10—
The Normative Thoughts of Parental Love, Part II
Other-restricting Normative Beliefs

1.

In the last chapter, I discussed the self-restricting normative thoughts of parental love. I showed in the process how the Aristotelian model can accommodate a strong sense of obligation—one that reflects a set of priorities that solves the integration problems of the agent of integrity who is both respectful and parentally loving. Moreover, the analysis was applied across a wide range of contexts and revealed that the regulative functions of parental love and self-respect are symmetrical in their regulative effects. The deficiency of the Kantian model was shown to be that it cannot accommodate the function of parental love as a regulative and determinate self-restricting norm.

In this chapter, I attempt to provide similar results regarding the other-restricting normative beliefs of the agent who is both parentally loving and respectful of others. Because other-restricting beliefs involve the interests of third parties, they are especially relevant to impartial norms. They raise the issue of how the respectful and parentally loving agent is to integrate impartial concern for third parties with the partial concern for a beloved child. This complicates matters because it requires an argument for the Aristotelian model that employs a two-pronged defense, one against Kantian consequentialism and another against Kantian deontology, both of which take internalist forms. Thus I begin with some discussion of how to understand that distinction, with the goal of laying some foundation for the arguments not only of this chapter but of later chapters as well. I argue that neither deontological nor consequentialist deliberative patterns, when given Kantian interpretations, are rational across the complete range of deliberative contexts faced by an agent of integrity who is both impartially

respectful of others and parentally loving. I also consider another alternative represented by Samuel Scheffler's notion of an agent-centered prerogative.^[1] But more on that later. In the process of the argument, I demonstrate the rationality of the Aristotelian model with its symmetrically regulating norms. I start, then, with some discussion of the difference between Kantian deontology and Kantian consequentialism.

2.

Taxonomies can be tiresome and irrelevant, but when it comes to evaluating rival conceptual schemes they are indispensable. As preparation for the argument for the Aristotelian model, I will try to keep the tedium to a minimum, consistent with what is necessary for the inquiry. Important here is the following general taxonomy of normative conceptual schemes:

As we gain some understanding of this general taxonomy, I will attempt to show how it can take Kantian forms. Thus there will be Kantian consequentialism of both direct and indirect versions: Kantian nonconsequentialism of the prerogative and the deontological sort; and Kantian deontology of an action-type and a personal relations sort.

An impartial consequentialist conceptual scheme is one that has both a conception of value and a conception of obligation. (Hereafter, when I refer to consequentialism I will mean impartial consequentialism, unless otherwise noted.) The conception of obligation states that what one ought to do is to maximize good consequences, to act in a way that is most likely to result in the best available state of affairs. The conception of value is one that provides impartial criteria for ranking consequences or states of affairs from best to worst. Thus the theory of value makes possible the implementation of the theory of obligation: Ranking consequences or states of affairs from best to worst is necessary for maximizing good consequences. Utilitarianism is, of course, the clearest case of impartial consequentialism.

So understood, consequentialism comes in both direct and indirect ver-

[1] . See Samuel Scheffler, The Rejection of Consequentialism (Oxford: Oxford University Press, 1982); George W. Harris, "A Paradoxical Departure from Consequentialism," Journal of Philosophy 86, no. 2 (February 1989): 90–102.

sions. Direct versions construe good consequences as the causal consequence of individual acts; indirect versions construe good consequences as the causal consequence of any one thing or a combination of a variety of things. Most prominent among these are either the adoption of a set of rules for a general practice or the development of a set of character traits for living a life. In each case, the general practice and the set of character traits are determined to be "optimific" in terms of whether they result in the best state of affairs on impartial criteria,

With these distinctions, we can formulate two kinds of consequentialist deliberative patterns, patterns of deliberation that result in practical conclusions. A direct consequentialist deliberative pattern has the form, A ought to do x because x is likely to result in the best available state of affairs on some impartial value, V. With one caveat, an indirect consequentialist deliberative pattern has a rather complicated form, which can vary depending on the pattern of indirection. If formulated in terms of rules for a practice, it can be stated as: A ought to do x because x is required by a rule in a rule-governed practice that is most likely to result in the best available state of affairs on some impartial value, V. Stated in terms of a set of character traits, it can read: A ought to do x because x is necessary to develop and maintain a set of character traits that are essential to a way of life that contributes to the impartially optimific state of affairs according to V. The caveat is that there are indirect versions of consequentialism that would deny that they have this form. According to some, indirect consequentialism does not provide a decision procedure and hence does not recommend a deliberative pattern. Rather, it provides an account of moral "objectivity," an account for evaluating the truth of moral claims.^[2] I believe there are several problems with this view, but they need not concern us here. What we are trying to do here is to evaluate different conceptions of morality that are based on practical reason, not on some independent notion of moral objectivity. On conceptions of morality that are based on practical reason, the objectivity or truth of a normative claim is cashed out in terms of whether it meets conditions of rationality. Thus if indirect consequentialism is to be a competitor for a conception of practical reason, then it must recommend indirection as a deliberative pattern.

Nonconsequentialism, then, is the view that it is not always, either directly or indirectly, obligatory to act in a way that is designed to bring about the best available state of affairs on impartial criteria. Here we will consider

[2] . If I am not mistaken, this is Richard Brandt's view. See his A Theory of the Good and the Right (New York: Oxford University Press, 1979), 183–99.

two versions. The first is prerogative nonconsequentialism, which is a weak denial of consequentialism. Introduced by Samuel Scheffler, it asserts that while it is sometimes permissible to act in ways that are not impartially optimific, it is never wrong to act in an impartially maximizing way.^[3] This ensures that sometimes (though not always) an agent has the "prerogative" either to maximize on impartial criteria or to pursue more personal projects. Prerogative nonconsequentialism is motivated by the thought that sometimes consequentialism can be too demanding for an individual agent. Thus a prerogative deliberative pattern has the form, A may do x rather than y because y is too demanding. The second version of nonconsequentialism is deontology, which is a strong denial of consequentialism in that it asserts that it is sometimes wrong to act in ways that are designed to bring about impartially specified states of affairs. In asserting this, deontology employs what has become known as agent-centered or deontological restrictions, moral restrictions on the agent from acting on impartial, consequentialist criteria. Deontology, then, is motivated by the thought that consequentialism can demand the wrong thing. We cannot say much yet in general about what a deontological deliberative pattern is since different versions of deontology might arrive at agent-centered restrictions through different deliberative routes. As we will see, the Aristotelian model is a form of deontology that includes agent-centered restrictions significantly different from any Kantian model. What we can say at this point is that a deontological deliberative pattern of whatever variety has the form, A ought not to act on impartial maximizing criteria because doing so would be wrong. However, the rationale for thinking that impartial maximizing would be wrong is not revealed in this form. What we need is a form that reflects the rationale behind the restriction, and this we can formulate only on different versions of deontology. We can note, however, that the rationale for the Kantian version must be revealed through the CI procedure on either the procedural or substantive interpretation of that procedure. In the former case, it must be revealed through the universalizability tests of the first two formulations of the categorical imperative. In the latter case, the rationale must be accessible through practical judgment guided by respect for human dignity, where human dignity is absolute, nonscalar, nonadditive, and immune to trade-offs in terms of other values. Later, I will argue that it is difficult to see what that procedure is in some relevant contexts and

[3] . See Scheffler, The Rejection of Consequentialism .

in others that practical judgment is clearly inconsistent with affording human dignity such a value. This will be to argue that whatever latitude for judgment is left on the substantive Kantian view, the requirement that such judgment be guided by thoughts of human dignity so conceived is incompatible with the commitments of the agent of integrity in the thick sense.

Finally, deontology can employ different kinds of restrictions: actiontype restrictions or personal relations restrictions. Action-type restrictions specify action-types that are not to be engaged in, even if they will bring about a better state of affairs on impartial criteria. For example, one might believe that promoting good consequences, such as social utility, is a good thing, even obligatory in some contexts, but that it is wrong to do so if doing so requires one to break a promise, tell a lie, or torture an innocent person. Here breaking-a-promise, telling-a-lie, torturing-an-innocent-person are all action-types. An action-type deliberative pattern has the form that A ought or ought not to do x because of its action-type, regardless of whether it is impartially optimific.

Personal relations restrictions, on the other hand, are restrictions against impartial maximization based on the agent's personal relations. For example, a personal relations restriction might require you to do something for the sake of a loved one, even where there are other acts available to you that would bring about a better state of affairs on impartial criteria. In this chapter, the major focus is on whether there are such restrictions regarding parental love and what account is to be given of them. I argue that the Aristotelian model provides the best account, for it provides a rationale that is elusive on Kantian versions of deontology. Logically, such a restriction takes the form of an other-restricting normative belief, A ought to do x for B rather than do y for C, where what can be done for C (and others) would result in a better state of affairs from an impartial point of view.

As I understand at least some versions of Kantian deontology, actiontype restrictions are generated by an application of the CI procedure sometimes called the conceivability test. According to these versions, we can sometimes conclude that impartial maximizing would be wrong because the maxim that would allow it fails the conceivability test. According to this test, a maxim is not to be acted on unless one could conceive of that maxim as a law in a system of nature where that system could sustain itself. Put another way, if the system of nature could not be conceived as self-sustaining because it includes that maxim, then that maxim is not to be acted on. There has been much dispute about how to interpret this test, but I will employ Korsgaard's practical contradiction interpretation, even though she denies

that the test generates duties against action-types.^[4] I employ this interpretation because it seems to me the right one for the reasons she gives, though there is clearly room for debate on this point.^[5] Whether it has the connection to action-types is a further question I will come to later. On this interpretation, there is a special kind of contradiction involved in the attempt to universalize certain maxims of action. The contradiction is not strictly that of contradictory beliefs, as in, say, formal logic or some purely theoretical enterprise. Rather, it is a practical contradiction in the following sense. We are supposed to imagine that we live in a universe in which the maxims we act on actually turn into laws of nature. Thus if we act on a maxim, all our future actions as well as all the actions of other persons are governed by these natural laws that we create through our own willings. Some such maxims would be self-defeating in the sense that they would will a certain purpose but that purpose could not be achieved because of the content of the maxim. Kant's example involves false promises. He imagines a person acting on the maxim of borrowing money knowing that he will never pay it back.^[6] The maxim is supposed to be something to the effect that the agent intends to borrow money promising to pay it back but with the intent not to do so if doing so is at all inconvenient. In a world where our maxims become natural laws once acted on, this maxim would be self-defeating because it would result in the downfall of the promise-making/ promise-keeping convention once everyone realized that one could not keep a promise were it at all inconvenient to do so. More specifically, in the present case, the purpose of borrowing money, which is the goal of the maxim, could not be achieved through that maxim in our imaginary world because everyone would know that it is a law of nature that the money could not be repaid and consequently no one would make the loan. Hence if it can be determined that false promises fail this test, then those (unlike Korsgaard) who want to put the emphasis on action-types would say that any maximization pattern that would allow false promises is a prohibited pattern because of the action-types they involve. The same might be said for

[4] . For a discussion of alternative interpretations of the test—the Logical Contradiction Interpretation and the Teleological Contradiction Interpretation—as well as the Practical Contradiction Interpretation, see Christine M. Korsgaard, Creating the Kingdom of Ends (Cambridge: Cambridge University Press, 1996), 77–105.

[5] . For an interesting discussion, see Nancy Sherman, Making a Necessity of Virtue: Aristotle and Kant on Virtue (Cambridge: Cambridge University Press, 1997), 305–11.

[6] . Immanuel Kant, Groundwork of the Metaphysic of Morals , trans. H. J. Paton (New York: Harper Torchbooks, 1964), 89, 90.

lying, cheating, violating contracts, and torturing innocent persons. I will say more about this later, but let us call a deliberative pattern of this sort a perfect duty deliberative pattern, since the conceivability test was said by Kant to generate perfect duties.

Most Kantians also want to assert that we have special obligations to our loved ones, but it is unclear how these obligations are generated. One possibility, of course, is that they are action-type restrictions. If so, the CI procedure in its deployment of the conceivability test must reveal what the content of these obligations are and that they are perfect duties. Another possibility is that personal relations restrictions are generated by another implementation of the CI procedure sometimes called the volitional test. In order to understand this test, we must assume that it is applied to maxims that pass the conceivability test. For Kant, duties or obligations are always generated when a maxim fails a test. Should a maxim fail the conceivability test, there is no need to employ the volitional test, for that maxim has already been shown to be one that cannot be universalized. However, it is possible that a maxim could pass the conceivability test yet still be one that is not universalizable. The volitional test is supposed to show this.

Very generally, the volitional test requires us to imagine that a maxim could be a law in a possible, self-sustaining system of nature but that somehow a rational agent could not will to live in that system of nature. Though there is a good deal of dispute about what this comes to, I will interpret the test in the following way. Kant is not having us imagine that a self-interested rational agent could not will to live in this system of nature. Rather, he is having us imagine that if the agent acts on a particular maxim, then that maxim will become a natural law that will govern the actions of all rational agents who live in that system of nature. Since rational agents are all motivated by impartial respect for persons, the construction of a system of nature through one's will must be such that it is consistent with the willings of every other rational being who must live in the same environment. Hence what I can will for myself without concern for others is not the issue, but what I can will for myself and any other rational being who is affected by my maxims as possible natural laws.

Kant believed that this version of the CI procedure generated a duty of mutual aid.^[7] Here actions are not required because they are of a type that can be evaluated for the kind of inconsistency found in false promises. There are many ways to render aid. For this reason, there is a great deal

[7] . Ibid., 90, 91.

more latitude in just what is required in the way of specific acts by this test.^[8] Thus maxims are prohibited by this test not because they attach to actiontypes but because they are inconsistent with the motivation of being concerned with the agency of impurely rational beings, beings who are both rational and have inclinations that can be thwarted in various ways. This is why Kant called obligations generated by the volitional test imperfect duties. I propose, then, that we understand a deliberative pattern that employs this form of the CI procedure an imperfect duties deliberative pattern. Later, I will argue that Kantians have not shown how such a pattern generates agent-centered restrictions involving personal relations, including those of parental love. Either they have failed to show on a purely procedural interpretation of the test how this is done, or they have failed to show on a substantive interpretation how such restrictions are rational when judgment is guided by the thought that human dignity is absolute, nonscalar, nonadditive, and immune to trade-offs in terms of other values.

Finally, I must say something about Kantian consequentialism. Historically, Kantian deontology has been contrasted with utilitarianism, and it has been thought that it is sufficient to be a deontologist that one believes that it is sometimes wrong to seek to maximize social utility, either directly or indirectly. It should be easy to see that identifying deontology with this belief is a mistake. One could believe that the wrongness of pursuing one impartial goal is made wrong by the priority of some other impartial goal. If this is the reasoning behind the rejection of utilitarianism, then it is not one that leads to deontology. For deontology asserts that it is sometimes wrong to seek to promote impartial goals, no matter what they are. Kantian consequentialism rejects utilitarianism, not because it is consequentialist, but because it employs the wrong criteria for ranking states of affairs from best to worst.

David Cummiskey has defended a view of this sort in his book, Kantian Consequentialism . As it turns out, on Cummiskey's view, a Kantian consequentialist deliberative pattern is two-tiered: it aims at promoting the conditions for the exercise of rational agency throughout the network of rational agents, and it aims at promoting the happiness of rational beings. Actually, Cummiskey says that this two-tiered view involves three things: "First, each agent should promote the conditions necessary for forming, revising, and effectively pursuing a conception of the good. Second, since all agents are equal, each agent should adjust his or her personal ends in light

[8] . I will discuss the latitude of imperfect duties in more detail in chapter 18 when I discuss Marcia Baron's work in more detail.

of the equal status of all other agents and their ends. Third, since others' ends are just as important as my own, each agent should recognize the equal value of the ends of others, and thus promote the happiness of others—provided that the ends of others are not inconsistent with the previous two conditions and are thus permissible." He says also that "in deciding difficult cases of conflict between individuals . . . we must adjudicate in light of the equal status of each person and his or her interest in realizing a self-chosen conception of the good."^[9]

Though Cummiskey does not give us a crystal-clear picture of how these deliberations in difficult cases are to take place, the general spirit of the view is clear enough. In the general context of duties to others, obligations derive from a concern either to respect autonomy or to promote happiness, with the former ordinally prior to the latter. Adjusted to accommodate the distinctions regarding categorical and noncategorical interests, Cummiskey's view can be interpreted to distinguish between two Kantian deliberative patterns. The first I call a Kantian ordinal pattern. The idea is that when rational agents are parties to a conflict their interests are to be lexically ordered with categorical interests being the most important, important noncategorical interests the next important, and minor interests the least important. The ordinal pattern requires that, everything else being equal, the interests higher up the scale are to be given priority in deciding what to do because they are more central to the autonomy of a rational agent's life plan. The second pattern I call a Kantian maximization pattern. This pattern comes into play only when conflicts are between interests on the same level of the lexical ordering. It requires that when interests on the same level of the ordinal scale conflict, then one ought to act in a way that is most likely to satisfy the highest number of interests at that level without regard to persons. In what follows, I will argue that neither of these patterns is rational across the complete range of contexts a respectful and personal loving agent might face.

Before turning to the issue of the other-restricting beliefs of parental love, let me indicate Cummiskey's deepest worries about Kantian deontology. His major worry is about action-type deontological restrictions. The worry, expressed by many others, is that it is difficult to see what rationale such restrictions could have. Restrictions against torturing innocent persons is a good example. Any moral person is concerned to avoid and prevent the torture of innocent persons. But suppose that there are only two

[9] . David Cummiskey, Kantian Consequentialism (New York: Oxford University Press, 1996), 98, 98–99.

options: You can prevent either the torture of one innocent person or the torture of many other innocent persons, and everything else is equal. On what grounds could it be rational to choose to prevent the torture of only the one person? If it is the fact that an action is of a certain type that gives one reasons for thinking it to be wrong or irrational, then how could it not be worse or more irrational to allow for more instances of it, everything else being equal? I cannot see how to reject the implications of this question. For this reason, I am as skeptical of action-type deontologies as Cummiskey. More crucial, however, is the fact that if he is right about the irrationality of action-type deontology, then there is something incoherent about the employment of the CI procedure in testing for perfect duties if it tests for action-types. For there is no maxim involving action-types that would fail the conceivability test, if that test is to be understood as generating moral prohibitions against action-types. Why? Because it would always be rational on Cummiskey's view to perform an action-type to prevent further actions of that type. In this, I think he is right. Moreover, he is right that Kantian action-type deontologists do not address this issue.^[10] But it does not follow that deontology has been refuted, because there remains the issue of a personal relations deontology. In what follows, then, I will give an account of the other-restricting normative beliefs of parental love that rejects action-type deontology of any sort for Cummiskey's reasons and leave it to those who would endorse such deontology to give us reasons for changing our minds. This leaves open the possibility that the conceivability test can be seen not to generate duties against action-types but should be understood in a way that avoids consequentialism, provides a basis for personal relations restrictions in the conceivability test, and has a clear rationale. One of my main tasks in the rest of this chapter, then, is to show that none of the following deliberative patterns—a Kantian ordinal pattern, a Kantian maximization pattern, and a Kantian imperfect duties pattern—is rational across the complete relevant range of deliberative contexts. I will also consider a utility maximization pattern, as well as a prerogative pattern. Though there are no Kantians who would defend all these patterns, the point is to consider a wide range of deliberative patterns that might be called "Kantian" and argue that they all fail in important ways. In this way, I avoid the task of determining what the "real" Kantian view is.

Finally, I can reject Cummiskey's claim that only an indirect justification is possible for our deontological intuitions, if I can show that there is a ra-

[10] . Ibid., 11–15, 92, 93; and see Korsgaard, Creating the Kingdom of Ends , 100–101.

tionale for them that does not fall prey to the problems of action-type deontology.^[11] I argue that the Aristotelian model provides a direct rationale for personal relations restrictions.

3.

Recall that an other-restricting normative belief concerns the interests of third parties. "A ought to do x for the sake of B rather than do y for the sake of C" is the logical form of such a belief. If the restriction is such that doing y for C (and others) would result in a better state of affairs on impartial criteria than doing x for B, then it is a personal relations deontological restriction. In this section, I address the integration and priorities problems as they arise across a significant range of contexts. After doing so, I address Stocker's problem and how the sense of obligation arises in these contexts.

Here the integration problem is this: How does one integrate the various concerns in the love for a child with the various concerns for others, especially those of respect? Again, this problem arises only when we recognize that the agent has a multiplicity of goods that appear within his or her deliberative field that must be integrated within a coherent way of life. The solution requires some reasonably settled set of priorities that allows for integrating these concerns across deliberative contexts.

To achieve integration, three alternative sets of priorities are ruled out. First, lexical priority to the interests of the child over the interests of others regardless of deliberative context leaves insufficient room for respect for others. If you cater to your child's whims while others suffer miserably in ways you could easily prevent, you are not a respectful person. My previous analysis of respect should have made this clear. Reflective endorsement of a way of life in which this set of priorities held would not be possible for an agent of integrity in the thick sense, because the criterion of finality would rule it out. For a respectful agent, other respectable people must appear as ends in the way of life that meets the finality criterion, and, on this set of priorities, they do not. Second, the same kind of priority to

[11] . Cummiskey says, "If one rejects deontological intuitionism, then an indirect consequentialist account of our deontological intuitions and commitments is the only reasonable justification that remains." See Kantian Consequentialism , 100. I cannot see that Cummiskey ever gives an argument for this conclusion. First, it might be that there are only two possible justifications for deontological intuitions but neither of them succeeds. Second, what is the argument that there are only two reasonable alternatives? I intend to show here that an Aristotelian conceptual scheme can provide a direct justification for personal relations restrictions.

the interests of others over those of the child regardless of context leaves insufficient room for parental love. What would it be to love your child and always put the interests of others above those of your child? Even Cummiskey admits this, although his justification for some of the most important cases is only indirectly a function of the concern to maximize the network of Kantian autonomy.^[12] But on the Aristotelian view, this set of priorities is ruled out because of the criteria of finality and self-sufficiency: finality, because the child does not have the proper place as a beloved end within the way of life ordered by these priorities; and self-sufficiency, because that way of life is missing something good, namely, the child as a beloved end. A third alternative—the equal treatment of others and the child across deliberative contexts—is also to be rejected. No plausible conception of respect requires this, and no plausible conception of parental love permits it. The general analysis of personal love is sufficient to establish this point. The child appears within the respectful and parentally loving agent's deliberative field as important in ways that preclude such an ordering of his or her priorities. Thus the respectful and parentally loving agent is able to order his or her priorities without the aid of an impartial decision procedure in a way that rules out all three alternatives. Rather, the priorities are set by the vision of a way of life that makes a place within that way of life for the goods that appear within the agent's deliberative field. This is just what the Aristotelian conception of practical reason promised to provide.

Yet, if these priorities are to be rejected, what are the priorities of an agent who has parental love and respect for a child and is respectful of others? What do these priorities show about the symmetry of regulative function between parental love, impartial respect, and impartial sympathy for others? And how can a successful solution to the priorities problem show that the rightness of the agent's actions is the nonaccidental result of his or her motives? To answer these questions, we must consider a range of contexts and the deliberative patterns that are rational in them.

1. When the interests at stake for the child are categorical, they have first priority over any noncategorical interests of others who are respected but not loved by the agent. This is true no matter how many noncategorical interests are at stake for others and no matter how many other persons have such interests at stake. To make this point vivid, consider whether you would sacrifice the fundamental psychological well-being of your child to

[12] . Cummiskey, Kantian Consequentialism , 11–15, 146 ff.

prevent the temporary, minor irritation of everyone on earth. Or to make the point more strongly, what would you do to protect the fundamental psychological well-being of your child in the way of causing others the kind of temporary depression that might come with the frustration of an important noncategorical interest? If we take temporary but nondebilitating depression as the strongest indicator that an important noncategorical interest has been frustrated, I cannot see how the concern to prevent such depression in itself, without regard to its impact on the categorical interests of others, could ever give a loving parent reasons for sacrificing the things crucial to the child's finding life meaningful, no matter how many people were so depressed. Suppose, for example, that sexually abusing your child, with the kind of deep long-lasting trauma caused by such an experience, would prevent the rest of the human population from being depressed for, say, a month. Or imagine the kind of deep psychological effects on your child of your locking him or her in a closet for a month with only the most sparse diet. Would you do this if the only bad effects were confined to your child and the only good effects were the relief of the kind of nonclinical depression alluded to for the rest of the human population for a month? These considerations should establish that a loving parent does not deliberate about utility calculations in such contexts. The loving parent's deliberations do not reflect a utility maximization pattern. The reason for this is that the place the child has in the parent's deliberations is determined by parental love. For on a way of life ordered by the set of priorities leading to the utility calculus the child could not have a place in that way of life as a beloved end, and the finality criterion would not be satisfied. Nor would that way of life be self-sufficient in even a minimal sense, since it would omit a good that is categorical from the parent's point of view. Thus the priorities leading to the utility calculus are rejected because they would not order a life in which the agent's ends would have their perceived value. In rejecting these priorities, the agent does the right thing in a way that the rightness of his or her action is the nonaccidental result of its motive. That is, the agent orders his or her priorities and the actions entailed by them in a way that takes into account all the relevant values from the widest possible position of reflective endorsement. Although this might seem a case of parental love for your child regulating your impartial sympathy for others, I will give reasons shortly for thinking that sympathy and love would converge in these cases.

Certainly the priorities of simple respect and parental love would converge. For if you respect a third party and both respect and love your child,

you will be disposed to give priority to your child's interests in any case in which the interest at stake for your child is categorical and the third party's is noncategorical. And it will not matter to you how many minor interests of others, including the third party's, are at stake. Nor would you as respectful agent deliberate about utility payoffs in such contexts, even if your child was not involved. As a respectful person, let alone a loving one, you would not deliberate about utility payoffs when faced with a conflict of this sort between the interests of two children neither of which were your own and to which you were not specially related. Why? Because the priorities that lead to the utility calculus clearly preclude the kind of value placed on those worthy of respect who are to be treated as ends. This is how respectable people appear within the deliberative field of a respectful person. Though this conclusion is not sufficient to establish that respect attaches to a conception of human dignity that is absolute, nonscalar, and immune to trade-offs, it is sufficient to establish that respect reflects a commitment to a value that is immune to trade-offs of this sort, as is parental love. Thus parental love and simple respect do not differ in their priorities in this regard, and there is no need for regulation between them, which is to say that the deliberative patterns of parental love and a Kantian ordinal pattern would converge in these contexts. Of course, the Kantian and the Aristotelian reach these priorities through different routes: allegedly the Kantian through the CI procedure and the Aristotelian through the criteria of finality and self-sufficiency as applied to how a set of priorities orders goods within a way of life. But how is it that one can deny that for the Aristotelian the rightness of action is the nonaccidental result of its motive, despite the fact that the Aristotelian's deliberations do not employ an impartial decision procedure?

A similar analysis applies, I believe, regarding impartial sympathy, regardless of what many direct utilitarians say about maximizing happiness. Suppose we could determine through interpersonal comparisons of utility that if you do x C will get 10 units of utility, your child, B, will get 1, and that there are twenty other people affected by your doing x. Suppose that they receive 10 units each. If you are impartially sympathetic, say some utilitarians, you will be disposed to do x if you believe that there is no alternative with a better overall sum of utility. But suppose that if you do y, then your child will receive 2 units of utility and C and the others will receive 5 units. According to either total sum or average utilitarianism, if you are impartially sympathetic, you will believe that you ought to do x rather than y. But I do not believe that you will necessarily have such a sense

of priorities, even discounting your parental love for B, but certainly not where your love for your child is included.

Consider first the case in which B is just one among others and not your child, that your concern for her is impartial sympathy. What if you believe correctly that 1 unit of utility measures an attachment to life that makes it minimally worth living? Anything less and life becomes pointless. Anything more only adds to a meaningful but far from perfect life. Add to this that you correctly believe that 5 units measure the satisfaction of a life with great meaning and that 10 units measure a life of ecstasy. It is anything but clear that your sympathy would lead you to sacrifice the meaningfulness of B's life for the increase from great meaning to ecstasy in the lives of the others. But there are two things that are clear. First, not just any marginal increase (either in average or total sum) of utility would allow you as a sympathetic person to do x rather than y. The base concern for the categorical interests of one person over the noncategorical interests of others would not allow you to make such a sacrifice. Second, this is most certainly the case if you have either simple respect or personal love for B.

On the first point regarding sympathy alone: Just imagine cases in which the total or average fluctuates only marginally in the meaning assigned to the units of utility. What this shows is that the solution to the problem of interpersonal comparisons of utility does not solve the problem of ranking states of affairs from best to worst from an impartial point of view. Being able to assign cardinal measurements to utility received for each person affected does not tell us what to do with those numbers in evaluating overall states of affairs. Do we add them, add and divide, or do we need some interpretive device to give them significance for evaluating states of affairs? I believe that the distinctions involving categorical and noncategorical interests are needed even to understand the concept of impartial sympathy. In this sense, I have serious doubts that the utilitarian's conception of impartial sympathy is adequate to give us a notion of social utility.^[13]

[13] . See Derek Parfit's discussion of the repugnant conclusion in Reasons and Persons (London: Clarendon Press, 1984), 381–90.

What is important is that the mathematical model for social utility track the concept we employ in judging the well-being of one population over another. Neither average nor total sum utilitarianism does this very well, once one interprets the numbers in a way that makes vivid what the numbers might mean. That there might be some mathematical model that captures our notion of the well-being of a population I do not deny (though I do not affirm it either), but the math tracks the concept rather than the other way around. And the concept is not revealed in thenumbers produced by the solution to the problem of interpersonal comparisons of utility.

On the second point in which either respect or love is involved: Good people who are either loving parents or respectful of others, or both, do not participate in some kinds of consequentialist thinking (which is not to deny that they participate in others). This shows, I believe, that there need be no conflict between Kantians and the dispositions of parental love when it comes to this kind of conflict. What I mean is that the priorities of the parentally loving but respectful person are extensionally equivalent on the Kantian ordinal pattern and the Aristotelian model for these contexts. The difference is that this set of priorities for the Aristotelian is a function of reflection on whether a set of priorities can make an adequate place within a way of life for the things that appear good within the agent's deliberative field; whereas on the Kantian view it is a function of the employment of the CI procedure on a certain interpretation. In order to understand the latter, a richer account of what it is to respect another as an end is needed than is sometimes found among Kantians. There must be a clearer understanding of how categorical interests are crucial to personhood and how personal love often allows one to be the person one is.

To this point, then, we have evaluated the rationality of deliberative patterns for contexts in which the categorical interests of the beloved child are in conflict with the noncategorical interests of third parties. We have seen that for a parentally loving and respectful parent, the categorical interests of the child have first priority over the noncategorical interests of others. This is what we would expect on an Aristotelian view, but it also converges with a Kantian ordinal pattern. But what about the rationality of the other three deliberative patterns, the Kantian maximization pattern, the imperfect duties pattern, and the prerogative pattern? We need not spend much time on these for three reasons. First, the Kantian maximization pattern applies only in contexts in which there are conflicts of interests on the same level within the lexical ordering of the categorical/noncategorical scale, and this is not a context of that sort. Second, since we have a plausible Kantian account for these contexts, there is no need to evaluate the imperfect duties pattern. And third, this is not a context in which the loving parent feels that a reverse set of priorities would be rational and permissible, which rules out the prerogative pattern.

2. When the interests at stake for the child and the interests at stake for others are important noncategorical interests, numbers of people involved

or interests at stake might matter. If it is only a conflict between your child's interests and C's interests, and everything else is equal, then the numbers will not factor. As a loving parent, you will give priority to your child's interests. This follows from the general analysis of personal love as an intentional dispositional state and how the beloved appears within the deliberative field of an agent who is both respectful of others and parentally loving. Such a set of priorities orders a way of life in which C appears as an end worthy of respect and your child as a beloved end. Both finality and self-sufficiency are satisfied on the Aristotelian view, and no one is treated as a mere means on the Kantian view.

But if there are large numbers of other respectable people who have equally important (though noncategorical) interests at stake, what are we then to say about the priorities of the loving parent who is also respectful and sympathetic toward others? Suppose, for example, that if you were to contribute to a Christmas fund for disadvantaged children you could be practically certain that the amount of your contribution would bring a certain amount of joy to a large number of children during the holidays. If you do not contribute, the same number of children will be somewhat depressed for a week or so. Your child, however, has her heart set on a particular, relatively expensive gift. If she does not get it, she will be somewhat depressed for a similar amount of time. In neither case—neither the disadvantaged children nor your daughter—will there be anything like agent breakdown as a result of disappointment; still, the disappointment is no minor matter, which is demonstrated by the depression. The reason the same sum of money will meet either the interests of the disadvantaged children or the interest of your own child is explained by the different expectations generated by social and economic background. Your child, say, wants a new, upgraded computer, and the other children want inexpensive dolls. What we are to imagine in the case of your daughter is that the difference between her old computer and the new one is understandably significant for her, no mere indulgence, but not crucial in ways that would make the interest in it seem more than the other children's interests in the dolls. We should also imagine the circumstance as being rare. For too many instances of deferring to others in this way might undermine the relationship between the parent and the child and hence would introduce the interest in the relationship into the equation, and that interest is categorical. What I have in mind is the fact that some acts represent patterns that express deep commitments and sentiments, and establishing and maintaining such patterns are crucial to the categorical interests involved in loving relationships. Here I am try-

ing to imagine a case where doing something for others would not undermine such patterns but would nonetheless come at some cost to a loved one. The purchase of that cost, however, is not the prevention of catastrophe or tragedy for others; rather, it is something of real importance, but not categorically so.

What would we make of a parent who considered the numbers in such contexts and eventually yielded to the interests of others? I, myself, do not see that such a parent would be unloving for having such priorities, though I must report that some people I know who clearly care a great deal about other people and who love their children cannot see how a loving parent could be moved by such considerations. This gives me reservations about asserting that the numbers must eventually add up or the person is not respectful and sympathetic toward the other children. I suspect this says something about the underdeterminate nature of our conceptions of parental love, on the one hand, and simple respect and sympathy, on the other. Perhaps it says something about getting a clear picture of just what is at stake in such contexts.

At any rate, I want to say rather tentatively that I can see nothing contrary to parental love in making such calculations in some contexts and eventually acting on them. A set of priorities that would allow for such calculations in such circumstances would not to my mind preclude a place to the child as a beloved end within a loving way of life. For example, one can imagine explaining to one's child the reason for going with the larger numbers in some cases, her understanding such a reason, and her still being disappointed in (though not disapproving of) the results. Examples of this sort show the regulating influence of simple respect and sympathy on the priorities of parental love. (Moreover, the fact that thoughts about these contexts are tentative shows the same regulative tension.) But precision here seems to be both unattainable and undesirable. Both respect and parental love are underdeterminate norms. Just how much others' important noncategorical interests can be trumped (if at all) by the same kinds of interests of a beloved child from the point of view of a loving parent is not subject to precise formulation. It will certainly not be at the precise point of marginal differences, for, as we have seen, the beloved child does not get into the way of life ordered by those priorities in a way that matches how she appears within the parent's deliberative field. Moreover, others do find a place as ends within a way of life that does not take the borders of marginal difference as crucial. Where others do not find their place as ends worthy of respect and sympathy is where the numbers never seem to matter, regardless

of how high they are and no matter how crucial the interests. These observations suggest the symmetrically regulating influence of parental love and simple respect on each other in the priorities of a person who is both a loving parent and a respectful person. All this seems to be explained by the fact of reflection on these priorities and how they order the goods that appear within the deliberative field of one who is both respectful of others and parentally loving. The underdeterminate nature of our concepts here seems easily explained by the Aristotelian model of practical reason. The internalist Kantians, especially advocates of the deontological versions of that view, must either affirm this result or deny it. In either case, they must show how the CI procedure, on either the purely procedural interpretation or the substantive interpretation, leads to what they take to be the correct result. If they affirm this result, they need to show how respect is operating as an asymmetrical regulative norm, which seems puzzling. To show this on the substantive interpretation of imperfect duties, they must show how the concept of human dignity as absolute, nonscalar, nonadditive, and immune to trade-offs properly restricts the latitude of the agent's reasoning. They cannot merely allude to the latitude of imperfect duties. If they deny this result, they need to show what is irrational about such a parent.

These observations should be sufficient to reject a utilitarian maximization pattern for these contexts. Slight marginal increases in utility at this level do not give a loving parent reasons for favoring others over a beloved child. How could they and the child's interests appear within the parent's deliberative field to be more important, even significantly more important, than the interests of others? Nor can I see any reason for thinking that a Kantian maximization pattern is any more rational than the utilitarian pattern. It is only when increases in utility or autonomy go well past the marginal borders that the loving parent's priorities might change. A prerogative pattern does not seem to account for these priorities, because the loving parent in these contexts seems to feel that either the numbers add up to the point of being compelling or they do not. Either way, the parent does not feel that it would be equally rational to reverse the priorities.

We are left, then, with a Kantian ordinal pattern and an imperfect duties pattern. The Kantian ordinal pattern does not apply because the conflicts are within the same level on the lexical scale. What about the imperfect duties pattern? The problem here is that we do not have an account from Kantians of how this pattern is supposed to work in contexts of this sort.

The person who has written most from a Kantian perspective about children has been Onora O'Neill, but she does not address the issue of how the

CI procedure generates parental obligations. She has discussed parental obligations in "Begetting, Bearing, and Rearing," in Having Children .^[14] However, she does not discuss Kant in that work. She also discusses children in Constructions of Reason , chapter10, "Children's Rights and Children's Lives."^[15] There she is concerned to show that a rights-based account of obligations to children is incomplete and blurred because it fails to take into account imperfect duties. In closing she says, "There are good reasons to think that paternalism may be much of what is ethically required in dealing with children, even if it is inadequate in dealings with mature and maturing minors. Nothing is lost in debates about the allocation of obligations to children between families and public institutions if we do not suppose that fundamental rights are the basis of those obligations. However, a fuller account of fundamental obligations to children and of their appropriate institutionalization in families and in public institutions is a further story. The task of this essay has been to show why that story needs telling."^[16] This is, as far as I know, a promissory note as yet unfulfilled by either O'Neill or any other Kantian.^[17] Nor is there in her discussion an indication of how the CI procedure would apply in the various kinds of contexts we are considering here. Recall from chapter 2 that this must be done in a way that does not involve the filtering of respect through considerations of parental

[14] . See Onora O'Neill, "Begetting, Bearing, and Rearing," in Having Children: Philosophical and Legal Reflections on Parenthood , ed. Onora O'Neill and William Ruddick (New York: Oxford University Press, 1979), 25–38.

[15] . Onora O'Neill, Constructions of Reason: Explorations of Kant Practical Philosophy (Cambridge: Cambridge University Press, 1989), 187–205.

[16] . Ibid., 205–6.

[17] . In Towards Justice and Virtue: A Constructive Account of Practical Reason (Cambridge: Cambridge University Press, 1996), O'Neill says the following:

Good parents usually think they are bound in many ways to their children. They will take it that they owe their children (and others) certain obligations which they also owe all others, to whose observance their children (like others) have rights: for example, they will take it that they are obliged not to injure. These are the universal perfect obligations , which abusing parents, and other child abusers, violate. They will also take it that they owe their children care and support, which their particular children, but not all children, let alone all others, have a right to receive from them: these are the special perfect obligations that neglectful parents violate. They will also take it that they owe their children, as they owe others, a measure of courtesy and concern: these universal imperfect obligations are virtues that rigid or fanatical or cold people lack, and whose lack rigid, fanatical or cold parents inflict specifically on their children, even when they scrupulously observe all their children's rights. Finally, good parents will take it that they owe their children certain sorts of love, attention and support which they do not owe to all, which are quite specific to the relationship to the child, but to which their children have no right. Certain sorts of fun, warmth and encouragement might come under this heading: these are special imperfect obligations which parents see as part of what is required of good parents, and which may be neglected even by parents who both observe their children's universal and special rights scrupulously and accord their children the same help and concern they habitually show all others–but no more. (151–52)

What is said here is anything but sufficient to fill the aforementioned promissory note. This is nothing more than a taxonomy of parental obligations. We are told nothing about how this taxonomy works out in detail in different contexts; nor are we told anything about how to get from the CI procedure to an understanding of how obligations falling under the categories of this taxonomy take shape in particular contexts. It is one thing to provide a taxonomy and quite another to show how that taxonomy reflects a theory of practical reason that is both context-sensitive and contextually thorough.

love. Moreover, it must be done in a way that affords human dignity a value that is absolute, nonscalar, nonadditive, and immune to trade-offs. It is difficult, then, to see how Kantians can be confident of their theory without having worked out these details. Indeed, it is as puzzling as how a scientist could be confident of a test for a hypothesis without controls. I will come back to this issue when I consider the normative thoughts of friendship. But I doubt very seriously that the CI procedure could succeed here, because these are contexts in which parental love and impartial respect are symmetrical in their regulative effects. And whatever the interpretation of the CI procedure, it cannot allow for such symmetry of regulative function. If this is true, then these are contexts in which the Aristotelian model provides an account of the rightness of action being the nonaccidental result of its motive where the CI procedure does not.

3. One might think that a similar analysis is true regarding conflicts between the minor interests of the child and others. There might be contexts in which as a loving parent you are in a special role vis-à-vis your children and other children and because of that context the numbers might matter. In the ordinary life of a parent, however, unmindfulness of the minor interests of other children and attention to the minor interests of one's own is certainly no mark of a lack of either sympathy or respect for others. Theories that try to fine-tune normative conceptions of respect and sympathy to cover cases of this sort do so only arbitrarily and at the cost of showing how silly theories can become. If this is correct, then, in a small way, these contexts reveal the regulative effect of parental love on our conceptions of simple respect and sympathy.

4. Conflicts between the categorical interests of the child and the categorical interests of others who are respected but not loved raise special difficulties. Only in very extreme cases can numbers matter for the loving parent, but at some point the numbers will have to add up if the parent is respectful of others. Often consequentialists offer as argument against nonconsequentialist forms of reasoning that if things get bad enough it is

absurd to stand by certain personal commitments. Of course it is, but to offer this as an argument is puzzling. There is a difference between impartialist maximizing rationality, impartialist minimizing rationality, and what might be called catastrophic reasoning. It does not follow that if I am willing to sacrifice my life to avoid a catastrophe for others that I am willing to sacrifice it either to maximize some impartial good or to minimize some social evil. The same applies regarding a loving parent's willingness to sacrifice the categorical interests of a beloved child. But, in such catastrophic cases, it is important to note that the loving parent is often likely to experience a deliberative impasse.

In contexts in which you are a loving parent and C is someone for whom you have only simple respect, if your child's categorical interests conflict with C's, and everything else is equal, you will give priority to your child's interests and have great sadness at C's loss. We can explain both the choice to favor your own child and the sadness on the Aristotelian model. The context is one in which there is no alternative that will save all the goods that are within your deliberative field; hence you must choose the most final of less than ideal choices. Because your child's categorical interests appear within your deliberative field as more important than those of the other person, the rational choice is the one that preserves your child's categorical interests. The sadness is explained by the wish for an alternative that would preserve the categorical interests of everyone involved.

A similar analysis is true, I believe, even if a significant number of losses of this sort accrue to other people, except that there will be greater sadness involved. This is the reaction we expect from loving parents who are respectful of others. It testifies to the regulative tension between parental love and respect for others, which would not be true if we thought of respect or parental love as asymmetrical in their regulative functions. Were parental love asymmetrical in its regulative function, we would have a difficult time explaining the sadness, and were impartial respect asymmetrical, we would have a difficult time explaining the choice in these contexts.

But if we think of agents in such contexts as respectful, at some point we expect them to have difficulties giving priority to their children. This might simply be due to the large numbers involved, or it might be the numbers plus a relative difference in the character of one's children versus the character and interests of others. If you have a bad child, you might love the child and still favor its categorical interests over the categorical interests of other respectable people. In fact, unless your child were totally evil, this seems certainly the case. But it would be inconsistent with the disposition of simple respect to be as disposed to allow for the categorical sacrifices

of others for the sake of such a child. Further, to the extent that parental love would allow the agent to permit such sacrifices, the sadness would be greater. Both the contexts leading to sadness and those in which the threshold reverses the priorities make evident the regulative influence of respect on our conception of personal love. But the lack of bright-line criteria for drawing the line signals the regulative influence of parental love on our conception of impartial respect. The thought process that reveals that this is true is simply the projection of how a set of priorities would structure a way of life relative to the goods that appear within one's deliberative field. If seeing what life would be like for persons with the character of loving, respectful parents on each of the alternatives does not settle the issue, there simply is no rational way to settle it. If Kantians are to deny this, they must show precisely how their theory avoids it; they cannot simply assert that this objection is a caricature of Kant. Nor can they merely allude to the latitude of imperfect duties, because there are competing accounts of the latitude allowed within practical reason: one Kantian and one Aristotelian. The former must explain the latitude in a way that is guided by the value of human dignity as absolute, nonscalar, nonadditive, and immune to trade-offs, which seems clearly false in these contexts. The latter explains the latitude in terms of the criteria of finality and self-sufficiency and the need to balance a variety of goods that symmetrically regulate each other.

There are rather mundane as well as dramatic situations involving the contexts in question that illustrate the priorities of a loving and respectful parent in this regard. The mundane situations are those everyday situations in which we apportion a percentage of our budgets to assist others. Most of us do not have a conception of respect such that, even on reflection, we would view ourselves as disrespectful of others unless we would be willing to reduce the well-being of our children to the point of diminishing marginal utility (or some Kantian standard) in the service of the categorical well-being of others. That we admire people who make exceptional sacrifices in this regard is clear, but it is not lack of respect for others that keeps loving parents from requiring categorical sacrifices of their children in this regard. What we would be willing on reflection to do with our everyday lives confirms this in rather mundane and undramatic ways. Of course, there may be those who do think that this is disrespectful, but then it is unclear what their conception of a loving parent is. Unless we are to reject the view that loving parents are in that regard good people, it is simply philosophically unacceptable to assert that some behavior is disrespectful unless one can say how avoiding such behavior is consistent with being a loving

parent. Kantians owe us a carefully worked out conception of what a loving parent is before they can tell us what respect for others requires of a loving parent. I cannot see that this has even been attempted.^[18] Morever, they must show how that account is consistent with a view of respect for human dignity as a value that is absolute, nonscalar, and immune to trade-offs in terms of other values. They cannot simply allude to latitude allowed by the universalizability tests. But if anything could be clearer than that these contexts show that respectful agents do allow for trade-offs, it is difficult to imagine what it would be. That parental love involves giving priority to the categorical interests of one's child over others in this way, at the cost of the very meaning of life to others beyond marginal differences in numbers, shows that respect does not function as absolute, nonscalar, nonadditive, and immune to trade-offs. So whatever latitude is involved in these contexts, it is not governed by the concept of respect for human dignity claimed by Kantians.

The point can be seen more dramatically by thinking about people who are put in witness protection programs in order to protect their families. Imagine being in a position to provide information that might undermine a drug cartel and the human misery such cartels inflict on large numbers of people. There can be no doubt that viewed impartially your own life and the life of your children are not as important as the many lives you might save through cooperation. But suppose that testifying would expose your children to very real and horrible dangers of recrimination by the cartel. You might be quite willing to risk recriminations against yourself, but would you be willing to risk the categorical well-being of your children, regardless of how small the numbers beyond the threshold of impartial value

[18] . In this regard, I am expressing a frustration with Kantian ethicists similar to the frustration expressed by Annette Baier. One frustration she has expressed is that Kantians do not seem interested in telling us very much about parent/child relations. Of course, she doubts that the Kantian framework is at all equipped in a way that would allow them to do so, primarily because she believes that the role of trust in the moral life is crucial, especially to women's experience, and that when contracts, promises, and voluntary exchanges are taken as the primary models for moral understanding, we lose sight of the kinds of trust that are essential to much of personal life. Moreover, I believe she is right that to gain a conceptual scheme that is more sensitive to matters of trust, we need to appeal to a tradition that makes much more out of human sentiments than the Kantian tradition can. Moreover, I believe this is as true for loving fathers as for loving mothers. For these reasons, I think Baier has done far more to shed light on what it is to be a good parent than anything found among Kantian writers. See especially her book, Moral Prejudices (Cambridge, Mass.: Harvard University Press, 1995).

that you might save by testifying? In fact, I think such considerations reveal that your insistence on witness protection as a condition for testifying when even some rather large numbers are involved illustrates how much more important one's children are to a loving parent than those for whom the parent has only impartial respect. We would not think of you as at all disrespectful were you to insist on witness protection as a condition of your testifying. Those who want to insist that you would be disrespectful inherit the burden of giving us a clear picture of what a loving parent is.

On the other hand, there are dramatic cases that illustrate that at some point the consequences to others of favoring the categorical well-being of one's own children would be disrespectful and uncaring. Reflections on actual recent events in Bosnia are illuminating in this regard. In A Bridge Betrayed , Michael Sells provides a vivid account of the kind of context faced by many Serbs in which the Serbian army was carrying out a policy of genocide against Muslims. First, he reports:

Serbs who refused to participate in the persecution of Muslims were killed. In a Serb-army occupied area of Sarajevo, Serb militants killed a Serb officer who objected to atrocities against civilians; they left his body on the street for over a week as an object lesson. During one of the "selections" carried out by Serb militants in Sarajevo, an old Serb named Ljubo objected to being separated out from his Muslims friends and neighbors; they beat him to death on the spot. In Zvornik, Serb militiamen slit the throat of a seventeen-year-old Serb girl who protested the shooting of Muslim civilians. In the Prijedor region, Serb militants put Serbs accused of helping non-Serb neighbors into the camps with those they tried to help. Similar incidents occurred throughout areas controlled by the Bosnian military.^[19]

And later:

When fighting broke out in Croatia in 1991, Serb irregular militiamen wore ski-masks or face paint. Survivors of atrocities reported trying to discern the accent of their masked torturers to determine where they came from. Sometimes a victim would recognize the voice of a neighbor. In many cases the man behind the mask was content to allow his identity to be known through his voice, and in some cases even taunted his victim with the fact that they knew each other.^[20]

[19] . Michael A. Sells, The Bridge Betrayed: Religion and Genocide in Bosnia (Berkeley: University of California Press, 1996), 73.

[20] . Ibid., 77.

Yet even in the face of such terror, some Serbs came to the aid of Muslims at the cost of putting their families at risk. In this regard, Sells reports:

Despite the effort of Serb religious nationalists to dehumanize both their own population and their target population, many Serbs have resisted and kept their humanity. Bosnian Muslim survivors commonly reported that a Serb or (in the case of Croat extremist violence) a Croat helped them escape. A soldier or border guard may have turned a blind eye as a Bosnian slipped away from an atrocity or fled to safe territory. A family might shelter a fugitive in their home, at great risk. A Muslim survivor of the killing camp at Susica mentioned that a Serb priest tried to help him. Bogdan Bogdanovic, the Serb former mayor of Belgrade, has spoken out courageously against the systematic annihilation of mosques and other cultural monuments. Many of the stories of resistance and courage cannot be told at this time, because the resisters or their families are still vulnerable to reprisals.^[21]

What we have in this kind of context is a catastrophe of monumental proportions, and it is here that we can make sense of a loving parent being willing to put the categorical well-being of his or her children at risk for the sake of others. The regulative influence of respect on parental love is seen in a very graphic way. But these contexts are not contexts involving the marginal borders of impartial considerations; they are catastrophic.

Thus the loss that other respectable people might have to bear in order for the loving and respectful parent to favor his or her child's categorical interests might become so acute that there can be no question that the interests of others prevail. Sells's observations about Bosnia show this. But if a loving parent is forced into catastrophic contexts like those in Bosnia, he or she—as a loving parent and as a respectful agent—will experience a categorical conflict. Even if there is a deliberative resolution, it will come at a categorical cost to the agent, and such costs always involve some degree of loss of integrity, should tragedy result. One might be the loss of the will to live. Justifying thoughts are not always consoling thoughts, and that one has had to sacrifice one's child for the sake of others might be justifying, but it is not consoling for a loving parent. Reflection might reveal that there is a way of life that one can endorse from among the alternatives in the context, but that way of life will fail in a significant way to be self-sufficient. It will lack a good fundamental to the meaning the agent sees in the best life. The parent's deliberative field will be haunted by the ghost of the child for some time to come. Thus, even in extreme cases, it is the exceptional per-

[21] . Ibid., 78.

son who is a loving parent who can have the normative thought that he or she ought to sacrifice a beloved child. The inability to do so even in these extreme cases is more a sign of the limits of human agency than a lack of respect.

Here it is important to note a difference between the priorities inherent in simple respect and parental love. If Cummiskey is correct (and I think he is), simple respect, considered in itself, can be forced into consequentialist reasoning that is not open to parental love. If you, only as a respectful agent concerned with the autonomy of others, can do x or y, where y will result in marginally less intrusions of individual autonomy than x to a larger number of people, and everything else is equal, then you will believe that you ought to do y. When, for example, rights of equal weight conflict, the Kantian can no longer resort to ranking rights. The only rational pattern left is optimizing across persons, a Kantian maximization pattern. Until Kantian deontologists have shown us a clear rationale for avoiding this conclusion, we are justified in believing it. Our belief that saving many human lives at the expense of far fewer innocent human lives is stronger than our belief that human dignity is absolute, nonscalar, nonadditive, and immune to trade-offs. In fact, I do not believe there is any evidence that we believe the latter. The evidence of moral reflection supports only the view that the value we place on human dignity resists some kinds of trade-offs. But parental love, considered in itself, does not allow such optimizing for the sake of others. To do so would leave no place in a way of life for a fundamental good that appears within the loving parent's deliberative field. Nor, except for the catastrophic contexts, does parental love allow for impartial consequentialist thinking even when combined with the dispositions of respect. For this reason, the agent who is personally loving and respectful of self and others has other-restricting normative beliefs that reflect thoughts of agent-centered restrictions. Thus parental love is not consistent with a conception of simple respect where simple respect is conceived in isolation from other dispositions and norms an agent might have. This, I believe, is why the Kantians give us a distorted picture of what simple respect is for human agents. It is not for a value that is absolute, nonscalar, nonadditive, and immune to trade-offs in terms of other values. Cummiskey, then, is wrong to think that these personal relations restrictions have no direct rationale. They are the direct result of reflective endorsement: The loving and respectful agent sees the priorities embedded within these restrictions as ordering the goods of his or her deliberative field in a way of life that is the most final and self-sufficient among the alternatives. Moreover, were simple respect taken as an asymmetrical regulative norm that required the

Kantian maximization pattern it would not guarantee the rightness of action in these contexts, for it would set the agent's priorities in the wrong way. On the other hand, where simple respect and parental love are symmetrical in their regulative functions, as they are on the Aristotelian view, priorities are set in a way that the agent's actions are the nonaccidental result of his or her motives.

In any event, the loving parent's dispositions toward a child and respectful dispositions toward other respectable people are not egalitarian, in either a Kantian or a utilitarian sense. The latitude of the loving and respectful parent's deliberations is not restricted by the concern for human dignity as the kind of value Kantians assert it to be. Nevertheless, the priorities described here allow and require considerable weight to be given to the interests of other respectable people. They certainly require that the parent be willing to sacrifice important noncategorical interests of the child, even for noncategorical interests of others. Under extreme conditions, there is even some room for the horrifying thought that the categorical interests of the child are to be sacrificed. Room for such thoughts, however, is limited by the place of parental love in an agent's life. Still, this set of priorities hardly reflects an overly conservative commitment to the welfare of those an agent is related to in only agent-neutral or impartial ways. To this extent worries that any view that takes its departure from Aristotle will be overly conservative and lack progressive vision are unfounded. It is the symmetry of regulative effect as exemplified in these contexts that constitutes the Aristotelian yet progressive view. And note that these priorities are arrived at without the need for an impartial decision procedure and that the rightness of the agent's actions is the nonaccidental result of his or her motives. Thus the account of full rationality and reflective endorsement in these contexts confirms the Aristotelian conception of practical reason.

I turn now to a discussion of Stocker's problem and how parental love can generate a sense of obligation in terms of other-restricting normative beliefs.

4.

Without other-restricting normative beliefs, it is impossible for us to make sense of the place a child has in the sentiments and dispositions of the loving parent. The previous discussion has established that these beliefs reflect a symmetrical regulative function between parental love and impartial respect. It remains, then, for me to show how these normative beliefs can be deontic beliefs and carry a strong sense of obligation. To see this, we must

understand how your respectful interest in another person can act as a contrary inclination to your parental interest in your child. This is where you have, relative to your child and the other person, an other-restricting normative belief. In what circumstances, then, are these normative beliefs deontic beliefs?

It is easy enough to imagine cases in which you would feel contrary inclination to favor your child over another person where you have an other-restricting normative belief regarding the other person. A good example is where you are the teacher of both and are grading their papers or deciding on academic awards. The point of mentioning this is to show that one other-regarding interest can act as a contrary inclination to another. In this case, parental love for your child acts as a contrary inclination to simple respect and esteem for the other person. But the kind of case we are seeking is different; it is either one of two kinds. The first is where an impartial interest (respect, esteem, sympathy) acts as a contrary inclination to an agent-centered one (parental love). The second is where a partial interest (parental love) acts as a contrary inclination to another partial interest, that is, to some other form of personal love.

The first kind of example might involve situations where the exhilaration of working in the cause of justice is pitted against the sometimes humdrum of parenting.^[22] Here we must imagine that in such a situation you truly care about the other person's just treatment rather than about your own moral self-aggrandizement. It is the simple respect for the other person plus the numbing effects of routine parenting that have led you to the neglect of parental commitment. Strongly inclined to help the other person, you are reminded of your child's plight. You then come to believe that you ought to do x for the sake of your child rather than do y for the sake of the other person. On some occasions of such reminders, your normative belief will not become a deontic belief. For sometimes the reminder will evoke feelings of love that will eliminate any competition between the partial interest in your child and the impartial one in the other person. But these reminders do not always eliminate the felt competition between these interests.

You might, for example, feel the competition between the two interests, even after recognizing that the parental commitment to your child comes

[22] . The film A World Apart deals with a woman and her relationship to her daughter as conflicts arise as a result of her commitment against the injustices of the apartheid policies of South Africa. See Chris Menges, A World Apart (Atlantic Entertainment and British Screen, 1988).

before the concern for the other person. You might especially feel this way when caring for your child is particularly burdensome and attending to the concern for the other is not as personally taxing. This is often true of the commitments of personal relationships. The concerns that often lead good people to do things that are wrong are not like those that lead others (especially the elusive egoist) to do things that are wrong. It is a mistake born of a certain way of reading Kantian thought to overlook this. In itself, it is a praiseworthy concern that leads the biased parent to sometimes wrongly favor his or her own child when he or she should not. To remove the objectionable element in the bias, it is not love that needs to be subtracted but respect that needs to be added. Likewise, it is an intrinsically praiseworthy concern for another child or person that might sometimes lead a parent to wrongly favor the third party over the child. This is especially true in the humdrum-type contexts. Respect does not need to be subtracted, but love brought to the forefront.

When you are reminded of your child's plight and feel a conflict between your interests in your child and the other person, the conflict does not show that you do not love your child. Rather, it is your parental love for your child that acts as a check on the contrary inclination to help the other person instead. Suppose you are currently justifiably frustrated with your child. You might need to be reminded that this is indeed your child in order to avoid the kind of neglect that can come about through what are otherwise admirable motives.

Here it is your love for your child that allows you to check the contrary inclination to help the other person. The competing nature of this contrary inclination is properly analyzed as a combination of two factors. One is your simple respect for the other person; the other is your justifiable frustration with your child. But the frustration with your child is not itself a motive. Rather, it is a psychological influence that serves to detract attention from the rightful place of your child in your sentiments and commitments. This is contrasted with a case in which you might help the other person rather than your child for the express purpose of venting the frustration with your child. Parental love comes in to deal with the frustration in such a way that allows you not to eliminate it but to check it. This is why the interest in the other person remains a psychologically competing interest. It is also why your other-restricting normative belief regarding the interests of your child and the other person is a deontic belief for you, because it results in your having a felt sense of obligation in this and similar contexts.

Of course, if you always find the parental routine numbing and frustrat-

ing so that the concern for others leads to your neglecting your child, we must question your parental love. But the presence of some such frustration goes with being psychologically normal, as does the ability to manage these effects with being a normal, loving parent. Thus without the ability to check the contrary inclinations of a normal person with impartial concerns for others, your sentiments, commitments, and dispositions would be incomplete. They would be incomplete as those of a loving parent and would not properly reflect the place of your child in your life. Perhaps this is why patience is such an indispensable parental virtue.

Another way in which contrary inclination might arise regarding an other-restricting normative belief is where the scope restriction applies to the interests of another loved one. You might, for example, have personal love for both your child and your spouse and have the other-restricting normative belief that you ought to do something for the sake of your child rather than do something for the sake of your spouse. There will certainly be situations where the above belief will be accompanied with frustration with the child. And it is your parental love that will act as a check on the contrary inclination of your romantic love for your spouse. Your spouse, for example, might want to spend the evening at the cinema, whereas your child might need help preparing for an algebra exam, partly due to the fact that she has been negligent in her studies. That your daughter has been negligent is frustrating to both you and your spouse, but certainly in some such cases there could be no doubt that the child's interest comes first. Of course, as a loving spouse who is also a good parent, your spouse will understand this. In other cases, he or she rightfully and lovingly will not.

5.

I conclude, then, that if you have parental love for your child, you will, in many normal contexts, have both self-restricting and other-restricting normative beliefs regarding your child. Often these will be deontic beliefs, resulting in your having a felt sense of obligation. The significance of this is that there is a set of normative beliefs that comes with the disposition of personal love in its parental form. This precludes a person from having the dispositions of parental love and giving lexical priority to impartial concerns in all contexts in which they conflict with partial ones. In fact, the dispositional sensitivity of parental love is such that it requires, in some normal contexts, a reverse ordering of priorities. But contrary to the worry of some, parental love, despite its favoring of the child, leaves significant room for commitment and concern for others. Moreover, respect requires

it, though not in the ways or for the reasons that the Kantian or the utilitarian would require. Thus the priorities reflected in the normative thoughts of an agent who is both a loving parent and a respectful person reveal the symmetry between the agent's impartial and partial regulative norms. And the agent's deliberations that reveal the structure of these norms take place without an impartial decision procedure. Rather, they are the result of rational reflection by an agent with a respectful and loving character on priorities that structure a way of life in terms of finality and self-sufficiency. This argues for the Aristotelian conception of practical reason as it applies to the agent of integrity in the thick sense as thus far understood.

This completes my sketch of a solution to Stocker's problem, the integration problem, and the priorities problem as they apply to parental love. The sketch not only indicates what the priorities of a loving parent who is respectful of self and others cannot be; it also provides an outline of what they must be. In this regard, it provides an account of how parental love and respect together, as forms of direct concern for others, generate the normative beliefs of a respectful and loving parent. Finally, it gives an account of the motivational match between concern and normative belief in a way that is consistent with a sense of obligation that avoids the problems associated with acting on principles foreign to that sense. In Herman's terms, right action is the nonaccidental result of its motive. In the next several chapters, we will see how this analysis applies to other forms of personal love. These are forms of what I call peer love: friendship and neighborly love.