Constraining Rights and the Value of Status
Constraining Rights and the Value of Status
Abstract and Keywords
Examine the question of whether it is morally permissible to treat people in ways ruled out by the Principle of Permissible Harm (PPH; this was introduced in Ch. 7 and provides an account of certain restrictions/constraints on killing) only for the sake of minimizing violations of the PPH itself, or whether there is a constraint on doing this. Having considered alternative grounds for a constraint in Ch. 9, Ch. 10 fleshes out a victim‐focussed, agent‐neutral, rights‐based view founded in a strengthened PPH right (constraint), which protects against minimizing violations of the right by violating the right; consideration is given to whether and in what sense minimizing violations of PPH rights by violating them would be both strictly irrational and also exhibit lack of concern for the right. Rejecting this as a route to founding the constraint (on grounds that strict irrationality would arise only if there were already a constraint (or an absolute right not to be killed)), a constraint is generated by focusing on a concern that is at the heart of the PPH and applying it to the pursuit of any goal (utility or minimization of rights violations), the chapter considers how the permissibility of minimization would alter every person's status, and examines the distinction between eliminating a right, violating it, and infringing it, focussing on the significance of negative residues of, and compensation for, rights violations. An exploration is made of whether the structure of deontological and consequentialist theories can be brought closer together via the agent‐neutral value of an inviolable status (of a certain sort), though a distinction is made between the irrationality argument against minimizing the violation of constraints, and support for a concept of the person as strongly inviolable. It is also considered whether creatures who are inviolable are therefore more important entities whose existence makes the world a better place and whether belief in a constraint affects both how good the world is and the effect of acts done in accord with or in opposition to the constraint; further examination is made of the futility of permitting minimization of rights violations by violating rights (‘futilitarianism’) by contrasting the role of utility vs rights per se in motivating minimizing.
Keywords: agent‐neutral views, consequentialism, constraints on killing, deontology, inviolable status, minimization of rights violations, Permissible Harm, Principle of Permissible Harm, restrictions on killing, rights elimination, rights infringement, rights violations, rights, rights‐based views, status, value of status, victim‐focussed views
. . . from the fulfillment of this wish he can expect no gratification of his sensuous desires and consequently no state which would satisfy any of his . . . inclinations, . . . all he can expect is a greater inner worth of his own person.
Immanuel Kant, Groundwork of the Metaphysics of Morals
Common Foundation for Prerogatives, Restrictions, and Constraints
In the two previous chapters we described some attempts to decide which potential victim(s) should be protected in the Constraint Case in which one person could be killed to stop five from being killed to save yet others. We have also considered ways in which to justify the rights as restrictions presented by the Principle of Permissible Harm itself. We did this in order to deal with Scheffler's argument that, if it is no more important to prevent the violation of a right than to prevent a comparable harm brought about by natural causes, then if we may violate a right to minimize violations of comparable rights, we may violate that right in order to minimize disutility. This is essentially an attack on the correctness of the PPH or any form of restriction. We dealt with these issues primarily on the temporary assumption that the single potential victim who will lose life he would otherwise retain independently of imposition on us or other potential victims ran no risk of being one among the greater number of victims.
The ways to justify constraints when this assumption holds according to the strategies discussed in chapters 8 and 9, are (1) to show that some person differs from all the other potential victims either in himself or in the relationship in which he stands to other potential victims (this is a victim‐focused approach), or (2) to show that the relation in which the agent stands to his victim is different from his relation to other victims as an agent‐relative approach, or (3) to show that the content of agency is immoral (an agent‐focused approach), or (4) to show that the agent's awareness of himself limits his conduct. Option (4) was taken to include the agent's noting that his victim cares more for his own survival than he does for the survival of others, or (p.260) cares more that harm to him should not motivate action than that harm to others not do so. This consideration was included in an agent‐concerned view because, although each potential victim may, according to this theory, think the same, it is the agent's own victim whose concern looms largest to him.
My suggestion is that these approaches to selecting the potential victim who is protected are incorrect. Let us summarize briefly the alternative position that will be presented in detail below. To understand why we may not kill the single person, we need not look for a distinguishing asymmetry that makes that person different from all other potential victims. He may have the same characteristics and stand in the same relations as the others, but these properties and relations nonetheless constitute an impassable barrier—a restricting and constraining right—which we come up against when we would have to harm him for the sake of others. Each one of the potential victims has this same property, but this does not mean there is any problem in choosing whether to kill the one for the sake of the five. Further, this property need not be connected to the victim's caring more that he not be violated than that others not be. When I say that the constraining right does not rest on an asymmetry, I mean that there is not necessarily an asymmetry between the position the single person will be put in vis‐à‐vis the five for whom he would be sacrificed, and the position those five will be placed in vis‐à‐vis the persons they are sacrificed for. The constraining right may, however, be crucially tied to asymmetries in the relations between the one and the five for whom he would be sacrificed, as described by the PPH. For example, the single person would be used for the sake of the five as they would not be for his. These asymmetries give rise to a restriction, that is, a right against ordinary welfare maximization. A strengthened restriction gives rise to a constraint against minimizing violations of the restriction by violating the restriction itself.1 The point here is that if it were not for this strengthened restriction, the victim would be treated in the same inappropriate way that is proscribed by the PPH, only for a different goal, that is, to protect other people from rights violations. He ought not be so treated.
Notice that there is a difference between saying on the one hand that each person has a personal point of view, leading him to care more for his projects than for those of others, incorporating this fact (in some way) into a defense of the person's right not to be used for others, and claiming on the other hand that this right truly constrains, when others' rights are also at stake, because the person we confront cares more about his rights than about the rights of others. The former use of the personal point of view could be correct and the latter still be incorrect.
If it is true that the single person has no distinguishing features that protect him but not the others, and if, nonetheless, it is wrong to harm him for the sake of others, then all the potential victims have a property that gives rise to a constraining right, a right that would not exist in any of them if it were permissible to kill the one to minimize rights violations in the others. The constraining right each has, furthermore, expresses a new value, the conception (p.261) of the person as inviolable in a way he would not be without it. It is even a sublime and elevated conception of the person. This value can be the basis of a non‐agent relative rights‐based account of the protection afforded the person against being used to minimize rights violations. That is, there seem to be two different routes to take when one sees that the single individual has a right. In both routes I come up against his right if I try to kill him, and this stops me. The five will also be killed but not relative to me; I let them die. The agent‐relative route interprets all this to mean that my basic duty is that I not kill. According to the second route when I see that the one has a right not to be killed, I should see that if it were permissible to kill him, this would mean that each of the five is also killable to save others from being killed (even though they wouldn't be killed if I saved them). By contrast, if the five are killed because it is impermissible to kill the one, this does not mean that morality endorses the permissibility of their being killed. Now suppose we care about the status of persons as inviolable (we care about nonkillability, not only not being killed), and each of us has a duty to promote this value. Suppose we are tempted to kill the one to promote this value. When we realize that we cannot promote the value by doing this—indeed the permissibility of doing it would deny the existence of the valuable status in everyone—we realize it is impermissible to kill the first person we come up against. Here concern for a victim‐focused, agent‐neutral value that all have a duty to promote, leads to a derivative (versus basic) duty that I not kill even if others do. (Notice that this implies a non‐agent relative reason why permissible killing could be more serious than permissible letting die (of harm): the former endorses violability, and expresses a different conception of the status of persons, the latter does not endorse violability. The factors pointed to in chapters one through four do suggest why what we do and what happens to someone is morally different if we kill rather than let die. But even if someone will have been killed by someone else if we let him die, there is a difference in what is true of the victim if killing him is permissible from what is true of him if letting him die is permissible. This may justify doing more to avoid killing than to avoid letting die of harm.) The first, agent‐relative, route begins by assuming that the one person has some right not to be killed that I come up against, but it is only the fact that I would have to kill that makes it true that the single person should not be killed even when preventing five from being killed is at issue. The second route (via concern for a certain status) gives an account of why the one person has the right not to be killed even when the five would be killed, a constraint that I then come up against. We can see the right of the single person not to be killed to save others as an expression of a certain status he and all persons have; the impermissibility of my killing him is implied by the requirement that I respect this right by doing what it tells me to do when I come up against it.2 This is a victim‐focused approach, which provides an answer to the Selection and Value Problems.
Are restrictions and constraints connected with a prerogative? It seems wrong to break the connection between prerogative, restriction, and constraint (p.262) altogether, even though, as noted earlier, sometimes the prerogative is not present when the others are and vice versa. This may be because, although the restriction and the constraint do not stem directly from the prerogative, in some cases they all stem from the same source, namely, a person's entitlement to control something, because he is not merely a means to the greater good. The prerogative does not, I have argued, derive only from the fact that persons have a personal point of view and are motivated to act from it. It derives from the fact that persons, that is, rational beings, who are capable of a personal point of view, are not to be always used as mere tools to the greater good if they do not choose that as their goal. They are ends‐in‐themselves. (Nonrational beings with a personal point of view are not necessarily accorded such treatment.)
Contrary to Scheffler, this same ground is available for restraints and constraints. In those cases in which there is no restriction on taking what someone may refuse to give, but still a prerogative, the prerogative may be based solely on the greater strain of being the cause of one's own loss rather than suffering the loss at someone else's hands. In cases where there is no prerogative (i.e., there are obligations), but there are restrictions and constraints, some extra value is seen to come from an individual's fulfilling his own responsibility to act.
Let us grant that, from a formal point of view, having a constraining right strengthens a right and shows deeper commitment to avoiding the inappropriate relationship between persons and other ends proscribed by the PPH. Let us also suppose that it expresses a more elevated status of the person it protects. (We will need separate discussion to show these things.) Could it be a characteristic of morality itself, rather than of persons, that unavoidably implies a system with such a strengthening mechanism, so that a constraint, for example, is in this way not merely an optional feature of a moral system, but a requirement? Arguing that there is a characteristic of morality itself that requires a certain view of people, where this does not just mean that morality recognizes certain characteristics of people as giving rise to prerogatives, restricting rights and constraints, involves the claim that we could not carry out the demands of morality (be moral agents) unless we had certain rights, constraints, and prerogatives.3 So to the extent that we must be moral, we must have these. I do not believe that this way of arguing for the existence of rights and prerogatives will succeed. It seems much more likely that it is the characteristics of people that give rise to the rights, constraints, and prerogatives to begin with.
Irrational Ways to Show Concern
Let us now consider in more detail whether rights‐as‐constraints are rationally possible or whether caring about rights and the persons they protect commits us to minimizing their violation, that is, commits us to having a utilitarianism (p.263) of rights (as Nozick calls it),4 or making of rights violations an agent‐neutral disvalue (as Kagan puts it). This constraint could be analyzed as having two components—a core, which is an ordinary right not to be killed to save other lives as the PPH describes, plus a strengthening component, which prohibits killing to prevent violations of the same or a less significant right. Together these two components make the constraint a sort of super right. I will first discuss this issue in cases that are to be imagined as involving a single potential victim of whom it is known that it is impossible that he should ever himself have faced the threat of death that the five now face, or even have been in any other life‐threatening position aside from our threat to kill him now. The aim is to deal most straightforwardly with the case in which someone's sacrifice would have to be justified solely in terms of protecting rights of others, rather than minimizing the chances of his own rights' being violated.
The basic thrust of the rights‐as‐constraints view is some sort of non‐substitutability that is ascribed to each of the individual would‐be victims. If in the Constraint Case the five who will be killed if the one is not have their non‐substitutability ignored by being killed in order to save others, this would not alter the fact that the single person should not be substituted for them by any agent.
One strategy for defending the coherence of such a view is to describe a possible right whose violation it would be irrational to permit us to minimize by violating a comparable right. Such permission for minimization would be irrational because what would motivate it would be concern for the right, and this concern could not rationally be served by violation of that right.
One point this strategy emphasizes is that if caring about a right and the status of the person it expresses involves heeding what it demands, then minimization‐by‐violation does not show concern for that right since it will not heed the right. A second point is that if morality sought and permitted minimization of the violations of a certain right by transgression of that very right, it would essentially eliminate that right and the concept of the person it expresses from the moral system. Minimization‐by‐violation would be futile as a way of showing concern for rights; it would be a “futilitarianism” of rights.
Before proceeding, however, notice that one might respond to the question “If you care about rights, why don't you minimize their violation?” by stopping the “caring” approach in its tracks. One might simply say that it is not because one cares about rights that one does not transgress the right of one person in order to minimize rights violations. Rather, it is because one respects rights. As noted above (p. 261), I have chosen to see whether the notions of caring and respecting can be interwoven, to see whether, for example, respecting the single person's right could be seen as the best way to care about a right, and a certain view of the person with which the right expresses. I will distinguish this sense of caring from other, more common senses, below. One might also respond to the question “If you care about rights, why don't you minimize their violation?” by emphasizing that it is persons as ends‐in‐themselves, not their rights per se, that one should care about, and any answer to the question (p.264) must be consistent with this second concern. We shall try to remain conscious of this point.
An Argument for the Irrationality of Minimization
Let us begin by supposing (for the sake of argument) that there is a PPH right. Now consider the idea of a specified right, by which I mean a right that implicitly or explicitly tells us to what extent the individual is protected from transgressions against her, a right that does not leave this matter of extent entirely vague. The PPH itself is, in a sense, already a specified right: it specifies how someone may and may not be treated. This is manner specification. By specified right here I mean the further specification of the numbers of lives we forgo in order not to treat people in violation of the PPH. This is numerical threshold specification. Suppose the right not to be killed in violation of the PPH were absolute, and that this meant that we had a right not to be killed for utility maximizing, for example, in the Transplant Case, no matter how many other people could be saved from being killed by our being killed. It might be suggested that to say it is permissible to violate one person's absolute right not to be killed for the sake of utility maximizing, in order to prevent the violation of other people's comparable absolute rights, is just to deny that people have such an absolute right. If it were permissible to minimize violations, fewer people might in fact die of violations of this absolute right, but people would no longer really have the absolute right not to be killed for utility maximizing. This further implies that they would not have the status of persons who had this degree of inviolability. (Or at least so the argument we are examining claims.)
Suppose the right not to be killed in transgression of the PPH is not absolute. For example, suppose (just for the sake of argument) that the PPH right specifically forbids killing one person as a means to saving up to ten others, but permits such a killing when more could be saved. Given a situation in which five people will be killed in violation of their nonabsolute PPH right unless we kill one, the decision to minimize the violation of the PPH right, out of concern for the PPH, by killing the one would, it is suggested, be confused. This is because the PPH right specifically excludes this as a correct course. If we care about the right, and the degree of inviolability of the person it represents, why should we act against its clear demands? (This is the first point about not heeding a right.)
Furthermore (this is the second point), for morality to endorse an act against the clear demand of the right seems to involve eliminating this version of the PPH from morality, and recognizing a concept of the person that makes it permissible for a person to be used in this way. Not minimizing violations is, therefore, in a sense to minimize violations (and show maximal concern for the right and the inviolability of the person it expresses) consistent with the right and that degree of inviolability being a part of morality at all.
Something seems to be wrong with this argument. We wish to distinguish between the restriction, as represented by the PPH, and the constraint. The PPH insists only on not doing certain things to maximize lives saved (or utility, in general). Suppose we kill one person to prevent two others from being killed in violation of their PPH right. This means the two might otherwise be killed to maximize lives saved, but the one person we kill would be killed in order to prevent the violation of two rights. It might also be true that we do save two lives at the expense of only one, but that is not our aim in acting. We aim to protect rights, it is said.5 Therefore, if intention is relevant, we do not aim to violate the PPH (which is concerned with not killing to maximize utility), and we may not actually violate it, since we do not act against its stricture. Hence, if what morality endorsed (permitted) was not a transgression of the PPH, it would not, in permitting the killing of the one to minimize rights violations, eliminate the PPH‐right from morality. It would not say it was, after all, sometimes all right to kill one as a means of saving simply more lives.
Suppose it is true that the PPH is not transgressed, strictly speaking, when we kill the one in the Constraint Case, because it explicitly makes reference to utility or saving lives. Then, killing the one to prevent transgressions of the PPH will not involve minimizing PPH‐rights violations by violating the same right, that is, a right that forbids killing people (in a certain way) to minimize lives lost.
A Constraint Already
If the PPH were already an absolute constraint (i.e., if its subject matter dealt specifically with forbidding the killing of people to minimize rights violations), we could tell a different story. Then it would be self‐contradictory to say that it is morally permissible to minimize violations of the constraint for the sake of showing concern and respect for it, or because one cared about it and the concept of the person as inviolable in the way the constraint expresses. For example, suppose there is a constraint not to be treated in the manner prohibited by the PPH (e.g., to be harmed as a means) in order to prevent violations of rights. To endorse minimizing violations of the constraint by permitting transgression of that constraint, would amount to the rejection of the constraint.6 Indeed, if the constraint did not prohibit using people to maximize utility but only prohibited using people to minimize rights violations, it would make more sense as a formal matter to permit transgressing such a constraint for the sake of maximizing utility than to permit transgressing it for the sake of showing respect for the constraint or the degrees of inviolability it represents.7 But we were trying to argue from the (hypothesized) existence of a restraining right to the existence of a constraint that would prevent killing one to minimize violations of the restraining right, and (p.266) we will have failed to do so by this “inconsistency” argument if we have to assume what we were trying to prove, namely the constraint.
Suppose that, strictly speaking, killing one to save others from being killed (to save yet others) is not minimization of PPH‐rights violations by violation of the PPH. There will still be other rights whose violation we could try minimizing by violating an instance of the right. These will be either absolute rights not to be killed or rights that specifically direct us not to kill to prevent violations of rights. To permit violating such rights in order to minimize violations of the right as a sign of concern and respect for the right will be irrational, since such permissibility will go against what the right itself directs. It will also make impossible the existence of a certain value, that is, whatever degree of inviolability of the person the right represents. However, the reason to endorse such rights—which are already constraining rights—is not that without them mere PPH‐restraining rights cannot exist (as the first argument claimed), but rather that without them the status of a highly inviolable person, represented by the constraining right, and whatever value this has, cannot exist.
This has been a brief summary of an argument that I shall now describe in more detail. We can begin by restating all that was said (mistakenly) in connection with the PPH. Let us begin by considering the idea of a specified right. By a specified right, I here mean a right that implicitly or explicitly tells us to what numerical extent the individual is protected from transgressions against him, a right that does not leave this matter of extent entirely vague. This could be an absolute right or one with a numerical threshold. Suppose we had such a right not to be killed in certain ways, as suggested by the PPH, but it was not limited to a prohibition on killing in these ways for greater utility. This right has the same formal structure as the PPH—that is, the manner in which one is killed, the relation between victim and beneficiaries that are thought significant are the same—but it protects us against being killed in violation of the PPH's formal structure for the additional end of minimizing rights violations. This right has the manner requirement of the PPH at its core, but not necessarily the whole PPH where this implies the prohibition on utility maximizing. Why might we think there is such a right? Because, if the way of harming people prohibited by the PPH is objectionable, then we show just how objectionable it is by excluding that way of harming for other goals, utility or minimizing rights violation. This would also make people more inviolable. Suppose this right which is a constraint were absolute, and that we had a right not to be killed in the way we would have to be killed in the Transplant Case, no matter how many other people could be saved from being killed by our being killed. To say it is permissible to violate one person's absolute constraint not to be killed in order to prevent the violation of other people's rights (one of which is a constraint) eliminates from morality that (p.267) absolute constraint. This is because, if that killing were permissible, it would no longer be true that morality insists that it is absolutely wrong to kill a person who would otherwise live, in order to minimize rights violations. Under such an interpretation, if we minimized violations of the absolute constraint, fewer people might in fact die of violations of this constraint, but people would no longer really be protected by the absolute constraint on being killed. This further implies that they would not have the status of persons who had this degree of inviolability.
Suppose the constraint on being killed (with the formal structure of the PPH but ruling out killing for ends besides utility) is not absolute, but specified numerically. For example, suppose (just for the sake of argument) that it specifically forbids killing one person as a means of preventing the constraint on being killed from being violated in up to ten others, even when they would be killed as a means of saving less than ten others. However, it permits such a killing when more comparable rights violations could be prevented. Given a situation in which five people will be killed in violation of this nonabsolute constraint unless we kill one, the decision to minimize the violations by killing the one would, it is suggested, be confused. This is because the constraint specifically excludes this as a correct course. If we care about the constraint, why should we act against its clear demands? This is the first point about the irrationality of trying to show concern for the constraint by not heeding it.
Furthermore, it is suggested, for it to be permissible to act in this way seems to involve eliminating this constraint and the status of the person it expresses from morality. Not permitting minimizing violations is, therefore, to show maximal concern for the right and the status, consistent with the right and the status existing at all.
What if the right is not specified with a threshold nor specified as absolute? That is, what if there is a right not to be killed that focuses on the same formal properties of the PPH (i.e., the manner in which one is killed, the relation between victim and beneficiaries, etc.) and that prohibits killing for the sake of rights in addition to prohibiting killing for utility. This constraint, however, does not include any explicit specifications of how many rights violations can be tolerated out of respect for the constraint. Suppose that, on a given occasion, we decide that in order to save a thousand people from having their constraints violated, we will transgress this constraint of one person, and it is permissible to do so. One understanding of this decision is that we have now specified the strength of the constraint, and so removed the vagueness that existed before. Since there was no specification to begin with, however, granting permission to minimize violations of the right does not eliminate the right from morality. If it wasn't specified to begin with, there is no elimination by, in essence, specifying for the first time.
However, notice the following complication. Consider, first, an unspecified right not to be killed. Suppose we decide it is permissible to kill someone as a means of saving one thousand people from being killed. It may be said that the right of the one not to be so used still exists and is recognized even in the breach. That is, without thinking that we act impermissibly, if we kill the one we may think that there is a negative residue produced by our acting, for which we would owe compensation to the victim. The claim is that there would be no such negative residue unless the victim still had a right not to be so used.8 Deciding that it is permissible to kill the person in this case does not conflict with the claim that he has a right not to be used to save a thousand people from having their rights transgressed, though we may permissibly infringe the right. Permission, or moral endorsement need not imply that people do not truly have the right and the accompanying status (and whatever properties it seems from or is associated with).
On this view, we have not, in fact, specified that the victim's right is not absolute (or we have not specified his right at a lower threshold than it is truly specified at) by granting the permissibility of killing. But, if all this is true, why, in the case of a specified right not to be used in a certain manner to save one thousand people who will otherwise themselves be so used, might it not also be said that the permissibility of killing the one to save the thousand from such killing would not—contrary to what was asserted above—involve a denial of the right not to be killed to save the thousand from such killing? Can we not here also use the language of permissibly infringing a right and owing compensation for infringing a right that we recognize is still in force? Such infringeable rights will be weaker than non‐infringeable ones, reflecting a concept of the person as more violable; this could be a reason to think they don't exist. But with an infringeable right, there will still be some right. In fact, having an infringeable right not to be used to minimize the violation of the comparable rights of 1,001 people reflects a concept of the person as more inviolable than having no right at all not to be sacrificed when preventing 1,001 rights violations is at stake. (The latter is the implication of having a non‐infringeable right not to be violated for up to a thousand rights.)
I believe that the crucial question about this proposal, so far as the present topic is concerned, is whether it is permissible to infringe the constraint that is specified only for protecting rights for some reason other than concern for rights violations, for example, utility, or whether concern for one thousand rights violations qua one thousand rights violations may permissibly prompt the infringement of the specified constraint as well. Similarly, the claim that permitting transgression of an absolute right not to be killed for minimizing rights violations in order to minimize rights violations eliminates the absolute right is consistent with the view that utility (or saving lives) is a permissible reason for infringing the right. It is also true that, if grounds offered for the infringement are the very factors that a specified right excludes as grounds for its infringement, that is, if it specifically says or implies “that we ought not kill (p.269) one to save one thousand from being improperly killed,” then the infringement cannot be done out of respect and concern for the right itself or the status of people which it expresses. Disregarding the instructions of the right shows disrespect for the right, not respect for it.
To make this issue clearer, let us consider in greater detail the relation between specification, elimination, and infringement. Suppose someone, A, has a right which is described as the “right to her house,” without any explicit specification of threshold in the content of the right. It is also agreed that if someone, B, would die of exposure unless he breaks into the house in the absence of the owner, it is permissible for him to do so, although he may then be liable for compensating the owner. The favored way of describing this is that A's right to her house is infringeable; the fact that she is owed compensation shows that she had a right and that she was done some wrong in the course of someone's doing the overall right act. Suppose this is the correct description of the case. May we at least conclude from its being permissible for B to enter the house in A's absence that, whatever right the owner had, she had no right to exclude the person in dire circumstances when she was not using her house? If so, then it is not true that the reason compensation is owed is that the right to exclude the person in dire circumstances was infringed. We can also conclude that the fact that helping the person in dire circumstances is sufficient reason for infringing A's right to her house is just what implies that there is no right to exclude the person in dire circumstances. That is, more generally, if reason x overrides the right to y, there is no right to y even if x. If a reason is grounds for permissibly infringing the right, then a right that was specified to include in its content the denial of the reason would be an invalid right. That is, this right must be eliminated from morality.
If this is all true, what can we say about the claim that it is permissible to infringe a right that says someone is not to be killed, when it would be “infringed” for the sake of preventing a thousand (or fewer) from being killed? If a reason allows us to permissibly infringe a right, then the denial of the reason cannot be part of the content of a valid right. So if it is permissible to kill one for the sake of saving a thousand from being killed, we must eliminate, not merely infringe, a right that says someone is not to be killed even to save a thousand being killed.
There would, however, be a difference between our case in which someone's right to her house is permissibly infringed and cases in which someone would (by hypothesis) be permissibly killed to save a greater number from being killed. This difference might suggest a counterargument. In the house case, I believe, we think there is no right on the owner's part to resist the imposition. (This does not necessarily imply that there is a duty on her part to donate her residence.) In the killing case, even if we thought it was permissible to kill someone to prevent a certain number of killings, we would, I believe, think it permissible for the person to try to resist. Would the permissibility of resistance signify that he has after all a valid, specified right not to be killed to prevent a thousand killings, though it is permissible to kill him solely to prevent a thousand killings? I think it only signifies that he has a valid right to (p.270) resist. It does not signify that he has a specified right not to be killed to prevent a thousand from being killed, and that we only infringe the right if we kill him to save a thousand from being killed.
Nevertheless, I repeat, that the right specifically prohibits killing to save one thousand from rights violations does not mean that one could not permissibly infringe the right for the sake of saving the one thousand lives, that is, out of concern for life rather than for rights. This would, however, violate a specified PPH. We shall consider this further below (p. 282).
Notice that if there truly are numerically specified (rather than absolute) rights, this could be consistent with the existence of a negative residue in cases where violating a person is specified (explicitly or implicitly) as permissible. The negative residue can express the sense that “it's a shame” people do not have a right to greater inviolability, and if they do not have a right to such greater degrees of inviolability, at least they have a right to be compensated for this sad truth. Put somewhat differently, suppose the list of reasons that make infringing a completely non‐specified right permissible implies a set of invalid would‐be specified rights (which is what has been argued above). Nevertheless, suppose trying to compensate is required even where infringement of a non‐specified right is permissible (though this does not mean infringement is not permissible if compensation is impossible). Then it could be said that what someone has a right to is that there be transgression for a certain reason only if there is an attempt to provide compensation.
This analysis of when permissible infringement is consistent with continued endorsement of a right shows that this topic is related to the criticism of views (e.g., Sen's) that say that we may permissibly transgress rights to minimize rights violations but not for the sake of utility.
A Hierarchical System
A prohibition on minimizing rights violations by violating the right in question produces what might be called a hierarchy effect among rights violations. If someone is interested in minimizing, he counts violations of comparable rights. But when the absolute or specified right tells us that in this situation all things considered it would be wrong to transgress a right in order to save other people from comparable rights violations, we are not at liberty to say to ourselves, “The violation of this one person's right would be one wrong, but the violation of the rights of five others would be five wrongs, so better one wrong than five.” When the wrong involves doing what all things considered the right says we must not do, not doing it dominates, if we are to show respect and concern for the right and for the idea of the person it expresses.9 It is because of their specific exclusion of these other killings as justified reasons for infringing the right that in deciding whether to kill the one to save the greater number from having their rights infringed, we are not faced with a dilemmatic situation. So far as concern over rights (even when utility is associated with the transgression of rights) there is no dilemma, where dilemma is defined as a choice situation in which whatever we do, we do something (p.271) wrong. (There may also be no dilemma because the duty not to harm is stronger than the duty to aid.)
Once we understand why minimization of violations of a particular right by violating a comparable right is wrong, the additional supposed problem of how one selects, among all the individuals who have rights, the one who will not be killed—the Selection Problem—seems straightforwardly answerable: we select not to kill anyone we would have to kill in violation of the constraint. This is so, not because we, standing in a special relation to the person's right, generate a constraint for the first time. Rather, it is because we come up against the constraint of the right and the properties of a person that give rise both to his being inviolable in this way and to such a constraint. Analogously, we need not think that the power of a brick wall to constrain us is a new physical‐chemical output of the interaction between intrinsically weak building materials and our body. Rather, whichever intrinsically constraining brick wall I meet first will constrain me. I will be there when it constrains me, but not because I help produce the constraint.
The Significance of Permission and Status
It is important to see that it is the permission to kill, the endorsement of killing, not any actual killing, that can eliminate certain rights not to be killed and a status of high inviolability from the moral system. If the comparable right is violated many times over, this does not involve morality endorsing such violation; though many people may be killed, we do not say that it was correct that any of them were killed, which is what permission would involve. More people may die, but the conception of each person that is morally endorsed involves a higher degree of inviolability. We have already noted two different reasons that might be given for endorsing a transgression of the one. The first is misguided concern for the right and the inviolability of persons it expresses, that would produce a system that eliminates the right and the inviolable status of the person. A second reason for endorsing transgression of the one is greater concern for saving lives than for rights. In this second case, it has been suggested that the permissibility of the killing may not involve a denial of the constraint. To repeat: this is because infringing the right for the sake of some competing value, for example, utility, can leave the right still, existing, if the right itself excluded only minimizing rights violations. Permitting violation of the constraint for the sake of utility is not as formally irrational as endorsing violation for the sake of respect and concern for the right. (This does not mean that endorsing violation for the sake of utility is correct. It will, of course conflict with the PPH right which says we should not kill persons to maximize utility and it denies a highly inviolable status to persons. We discuss formal irrationality again at p. 282.) Suppose that right, unspecified or specified, were not eliminated by permitting its transgression for the (p.272) sake of minimizing rights violations; it was only permissibly infringed. This would still reveal that the person had a less inviolable status. In this case the right would no longer be the sole expression of the status of the person; the degree of infringeability would express this as well.
On some views, non‐consequentialist constraints are best characterized as concerned with the distinction between what I as an agent do and what happens; I must not do something, even if this means that something else happens. Without denying that, in a non‐agent‐concerned sense, this is true, the view being presented here emphasizes a different distinction, that between what it is permissible to do to people (i.e., their status), and what happens to people (their fate, which may involve more rights violated). What I do, rather than what happens, is important when it reveals what it is permissible to do, that is, what the status of a person is. The realm of status is not what happens to people. If many are killed in violation of their rights because we may not kill one to save them, their status as individuals who should not be killed does not change. If it were permitted to kill the one to save them, their status would change. We may be concerned about what happens, but be unwilling to prevent it in a way that is only consistent with a change in status. It is a mistake to see an opposition between the rights of the one person and the rights of all others, since the status of everyone is affected by the way it is permissible to treat one person.
The Significance and Type of Inviolability
Suppose the correct constraint prohibits many types of killing even for very good causes. If we are inviolable in this way, we are more important creatures than more violable ones; this higher status is in itself a benefit to us. In consequence, we are creatures whose interests as recipients of such ordinary benefits as utility may be more worth serving. (It is having the status itself which is a benefit, not just its being respected. If only actual appropriate respect were a benefit, then one would be better off in a world in which one had a low status but this was fully respected than in a world in which one had a higher status, though it was not fully respected. If having a higher status were not a good to the person who has it, we should be indifferent between having a lower status but being treated better than we deserved (to level x) and having a higher status fully respected (to level x). Having the status is a benefit, in part, because it makes one worthy of respect, owed respect. Without this, respect becomes merely a psychological attitude others take toward one.) Furthermore, the world is, in a sense, a better place for having more important creatures in it. Our having higher status is a benefit to the world. In this sense, the inviolable status (against being harmed in a certain way) of any potential victim can be taken to be an agent‐neutral value, and as providing a solution to the Value Problem. This is, however, a nonconsequential value. It does not follow (causally or noncausally) upon any act of ours, it is not something we must aim to bring about as a consequence of our acts. It is already present in the status that persons have. Respecting it provides the background against which we may then seek the welfare of persons or pursue other values. It is not (p.273) our duty to bring about the existence of such valuable persons, but only to respect the constraints that express the presence of value. Kagan claims that the only sense in which we can show disrespect for people is by using them in an unjustified way.10 Hence, if it is justified to kill one to save five, we will not be showing disrespect for the one if we so use him. But there is another sense of disrespect tied to the fact that we owe people more respect than we owe to other creatures, even though we also should not treat these other creatures in an unjustified way. And this other sense of disrespect reflects, I believe, a failure to heed the greater inviolability of persons.11
Individuals whose rights stand as a barrier to action are more potent individuals than they would be otherwise. There being rights and constraints with high thresholds is a mark that the person who has them is a stronger, more valuable type of thing, even if they prevent us from stopping more transgressions. It is true of all those who die because we cannot save them as well as of those who are not violated, that they have these rights. By analogy, a stronger, more impressive wall is one that we will not be able to pass through, even to prevent the destruction of comparable walls behind it. One would expect that the highest values or rights are intrinsically such that it would be wrong to minimize even a great many violations of them, either for the sake of rights or the sake of utility, by transgressing them; they would not be so almighty if they could be transgressed.
Suppose such a conception of persons involving a sort of inviolability, if true, would give persons great worth. What reason do we have to believe such a conception of persons is true? That is, why should we believe that the best is true, or even that what it would be best for us to believe about ourselves is true? Is it possible that in this area we can bootstrap ourselves into significance: If someone thinks he is inviolable, then he is? But thinking may make it so only because our capacity for such a self‐conception indicates something else about us—for example, that we are rational beings—which itself renders us worthy of certain sorts of protections, even if we do not believe ourselves worthy.
In any case, it is certain properties (here not enumerated) that we have as individuals that would account for our status. (I mean to emphasize “as individuals” quite independent of any relations we are, or are not, part of.) Further, it need not be that having a personal point of view alone makes one inviolable, for those whose reasons are never out of synchrony with the impartial point of view could be inviolable (in a certain way) in virtue of, for example, being creatures who act for reasons. They would just waive their right not to be violated for the greatest good. If creatures who act for reasons happen to have a personal point of view, the factors seen from that point of view can be legitimate reasons as well, meriting noninterference. Even if they are not legitimate as reasons, actions prompted by them, in such a creature, could give one a reason for abiding by a constraint.
But recall that all this talk about inviolability is only rough. It is inviolability in a certain way that is at issue. Consider the right not to be harmed in order to produce greater utility (rather than the constraint on harming someone to reduce violation of rights). This right is not absolute. This does not (p.274) mean only that there might be some number of lives (threatened by natural disease) such that we could kill one to save them, in the way the Transplant Case requires. It also means that we may kill one even to save two in the way the Trolley Case requires. Now the conception of the person could contain a higher degree of inviolability if it were not even permissible to bring about death in the way we do it in the Trolley Case (or in self‐defense cases, etc.). Yet we think, presumably, that it is morally permissible to do these things. Persons are not absolutely inviolable, and so the fact that the constraint protects their inviolability does not show that it does so to the correct or to the highest degree. Indeed, if killing were impermissible in the Trolley Case, but there were no constraint on killing the one to minimize violations of the rights of others, then quantitatively it could be that the same degree of inviolability might be expressed. Simple talk about inviolability is not enough. Restrictions and constraints are better explained by inviolability against impositions that create inappropriate relations between victim and beneficiaries.
Suppose the right (against being used to maximize utility) whose violation we contemplate minimizing has its content given by what I have referred to as the PPH. Then if we treat people in the manner ruled out by the PPH right in order to minimize violations of it, we will be harming someone in the very way it rules out, even if we do this for the sake of something other than utility maximization. (We do it to minimize rights violations.) If this way of harming certain persons is wrong, then we should not do it even to minimize occurrences of this way of harming them. It is not inviolability against harm per se that is represented by the constraint; it is the person's inviolability against this way of being harmed.
The Inviolability of Persons and the Inviolability of Rights
I must emphasize that a component of the argument against minimizing rights violations that I have given is so formal that it itself does little to answer the questions of why the manner restriction at the core of the constraint has the specific content it does have and what the numerical threshold is on the constraint. This formal argument is merely concerned with the elimination of a specified constraint (of any sort) by the permissibility of minimizing violations of it by transgressing it. This formal component does only a little to directly support the view that persons have a high degree of inviolability. Our argument for a high degree of inviolability comes in part in deciding that the PPH is correct, since the constraint just has the manner restriction of the PPH extended to deal with protection against minimizing rights violations. The correctness of this extension itself contributes to higher inviolability. The high degree of inviolability also results from it being correct to set a high threshold for permissible transgressions. Once the constraint with this content is in place on these grounds, we can argue that trying to minimize violations of such constraints by violating them fails to show respect and concern for the constraint. This last step is mostly an argument for the inviolability of the right (p.275) rather than the person.12 A person can only have the status that the constraint expresses if it is irrational to violate the constraint to minimize violations but this irrationality is not the foundation of the constraint. One might reject the PPH I have suggested and have a more consequentialist description of a core right allowing killing in the Transplant Case and setting a low threshold for utility maximizing, and use this to generate a constraint. Whatever PPH right is generated, it could be said that we show greater concern for its manner restrictions if we do not engage in behavior it prohibits for the goal of minimizing rights violations. This generates the constraint. Once a specified constraint is in place it will make no sense to permit us to violate the constraint to minimize violations of it. This is the formal part of the argument. If it is permissible to violate the constraint people will not have the same status it expresses—even if this is low inviolability—and the right will be eliminated, even if fewer violations of people occur.
Indeed, because respecting a right and respecting a person can differ, by minimizing we might show more respect for a right that expresses lower inviolability of the person and show less respect for a right that expresses higher inviolability of the person. For example, suppose R1 tells us that someone has a right not to be killed to save up to ten other people from being killed. R2 tells us that someone else has a right not to be killed to save up to a million people from being killed. If we would kill the person covered by R1 to save nine people, we show only minimal disrespect for the right, but the inviolability of the person is quite low. If we would kill the person covered by R2 to save five hundred thousand people but not fewer, we show great disrespect for the right, but the status of the person is highly inviolable.
In sum, the account of the constraint as an expression of a new value, namely a person with great significance, because highly inviolable, should be distinguished from the futility (irrationality) argument. That argument is concerned with self‐defeatingness, that is, the futility of trying to show respect for a specified constraining right by violating it, even to minimize violations. The irrationality argument claimed that utilitarianism of rights is a futilitarianism of rights. The argument which points to the value of the status of inviolability independent of what happens provides a deeper, more fundamental reason for not violating a right to minimize rights violations. It alone explains why there is a strong constraining right in the first place, by pointing to the significance of the status of inviolability.
Measures of Significance
There are at least two possible questions that can be raised about the view that higher inviolability is connected to the greater significance of persons.13 Why is one not a more significant type of creature if one is less inviolable but more free to violate, and why is one not a less significant creature because one is not the sort for whose sake one may permissibly violate others? Being free to violate and having high saveability are both statuses that can hold even if they are not reflected in what happens. Indeed, we must be careful to separate the (p.276) concern that (1) people have either of these two statuses to a greater degree than inviolability, from the concern that (2) what happens to people (whether they are actually saved) is more important than the status people have (e.g., inviolability). (We might imagine, for example, that as one's saveability went up, the actual numbers of people helped when in need went down.) Notice also that every degree of inviolability has its complementary degree of saveability; the question is what sort of combination of the two represents high status. Consider the first question. My sense is that if the person who gains legitimate liberty to violate is one of those who loses inviolability as well, he has come down on the ladder of significance. One does not gain anything in the way of status by being at liberty to violate a creature whose inviolability is less, and one will have lost inviolability oneself. Furthermore, the liberty one gains is not even a power that must be respected; that is, it is not greater authority. It can be resisted and for this reason also does not elevate its owner. If one is not of the same type as the creature whose inviolability is less, one's status still does not go up intrinsically merely because the status of another creature goes down so that one may violate it. It is only if one's status increases intrinsically to the point that one could permissibly violate someone, himself unchanged, whom one was not formerly permitted to violate, that one might seem to gain in importance. But even here one must be careful. If a creature is truly inviolable, how can one's own importance be less if one is not permitted to violate it than if one is? To say that one's importance is less and could still be increased is like saying that God's power is limited because he cannot change the laws of mathematics.
The second question is more difficult, I believe, since it does seem that if a creature is significant, it is wrong to abandon it and that some costs should be imposed on others to help it. Am I even more significant if I must be aided but I may leave others unaided? If we are all the same type of creature, universalization would apply; there could be no such comparative difference in significance: if I may leave others unaided, they may leave me unaided. One crucial issue in the use of constraints is whether we may impose a loss of a certain significant size on someone for the sake of stopping a comparable loss, only to each of several others. But sometimes constraints protect individuals from suffering losses that are significant, but still less than the losses others will suffer. Even in these cases, we need a good justification for why—as I believe is true—our own status as a person goes down rather than up if a great deal may permissibly be done to persons in order to save us. If our status is connected to being inviolable in a certain way, then it would be no wonder that the status of those saved would be decreased if everyone, including those saved, were violable in order to save persons. Suppose the high status of persons, making it important to save them, was held to stem from some other property than their inviolability. Then, to say that their status would go down if they were violable (in certain ways) even if they were more saveable is to claim that the properties that underlie one's inviolability are more important than those that give rise to any other significance one has. Might it be to claim that having a rational will, whose consent we must seek when interfering with (p.277) what a person has independently of imposition on us, will give a person higher status than being a complex, feeling creature who cares about whether it lives or dies? But what if it is only a question of the saveability of individuals whose rational wills are being interfered with? If these people are highly saveable, they have a claim on what others have independently of imposition on them. This denies a high degree of personal sovereignty and also denies the significance of its ground—which is also the ground of the moral difference between harming and not aiding: the existence of separate persons with rational wills.
A Better World, the Status of Persons, and Non‐Consequential Value
What further values does the sort of high inviolability represented by the constraint built on the PPH express? It may be said that if the single person is not killed to save the five in the Constraint Case, the world will be a worse place (at least in some ways) from the point of view of the rights theory itself—a world in which more rights violations occur, and for worse reasons, than would occur if one agent, trying to correct a bad outcome, killed one person. As an indication of this it might be said that we would punish an agent who kills a single person to save five others from being killed less severely than an agent who maliciously (or non‐maliciously) kills five (even to save yet others). Furthermore, it might be said, if we compare a world in which a morally deluded person, trying to do his best, kills one person to save others, with a world in which five people are maliciously killed or killed to save others, we must conclude the first world is better. The motivation of the killer is not bad, and we have a delusion leading to a rights violation. This contrasts with either maliciousness leading to five such violations, or five people dying as a result of a moral delusion.
We must be careful in doing this comparison. We are comparing a world in which one person, morally deluded and trying to do the right thing, kills one person, with a world in which five are killed, through a similarly morally deluded act or through a maliciously motivated one. This way of putting the comparison may be deceptive, in part because it suggests that we have a situation structurally analogous to a redirected‐threat case: we are offered a choice between outcome A in which someone kills one and outcome B in which someone kills five. In the redirected‐threat case, the failure of outcome B to come about is not a causal effect of outcome A's coming about. Outcome A is merely a substitute for outcome B. What the redirected‐threat picture omits about our comparison is that the prevention of the outcome in which five are killed is a causal effect of the one being killed in a way, which is a violation of the constraint form of the PPH. In comparing the outcomes we must not omit the causal connection from the picture.
One response to the claim that it is a worse world is just to say that doing what is right in not killing the one does not necessarily produce the best state of affairs, even from the perspective of concern for rights.14
A second response, however, is that, at the very least, the world will not be a worse place if a constraint is in fact merited because a constraining right that (p.278) expresses the right sort of inviolability produces a status for a person that is more sublime and elevated than a status that lacks it. This is an additional value that enters the world if people are worthy of inviolability. As was said earlier, this is a non‐consequential value. Its presence does not follow from any act, but is already present in the status that persons have. The tendency to say that the world would be a better place if, for good motives, an agent killed one person to save others from killing five, arises most clearly when we think of the agent who kills the one person as doing what is nevertheless wrong, and violating the constraining right. This is because it is compatible with a highly inviolable status of persons. The killing occurs against a background in which it is the wrong thing to do. By contrast, if he did the right thing, morality would endorse and permit the act. To say that the agent did the right (not merely excusable) thing in killing the single person is either to say that there is no constraining right and so people have a lower degree of inviolability, or to say that the right exists, but is too weak to stand up to much and hence that people do not deserve to have their inviolability taken too seriously. A world in which these things were in fact true could, in fact, be a worse world.
There is an additional point: If we also believe that the agent who would kill the one person does the wrong thing, we are not deluded about the truth. This may also play a role in our thinking the world better if he acts: for if we were all deluded in thinking he did the right thing, then our living with an incorrect as well as intrinsically less elevated conception of ourselves might make it a worse world. So if it is better that he kill the one—and it is not clear that it is—this is so because it remains wrong for him to do this and we don't believe it is right. (We shall discuss the issue of wrong and worse beliefs further later.)
Note that the points we have been making here are different from, though related to those on p. 272 and to a point that Thomas Nagel emphasizes:15 it is a better world if it is possible for wrongs to be done to us. For if it were not possible for the wrong of violation of the constraint to occur because we were not worthy of a constraint, then, although no wrong would occur if we were to minimize rights violations, it would be a worse world because we would have a lower status. (In other words, a world with only lions is worse than one with people in it, even if the people are improperly attacked and the lions couldn't be.) This is, I believe, analogous to Mill's claim that it is better to be Socrates unhappy than a pig satisfied, and even more to Kant's point that it is more important to be—i.e., have the status of a creature who is—worthy of happiness than to actually be happy, even, we might add, if having the higher status were the only thing that gave rise to the possibility of unhappiness. You can be subject to certain evils only if you have a high status.
In these remarks we have emphasized the good of inviolability and, perhaps, general belief in this truth. We have still to consider further whether, in a world with such entitlements and beliefs, it is better if one person acts in deluded fashion. Before doing this, however, we should make some additional points about value.
The analysis I have offered suggests that a nonconsequentialist, agent‐neutral analysis of a constraining right based on the PPH could be given. That is, the constraint need not be based on an agent's perception that harm to the potential victim of his current act has a magnified value; nor is it based fundamentally on the agent‐relative fact that I would kill my victim but only let die someone else's. Rather, the constraint may be based on the fact that if the agent were permitted to harm the one, this would defeat the agent‐neutral value (i.e., a value all should promote and protect) of a certain sort of inviolability of any of the potential victims. Being let die of a killing does not threaten this value. The agent's victim is special only in that he will be there when the agent makes his decision. The agent's own act is special only in that it makes him come up against the constraining right. As already noted, this makes the constraint victim‐focused rather than agent‐focused or merely agent‐relative; it is fundamentally about the potential victim and the fact that he is objectively an end‐in‐itself. The victim himself need only say that no one should be treated in a way that transgresses such a right; he need not say that his own fate is magnified for him.
Another way to emphasize that we offer a victim‐focused, not an agent‐relative, account of the constraint, and a proposal for agent‐neutral value is to note that we should morally object to a consequentialist natural order with no human agent in it that proceeded as follows: whenever several people were to have their PPH‐rights violated, another fell ill solely in order that his organs be available to save them. (The natural order here imagined is sensitive to moral wrongs; several people dying of natural causes would not trigger this odd illness.)
We can, therefore, think of the constraint as a way of expressing an agent‐neutral value: not the number of lives saved, but the maintenance of a conception of the person as strongly inviolable. As noted earlier, this is nonconsequential value. Its presence does not follow from any act, but it is already present in the status that persons have that provides the fixed background against which we may then seek their welfare or pursue other values. It is because of the constraints, or limits, morality places on us that the depths of morality, that is, the representation of the significance of persons, is possible. These limits, therefore, make possible the depths.
This way of explaining the constraint reduces the distinction between consequentialist and non‐consequentialist ethical theories in at least one way. The non‐consequentialists are identified as putting “the right before the good”; that is, the theory of right acts before the theory of value. Our way of explaining the constraint locates some new value which is such that its maximal presence is achieved only if there are constraints expressing high inviolability and it is not always permissible to act so as to violate rights to minimize rights violations. However, consequentialism is concerned with bringing about what some theory of value determines to be the best state of affairs. If the high (p.280) status of persons is a good whose existence is reflected in constraints, then, in this sense, constraints are connected with a good and a good state of affairs, but this good state of affairs is not brought about causally by any act that seeks to maximize the value (or even by its being true that people are protected by constraints). It is rational to act for the reasons that there are, and if persons are inviolable (in a certain way) this gives us a reason, which we may recognize, to act in accord with this truth, even if we do not thereby produce anything of value as a consequence. The agent‐neutral value I have emphasized can also reduce the gap between theories that are phrased in terms of respect for persons or rights (leading us to heed them) as opposed to those that emphasize caring for a value. If we care about this new value (whose existence can be expressed by the constraint), we act by heeding the rights of persons.16
Belief, Truth, and Action
We have been imagining how a certain value is present in the world if it is true that there are moral constraints, making it a better world. We have also assumed that we believe that there are such constraints. But these two conditions can pull apart.17 What we have said above argues for the view that it would be a worse world if there were, in reality, no constraints. The hypothesis is that no world can be the best unless it is impermissible in that world to kill one person to save five even from being impermissibly killed. But we also considered the view—and here we return to deal with the question that prompted it—that so long as it is true that there is a constraint, it might be a better world if one morally deluded person killed one person to save the five. (If we are not consequentialists, we do not have to believe that it is permissible to kill the one if it will be a better world if he is killed.) What of a world in which not just one person, but none of us believe the truth that people have constraining rights? Would the world in which the one is killed to save the five be worse than the alternative in which he is not killed?
To help us consider these issues it will be useful to outline some of the possible relevant configurations of the world.
A. (1) In reality, there is no constraint; and
(2) We don't believe there is a constraint, and either
(a) his belief leads someone to kill the one, or
(b) his belief leads someone not to kill the one.
B. (1) In reality there is no constraint;
(2) We believe there is a constraint, and either
(a) his belief leads someone to kill the one, or
(b) his belief leads someone not to kill the one.
C. (1) In reality, there is a constraint; and
(2) We don't believe there is a constraint, and either
(a) his delusion leads someone to kill the one, or
(b) his delusion leads someone not to kill the one.
D. (1) In reality there is a constraint.
(2) We believe there is a constraint; and either
(a) his delusion leads someone to kill the one, or
(b) his delusion leads someone not to kill the one.
In some sense the issue with which we are concerned would be more easily resolved if we did not believe in moral reality antecedent to our beliefs, for then our beliefs about what rights people have might be thought to be necessary for making people have what rights they have. Our beliefs would be crucial for constructing moral reality, bootstrapping ourselves to status. Then its being better for there actually to be people who have constraining rights would imply that it would be better if we believed that people had constraining rights.
Assuming there is a moral reality independent of our views about ourselves, what can we say about these alternatives? How do A and B compare? Here the question is, assuming that it would be better if we really were the sorts of creatures who deserved to be protected by a constraint, though we are not such creatures, would it be better if we had a pleasant illusion or if we knew the truth? If it is better to have the illusion, is it better for someone to act on it and treat people better than they deserve to be treated? Is it better to treat them better than they deserve to be treated by way of the illusory belief in a good reality (via some fluke or persistent error) or to treat them appropriately by way of a true belief in a worse reality? In C, if we just consider (1) and (2), we see that there is some negative value attached to not knowing the truth, being deluded about ourselves, especially since the truth is about a better world. Suppose, however, that C(1) and C(2) are combined with option (b). If not believing truth about a good world led someone (somehow) to act in accord with the truth, this might be acceptable, and preferable to a world in which someone did not act in accord with what was, indeed, the truth, even though he knew the truth. (In this case, truth is assumed to be “don't kill the one to save the five.”) Therefore, whatever negative weight comes from having a false belief, especially one with a bad content, is tolerable if it (somehow) leads to or is compatible with the right acts.
The combination of C(1), C(2), and (a) is what would exist if we thought it permissible to kill when it wasn't in fact permissible, and someone did kill. Suppose it is true that the negative weight of having a false belief about a good truth does not outweigh the positive weight of doing the right act (of not killing the one). Could the negative weight of our all being deluded about the good truth, all by itself, be large enough to make a world in which the wrong act of killing the one is done, a worse world? We ask this question, because earlier it was suggested that one agent's being deluded and killing a person to save the five might not make for a worse world only because his behavior does not imply that there is general delusion about right and wrong. (An alternative interpretation is that it is not a worse world if he kills because, in fact, he shouldn't, that is, it is true people are inviolable and this makes the world better, not people's beliefs about it.) This implies that doing the wrong act by (p.282) itself is not making the world worse, since it minimizes rights violations. Would not a general delusion (even unaccompanied by action) make the world worse? In a world of general delusion, we lose two goods: (1) an elevated conception of ourselves, and (2) having our beliefs track the truth. If general delusion, as embodied in the moral system in which we believed, did not make the world worse, how could everyone's not endorsing the killing be what accounts for the world not being worse if the one kills to save the five rather than if he doesn't? But suppose it would be a worse world in which everyone, not just one person, were deluded about a good truth, and a killing were done because of the one's delusion. This does not mean that the presence of mass delusion is the only thing that makes a world in which one deluded person kills to save five a worse world than one in which no one is deluded and no one kills, nor does it mean that a world without mass delusion but a killing to save five is a better world than one with mass delusion and no such killing.
So we might backtrack and say that it is a worse world even when only the one person is deluded and he kills to save the five, let alone when all are deluded and all kill in this way. It is worse because events are not in accord with the truths of moral reality (i.e., with the reasons there are for doing things). Morality says the one person should not be killed to prevent the five killings. His being killed by the deluded person, not the person's being deluded, makes the world worse. This makes the worseness of the world simply a function of a wrong act taking place in it. Indeed, I think it is true that we would not be disappointed if the deluded agent suddenly lost his delusions and respected the constraint. The improvement might be due, in part, to the fact that a person had come to appreciate moral reality, but more to the fact that his acts were in accord with moral reality. This is buttressed by its being better if everyone through delusion denied a good truth, but they acted in accord with it; right action can speak louder than general correct beliefs. Also note that knowledge of a truth about a good world combined with action against it, [D(a)], can be worse than a false belief about a good world combined with action against the good truth, [C(a)]. There is an interaction effect; we do not give high marks for knowing a truth that is good when one can act against it. A misguided mind that leads to wrong acts is in some ways less disturbing than a clear‐seeing person whose acts go contrary to the truth.
“Futilitarianism” and Rights to Be Aided
I suggested that it is more likely to be formally rational to transgress a specified constraint in order to maximize things different from it, such as utility, than in order to minimize violations of comparable constraints. This may be true even when suffering a constraint transgression is more serious for someone than suffering the loss of utility alone, and even though we would do more to prevent constraints being violated than to prevent loss of utility if we did not have to violate a comparable right to do so. A basis for nonformal, (p.283) substantive rationality of this claim is that within a subset of values, tradeoffs between components of the subset may be prohibited, yet the subset may conflict with some other subset of values with which tradeoffs make more sense. Perhaps an analogy is that it could be better to violate a religious rule to save the life of a nonbeliever (that is, someone who does not participate in the religious subsystem) than to save the life of a believer. This is true even if the believer is considered, and actually is, more valuable than the nonbeliever.18 Without endorsing the permissibility of violating such rights for the sake of maximizing utility, I suggested that violating a specified constraining right for the sake of comparable rights is a form of “futilitarianism.” What of a nonspecified right? It will not be strictly irrational to transgress it for the sake of rights; we may have a case of permissible infringement. But if the strength of the right that each has is a measure of how inviolable the person is, then there may not be many permissible infringements. This might be even more true if we aimed to minimize rights violations than if we acted for other values.
Related to the “futilitarianism” point is another way of arguing for the conclusion that concern for constraining rights does not imply minimizing violations of them by transgressing comparable rights. The point here is that concern for rights per se, independent of their relation to utility, would never prompt such minimization.
For example, suppose people sometimes have a right to be aided by others. It is then necessary to explain why it is permissible to refuse to aid the first person we meet who has a strict right to our aid in order to aid the five people we see further down the road, each of whom also has such a right to our aid.19 This looks like minimizing rights violations. (Might we also abandon the person to whom we are obligated in order to help five who will suffer because someone else broke her commitment to them? I believe so.) If the single person has a claim to something (our aid), why is it permissible to deprive him of it in order to satisfy the five's similarly valid claims? How is this consistent with the possibility that neither I nor the five have any right to use what belongs to the single person (e.g., his device) to save others?
In search of an answer to this question, let us consider the Priest Case, in which a positive right is constraining against minimization: I argue that if we meet one priest, we should bow to him even if it means not rescuing five priests who will not be bowed to by others. By contrast, we said that we could ignore the first person who needed lifesaving aid in order to save those down the road. In the Priest Case, utility considerations are absent, that is, the five priests lose no welfare in not being bowed to. This suggests that, in the absence of utility considerations, one would never think that respect for rights alone called for violating one right in order to minimize disrespect for rights.20
By contrast, the cases in which each person has a claim to be given aid but the aider cannot help everyone and significant welfare for each is at stake, are like cases in which too much stock gets sold in the same thing. Each person was meant to have the thing in question. I believe that one way to handle these cases which accounts for the permissibility of ignoring the first needy person (p.284) in order to save the later five, is to pay attention to numbers.21 Each person with a claim must confront every other person with a claim. This means either multiple tosses of a coin or the balancing of one claim by a confrontation with an equal and opposite claim. In such confrontations we do not take away from someone his claim to something. Rather, he uses it up in a confrontation with his equal and opposite number. (In cases in which everyone buys a chance to get something, but it was never intended that everyone get to have the thing, there is no loss to the losers of that to which they had a claim, so the problem is somewhat different.)
Having a claim to something (e.g., my aid) that conflicts with others' claims to it, generates this required confrontation. When one has the only right to something (e.g., one's own body), there is no reason to enter into the confrontation procedure with others. This is what distinguishes the case of positive rights from those of negative rights, at least when the positive rights involve significant welfare. So it seems that when no utility is at stake, the positive right may be a constraint in the way a negative right is, and less often otherwise.
It has been said that the solution to the Selection Problem is victim‐focused rather than agent‐focused and that it is possible to understand a constraint as arising from concern for the manner restriction in a PPH right and what it expresses about persons, recognizing that to permit certain transgressions is to concede that a right and the status it seems to signify do not really exist. Let us consider how the treatment of positive rights shows the limits of these points.
It has been argued that we must attend in some way to the non‐derivative, non‐substitutability of individual victims. This non‐substitutability is said to be non‐derivative because it does not depend on the non‐substitutability of agents or agent's time slices. Still, in this victim‐focused view, the agent is called upon to attend to the victim primarily by himself not violating a non‐substitution requirement, rather than by preventing its violation by others even when preventing it would involve no transgression of rights by him. One way a rights‐as‐constraint view may justify this is to note that someone's status of inviolability is not affected by his being left to be violated. But not limiting someone's rights against an agent for help will affect the agent's status of inviolability, insofar as this involves his right to be (to a certain degree at least) free from others interfering with him for the sake of helping someone be better off than they would have been independently of his aid. This is another aspect of separateness of persons.
However, we noted previously that if inviolability is related to the idea of significance, there may have to be some balancing between the idea of inviolability and saveability which is concerned with promoting the ends of such significant beings. If this is so, does it imply that those who are ends‐in‐themselves may also, in part, be, on occasion, pure means‐in‐themselves: though their goals are in no way promoted by aiding, they must give aid?22 That is, is it this elevated status that itself makes possible and also requires the (p.285) role as pure means on occasion? If inviolability of a certain sort stems from someone's having a rational will, the duty to aid may come rather from the capacity for rational reflection, which gives rise to an objective view about the importance of others and the adoption of their goals as ours.
Any position on whether there is a duty to aid should be distinguished from a position that says that while it is impermissible for an agent to kill the single person to minimize violations of comparable rights, he may nevertheless allow others to behave in this way with the intention that they minimize rights violations. The latter is not permitted on the view I describe, since it involves intending that someone else kill (which may be tied to willing the loss of what the victim would have independently of imposition on us) and endorsing the killing as morally correct. However, it might be permissible to leave the one to be killed if this is a side effect of not wanting to make large efforts to aid, or the side effect of being busy doing something else significant. This is a matter of the moral relevance of (something like) the intention/foresight distinction.
How does all this connect with the issue of caring about rights? In the ordinary sense of caring about a right, a “caring analysis” should lead us to prevent by innocent means the rights violation of others. This conclusion is tempered by the fact that both requiring an agent not to kill and not requiring him to prevent others from killing (i.e., giving him a prerogative not to aid) seem susceptible to a similar interpretation. That is, suppose the conception of persons as separate is expressed by noninterference and by not enforcing association through duties to aid. Then, as suggested earlier, caring about this conception of the person will generate both rights not to be harmed and permissions not to aid in the moral system. There is no conflict between this and endorsing—caring that people have—strong negative rights, since letting die does not endorse the killability of the victim, even if it lets him be killed.
Furthermore, there is a minimization scenario that can be envisioned here as well: suppose more people will have their right not to give aid violated unless one person has his right not to give aid violated. Then it can be in the service of making possible the existence of a strengthened right not to aid and a strengthened conception of persons as separate, that morality does not permit minimization. That is, we do not oblige one person to aid, even in order to prevent others from being forced to aid. If we care about separateness, and its purest expression tells us not to violate the separateness of one to save the separateness of five, then we show concern for the ideal of separateness by acting as it says we should. As has already been noted, if prerogatives are related to the ideal of a person as an end‐in‐itself, having a point even if she does not serve the greater good when the greater good could be served, obliging someone to minimize occasions on which people do not get to exercise a prerogative defeats this ideal.
Notice that I am not concerned here with making the right not to aid absolute. The claim is only that the permission to transgress an absolute right or a specified right not to aid for the sake of preventing more violations of comparable rights would eliminate the right as part of morality. Neither the (p.286) one whose right we transgress nor the greater number we save will have this moral right. The permission to infringe a general right also reduces the status of individuals as separate, even if it does not eliminate the general right.
Summary and Conclusion
I conclude that it is not unreasonable to think that there is a connection between prerogatives, restrictions, and constraints, though the prerogative may have to be grounded differently than Scheffler thinks. That is, the prerogative may not derive simply from the significance things have from the personal point of view and our tendency to act from the personal point of view. Rather, it derives from some other ground which entitles one to control certain things. In particular, I have focused on the idea that one has a prerogative because a person is not a tool for the greater good, but is an end‐in‐himself for reasons other than his having a tendency to act from a personal point of view.
Scheffler says about his version of the prerogative that in order to introduce the idea of the personal point of view at a basic level in the moral system, it is introduced unmediated by maximization (or minimization). We are not required to minimize acts contrary to the personal point of view. Another way of saying this is that someone's personal point of view is permitted to block out consideration of the personal perspective of others in the aggregate, at least sometimes. On Scheffler's view, we make the personal point of view more important by permitting the person to try to act on it regardless of maximization or minimization. Yet, we have argued, to the extent that the status expressed by the prerogative would not exist if minimization of violations were permitted, we do “maximize” the presence of some impersonal value (this status) by having a prerogative.
Likewise, if we are not permitted to violate certain rights in order to minimize occasions in which comparable rights will be violated, to the extent to which this super right is based on the PPH, it introduces the idea of a highly inviolable person. This too is an aspect of the conception of the person as an end‐in‐itself. In both cases, we could say we “maximize” the presence of an inviolable person and a person who may try to exercise control over something, since these types of persons would not exist if it were permissible to minimize interference with self‐control or minimize rights violations by violating rights.
We can see that the problem of finding an asymmetry between potential victims in the case of rights (a problem thought not to exist for prerogatives) does not exist, once we understand the very idea of the constraining right. Each person is protected; so any person we would act against is protected. The five who are left to die do not have a claim to the use of the single person to save them; so if they can be saved only by using him, we do no wrong in not saving them. Nor does this behavior imply that their status is lower. The type of argument I have given for the existence of constraints on killing could be offered to account for other constraints. For example, it can be argued that permission to interfere with the free speech of one person for the sake of (p.287) minimizing violations of comparable free speech is inconsistent with the status of persons as bearers of the strong forms of free speech rights.23
We have distinguished between two questions: Can we give a justification for a constraint on violating rights to prevent other rights violations? and Can we give a justification for a restriction on maximizing utility? The second question deals with whether we can justify a restriction on killing one when we could save more lives from natural disaster.
It may seem that we should have answered this second question first, since we can most sensibly generate a right that we should not transgress in order to minimize rights violations by building on a restriction on maximizing utility or lives saved. Nevertheless, we began in this chapter by trying to answer the first question, and even suggested that we might, as formal matter, violate the constraint for the sake of utility. We can deal with the second question, essentially by reminding the reader of our previous discussion. The PPH is defended by showing why bringing about good by harming people in a certain way is wrong. We tried to begin such a defense in chapter 7. Objecting most seriously to the inappropriate relation that violating the PPH involves gives rise to a prohibition on bringing the relation about for any good, utility or maximal rights protection. This is a constraint, and it expresses a certain sort of inviolability of the person.
(1.) This way of putting the problem is not quite right, as we shall see in later discussion.
(2.) Notice that traditional Kantian universalization arguments claim that I would be deprived of value if it is true that others, who are like me, have no value. The argument I have presented claims that if it is true that the one person does not have a certain status, then others who are like him, do not have it either. Might a parallel analysis, speaking of the agent, be given for an agent's unwillingness to act in a certain way? What is meant by saying “I stand in a different relation to a victim or an act when they are mine, and therefore I won't kill to prevent others from killing” is neither “I care more about what is part of my life than about what is part of the lives of others,” nor “Each person must be morally responsible for his own conduct (or intentions) and not for that of others.” Rather, what is meant is that, because I am considering which will be my act and my victim, I will come up against the barrier‐like content of the act (comparable to the barrier‐like property of the victim). It being my act just tells me that I will be situated at the barrier rather than elsewhere. We have here a duty‐based, agent‐focused theory but not an agent‐concerned one, since it does not make the content of the constraint depend on seeing oneself in the picture or on magnification. But how is this read in a way which is not just basically agent relative? What agent‐neutral value is at stake? A certain type of agency is inappropriate. Recognizing this is inconsistent with endorsing engaging in the agency to minimize instances of this agency; if we allow others to engage in the agency, we do not thereby endorse such acts. Therefore, each has a derivative duty not to engage in this type of agency.
(4.) In Anarchy, State and Utopia.
(5.) Later, we shall examine whether we would in fact care about minimizing rights violations if utility were not attached to them at all. Then, the question arises whether, if utility plays some role in our decision to kill the one, killing him violates the PPH.
(6.) It would be a mistake to describe the constraint as telling us not to violate a right—rather than not to treat someone in a manner prohibited by the PPH—in order to prevent violations of rights. Still we can derive this (which we can call) constraint' from the constraint as follows: If someone should not be treated in the manner prohibited by the PPH for the sake of minimizing rights violations (this is the constraint), then they have a right this should not be done, and we must not violate this right in order to minimize rights violations (constraint').
(7.) The position developed here contrasts with that developed by Amartya Sen in “Rights and Agency,” Philosophy & Public Affairs 11 (Winter 1982: 3–39), where he argues that we might violate rights to minimize rights violations, but not to maximize utility, if it were not for the fact that from a position‐relative consequentialist point of view, the outcome is worse from an agent's position if he violates a right than if more rights are violated by others.
(8.) These views on infringing and compensation are owed to Judith Thomson and Joel Feinberg (whose exposure case we discuss on p. 268).
(9.) Of course, if it is all things considered wrong to kill the one to save the five from themselves being used to save others, it will still be all things considered a wrong for the five to be so used. So it will be all things considered right to permit to continue what is all things considered still a wrong. Is it a worse wrong? We shall discuss this later (pp. 276–78 and 279–82).
(11.) Those who argue that a constraint arises only from the quality of agency might give a somewhat different reason why the world will be worse if morality permitted killing one to minimize rights violations: It would be a better world in which there are more deaths and more rights violations and no person is permitted to be a killer, at least in this way. The five are, in effect, exchanging their lives to retain a certain conception of persons as agents. I do not find this agent‐focused explanation of value plausible.
(12.) Amy Guttman and especially Arthur Applebaum helped to make this clearer to me.
(14.) Virginia Held has suggested that if there are going to be some unavoidable rights violations no matter what we do, we should see to it that the comparable violations are distributed fairly, and on these grounds kill one to prevent more violations to the greater number. But this position fails to recognize that if we take rights seriously it may not be permissible to bring about what may in fact be the fairer state of affairs by violating someone's right. The best end‐state of affairs—from the point of view of fairness—cannot be brought about by violating the rights of one person, if to take the right, which is everyone's right, and the status it expresses, which is everyone's status, seriously involves its being constraining. In the same way, an agent may refuse to do an act that produces a fairer state of the world, or someone may have a prerogative not to make a big sacrifice to produce a fairer state of the world. However, if someone receives the benefits of an unfair state of the world, it may be permissible to take that benefit away to undo the unfairness.
(15.) In his discussion of my views in “The Values of Inviolability” Revue de Metaphysique et de Morale No. 2/1994.
(16.) At this point, I believe it is worth repeating some remarks made earlier, so as to contrast the account of constraints given by the so‐called expressive theory of rationality, as described by Elizabeth Anderson (in Value in Ethics and Economics, Cambridge, Mass.: Harvard University Press, 1993). See note 10, pp. 255–56.
(17.) This section was prompted by questions raised by Derek Parfit on the difference between belief in constraints and their objective reality.
(18.) Admittedly it is hard, substantively, to argue that we can transgress rights for something less, rather than more, important. It will often show respect for persons to just respecify the constraint so that it represents a lower degree of inviolability when minimization of rights violations is at stake.
(19.) Notice that this is a different case from one in which we are first involved in helping one person and must decide whether to abandon him to help five. It is also different from one in which we confront the one and five simultaneously. The latter case involves direction of our aid, which is different from breaking a tie of aid already established to one person. It also differs from doing nothing when one could do something to aid for the sake of helping others much later on. (On these other types of cases, see Morality, Mortality, Vol. 1. [New York, Oxford University Press, 1993].)
(20.) Still, this is consistent with utility loss triggered by rights violations being considered more serious than utility loss alone.
(22.) Is there a real moral difference between saying, as Kant does, that one has a duty to make the ends of other persons one's own ends, and saying that on occasion one has a duty to be a means to others' ends? Here are cases which highlight the difference: (1) I am opposed to A's goals but because I promised B, I let A use me. (2) I understand and approve of A's goals and hope he achieves them, and so lend him my services.
(23.) We might even see in a common argument against capital punishment a version of my argument for constraints: If the State kills a killer in order to prevent more killings by others, it endorses killing and this endorsement implies that people are less inviolable; if it does not prevent many killings the State nevertheless does not endorse them. (Of course, in this context the argument may be wrong, since guilty persons may not be as inviolable as innocent bystanders.)