1 For example, in my Morality, Mortality, Vol. I (New York: Oxford University Press, 1993). All references in the text to my “earlier work” will refer to this volume unless otherwise noted.
2 I have said that all candidates are to be imagined as equal in respects other than expected life years due to the resource. It is still possible that if they are equally young, the value to each candidate of the additional years will vary from what it would be if they were all equally old. At minimum, I only require one case in which the claim I have just described in the text holds—that is, the claim about the difference between the effect on interpersonal allocation of interpersonally aggregating (by giving five years to G) versus intrapersonally aggregating (by giving five years to B). Notice that even if we thought that B should be given a somewhat higher chance of getting the resource than A, this would still differentiate the effect of the extra five years in Case 1 from its effect in Case 2, where I think it could reasonably determine the outcome in which we help F and G.
3 Notice that the DALY does not seem to be a well-being measure but rather a record of a physical defect, whether or not compensation makes it have little impact on well-being. If this is true, it raises concern about the use of the measure. For example, suppose one society decides to spend x amount to compensate for a disability and another society spends the same amount to cure the disability. Well-being levels might be the same in the two societies, but one's DALY rating could be much higher than the other's. In such a case, should the DALY rating be at all relevant for deciding whom to help? I shall ignore the issue in this essay and assume (for the most part) that higher DALY ratings reflect differences in well-being or other forms of goodness in a life.
4 For example, in Peter Singer, John McKie, Helga Kuhse, and Jeff Richardson, “Double Jeopardy and the Use of QALYs in Health Care Allocation,” in Kuhse, Helga, ed., Unsanctifying Human Life: Essays on Ethics (Oxford: Blackwell, 2002); first published in the Journal of Medical Ethics 21 (1995): 144–50. All references to Singer are to this essay.
5 Singer grants that other things—including the long-term effects on society of such a policy—may mitigate against choosing a candidate who promises to have a better outcome. I shall return to this point below.
6 Singer et al., “Double Jeopardy,” 284–85.
7 I have modified the figures, but a discussion like this is present in ibid., 286.
9 Carlos Soto, “Choosing Whom to Aid,” unpublished paper.
10 He says this in his response to my “Faminine Ethics: Peter Singer's Ethical Theory,” in Jamieson, D., ed., Singer and His Critics (Oxford: Blackwell, 1999).
11 Singer, Peter, Practical Ethics, 2d ed. (New York: Cambridge University Press, 1993), 230–31.
12 I made these points in my “Faminine Ethics: Peter Singer's Ethical Theory,” 162–208.
13 Singer et al., “Double Jeopardy,” 286. I have already noted that even someone who would take need into account might object to counting differences in past quality of life in deciding on need, if the past has been adequate conscious life.
15 The impulse to compensate the disabled person would be especially strong, if quality of life were generally relevant to allocation decisions, when the disabled person would also be cured and live a normal life in the future. Yet Singer's forward-looking view also implies not giving preference to such a person in a contest with someone who had always led a normal life and would continue to do so. (But Carlos Soto raises other objections to “compensating” the nonrecently disabled in his “Choosing Whom to Aid.”)
16 Singer et al., “Double Jeopardy,” 292.
19 Material in this section is based on, but adds new arguments to, a section of my essay “Deciding Whom to Help, Health-Adjusted Life Years, and Disabilities,” in Anand, Sudhir, Peter, Fabienne, and Sen, Amartya, Public Health, Ethics, and Equity (New York: Oxford University Press, 2005).
20 See my “Disability, Discrimination, and the Principle of Irrelevant Goods,” unpublished paper, from which material in this section also draws.
21 For discussion of the moral difference that distributing rather than concentrating the benefit can make, see the discussion of Cases 1 and 2 above.
22 I first discussed cases of this sort in arguing against Thomas Scanlon's view that it is only how the treatment we have to give bears on the problem it is meant for in each candidate that is relevant to its allocation, in my essay “Owing, Justifying, and Rejecting,” Mind 111, no. 442 (2002): 323–54. Different issues about whether to give priority to helping the worse-off person would arise if the person without the hand had lived without it for a long time in the past as well (or even instead). I shall not discuss these different issues in this essay.
23 However, defending such a reason would require showing that it was consistent (a) not to hold someone's disability against him because it is irrelevant and (b) to hold his disability in his favor because it is a relevant burden. For if (a) is correct, why is it not also correct that the nondisabled person has an extra good that is irrelevant when it comes to choosing against him in the matter of preventing gastritis? I was prompted to think of the issue by Carlos Soto's work on the tension between compensating for past paraplegia but ignoring it in future outcomes in life-and-death cases. See his “Choosing Whom to Aid.”
24 Presumably, we are not interested merely in creating perfect specimens, as this would commit us to curing minor conditions in the almost well rather than making big differences to those who are very badly off though they will never be perfect.
25 I thank Susan Wolf, David Sussman, and other members of Philamore for discussion of this point. It is worth pointing out, as an analogy to someone's concern for his own life, a parent's concern for the life of his child. One child may have many more good traits than another. A parent could prefer that his child have better traits. He might even allow the child to take certain risks in order to become better for the child's own good. But that does not mean that a parent would make greater sacrifices to save the life of the child when he has better traits than when he has less-good traits. And this may be due to love for the actual child, rather than from duty.
26 Singer et al., “Double Jeopardy,” 286.
27 Though this view might be supported in other ways as well.
28 However, it is worth emphasizing a difference in the causal route in the intrapersonal and interpersonal cases. Intrapersonally, a person may give up, or risk giving up, something of his in order to get something better for himself. The exact parallel in the interpersonal context would be that someone gives up something, or risks giving it up, in order to cause someone else to get something. Here the “in order” involves a causal relation. Someone might object to a principle permitting the latter and yet distinguish it morally from what happens when we outright give a scarce resource to one person with the result that another person loses his life or loses a chance to keep it. In this case, the loss, or risk of it, for the second person does not cause the gain to the first person. Without denying the moral significance of this difference, what I am discussing throughout this essay is the possibility that one person could object to our concluding, on the basis of the reasonableness of his intrapersonal sacrifice of something, that it is permissible to outright give a scarce resource to another person when it results in the first person's losing, or risking the loss of, that same type of thing.
29 This claim may be in some tension with the view that the badness of death for someone is a function of the goods of which it deprives him.
30 I owe this point to David Sussman.
31 This argument implies that when .5 by itself is compared with 0, it is worth some maximal sacrifice x, and when 1 by itself is compared with 0, it too is worth x. This suggests that by the measure of sacrifice, and relative to 0, .5 equals 1. But that does not mean that .5 is equivalent to 1 per se on all measures. For example, we could be willing to give up .5 for 1 and also willing to risk going from .5 to 0 to get 1, but (obviously) not willing to risk going from 1 to 0 to get .5. This argument could even be taken to generate an intransitivity: that is, .5 equals x, and 1 equals x, but .5 does not equal 1. But the supposed intransitivity is explicable because of the effects of different contexts, where different alternatives are available: .5 is worth x in a context where 0 is the only alternative; 1 is always worth x; .5 is not worth as much as 1 when both are alternatives to 0 for a given person.
32 This issue is discussed by Parfit, Derek in his Reasons and Persons (New York: Oxford University Press, 1981), and by Chang, Ruth and others in Incommensurability, Incomparability, and Practical Reasoning, ed. Chang, Ruth (Cambridge, MA: Harvard University Press, 1997).
33 Ruth Chang has suggested (in conversation) that the separateness of persons serves as insulation in interpersonal allocation even when it is not reasonable for one person to care about his sufficiently good only option as much as another person cares about his better option. This is certainly true when we would have to take away from someone something that is his, in order to help someone else. I believe she is suggesting that it may be true even when we deny him help, as in the cases we are discussing. This would mean that the explanation of the fact that an additional good that matters in the intrapersonal context does not matter in an interpersonal context can be independent of the distinction between “caring for” and “caring about.” The possibility of such an alternative explanation was suggested in note 27 above.
34 The analogous contrast in the case of income is between the following two choice-sets: (1) One person can avoid extreme poverty by having $50,000 a year and another by having a million a year; (2) One person can avoid extreme poverty for five years and another for fifty years (whether at $50,000 a year or a million). The analogous conclusion would be that it could be reasonable to help the second person in (2), even if both should have an equal chance in (1).
35 How long someone will live is not a quality of a life such that someone might say that we are discriminating against those who will not live long (or have the property of short-livedness), at least when we are deciding so as to determine whether there will be a long or a short life. A context in which we would be discriminating against those who will not live long is when we refuse them admission to a park, for example, more than we refuse admission to those who will live a long time, even though both can make equally good use of the park.
36 This revises the view I presented in my Morality, Mortality, Vol. 1.
37 Hence, this implication departs from the view I defended in Morality, Mortality, Vol. 1, that in life-or-death situations, quality of life (above a certain minimal level) should not affect our decision about who gets a scarce lifesaving resource, given that each person wants the resource.
38 If this were not true, we might imagine that the drug prevents impending blindness and deafness, on the supposition that these together are worse than paraplegia.
39 I will here deal only with cases in which treatment also causes the side effect of a cure for a disability in a recently disabled person. A fuller range of cases would also include ones in which a treatment also causes the side effect of a disability in a previously nondisabled person. I discuss the fuller range of cases in “Deciding Whom to Help, Health Adjusted Life-Years, and Disabilities,” and in “Disability, Discrimination, and Irrelevant Goods.”
40 I am using paraplegia in discussing the Switch Cases. But it is possible that only if we could cure a worse disability (e.g., quadriplegia) would it be appropriate to respond differently to cases in which a better outcome occurs as a result of our curing a disability as opposed to our saving a person who is not disabled. I am concerned only with whether such causation ever matters morally, not so much when it does.
41 I discussed it in my “Deciding Whom to Help, Health-Adjusted Life Years, and Disabilities,” and in my “Disability, Discrimination, and Irrelevant Goods.”
42 We shall consider G and H from Case 10 again below. The Switch-and-Reduce Cases are derived from my initially considering a scenario in which a paralyzed person who will remain paralyzed and will live twenty years gets a lifesaving resource rather than someone who will remain unparalyzed but live for only five years because we abide by one of the principles discussed in Section IV.B. After getting the resource, the paralyzed person opts for a separate surgery that he knows will cure his paralysis but reduce his life expectancy to five years. Presumably, we would not want to rule out his possibly reasonable choice of the surgery simply on the ground that he is opting for an end state identical to the one that resulted in someone else's being deprived of an equal chance for the scarce resource relative to him. Yet the case seemed problematic to me.
43 Notice that the fact that piggyback U adds positive weight when the candidate will live five years cannot be accounted for by the U property's standing out in this particular context but not when a candidate can gain twenty years. For if a P candidate could also gain only five years, the Nonconstant Role Principle would recommend equal chances for a P candidate and a U candidate, with U adding no additional positive weight.
44 This was pointed out to me by Shelly Kagan.
45 Here is a way in which we should not explain our judgments about Case 11: we should not disaggregate interpersonally and turn the greater difference we make intrapersonally into a greater difference we make interpersonally. If we disaggregate, we claim that the two-person choice case is like the three-person choice case. That is, Case 11 is treated as though it were equivalent to the following case:
In this disaggregated case, we choose between saving the life of an unparalyzed person or saving the life of a paraplegic (who remains such) and also curing paraplegia in a third person who does not need his life saved. In this case, where the additional good is distributed over a third person, I believe the extra good done could permissibly determine our choice of whom to help. But if this conclusion carried over to Case 11, it would imply that we should save I (who is paralyzed) rather than J (who is not), and I think this is wrong. (We already saw in discussing Case 1 and Case 2 that turning a case of a concentrated aggregate of goods into a case of a distributed aggregate of goods need not lead to the same interpersonal allocation decision. In that set of cases, the outcome in the two-person case differed from what it was in the three-person case: A would live ten years and B fifteen years. In Case 11, the outcomes of I and J are the same.)
46 It is very important to repeat that I continue to imagine that the disabled person has (only recently) become disabled, and also that the nondisabled person has always been nondisabled. I do this in order to avoid the issue of one candidate's being needier than another because, if not aided, he will have lived a worse life, other things equal, in having lived for a long time as disabled. If we were to give some priority to helping the person who would be worse off if not aided, this might be a reason to hold someone's having lived a long time nondisabled against him in a choice between him and the disabled person (though not if only quantity- and not quality-of-life considerations should play a role in the evaluation of need in life-and-death choices). But I have constructed the cases so that this is not a factor.
47 I owe this point to Gideon Yaffe.
48 The objection was raised by Douglas MacLean and John Broome.
49 I owe this point to James Nelson.
50 I discuss how to manage apparent intransitivities that may result from using this procedure in my essay “Disability, Discrimination, and Irrelevant Goods.”
51 See Scanlon, Thomas, “Contractualism and Utilitarianism,” in Sen, Amartya and Williams, Bernard, eds., Utilitarianism and Beyond (Cambridge: Cambridge University Press, 1982). Scanlon discusses Harsanyi, John, “Cardinal Welfare, Individualistic Ethics, and Interpersonal Comparisons of Utility,” Journal of Political Economy 63, no. 4 (1955): 309–21. Singer's approach to the veil of ignorance seems like Harsanyi's. My discussion in this subsection of the essay is based on my comments on Gibbard, Allan's Reconciling Our Aims: The Tanner Lectures in Human Values, University of California, Berkeley, 2006 (New York: Oxford University Press, ).
52 See Scanlon, Thomas, What We Owe to Each Other (Cambridge, MA: Harvard University Press, 1999), chap. 5.
53 Scanlon follows Thomas Nagel, who argues for pairwise comparison as coming closest to being the correct way to combine people's different interests in an outcome. See Nagel, , “Equality,” in his Mortal Questions (Cambridge: Cambridge University Press, 1995).
54 I shall have more to say below about what sort of “chances to be in any of many positions” are relevant to focusing on (1).
55 Indeed, Scanlon believes that forcing people to identify in this way with each person is why Rawls uses a thick veil of ignorance (excluding probability calculations). See Scanlon's “Contractualism and Utilitarianism.” Singer, like Harsanyi, does not object to probability calculations, as can be seen in his discussion of the morality of slavery (Singer et al., “Double Jeopardy,” 291).