Causing People to Exist and Saving People’s Lives

Abstract

Most people are skeptical of the claim that the expectation that a person would have a life that would be well worth living provides a reason to cause that person to exist. In this essay I argue that to cause such a person to exist would be to confer a benefit of a noncomparative kind and that there is a moral reason to bestow benefits of this kind. But this conclusion raises many problems, among which is that it must be determined how the benefits conferred on people by causing them to exist weigh against comparable benefits conferred on existing people. In particular, might the reason to cause people to exist ever outweigh the reason to save the lives of existing people?

Possible People

Suppose that whatever one does a new person will come into existence. But one can determine who this person will be by either doing or not doing act A.

Possible People

$$ \begin{array}{*{20}c} {{\text{A:}}\,{\text{P}}_{ 1} \,{\text{will}}\,{\text{later}}\,{\text{exist}}\,{\text{and}}\,{\text{live}}\,{\text{to}}\, 80} \hfill & {{\text{P}}_{ 2} \,{\text{will}}\,{\text{never}}\,{\text{exist}}} \hfill \\ {{\text{Not}}\,{\text{A:}}\,{\text{P}}_{ 1} \,{\text{will}}\,{\text{never}}\,{\text{exist}}} \hfill & {{\text{P}}_{ 2} \,{\text{will}}\,{\text{later}}\,{\text{exist}}\,{\text{and}}\,{\text{live}}\,{\text{to}}\, 60} \hfill \\ \end{array} $$

Suppose that these two people would be highly and equally well off while they were alive, that there is no cost to doing A, and that the effects that the existence of one of these people would have on the lives of others would be neither better nor worse, overall, than the effects of the existence of the other. Because a life that is well worth living that lasts 80 years is better than an otherwise comparable life that ends after 60 years, it seems that the outcome in which P₁ comes into existence is better than that in which P₂ comes into existence. And if one outcome is better than another, there is a reason, from an impartial point of view, to prefer it and to bring it about if one can—though this reason might of course be outweighed by some other conflicting reason. In this case, however, there are no other relevant differences between doing A and not doing A. It seems, therefore, that in Possible People, one has an unopposed reason, and seemingly a moral reason, to do A.

The choice between doing A and not doing A is what Derek Parfit calls a “Same Number Choice” (Parfit 1984, p. 356).¹ Although different people would exist in the different outcomes of the choice, the same number of people would exist in each outcome. Parfit suggests that Same Number Choices should be evaluated by reference to the “The Same Number Quality Claim”:

If in either of two possible outcomes the same number of people would ever live, it would be worse if those who live are worse off, or have a lower quality of life, than those who would have lived. (Parfit 1984, p. 360)

This principle implies the intuitive view that it is better to do A than not. Although it is concerned solely with the comparative evaluation of outcomes, it is directly relevant to what we have reason to do, provided that we understand “good,” “bad,” “better,” and “worse” in what Parfit calls the reason-implying sense (Parfit 2011, p. 38).

To describe certain relevant features of this choice, it will be helpful to introduce both some assumptions and some further terms. Although this is controversial, I will assume that it cannot be better or worse for a person to be caused to exist. This is because “better for” and “worse for” are essentially comparative. The claim that to cause a person to exist would be better for that person thus entails that it would be worse for that person never to exist. But nothing can be worse, or bad, for someone who never exists. People who never exist cannot be victims of misfortune or the beneficiaries of good fortune. There is, however, a corresponding claim that does make sense. It is coherent, and plausible, to claim that to cause a person to exist is good for that person when the intrinsically good elements of the person’s life more than compensate for the intrinsically bad elements. Because the claim is only that this would be good for the person and not that it would be better for the person, there is no implication that it would be bad, or worse, for this person never to exist.²

Because causing a person to exist can be good or bad for that person, it also makes sense to apply the terms “benefit” and “harm” to instances of causing people to exist. If causing a person to exist is good for that person, it is convenient and perspicuous to say that causing that person to exist benefits him or her. Similarly, causing a person to exist harms that person if it is bad for him or her because the intrinsically bad aspects of the life are not compensated for by the good.

I will refer to benefits that are bestowed and harms that are inflicted by causing a person to exist as existential benefits and harms. While rather pretentious sounding, this label is preferable to “procreative” benefits or harms, as it better describes the results of acts that are a necessary part of the cause of a person’s existence but are not literally procreative. (To say that an act is a necessary part of the cause of a person’s existence is not to say that the person could not have existed in the absence of that act but only that he would not have existed.) Existential benefits and harms contrast with ordinary benefits and harms, which are bestowed or inflicted on existing or future people, whose existence is independent of the act that causes or constitutes the benefit or harm.

Existential benefits and harms are usually noncomparative, in the sense that they are good or bad, but not better or worse, for their subjects. But there are exceptions, as in the following example.

Same Possible Person

One must choose one of three options. One can

• (1) cause a person, P, to exist with a life that will last 80 years,

• (2) cause the same person to exist with a life that will end after 60 years, or

• (3) not cause anyone to exist.

Either option for causing P to exist would bestow an existential benefit on him, since in either case his life would, we may assume, be well worth living. But the first option would bestow a greater existential benefit on him than the second option would. While neither of these existential benefits would be better for him than never existing at all, the greater existential benefit would be better for him than the lesser existential benefit. Correlatively, the lesser existential benefit would be worse for him than the greater. Some existential benefits and harms are, therefore, not merely good or bad for people but also better or worse for them, at least in relation to other outcomes in which those same people would exist. Yet existential benefits that have this comparative dimension are relatively rare. This is because changes in the conditions that cause a person to exist that affect the character of the life that is thereby created also tend to result in the existence of a different person from the one who would have existed in the absence of those changes. This is the essence of what Parfit calls the Non-Identity Problem (Parfit 1984, p. 359).

In contrast, ordinary benefits and harms are usually comparative in the obvious sense that failing to bestow an ordinary benefit would be worse for the potential beneficiary and failing to inflict an ordinary harm would be better for the potential victim. But here too there may be exceptions. But to suggest an example, I must first note some distinctions among concepts of harm.

There are several different analyses of the concept of harm. First, an act harms a person in the counterfactual comparative sense if it is worse for that person than what would, or could, have happened had it not been done, or had it not been done and some other particular act been done instead. Second, an act harms someone in the temporal comparative sense if it makes that person worse off than he or she was before it was done. Finally, an act is intrinsically harmful if it causes a person to be in an objectively deprived or intrinsically bad state.³

Consider now an amputation of a person’s limb that is necessary to save that person’s life. This seems to be an ordinary harm in that there is a sense in which it is bad for the person. After the amputation, this person may seem to be objectively deprived, in that she now lacks a limb that was good for her. In one respect, therefore, she is also worse off now than she was before—though in another respect, of course, she is better off, as she now has the prospect of a longer life. Yet what thus seems to be an ordinary harm may seem not to be worse for this person in the dominant, counterfactual comparative sense, for the amputation is not worse for her than what would have happened had it not been done. It is, indeed, in this sense a benefit. It is therefore tempting to conclude that the amputation is an ordinary harm only in some intrinsic sense.⁴ Yet it may be possible for the counterfactual comparative account to explain in what sense it is a harm. It is a harm in that it is worse than what would have happened to the person if the life-threatening condition had not arisen.

It may also seem that there can be ordinary benefits, either in the temporal comparative sense or in some intrinsic sense, that are not better in the counterfactual comparative sense for the person on whom they are bestowed because they prevent or exclude a benefit that would be greater in any of these senses. These are effects that seem to be either intrinsic benefits or benefits in the temporal comparative sense, but harms in the counterfactual comparative sense. But, again, it may be possible for the counterfactual comparative account to explain the sense in which these effects are benefits—namely, they would be better for the person than what would have happened if there had been no prospect of the greater benefit. (Even if the counterfactual comparative account cannot adequately explain all ordinary benefits and harms, it is nevertheless true that all ordinary benefits and harms have a comparative dimension that noncomparative existential benefits and harms lack. For example, even if there is no sense in which an ordinary benefit that excludes a greater benefit is better for the person who receives it, it is worse for that person. Similarly, an ordinary benefit that excludes an equivalent benefit is equally good for the recipient, even if not better.)

Even if the counterfactual comparative account can provide an adequate analysis of all ordinary benefits and harms, including the harm (or, in some cases, benefit) of death, a full account of benefit and harm will have to recognize both comparative and noncomparative benefits and harms, since most existential benefits and harms are essentially noncomparative. If, however, there are some ordinary benefits or harms that are essentially intrinsic, and cannot be explained in counterfactual comparative terms, then a full account of benefit and harm must recognize that while most ordinary benefits and harms, as well as some existential benefits and harms, are comparative, there are nevertheless some ordinary benefits and harms, and many existential benefits and harms, that are essentially noncomparative. Thus, whether or not all ordinary benefits and harms are comparative, we must accept that there are both comparative and noncomparative benefits and harms.

Assuming that an act that causes a person to exist, or is a necessary part of the cause of the person’s existence, can benefit or harm that person, there is then the further issue of the extent to which the person is benefited or harmed by the act. Some philosophers assume that all the good and bad constituents of the life are attributable to the act, so that the net benefit is the amount by which the good elements of the entire life outweigh the bad (see, for example, Harman 2009, p. 143). Others assume that only some of the good and bad elements of a life count as benefits and harms that are attributable to the act that caused the person to exist. Suppose, for example, that a doctor’s negligence in prescribing a risky fertility drug for my mother not only enabled her to conceive a child earlier than she otherwise would have been able to but also caused her child—me—to have a painful congenital condition. Some philosophers claim that while the harms I suffer from that condition are attributable to the doctor’s action, my stubbing my toe just now is not, even though I would not have stubbed my toe had the doctor not acted negligently, for in that case I would never have existed.⁵

Suppose that someone gives me $10 each day for a year. Each of these gifts is a benefit, but the sum of them—the total $3,650—is also describable as “a benefit.” Existential benefits can be referred to in the same way. Each discrete benefit within a person’s life that is attributable to that person’s having been caused to exist is an existential benefit. But so is the sum of those individual benefits. And if those philosophers are right who claim that all the benefits in a person’s life are attributable to acts necessary for that person’s existence, then the person’s life as a whole may be considered an existential benefit as well, provided that it is worth living.

Suppose that it can be correct to attribute each discrete benefit a person receives in life to each act—of which there are indefinitely many—that was a necessary part of the cause of that person’s existence (see the discussion in Parfit 1984, Section 25). In that case, the concepts of existential and ordinary benefits must be “act-relative.” Suppose that the proximate cause of my receiving a benefit is that a friend confers it on me. That benefit is an ordinary benefit relative to my friend’s act but an existential benefit relative to those acts that were necessary for my coming into existence.

Having explained certain assumptions and defined some relevant terms, I can now better explain my claim about the choice between doing A and not doing A. My claim that it is better to do A is an implication of the more general claim that it is better to bestow a greater existential benefit on one person than to allow a lesser existential benefit to go to a different person, if there are no other relevant differences between the options.

Possible People and Future People

Suppose next that, in Future or Possible, one can do either A or B but not both. If one does B, a person, P_F, who does not exist now but will later exist independently of what one does, will live to 80 rather than to 60. That is, doing B now will prevent this future person from dying when he is 60, thereby enabling him to live to 80. The possible outcomes of one’s choice are these:

$$ \begin{array}{*{20}c} {{\text{A:}}\,{\text{P}}_{ 1} \,{\text{will}}\,{\text{later}}\,{\text{exist}}} \hfill & {{\text{P}}_{ 2} \,{\text{will}}\,{\text{never}}\,{\text{exist}}} \hfill & {{\text{P}}_{\text{F}} \,{\text{will}}\,{\text{later}}\,{\text{exist}}} \hfill \\ {\quad \;{\text{and}}\,{\text{live}}\,{\text{to}}\, 80} \hfill & {} \hfill & {{\text{and}}\,{\text{live}}\,{\text{to}}\, 60} \hfill \\ {{\text{B:}}\,{\text{P}}_{ 1} \,{\text{will}}\,{\text{never}}\,{\text{exist}}} \hfill & {{\text{P}}_{ 2} \,{\text{will}}\,{\text{later}}\,{\text{exist}}} \hfill & {{\text{P}}_{\text{F}} \,{\text{will}}\,{\text{later}}\,{\text{exist}}} \hfill \\ {} \hfill & {{\text{and}}\,{\text{live}}\,{\text{to}}\, 60} \hfill & {{\text{and}}\,{\text{live}}\,{\text{to}}\, 80} \hfill \\ \end{array} $$

Doing A would bestow a greater existential benefit on one person rather than allowing a lesser existential benefit to go to another person. Doing B would bestow an ordinary benefit comparable in magnitude to the difference between the two existential benefits.

If one does A, that will make it the case that a person will exist and live to 80 rather than that a person will exist and live to 60. The same will be true if one does B. The difference is that if one does A, the person who will live to 80 (P₁) will be a different person from the person who would have lived to be 60 (P₂), whereas if one does B, the person who will live to 80 and the person who would have lived to 60 will be the same person. Because of this, the failure to do B will be worse for that person. But the failure to do A will not be worse for the person who will exist and live to 60. Indeed, the failure to do A will not be worse for anyone who ever lives.

According to Parfit’s No-Difference View, this difference between A and B is insignificant (1984: 366–369).⁶ None of the people whose existence or longevity would be determined by the choice between A and B now exists. Whether one does A or B, two new people will exist. If one does A, a person will later exist who will live to be 80 and another will exist who will live to be 60. The same will be true if one does B. It makes no difference, according to Parfit, that doing A would be worse for one person and better for no one, while doing B would be better for one person and worse for no one.

Described that way, however, the difference seems to matter. But Parfit presents an example, the Medical Programs Case, to show that intuitively it does not matter (Parfit 1984, pp. 366–369). In that example, one of two medical programs must be canceled. If Preconception Testing is canceled, 1,000 disabled children will later exist rather than 1,000 different children who would not have been disabled. If Pregnancy Testing is canceled, 1,000 future children will be born disabled whose disabilities could have been prevented through prenatal treatment. Cancellation of either program, therefore, will result in the future existence of 1,000 disabled children rather than 1,000 children without a disability. Assume that the effects on other people would be the same in either case. Many people share Parfit’s intuition that, if there are no other relevant differences, it does not matter which program is canceled. This intuition might be supported by the fact that, if only 999 instances of disability could be prevented by Pregnancy Testing while 1,000 could be prevented by Preconception Testing, it seems that it would be better to cancel Pregnancy Testing, even though there would be 999 people for whom that would be worse.

Future or Possible is modeled on the Medical Programs Case. The only two possibly significant differences are that in the Medical Programs Case the number of people in each outcome is greater and the relevant harm is disability rather than reduced longevity. (It might be thought significant that Pregnancy Testing involves medical treatment of fetuses while doing B extends the life of an adult. But the benefit of Pregnancy Testing spans an entire life and it could be stipulated that B operates causally while the beneficiary is a fetus, so that the increased longevity is, like the prevention of disability, a consequence of what is done to a fetus.) It seems, therefore, that those who agree with Parfit that it makes no difference which program is canceled should also accept that it makes no difference in Future or Possible whether one does A or B. When there is literally no difference evaluatively between two acts or outcomes, those acts or outcomes are evaluatively equivalent. According to the No-Difference View, therefore, doing A and doing B are equally good. This judgment seems intuitively plausible.

Future People and Existing People

Suppose that, in Future or Existing, one can do either B or C but not both. If one does C, an existing person, P_E, will live to 80 rather than to 60. The possible outcomes of one’s choice are these:

$$ \begin{array}{*{20}c} {{\text{B:}}\,{\text{P}}_{\text{F}} \,{\text{will}}\,{\text{later}}\,{\text{exist}}\,{\text{and}}\,{\text{live}}\,{\text{to}}\, 80} \hfill & {{\text{P}}_{\text{E}} \,{\text{will}}\,{\text{live}}\,{\text{to}}\, 60} \hfill \\ {{\text{C:}}\,{\text{P}}_{\text{F}} \,{\text{will}}\,{\text{later}}\,{\text{exist}}\,{\text{and}}\,{\text{live}}\,{\text{to}}\, 60} \hfill & {{\text{P}}_{\text{E}} \,{\text{will}}\,{\text{live}}\,{\text{to}}\, 80} \hfill \\ \end{array} $$

In this case, one can either prevent a future person from dying at 60 or prevent an existing person from dying at 60. In order for the options to be as similar as possible, assume that the only difference in the relation that one bears to the two people is that P_E exists at the time one must act. One has no more knowledge of P_E than one has of P_F, since one has no knowledge of either apart from the fact that one exists now while the other does not but will exist later. Thus, at the time at which one must choose whether to do B or C, P_E might be a late-term fetus or infant, or he or she might be 60 and close to death. If one knew at the time at which one had to make the choice that P_E was then on the verge of death, that might make it seem more important to do C.⁷ But I would like to exclude that possible reason for thinking that it would be better to do C rather than B. The difference I want to highlight is simply that between a presently existing person and a person who does not now exist but will in the future. Hence the stipulation that all that one knows is that if one does C, that will cause P_E to live to be 80 rather than die at 60, while if one does B, that will cause P_F to live to be 80 rather than die at 60. This is of course consistent with the earlier stipulation that the act that extends a person’s life from 60 to 80 years could operate causally at any point in the person’s life. While most ways of extending a person’s life count as saving the person’s life, I will in general refer to “extending” rather than “saving” both to avoid any possible positive associations that the term “saving” might have and to allow for the possibility that there are ways of extending a person’s life that do not count as saving it.

Future or Existing is what Parfit calls a “Same People Choice”: all and only the same people exist in both outcomes. It is a choice between bestowing an ordinary benefit on one person and bestowing an equal ordinary benefit on another person. The only question is whether it matters that one of these people exists at the time one must choose while the other will exist only later.

Many people find it intuitively plausible to suppose that doing B and doing C are equally good. Many people also believe that it is permissible to do either. To suppose that it matters more to extend P_E’s life than to extend the life of P_F by the same amount would be relevantly similar to supposing that it matters more to extend the life of a person who is physically proximate than to do the same for a person who is farther away, when this difference in spatial proximity is the only difference in the way one is related to them. As Parfit observes, “remoteness in time has, in itself, no more significance than remoteness in space” (Parfit 1984, p. 357).

Existing People and Possible People

Suppose that, in Existing or Possible, one can do either A or C but not both. The possible outcomes of one’s choice are these:

$$ \begin{array}{*{20}c} {{\text{A:}}\,{\text{P}}_{ 1} \,{\text{will}}\,{\text{later}}\,{\text{exist}}} \hfill & {{\text{P}}_{ 2} \,{\text{will}}\,{\text{never}}\,{\text{exist}}} \hfill & {{\text{P}}_{\text{E}} \,{\text{will}}\,{\text{live}}\,{\text{to}}\, 60} \hfill \\ {\quad \;{\text{and}}\,{\text{live}}\,{\text{to}}\, 80} \hfill & {} \hfill & {} \hfill \\ {{\text{C:}}\,{\text{P}}_{ 1} \,{\text{will}}\,{\text{never}}\,{\text{exist}}} \hfill & {{\text{P}}_{ 2} \,{\text{will}}\,{\text{later}}\,{\text{exist}}} \hfill & {{\text{P}}_{\text{E}} \,{\text{will}}\,{\text{live}}\,{\text{to}}\, 80} \hfill \\ {} \hfill & {{\text{and}}\,{\text{live}}\,{\text{to}}\, 60} \hfill & {} \hfill \\ \end{array} $$

This is a Same Number Choice, since whether one chooses to do A or C, the same number of people will ever exist. Indeed, even if there is the third option of doing neither A nor C, one still faces a Same Number Choice provided that in exercising that third option one does not cause a further individual to exist. Furthermore, when applied to the choice between A and C, the No-Difference View implies that it is morally irrelevant that there is someone for whom the failure to do C would be worse while there is no one for whom the failure to do A would be worse.

But I suspect that in this choice, many people would find this difference to be intuitively morally significant. They would think that it is better, in effect, to extend the life of an existing person than to ensure that a person who will have a longer life comes into existence rather than a person who would have a shorter life. And they would think that, given the choice, one ought to extend the life of the existing person—that is, that it would be wrong to allow an existing person to die when he could live an additional 20 years, in order instead to do what would cause a longer-lived person to come into existence rather than a different, shorter-lived person. This is my own untutored intuitive response. Many people, I suspect, would have the same view even if the example were altered so that doing C would enable P_E to live only to 70 rather than 80.

But now there is a problem. I have suggested that doing B is equally good as doing A and that doing C is equally good as doing B. This implies that doing A is equally good as doing C; for doing C is equally good as doing B, which is equally good as doing A. But it seems that doing C is better than doing A. Or, starting from the other end, if doing C is better than doing A and doing B is equally good as doing C, it seems that doing B should be better than doing A. But it seems that it is not. In that choice, the No-Difference View seems intuitively plausible. But these inferences challenge the No-Difference View if we find it difficult to accept the conclusion that to cause P₁ to exist rather than to cause or allow P₂ to exist is equally good as enabling P_E to live to 80 rather than dying at 60.

One might object that these inferences cannot be drawn from the discrete pairwise comparisons I have made—doing A with doing B, doing B with doing C, and so on—since comparisons between acts may be affected by which other acts are possible. The relevant question is whether the problem cited in the previous paragraph arises when all three acts are possible, as follows.

$$ \begin{array}{*{20}c} {{\text{A:}}\,{\text{P}}_{ 1} \,{\text{will}}\,{\text{exist}}} \hfill & {{\text{P}}_{ 2} \,{\text{will}}\,{\text{never}}\,{\text{exist}}} \hfill & {{\text{P}}_{\text{F}} \,{\text{will}}\,{\text{later}}} \hfill & {{\text{P}}_{\text{E}} \,{\text{will}}\,{\text{live}}\,{\text{to}}\, 60} \hfill \\ {\quad \;{\text{and}}\,{\text{live}}\,{\text{to}}\, 80} \hfill & {} \hfill & {{\text{live to 6}}0} \hfill & {} \hfill \\ {{\text{B:}}\,{\text{P}}_{ 1} \,{\text{will}}\,{\text{never}}\,{\text{exist}}} \hfill & {{\text{P}}_{ 2} \,{\text{will}}\,{\text{exist}}} \hfill & {{\text{P}}_{\text{F}} \,{\text{will}}\,{\text{later}}} \hfill & {{\text{P}}_{\text{E}} \,{\text{will}}\,{\text{live}}\,{\text{to}}\, 60} \hfill \\ {} \hfill & {{\text{and}}\,{\text{live}}\,{\text{to}}\, 60} \hfill & {{\text{live}}\,{\text{to}}\, 80} \hfill & {} \hfill \\ {{\text{C:}}\,{\text{P}}_{ 1} \,{\text{will}}\,{\text{never}}\,{\text{exist}}} \hfill & {{\text{P}}_{ 2} \,{\text{will}}\,{\text{exist}}} \hfill & {{\text{P}}_{\text{F}} \,{\text{will}}\,{\text{later}}} \hfill & {{\text{P}}_{\text{E}} \,{\text{will}}\,{\text{live}}\,{\text{to}}\, 80} \hfill \\ {} \hfill & {{\text{and}}\,{\text{live}}\,{\text{to}}\, 60} \hfill & {{\text{live}}\,{\text{to}}\, 60} \hfill & {} \hfill \\ \end{array} $$

It seems that the same reasoning applies even when it is possible to do any one of the three acts. It seems intuitively plausible that doing A and doing B are equally good, that doing B and doing C are equally good, but that doing C is better than doing A. So the problem I identified remains.

Although I began with the claim that A is better than not-A, this claim is unnecessary for the challenge to the No-Difference View. Yet it does support that challenge. For if A is not better than not-A, that suggests that A is not equally good as B and hence that even if B is equally good as C, A is not equally good as C. So the challenge to the No-Difference View would be weakened if there were reason to believe that it is not better to do A than not to do A.

Many people believe that potential parents are not required to choose, from among the possible children they might have, the one with the best expected life.⁸ Some believe that there is not even a defeasible, pro tanto moral reason to choose to have the child with the best expected life. And virtually everyone would accept that it is unobjectionable to breed an animal rather than have a child, thereby causing a less well-off individual to exist rather than a substantially better-off individual (see McMahan 1998, pp. 233–234). One might suppose that these facts suggest that many or most people would deny that it is always better, other things being equal, to cause a better-off individual to exist than to cause or allow a less well-off individual to exist. If so, perhaps it is not better to do A than not to do A.

But many of those who claim that there is not always a requirement, or even a reason, to have a child with a better life rather than a different child with a less good life would not reject the claim that there is a reason to do A rather than not to do A. What they deny is a certain perfectionist or elitist view according to which it would be objectionable, for example, to have a child with an IQ of 140 when it would be possible instead to have a different child with an IQ of 145. Many people who deny that there is a requirement, or even a reason, to have the child with the higher IQ would accept that there is a reason to do A rather than not to do A, as the difference between P₁’s life and P₂’s life is substantial: 20 years of life that would be worth living.

One might argue that merely evaluative claims do not entail normative claims, so that it is possible for one outcome, such as the existence of P₁, to be better from an impartial point of view than another, such as the existence of P₂, without there being any reason to prefer it or to cause it to occur instead of the less good outcome. But I do not see how this could be; hence my use of “good” and its comparative cognates in the reason-implying sense. I also do not see how the existence of P₁ could fail to be better than the existence of P₂, if the only morally relevant difference between the existence of one and the existence of the other is that P₁ would have 20 more years of life that would be worth living. It therefore seems to me that the explanation of why people are not in general morally required to have children is that the reason to bestow noncomparative benefits by causing people to exist is normally outweighed by the reasons that people may have to do other things that do not involve having children. But this suggests that the reason to bestow noncomparative benefits rather than not to bestow such benefits is relatively weak—weaker, perhaps, than the reason to bestow greater rather than lesser noncomparative benefits. I will return to this later.

While the claim that it is better to do A than not to do A is important, it is not, as I noted, part of the simple argument I have given for the conclusion that doing A is equally good as doing C. Because the choice between doing A and doing C is a Same Number Choice, the claim that doing A is equally good as doing C is directly implied both by Parfit’s Same Number Quality Claim and by his No-Difference View, assuming that the latter applies to comparisons involving existing people as well as to those involving future people. If we intuitively find that doing A is not as good as doing C, then that challenges both of Parfit’s claims. If we considered only Existing or Possible, intuition might prompt us to reject both these claims. But I have given a simple argument, which appeals to other intuitions, for the claim that in Existing or Possible, doing A is equally good as doing C. I have argued that this claim is implied by the plausible claim that doing A is equally good as doing B and that doing B is equally good as doing C. It seems, therefore, that if we are to reject the claim that doing A is equally good as doing C, we must also deny either that A is equally good as B or that B is equally good as C.

It seems highly implausible to deny the latter. For the only difference between doing B and doing C is that the beneficiary of doing C exists at the time when one must choose between the two acts, while the beneficiary of doing B does not. It might be claimed that coexistence in time, even for a very brief period, is a special relation that provides grounds for legitimate partiality toward P_E. But that stretches the notion of a special relation to the point of vacuity. And, in any case, the distinction between an existing person and a future person is in this instance a matter of whether that person exists at the time at which one must choose between doing B and doing C. That leaves it open that one may later overlap in time with P_F, in which case one could anticipate being “specially” related through coexistence to P_F as well as to P_E.

This is not to say that our intuitions are not influenced by the difference between existing people and future people. It is just to say that, on reflection, we are reluctant to endorse the view that existing people matter more than future people, just as we are reluctant to endorse the idea that physical proximity could matter morally, even though it may affect our emotional reactions on some occasions.⁹

Is There an Asymmetry Between Comparative and Noncomparative Benefits?

The other option is to deny that, in Future or Possible, doing A is as good as doing B. Recall that Future or Possible is just the Medical Programs Case writ small and that it is the intuition that Parfit hopes to elicit from the latter that supports the No-Difference View. So to deny that doing A is as good as doing B is tantamount to affirming that there is a relevant difference between the two medical programs and thus is also tantamount to rejecting the No-Difference View.

One might base the denial that doing A is as good as doing B on the claim that existential benefits matter less than ordinary benefits—that is, that an existential benefit contributes less to making the outcome better than an equivalent ordinary benefit and that there is a correspondingly weaker reason to bestow the existential benefit than there is to bestow the equivalent ordinary benefit. For the additional benefit that would be bestowed by doing A is an existential benefit while the benefit that would be bestowed by doing B is an ordinary benefit. Some philosophers in fact claim that existential benefits do not matter at all—that their occurrence does not make the outcome better and that there is no moral reason to bestow them. (Some indeed, deny that there can be existential benefits.¹⁰) There is intuitive support for this claim in the common belief that the expectation that a person would have a life worth living does not by itself provide a moral reason to cause that person to exist. (This belief constitutes one half of the view sometimes referred to as the Asymmetry. The other half is that the expectation that a person would have a life in which the intrinsically bad elements outweigh the good does provide a moral reason not to cause that person to exist.) Yet the idea that existential benefits do not matter at all seems to have the counterintuitive implication that it is generally bad, in practice, to cause people to exist, for example by having children. For if we think, as most people do, that existential harms do matter, and if causing people to exist always involves the infliction of existential harms, then it seems that for it to be permissible to cause a person to exist, that act must also have good effects that outweigh the existential harms. But if existential benefits do not matter at all and so cannot counterbalance or compensate for the existential harms, it may seem that whether it can be permissible to cause a person to exist will depend on whether doing so would produce ordinary benefits for existing and future people sufficient to outweigh the existential harms, which they might do only rarely. I will return to this problem below.

The idea that benefits matter less if they are existential is challenged by Same Possible Person. The benefit of the longer life for P is an existential benefit, but it seems a mistake to discount it. For it seems that to cause P to exist with the life that will last 80 years (henceforth “cause P-80 to exist”) rather than to cause P to exist with the life that will last for only 60 years (henceforth “cause P-60 to exist”) matters just as much as enabling a future person to live for 80 years rather than 60, which in turn matters just as much as enabling an existing person to live for 80 rather than 60 years.

There is, however, a question about what the relevant comparison is here. What exactly is the analogue in the case of an existing person of causing P-80 to exist? Or the analogue of causing P-60 to exist? If causing P-80 to exist is, as I suggested, like enabling an existing person to live to 80 rather than 60, then causing P-60 to exist should be like failing to enable an existing person to live to 80 rather than 60. This assumes that the default in the case of the existing person—that is, what would happen with no intervention—is living to 60. But if one causes P-60 to exist, one has caused him to live to 60 rather than causing him to live to 80. But causing an existing person to live to 60 rather than to 80 could be an instance of killing, or an act that is relevantly like killing. If the appropriate comparison with causing P-60 to exist rather than causing P-80 to exist is killing an existing person at 60, then it seems plausible to suppose that the existential benefits in Same Possible Person must be discounted. For causing P-60 to exist rather than P-80 does seem less bad than killing an existing person at 60.

It seems, however, that neither failing to enable an existing person to live to 80 rather than 60 nor killing an existing person at 60 is precisely analogous to causing P-60 to exist rather than P-80. This is because the “defaults,” or what would happen in the absence of intervention, are different in the three cases. The implied default in failing to enable a person to live to 80 rather than 60 is that the person will live to 60. And the implied default in causing a person to live to 60 rather than 80 is that the person will live to 80. What the default is determines the nature of one’s agency—for example, whether one causes or allows a person to die. And the nature of the agency affects the morality of the action. But in Same Possible Person, there is no single implied default. There are, for example, two possible alternatives to causing P-60 to exist: causing P-80 to exist and not causing P to exist at all. Of these, the latter is the natural default, for it best captures the idea of one’s doing nothing, or not intervening at all. Because there is no option of allowing it to be the case that an existing person never exists, there can be no unique parallel in the case of an existing person to causing P-60 to exist or to causing P-80 to exist. It therefore seems impossible to test for the intuitive plausibility of discounting existential benefits in Same Possible Person by comparing the options in that case with parallel options involving an existing person.

Suppose, however, that one causes P-60 to exist. Putting aside issues of agency, the absence of the additional 20 years of good life that P could have had—that is, the absence of that additional existential benefit—seems to matter just as much as the absence in an existing person’s life of an ordinary benefit of an additional 20 years that was possible for him. This may be because in this case the existential benefit, unlike most existential benefits but like most ordinary benefits, has a comparative dimension. P’s getting the 60-year life is worse for him than getting the 80-year life would have been. Perhaps the relevant difference between types of benefit is not between existential benefits and ordinary benefits but, as some people have thought, between benefits with a comparative dimension and those that are essentially noncomparative.¹¹ Perhaps the appeal to an asymmetry between comparative and noncomparative benefits is what is needed to avoid the conclusion in Existing or Possible that to extend the life of an existing person, P_E, by 20 years is not better than to cause P₁ to exist rather than allowing P₂ to exist instead. For the benefit to P_E is comparative, while the benefit to P₁ is noncomparative.

Still, as I conceded earlier, the No-Difference View seems plausible in Future or Possible. In that case it seems to make no difference that not doing B would be worse for someone, while not doing A would not be worse for anyone. And that suggests that the distinction between comparative and noncomparative benefits has no moral significance. But the No-Difference View’s initial intuitive appeal in that case might be countered by considering the thoughts that might be legitimately available to the people in the different outcomes of this choice (McMahan 1998, p. 243). Suppose that one does A rather than B, thereby causing P₁ to exist and live to 80 rather than allowing P₂ to exist and live to 60, but also failing to enable a future person, P_F, to live to 80 rather than 60. P_F can correctly think that “the choice of A rather than B was worse for me, for I will now die 20 years earlier than I would have if B had been done instead.” And P₁ can think that “it was good for me that A was done because I will live for 80 years, whereas if B had been done I would never have existed at all.” Suppose that doing B rather than A would not have been worse for the agent or for any other existing or future person. In that case the future person is entitled to the thought that, if B had been done, that would not have been worse for anyone but would have enabled him to live an additional 20 years. That is a bitter thought. Note, however, that the thought I cited as available to P₁ does not specially favor doing A. For P₁ can have the same thought about the indefinitely many acts to which, because of the Non-Identity Problem, he owes his existence.

Next consider the thoughts that the people can legitimately have if B is done rather than A. The future person can correctly think that “the choice of B rather than A was better for me because I will now live to 80 rather than dying at 60.” And P₂ can think that “it was good for me that B was done because I will live for 60 years, whereas if A had been done I would never have existed at all.” The choice of B is good or better for everyone. No one is entitled to bitter thoughts.

These are striking differences between the outcome of doing A and the outcome of doing B. They suggest that there may well be an evaluative or moral asymmetry between comparative and noncomparative benefits. Such an asymmetry might be quite strong. It might be, for example, that noncomparative benefits do not matter at all—that they do not make the outcome better and that there is no moral reason to bestow them. This would support the view to which I referred above as the Asymmetry. But it is incompatible with the intuition that in Possible People there is a reason to cause P₁ to exist rather than to allow P₂ to exist. Alternatively, it might be that the value of noncomparative benefits is discounted to some extent in relation to the value of equivalent comparative benefits. Even a comparatively slight asymmetry would undermine the simple argument I presented earlier for the intuitively problematic conclusion that in Existing or Possible, doing A is equally good as doing C (that is, causing P₁ to exist in place of P₂ is equally good as extending P_E’s life by 20 years). For if there is an asymmetry between comparative and noncomparative benefits, that would imply that in Future or Possible, doing B would be better than doing A. This is because the benefit of an additional 20 years for P_F would be a comparative benefit, while the benefit to P₁, which would be greater by 20 years than the benefit to P₂, would be a noncomparative benefit. Thus, if doing B is better than doing A, and doing B and doing C are equally good, it is possible that in Existing or Possible, doing C is better than doing A—that is, that it is better to extend the life of P_E.

It is worth emphasizing the precise nature of the suggested asymmetry. The benefits that would be discounted are those that altogether lack a comparative dimension. They therefore do not include those ordinary benefits to which I referred earlier that are not better for those who receive them—namely, those benefits that exclude a greater a benefit. Such ordinary benefits do have a comparative dimension, as they are worse for those who receive them than receiving a greater benefit would be. The only benefits that are fully noncomparative are existential benefits. As Same Possible Person shows, not all existential benefits are noncomparative in the sense of lacking any comparative dimension. Yet, as I observed earlier, existential benefits that have a comparative dimension are relatively rare because of the Non-Identity Problem. So the suggestion is that existential benefits that lack a comparative dimension, which include the great majority of existential benefits, are to be discounted, while all ordinary benefits and those few existential benefits with a comparative dimension are not.

Here is a further test of our intuitions about the significance of comparative and noncomparative benefits. In Same Possible Person, both the benefits one might bestow are existential benefits but both have a comparative dimension. Consider next a parallel choice in which the benefits one might bestow are existential benefits of the same magnitude as those in Same Possible Person but without a comparative dimension. Recall the first case we considered, Possible People, in which P₂, who will live to 60, will come into existence unless one causes P₁, who will live to 80, to exist instead. We can alter this case so that the default—that is, what will happen if one does not act—is not that P₂ will exist but that no one will come into existence. This yields:

Different Possible People

One must choose one of three options. One can

• (1) cause P₁ to exist,

• (2) cause P₂ to exist, or

• (3) not cause anyone to exist.

This case is like Same Possible Person except that the person with the longer life and the person with the shorter life are different people rather than the same person who lives longer in one outcome than in the other. Because in each case there is the option of not causing anyone to exist, these are both Different Number Choices.

Suppose that in this case one causes P₂ to exist. And suppose that in Same Possible Person one causes P-60 to exist. Both these choices seem worse than the alternative in which a person would have existed and lived to 80. But are they equally bad? In Different Possible People, one bestows a lesser noncomparative benefit. In Same Possible Person one bestows an equivalently lesser comparative benefit. If there is an asymmetry between comparative and noncomparative benefits, the bestowal of the lesser benefit should matter less in Different Possible People. And this may seem intuitively plausible if we consider the thoughts available to the people in the outcomes of the two choices. In Same Possible Person, P-60 can reflect that causing him to exist with the 60-year life was worse for him than causing him to exist with the 80-year life would have been. But in Different Possible People, the corresponding thought to which P₂ is entitled is only that causing him to exist was good for him. These facts provide some intuitive support for the idea that the failure to bestow the greater noncomparative benefit is less bad than the failure to bestow the equivalently greater comparative benefit.

If one finds the idea that there is an asymmetry between comparative and noncomparative benefits plausible, one may wonder whether there is a corresponding asymmetry between comparative and noncomparative harms. That seems unlikely. While many might find it acceptable to suppose that noncomparative benefits do not matter at all, it is wholly implausible to suppose that noncomparative harms do not matter at all. If that were true, it would not be bad to cause a person to exist whose life would contain nothing but agony. This leaves it open whether there might be a lesser asymmetry between comparative and noncomparative harms. But if there is, it is unlikely to be as significant as that which might plausibly be supposed to exist between comparative and noncomparative benefits. That is, if noncomparative harms are discounted at all, they cannot plausibly be discounted to the extent that one might find acceptable in the case of noncomparative benefits. I will return to this point below.

That the asymmetry I have suggested is only between comparative and noncomparative benefits, and not between comparative and noncomparative harms, shows that the suggestion is not equivalent to the claim that existing and future people matter more than possible people. Some philosophers have sought to defend the view that while our choices must be constrained by respect for the rights and well-being of existing and future people, the possible well-being of possible people—those whose existence depends on the outcome of our choice—does not matter at all. Others have argued that, while the well-being of possible people has some weight, it matters less than the well-being of existing and future people. The suggestion I have made, in contrast, is about benefits only, not about the distinction between actual and possible people more generally. It is therefore compatible with the view that harms caused by causing possible people to exist matter as much, or nearly as much, as equivalent harms caused to existing and future people. And it holds that some benefits bestowed by causing possible people to exist—namely, those with a comparative dimension, such as those that might be bestowed in Same Possible Person—can matter, at least in some ways, as much as equivalent benefits bestowed on existing or future people.

Reason-Giving Weight and Canceling Weight

One may wonder, however, about the implications for the morality of ordinary procreation of the claim that noncomparative benefits have less weight than equivalent comparative benefits. If noncomparative benefits do not matter, or have no weight at all, then the fact that causing a person to exist would benefit her provides no reason at all to cause her to exist. That there is no such reason is probably the common sense view. Yet I have suggested that noncomparative benefits do matter—for example, in Possible People, when one can choose whether to cause P₁ to exist or to allow P₂ to exist instead. But the further suggestion that they have less weight than equivalent comparative benefits seems to imply that, in general, the reason to bestow a benefit by causing a person to exist is weaker than the reason to bestow an equivalent comparative benefit on an existing or future person. This implication seems highly plausible.

Yet there are two different ways in which benefits caused by causing people to exist may be relevant to the morality of procreation. In the preceding paragraph, I was discussing what might be called their reason-giving weight—that is, whether and to what extent they provide a moral reason to cause an individual to exist.¹² If, for example, noncomparative benefits have half the reason-giving weight of equivalent comparative benefits, there would be a greater reason, other things being equal, to benefit an existing person by extending his life than there would be to bestow a benefit that would be almost but not quite twice as great by causing a person to exist.

There is, however, another, different way in which noncomparative benefits may matter. They may weigh against, and cancel out, corresponding noncomparative harms. We know that in causing people to exist, we will inevitably cause them to suffer certain noncomparative harms. These inevitable harms count against causing people to exist. We can imagine cases in which these harms would be so bad that it would be impermissible to cause the person who would suffer them to exist. Yet almost everyone accepts that it is generally permissible to cause people to exist, for example, by having children. But this is not because most people believe that the noncomparative harms inflicted on the children by causing them to exist are outweighed by the benefits to others of the children’s existence. It is instead because they believe that the noncomparative benefits of being caused to exist nearly always outweigh the noncomparative harms, in the same way, or much the same way, that the comparative benefits that a person derives from having his life saved usually outweigh the comparative harms that he will have to endure as a result of being saved. We can call the weight that noncomparative benefits have in canceling the noncomparative harms suffered by the same person their canceling weight.

Reason-giving weight and canceling weight are quite distinct and independent ways in which benefits might matter (or indeed ways in which harms might matter, as the same distinction can be drawn among harms, including noncomparative harms). It is possible that noncomparative benefits may have no reason-giving weight but full canceling weight, or discounted reason-giving weight but full canceling weight, or heavily discounted reason-giving weight and slightly discounted canceling weight, and so on. There are many possibilities, some of which correspond to positions that have been defended in the literature. If we thought, for example, that noncomparative benefits have no reason-giving weight but full canceling weight, that would support the common sense view that while it is almost always permissible to have a child, the only positive moral reasons to have a child derive from the interests of people who already exist or will exist independently of whether one has the child.

Common sense views of procreation presuppose that even if noncomparative benefits have discounted reason-giving weight, or no reason-giving weight at all, they nevertheless have full, or nearly full, canceling weight—that is, a noncomparative benefit cancels a noncomparative harm of equivalent or comparable magnitude. Yet the suggestion that noncomparative benefits matter less than equivalent comparative benefits can be interpreted as applying to their reason-giving weight, their canceling weight, or both. Some philosophers have argued for “critical level” views of the morality of causing people to exist. They claim that if a person’s life would be below some threshold on the scale that measures positive well-being, it would be bad to cause that person to exist even though her life would be worth living—that is, even though the benefits in the life would outweigh the harms (see, for example, Blackorby et al. 1997; Kamm 1992, pp. 132–133). These people’s views could be supported by the claim that both the reason-giving weight and the canceling weight of noncomparative benefits must be discounted, perhaps to the same degree, in relation to the weight of equivalent comparative benefits. But to me it seems more plausible to restrict the discounting to the reason-giving weight of noncomparative benefits. A more precise statement of my suggestion is thus that the reason-giving weight of noncomparative benefits is less than that of equivalent comparative benefits.

One philosopher who claims that noncomparative benefits have not only no reason-giving weight but also no canceling weight is David Benatar. According to Benatar, the absence of a benefit “is not bad unless there is somebody for whom this absence is a deprivation” (Benatar 2006, p. 30). He further assumes that there is no reason to bestow a benefit if its absence would not be bad. From this he then infers that “the positive features of life [that is, the benefits], although good for those who exist, cannot justify the negative features [the harms] that accompany them. Their absence would not have been a deprivation for one who never came into existence” (Benatar 2006, p. 6). This is his way of stating that noncomparative existential benefits do not weigh against or cancel out corresponding noncomparative existential harms to the same person. On this view, the noncomparative harms inflicted by causing people to exist provide a reason not to cause people to exist, while the noncomparative benefits bestowed by causing people to exist not only provide no reason to cause people to exist but also fail to weigh against and cancel out corresponding harms. Noncomparative harms can be canceled and outweighed only by comparative benefits to existing and perhaps future people. On Benatar’s view, therefore, if a person’s life would contain any harms at all, it would be bad, and perhaps impermissible, to cause that person to exist, no matter how much benefit her life would contain, unless the harms were morally outweighed by benefits to others. It is, moreover, open to Benatar to appeal here to the common view that the infliction of a harm on one person can be morally outweighed only by the conferral of substantially greater benefits on others.

Benatar’s view can be interpreted in such a way that it does not support his conclusion. Consider a person who has been caused to exist—yourself, for example. It seems that the noncomparative benefits attributable to your having been caused to exist do have canceling weight because their absence would have been a deprivation to you in this sense: that it would have been worse for you to exist with the harms in your life but without the benefits. This is clearly not what he considers to be the relevant comparison. Yet this is one of the ways in which the relevant benefits might not have occurred. It therefore seems legitimate to compare the occurrence of the benefits with this way of their not occurring to determine whether their absence would be a deprivation.

I will not, however, pursue this line of objection. For when understood in this way, these existential benefits have a comparative dimension, like the existential benefits in Same Possible Person. Existential benefits are noncomparative only when the alternative to their bestowal is that the potential beneficiary never exists. Suppose that is true in the cases with which Benatar is concerned. In that case, his claim is that those benefits have no canceling weight. In determining whether to cause people to exist, the only effects on those people that matter are the harms they would suffer. Since harms are inevitable in life, Benatar concludes that if we take into account only the effects on the people who would be caused to exist, it is always bad to cause or, within relevant moral constraints, to allow people to come into existence.

This conclusion, which Benatar accepts, is sufficiently counterintuitive that few people have been willing to accept his argument. But his position has a further implication that I doubt that even he can really accept. Suppose that one can provide effective birth control to only one of two couples. It is predictable that, in the absence of birth control, each couple will eventually have a child. Genetic testing has been done on the members of each couple and has revealed that the members of one couple have a rare combination of genes that will cause any child they have to have an extremely painful though relatively short life. If they have a child, that child will live for about 2 years in unremitting pain that, together with inevitable mental defects, will virtually exclude the possibility of its having any satisfying experiences. If, by contrast, the other couple have a child, there is a high probability that it would live for at least 80 years, during which the benefits in the life would greatly exceed the harms. That is, if this second couple have a child, it will live a long and happy life—an unusually good life by current standards. Yet the total harm that the second child would suffer over more than 80 years would be slightly greater than that which the first child would suffer in just 2 years. According to Benatar’s argument, in neither case would the benefits in the life weigh against the harms. Because the harms in the longer life would exceed those in the shorter, one ought to give the birth control to the couple whose child would live the longer life, assuming for the sake of argument that the effects on other people would be comparable in each case (perhaps because each couple lives in a remote and isolated area). In effect, Benatar’s view implies that, if other things are equal, one ought to do what would result in the existence of a human being whose life would contain almost nothing but suffering rather than what would result in the existence of a person who would have a long life at an overall high level of well-being.

In my judgment, we must conclude from this that Benatar is mistaken to think that noncomparative benefits have no canceling weight. It seems, moreover, that some of the cases we have considered show that noncomparative benefits also have reason-giving weight. Consider again Different Possible People, which is like Possible People except that the default option is that no new person will come into existence rather than that P₂ will come into existence. If noncomparative benefits have full canceling weight, or even just significant canceling weight, it would be good for P₁ to exist and good for P₂ to exist and thus it would be permissible to cause either to exist, considered on his own. For in each case the noncomparative benefits would cancel the noncomparative harms, thereby leaving a significant net benefit to each person. Yet it seems that if one has decided to cause someone to exist, it ought to be P₁ rather than P₂, other things being equal. One has reason, in other words, to cause P₁ to exist rather than P₂. This is because of the additional noncomparative benefits that P₁’s life would contain that P₂’s would not. But if noncomparative benefits had canceling weight only, those additional benefits would be irrelevant. There are no further harms for them to cancel, so they should make no difference to the choice between P₁ and P₂. But they do make a difference: They provide a reason for causing P₁ to exist rather than P₂. They therefore have reason-giving weight, even if that weight is discounted relative to the reason-giving weight of equivalent comparative benefits (or equivalent noncomparative harms).

Greater or Lesser, Some or None

It may seem, however, that noncomparative benefits have reason-giving weight only in choices between a greater benefit and a lesser benefit. In Possible People, for example, there is a reason to cause P₁ to exist rather than to allow P₂ to come into existence instead. But in Different Possible People, it may seem that there is a reason to cause P₁ to exist only if one decides to cause someone to exist, so that the alternative has become to cause P₂ to exist. Most people do seem to think that noncomparative benefits have reason-giving weight only in this rather peculiar way—namely, that they provide a reason to cause a person to exist when the alternative is to cause or allow someone else to exist who would be worse off, but not when the alternative is neither to cause nor to allow any new person to exist. Thus, although most people seem to believe that in Different Possible People there is a reason to cause P₁ to exist rather than to cause P ₂ to exist, it seems that they also think that there is no reason to cause P₁ to exist rather than not to cause anyone to exist. It seems, that is, that most people accept that noncomparative benefits have reason-giving weight only conditionally—only if someone is going to come into existence, either because one cannot prevent that from happening (as in Possible People) or because one has decided to cause someone to exist (as in Different Possible People).

Similarly, most people seem to believe that in Same Possible Person, there is no reason to cause P to exist but that, if one decides to cause him to exist, one ought, other things being equal, to cause him to exist with the life that will last 80 years rather than the life that will last only 60 years. In this case, the benefit of causing him to exist with either life is, if one may put it this way, comparative in relation to causing him to exist with the other life, but noncomparative in relation to not causing him to exist at all.

But these beliefs are puzzling. Suppose that, as the common view suggests, there is initially no reason in Different Possible People to cause either P₁ or P₂ to exist. But if one decides, arbitrarily, or on a whim, to cause someone to exist, the common view implies that that decision somehow creates a reason to cause P₁ to exist. The noncomparative benefit to P₁ does not matter when the alternative is that no noncomparative benefit will be bestowed. But that benefit to P₁ does matter if an arbitrary decision makes it the case that the alternative is the bestowal of a lesser noncomparative benefit on a different person. But how can a decision made simply on a whim create a reason that did not exist antecedently to cause P₁ to exist? More generally, how can noncomparative benefits have reason-giving weight in choices between greater and lesser, but not in choices between some and none? (Compare McMahan 1998, Section IV.) If there is no reason to bestow a noncomparative benefit on its own, how can there be a reason to bestow it rather than a lesser benefit?

These cases in which noncomparative benefits seem to have reason-giving weight when the alternative is to bestow a lesser noncomparative benefit but not when the alternative is to bestow no benefit at all are quite unlike a superficially similar case of Parfit’s in which one can save one of a person’s arms at significant cost to oneself, do nothing, or save both that person’s arms at the same cost to oneself (Parfit 1982, p. 131). Parfit claims, plausibly, that it can be permissible to save neither arm, but that if one decides to help this person, one must save both arms rather than only one. As in Different Possible People, the requirement to bestow the greater benefit is conditional on a decision to confer some benefit. But despite the fact that it is permissible to bestow neither benefit in Parfit’s case, this is not because there is no reason to bestow either rather than neither. There is a strong reason to bestow each benefit rather than no benefit. It is just that that reason can be overridden by the cost to the agent.

Jonathan Glover also once observed that a policy of picking the best apples does not entail that one has a reason to pick as many as possible (Glover 1977, p. 69). That is clearly true, since the reason for picking apples is instrumental and there is a limit to the instrumental value of apples. What he could not have said is that having a reason to pick the best apples does not entail that one has a reason to pick apples. If one had no reason to pick apples, one would have no reason to pick better rather than worse apples.

The common assumption that noncomparative benefits have reason-giving weight in choices between greater and lesser but not in choices between some and none may imply an absurd conclusion. Consider two choices, one between a greater and a lesser noncomparative benefit, the other between a noncomparative benefit equivalent to the greater benefit in the first choice and no such benefit at all. The first of these choices can simply be the original Possible People:

$$ \begin{array}{*{20}c} {{\text{A:}}\,{\text{P}}_{ 1} \,{\text{will}}\,{\text{later}}\,{\text{exist}}\,{\text{and}}\,{\text{live}}\,{\text{to}}\, 80} \hfill & {{\text{P}}_{ 2} \,{\text{will}}\,{\text{never}}\,{\text{exist}}} \hfill \\ {{\text{Not}}\,{\text{A:}}\,{\text{P}}_{ 1} \,{\text{will}}\,{\text{never}}\,{\text{exist}}} \hfill & {{\text{P}}_{ 2} \,{\text{will}}\,{\text{later}}\,{\text{exist}}\,{\text{and}}\,{\text{live}}\,{\text{to}}\, 60} \hfill \\ \end{array} $$

The second choice can be called Possible Person:

$$ \begin{array}{*{20}c} {{\text{D:}}\,{\text{P}}_{ 3} \,{\text{will}}\,{\text{exist}}\,{\text{and}}\,{\text{live}}\,{\text{to}}\, 80} \hfill \\ {{\text{Not D:}}\,{\text{P}}_{ 3} \,{\text{will}}\,{\text{never}}\,{\text{exist}}} \hfill \\ \end{array} $$

Suppose that Possible Person is parallel to Possible People in that there is no cost to doing D and that the effects of doing D on existing and future people would be neither better nor worse overall than the effects of not doing D. Suppose further that P₃ would enjoy the same high quality of life that P₁ would enjoy. Their lives would, indeed, be equivalent in all morally significant respects. The only difference is that they would be different persons. Suppose, moreover that these two choices—in Possible People and Possible Person—are made in isolation. They are not, in other words, mutually exclusive: doing D does not exclude doing A, and vice versa. The two choices are instead wholly unrelated. One might imagine them as choices faced by different agents at different times. Call the comparison between these two choices the A–D Comparison.

In Possible People, it is better to do A than not to do A—that is, it is better to cause P₁ to exist than to allow P₂ to exist instead. Most people, it seems, accept this. In contrast, most people deny that in Possible Person it is better to do D than not to do D. They deny that there is a moral reason to cause P₃ to exist rather than not to cause anyone to exist. (This denial is one of the two claims that together constitute the Asymmetry.) Thus, in describing what he takes to be the common view of the value of human life, Ronald Dworkin comments that it is not regarded as the sort of value that there is a reason to create, or to create more of. “It is not important,” he writes, “that there be more people. But once a human life has begun, it is very important that it flourish and not be wasted”—that is, that it continue (Dworkin 1993, p. 74).

Yet the actual consequence of doing A and the actual consequence of doing D are equivalent in all morally significant respects. There is no relevant difference between the existence of P₁ and the existence of P₃. But if doing A and doing D result in morally equivalent states of the world, and not doing A is worse than doing A, it seems to follow that not doing A must also be worse than doing D. Most people believe, however, that not doing D is not worse than doing D. But if not doing D is not worse than doing D, yet not doing A is worse than doing D, it seems to follow that not doing A is worse than not doing D—that is, that allowing P₂ to exist is worse than causing no one to exist. But that is false. It is not worse, if other things are equal, to allow a person to exist and live for 60 years with a high quality of life than not to cause any person to exist at all.

If this argument is correct, it seems that we should accept either that it is not better to do A than not to do A, or that it is better to do D than not to do D. For if, for example, it is better to do D than not to do D, the A–D Comparison does not then imply the absurd claim that it is worse to allow P₂ to exist than not to cause anyone to exist. For if doing D is equivalent in its results to doing A, and if doing A is better than not doing A, so that doing D is better than not doing A, and if doing D is better than not doing D, then nothing at all follows about the relation between not doing D and not doing A. Both are worse than doing either A or doing D, but nothing follows about whether one is worse than the other.

Of the two options, it seems more plausible to accept that doing D is better than not doing D. For it is plausible to suppose that noncomparative benefits have some reason-giving weight in all contexts, even if they do not have the same reason-giving weight as equivalent comparative benefits. That this is plausible is supported by common sense intuitions about the prospect of human extinction. Most of us believe that the extinction of the human species would be the worst of those possible tragedies that have more than a negligible probability of actually occurring. The badness of extinction can of course be explained in part by the effects on the members of the final generation, and in part by the retroactive diminution of the meaning or significance of acts done by members of previous generations in the expectation that they were contributing to efforts that would continue indefinitely into the future. But there seems intuitively to be much more to the badness of extinction than this. As long as extinction can be deferred, human life, and posthuman life, can continue indefinitely, with unimaginable numbers of people enjoying the goods of life, which might in time become vastly superior to the goods accessible to human beings thus far, just as the goods accessible to us are vastly superior to those that were accessible to our remote evolutionary ancestors. To most of us, it is appalling to think that instead of this incalculable number of people enjoying these incalculable benefits, there might instead be only the emptiness of a world devoid of consciousness.¹³

To accept that noncomparative benefits have reason-giving weight even in choices between some and none—that is, to accept that there is a reason to cause people to exist if their lives would be worth living—is not to accept any particular view about the weight that noncomparative benefits have in relation to equivalent comparative benefits. Even so, to acknowledge that they have any reason-giving weight at all is to raise certain difficult problems, as I will indicate in Section 9.

Objections to the Argument Based on the A–D Comparison

It is therefore worth considering what might be said against the argument I have given on the basis of the A–D Comparison. One obvious objection to that argument is that my claim that doing A and doing D are equally good ignores an essential dimension of the evaluation of the two acts. For I appealed only to the actual outcomes of those acts—the existence of P₁ in the case of doing A and the existence of P₃ in the case of doing D. But this fails to consider what would happen in each case if the act were not done. And that is clearly relevant to how good it is that each act is done. Suppose that if A were not done, a million people would die, whereas not doing D would have no effect on whether anyone would die. If A and D are both done, the only way in which the world differs after each act from the way if was before is that it contains one more person. But clearly it does not follow that in these circumstances, doing A and doing D are equally good.

As we have seen, it seems that most people accept that doing A is better than not doing A, but deny that doing D is better than not doing D. These evaluations suggest that most people accept that it is better to do A than to do D, considering each choice in isolation from the other and taking into account what would happen if each act were not done. But is this common view actually defensible? I have noted that the actual consequences of doing A and of doing D are relevantly equivalent. So if it is better or more important to do A than to do D, that must be because of what would have happened if each had not been done. That is, doing A must be better than doing D because the alternative to doing A would be worse than the alternative to doing D. But to accept that is to directly embrace the absurd conclusion I drew from the A–D Comparison—namely, that allowing P₂ to exist is worse than not causing anyone to exist. This first objection to the argument based on the A–D Comparison therefore fails.

A second objection is that there is a fundamental difference between Possible People and Possible Person that prevents us from drawing any conclusions from the comparison between them, at least until further work is done in moral theory. This difference is that Possible People is a Same Number Choice while Possible Person is a Different Number Choice. According to Parfit, conclusions we might draw about Same Number Choices cannot be straightforwardly extrapolated to Different Number Choices. He argues in Reasons and Persons that we must find a principle, Theory X, that implies the Same Number Quality Claim and is compatible with the No-Difference View but also applies to Different Number Choices (Parfit 1984, p. 361). Yet he has thus far been unable to find Theory X and no one else has succeeded where he has failed.

One reason why Different Number Choices are so much more problematic than Same Number Choices is obvious. When the number of people in the different possible outcomes of a choice is the same, the standard measures of how well off the people in the different outcomes are coincide. For example, an increase in total well-being is also a corresponding increase in average well-being. Thus, when the Same Number Quality Claim, which applies only in Same Number Choices, states that an outcome in which people are worse off is worse, there is no uncertainty about which of the possible outcomes is the one in which people are worse off, even though there are different people in the different outcomes. (It is perhaps worth noting that certain other principles could conflict with the Same Number Quality Claim. Distributional principles, such as principles of equality or priority, could imply that an outcome that is worse according to the Same Number Quality Claim is actually better.) Yet when there are differences in the number of people in the different possible outcomes, standard measures of collective well-being may diverge. It may no longer be obvious or uncontroversial which group is worse off. For this reason, principles such as the Same Number Quality Claim, which seem highly plausible when applied to Same Number Choices, such as Possible People, cease to have any clear application in Different Number Choices. Furthermore, outcomes with different numbers of people tend to be less precisely comparable than outcomes with the same number of people.

Yet it seems that these ought not to be significant problems in a simple Different Number Choice such as Possible Person, in which all relevant factors are the same in the two possible outcomes except that in one there is an additional person who will have a life that is well worth living and will last for 80 years. If we accept that Possible People shows that noncomparative benefits matter and have reason-giving weight, then the presumption must be that they have reason-giving weight in Possible Person as well. The burden of justification thus seems to lie with the common sense view. An explanation is required of why the additional noncomparative benefits to P₁ provide a reason to cause him to exist in place of P₂ while the possibility of bestowing equivalent noncomparative benefits on P₃ provides no reason to cause him to exist.

We can, of course, imagine cases in which there would be a reason to cause P₁ to exist but might not be a reason to cause P₃ to exist. If, for example, everyone in the world had the same level of well-being that P₁ and P₃ would have, it might be objectionable on grounds of equality to allow P₂ to come into existence. Hence there could be a reason to prevent that by causing P₁ to exist instead, even if there were no reason to cause P₃ to exist. This would support our intuition in Possible People without implying that it would be better to cause P₃ to exist in Possible Person. Yet while this would support our intuitive belief, it would not explain it in a plausible way. For the suggestion here is that the reason to do A is not that P₁ would benefit more from existence than P₂ would, but that the existence of P₂ would be bad, for reasons of equality. Furthermore, reasons of equality do not provide a general justification for doing A rather than not doing A. If, for example, everyone in the world had the level of well-being that P₂ would have, reasons of equality might imply that it would be better not to do A, and indeed not to do D either.

It seems, therefore, that the fact that Possible People is a Same Number Choice while Possible Person is a Different Number Choice provides no obvious reason to deny that whatever explains why in most circumstances it is better to do A than not to do A also justifies and explains the claim that in most circumstances it is better to do D than not to do D. In particular, it provides no reason to suppose that in Possible Person, which is an extremely simple Different Number Choice, the outcome in which P₃ exists is not the one in which people are better off. It certainly provides no reason to think that this outcome is the one in which people are worse off.

There is, however, one further type of objection to the argument I developed on the basis of the A–D Comparison that seems quite powerful. This type of objection consists of challenges to the precision of the relations that I have claimed hold between the various acts. For example, one of the premises in the argument I gave is that doing D is not better than not doing D, hence not doing D is not worse than doing D. As Parfit points out, “not worse than” is sometimes an imprecise relation.¹⁴ If the most that can be claimed is that not doing D is imprecisely not worse than doing D, it does not follow that not doing D is at least equally good as doing D. In that case, the fact that not doing D is not worse than D, which is better than not doing A, may not entail that not doing D is better than not doing A.

A further objection of this type is that there is no warrant for the precision in my earlier claim that doing A is equally good as doing D. The most that is warranted, one might argue, is the imprecise claim that neither doing A nor doing D is better or worse than doing the other. If that is correct, then the claim that not doing A is worse than doing A does not entail that not doing A is worse than doing D. For it seems possible, because of the imprecision in the comparisons, that not doing A is worse than doing A, that doing D is neither better nor worse than doing A, and yet that not doing A is not worse than not doing D. In short, if there is no basis for the claim that not doing A is worse than doing D, then even if not doing D is not worse than doing D, it does not follow that not doing A (that is, allowing P₂ to exist) is worse than not doing D (not causing anyone to exist).

I am uncertain whether there is imprecision in these relations that are asserted to obtain in the premises of the argument based on the A–D Comparison. But even if there is, the challenge remains to explain how it could be that noncomparative benefits have reason-giving weight in choices between greater and lesser but not in choices between some and none. Given that noncomparative benefits clearly do have reason-giving weight in choices between greater and lesser (such as Possible People), the presumption seems to be that they also have reason-giving weight in choices between some and none—that is, that there is a reason bestow noncomparative benefits by causing people to exist even when the alternative is not to cause anyone to exist. That presumption stands unless one can find a good reason to believe that noncomparative benefits have reason-giving weight in choices between greater and lesser but not in choices between some and none.

Choices Between Causing People to Exist and Saving People’s Lives

It may help to review the overall argument to this point. If, as virtually everyone believes, it is permissible to have a child in the absence of some unusual reason not to, then noncomparative benefits seem to have full canceling weight—that is, they weigh against and cancel corresponding noncomparative harms in the same way that comparative benefits weigh against and cancel comparative harms. Many people, of course, believe that there is a moral asymmetry between comparative benefits and comparative harms, in that for it to be permissible to harm a person without her consent in the process of bestowing a benefit on her, the benefit must significantly outweigh the harm (that is, the benefit must be good for her by significantly more than the harm would be bad for her). Those who accept this view about comparative benefits and harms may accept a similar view about noncomparative benefits and harms.

Possible People supports the view that noncomparative benefits have reason-giving weight as well as canceling weight. When we initially consider the A–D Comparison, however, our intuitions may suggest that while noncomparative benefits have reason-giving weight in choices, such as that in Possible People, between bestowing a greater benefit and bestowing a lesser benefit, they do not have reason-giving weight in choices between bestowing a great noncomparative benefit and not bestowing any such benefit at all—that is, in choices between causing a person to exist and not causing anyone to exist. Although I developed an argument to show that that view is indefensible, it may be that the argument fails because of the imprecision of the comparative evaluations that constitute its premises. Yet even if that particular argument fails, it nevertheless seems arbitrary to suppose that noncomparative benefits have reason-giving weight in choices between greater and lesser but not in choices between some and none.

Assuming that noncomparative benefits have reason-giving weight in at least some choices, do they have the same weight as equivalent comparative benefits? The No-Difference View asserts that they do, though Parfit’s discussion of that view is confined to Same Number Choices, which are choices between greater and lesser. I suggested, however, on the basis of Existing or Possible, that it is intuitively plausible to suppose that noncomparative benefits have less reason-giving weight than equivalent comparative benefits.

Our intuitive reluctance to accept that noncomparative benefits have reason-giving weight in choices between some and none suggests a further possible view, which is worth noting. According to this view, noncomparative benefits have discounted reason-giving weight in choices between greater and lesser. Hence in Existing or Possible, it is better to do C than to do A—that is, it is better to confer the comparative benefit of an additional 20 years of life on an existing person than it is to confer a greater noncomparative benefit by causing a person to exist who will live 20 years longer than a different person who would otherwise come into existence. Yet in choices between some and none, noncomparative benefits are discounted even more than they are in choices between greater and lesser. Hence the reason to do A in Possible People might be stronger than the reason to do D in Possible Person, despite the fact that doing A would confer a lesser net noncomparative benefit than doing D would.

Irrespective of whether we think that noncomparative benefits have less reason-giving weight in choices between some and none, it is, as I have suggested, difficult to see how they could have no reason-giving weight in such choices when they do have some reason-giving weight in choices between greater and lesser. Yet if it is conceded that they have reason-giving weight in choices between causing a person to exist and not causing anyone to exist, various problems arise. For the reason to bestow noncomparative benefits by causing people to exist can conflict with reasons to bestow comparative benefits on existing people, or with reasons to prevent existing people from losing comparative benefits, or even with reasons not to cause or allow existing people to suffer comparative harms.

This problem is most acute if we accept the No-Difference View, which claims, in effect, that there is no moral asymmetry between comparative and noncomparative benefits. The No-Difference View does not, of course, directly imply that there is any reason to confer noncomparative benefits by causing people to exist when otherwise no new person would come into existence (that is, in choices between some and none). But Parfit defends the No-Difference View by reference to Same Number Choices such as the Medical Programs Case, of which Future or Possible is an analogue, and he assumes that it will be implied by Theory X, which will extend the Same Number Quality Claim so that it will apply to Different Number Choices. So Parfit assumes that even in Different Number Choices, the fact that an outcome is worse for a particular person makes no difference, and this is tantamount to assuming that there is no moral asymmetry between comparative and noncomparative benefits, or between comparative and noncomparative harms.

Suppose, then, that one must choose between causing a person to exist and saving an existing person’s life. Suppose, for example, that one can either cause a person to exist who would live to be 60 or save the life of a 20-year-old, enabling him to live to 80. The first of these options would confer a great noncomparative benefit, while the second would confer an equivalent comparative benefit. Everyone, I assume, would accept that the second option is better, and that one has a stronger moral reason to save the existing person than to cause a person to exist. While the No-Difference View does not imply the contrary claim, it does deny that the intuition that virtually everyone has about this choice can be defended on the ground that the failure to save the existing person would be worse for that person, or on the ground that there would be a victim of the failure to save the person but not of the failure to cause a person to exist.¹⁵

This problem is mitigated but not eliminated if we reject the No-Difference View and instead accept that the reason-giving weight of noncomparative benefits is discounted relative to that of equivalent comparative benefits. According to this view, it is better in the case just presented to save the existing person than to cause a person to exist. But if the numbers were different, this view too would imply that it is better to cause a person to exist. Suppose, for example, that the person one might cause to exist would live for 80 years while the person one might save would live only two more years. In that case, unless the discounting factor is exceedingly high, even a view that discounts the reason-giving weight of noncomparative benefits should imply that it is better to cause the person to exist than to save the existing person. This remains true even if noncomparative benefits are doubly discounted, once for being noncomparative and again because the alternative to causing them is that there will be no noncomparative benefits at all. This problem arises for any view that concedes that there is a reason to confer noncomparative benefits by causing a person to exist even when the alternative is not to cause anyone to exist.

Parfit has suggested that even a wholly impersonal moral theory can recognize one respect in which it is better, other things being equal, to save a person’s life than to bestow an equivalent benefit by causing a person to exist.¹⁶ This is that the benefit conferred in the first way would go to a population with fewer people than the benefit conferred in the second way, since the second way would expand the size of the population. When a benefit of a fixed size goes to a population with fewer people, it makes people better off on average than it would if it went to a population with more people. This may matter even if it is a mistake to suppose that in general one ought to maximize average well-being. This claim is limited, of course, to choices between saving a person in a given population and causing a person to exist who would be a member of that same population. But in practice these are the important choices and in any case one could claim that the relevant population is the set of all the people who ever live.

While I think this point is correct, it seems to have little significance in a choice between saving one person and causing one person to exist. The difference that such a choice would make to the level of well-being per person in a world as highly populated as ours is negligible. And the difference it would make if the relevant population is all the people who ever live is less than negligible.

Before concluding, I will mention one further set of problems raised by the claim that noncomparative benefits have reason-giving weight even in choices between some and none. Suppose that one can do either of the following acts but not both.

(E):

Cause an individual to exist who will live for 80 years with a life well worth living.

(F):

Save a different individual’s life immediately after he has begun to exist, thereby enabling him to life for 80 years with a life well worth living.

Almost everyone, it seems, believes that we begin to exist prior to birth. Most people seem to believe that we begin to exist at conception, though others believe that this happens when significant cell differentiation occurs, when twinning ceases to be possible, when brain activity begins, or when the capacity for consciousness first appears. All such views imply that we begin to exist with no psychological capacities or with extremely rudimentary psychological capacities.

If one does E, one will confer a great noncomparative benefit. If one does F, one will confer an equivalent comparative benefit. If comparative benefits have substantially more reason-giving weight than equivalent noncomparative benefits have in some-or-none choices, as the previous choice between saving a person and causing a person to exist suggests, it would be much better to do F, and one’s reason to do F would be much stronger than one’s reason to do E. But if we are wholly unconscious, or only dimly conscious, when we begin to exist (and for a considerable time after that), it may seem that failing to save a zygote or embryo is not much worse, or not worse at all, than failing to cause a person to exist. After all, around two-thirds of naturally conceived human embryos die spontaneously in the first few months of pregnancy, yet no scientifically advanced society has devoted more than a trifling amount of effort or resources to try to prevent these many deaths from occurring. And one might ask oneself whether from one’s present perspective one can find any relevant difference between the possibility that one might not have been caused to exist and the possibility that one might have died immediately after beginning to exist. These reflections may suggest that noncomparative benefits matter almost as much as equivalent comparative benefits.

But this would be a mistaken inference. The truth is instead that equivalent comparative benefits do not always have equivalent weight. In particular, given that we begin to exist with rudimentary psychological capacities, or no psychological capacities at all, the benefit that an embryo or fetus receives in having its life saved must be heavily discounted for the weakness of the psychological relations it bears to itself in the future. I cannot elaborate on this suggestion here, though I have defended it at length elsewhere (McMahan 2002, Chapters 1, 2, and 4). The point is that the comparative benefit one bestows in saving the life of a fetus has substantially less weight than a lesser comparative benefit one might bestow in saving the life of a 20-year-old. Hence even if causing an individual to exist matters almost as much as saving the life of an individual who has just begun to exist, that is compatible with the fact that causing an individual to exist matters substantially less than saving the life of an older child or adult.

This does, however, leave a residual problem, at least for those who believe that abortion can be permissible in a wide range of cases, including relatively late in pregnancy. No one supposes that the comparative benefit to a fetus of saving its life could matter less than an equivalent noncomparative benefit that could be produced by causing a person to exist. Thus, if noncomparative benefits have significant reason-giving weight in some-or-none choices, there must also be a significant reason to save the life of a fetus, and, assuming there is an asymmetry between doing harm and allowing harm to occur, an even more significant reason not to kill a fetus via abortion.¹⁷ Defenders of abortion must presumably claim that the benefit to the fetus of being enabled or allowed to continue to live is outweighed in most cases by conflicting considerations. But this is tantamount to conceding that this comparative benefit has relatively little weight, which in turn implies that an equivalent noncomparative benefits cannot have more weight, and presumably has even less.

While this may seem intuitively plausible, it has to be reconciled with the reasons we have discussed for accepting that noncomparative benefits have significant weight. Consider, for example, the following variant of Future or Possible.

$$ \begin{array}{*{20}c} {{\text{A:}}\,{\text{P}}_{ 1} \,{\text{will}}\,{\text{later}}\,{\text{exist}}} \hfill & {{\text{P}}_{ 2} \,{\text{will}}\,{\text{never}}\,{\text{exist}}} \hfill & {{\text{P}}_{\text{F}} \,{\text{will}}\,{\text{later}}\,{\text{exist}}} \hfill \\ {\quad \;{\text{and}}\,{\text{live}}\,{\text{to}}\, 80} \hfill & {} \hfill & {{\text{and}}\,{\text{live}}\,{\text{to}}\, 60} \hfill \\ {{\text{B:}}\,{\text{P}}_{ 1} \,{\text{will}}\,{\text{never}}\,{\text{exist}}} \hfill & {{\text{P}}_{ 2} \,{\text{will}}\,{\text{later}}\,{\text{exist}}} \hfill & {{\text{P}}_{\text{F}} \,{\text{will}}\,{\text{later}}\,{\text{exist}}} \hfill \\ {} \hfill & {{\text{and}}\,{\text{live}}\,{\text{to}}\, 40} \hfill & {{\text{and}}\,{\text{live}}\,{\text{to}}\, 6 5} \hfill \\ \end{array} $$

If noncomparative benefits have relatively little weight compared with equivalent comparative benefits, one’s reason to do B could be stronger than one’s reason to do A. I find that difficult to accept.

One final problem worth noting is that the claim that there is a general reason to confer noncomparative benefits by causing people to exist not only implies that in some cases it can be better to cause people to exist than to save people’s lives but also seems to imply, in other cases, what Parfit calls the Repugnant Conclusion (though Parfit has well-known arguments that lead to that conclusion even without the assumption that there is a general reason to cause people to exist if their lives would be worth living) (Parfit 1984, Chapters 17 and 19). So the implications of the claim that noncomparative benefits have reason-giving weight in choices between some and none include some that virtually everyone will find counterintuitive.

This is a depressing point on which to conclude, though perhaps not disappointing, as I had no real hope of solving any of the problems in this area of moral theory when I began. Problems in the morality of causing people to exist seem to me the most difficult and intractable of all the problems of which I am aware in normative and practical ethics. They suggest that it is a real possibility that any moral theory that is both complete and coherent will have implications that are intuitively intolerable. It is these problems, therefore, rather than arguments in metaethics about the queerness of objective values, the connections between normativity and motivation, and so on, that seem to me to pose the greatest challenge to realism in ethics.