rejection of the strict egalitarian model has been hasty, for there might be an indirect route to the priorities reflected in our example within the strict egalitarian model itself. It should be noted that the strict egalitarian model accommodates a consequentialist framework very well, though not utilitarian in nature. When problems can be solved simply by appeal to the lexical ordering of interests, maximizing is not at issue. But when conflicts are within one category of interests on the lexical scale, resolving disputes by a maximizing strategy is, from an impartial viewpoint, quite rational, everything else being equal. Recognizing this, one might argue that, given human variables, the maximal degree of equality achievable among persons is best brought about when loving parents act on the priorities reflected in our example and in our claim regarding regulative effect than when they act directly on the strict egalitarian conception of impartial respect.
This is an old story originally employed by utilitarians and still with much currency among many moral philosophers today.[23] I will not pursue it very far, except to point out that it runs directly into Stocker's problem.
[23] . See Peter Railton, "Alienation, Consequentialism, and the Demands of Morality," in Friendship: A Philosophical Reader , ed. Neera Kapur Badhwar (Ithaca: Cornell University Press, 1993), 211–44; and Cummiskey, Kantian Consequentialism .
Herman seems to want to avoid such a strategy just because she recognizes that it would result in an objectionable moral psychology.[24] Just as people who care about animals in ways that we respect do not care for them indirectly but directly, loving parents care for their children directly and their reasons for acting on their behalf are direct. To endorse indirection here is to revise or abandon our conceptual scheme, and recognizing this was part of the original motivation Kantians had for bringing attention to regulative norms in the first place. Resorting to indirection, therefore, seems to admit the defeat of the original strategy and to sacrifice its original insight, which is considerable.
At this point, then, we can say that whatever the Kantian reply is it must satisfy the following conditions: it must (i) secure the results of our example, (ii) reflect a more substantial conception of respect than found in the Noninterference Model, (iii) avoid the results of the direct application of Strict Egalitarianism, (iv) avoid the distorting effects of Indirect Egalitarianism, and (v) achieve all this without filtering the concept of impartial respect through considerations of parental love.
8.
How to construct the constructivist model is problematic in itself. The reasons are that Rawls himself did not intend the model to cover the entire scope of morality and other attempts have either been found inadequate or incomplete. There was an early attempt by David A. J. Richards in A Theory of Reasons for Action , which has received little support, and a later sketch by Thomas Hill, Jr., which by his own admission is incomplete.[25] Neither has been taken up by recent writers in defense of Kantian impartiality against partialist attacks. Rawls's model is designed for the choice of principles governing basic social institutions. The problem here involves norms that govern one's personal life, in many ways a much messier affair, and capturing these complex nuances is the major part of the difficulty. The best that I can do here is to construct the model in a way that avoids caricature and is sufficient to provide guidance in regard to evaluating the current issue.
[24] . Herman, The Practice of Moral Judgment , 161–62, 173.
[25] . Thomas Hill sketches an attempt at expanding Rawls's constructivist model, but I do not see how to employ it in the current context. What I do in what follows is an attempt to alter the model in reasonable ways to accommodate the issue. See David A. J. Richards, A Theory of Reasons for Action (Oxford: Clarendon Press, 1971); Hill, Dignity and Practical Reason in Kant's Moral Theory , 243–50.
I will be guided by two crucial thoughts. First, not only should the model yield a result that is rational given the constraints of the choice situation; it must yield the result that it is rational for persons to act on the norms that would be chosen in the hypothetical situation in actual situations with all the known facts. The model is simply a means of elucidating our deepest values. This, I believe, is a feature of Rawls's view as well. Otherwise, the norms are not rational for agents like us but for some other kind of agents. Second, the model must not obviously filter for parental love.
We must characterize the choice situation in terms of an original position in which rational agents must choose from among a list of principles to govern cases. The original position will include a veil that filters out all irrelevant specific knowledge and allows only relevant general knowledge. Much of the problem is in understanding what is to count as irrelevant specific knowledge and relevant general knowledge. I will consider three different characterizations in this regard and evaluate each, arguing that none is successful.
I will assume aversive attitudes toward risk for those in the original position, for without them we are not likely to get anything like a Kantian result. For similar reasons, I will assume the Rawlsian maximin strategy of avoiding worst outcomes, which assumes that worst outcomes are most probable. If we assume equiprobability of alternative outcomes under conditions of uncertainty, we will end up with something like utilitarianism and be far from anything Kantian, as John Harsanyi has already shown.[26]
Now imagine that these reasoners are presented with a choice of norms to govern their lives once the veil is lifted. Here we are concerned with the rationality of a choice of those in the original position between the following alternatives.
a. Simple Noninterference with autonomy;
b. Direct, Strict Egalitarianism;
c. Indirect Egalitarianism; or
d. Limited Partialism (the priorities involved in our education example).
A necessary condition for the Kantian rejection of our example as establishing the symmetry of regulative functions between impartial respect and
[26] . John Harsanyi, "Morality and the Theory of Rational Behavior," in Utilitarianism and Beyond , ed. Amar tya Sen and Bernard Williams (Cambridge: Cambridge University Press, 1982), 39–62.
parental love is that the reasoners would choose (a) Limited Partialism. It is necessary because Noninterference is not robust enough, Direct, Strict Egalitarianism does not establish the results of our example, and Indirect Egalitarianism runs afoul of Stocker's problem. Allegedly, it would be sufficient, baring other difficulties, because it would establish those priorities through a procedure that did not build the partial norm of parental love into the decision procedure itself.
First Construction
The first way of constructing the model employs all the above conditions and assumes that all the parties in the original position are parentally loving and that the veil of ignorance does not filter out this specific knowledge. Now it does not take much imagination to see that constructing the model in this way will lead to the desired result of their selecting Limited Partialism over the other principles. But, then, it is equally easy to see that the first construction filters for parental love in a way that does not serve the Kantian view.
We should not think that, relative to the current issue of partial norms, the constructivist model can be treated the way it was relative to the issue of the symmetrical functions of sympathy and respect. The reason is that we must distinguish between being self-interested, which all the parties to the choice are, and different psychologies that embed different partial norms. When we make the reasoners self-interested behind the veil of ignorance, we effectively make them sympathetic reasoners. Thus we get conceptions of respect and sympathy that are symmetrical in their regulative functions relative to each other. For the moment, I propose that we bracket this philosophical conclusion about the symmetrical function of respect and sympathy and consider other results of the model. Relevant here is the fact that the reasoners in the current case will only know that among the different psychologies they might have after the veil is lifted is one that includes parental love. It might also exclude it, ratherdrastically. If we are to capture what it is to respect people with different ways of life autonomously chosen, we cannot favor one way of life over another in the way that the choice situation is set up. Thus there are many ways of life with many attendant psychologies, and the model must honor this if it is to remain Kantian. Accordingly, the reasoners in the original position are sympathetic, but they are not parentally loving. Of utmost importance is that the original position be set up in a way that makes the reasoners sympathetic and respectful but neutral between conceptions of ways of life that include
parental aspirations versus conceptions that make little or no place for such relations. Behind the veil, the reasoners are not psychologically constituted in any way specific to a way of life. They know only that when the veil is lifted they will have some specific psychology fitted to some particular way of life, ranging from highly communal ways of life to highly individualistic ones that include no love for children. For this reason, the first construction is to be rejected: though it will yield the right decision, it is set up in a way that it filters for parental love.
Second Construction
Now consider a second construction in which the veil of ignorance filters out all specific knowledge of individual psychology and that all the other features of the first construction remain.
Enough has been said, I believe, to rule out Noninterference as a rational choice, but I believe that Limited Partialism is irrational as well, given the choice assumptions. The reason is quite simple in the end: Direct Strict Egalitarianism ensures best against worst outcomes. To be sure, there are many contexts in which Limited Partialism and Strict Egalitarianism will yield consistent results. Since Limited Partialism has been filtered through the regulatory functions of a conception of respect, it will be consistent with those contexts in which the interests of strangers take priority over the interests of loved ones, namely, those contexts in which the stranger's interests are higher up the lexical scale than those of the loved one. Moreover, impartial respect will require that priority in these contexts is given to the interests of strangers. Yet if the order is reversed, if the interests of loved ones are higher up the scale than the strangers, loved ones will get priority. This is true on both Limited Partialism, as I intend it here, and Strict Egalitarianism.
They will differ where the conflicts involve interests within the same category on the lexical scale, most poignantly, when the conflicts involve first order interests. Strict Egalitarianism will maximize here. It will require acting in a way that brings about the greatest overall equality, which means giving priority to that option that will accommodate the most persons with their first order interests satisfied. Limited Egalitarianism will not maximize, though it will be sensitive to avoiding what we can call catastrophic outcomes, outcomes where extremely large numbers of people have their first order interests and hopes destroyed. To test this last point, merely think of what it would be like to reject the counterfactual that one would sacrifice one's self and loved ones in order to save the world. This is different from
rejecting the counterfactual that one would sacrifice one's own first order interests and those of one's loved ones in order to marginally increase overall equality. Strict Egalitarianism and Limited Partialism are different in this important regard. What remains is the issue of worst outcomes.
If the reasoners are risk averse and maximin strategists, they will choose that option that is most probable to minimize worst outcomes for themselves once the veil is lifted. Strict Egalitarianism is preferable to Limited Partialism in this regard, for it will minimize the number of people who have their first order interests unsatisfied where the conflict is between first order interests alone. Thus, barring other considerations, the parties can know through purely general knowledge that it is less probable on the Strict Egalitarian view than on Limited Partialism that they will be victims of social inequalities.
But what about these other considerations? Might they not be such as to turn things in favor of Limited Partialism from the point of view of the reasoners in the original position? Now it might be argued that from purely general knowledge the reasoners can know a fact of developmental psychology, namely, that it is highly unlikely that anyone deprived of the benefits of some degree of parental favoritism of the sort reflected in our example (though not necessarily involving the issue of education) will develop in a way that will make him or her an autonomous chooser of a way of life. If this is true, then the autonomous choice of even a highly individualistic way of life requires the benefits of Limited Partialism. Thus from general knowledge from behind the veil of ignorance the reasoners would choose Limited Partialism and reject Strict Egalitarianism.[27]
The response to this is that even assuming that the reasoners in the original position would make such a choice with such general knowledge, this construal of the rationality of Limited Partialism is the result of indirection. It is, in fact, Indirect Egalitarianism. From the point of view of the original position, the concern for parental interests is, on this argument, an indirect concern for equality. But, once again, this way of arguing defeats the very purpose of introducing the concept of regulative norms in the first place. It was, after all, the original purpose of introducing the concept of regulative norms to allow for the rationality of direct partial concerns. To resort to indirection is to admit that the notion of regulative norms cannot deliver on what was originally promising about it.
[27] . I would like to thank Dwight Furrow for discussion on this point.
Third Construction
The third construction alters the second without admitting that it is a general fact of developmental psychology that to become autonomous choosers requires parental favoritism in children's upbringing of the sort displayed in our example. It only admits that it might be in certain cultures and that the reasoners can know this. Moreover, the reasoners are allowed to know this specific fact of social psychology that theirs is such a culture. Now, some might object that the construction cannot include this and remain Kantian. But this will not be my objection. I see no problem at all with allowing this within the Kantian framework. And with this assumption, there is a way of arguing directly to Limited Partialism. It is this: The reasoners in the original position reason disjunctively that either they will have parental interests of the sort involved in parental love or they will need the benefits of parental favoritism to develop into autonomous choosers of a way of life. Either way, Limited Partialism is preferable to the other principles on the list. What this construction captures, I believe, is that we respect people who are loving parents and people who have no interest in having children. The disjunctive reasoning captures this, in my opinion; hence there is nothing objectionable in the indirect reasoning of the second disjunct.
What, then, could possibly be wrong with the third construction? The only answer could be that it involves filtering. The Kantian must argue that the model is designed to take into account the special circumstances of human agents, and the non-Kantian must argue that the model filters in the first disjunct for parental love. If we confine ourselves to the current issue and ignore the issue of the regulative influence of impartial sympathy in the previous discussion in the case of both lower animals and the Duty of Beneficence, it might seem that there is a standoff, that there is no way to choose between these accounts. I believe, however, that the observations about sympathy for lower animals and the argument for the regulating effect of sympathy on our conception of respect for human beings themselves should lead us to see that the model filters rather than applies to the special circumstances of human beings. The philosophical account in all these cases is, on such a construal, the same. There is no philosophical motive to change the structure of the account. The way in which we generalize from the previous accounts of sympathy (first, for lower animals and respect for persons and, second, for sympathy for persons and respect for persons) to the account of parental love and respect for persons is that there is no philosoph-
ical motive to change the structure of the accounts. Once we recognize that the symmetrical nature of the regulative functions between sympathy and respect is the result of filtering, we have no philosophical motive for not thinking of the regulative relations between parental love and respect for persons in the same way.
Thus we have not discovered in our three versions of the constructivist model one that will both provide the right decision and avoid the other pitfalls. Now, it might be that there is some other construction that will do the work, but it can hardly be said that the argument here has relied on caricature, unless it can be shown that there is a clear model that will provide the desired result and that I should have seen rather clearly what it is. Baron has a view of imperfect duties that, I believe, is intended to cover such cases, but she has not explained the casuistry necessary to get from the general notion of imperfect duties to a solution to cases like the one presented here.[28] Until she addresses these issues, her view of imperfect duties is as yet only a gesture of a response to the current argument, however adequate it is to Stocker.
Nor will it do to say that problems of this sort would not occur in an ideally just world, that in a just world loving parents would not have to make such decisions. First, I am not at all sure that this is true, but I am sure that the ethics we need is for a world in which such decisions have to be made. One might object that ethics needs to deal with nonideal circumstances, but this does not show that we can argue straightforwardly from what it seems right to do in nonideal circumstances to the principles that justify norms. Nothing in this area, it might be said, is as simple as I make out. My response is that I am not the one who is assuming that things are simple. Along with Isaiah Berlin, I do not believe that there is any coherent notion of an ideally just world. I might be wrong about that (as Berlin might have been) but it is just simpleminded to assume that such a world is coherent without giving an account of its coherence, and Kantians have not given such an account. To allude to an ideally just world to avoid the present difficulty, then, is merely to gesture in the direction of a response, and it is this sort of gesturing that non-Kantians want Kantians to avoid as much as Kantians want their critics to avoid caricature. Moreover, even if an ideally just world is possible (which, again, I doubt), what is its relevance to nonideal circumstances? We do not live in that world, and even if it is possible it is not clear in what sense it is normatively relevant. In any event, to ap-
[28] . Baron, Kantian Ethics Almost Without Apology .
peal to an ideal world either without an account of its possibility or without an account of the path from its possibility (routed through current circumstances) to its actuality is to refuse to consider other alternatives.
9.
Finally, someone might object that my argument only shows that impartial respect is not a norm that is asymmetrical in its regulative effects, but it does not follow from my argument that there is no impartial norm that has such a function. Here the thought is that the Kantians have erred simply in locating the wrong impartial norm.
My response in this regard is brief and to the point. First, this objection is not open to the Kantian. Thus even if I have not established the larger point, I have nevertheless made an important one. Second, the objection itself might arise from two sources, both of which are questionable. One source might be an a priori commitment to the notion that the unity of practical reason requires a norm that is asymmetrical in its regulative effects. But, of course, the veridicality of such a commitment is anything but self-evident, nor can I see that it is even plausible. Nevertheless, if it can be shown that the very notion of the unity of practical reason requires a norm that is asymmetrical in its regulative functions, I will, of course, retract my larger claim. The burden, however, is on those who are committed to the a priori view. The other source of the objection might be some concrete suggestion about a particular norm that is claimed to have the relevant asymmetrical function. If so, then it will have to be demonstrated that the suggested norm meets the requirements of the criteria set out here. I have found no such norm, but I will gladly accept that there is one if an argument can be presented that demonstrates its presence within the practical reason of those most admirable. However, once again, the burden is on those who would claim that there is such a norm.
10.
What are we to make of these arguments? First, we should note that they strongly suggest that central features of our normative framework do not fit well within the Enlightenment tradition as it is tied to asymmetrically dominant, impartial norms, especially the Kantian version of that tradition. Second, we should note that our considered moral judgments as they are reflected here support the second condition of the Aristotelian conception of practical reason: All norms within practical reason are symmetrical in their regulative functions. However, this is not enough to establish the
Aristotelian conception of the integrative function of consciousness. For this, we need to be able to show how the features of our normative framework have a rational basis in the goods that make life meaningful for us from our own points of view and how reflective endorsement by means of the criteria of finality and self-sufficiency leads to these norms. We will then be in a position to employ the integration test. Before we can do this, we need to understand how the thin conception of integrity illuminates the relationships between the goods of an agent's life and his or her reasons for action.
3—
The Thin Conception of Integrity and the Integration Test
1.
Any conception of morality that begins with practical reason, whether Kantian or Aristotelian, inherits several burdens. One is the assignment of a functional role to practical reason within an overall psychology. As we have seen, that role for both the Kantian and the Aristotelian models must be understood in terms of the norms of practical reason and their regulative functions. Another burden is to provide the relevant test for whether the function assigned to practical reason is actually instantiated within the psychology of human beings. I have claimed that the integration test is the appropriate test for determining whether any candidate for meeting the first burden is successful. I have also claimed that in order to run the test, we need an adequate philosophical psychology. It is the purpose of this chapter and the next to provide such a psychology and to show how it is responsive to the integration test. The basic categories of the psychology are set out in this chapter and related to how they facilitate the integration test. In the next chapter, the categories are put to use in a general conception of practical deliberation.
The guiding thought here is that an adequate philosophical psychology is one that allows us to have an understanding of practical reason in human beings who are agents of integrity in the thin sense. The overall function of practical reason, then, is to be spelled out in these terms. The categories of the philosophical psychology are to be accepted only to the extent to which they illuminate how practical reason could have a regulative function within the psychology of a rational human agent who has integrity in the thin sense.
On any plausible view, there can be many different kinds of persons with
many different kinds of character who are agents of integrity. Mother Teresa and Eleanor Roosevelt, Martin Luther King, Jr., and Malcolm X, Mahatma Gandhi and Field Marshal Rommel, Voltaire and Sir Thomas More—all have been noted for their integrity, and all are very different sorts of people. But to qualify as agents of integrity at all, they must satisfy the minimum criteria of the thin conception: They must have a sufficiently unified personality to have at least one and no more than one basic "self," a level of self-knowledge inconsistent with a life of self-deception regarding their basic sense of self, the strength of character to meet significant challenges to that which makes their sense of self possible, and a sense of self-worth as separate and numerically distinct persons. And, as ongoing agents, they must have positive reasons for living. The goal here is to provide a general way of understanding how the basic elements of the thin conception of integrity are possible. For it is only within the conditions for the possibility of integrity in this sense that a functional role for practical reason can be assigned. It is because that role is an integrative task that the appropriate test for competing conceptions of practical reason is the integration test.
Finally, we should expect an adequate philosophical psychology to include categories that empirically accommodate facts about both cognitive and clinical dimensions of human psychology. What is needed is a philosophical psychology that reflects the psychology of a personality in which practical rationality has a definitive role. No psychology that included only cognitive categories could yield a conception of a human person. For this, we need a philosophical psychology that can yield a notion of practical reason with pathos . We need categories that illuminate how a psychology can have energy and a core, how it can flourish and how it can break down. And it is ultimately by empirical observations of how a personality so conceived is successfully integrated that we can apply the integration test to adjudicate between rival conceptions of rational agency.
2.
Among the most fundamental things to understand about a human agent of integrity is that such an agent has interests and has a way of setting priorities among them. Without a strong sense of priorities, the basic elements would be impossible, and, without interests, the basic elements, even if possible, would have no point. Thus it is in terms of interests that a psychology has pathos, and, as we will see, it is in terms of distinctions among interests that a psychology has the structure of a personality.
Elementary among these interests are vital interests and the hedonic interest in avoiding unnecessary pain. Vital interests are the interests in biological life and the necessary means to it. A bare interest in biological life would be an interest in being alive in the biological sense without any consideration for the quality of that life. But it is doubtful that anyone has such an interest. If we distinguish between a person's "taking an interest in" something and something "being in the interest of" a person, it is plausible to say that a person has vital interests—in the sense of "taking an interest in" biological life and the necessary means to it—only if a person has other kinds of interests. Also, it seems plausible to say that a potential person (such as a fetus) has vital interests—in the sense of something "being in the interest of" that potential person—only if it potentially has other nonbiological kinds of interests.
If we were to imagine a human fetus developing into an insect rather than a person under drastic environmental changes, we would hardly maintain under these conditions that its genetic endowment suffered no loss of importance. Also, if we were to discover some new biological species, we would conclude that its life was of intrinsic rather than mere instrumental value only if we held some beliefs about the quality of the life it could live. If it could live only a life of excruciating pain, there would be little sense of intrinsic loss and a great sense of relief on its extinction.
The life of mere excruciating pain, then, is reason enough for not taking an interest in biological life. Perhaps this is why burning in an eternal fire is the most vivid picture of hell. On the other hand, freedom from pain alone does not generate vitality. Schopenhauer seems to have thought that many people will to go on living in the most dire circumstances, even when they have given up hope.[1] But if it is not hope that keeps them going but mere habit, then it is not for something, let alone for a reason, that they go on living. Yet a person of integrity is a person who, under normal circumstances, has reasons for living—reasons founded in interests other than the interest in avoiding unnecessary pain.
Having positive reasons for living, however, is not sufficient to explain the basic elements of integrity. What we need is a way of understanding the psychological structure among an agent's interests before we can see how the elements of integrity can emerge. For this we need to understand the
[1] . See Arthur Schopenhauer, The Will to Live: Selected Writings , ed. Richard Taylor (New York: F. Ungar, 1967), and The World as Will and Representation , trans. E. F. G. Payne (Indian Hills, Colo.: Falcon's Wing Press, 1958).
relationship between the agent's interests and the agent's sense of priorities as they relate to his or her reasons for living.
3.
The interests of the person of integrity associated with his or her reasons for living can be called categorical interests. They are interests apart from which there is a serious loss in the unity and meaning of one's life from one's own point of view. Sometimes this loss of meaning results in a loss of the will to live at all.
Aristotle shows a keen awareness of these kinds of interests in the Eudemian Ethics , book 1:5, where he says:
About many things it is not easy to judge correctly, but it is especially difficult to do so in regard to that which everyone thinks is most easy and within anyone's capacity to know; namely, which of the things in life is worth choosing, and such that one who obtains it will have his desire fulfilled. After all, many things that happen are such as to induce people to abandon life—disease, extremes of pain, storms, for example; so that it is evident that, on account of those things at any rate, it would, given the choice, have been worth choosing not to be born in the first place. Again, [there is] the life which men lead while they are still children. For no one in his right mind would tolerate a return to that sort of existence. Moreover, many of the things that involve neither pleasure nor pain, or involve pleasure, but of a reprehensible sort, are enough to make not existing at all preferable to being alive. In general, if we put together all the things that everyone does or undergoes, but not voluntarily (because they are not done or undergone for their own sake), and an infinite stretch of time were provided in addition, no one would choose in order to have them to be alive, rather than not. Nor again would anyone who was not a complete slave prefer to live solely for the pleasure associated with nutrition and sex, if all the pleasures were removed that knowing or seeing or any of the other senses bestow upon human beings; for it is evident that, for a man who made such a choice as this for himself, it would make no difference whether he were born a beast or a man. Certainly the ox in Egypt, which they honor as the god Apis, has a greater abundance of several of such things than many sovereigns. Similarly, no one would prefer life for the pleasure of sleep; for what difference is there between sleeping without ever waking from one's first day to one's last, over a period of ten thousand years—or however many one likes—and living the life of a plant? . . .
They say that Anaxagoras, when someone raised just these puzzles
and asked him what it was for which a person would choose to be born rather than not, answered that it would be "in order to apprehend the heavens and the order in the whole universe." (EE 1215b:15-1216a:15)[2]
Notice that the concern in this passage is not with what makes one kind of life more worth living than another but with what makes living preferable to death. Aristotle is here concerned with the most fundamental interests that make the prospects of life at all alluring. To be sure, we would not all give Anaxagoras's answer, but the person of integrity has some such answer.
Following Bernard Williams, we may call the objects of categorical interests ground projects.[3] They are ground projects because they serve to ground the unity and meaning of one's life from one's own point of view. In fact, they are the grounding necessary for a person's having a point of view at all such that we can attribute one basic, significantly integrated self to that person. This is in contrast to attributing no or many selves to a human. And the person of integrity is the human to whom we attribute at least one, and no more than one, basic, significantly integrated self.
Of course, people do not, in any self-conscious way, usually think of those things that play the role of ground projects as being the means to unity and meaning in their lives. They simply care about things in a way that brings focus and meaning. Something is valued categorically in this sense, then, when it is what life is most centrally about for a person and is such that its loss brings with it, at minimum, a serious loss in the coherence of life or, at maximum, a loss in the allure of life itself. The objects of categorical interests, then, appear within one's deliberative field in a very fundamental way (and depending on one's character they attach one to items in the world, a defining mark of objective eudaimonism).
Imagine a woman dedicated to being an excellent surgeon to the exclusion of all else. As far back as she can remember, she has always pursued
[2] . Aristotle, Eudemian Ethics: Books I, II, and VIII , trans. Michael Woods, Clarendon Aristotle Series (Oxford: Oxford University Press, 1992), 4–5.
[3] . See Bernard Williams, Problems of the Self (Cambridge: Cambridge University Press, 1973), Moral Luck (Cambridge: Cambridge University Press, 1981), and Ethics and the Limits of Philosophy (Cambridge, Mass.: Harvard University Press, 1985). See also Williams's use of the concept of categorical desires in "Persons, Character, and Morality," in Moral Luck , esp. p. 11. I attempt to develop the notion of categorical interests in a way that is more closely tied both to psychological phenomena and to different kinds of reasons for action than I find in Williams and to show that we can say more about our normative concepts and how we think about resolving disputes than Williams does.
this goal, sacrificing many other valuable aspects of life along the way. After several years of medical practice and at the pinnacle of her success, she discovers that within a short time she will be permanently blind.
The trauma of her discovery is easy to imagine. Her life is in shambles because she has lost her ground project and with it her point of view toward her life and its future. A guiding perspective is impossible because she lacks a point of view from which to assess her life, except in terms of what has been lost. The thought, "I am the person who has lost that which is most important to me, my surgical practice," is what we may call her identifying thought, a thought that need not be self-conscious but implicit within a person's overall psychology. It is the kind of thought that allows a person to express identification with his or her life and its components. In this sense, an identifying thought is an answer to the question, Who am I?
Of course, the question, Who am I? is not unambiguous. It is important, then, to be as clear as possible on what it does and does not mean. It is not a question regarding the identification and reidentification of particulars, as applied to persons. The person who suffers puzzlement regarding the question as intended here is not confused about whether she is Sue or Sarah or Jane or Jill. She might know very well that she is Sue and nonetheless be puzzled in a way that raises the question, Who am I?
Rather than reflect puzzlement over the reidentification of particulars in this sense, the question reflects a different quandary. It is the quandary over how to identify with one's life so that it provides a sense of who one is in terms of which one's past, present, and future are comprehensible as a significantly integrated whole. The puzzlement arises in contexts in which an agent faces the future without adequate direction. The need is to bring to bear thoughts of oneself and what one values to resolve the issue of direction, where there is an absence of the relevant sorts of thoughts that will provide this guidance into the future. The thoughts that do provide this sense of self and that prevent or resolve such puzzlement I am calling the agent's identifying thoughts.
Further clarification of what is meant by an agent's identifying thoughts is revealed in the experience of emotions of self-assessment.[4] A positive example is that of pride. It is in terms of the thought that one has exceeded some minimally acceptable standard of expectations that one experiences the emotion of pride. As such, it is a thought that identifies the agent as one worthy of special self-affirmation. A negative example is that of self-
[4] . see Gabriele Taylor, Pride, Shame, and Guilt: Emotions of Self-Assessment (Oxford: Clarendon Press, 1985).
contempt, where the thought that one is a person of a certain sort identifies one as unworthy of respect.
As with emotions of self-assessment, certain interests are possible only with identifying thoughts that reflect a conception of oneself as a significantly unified personality over time. The interest in being a good parent is an excellent example. Absent other concerns, a person having the interests and character of a good parent would be puzzled regarding the direction of life at the unexpected loss of his or her child. Other interests, however, cannot play this role in a person's identifying thoughts. Under normal conditions, the thought that one has not had quite enough sleep does not raise the puzzlement over who one is and the direction of life, even where there is some interest in a bit more sleep.
Note that a person, such as the woman in our example, who has completely lost her ground project has only a backward-looking identifying thought. As such, it gives her no direction for her life. It gives her no identity in terms of her future. It tells her who she was rather than who she is and who she is to be. Such disarray in terms of her identifying thoughts reveals that her loss was that of a ground project, rather than something of less importance.
To have reasons for living, then, she needs an identifying thought that directs her toward the future. In this sense, her identifying thoughts must be motivating ones, the kinds of thoughts generated by a positive interest in life. Unless motivating thoughts emerge, there will soon be no identifying ones. For the emergence of new identifying thoughts requires the appearance in her life of a new ground project in which she has a categorical interest.
Not all identifying thoughts, however, are either backward looking or based on motivating categorical interests. Some such thoughts express categorical aversions. These are aversions to actions or ways of life that have serious consequences for the agent. The consequences are that if the person performed such actions or participated in these ways of life serious disunity of the self and loss of meaning in the person's life would result. This might extend to a loss of a will to live at all. The cliché, "I would rather be dead than red," is putatively an identifying thought that expresses a categorical aversion. But like purely backward-looking identifying thoughts, these thoughts do not express reasons for living. Only motivating categorical interests in ground projects can provide this. A person of integrity, as we will see in later chapters, has both categorical interests and categorical aversions. Yet one's life has categorical value from one's own point of view only if one has a categorical motivating interest in a ground project. Within such a
project and among such interests might be the interest in being an agent of pure practical reason and governing one's life in accordance with the CI procedure. If so, then such an interest serves the most fundamental integrative task of that psychology. There is nothing, then, about the concepts of categorical interests and categorical aversions as employed here that rules out Kantian internalism in advance of the application of the integration test.
4.
It would be a mistake, however, to think that since ground projects are the objects of categorical interests, they are permanently fixed in the life of the agent. Ground projects can and do evolve and change over time. Such evolution, however, is never a matter of simple choice by the agent. Views of human agency that treat human choice as though everything about the agent is under the power of the human will are simply flights of fantasy. Nevertheless, the agent's choices, as well as other factors influencing the agent's life, can bring about fundamental changes in what life is most about for the agent. In this sense, ground projects can be malleable, but, even in exceptional cases, only to a degree.
Nor should we think that ground projects are so rigidly placed within the priorities of any agent that recovery from the loss of such a project is simply out of the question. It might be that for some agents recovery is possible and for others it is not. Here it is important to distinguish different descriptions of a loss a person might suffer. To describe a loss as irretrievable is to say that the agent cannot get the thing that is lost back. Immortality aside, the loss of life is irretrievable, as is the loss of innocence, virginity, youth, and some forms of ignorance. On the other hand, some descriptions apply more to the effect of the loss on the agent than to the thing that is lost. To say that a loss is irretrievable is not in itself to say anything about the effects of the loss on the agent. People, for example, might respond differently to the loss of virginity—some with joy, some with guilt, and others with indifference. But, to describe a loss as "debilitating," for instance, is to say something about the agent, and, of course, some irretrievable losses can be debilitating—and to different degrees. Some such losses are ones from which an agent can never recover, while others are those from which an agent can recover, though with difficulty. The gauge of debility is the loss of integrity in one of the ways in which that is possible, some of which are permanent and thus mortal losses of integrity. Here we are clearly dealing with the possibilities of pathos.
Imagine that through support from loved ones, or through therapy, or
something of that sort, the fictional surgeon recovers. What she lost was irretrievable, yet she recovers by taking on some other interest that is motivating in that it gives her reasons for living and provides her with unity and meaning in her life. She is able to "put her life back together again," which is a necessary condition for her being an ongoing agent of integrity.
If she is able to do this, the loss of her original ground project proved to be a debilitating loss but not a mortal one. If the loss of the original project was such that the woman was psychologically unable to put her life back together in a coherent way, the loss of the ground project was a mortal loss. If it resulted in the loss of the will to live or left her with a permanently and seriously disfigured sense of self, it was likewise a mortal loss of her integrity.
There are many ways in which an agent can suffer a mortal loss of integrity, and it is one of the goals of this book to explore the significance of this. I will argue that personal integrity is far more social and far less individualistic than one might think from the original surgeon's example. But this argument will come only with the analysis of the thick conception of integrity. The point here is that it is only in terms of categories of pathos that we are able to make sense of a psychology in which integrity resides.
5.
An important distinction for understanding degrees of loss concerns the relative simplicity or complexity of ground projects. So far, I have been speaking as though ground projects were simple and that loss of a ground project was necessarily complete. But this is misleading. Such projects can vary greatly in complexity. The surgeon's project, for example, was a simple one. A more complex and more normal project would include several components. Imagine an altered version of the surgeon example in which the surgeon has children. If on the discovery of her impending blindness the surgeon's interest in life is unaffected by any interest she might have in her children, then her children do not figure into her ground project. Suppose, however, that her love for her children were to aid her in surviving her sense of loss. This does not mean that her surgical practice was not a component of her ground project. It was as long as its loss resulted in a serious, though perhaps not a complete, loss of unity and meaning in her life. A project is a component in a larger ground project if its loss affects the identifying thoughts of the person and this result is viewed with severe regret from that person's point of view. In this way, we can clearly speak of distinguishable components of a complex ground project and the loss of a ground
project as partial or complete. We also may clearly speak of either a component or the entire project in terms of whether its loss is mortally threatening to the agent.
6.
An understanding of categorical interests and aversions, then, is central to an understanding of the psychological structure of the person of integrity as a rational agent. But to further clarify the concept of integrity, we need to understand how noncategorical interests shape the priorities of a human agent. For our purposes, we may distinguish two types. If an interest is such that its frustration has no effect on the identifying thoughts of the agent but does lead to serious disappointment, it is an important noncategorical interest. Frustrations of these kinds of interests, though painful, do not seriously threaten the unity and meaning of a person's life. If an interest is such that its frustration leads to little disappointment to the agent, it is a minor noncategorical interest.
Reflection on the difference between many cases of embarrassment and severe guilt is revealing in regard to the distinction between important noncategorical interests and those that are categorical. The emotional response of severe guilt strikes deep at the structure of a personality, reflecting the workings of an interest fundamental to the psychology in which it is embedded. Indeed, often such guilt results in debilitating depression. On the other hand, embarrassment does not usually have this feature. True, embarrassment is structurally revealing, but it is not usually associated with serious depression or something of that psychological depth. Yet none of us likes to be embarrassed, and in most cases we think it important to avoid. Typically, embarrassment of the more severe sort temporarily disorients us at most; it does not threaten the very unity and meaning of our lives, even if an embarrassing episode causes some short-term depression. The difference between a categorical interest and an important noncategorical interest then is a structural one.
An additional comment on the experience of pain will perhaps further clarify the distinction. Unless intense physical pain is prolonged, it does not in itself typically affect the view one has of oneself and of one's life prospects. It does not typically affect a person's identifying thoughts. Pain of this sort differs, then, from emotional pain that does affect the view one has of oneself and one's life. To feel the pain of guilt or of humiliation is structurally different from mere physical pain, even when the former is far less intense than the latter. This is not simply because the emotional pains are
more likely to be of greater duration, for the pain of even temporary embarrassment is also different from physical pain in that the former involves the self in a different way than does the latter.[5] Categorical interests function at the core of a psychology in a way that noncategorical interests do not, and important noncategorical interests play a structural role that minor interests do not. Without awareness of these differences, there is a failure to understand how priorities function in the life of an agent. Thus it is through interests that goods are presented as ends within an agent's deliberative field, but the sense of how important those goods are to an agent's depends on the kinds of interests through which they enter the field.
7.
Perhaps it would be wise here to consolidate the previous observations regarding the thin conception of integrity within a psychological setting and then to relate these observations to both Aristotle's "function argument" and the integration test. To consolidate we need to index both the kinds of interests that might be affected within an agent's life and the kinds of effects on the agent when those interests are frustrated. I will describe the latter in terms of agent states, states the agent might be in as a result of the frustration of the interests in the other part of the index. Consider, then, the following chart:
| |||||||||||||||||||||
[5] . Of course, if intense enough, physical pain can be the object of a categorical aversion, in which case it does have structural significance.
What I am aiming for is a conception of human interests as those interests are structured within a personality. Without the distinctions in the left column, we are unable to understand the interests as those of a person, especially those of a person of integrity. Notice that neither Bentham's categories of intensity and duration nor cardinal measurements of individual utility reveal anything resembling a personality. Without the list of the agent states in the right column, we are unable to fully understand the distinctions in the left column as they apply to human personalities. Agent breakdown, then, can be viewed as involving changes in agent states that occur sometimes as a result of the frustration of different kinds of interests the agent has.[6] (Other times, of course, such breakdown occurs as a result of nonintentional states of the human body.) When these breakdowns occur at the core, they reflect the disintegration of personality.
Now, how does all this relate to Aristotle's function argument and to the integration test? In the Nicomachean Ethics , book 1:7:1097b: 21–24, Aristotle says, "To call happiness the highest good is perhaps a little trite, and a clearer account of what it is, is still required. Perhaps this is best done by first ascertaining the proper function of man." The word translated as "function" is ergon , which means "activity." Reflecting his biological bent, Aristotle claims that every natural kind has an activity that is natural to it, that expresses its most fundamental nature, and the good for any natural kind is to be able to express its nature by engaging in those activities that allow it to survive and flourish in its natural environment. He comes to the conclusion that the good for man is "activity of the soul in accordance with virtue" (NE 1098:a15). If we demythologize "soul" to mean "psychology," we get the notion that eudaimonia is an active life that is expressive of a psychology that is most natural to humans. Though we should be cautious about Aristotle's assumptions about the species-wide good, especially in regard to humans, demythologizing the soul in this way in our understanding of human well-being opens the way to empirical tests for whether some way of life is as a matter of fact good for a person of a certain kind. Just as an arid environment is not good for an acorn because it will not be conducive to the kind of flourishing that we observe in healthy oaks, some ways of life characterized by their activities are not good for most humans. The function argument in Aristotle is supposed to ground some sort of empirical testing for whether a way of life is one that fits the nature of humans
[6] . I discuss what I call integral breakdown in considerable detail in my Dignity and Vulnerability: Strength and Quality of Character (Berkeley: University of California Press, 1997).
and hence is good for them. I suggest that we understand the function argument in terms of the integration test and the latter in terms of how a psychology can break down.
The consolidation of our philosophical psychology, then, provides us not only with an understanding of how a psychology can have pathos and structure but also with an understanding of how we can test for claims about the integrative function of practical reason. One thing we see is that the unity of a personality is made possible by the fact that some interests regulate other interests in their demands on the overall psychology. In this sense, there is a hierarchy within a personality: Categorical interests and aversions are dominant relative to noncategorical interests and aversions. It might seem to follow from this that if norms are, in part, a function of interests, then necessarily there will be regulative norms that are asymmetrical in their regulative functions. But this thought is based on conceptual confusion. True, categorical interests asymmetrically regulate noncategorical interests, but it does not follow from this that there are any categorical interests that are not regulated by other categorical interests. This means that it is possible that there are no partial or impartial interests that are unregulated by other partial and impartial interests. Looking back at chapter 2, this is what we would expect. Moreover, by viewing our experience from these categories, we begin to understand how practical reason can have an integrative function within a psychology by construing practical reason as essentially connected to what gives a psychology its pathos. And most important, we can see how to test for whether integration has been achieved. Do the occurrent interests of the agent allow for the agent's flourishing, or do they lead to agent breakdown? These are empirical questions answerable by sophisticated social science in terms of the integration test guided by an adequate philosophical psychology. It is in terms of such a test that we can determine whether the value of rational agency in Kant's sense or Aristotle's sense is among our deepest values.
We should, however, be careful to distinguish on the one hand between the integration test and the function argument, which take place from the third-person point of view, and practical deliberation on the other, which takes place from the first-person point of view. Neither the function argument nor the integration test is an explanation or model for how to deliberate or to reason practically: Deliberative models have justificatory functions. Rather both are naturalistic accounts of why we deliberate and reason in the practical way that we do, and hence they have explanatory functions. When we confuse the latter with the former, as Aristotle himself sometimes does, our understanding of practical reason itself becomes
confused. That we are social animals is on both Aristotle's and my view an explanation for why we sometimes reason in the way that we do. This does not mean that it is a factual premise in our deliberations that we are social animals. We do not derive values from facts in that or any other way. It is one thing, then, to give a naturalistic account of why the most admirable persons reason in the way they do; it is quite another to give an account of how they deliberate from the first-person point of view. All that is required philosophically is that the explanatory account match in a coherent way the justificatory account. It does this by being explanatorily accurate in regard to persons of character we admire. But admiration comes from the firstperson point of view. If we do not admire and desire to emulate persons of a certain character, there are no merely factual beliefs that will lead us to a normative guide. Still it remains true that there is some naturalistic explanation for why we admire some kinds of persons and not others.
In its most sophisticated form, the integration test is administered by the most developed social science, which was not available to Aristotle. It is for this reason that I suggest we understand the function argument in terms of the integration test. If we do, I think we arrive at a more fine grained account of practical reason than Aristotle himself did, simply because we have a more sophisticated test. Here the best clinical psychology can both diagnose agent breakdown and, in those cases in which it is true, trace its causes to the values influencing the agent's life. For example, clinical depression might be diagnosed and traced to the misplaced values of adolescence; it might also be traced to tragic losses that are nearly impossible for anyone to handle. But as Korsgaard rightly points out, the practical point of view for any agent is the first-person point of view, not the point of view of social science. How, then, does failure to pass the integration test appear from the point of view of the first person? A brief answer is that, under the appropriate conditions, the agent experiences deliberative difficulties with the relevant clinical consequences. Much more will be said about the kind of difficulties I have in mind in chapter 4.
Here, however, several things need to be explained, and to do so I will appeal to some thoughts of Charles Sanders Peirce. In "The Fixation of Belief," Peirce gave a sketch of how we should think about belief naturalistically.[7] On his account, our beliefs serve us in the task of integrating our lives in relationship to our environment. The moment of doubt arises when we face a problematic situation, a situation in which the network of our be-
[7] . See Charles Sanders Peirce, "The Fixation of Belief," in Philosophical Writings of Peirce , ed. Justus Buchler (New York: Dover Publications, 1955), 5–22.
liefs is no longer sufficient for unifying our experience in a way that can provide cognitive guidance. We become baffled in our attempts to accommodate the problems of experience, due to the poverty of our concepts. We regain confidence when we form concepts that can project us into the future in a way that gets us past the problems that we can foresee on the horizon. Inquiry is a way of fixating belief, relative to whatever cognitive problems have emerged for us. As I understand his view, we are justified in a reflective endorsement of a way of conceptualizing things to the extent to which we can imaginatively project a cognitive understanding of our experience in which problematic situations, occasions for doubt, do not occur. Now I want to think of practical deliberation in an analogous way. Our practical reflective lives begin with a pattern of normative beliefs. Eventually, life presents us with a problematic situation, a situation in which the network of our normative beliefs, a network of beliefs about how to live, is no longer capable of organizing our practical experience. Instead of cognitive bafflement, what we experience is puzzlement over the direction of life. It is the concern over the direction of life that gives rise to the problem of reflective endorsement in the first place. And here I am talking about the first-person point of view. What will count as a solution to this problem, therefore, must fit the problem itself, namely, how to get from here into the future within a life. Thus the revision of one's values must be such that those values are placed within a conception of a life in which the problematic situation no longer appears. The very nature of the problem of reflective endorsement from the practical point of view dictates that the widest reflective standpoint is the point of view of one's life as a whole. For it is a problem about life that requires a solution, and the solution is adequate only insofar as it gives coherent meaning and guidance to a life. A set of values, then, fails from the first-person point of view when it presents the agent with a deliberative field the agent cannot integrate in terms of its implications for a way of life.
But what are the appropriate conditions under which a set of values fails to provide a basis for integration from the first-person point of view? One condition is that reflective consideration of the problematic situation is not unduly hindered by self-deception or prevented by the lack of character to face the facts. These are conditions imposed by the thin conception of integrity. Another condition is that the agent takes the point of view that allows for the imaginative projection of a way of life in which all the relevant values are given their chance to appear as the ends they are taken to be within the agent's deliberative field. It is for this reason that the Aristotelian criteria of finality and self-sufficiency are relevant; they fit the kind of prob-
lematic situations that give rise to the tasks of practical reason. The final condition is that the agent cannot project a way of life in which all the values of the agent can find a place in a set of priorities the agent can endorse. Thus the failure of a set of values to pass the integration test from the personal point of view takes the form of deliberative difficulties in the face of the alternative ways of life available to the agent from his or her own point of view. And, of course, such difficulties come with the attendant clinical implications for his or her psychology. I will say more in the next chapter on these deliberative difficulties, but here the main point is that the task of practical reason is a problem the nature of which requires the point of view of a life as a whole as the widest possible reflective position an agent can take in regard to it. This should reinforce the view that if the CI procedure is to be reflectively endorsed, it must be endorsed from the point of view of a life as a whole, from the viewpoint of a way of life, thus making the criteria of finality and self-sufficiency central to the tasks of practical reason.
8.
Before concluding my comments on the interests and priorities of the person of integrity, I must say something about the essential concept of commitment. I must say what it is and why the concepts of human agency, human character, and personal integrity, whether moral or nonmoral, require it.
Perhaps there could be an environment where there were no possible challenges to human interests, where frustrated interest would not be possible. But it is not possible that human beings could be agents in such an environment, and it is the concept of agency that is our subject. The reason that humans could not be agents in such an environment is because they could only be passive regarding their interests. That is, they could not pursue their interests. Nothing they could possibly do would count as not satisfying their interests, and in this sense there would be no alternatives. Without the possibility of challenge, the frustration of interest does not make sense, and without the possibility of the frustration of interest, the concept of agency does not make sense.
Yet it might be thought that there can be agency in an environment in which frustrated interest is possible but never actual. In heaven, that is, things are arranged in a way that renders the anticipation of such frustration practically nil. In such a heaven the likelihood of frustration attached to any alternative available to an agent is improbable beyond the point of psychological significance. But if this is true, all available alternatives stand
on a par; so in what sense are they alternatives? How are they alternatives if they in no way constitute a greater or lesser threat to the objects of the agent's interests? Or more positively, how are they alternatives if they are all equally satisfactory. Among other things, such an environment (like the environment in which frustrated interest is not possible) would render the formation of intentions—a concept crucial to the notion of agency—problematic. The concept of agency, then, does not make sense in an environment in which challenges to interests are not possible. Nor does it make sense in an environment in which such challenges are practically remote beyond the point of psychological significance. Therefore, the concept of agency requires the concept of an environment in which challenge to interests is to some significant degree likely.
This, of course, is true of the natural environment in which humans live. Thus the analysis of human agency and human integrity is in this sense naturalistic: It is an analysis of human agency in terms of a human being (i) as an inhabitant of a native environment, (ii) as having different kinds of interests in itself and that environment, and (iii) as having different capacities for responding and adapting to that environment.
Commitment is the human capacity that enables us, as agents, to meet the challenges to our interests.[8] Whether as a moral or as a nonmoral agent, to be an agent of integrity one must be able to meet these challenges with some degree of success. The scale that gauges the success required for agent-integrity is one that by now should be familiar. It is the ability to retain the degree of coherence in one's life to qualify as having one rather than no or several selves and to have reasons for living. This is why agents must be committed to categorical interests in a way that they are not to other interests if they are to survive as agents of integrity.
If human agents are to survive with their integrity, then, they must be committed to their ground projects categorically. This is the foundation for their identifying thoughts. Thus, given the human condition, there are no stable identifying thoughts without categorical commitments.[9] And it is through the capacity for categorical commitment that categorical interests can have the standing of norms. For it is the mark of an interest that it expresses a norm that it is held with some degree of commitment; otherwise,
[8] . I avoid here the dispute between deterministic and libertarian accounts of commitment and human volition.
[9] . See John Kekes's reference to identity conferring commitments in "Constancy and Purity," Mind , no. 92 (1983): 499–518.
it is only an interest. Moreover, it is the mark of a commitment that it functions as a norm in an agent's practical reasoning. Thus we can speak of categorical and noncategorical norms.
The concept of categorical commitments and norms helps to explain two important features of agent-integrity. The first is the unity of the self necessary for an ongoing agent. The second is the strength of character required for the survival of that unified self as an agent meeting challenges to its interests in its environment. It does not follow, of course, that if an agent has integrity in this sense he or she is thereby wholly admirable. For it is possible that the person is categorically committed to the despicable. Admirable integrity is a subject for consideration in later chapters. But here it is important that unless a human being has integrity in at least this sense he or she cannot be an admirable agent simply because there is not an agent with a character. It is only by seeing how norms are expressed in the structure of a character that we can assign them a functional role within agency. For this, we need the distinctions regarding categorical and noncategorical interests and how they function as norms as expressed in the commitments of a structured psychology. In parts 2 through 4 the inquiry involves the structure of this psychology as it is revealed in the agent whose character expresses the thick conception of integrity.
9.
It has been argued by some that to be an agent of integrity is to have a sense of self such that one sees one's commitments as one's own. To be an agent of integrity I must be conscious at some level of awareness of a distinction between the categorical worthiness of some project and my commitment to that project. If this is true, it would offer some explanation for the importance of the "separateness" or the "distinctness" of persons taken as a foundational moral value by Kantians.
The claim seems correct for two reasons. First, it is necessary for any concept one might have of oneself as a human agent. If Sue's awareness of her behavior lacks a sense of "herself-as-bringing-it-about-that" such and such happens, then she is not aware of herself as an agent to whom we can ascribe responsibility. Also, if she is unaware of "herself-as-being-affected" by events, it is difficult to see how she could see herself as having any interests.
The second reason that awareness of one's interests and actions as one's own is a fundamental feature of integrity is this: It is necessary to make co-
herent many of the interests that are of categorical value to human beings. If I am committed to a shared relationship with someone, I am not only aware of myself in a certain role; I am acutely aware of its being my role and of the importance of my fulfilling that role.[10] What would it be to see myself as a parent, to be very concerned that the needs of my children are met, but be indifferent to who meets those needs as long as they are met? Nor is my commitment to my own work reducible to the concern that the work gets done; it includes the thought that I do it. I simply cannot have these interests if I do not have an awareness of myself as a separate and distinct person and assign some importance to the fact of this distinctness. This sense of distinctness, then, is crucial to the identifying thoughts of the agent whose self is at risk in a challenging environment, for without a personal attachment to one's own life such survival is not an issue.
Moreover, it is categorical value that gives survival a point. Since humans do not have a bare interest in living, survival becomes an issue only when a person is within an environment of a certain sort: one in which there are things one can take a categorical interest in and to which one can make a categorical commitment. The first challenge to the possibility of integrity, then, is the challenge of nihilism, the view that life is not worth living, that suicide is—to answer Camus's question—the only course of action.[11] Since it is ludicrous to think of nihilism as an a priori doctrine, meeting its challenge requires an understanding of the things that can provide categorical value for humans, things that can appear within one's deliberative field that give one reasons for living. We can meet this challenge if we can satisfy two conditions: first, specifying what things must be within an environment to make survival an issue for humans, and second, showing that these things are at least sometimes within our environment.
If nihilism can be refuted and survival has a point, there remain other challenges to integrity, challenges regarding human capacities. Failure of these capacities is perhaps the most illuminating way to think about them. The first is the failure in the capacity to make any commitments at all. In such cases, the agent has interests—perhaps even categorical ones—but is completely unable to meet any challenges to those interests. Thus the agent
[10] . Acute awareness, here, should not be construed too strongly. As James Harris has pointed out to me, there must be some room for thoughts to the effect that "I didn't realize that I was in love with you, or how much you meant to me." Still it would be more than just odd to live a life in which one could say at the end, "I didn't realize that any of you mattered to me."
[11] . See Albert Camus, The Myth of Sisyphus and Other Essays , trans. Justin O'Brien (New York: Vintage Books, 1991).
is constantly abandoning and changing his or her interests the moment a challenge arises. Self-awareness is merely that of various episodes connected at best only in memory. In these cases, the failure is a failure of character, an inability to live with any priorities at all. It is the failure of having a psychology in which there are interests but no norms. Also, there are failures related to commitment to too many things, to the lack of capacity for self-denial, and to commitment to conflicting projects. Finally, there are failures that are the result not of the lack of normal capacities but of bad luck.
10.
Whether, then, one is Socrates or Galileo, Richard Nixon or Winston Churchill, van Gogh or Gauguin, or Uncle Ralph or Aunt Clara, one must have a sense of the categorical and the noncategorical to have the basic elements of integrity in the thin sense. For without such a sense of the categorical and the commitments that attend it, there is no way for a sense of priorities to emerge that explains the emergence of the basic elements of integrity. The thin conception of integrity, therefore, is essential to an adequate moral psychology. For it is within a sense of priorities psychologically conceived that we are able to understand human agency and any place that morality and an agent's norms might have within it. These thoughts should provide a guide for thought about practical deliberation, the subject to which I now turn.
4—
An Integrity-Sensitive Conception of Human Agency, Practical Reason, and Morality
Understanding the place of morality within the life of a human agent requires an integrity-sensitive conception of both human agency and morality. This, in turn, requires an understanding of how, as an agent, one's reasons for action emerge from the meaning of one's life from one's own point of view. This follows from the fact that one's categorical interests are both those in terms of which an agent finds life meaningful and those in terms of which the structure of one's psychology is built. And, as we have seen, this requirement does not in itself rule out the Kantian CI procedure from being defensible in terms of our categorical values. Thus an adequate philosophical conception of morality must be true to whatever the relationship is between an agent's reasons for action and the agent's own perceived good. In this chapter, I discuss a thin conception of the good for an agent of integrity, the implications of this thin conception for an account of an agent's reasons for action, and what this conception of the good requires of a moral conception regarding its sensitivity to the integrity issue.
1.
Just as nihilism is absurd as an a priori view of the human condition, so is an a priori answer to the question of what things are of categorical value to humans. The answer to such a question must be found in what humans actually find under reasonably favorable conditions to make life worth living. In this sense, though perhaps not in others, categorical value is not an issue of justification but of explanation.
Consider in this regard how we are both like and unlike other animals. It is natural for a dog to bark, but the dog's bark can be eliminated without
eliminating the dog. This might be done for reasons of social utility, human rights, or the will of God. But to eliminate the dog's bark is to disfigure the dog, even if such disfigurement is justified by other considerations. It is, of course, one thing to justify disfiguring the dog but quite another to require of the dog that it should justify its bark. Dogs are just the kinds of natural organisms that find barking natural. They also find copulating and fighting rather natural. Still these things too can be eliminated and perhaps on justifiable grounds. But if the dog's bark and its sexual and predatory instincts are eliminated, the disfigurement borders on eliminating that which makes a dog a dog. There are just some interests, then, a dog has because it is a dog. No sense is to be made of a dog's having and pursuing these interests for a reason or being justified in it.
Of course, a dog is not a reason-giving animal and hence does not live in terms of justifications. We, on the other hand, are reason giving and do live in terms of justifications. But however reason giving we are, we are still animals. Taking offense at this is simply taking offense at the truth. To be sure, the forms of meaningful life of which humans are capable are diverse in ways that the good for dogs is rather uniform for the entire species. But the central point remains that our reason giving proceeds outward from our categorical interests, which have their substance in whatever source of human meaning it is that has its deepest grip on us.
The call for justification, then, is at least unclear when it comes to reasons for living and to Camus's worries about the worth of life. It is at least worth asking what one is wondering about regarding the meaning of human lives where humans have self-respect, are involved in good personal relations, have some intrinsically meaningful work to do, have a healthy amount of play and leisure, are not suffering from serious physical or psychological abnormalities, and seem to take up their activities with vigor. These things do not justify their lives but are the values in terms of which human justifications normally proceed, at least for the overwhelming majority. Remove these things and most humans simply struggle to take any interest in life at all. Thus it is only from the things that add meaning that practical justifications emanate.
Someone might object, however, that what I have just said is simply false, that there are people who have all the things I have just mentioned in their lives but still wonder if their lives are meaningful. According to one version of this objection, these people, on looking out over their deliberative fields, are asking the question, Is this all there is? Such a question expresses a dissatisfaction with life because it lacks something; something is
missing. In Aristotelian terms, the goods of life are not self-sufficient; in fact, they are not minimally sufficient to make life worthy of being chosen for itself.
What this "something" might be that could be added to life in addition to the things already mentioned to make life worthy of choice is unclear in most cases. Still it is possible. Since I have not mentioned a relationship with God, this is certainly one possibility. Other possibilities might present themselves, though it is difficult to think of clear cases. Not enough security or not enough excitement come to mind. Yet it is difficult to see that these things would not be included within good personal relations and intrinsically meaningful activities of work and play. But whatever this something that is lacking might be it must be something of categorical value if the question about the meaning of life is deep enough to seriously disturb the person who asks the question. So understood, this version of the objection turns out to be no objection at all to the general point about justification: Justification proceeds from what is most meaningful for us; it does not precede it. For most people, I believe,meaning derives from the elements I have mentioned and are included within the thick conception of integrity. I will argue this in more detail in parts 2 through 4.
Another version of the objection construes the wonder about the meaning of life differently. According to this objection, some people who have all that I have mentioned in their lives still wonder whether the substantive focus of their lives is justified from some objective point of view. In yearning for a meaningful life, they seek not only the things above but these plus an objective justification for living this kind of life. Without the belief that such a life is justified, they do not find life meaningful.
One theoretical way of construing this line of thought about the meaningfulness of one's life is in terms of some conception of what value is. Consider how we might think of questions about the meaningfulness of our lives from the perspective of a certain view of intrinsic value. I have in mind the view of intrinsic value begun by the intuitionists (Brentano, Moore, Ross, Ewing, etc.) and developed by Noah Lemos in his book, Intrinsic Value .[1] According to this view, there is a kind of intrinsic value that attaches to states of affairs that is nonrelational which contrasts with both instrumental value and relational intrinsic value. Some things are instru-
[1] . Noah Lemos, Intrinsic Value: Concept and Warrant (New York: Cambridge University Press, 1994). Also, see my review of Lemos's book in Mind 105, no. 418 (April 1996): 496–503.
mentally good for something; others are intrinsically good for someone; and others are intrinsically good period. Hammers are instrumentally good for driving nails; playing the piano is intrinsically good for the music lover; and (perhaps) the flourishing of a tropical forest is intrinsically good period, regardless of whether there is anyone around to appreciate it. Of course, it is controversial that there is any such value as being intrinsically good period, but this is not something I am concerned to challenge here. The point here is to assume there is such value for the moment to see what relevance it might have to the question of the meaningfulness of one's life from one's own point of view and the issue of justification. In this regard, the point here is similar to the one we considered in the introduction that Aristotle made against Plato concerning the relevant sense of good: What sense of value is relevant to how we live our lives?
According to Lemos, intrinsic value is not only nonrelational, its existence is independent of any psychological attitude. This is true despite the fact that Lemos explicates the concept of intrinsic value in terms of a state of affair's being intrinsically worthy of love. He says,
(P2) p is intrinsically worthy of love if and only if p is necessarily such that, for any x, the contemplation of just p by x requires that x love p and not hate p.[2]
Bringing in love in this way to explicate the concept of intrinsic value would seem to contradict both the claim that such value is independent of any psychological attitude and the claim that it is nonrelational. When properly understood, however, the concept of being intrinsically worthy of love deepens our understanding of the nonrelational, psychological independence of intrinsic value. Lemos insists that the word requires in (P2) cannot be read causally but normatively. Thus (P2) is logically consistent with "x hates p even when contemplating just p." When x contemplates just p what x sees is that loving p is fitting. This does not mean that contemplating just p causes x to love p. Indeed, to see that love is fitting is to perceive a state of affairs, a fact, namely, that p is intrinsically good. Intrinsic value, on this view, then, is another fact about the world, and judgments about intrinsic value are about such facts. Not only is the fact "the tropical forest is flourishing" intrinsically valuable, but "the flourishing of the tropical forest is intrinsically valuable" is a fact. Moreover, it is a fact that obtains in-
[2] , Lemos, Intrinsic Value , 12.
dependent of anyone's believing it to obtain, and our belief that it obtains is true only if it obtains.
Though Lemos does not raise the question of the meaningfulness of life, we might understand what it is to justify one's life in terms of this concept of intrinsic value. Suppose one has one's self-respect, sympathy and concern for others, and love in a variety of forms for others and is committed to excellence at what one does. Assume also that one is in reasonably good physical and psychological health and that things are going well for those about whom one cares where one cares in the ways described. Further, assume that one is able to achieve some level of excellence acceptable from one's own point of view. Finally, assume that one has all this with the basic elements of integrity intact. Now, on the view under consideration, one might still wonder about the meaningfulness of one's life by wondering if it is justified. One would be wondering about whether the sentence "My life so described is intrinsically good" is true, that is, whether this sentence reports a fact. One would not, on this view, necessarily be wondering if one could on reflection endorse or approve of one's life.
So construed, if this is my worry, the wonder about justification is decidedly not about how to live my life but about the epistemic justification for believing that a certain fact obtains, namely, that my life is intrinsically valuable. Suppose this fact obtains and I justifiably believe that it does. There remains the question of how to live my life, unless I am psychologically constituted in a way that I find life meaningful only if I believe that I am justified in believing as fact that my life is intrinsically valuable in this sense. Therefore, even if we accept this ontology of value, there remains the issue of how such value bears on practical reason. There is a gap, on this view, between believing that something is intrinsically valuable and taking such a fact as even remotely of practical significance. For it is a contingent matter, on this view, that I assert that p is valuable and that I endorse or approve of or value p.[3] I might on this view care deeply that my beliefs about intrinsic value be justified but care not at all that my life is intrinsically valuable in this sense. Indeed, believing, on this view, that my
[3] . In this regard, Lemos says, "Even if it were true that whoever accepts that a thing is intrinsically good also approves or desires that thing, it is possible that he could conceive or entertain that thing's being intrinsically good without conceiving of his approving or desiring that thing," Intrinsic Value , 115. There is then no essential connection, on this view, between x's believing that p is intrinsically valuable and x's valuing p. Indeed, one could believe that p is intrinsically valuable and not value p at all.
life is not intrinsically valuable in this sense gives me no reason whatever for changing my life. For on this view, to be aware that something is good and of intrinsic value is not to be aware of something as appearing within one's deliberative field.
I cannot see that when most people are wondering about the meaningfulness of their lives they are worrying about whether it is a fact that their lives are intrinsically valuable in this sense. But if there are those who are and if their worry in this regard is at the center of their lives, then they have categorical values regarding epistemic justification that others lack. When their beliefs about intrinsic value affect their lives, it is not simply because they believe certain propositions concerning value to be true but because certain values are categorically instantiated in their affective (not merely cognitive) psychologies. What intuitionism lacks, then, is a conception of value that is relevant to practical reason. I believe the same applies to any externalist account of morality or what it is to have reasons for action, including traditional Kantianism. And this is one argument in favor of Kantian internalism.[4]
There is a difference between a moral epistemology, a theory of which is included in the second part of Intrinsic Value , and a theory of practical reason. At least, there is a difference on the intuitionist's view. Judgments of intrinsic value are on Lemos's view like axioms except that they have only modest a priori justification,[5] which is to say that they are defeasible judgments about necessary truths regarding value. As axiom-like, however, they are not practical norms. They cannot be, given that the assertion by a speaker that p is intrinsically valuable is not an endorsement or expression of approval of p; prescriptivity is thus external to the judgment that something is intrinsically good. To accept an axiom as a practical norm, however, is to see it as prescriptive, and it is only so that an axiom has a place in practical reason.
Why we should care about intrinsic value is, according to the intuitionist view, a separate issue from whether our value beliefs are epistemically justified. It is this separate issue that is both about practical reason and
[4] . Christine Korsgaard also faults intuitionism along these lines in Creating the Kingdom of Ends (Cambridge: Cambridge University Press, 1996), 52, 66, 228, 246–47. She similarly rejects realism in The Sources of Normativity (Cambridge: Cambridge University Press, 1996), and, I believe, decisively wins the argument against Thomas Nagel there. For the exchange between Korsgaard and Nagel, see The Sources of Normativity , 200–9, 219–58.
[5] . They are not axioms in the strong sense of having strong a priori justification.
about the meaningfulness of our lives. And what can be said in this regard about intuitionism can be said for any conception of "objective" value that is independent of the facts about human psychology.
It is one thing, then, to wonder if one is living according to one's deepest values or to wonder what one's deepest values are, or even whether one's life is anchored in some deep set of values at all. It is quite another to wonder with any sense of urgency if one's life is justified after careful reflection on one's deepest values where one finds that one's life is going well in terms of those values. What could one be wondering about? Here it is crucial to note that this is anything but a rhetorical question. It is rather a request for the hard philosophical work of providing a clear understanding of what the question of meaning is calling for and what its connection is with the issue of justification. Without the fruit of such work, the question of meaning seems itself rhetorical, even superficial. So when I say that ultimately categorical values are not in need of justification, I am not simply trying to circumvent a difficult philosophical problem. I am asserting that there is a clear understanding of practical justification that proceeds from categorical values rather than to them and that it is unclear what the issue of meaning is that motivates the denial of this conception. As we will see, there is some limited room for deliberation about one's categorical values, but these facts do not deny what has been said about practical justification and its structure. Contra Aristotle, we can deliberate about ends but only in a limited and special way.[6]
There is, of course, a sense in which categorical values are justified. They are justified from the agent's most reflective point of view, namely, the point of view of life as a whole. For this to be the case, those values must be such that they provide the basis for an agent's perceiving a way of life in which those values have a coherent place and that way of life is chosen for itself. But this does not mean that there is some point of view outside of any particular agent's own point of view from which the agent derives a justification for his or her deepest values. In this sense, explanation and justification converge. From the third-person point of view, what explains the rationality of one's actions are the very same factors that justify it from the first-
[6] . The best discussion I have seen in the literature on the possibility of revising final ends through deliberation is by Henry Richardson. There may be some difference between Richardson and me on what counts as a matter of deliberation, but on the view I defend there can be a change in one's final ends. How much change and how such change is effected by deliberation turns somewhat on what one counts as a matter of deliberation. See Henry S. Richardson, Practical Reasoning about Final Ends (Cambridge: Cambridge University Press, 1994).
person point of view, namely, a coherent set of values instantiated within a psychology with integrity. Hence the kind of naturalism suggested here does not commit the naturalistic fallacy: Practical judgments are normative through and through, but the explanation for why we value what we do is to be found in a naturalistic explanation of our psychologies.
2.
If what I have just said is true, that practical deliberation proceeds from our categorical values rather than to them, we are in a position to say something more about the notion of an action being rational, all things considered. The "all things considered" clause must always be understood as relative to a point of view, namely, the point of view of whatever the deepest values are of a particular agent. If Kantian internalism is correct, then the integration test should reveal that for agents who pass the test, their deepest concern is that their actions conform to the requirements of the categorical imperative. If the Aristotelian model is correct, then the integration test will reveal that for large numbers of people who pass the test, they will be agents of integrity in the thick sense whose practical norms are symmetrical in their regulative functions. Moreover, we should make sense of their practical reasoning in terms of their actions and their reasons for them as proceeding from their character. This means that they will see an action as rational, all things considered, because of its place in a way of life. And this life they will see as a whole in which none of its parts provide a basis for their taking as good, in Herman's sense, something that could provide a reason that is asymmetrically regulative of their other concerns. In either case, we must understand the notion of rationality, all things considered, as proceeding from those goods apart from which an agent could not find life at least minimally meaningful from his or her own point of view. And, of course, this means that the point of view of reflective endorsement is the point of view of life as a whole.
Now there is some irony in this for Kantians. To be committed to the view that practical reason proceeds from the bases of meaning in an agent's life from his or her own point of view already leans in the direction of the Aristotelian model. This is because on the view that practical deliberation proceeds from rather than to categorical values, it could never be rational, all things considered, to act in a way that is contrary to the most fundamental goods of a way of life valued categorically by an agent, where that agent had a clear view of that way of life in which his or her actions are seen as parts. Thus there is a connection between the insistence on meaning and
having reasons and that part of the Aristotelian model that insists on a certain notion of rationality, all things considered. It is this: An act could never be rational, all things considered, if the act were contrary to a way of life consistent with the deepest values of an agent's psychology. Rational rejection of a deliberative alternative would occur where the agent's imaginative projection of the place of an act within an overall way of life was an accurate projection and the agent would in the light of this projection reject that way of life and the actions it involves. It is only Kantian externalism that can reject this notion of rationality, all things considered, since it is only this brand of Kantianism that rejects the connection between meaning and having reasons. Kantian internalism, then, cannot maintain that the results of the CI procedure are necessarily rational, all things considered, and accept the integration test with its connection between meaning and reason. Proponents of that view can only maintain that as a matter of fact the CI procedure will result in full rationality, all things considered. I will argue that CI rationality is not even contingently rational, all things considered, and in so doing I will employ the only notion of full rationality consistent with an essential connection between meaning and reason, namely, reflective endorsement from the point of view of life as a whole. Thus pressure is brought on the Kantian internalists either to defend externalism or to accept the Aristotelian model with its emphasis on the meaningfulness of an agent's life from his or her own point of view. The latter leads directly to the criteria of finality and self-sufficiency in its conception of reflective endorsement.
3.
It is, of course, another issue whether all ways of life worth living for humans are moral ways of life. Nonetheless, one point cannot be overlooked: Just as other animals cannot have obligations to be nonanimal-like in their living, humans cannot have obligations to be nonhuman in theirs. Why? Because on the most plausible view of what it is for agents to have reasons in their living they could not possibly have reasons for this. Our reasons for action emerge from the sources that make human life meaningful from a human point of view. More specifically, a particular human cannot have an obligation to be nonhuman on whatever particular form of human good it is that gives his or her life its deepest meaning. A way of life, then, cannot be morally obligatory for a human unless it is a life that is at least minimally worth living for a human. In other terms: Without categorical value there are no categorical imperatives. Thus the meaning of morality finds it-
self within the meaning of human life, within the meaning of individual human lives.
Thus the point of any theory of human agency is not to reflect a conception of morality but to be as true as possible to the facts of human agency. Among those facts are those concerning the structural features of the interests and priorities of the person of integrity in the thin sense elaborated in the previous chapter. A test of any moral theory is that it must be true to these facts. Accordingly, I will argue for what I call an integrity-sensitive theory of human agency and place a conception of morality within it.[7] On this conception of morality, a way of life cannot be obligatory unless it is minimally worthwhile. This does not mean that one will believe that actions may never be required that will end one's life. Rather, it means that when death is required it is because there is no alternative that preserves a way of life that is minimally worthwhile from the agent's own point of view. There will be no alternative way of life that makes even a minimal place for the agent's ends so that life is worthy of choice in terms of finality and self-sufficiency.
Therefore, an integrity-sensitive theory places an agent's reasons for action within a conception of the agent's own good. Yet the primary emphasis is not on human flourishing, as in Aristotle, but on those goods that make life minimally worth living and the elements of integrity possible.[8] It is not that I believe that Aristotle was wrong about practical reason and considerations of the best way to live but that there is much to be learned about practical reason from thinking about a life that is minimally worth living. Accordingly, an integrity-sensitive theory, as I conceive it, places an emphasis on the survival of the agent of integrity. This has implications for the integration test and for conceptions of justification and obligation.
First, it means that the integration test will test for whether a particular
[7] . Some will object that what I suggest is not a conception of morality at all. Certainly, what I suggest is not a form of what Bernard Williams calls "the peculiar institution." Ethics and the Limits of Philosophy (Cambridge, Mass.: Harvard University Press, 1985). I cannot see, however, why the tradition being attacked by Williams there has any copyright on the term "morality." So I refuse to yield the term while recognizing that I am rejecting that tradition. The question is whether I give good enough reasons for this. See also Charles Taylor's discussion of Williams's views on "the peculiar institution": "A Most Peculiar Institution," in World, Mind, and Ethics , ed. J. E. J. Altham and Ross Harrison (Cambridge: Cambridge University Press, 1995), 132–55.
[8] . This is not to deny the Aristotelian conception of the integrative function of practical reason. It is merely to pay attention to those features of human well-being that give the pursuit of flourishing a point.
good has foundation in the agent's categorical values. The behavioral manifestations that a good is a categorical good will be exhibited in personality disintegration should that good be denied the agent. Or such disintegration will occur, as we will see, as a result of what I call categorical evils. The kinds of disintegration will be those alluded to in chapter 3.
Second, the emphasis on survival puts limits on issues of justification. A human does not need justification for being a human. Nor does a human need justification for surviving as a human , even where justification for surviving might be in order. Indeed, it is hard to see what such justification would come to. Since having reasons to survive involve the pursuit of a minimally worthwhile life, an integrity-sensitive theory of human agency must reject the view that a human needs a justification for pursuing a minimally worthwhile life. It must also reject the view that an agent needs a justification for not pursuing a life that is not minimally worthwhile from the agent's own point of view. It is this fact that leads to the Aristotelian interpretation of the "all things considered" clause of practical rationality. For it could never be practically rational to act contrary to that way of life that one believes is essential to those things one values most.
This leaves open, of course, the possibility of radical relativism regarding practical reason. So it does. It does not, however, either ensure such relativism or rule out a strong form of universalism. In this regard, one should view with suspicion any antirelativistic conception of practical reason that declares itself the only logical possibility. Similarly, one should view with suspicion any relativistic conception of practical reason that declares radical relativism true without appeal to the facts of actual human reasoning. When we look at those facts, I believe we discover that while relativism is true it is seldom radical, and when it is radical, it is not nearly so extensive as many declare it to be. Nevertheless, the recognition that one's reasons for action proceed from one's categorical interests does not rule out that Kantian internalism accurately accounts for the values of any agent who possesses the qualities of an agent of integrity in the thin sense.
The third implication of the emphasis on survival by these theories is that they place limits on an agent's obligations. According to some moral theories, that a way of life is not minimally worthwhile for a person is consistent with that person having the obligation to adopt it. Many consequentialist moral theories and traditional Kantianism are like this. But this cannot be true for any theory that is integrity sensitive. That a way of life is not minimally worthwhile for a person is inconsistent with that person having the obligation to adopt it. To deny this would be to assert that as an agent one feels obligated to take on a life on criteria independent of any
meaning it has from one's own point of view. To some, this denial will not sound so strange. But I hope to show that to deny this is to misconstrue the actual place of impartial norms in our lives and in the lives of those we admire most.
As with other animals, then, the issue of justifying certain interests is miscast when applied to humans. As is the issue of whether a human life is worth living when a human is able to live a life that is at least minimally worthwhile in a natural and social environment. Moreover, there are limits to how much disfigurement can take place from curtailing human interests without eliminating the human element altogether. This is the issue of the survival of human agency, which is central to an integrity-sensitive theory. It can be put in this way: What goods are there that if eliminated from a human life or if not available to a human in its environment would make life not worth living for that human?
4.
Clearly the answer to this question concerning survival should be expressed in categorical terms. Here I want to clarify what it is for a particular human to find that the object of a human interest is a categorical good.
To survive as persons of integrity agents must have practical wisdom, which includes a sense of their own good. They must recognize their good when it is available to them. In part, this consists in having a clear picture of what appears within their deliberative field in a way that they see everything that is there (self-sufficiency) and do not misconstrue the value of what is there (finality). But also in part, a person's own good consists in having a set of interests with the structure of the person of integrity in the thin sense. It consists in having the objects of those interests in the agent's life in the way set by the priorities of that structure embedded within their psychology as norms. Thus the recognition of one's own good comes with practical wisdom. This consists (in part) in having a clearheaded set of priorities that reflect the structural features of the interests of the person of integrity in the thin sense. It does not consist in every instance or even in most instances of thoughts of a meaningful life and the means to such a life. The thought of a meaningful life does not usually enter the recognition of one's own good in this way in everyday deliberations.
Nevertheless, the person of practical wisdom has the ability to envision different ways of life and to see that some actions and character traits are consistent with some ways of life and not others. Indeed, it is this capacity that allows the agent to make judgments of full rationality. Were the agent
either unable to place things within some overall way of life or unable to place more meaning on some ways of life than others, the agent would lack the capacity essential for rational agency. For the lack of such a capacity is a failure in the ability to relate meaning to action. To be sure, the person of practical wisdom is not constantly engaged in assessing the overall meaning of every action; just as the Kantian agent is not always thinking through the CI procedure. Nevertheless, the kind of reflectiveness that requires seeing meaning in the context of life is required for practical wisdom and rational agency itself. Thus without the ability to employ, even if only intuitively, the criteria of finality and self-sufficiency, the capacity for practical reason is compromised.
Practical wisdom, then, includes the ability to recognize the importance of things within a set of priorities that are one's own. Without this wisdom, persons may think they find that something has a certain degree of importance when it does not. That is, humans can be mistaken about matters of categorical and noncategorical value. Such mistakes are reflected in the degree to which humans can flourish and maintain the elements of integrity in their lives. Mistakes of this sort we may call mistaken findings.
One kind of mistaken finding regarding categorical value concerns mistakes to the effect that some things would make life worth living for a particular human agent under favorable conditions when in fact they would not. John, for example, might think that a life confined to drugs and casual sex is of categorical value, that it is worthy of choice for itself and is deficient in nothing. Yet within a lifestyle of this sort he might discover that life tends to lose its allure for him, that confusion results regarding his self-identity. He may, in fact, find that these things are categorical evils because of their threat to other things that are the most meaningful in his life. Here the process of living must refine his capacities for seeing his life as a whole and how things fit into a way of life. If it does not or if he discovers his mistaken finding too late, his prospects for a worthwhile life are bleak at best.
Similarly, John might have a worthwhile life and be mistaken about which of its elements are actually of categorical value and which of its elements are not. Failure of discernment in this regard is a lack of practical wisdom due to the inability to employ the finality criterion, and the security of the categorical goods of such a person's life is tenuous. The problem is that an agent of this sort lacks the capacity for seeing which things actually are a part of a way of life and which things are not. This is one kind of failure in the capacity for reflective endorsement, and, as such, it is a failure in practical reason.
John also might be mistaken about whether his life is worthwhile at all.
It may not include the things that he thinks it does, and he may be unable to face this fact. For example, he may think his wife loves him, unmistakenly find life not worth living without her love, and yet simply be mistaken about her affections for him. Moreover, he may not be able to see the truth of the matter because his fear of the loss of her love threatens his reasons for living. Mistaken findings of this sort undoubtedly involve more than a failure in practical wisdom. Here the agent can envision a worthwhile life, but because he does not satisfy the conditions of integrity, he cannot see that his way of life is not the way of life he envisions.
John might also be mistaken about the value of a way of life simply because he is too inexperienced to see what things he would value were he only to experience them. Acquired tastes are one kind of example, but there are others. Contrast acquiring a taste for wine, a taste for classical music, and coming to love a child. Some values once acquired might lead to an overall reordering of one's values and to rejecting one's previous way of life in favor of another. Acquiring a taste for wine does not usually lead to such overall reordering. On the other hand, if one is a dedicated music lover, acquiring a taste for classical music over a previous affinity, say, for at least some forms of popular music, might lead to some reordering. But surely coming to love a child where one had previously been indifferent to such love might lead one to change one's way of life in an extensive way. Indeed, it might lead one to see one's previous way of life as rather shallow, even where it previously had some depth.[9]
From this, we can see that the issue regarding the categorical value of any human good can be put in terms of whether judgments about it are based on mistaken findings. What is it, then, for a particular human good to be categorical for a particular agent? It is one that under adequately informed conditions of choice a particular human agent of integrity and practical wisdom would not unmistakenly find life worth living without and be able to maintain the basic elements of integrity. What is it for a particular human good to be a universal categorical good? It is one that under favorable conditions of choice no human agent of integrity and practical wisdom would unmistakenly find life worth living without it and be able to
[9] . Any plausible form of internalism must accommodate these facts, and I can see no reason why it cannot. Whether Williams's view can accommodate them is debated in the literature. See especially John McDowell, "Might There Be External Reasons," in Mind, World, and Ethics , 68–85; and E. J. Bond, Reason and Value (Cambridge: Cambridge University Press, 1983). Richardson's discussion of deliberation and final ends is also relevant. See Richardson, Practical Reasoning about Final Ends .
maintain the basic elements of integrity. It is a relative categorical good if some but not all adequately informed human agents of integrity and practical wisdom would make a similar finding regarding it. At this point, I leave open whether there are any universal categorical goods, and I consider this a requirement of inquiry concerning an integrity-sensitive conception of human agency and morality.
Also important in this context is an understanding of what it is to discover a mistaken finding. From the third-person point of view, it is to discover the source of the loss of unity, meaning, and reasons for living in an agent's life in the agent's commitments and norms. These are commitments and norms that cannot establish and sustain the agent's integrity. To make such a discovery is to discover that some goods or a constellation of goods are not ones that will provide a conception of rational agency that will pass the integration test for that person. From the first-person point of view, to discover a mistaken finding is to discover that one's priorities are such that they do not structure a way of life that is self-sufficient and final, that is inclusive enough regarding what one values and is worthy of choice for itself.
Thus the categorical value of any human good is solely determined by its status in a worthwhile human life. No otherworldly, nonhuman conception of value is being imported to determine the worthwhileness of any human life. Nor—and this is of the greatest importance—is there reason to worry on a related front: No conception of the human good is being imposed independent of the kind of life a particular person of integrity actually would find worthwhile under favorable conditions. The only restrictions besides those regarding unmistaken findings and favorable conditions are those regarding the thin elements of integrity. Nothing I say about that concept gives any a priori bias against any conception of the good life for humans, except those that revere alienation, insanity, and self-deception. For all I have argued to this point, as opposed to what I have suggested or asserted, there is no a priori reason for ruling out many possibilities, especially Kantian internalism. At any rate, there is a test for whether any particular set of priorities for any actual agent is based on mistaken findings. It is whether those priorities can be sustained with the elements of integrity intact in a way of life in actual historical circumstances over time. This is the integration test. But it should be emphasized that the integration test when employed from the third-person point of view is a device of theory. The agent discovers a mistaken finding from the first-person point of view by getting a full picture of the place of some consideration within an overall way of life and finding that that way of life has no meaning for
him or her. It is the philosophical account of agency that employs the integration test from the third-person perspective.
5.
I move now to some comments on what it is for a way of life to be an available alternative for the person of integrity and practical wisdom. The most fundamental condition is that the way of life must make possible the elements of integrity. Again, these are substantial unity of self, self-preserving character, awareness and importance of one's distinctness as an individual, and sufficient perspicuity of self-perception. It also must be a deliberative alternative for the agent in question. This means that the alternative must be one that can be considered as a practical possibility for the agent through the agent's own deliberations.
Doubtless, some important sense can be made of a way of life being "available" to an agent in a nondeliberative sense. But the current issue leads us to focus on contexts involving an agent's deliberations. It centers on how we are to give an account of how an agent's reasons for action emerge in the context of living a life. That is, we are focusing on that aspect of an agent's behavior that is deliberative. For this, we want an account of the agent's normative thoughts, how they are prescriptive for the agent, and how they emerge in the agent's life. These issues lead to the current focus on deliberative contexts and therefore on deliberative alternatives. It is to these contexts that an integrity-sensitive theory must be responsive in a way that other theories are not.
The contexts for a particular human agent's deliberating about alternative ways of life are many. But the most fundamental are those in which the agent is deliberating about whether to change his or her way of life. Of similar importance are those in which the agent is deliberating about the way of life to which he or she is to introduce others. The latter typically occur within parenthood, but I will focus on the first sort of deliberation. I choose this focus because it applies directly to decisions involving an agent's considering whether his or her way of life or a particular moral demand is justified.
Sarah is deliberating about changing her life because she is anxious about its moral standing in terms of the lives of others. Can she deliberate about a change in her categorical values and in what she sees as categorically good? If she can, we can make sense of the notion of deliberative contexts involving categorical change.
There are such contexts, but it is important to understand their limitations. Sarah can deliberate about a change in her categorical values but only to the degree to which she can entertain the possibility of retaining some stability to her identifying thoughts through the transition. Otherwise, she cannot entertain the change as a change in her life. This means that Sarah cannot deliberate about a complete change in her ground projects and her categorical values. Such deliberation is simply incoherent. One could never deliberate about a complete change in one's categorical values, because there would be no substantive basis for the deliberation.[10] Since one's categorical values function as the foundation for deliberation about changes in one's way of life, a change in these values precedes a change in categorical deliberations.
Incomplete categorical changes that result from deliberation are possible but only for people with complex ground projects. A deliberative incomplete categorical change occurs, then, from deliberations that reveal conflicts or shortcomings within one's categorical values. Also, the agent's deliberations about the resolution of such conflicts must be consistent with the agent maintaining his or her identifying thoughts through the transition. This is possible only if some categorical goods function as the basis for giving up or adjusting one's commitment to other perceived goods. In this
[10] . The difference between moral imagination and moral deliberation is an important one here. It is quite possible for a person who knows that Kennedy and Churchill are dead to imagine that they are still alive. But in an important sense it is not possible for such a person to believe that they are still alive. Why? Because imagination contrary to knowledge requires the playful suspension of belief. And suspending belief in this way is not the same as not believing. Similarly, it is possible to imagine oneself being an ax murderer, but it is quite another to enter into practical deliberations in the direction of the actions of such a person. In fact, such deliberations are simply psychologically impossible—however imaginable they might be—for the overwhelming majority of us.
Perhaps, then, I can imagine myself as a completely different person, though even this is fraught with difficulties. But even if I could imagine myself as a completely different person, this would not establish the fact that I can deliberate about becoming a completely different person. Often moral theory fails to take seriously the distinction between moral imagination and moral deliberation. The latter does not allow for the kind of abstraction from the concrete circumstances in the way that the former does. Moral imagination allows us a degree of abstract distance from our own conative and affective dispositions that is not possible in our practical deliberations, at least in regard to our most central dispositions. And though moral imagination has its own importance in the moral life, it can, if not tempered with real deliberation, lead to a life of mere fantasy.
See also Michael Sandel, Liberalism and the Limits of Justice (Cambridge: Cambridge University Press, 1982), the last chapter.
way, contrary to what Aristotle says, it is possible to deliberate about ends without turning them into means. In these contexts, the agent is moved to envision different ways of life, each of which excludes something treasured categorically by the agent. The agent will see one of these ways of life as the deliberative choice only if the agent has one categorical value that is more fundamental to the agent's psychology than the other. Envisioning different ways of life allows the agent to filter some values through other values and in some cases allows the agent to bring resolution to a problematic situation by projecting one alternative way of life as the agent's own. This occurs only when the envisioned way of life brings about a change in the agent's values and interests by adjusting them to each other within a way of life. And it is in this way that the agent's character determines the rational choice among ways of life, which is the most fundamental way in which practical reason is character-relative. Though there is significant room for an agent shaping his or her own character, these observations mark the most fundamental limit on such an ability.
I am not asserting here that there cannot be complete categorical changes in a person's way of life and consequently in one's identifying thoughts. Rather I am saying that complete categorical change is at best only partially deliberative. It is more fundamentally the result of nondeliberative evolutionary changes in one's categorical values.[11] Sarah's doubts about the moral standing of her way of life may very well indicate an evolutionary transition in her categorical values. This would explain why they are unsettling to but are not totally disruptive of her identifying thoughts.
Some categorical goods, however, are, if lost, irretrievable goods of a ground project, even where the project is complex rather than simple. If an agent rightly regards a particular good to be irretrievable and its loss a mortal one, no way of life that excludes that good is an alternative way of life for that agent. Nor is it a way of life the agent can consider in practical deliberation, and therefore it is not available to the agent in this subjective sense. No human agent, then, can deliberate to the normative thought of the obligation to adopt a way of life that threatens the irretrievable goods valued categorically by that agent. When there are conflicts between irretrievable categorical goods, there is deliberative impasse, and for problem-
[11] . John McDowell seems to have nondeliberative change in mind in "Deliberation and Moral Development in Aristotle's Ethics," in Aristotle, Kant, and the Stoics , ed. Stephen Engstrom and Jennifer Whiting (New York: Cambridge University Press, 1996), 19–35, as well as in "Might There Be External Reasons?"
atic situations of this sort there is no point of reflection from which rational endorsement of an alternative is possible.
Considerations, therefore, that require categorical changes in a way of life can have important but limited room in the deliberations of a human agent. But these considerations do not exhaust the field, for some changes are noncategorical changes. There is much more room for deliberative alternatives here, where we can begin to make sense of deliberations that require adopting one worthwhile way of life rather than another.
There is nothing about the concept of integrity and its elements that is threatened by a change in life that does not affect a person's core categorical values. Thus there is nothing about a person's integrity that precludes an agent's deliberating to the conclusion that another way of life requires noncategorical changes. In this sense, there are many ways of life that are deliberative alternatives for the person of integrity. Some of these might involve significant and important changes in one's life.
Later, I will argue for a thick conception of integrity that includes what can reasonably be called moral values among an admirable agent's categorical values and norms. Some are partial, and others are impartial. Yet a person's impartial values and norms are not among his or her reasons for living. But they are among the reasons for living one way rather than another. That is, there can be no sense made of practical deliberation in which an agent deliberates to the worthwhileness of his or her life on impartial norms alone. Still there is much in an agent's deliberating to the conclusion that one worthwhile way of life is preferable to another on impartial grounds. Finally, there is also the possibility of an agent's deliberating on impartial grounds to the conclusion that none of the alternative lives available are either worthwhile or permissible. Later, we will see how this is especially true of respectful people. What we will see is that their respectfulness involves categorical aversions to some ways of life and that these aversions give rise to norms that are strongly regulative of an agent's ability to see a way of life as even minimally meaningful. In themselves, these arguments will do a great deal of work against traditional Kantianism but little against Kantian internalism. The work against Kantian internalism will be done by arguments in parts 3 and 4 that the norms of practical reason are all symmetrical in their regulative functions.
6.
Thus far, I have tried to do several things regarding the claims of the introduction. In chapter 1, I made clear my methodological commitments and
argued that these commitments have a clear rationale that leads to an understanding of the internalism requirement in a way that entails the integration test as a means for rationally adjudicating between rival conceptions of rational agency. In chapter 2, I tried to establish that our actual employment of normative concepts reflects a conceptual scheme in which asymmetrically dominant norms are absent. Impartial norms like sympathy and respect regulate each other as well as partial norms like friendship and parental love. But, similarly, partial norms like friendship and parental love regulate each other as well as impartial norms like respect and sympathy. This suggests that if our conceptual scheme has any kind of rational foundation, then a central feature of both externalist and internalist impartialism (whether Kantian or non-Kantian) is to be rejected, namely, that the integrative function of practical reason is achieved by an impartial norm that is asymmetrical in its regulative function. It also suggests that the first part of the Aristotelian, inclusive-ends conception of practical reason is true, that all norms are symmetrical in their regulative functions. In chapter 3 and in this chapter, I have tried to introduce a conceptual apparatus for showing how in general the rational foundation for practical judgments is to be found in the goods that make life meaningful from an agent's own point of view. Categorical and noncategorical goods provide both the meaning of an agent's life from his or her own point of view and the foundation for practical judgments. Moreover, understanding the distinctions between categorical and noncategorical goods provides us with an account of how the thin conception of integrity is possible. Before I can conclude, however, that we should accept the Aristotelian conception of practical reason, I must show in some detail how the agent of integrity in the thick sense, with the goods that make his or her life meaningful, comes to have the kind of conceptual scheme found in chapter 2. This is the task of parts 2 through 4.