Jump to ContentJump to Main Navigation
The Non-Identity Problem and the Ethics of Future People$

David Boonin

Print publication date: 2014

Print ISBN-13: 9780199682935

Published to Oxford Scholarship Online: November 2014

DOI: 10.1093/acprof:oso/9780199682935.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2017. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 22 February 2017

(p.237) Appendices

(p.237) Appendices

Source:
The Non-Identity Problem and the Ethics of Future People
Publisher:
Oxford University Press

Appendix A: Two More Problems with the Incoherence Objection (2.1)

The incoherence objection rests on two claims: that we cannot coherently make judgments about whether existing or not existing is better or worse for Pebbles, and that coherently making such judgments is necessary in order for P1 to be true. I am inclined to reject both claims.

The reason I am inclined to reject the first claim, the claim that such comparisons are incoherent, is that there are other cases in which we seem to be able to compare how existence and non-existence would be for someone without lapsing into incoherence. One kind of case is that in which it seems intelligible to say of some already existing person that he would be better off dead. Suppose, for example, that Fred is on fire, that he is suffering excruciating pain, and that he will either die right away or will die after suffering five additional hours of unfathomable agony. In this case, it seems to make sense to say that Fred would be better off dying now rather than surviving only to endure another five hours of unimaginable suffering. And this seems to make sense not because it means that if Fred dies now he will then be in a condition that is better for him than the condition of being alive and suffering horribly, but rather because it means that from the point of view of his own interests, Fred should prefer not to exist in any condition at all for the next five hours rather than to exist in excruciating pain.

But if we can make sense of the claim that Fred would be better off dead without comparing how things would be for Fred existing and not existing, then we can also make sense of the claim that Pebbles would be better off being alive without comparing how things would be for Pebbles existing and not existing. Just as it can make sense to say of a particular person that it is better for him that he no longer exist than that he continue to exist given that his existence involves a certain amount of pain, so it can make sense to say of a person who would otherwise not exist that it is better—and better for her—to exist with a certain disability than not to exist at all. In both cases, what is meant is simply that from the point of view of the interests of the person who would exist in one but not the other of two possible states of affairs, one state of affairs is rationally preferable to the other.

A critic of this first response to the incoherence objection might claim that we can account for our ability to make this judgment in the case of Fred without comparing existence and non-existence. On this alternative account, what we really mean when we say that Fred would be better off dead is that Fred’s life, when taken as a whole, will (p.238) turn out to have gone better for Fred if he dies right now than Fred’s life as a whole will turn out to have gone for Fred if he goes on living for another five hours in terrible pain. This explanation of the judgment that we make when we say that Fred would be better off dead does not involve comparing existence and non-existence in terms of how they would be for Fred. It simply involves comparing two different ways that things could go for Fred over the course of the period during which he does exist in terms of which would be better for Fred. And so, on this account, the fact that we can coherently say that Fred would be better off dead provides no support for the claim that we can coherently compare existence and non-existence in the way that P1 seems to require us to do. I’m inclined to reject this alternative explanation of the judgment we make in the case of Fred. There is a difference between asking what would make Fred’s life on the whole turn out to have gone better and asking what would be better for Fred from here on out. In ordinary cases, these two criteria go hand in hand. Whatever makes the rest of a person’s life go better typically makes that person’s life overall go better, too. But in idiosyncratic cases, the two criteria can come apart. In such cases, most people seem to care about what will make the rest of a person’s life go better rather than what will make the person’s life as a whole turn out to have gone better. It thus seems plausible to suppose that this is the kind of judgment most people really mean to be making in the case of Fred.

Suppose, for example, that you wake up in a hospital with no memory of how you got there. A doctor then tells you that you are one of two new patients, that she doesn’t know which one you are, but that she knows that either one or the other of the following is true: (1) yesterday, you underwent a terribly painful procedure that caused you 1,000 units of pain and you were then given a drug to make you forget this horrible experience, or (2) tomorrow you will undergo a somewhat less painful procedure that will cause you 800 units of pain, after which you will be given a drug to make you forget this very bad but not quite horrible experience.1 As the doctor is examining your chart to determine which situation is the one that you are in, you ask yourself which answer you should rationally prefer. If what you should prefer is the answer that will mean that your life as a whole will turn out to have gone better, you should prefer the second answer: suffering 800 units of pain tomorrow rather than 1,000 units of pain yesterday will be better for you by that measure. If what you should prefer is the answer that will mean that things will go better for you from here on out, then you should prefer the first answer: suffering 1,000 units of pain yesterday rather than 800 units of pain tomorrow will be better for you by that measure.

Virtually everyone, I suspect, would strongly prefer to receive the first answer. They would much rather have the 1,000 units of pain behind them than the 800 units of pain in front of them. And I suspect that this would be true even if they could be given a pill in the meantime that would let them forget that they have the 800 units of pain in (p.239) front of them. If that’s right, then when the two criteria come apart, what most people really care about when they make judgments about what would be in a person’s interest is what would be best for them from here on out, rather than what would make their life as a whole turn out to have gone best. And this, in turn, undermines the alternative explanation of the judgment that most people would make in the case of Fred. When we ask which state of affairs would be better for Fred one hour from now, the state of affairs in which he is suffering horribly or the state of affairs in which he does not exist, we are not asking which state would make it the case that his life as a whole turns out to have gone better. We are asking which would be better from that point on. This is the judgment that strikes most people as coherent, and if it is coherent, then so is the comparison between existing and not existing that seems to be required by P1 of the non-identity argument.

But let’s suppose that I’m wrong about this and that the alternative explanation of our judgment in the case of Fred really does prevent that example from undermining the first claim made by the incoherence objection to P1. Even if it does, there is a second kind of case that is not subject to this kind of response: cases where we think someone would have been better off never having existed. Suppose, for example, that a single sperm and egg are in a Petri dish and will shortly come together to form a zygote. If they do, it is inevitable that the zygote will eventually develop into a child who will experience nothing but excruciating pain from the moment he first becomes conscious until he dies at the age of three. And for ease of exposition, let’s suppose that the prospective parents have already decided on a name: they will call their child Barney. It seems just as clear that Barney would be better off never being conceived as it does that Fred would be better off dead. If you were trying to make a decision from the point of view of the interests of the person who would come into existence if the sperm and egg came together, that is, it seems clear that the right thing to do would be to prevent the conception from occurring, rather than to say that the idea of making a decision from that point of view was incoherent. But the alternative explanation that interpreted our judgment about Fred in terms of which option would make Fred’s life as a whole go best can’t be applied to the case of Barney. In the case of Barney, we really do seem to be coherently comparing existence and non-existence. And so even if the case of Fred fails to show that we can coherently compare existence and non-existence in the way that P1 seems to require us to do, the case of Barney can still establish this and can thus undermine the first claim made by the incoherence objection.

Before turning to the objection’s second claim, it is worth briefly considering a further response that might be offered to my argument against the first. A proponent of the incoherence objection might agree that the first claim made by the objection is incompatible with the judgments that virtually everyone would make in the cases of Fred and Barney. But rather than saying so much the worse for the incoherence objection, this critic might say so much the worse for the judgments that most people would make. We might think we are saying something coherent when we say that Fred would be better off dead and that Barney would be better off never coming into existence, that (p.240) is, but perhaps we are mistaken about this. And if we are mistaken about this, then the conflict between these common judgments and the claim made by the incoherence objection fails to count against the objection.

I don’t see how we could decisively establish that our judgments in the cases of Fred and Barney are coherent, and so I don’t think that this response to my argument can be decisively overturned, though the burden of proof clearly lies with those who claim that the judgments are incoherent. But I don’t think that overturning the response is necessary in order to establish that the response would fail to ground a satisfactory solution to the non-identity problem. As I noted in section 1.3.3, a solution is not satisfactory if it eliminates the non-identity problem by replacing it with an even bigger problem. Such a solution would fail what I called the modesty requirement. And that, it seems to me, is what would happen if we were to rest a solution to the non-identity problem on the incoherence objection and rest the incoherence objection on the claim that our judgments about cases like Fred and Barney are mistaken. Even if such a solution had no further problems, and as I argue below and in section 2.1 it does have further problems, it would enable us to say that it is wrong for Wilma to conceive a blind child rather than a sighted child only by compelling us to deny that Fred would be better off dead and that Barney would have been better off never being conceived. These implications strike me as more implausible than the Implausible Conclusion that the solution is attempting to help us avoid. And so it seems, at least to me, that this approach to salvaging the incoherence objection is unsatisfactory.

I’ve argued so far that we should reject the first claim made by the incoherence objection. But let’s now suppose that the first claim made by the objection is correct and that it really is incoherent to insist that Pebbles is better off existing than not existing. Even if this is so, I’m inclined to think that the incoherence objection should still be rejected. This is because I’m inclined to think that the second claim made by the objection is false. The second claim maintains that P1 requires us to compare Pebbles existing and Pebbles not existing in terms of one being better or worse for her than the other. But P1 does not require this. P1 does not maintain that Pebbles is positively better off existing than not existing. It merely claims that it is not the case that she is worse off existing. And it seems to me that the claim that it is not the case that she is worse off existing can be true even if the claim that she is better off existing is incoherent. Consider, for example, the claim that “green is smaller than justice.” That claim strikes me as incoherent. It maintains that the relation “is smaller than” obtains between green and justice, and they are not the kinds of things between which that relation can obtain. But now consider the claim that “it is not the case that green is bigger than justice.” This claim strikes me as perfectly coherent. It does not maintain that a particular relation obtains between green and justice. It simply claims that it is not the case that a particular relation, the relation of being bigger than, does obtain between them. And it seems true that it is not the case that that relation obtains between them. In the same way, and for the same reason, the claim that it is incoherent to say that Pebbles is better off existing than not existing does nothing to undermine the truth of the claim that it (p.241) is not the case that Pebbles is worse off existing than not existing. And that is all that P1 says: Wilma’s act has not, in fact, made Pebbles worse off than she would otherwise have been.2 Even if the first claim made by the incoherence objection is true, then, its second claim seems to be false.

Appendix B: A Further Problem with Benatar’s Argument (2.5)

Benatar’s argument depends on the claim that his Relational Asymmetry Principle (RAP) provides the best explanation for the moral asymmetry between cases like that of the Blessed Couple and the Cursed Couple. I will reject this claim here by arguing that we can identify an alternative principle that does a better job of explaining the asymmetry than does Benatar’s principle and that produces a different result in the case of Wilma.3 We can do this if we start by appealing to a symmetric version of Benatar’s principle, one in which (1)–(3) are the same as in RAP but in which the claim made about the absence of pleasure in (4) is symmetric to the claim made by RAP (3) about the absence of pain. What I will call the Relational Symmetry Principle (RSP), then, amounts to taking RAP (1)–(3) and conjoining them with:

(4) the absence of pleasure is worse than the presence of pleasure if either (a) there is an actual person whose interests are better served by the presence of the pleasure or (b) the absence of the pleasure would require the absence of a person who would otherwise exist and whose potential interests are better served by the presence of the pleasure

If RAP (3) is logically coherent, then surely RSP (4) is logically coherent as well. If the absence of pain can be better than the presence of pain even when there is no actual person who enjoys the absence of pain, that is, then the absence of pleasure can be worse than the presence of pleasure even when there is no actual person who is deprived of the absent pleasure. Benatar himself concedes this much.4 My claim here will be not only that RSP is as coherent as RAP, but that RSP can be used to undermine Benatar’s argument.

(p.242) To see how this can be accomplished, let’s begin by considering what my Relational Symmetry Principle would entail about the case of the Cursed Couple and the Blessed Couple. If the Cursed Couple conceives, the resulting child’s life on the whole will contain zero units of pleasure and a million units of pain. RSP (3b) is identical to RAP (3b). It says that the absence of pain is better than the presence of pain if the presence of the pain would require the existence of a person who would not otherwise exist and whose potential interests are better served by the absence of the pain. And so RSP, just like RAP, makes it plausible to say that by conceiving, the Cursed Couple would make things worse for the Cursed Child. If the Blessed Couple conceives, on the other hand, the resulting child’s life on the whole will contain a million units of pleasure and zero units of pain. RAP (4) entailed that by not conceiving, the Blessed Couple would not make things worse for the Blessed Child. But RSP (4b) maintains that the absence of pleasure is worse than the presence of pleasure even if the absence of the pleasure would require the absence of a person who would otherwise exist and whose potential interests are better served by the presence of the pleasure. So my RSP, unlike Benatar’s RAP, entails that the Blessed Couple would make things worse for the Blessed Child by not conceiving.

Does this mean that the Relational Symmetry Principle is unable to account for the asymmetry between the Blessed Couple and the Cursed Couple? It does not. The asymmetry between the Blessed Couple and the Cursed Couple can be accounted for by appealing to an asymmetry between pleasure and pain that is different from the asymmetry that Benatar appeals to and that is fully consistent with affirming the Relational Symmetry Principle. The asymmetry is that with respect to pleasure, no actual person is made worse off by the decision about whether to conceive them regardless of whether the better or worse outcome is selected, but with respect to pain, whether an actual person is made worse off by the decision about whether to conceive them depends on whether the better or worse outcome is selected.

If the Blessed Couple decides to conceive the Blessed Child, for example, then things will be better from the point of view of the interests of the Blessed Child, and the Blessed Child will therefore not be made worse off by the decision. If the Blessed Couple decides not to conceive the Blessed Child, then things will be worse from the point of view of the interests of the Blessed Child, but the Blessed Child will not exist and so will not be an actual person who has been made worse off by the decision. Regardless of which choice the Blessed Couple makes, then, no actual person will be made worse off by their decision. But pain is different from pleasure in this respect. Whether an actual person will be made worse off by the Cursed Couple’s decision about whether to conceive depends on the choice they make. If the Cursed Couple decides not to conceive the Cursed Child, then things will be better from the point of view of the interests of the Cursed Child. No actual person will be made worse off by their decision both because there will be no actual person to be affected by their decision and because their decision made things better rather than worse. But if the Cursed Couple decides to conceive the Cursed Child, then things will be worse from (p.243) the point of view of the interests of the Cursed Child, and the Cursed Child will exist to be made worse off by this fact. There will therefore be an actual person who has been made worse off by their decision. While in the case of pleasure no actual person is made worse off regardless of whether the Blessed Couple chooses the better or worse outcome, then, in the case of pain, whether or not an actual person will be made worse off depends on whether the Cursed Couple chooses the better or worse outcome.

Because of this asymmetry between pleasure and pain, the Relational Symmetry Principle that I am proposing as an alternative to Benatar’s Relational Asymmetry Principle can account for the claim that it would be wrong for the Cursed Couple to conceive the Cursed Child but not wrong for the Blessed Couple not to conceive the Blessed Child. The Relational Symmetry Principle can do this by appealing to what I will call the Actual Persons Principle (APP), where by an “actual person” I mean a person who actually exists at some point in time:

APP: When choosing between two options, it is prima facie wrong to make the choice the acting on which will result in its being the case that there is an actual person for whom your act made things worse.

RSP entails that if the Cursed Couple conceives the Cursed Child, this will make things worse for the Cursed Child. Since the Cursed Child will exist, the Actual Persons Principle entails that this will make the Cursed Couple’s act prima face wrong. RSP also entails that if the Blessed Couple does not conceive the Blessed Child, this will make things worse for the Blessed Child. But since if the Blessed Couple does not conceive the Blessed Child, the Blessed Child will never exist, and since the Actual Persons Principle focuses only on the harmful effects on people who actually exist at some point in time, the principle will not entail that the Blessed Couple’s act is prima facie wrong. Thus, while Benatar’s Relational Asymmetry Principle can account for the asymmetry between the Blessed Couple and the Cursed Couple, my Relational Symmetry Principle can account for it as well, provided that it is conjoined with the Actual Persons Principle.

As we saw in section 2.5, Benatar’s RAP entails that Wilma harms Pebbles by conceiving her. What does my RSP entail about this case? Suppose, as we did in that section, that if Wilma conceives Pebbles, Pebbles’s life on the whole will contain 500,000 units of pleasure and 20,000 units of pain. As in the case of RAP, we can follow Benatar and see how the case looks from the point of view of RSP by considering the relevance of the pain that Pebbles would suffer and of the pleasure that Pebbles would enjoy independently. With respect to the 20,000 units of pain that Pebbles will suffer if Wilma conceives her, RSP entails that existing will be worse for Pebbles than not existing. RSP (3b), just like RAP (3b), maintains that the absence of pain is better than the presence of pain even if the presence of the pain would require the existence of a person who would not otherwise exist and whose potential interests are better served by the absence of the pain. So, with respect to the choice between existing and not existing, (p.244) RSP, just like RAP, treats the 20,000 units of pain as counting against bringing Pebbles into existence.

With respect to the 500,000 units of pleasure that Pebbles will enjoy if Wilma conceives her, RAP (4) maintains that the absence of pleasure is worse than the presence of pleasure only if there is an actual person whose interests are better served by the presence of the pleasure. Since if Pebbles is not conceived there will be no actual Pebbles who is deprived by the absence of the 500,000 units of pleasure, RAP maintains that the absence of pleasure does not make it worse for Pebbles not to exist than to exist. But my RSP is crucially different from Benatar’s RAP in precisely this respect. RSP (3) is identical to RAP (3), but where RAP (4) maintains that the absence of pleasure is worse than the presence of pleasure only if there is an actual person who is harmed by the absence of pleasure, RSP (4b) maintains that the absence of pleasure is worse than the presence of pleasure even if there is no actual person who is harmed by the absence of the pleasure because the absence of the pleasure requires the absence of the person who would otherwise exist with the pleasure. So RSP maintains that the absence of pleasure does make it worse for Pebbles not to exist than to exist. With respect to the choice between existing and not existing, the 500,000 units of pleasure does count in favor of bringing Pebbles into existence according to my RSP, while it does not count in favor of bringing Pebbles into existence according to Benatar’s RAP.

Regarding the 20,000 units of pain, then, RSP maintains that existence is worse than non-existence, and regarding the 500,000 units of pleasure, RSP maintains that existence is better than non-existence. Since the total amount of pleasure clearly outweighs the total amount of pain in this case, the magnitude of the advantage that existence has over non-existence is greater than the magnitude of the advantage that non-existence has over existence. And so, in the end, RSP entails that in the case of Pebbles, the advantages of existence clearly outweigh the advantages of non-existence. Benatar’s RAP entails that Wilma harms Pebbles by conceiving her. But my RSP does not entail this.

We now have two principles that are capable of accounting for our asymmetric judgment in the case of the Blessed Couple and the Cursed Couple: Benatar’s Relational Asymmetry Principle and my Relational Symmetry Principle. The question, then, is which principle provides the better account. There are three reasons to conclude that my Relational Symmetry Principle does.

The first reason to favor RSP over RAP is that RAP entails that Wilma harms Pebbles by conceiving her while RSP entails that Wilma does not harm Pebbles by conceiving her. It may seem that I am begging the question here. It may seem that I am saying, “We should believe that Wilma does not harm Pebbles by conceiving her because RSP entails that she doesn’t, and we should believe RSP because RSP entails that Wilma does not harm Pebbles by conceiving her.” That would, indeed, be circular. But that is not what I am saying. What I am saying is this: Benatar provides, as a reason for endorsing RAP, the claim that RAP accounts for the difference between the Blessed Couple and the Cursed Couple. Accepting the deeply counterintuitive claim that Wilma harms Pebbles by conceiving her is simply the price we have to pay in order to (p.245) preserve our judgments about the cases of the Blessed and Cursed Couples. But RSP shows that we can account for the judgments in these two cases without paying that price. If, among two competing explanations for the same set of judgments, one has fewer counterintuitive implications in additional cases than the other, then that is a legitimate, non-circular reason for preferring it. The case of Wilma and Pebbles therefore provides a legitimate, non-circular reason to prefer RSP over RAP. And, of course, once we have selected RSP over RAP for this legitimate, non-circular reason, there is no longer any reason to doubt our initial judgment that Wilma does not make Pebbles worse off by conceiving her.

A second reason to prefer my RSP over Benatar’s RAP is independent of RSP’s implications for the case of Wilma and Pebbles. Instead, it arises from the fact that while both RSP and RAP involve affirming the existence of an asymmetry between pleasure and pain, the asymmetry that RSP appeals to is perfectly straightforward while the asymmetry that RAP appeals to is quite counterintuitive. The asymmetry that my RSP appeals to arises directly from the fact that pleasure is good and pain is bad. Since pleasure is good, the worse outcome regarding pleasure is when it is absent. Since pain is bad, the worse outcome regarding pain is when it is present. There is nothing surprising or mysterious about this. In order for an actual person to be made worse off by something, they must actually exist at some point in time. There is nothing surprising or mysterious about this, either. But these two perfectly straightforward observations are enough to ground the asymmetry between pleasure and pain that my RSP appeals to in accounting for the asymmetry between the Blessed Couple and the Cursed Couple: since existing is the worse outcome in the case of the Cursed Child, there is an actual person who is made worse off if the Cursed Couple makes the worse choice. Since not existing is the worse outcome in the case of the Blessed Child, there is no actual person who is made worse off if the Blessed Couple makes the worse choice.

But the asymmetric claim that Benatar’s RAP appeals to should seem quite surprising. It seems odd to say that the absent pain of non-existence makes it better for the Cursed Child not to exist but that the absent pleasure of non-existence does not make it worse for the Blessed Child not to exist. How can the absence of something have an effect on the would-be child’s welfare in one case but not in the other? Benatar, of course, provides a reason for accepting this asymmetry: the claim that accepting it is necessary in order to account for the asymmetry between our moral judgments in the case of the Blessed Couple and the Cursed Couple. But my RSP shows that Benatar’s RAP is not necessary in order to account for this feature of our moral judgments. Once we see this, we lose our reason to endorse the asymmetry that Benatar appeals to. And once we lose our reason to endorse the asymmetry that is appealed to by RAP, the fact that the asymmetry itself is intuitively less plausible than the asymmetry appealed to by RSP provides a second independent reason to prefer my RSP over Benatar’s RAP.

A final reason to prefer RSP over RAP is also independent of its implications for the case of Wilma and Pebbles. It arises from the fact that while both RSP and RAP can produce the intuitively correct answers in the case of the Blessed Couple and the (p.246) Cursed Couple, there is an important difference between the two principles in terms of the additional assumptions that have to be accepted in order for them to be able to do so. Neither RSP nor RAP by themselves yield any conclusions about what it would be right or wrong to do. Neither principle makes any reference to right or wrong, obligatory or impermissible, should or should not. Instead, they make judgments about which states of affairs are good or bad, or better or worse. Without some sort of bridging principle, therefore, neither can justify any conclusions about what should or should not be done. And this is what generates the third reason to prefer my RSP over Benatar’s RAP: the bridging principle needed in order for RSP to account for our judgments in the Blessed Couple and Cursed Couple cases is superior to the bridging principle needed in order for RAP to do so.

As I noted earlier, RSP can account for the asymmetry between the Blessed Couple and the Cursed Couple by appealing to what I called the Actual Persons Principle.

APP: When choosing between two options, it is prima facie wrong to make the choice the acting on which will result in its being the case that there is an actual person for whom your act made things worse.

In order to account for the asymmetry between the Blessed Couple and the Cursed Couple by appealing to RAP, however, we must instead depend on what I will call the Actual and Possible Persons Principle (APPP), where by a “possible person” I mean a person who will never actually exist, but who would have actually existed had the other option been selected:

APPP: When choosing between two options, it is prima facie wrong to make the choice the acting on which will result in its being the case that there is an actual or possible person for whom your act made things worse.

That Benatar’s argument requires us to assume the Actual and Possible Persons Principle is clear from the following considerations. Benatar claims that we must say that the absent pleasures if the Blessed Child is not conceived do not make it worse for the child that he not be conceived. He claims that we must say that it is not worse for the Blessed Child that he not be conceived so that we can avoid having to say that it would be wrong for the Blessed Couple not to conceive the Blessed Child. But the claim “not conceiving the Blessed Child does not harm the child” is necessary to sustaining the claim “not conceiving the Blessed Child is not morally wrong” only if it would be morally wrong for the Blessed Couple to harm the Blessed Child by not conceiving him. The Actual Persons Principle does not entail that this would be wrong: even though not conceiving the Blessed Child would make things worse for the Blessed Child, not conceiving the Blessed Child would not be wrong because it would not result in an actual person for whom refraining from conceiving the Blessed Child had made things worse. Only if we include merely possible people in the scope of our considerations can Benatar’s argument for RAP succeed. And so Benatar’s argument requires us to accept the Actual and Possible Persons Principle.

(p.247) But there are two reasons to prefer the bridging principle required by my RSP to the bridging principle required by Benatar’s RAP. The first is that the RSP bridging principle is simpler. The Actual Persons Principle posits only one kind of entity whose interests we must consider: people who at some point actually exist. The Actual and Possible Persons Principle includes this kind of entity plus a second and very different kind of entity: those people who do not exist and who will never exist but who would have existed had a different choice been made. All else being equal, a simpler principle should be preferred to a more complex principle, and this provides one reason to prefer the bridging principle required by RSP to the bridging principle required by RAP.

The second reason to prefer the RSP bridging principle to the RAP bridging principle is that the RSP bridging principle appeals to a consideration whose moral salience is less controversial. The fact that an act would harm an actual person is generally recognized as relevant to the question of whether the act would be wrong. But the fact that an act would be worse for the interests of a merely possible person who will never exist if the act is performed is not already accepted to be relevant in the same way. Indeed, to say that we must consider the interests of people who will never exist is quite controversial. All else being equal, a principle whose salience is less controversial should be preferred to a principle whose salience is more controversial. And this provides a second reason to prefer the bridging principle required by my RSP to the bridging principle required by Benatar’s RAP.

Benatar, of course, might maintain that all else is not equal, that RAP itself has a great amount of explanatory power because it can account for our asymmetric moral judgments in the case of the Blessed Couple and the Cursed Couple. But, as I have already argued, RSP can account for these cases as well. If the bridging principle required by RSP is simpler and less controversially salient than that required by RAP, and if RSP has at least as much explanatory power as RAP, then we have a third and final reason to prefer RSP to RAP. Combined with the fact that RAP implausibly entails that Wilma makes Pebbles worse off by conceiving her while RSP more plausibly entails that she does not, and with the fact that the asymmetry that RSP appeals to is more plausible than the asymmetry that RAP appeals to, we have three good reasons to prefer RSP over RAP. And by justifying the choice of RSP over RAP, we justify rejecting Benatar’s argument for the claim that Wilma’s act of conceiving Pebbles makes Pebbles worse off than she would otherwise have been.

I have argued that we can justify rejecting Benatar’s argument by showing that RSP better accounts for our intuitions about the Blessed Couple and the Cursed Couple than does RAP and by showing that RSP entails that Wilma does not make Pebbles worse off than she would otherwise have been by conceiving her. I want now to consider two responses that might be offered on Benatar’s behalf. The first response arises from the fact that Benatar’s defense of RAP as a whole appeals to its ability to explain our asymmetric intuitions in four kinds of cases, not just the kind of case involved in the Blessed and Cursed Couples. Even if RSP proves superior to RAP in the context of the first kind of case, then, Benatar’s RAP might still prove superior overall. The (p.248) second response arises from the fact that Benatar briefly responds to an objection that is in some respects similar to the one that I have developed here. One might then ask whether his response would apply to my objection as well.

Let us begin with the supplemental cases that Benatar appeals to in supporting RAP. The first such case is this: “Whereas it is strange (if not incoherent) to give as a reason for having a child that the child one has will thereby be benefited, it is not strange to cite a potential child’s interests as a basis for avoiding bringing a child into existence.”5 RAP accounts for this asymmetry by maintaining that bringing a happy child into existence does not really make the child better off whereas bringing an unhappy child into existence really does make the child worse off. But RSP can account for this asymmetry just as easily. It can explain why it is not strange to cite a potential child’s interests as a reason not to conceive, as in the case of the Cursed Couple. It can explain this because it maintains that creating an unhappy child causes harm to an actual person and because it is not strange to cite the negative impact on the interests of an actual person as a reason not to bring the child into existence. And it can also explain why it is strange to give as a reason for conceiving a happy child that the child will thereby be benefited. The reason that giving this as a reason is strange is not that it involves saying something that isn’t true. According to RSP, it is true that a happy child benefits from being brought into existence. And so it is true that the Blessed Couple would benefit its child by conceiving it. The reason it would be strange for the Blessed Couple to give this as a reason for conceiving the Blessed Child is that, if they don’t bring him into existence, there won’t be any actual child whom they have failed to benefit. In cases where it clearly does make sense to cite as a reason for doing a particular action the fact that doing the action would benefit someone, that is, it makes sense because there is an actual person who stands to benefit from the action being done and who will have to do without the benefit if the action is not done. But when the Blessed Couple is deciding whether or not to conceive, there is no actual person who stands to benefit from their deciding to conceive and no actual person who will have to do without a benefit if they decide not to conceive. And there is nothing about RSP that prevents us from appealing to this consideration in explaining our intuitions in the first supplemental case that Benatar appeals to.

Benatar’s second supplemental case involves reasons for regret rather than reasons for action.6 If we bring an unhappy person into existence, we can later regret this for the sake of the unhappy person himself. If we decline to bring a happy person into existence, on the other hand, we may regret this for our own sake, but we cannot really regret it for the sake of the person we could have conceived. RAP is again able to account for this asymmetry. We have harmed the unhappy person by creating him, and so regret for this for his sake is appropriate, but we have not harmed the potential but not actual happy person by not creating him, and so regret for his sake is not appropriate. As Benatar puts it, “The reason why we do not lament our failure to bring (p.249) somebody [happy] into existence is because absent pleasures [unlike existing pains] are not bad.”7

But RSP can account for this asymmetry just as well, too. If the Cursed Couple conceives the Cursed Child, then RSP entails that they have made things worse for this child, just as RAP does. Since their child will be an actual person, there will therefore be an actual person they have harmed and thus an actual person for whose sake they can appropriately regret their action. If the Blessed Couple does not conceive the Blessed Child, then RSP, unlike RAP, does entail that they have made things worse for the Blessed Child. But since in that case the Blessed Child does not exist, there exists no one for whose sake they can regret their decision. And so, as on the RAP account, they can regret their choice for their own sake, but not for the sake of any actual person that they have harmed. The case of asymmetric sources of regret then, like the case of asymmetric reasons for action, provides no reason to prefer RAP over RSP.8

Finally, Benatar appeals to an asymmetry between our response to distant places where people are suffering and our response to distant places that are uninhabited.9 When we learn of a distant place where people are suffering, we are sad for those people. But when we learn, say, of an uninhabited island, we are not sad for the people who, if they had existed, would have happily lived there. RAP is once more able to account for this asymmetry: the suffering people are worse off because of their suffering but the non-existent people are not worse off because of their not existing. And so we are right to feel bad about the existing pains and not feel bad about the absent pleasures. But RSP can once again account for the asymmetry, too. We feel bad for the suffering of actual people not just because their suffering makes things worse for them but because their suffering is something bad that is actually happening to them. And while the non-existing happy people really would have benefited from being brought into existence, according to RSP, it is nonetheless the case that nothing bad is happening to them by their not having been brought into existence—indeed, nothing at all is happening to them since they don’t existand so their absent pleasures, while worse than existing pleasures, give us no reason to feel bad for them. There is simply no them for whom we can feel bad.

The result of considering all four cases that Benatar presents in defense of his argument, then, is that we have three reasons to favor RSP over RAPthe reason arising from the case of Wilma and Pebbles, the reason arising from the more plausible (p.250) asymmetry that RSP appeals to, and the reason arising from the more parsimonious and more salient bridging principle that RSP depends onand no reason to favor RAP over RSP, since all the cases that RAP can account for can also be accounted for by RSP. The first kind of response to my objection to Benatar’s argument, then, should be rejected.

The second kind of response to my objection arises from something that Benatar says about the distinction between positive and negative duties. Immediately after arguing that RAP is supported by its ability to explain our intuitions in cases like the Blessed and Cursed Couples, Benatar notes that someone might propose an alternative explanation that does not involve the asymmetry between RAP (3) and RAP (4). Specifically, he anticipates someone suggesting that the difference can be accounted for by appealing to the distinction between a negative duty not to cause harm and a positive duty to actively “bring about happiness.” If we have a duty of the former sort but not of the latter sort, then this might explain why the Cursed Couple has a duty not to conceive while the Blessed Couple does not have a duty to conceive. And it could explain this without appealing to the sort of asymmetry that Benatar’s principle involves. Benatar responds to this possibility as follows: “I agree that for those who deny that we have any positive duties, this would indeed be an alternative explanation to the one I have provided. However, even of those who do think that we have positive duties only a few also think that amongst these is a duty to bring happy people into existence.”10 Since virtually everyone who believes in positive duties will still deny that it would be wrong for the Blessed Couple not to conceive, that is, a response to Benatar’s argument that depends on denying the existence of such duties will prove unsatisfactory.

It may seem that this alternative account of why it is not wrong for the Blessed Couple not to conceive is the same as the alternative explanation that I have provided here, and that the objection that I have raised here is therefore one that Benatar has already satisfactorily answered. But this is not so. It is true that the RSP-based response that I have endorsed maintains that the Blessed Couple’s child will be better off if they conceive it than if they don’t. And it is also true that the RSP-based response maintains that the Blessed Couple nonetheless has no moral obligation to conceive the child. It is therefore true that the objection to Benatar that I have defended here maintains that the Blessed Couple has no moral obligation to do the act that would be better for the child they would conceive.

But it does not follow from this that the RSP-based objection to Benatar’s argument depends on the claim that we have no positive moral duties. The RSP-based account is compatible with the claim that we have a great many positive duties, provided only that they are positive duties that we have to other actual people, people with actual lives that will go worse for them in various ways if we refrain from benefiting them. If there is a child drowning in front of you, or an injured motorist lying by the side of the road, (p.251) or a person starving in some distant country, for example, the RSP-based objection to Benatar’s argument can allow that it would be morally wrong for you not to provide assistance. Indeed, even in cases where your assistance would simply make already happy people even happier, the RSP-based objection to Benatar’s argument is consistent with the claim that it would be morally wrong for you not to provide it. If an already very happy child is about to enjoy a delicious ice cream sundae, for example, and if it would taste even better if you were to let him have one of your extra cherries to put on top of it, then the RSP-based account could allow that it would be morally wrong for you not to donate the extra cherry you happen to have to the child.

The RSP-based account can allow that you would have positive duties in any or all of these cases because in these cases there is an actual person whose life will go less well if you do not benefit them. If the Blessed Couple declines to conceive, by contrast, there will be no actual person whose life will go less well as a result. And so the claim made by the RSP-based account that I have offered here, the claim that it is not immoral for the Blessed Couple to refrain from conceiving since doing so does not result in there being an actual person who is made worse off as a result, is consistent with (but does not require us to accept) the claim that we have a prima facie positive duty to benefit actual people and not just a negative duty not to harm them. It is not the fact that the Blessed Couple would (merely) benefit the Blessed Child by conceiving it that renders it morally acceptable for them not to conceive the child on this account. It is the fact that if they don’t conceive the child, there will be no actual person whose life will go less well as a result: no actual person that they have harmed and no actual person that they have failed to benefit.

Appendix C: Three More Problems with Harman’s Argument (3.3.1)

(1) Harman’s claim that the rape and Nazi prisoner cases pose a problem for the Counterfactual Account of harm depends on our accepting the claim that the victims are harmed by the acts described in these cases despite the fact that they are not made worse off by the acts than they would otherwise have been. But given that they are not made worse off than they would otherwise have been, it is not clear why we must accept the claim that they have been harmed by the acts that wronged them. Harman claims that in each of the two cases three things are true: that the subject of the act has a legitimate complaint about the act’s having been performed, that the act was morally impermissible, and that the act harmed the victim. Although she is not explicit about this, the argument for the claim that the acts harmed the victims seems to assume that everyone will agree with the first two claims, and that this will lead them to accept the third claim as a result: if we don’t think the person was harmed, after all, then how could we think that they have a legitimate complaint about the act and how could we think that the act was morally impermissible? But there is a reasonable answer to this (p.252) question, and thus a way that a defender of the Counterfactual Account could deny that the two cases Harman appeals to involve acts that harm their victims.

The answer to the question of how the acts could be impermissible if they do not harm their victims is that some acts are impermissible not because they are harmful, but because they are done without consent. If a patient refuses to consent to a given operation, for example, this can make it impermissible for the doctor to perform the operation and can give the patient a legitimate complaint if the operation is performed against his will, and this can be so even if the doctor and the patient both know that the operation will be beneficial to the patient. In the same way, and for the same sort of reason, if a woman refuses to consent to sexual intercourse with a particular man, this can make it impermissible for the man to have sexual intercourse with her and can give her a legitimate complaint if he has intercourse with her against her will, and this can be so even if the man’s act turns out to be beneficial to the woman rather than harmful and even if both of them knew that this would be the case. Since the absence of consent is sufficient to account for the wrongness of the acts in the rape and Nazi prisoner cases, and since the acts would therefore be impermissible regardless of whether they end up harming their victims, the fact that we take them to be impermissible does nothing to support the claim that we should also take them to be harmful.

Harman anticipates this objection. She responds as follows:11

I agree with the objector that lack of consent plays a crucial explanatory role in both cases. But I affirm that harm also plays a crucial explanatory role. Part of the explanation of why it is wrong to rape, and why it is wrong to imprison someone as the Nazis did, is how awful it is for the victim to go through, and to have gone through, these experiences. Part of what makes these experiences awful to go through is that [they] were not consented to. I claim that consent does play a crucial explanatory role, but that the harm to someone of going through these experiences also plays an important role in explaining what is wrong with these experiences.

But this response is unsatisfactory for two reasons.

First, Harman provides no reason to believe that harm plays a crucial role in explaining the wrongness of the acts in question. She does “affirm” and “claim” that harm plays such a role, but she offers no argument to show that it does. Second, the claim that harm plays a crucial role in accounting for the wrongness of the acts is either too strong to be plausible or too weak to support Harman’s conclusion. If the claim that harm plays a crucial role means that the presence of harm is necessary in order for the acts in question to be wrong, then the claim would be strong enough to support Harman’s conclusion. The acts in question are clearly wrong and so if harm is necessary in order for the acts to be wrong, then the acts are clearly harmful. But the claim that harm is necessary in order for the acts to be wrong entails that harmless variants of the acts would be permissible even in the absence of consent. And this result is unacceptable. It would be wrong for a doctor to perform a procedure on a competent adult patient (p.253) if the patient refused to consent to it, even if it was clear that the procedure would benefit the patient and even if the patient acknowledged that this was so. In the same way, it would be wrong to have sexual contact with an unwilling person or to confine a person against their will even in circumstances in which it was clear that the behavior in question would benefit them and even if they acknowledged that this were so. If, on the other hand, the claim that harm plays a crucial role in the wrongness of the acts in Harman’s cases means something other than that the presence of harm is necessary in order for the acts in question to be wrong, then our agreeing that the acts are wrong in these cases gives us no reason to agree that the acts are harmful. If the acts are wrong regardless of whether they are harmful, after all, then the fact that they are wrong provides no reason to believe that they are harmful.

(2) I believe that there is an additional reason to reject the Non-Comparative Account. The reason rests on a presumption that killing is a form of harming. If this is so, then a successful analysis of harming should be symmetric with a successful analysis of killing. This presumption might turn out to be mistaken. But in the absence of a good reason to think that killing and harming should turn out to have structurally different analyses, the fact that a proposed analysis of harming would generate an implausible parallel analysis of killing should count at least somewhat in favor of rejecting the proposed analysis of harming. And the Non-Comparative Account of harm has precisely this feature.

To see that this is so, consider what the analyses of killing would look like if they were parallel to the Non-Comparative Account and the Counterfactual Account. The Counterfactual Account treats harming as an essentially comparative matter: to harm a person is to put that person in a state that is worse for him in terms of how much well-being he has than the state that he would otherwise be in. A parallel analysis of killing would therefore be comparative in a parallel manner: to kill a person is to put that person in a state that is worse for him in terms of how alive he is than the state that he would otherwise be in. The Non-Comparative Account, on the other hand, is fundamentally non-comparative. It says that some states are simply bad states for people to be in, and that an act that causes a person to be in that state harms that person, regardless of what would have been the case had the act not been performed. A parallel analysis of killing would therefore be non-comparative in a parallel manner: to kill a person is to cause it to be the case that the person is dead, regardless of what would have been the case had the act not been performed. I will call the first account the Counterfactual Account of killing and the second account the Non-Comparative Account of killing. The question, then, is whether there is a difference in terms of the plausibility of these two accounts.

The Counterfactual Account of killing says that to kill a person is to cause him to be “less alive” than he would otherwise be. This certainly has a strange sound to it. But it sounds strange only if we take it to mean that being alive comes in degrees. Once we realize that the account does not require this to be the case, what the Counterfactual Account of killing says should seem perfectly straightforward: an act kills a person if it (p.254) causes that person to be dead when he would otherwise be alive. The Non-Comparative Account of killing, on the other hand, says that an act kills a person if it causes it to be the case that the person is dead, regardless of what would have been the case had the act not been performed. And this analysis has strongly counterintuitive implications in cases involving conception, at least if we follow Harman in assuming that “counterfactual dependence is sufficient for causation: I assume that if an action is such that if it had not been performed, a particular event would not have occurred, then the action causes the event.”12 For every death that occurs, after all, there is an act of conception such that if that act of conception had not occurred, the death would not have occurred. This means that the Non-Comparative Account of killing entails that the act of conceiving a person kills that person. And this strikes me as an implausible result. Since the Counterfactual Account of killing is therefore more plausible than the Non-Comparative Account of killing, and since there is a reasonable presumption in favor of there being a structural symmetry between the best analyses of harming and of killing, we have a good independent reason to prefer the Counterfactual Account of harm over the Non-Comparative Account of harm.

(3) The moral principle that Harman defends in her 2004 paper entails that it would be morally impermissible for you to give the ice cream sundae to Sally. In order to avoid this kind of result, Harman introduces a modified version of the principle in her 2009 paper, appealing to what she refers to there as (4) rather than (4*). And referring to a case that is structurally identical to my ice cream sundae case, she responds to the worry that the revision to her earlier principle is ad hoc as follows:13

I think this case does establish that (4*) is false. But it helps to motivate the possibility that (4) is true while (4*) is false. What strikes us about the case is that if we don’t harm Sally, then Sally loses out on the benefit. What moves us to think it may be permissible to harm Sally is the thought of what is bad for her about our not harming her. This brings out an important difference between actions that affect whether people exist and those that do not. If an action benefits someone who exists independently of the action, then there are two kinds of considerations in favor of the action:

– the action benefits the person

– if the action is not performed, the person is deprived of the benefits (that is, she exists but doesn’t get some benefits she could have had)

When a non-identity action benefits someone, only the first of these two considerations is available to tell in favor of the action; the second does not apply. But the second seems to play a crucial role in how, sometimes, the fact that we benefit someone can play a role in making harming that person permissible—and it plays a crucial role in why (4*) is false. Since the second consideration doesn’t apply to non-identity actions, this suggests that the reasons (4*) is false just don’t apply to (4).

Harman’s reply, in short, is that restricting the scope of (4*) is independently justified by the difference between cases like that of the ice cream sundae and cases like that of (p.255) Wilma. If you cause Sally to have a stomach ache by giving her the sundae, there are two kinds of consideration that will favor your act: she will benefit from the sundae if you give it to her and she will be deprived of it if you don’t. But if Wilma causes Pebbles to be blind by conceiving her, there will only be one kind of consideration in favor of her act: Pebbles will benefit from the good experiences her existence will make possible if Wilma conceives her, but she won’t be deprived of those experiences, or of anything else, if Wilma doesn’t conceive her.

But this response to the objection only pushes the question back a step. Rather than asking why the difference between (4*) and (4) is morally relevant, that is, the question now becomes why the difference between cases where both kinds of consideration favor a harmful act and cases where only the first kind of consideration favors a harmful act is morally relevant. Harman’s response is to claim that in the kind of case where both considerations are present, it is the second that is actually doing the work in accounting for the permissibility of the harmful act. Put in terms of the ice cream sundae case, she says that “what strikes us” about the case is that Sally will lose out on the ice cream if we don’t give it to her, not that Sally will greatly enjoy the ice cream if we do give it to her. She says that “what moves us to think it may be permissible” to give Sally the ice cream is “the thought of what is bad for her about our not” giving it to Sally, rather than the thought of what is good for her about our giving it to her. And she says that the fact that Sally would be deprived of a benefit if we don’t give her the ice cream “seems to play a crucial role” in making the act of giving her the sundae permissible. If Harman is correct about all of this, then there is indeed a reason to treat the difference between the two kinds of cases as morally relevant. If the second kind of consideration is needed in order to justify a harmful act, and if the second kind of consideration applies in ordinary cases but not in non-identity cases, then there is an independent justification for limiting the scope of the Alternative Possibility Principle in the way that (4) does so that it only covers non-identity cases.

But why should we think that Harman is correct about this? Why should we think that the benefit to Sally of enjoying the ice cream is not enough, by itself, to make it permissible for us to give it to her? Harman gives no reason for us to think this. She simply asserts that we do think it. But this is an unsatisfying response for three reasons. First, simply saying that people think this does not provide a justification for their thinking it. Why should the difference make a difference? Second, the claim that most people think this is implausible. Suppose, for example, that a mother takes her child on a roller coaster ride knowing that it will be exhilarating for him but also knowing that it will cause him to become a bit nauseated. If we asked the mother afterwards why she took the child on the ride, we would surely expect her to say something like this: “because I knew he would have such a great time on the roller coaster and that the benefits to him would far outweigh the costs.” It would be quite surprising if she instead said something like this: “because I knew it would be very bad for him to not go on the roller coaster, much worse for him than suffering a bit of nausea.” Rather than thinking that people would primarily appeal to the second sort of consideration in cases where (p.256) both considerations apply, then, it seems much more likely that most people would naturally appeal to the first.

But finally, and most importantly, there is no reason to treat the two considerations in the ice cream sundae case as if only one of them could provide a sufficient justification for the permissibility of performing the harmful act. Even if someone really is initially moved to think that it would be permissible to give Sally the ice cream by considering how bad it would be for Sally if she were deprived of the ice cream, this would provide no reason for them to doubt that the fact that Sally would benefit from getting the ice cream would also be a sufficient justification for the claim that it would be permissible to give it to her. There is no reason, that is, to deny that the ice cream sundae case is simply one in which the permissibility of the act is overdetermined. Suppose, to take a different kind of example, that I promise you that I will not murder Sally. In this case, you might well be moved to think that it would be impermissible for me to murder Sally by considering how bad it would be for Sally if I were to murder her. The fact that I promised you that I would not murder Sally would likely play no role in your thinking. But this would do nothing to undermine the claim that the fact that an act would break a promise I made to you could also be a sufficient reason to deem it impermissible.

Appendix D: One More Case to Test the Preventative Benefit Principle (3.3.2)

In the case where Richie’s dropping the gold cube on Joe confers a pure benefit on Joe, the harms and benefits are strikingly incommensurable: a broken arm on the one hand, a great deal of money on the other. So we can change the case where Larry’s rescuing Moe confers a preventative benefit on Moe by making it one in which the harms and benefits are equally incommensurable. Here is an example:

Drowning Money: Moe is asleep next to a large open box that contains a great deal of his money. A strong wind starts blowing, and the money will soon be blown into a nearby lake, where it will be ruined. Larry is unable to close the box, but Moe will be able to close it if Larry wakes Moe up. In order to wake Moe up, Larry will have to break Moe’s arm. Larry wakes Moe up. As a result, Moe is spared the loss of a great deal of his money and suffers a broken arm.

In this case, as in the original drowning case, Larry confers a preventative benefit on Moe. But in this case, unlike the original drowning case, the harm that Larry causes is not at all commensurable with the benefit he confers.

Virtually everyone, I assume, agrees that it is permissible for Larry to break Moe’s arm in the case where Moe himself is drowning. What about the case where it is Moe’s money that will drown if Larry doesn’t break Moe’s arm? I suspect that a number of people will think that it would be permissible for Larry to break Moe’s arm in this case as well. But that is not enough to help vindicate the Preventative Benefit Principle. After all, a number of people will probably think that it is permissible for Richie to break (p.257) Joe’s arm in the original gold cube case, too. The significance of changing the drowning case into the drowning money case rests on the answer to this question: to whatever extent you think the original drowning case is different from the gold cube case in terms of the permissibility of the acts in question, does changing the drowning case to the drowning money case eliminate that disparity? If you think that there is no difference in permissibility between the original drowning case and the gold cube case, after all, then Shiffrin’s argument won’t be able to get off the ground in the first place: there will be no difference in permissibility for either the difference in commensurability or the difference between pure and preventative benefits to be accounting for.

Suppose, though, that you think it’s just obvious that it is permissible for Larry to break Moe’s arm in the original drowning case but that while you are inclined to think that it is probably permissible for Richie to break Joe’s arm in the gold cube case, you don’t really think it’s just obvious that it would be permissible. In that case, you should consider whether this relatively small discrepancy between the two cases is eliminated by changing the drowning case to the drowning money case. If the change leaves you thinking that it is probably permissible for Larry to break Moe’s arm in the drowning money case but less obviously so than in the original drowning case, then the change in the drowning story again shows that it was the difference in terms of commensurability, not the difference in terms of pure and preventative benefits, that was making the original cases seem somewhat different to you. If you think it is clear that it is permissible for Larry to break Moe’s arm in the original drowning case and also clear that it is positively impermissible for Richie to break Joe’s arm in the gold cube case, then you should consider whether this large disparity is eliminated by changing the drowning case to the drowning money case. If the change leaves you thinking that it is clearly impermissible for Larry to break Moe’s arm in the drowning money case, then the change from the original drowning story again shows that it was the difference in terms of commensurability, not the difference in terms of pure and preventative benefits, that was making the original cases seem so different to you.

The argument at this point must rest on the particular reactions that it generates, but I find it hard to believe that many people will think that there is a significant difference between the permissibility of Larry’s breaking Moe’s arm in the drowning money case and Richie’s breaking Joe’s arm in the original gold cube case. And if that’s right, then this second way of testing my claim should produce the same result as the first: Shiffrin’s cases fail to provide sufficient warrant for endorsing the Preventative Benefit Principle.

Appendix E: Rejecting Five Objections to the Rights Waiver Argument (5.1.3)

The rights waiver argument maintains that even if Pebbles has a right that Wilma’s act would directly or indirectly violate, the existence of the right cannot be used to ground (p.258) a satisfactory solution to the non-identity problem because Pebbles would waive that right at the time of Wilma’s act if she could or because she will later waive the right when she comes to understand the choice that Wilma faced. Several objections have been raised against the rights waiver argument, but none of them is successful.

Jeffrey Reiman, for example, argues that

the choices in the non-identity cases are not analogous to the choice to perform life-saving surgery on an unconscious person. Consider that when we assume that the unconscious person would waive his right not to have surgery performed on him without his consent, we assume that the waiver releases us from wrongdoing at the moment we made the choice to operate. Such a waiver is a retroactive permission, not an after-the-fact pardon. It is tantamount to saying that had the person been conscious at the moment when the choice to operate had been made, he would then have consented to surgery. No such retroactive consent is possible, however, in the case of the people negatively affected by the choices made in the non-identity cases, because they did not exist when those choices were made. Since particular individuals only care about existing as those particulars once they already are alive, the argument that they would waive their rights fallaciously assumes that particular people have an interest before they are alive that they only can have once they are alive.14

There are three problems with this objection. First, the fact that we take the waiver in the surgery case to be effective at the moment that we act does nothing to favor a hypothetical present consent account of waivers over a subsequent actual consent account. When a parent has a child vaccinated, for example, she could appeal to the claim that the child would consent right now if he understood the relevant details, but she could just as easily appeal to the claim that the child will later come to consent once he understands the relevant details. At the time that she acts, it is just as true that the child will later consent as it is that the child would at that time consent. The fact that the waiver is operative at the time that she acts therefore does nothing to favor appealing to the latter consideration over the former. And if subsequent actual consent is sufficient to generate a valid rights waiver, then the fact that the unconscious patient already exists when the surgery is performed but the not-yet-conceived Pebbles does not exist when Wilma conceives her does nothing to undermine the rights waiver argument. The fact that Pebbles will later exist and consent to what Wilma has done is enough to make Wilma’s act permissible at the time that she does it.

But suppose that Reiman is correct in maintaining that a waiver of rights is valid only in cases where the person would actually consent to it at the time that the relevant act was performed. And suppose that he is also correct in maintaining that people have no interest in living until they already exist. If this is so, then a second problem arises. If Pebbles has no interest in her existence prior to the point at which she is conceived, then she presumably has no rights relating to her existence prior to that point either. But if Pebbles has no rights relating to her existence prior to the point at which she (p.259) exists, then Wilma can appeal to the waiver of Pebbles’s rights to justify conceiving her even if only hypothetical actual consent is good enough to render a waiver of rights valid. Before she exists, Pebbles has no rights that could be violated by Wilma’s act. As soon as she exists, she may have such rights, but as soon as she exists Wilma can appeal to the fact that she would consent to waive those rights given that she now exists. And so there is no point in time either before Wilma’s act or after it at which Pebbles has a right that she would not waive and that is violated by Wilma’s act.

Finally, Reiman’s objection rests on the claim that pre-existing Pebbles would waive whatever right Wilma’s act might be thought to violate only if she had a positive interest in being brought into existence. But this claim, too, can be challenged. Consider, for example, the case of trespassing. If you are unconscious and bleeding to death on your living room floor, it seems reasonable to suppose that you would waive your right against my trespassing so that I could walk into your house and save your life. If you are peacefully sleeping in your bed, it seems reasonable to suppose that you would not waive your right against my trespassing so that I could walk into your house and kill you. You would waive your right against my trespassing in cases where doing so would serve your interests, that is, but not in cases where doing so would hinder them. But what about cases in which you are completely indifferent to my trespassing? Suppose, for example, that I am walking home and that it would be a bit more convenient for me to take a few extremely light and gentle steps across your front lawn. Suppose that any harm that my act might cause to your interests would be precisely offset by corresponding benefits. If I very slightly damage a few blades of grass, for example, then I also help to kill a few weeds that were competing with the grass for nutrients. And suppose that there was no other reason that you would not want me to walk across a small part of your lawn, so that you are truly and completely indifferent between my briefly trespassing and my not trespassing at all. In this case, it seems reasonable to suppose that you would not object to my trespassing on your property and so would waive your right against my doing so. But if this is correct, then the fact that Pebbles has no positive interest in being brought into existence—if this is a fact—does not mean that Pebbles would not waive her right against Wilma conceiving her. She would still waive that right since she would have no reason to object to Wilma’s conceiving her, just as you would waive your right since you would have no reason to object to my briefly walking across your lawn. And this provides one final reason to reject Reiman’s first objection to the rights waiver argument.

Reiman has a second objection to the rights waiver argument. It runs as follows:

[I]‌t is unjust to make the price of some particular individual’s existing that that person waive his or her right not to be negatively affected. Then it is clear, however, that such a waiver would be obtained under duress: that is, by an unjust albeit retroactive threat, namely, the threat of not existing. Parfit is certainly right to think that the people produced as a result of the negative actions in the non-identity cases might be tempted to waive their rights since the alternative is not existing at all. A person who waives a right in the face of an unjust threat of not existing, (p.260) however, does not make a free and morally binding choice. Thus, even if the future people in the non-identity cases would waive their rights, that waiver would not be valid.15

The foundation of Reiman’s second objection seems reasonable. Suppose, for example, that I tell you that I will kill you unless you waive your right not to have your eyes cut out and that, as a result of my telling you this, you explicitly waive your right not to have your eyes cut out. In this case, it seems clear that your rights waiver would not be valid and that my act of cutting your eyes out would be morally wrong despite the fact that you had explicitly waived your right against my doing so. Or suppose that I cut your eyes out while you are under heavy sedation and that when you later confront me about it, I defend my act by saying that I would have killed you if I hadn’t cut your eyes out and that I knew that you would have agreed to waive your right not to have your eyes cut out given that the alternative was my killing you. Again, this rights waiver would not be valid. So Reiman is presumably right to say that a rights waiver made in the face of an unjust threat is not binding.

The case in which I will either cut your eyes out or kill you, moreover, is in one important respect similar to the case of Wilma and Pebbles. After I make my choice, either you will exist as a blind person or you will not exist at all. After Wilma makes her choice, either Pebbles will exist as a blind person or Pebbles will not exist at all. If the fact that you would waive your right not to be deprived of vision given the choices available to you is not enough to make your rights waiver a valid one, then, it may well seem that the fact that Pebbles would waive her right not to be deprived of vision given the choices available to her is not enough to make her rights waiver a valid one either. And if her rights waiver would not be a valid one, then Reiman’s second objection to the rights waiver argument must be accepted.

But now suppose that a diabolical villain has injected a lethal poison into your eyes, that you are fully conscious, that I tell you that you will die unless you let me cut your eyes out and that, as a result of my telling you this, you explicitly waive your right not to have your eyes cut out. In this case, it seems clear that your rights waiver would be valid and that my act of cutting your eyes out would not be morally wrong. It seems clear that my act would not be morally wrong, moreover, precisely because you had explicitly waived your right against having your eyes cut out. And your rights waiver would also seem to be perfectly valid in the case where you are under heavy sedation and I appeal to the fact that you would waive your right not to have your eyes cut out if you could or that you will waive it after the fact. These considerations, too, seem to show that my act would not be morally wrong under those circumstances.

Yet the diabolical villain case is like the case of Wilma and Pebbles in just the same way that the case in which I will either cut your eyes out or kill you is like the case of Wilma and Pebbles. After I make my choice in the case where a diabolical villain has injected poison into your eyes, either you will exist as a blind person or you will (p.261) not exist at all. After Wilma makes her choice in our original non-identity case, either Pebbles will exist as a blind person or Pebbles will not exist at all. And the fact that you would waive your right not to be deprived of vision given the choices available to you in the diabolical villain case really does seem to be enough to make your rights waiver a valid one in that case. In order to sustain Reiman’s second objection to the rights waiver argument, then, it is not enough to show that the Wilma case is in one way like the case in which I will either cut your eyes out or kill you. The objection would have to show that the Wilma case is more like that case than it is like the case of the diabolical villain. But in the end, it is the diabolical villain case that proves more like the case of Wilma. And this is why, in the end, Reiman’s second objection must also be rejected.

In order to see this, it is important first to be clear about the main difference between the case in which I will either cut your eyes out or kill you and the case in which the diabolical villain has injected the poison into your eyes. In both cases, I tell you that you will die if I don’t cut your eyes out. But in the former case, it is my fault that these are your only two options. In the latter case, it is not my fault. Prior to my arriving on the scene in the first case, there was a third option available to you: living and not being blind. This third option would have been better for you. I took that option away from you by threatening to kill you if you did not let me cut your eyes out. Removing this third option thus harmed you. And I had no right to take this third option away from you. Removing this third option thus wrongfully harmed you. But prior to my arriving on the scene in the second case, living and not being blind was not an option for you. Because the diabolical villain had already injected the poison into your eyes, living as a blind person or dying were the only two options you had when I arrived on the scene. When I tell you in that case that you will die if I don’t cut your eyes out, then, I am simply stating the facts as they exist independent of my actions. My doing so neither harms you nor wrongs you.

This difference between the two cases explains why your rights waiver is invalid in the first case but valid in the second. When you agree to let me cut your eyes out in the first case, you do this only because I have made this the best option available to you by unjustly eliminating a third option that would have been even better. When you agree to let me cut your eyes out in the second case, you do this because this is the best option available to you regardless of what I do and regardless of what I have done. I unjustly coerce you into agreeing to waive your rights in the first case, then, but I do not unjustly coerce you into agreeing to waive your rights in the second case. This is why your rights waiver is valid in the case of the diabolical villain but not in the case in which I threaten to kill you if you don’t let me cut your eyes out.

But given this understanding of the difference between the two cases, we can now see that the case of Wilma is like the case of the diabolical villain and unlike the case in which I threaten to kill you if you don’t let me cut your eyes out. Prior to Wilma’s making her decision about when to conceive, that is, living and not being blind was not an available option for Pebbles. Because of a medical condition that was not Wilma’s fault and basic facts about human reproduction that were beyond Wilma’s control, the only (p.262) two options available to Pebbles were being conceived in an incurably blind condition and never being conceived at all. To put things in the terms that Reiman uses in pressing his objection, Wilma does not make it the price of Pebbles’s existing that she waive her right not to be conceived as a blind person. That price was set by factors that were beyond Wilma’s control. This makes the case of Wilma like the case in which the diabolical villain has injected poison into your eyes and unlike the case in which I threaten to kill you if you don’t let me cut your eyes out. I do not make it the price of your continued existence that you let me cut your eyes out in the former case. I simply recognize that this is the price given the facts as they are. But I do make it the price of your continued existence that you let me cut your eyes out in the latter case, since in that case it is only my unjust threat to kill you that makes losing your eyes the price of existence for you. Since your rights waiver is valid in the case of the diabolical villain, and since Pebbles’s rights waiver in the Wilma case is morally on a par with your rights waiver in the case of the diabolical villain, Pebbles’s rights waiver in the Wilma case is valid as well. Reiman’s second objection to the rights waiver argument, then, like his first, is ultimately unsuccessful.

David Velleman has also raised two objections to the rights waiver argument. Both are presented in terms of the case of a fourteen-year-old girl who conceives a child and who, because she is so young, is unable to give her child as good a start in life as she would have given her child if she had waited to conceive until she was older. Velleman argues that this girl’s act of conceiving at such a young age violates her child’s right to be given the best start in life that that child’s parents are capable of providing. And he presents two objections to the claim that the girl’s act is rendered morally acceptable by the fact that the child would or will waive that right. I will discuss Velleman’s two objections to the rights waiver argument in terms of the particular right that he endorses, but the points I raise in response to his objections would apply to other rights as well.

Velleman’s first objection is similar to Reiman’s second objection:

Ordinarily, the prospect of waiving a right arises in the context of three possible outcomes. We can (1) retain the right in order to ensure either (a) that it will be fulfilled or (b) that we will have legitimate grounds to protest its nonfulfillment; or we can (2) waive the right. Entertaining all three outcomes, we may prefer to retain the right, even though we would prefer to waive it if outcome (1)(a) were excluded. That is, we may think that retaining the right for the sake of possibly having it fulfilled would be sensible, but that retaining it merely for the sake of having grounds for protest would be petty and foolish. Given our preferences, the party against whom we hold the right can induce us to waive it if he can manage to take outcome (1)(a) off the table. But surely a waiver obtained by such means would not be normatively valid. He cannot gain release from fulfilling our right by confronting us with the fact that he isn’t going to fulfill it, so that our only alternative to waiving the right is to retain it for the petty purpose of lodging a protest. To be sure, the child of a fourteen-year-old mother cannot exactly claim that she has taken outcome (1)(a) off the table: it was never on the table for this particular child. And yet the child may still waive his birthright because his only alternative is to complain that it cannot be fulfilled. And such a waiver is granted less voluntarily, because it is granted in the presence of fewer (p.263) relevant alternatives, than the waiver of a right that can still be fulfilled. Its validity is therefore questionable.16

If you try to waive a given right only because that right will not be fulfilled no matter what happens, in short, then the waiver of your right will not be valid. The fourteen-year-old girl’s child will not have his right to the best start in life that his parents could provide fulfilled no matter what happens. And so the child’s waiver of this right will not be valid.

There are two reasons to reject this objection. The first is that it rests on a mistaken view of the nature of the right that the child could have against his mother. Velleman describes the right as a right to the best start in life that the particular parents of a particular child could provide. This right can be understood in two ways. One is as a positive right. On this interpretation, you have the right to be given the best start in life that your parents can provide you. The other is as a negative right. On this interpretation, you have the right not to be given a start in life that is less than the best that your parents can provide you. Velleman’s objection to the rights waiver argument depends on the claim that the girl’s child has this right in the positive sense. This is because the objection depends on the claim that fulfilling the right is not a possibility in the case of this particular child and because this is true in the positive sense of the right but false in the negative sense of the right. In the positive sense of the right, it is true that there is nothing the girl can do that will fulfill the child’s right. If she conceives the child, he will not get the best start in life that she can provide, and if she doesn’t conceive the child, he will not get the best start in life that she can provide, or indeed any start in life at all. But in the negative sense of the right, there is something that the girl can do to fulfill the child’s right. If she refrains from conceiving this particular child, then this particular child will not be given a start in life that is less than the best that she would be able to provide because he will not be given any start in life at all. If the right is simply the negative right not to be conceived unless you will be given the best start in life that your parents can provide, that is, then the claim that Velleman’s objection rests on—the claim that there is no way for the girl to fulfill the right—is false. She could fulfill this right in the negative sense by not conceiving until she is older.17

(p.264) So Velleman’s first objection can succeed only if we attribute the right to this girl’s child in the positive sense. But even if children do have a right to the best start in life that their parents could provide, they can have this right only in the negative sense. This is because if a child has a particular right against his mother, then the mother has a corresponding obligation not to violate that right. If she has an obligation not to violate that right, then she ought to refrain from violating it. But if she ought to refrain from violating it, then it must be the case that she can refrain from violating it. If it is impossible for her to refrain from violating it, then it cannot be the case that she ought to refrain, and if it is not the case that she ought to refrain from violating it, then it is not the case that it is a right that the child has in the first place. But the fourteen-year-old girl cannot refrain from violating her child’s right to be given the best start in life that she can provide on the positive interpretation of that right. No matter what she does, this child will not be given such a start in life. Therefore, the child does not have this right in the positive sense.

My argument for this conclusion requires only two uncontroversial claims: that rights entail correlative duties and that ought implies can. Indeed, the claim that ought implies can is enough by itself to undermine Velleman’s objection here. Even if we were to allow that the girl’s child really can have a right that it is impossible for the girl not to violate, the fact that it would be impossible for the girl not to violate it would still be enough to show that it can’t be the case that she ought not to violate it. And if it is not the case that she ought not to violate it, then the fact that conceiving the child violates his rights—even if it is a fact—won’t be enough to justify the claim that conceiving the child is something she ought not to do. I’m inclined to favor the view on which the impossibility of fulfilling the purported right means that there is no such right. But either way, the appeal to such a right cannot ground a successful objection to the rights waver argument. And either way, the problem with the objection will arise in the case of any purported right that might be violated in non-identity cases. Pebbles cannot have a right to be born with vision since it is impossible for Wilma to refrain from violating that right, for example, but Pebbles can have a right not to be born without vision, since Wilma can refrain from violating that right by refraining from conceiving Pebbles. The future members of the wealthy society cannot have a right to be born and then not killed by the leaking toxic waste since it is impossible for them to be conceived and for the toxic waste not to leak, but they can have a right not to be born at all if they can only be born into a world in which toxic waste will subsequently leak and kill them. But if these are the rights that Pebbles and the future members of the wealthy society can have, then Velleman’s objection to the validity of the waiver of their rights will again be unsuccessful. In the only sense in which they can have the right in question, their waiver of the right will prove valid because they would waive it or will waive it in a context in which the right really could have been fulfilled.

But let’s suppose that Velleman is correct and that the only options available to the child in non-identity cases are waiving the right in question or retaining it solely in order to have legitimate grounds to protest its nonfulfillment. Even if this is so, there is (p.265) a second problem with Velleman’s first objection. Velleman claims that under such circumstances, “such a waiver is granted less voluntarily, because it is granted in the presence of fewer relevant alternatives, than the waiver of a right that can still be fulfilled.” But the fact that a person chooses between fewer options rather than more options does not make the choice he makes any less voluntary or the waiver of rights that results from his choice any less valid. If I buy a car from you when I have only two cars to choose from, the waiver of my right to the money I give you is just as voluntary and just as valid as it is if I buy a car from you when I have hundreds of cars to choose from.

Perhaps Velleman thinks the second option in the non-identity case is not a genuine option because “retaining [the right] merely for the sake of having grounds for protest would be petty and foolish.” This would mean that the fourteen-year-old girl’s child has only one real option, which really would make his “choice” involuntary. But this assertion is unjustified. Suppose a thief announces that he is going to take your money no matter what you do and that once he takes it, you will never get it back. Before taking it, though, he asks if you would voluntarily waive your right to the money. I suspect that most people would refuse to waive the right to their money if they found themselves in such circumstances and that they would think their reasons for doing so neither petty nor foolish. Even if retaining the grounds for a legitimate protest would not have any practical consequences, a fundamental matter of principle would be at stake. It is wrong for the thief to take your money and by waiving the right to it you would in effect be endorsing his behavior and rendering it permissible. If the fourteen-year-old girl’s child would waive his right to the best start in life his mother could provide even though the only alternative to waiving it was retaining it just so that he could have a moral complaint about his situation, then, this would still be enough to make the waiver of his right voluntary and valid, and thus enough to sustain the rights waiver argument.

Velleman’s second objection to the rights waiver argument is more fundamental. He denies that the child in question would waive his rights in the first place. Parfit at times motivates the rights waiver argument by appealing to the claim that if he is glad that someone did something to him, then he waives or would waive his right that the person not do that thing.18 But Velleman rejects this inference:

The fact that a child would be glad to have been born does not entail that he would excuse his mother from her procreative obligations. He can reasonably say to his mother, “I’m glad that I was born, but you were wrong to have a child in my case.” Not only can he reasonably say this; he probably will say it, once he realizes that other children have been given, and sensibly regard themselves as entitled to, the best start in life that their parents could provide to a child. He will continue to assert his birthright, despite being glad that he was born.19

(p.266) If being glad that he was conceived does not mean that he would waive the right that was violated by conceiving him, then the rights waiver argument seems to be cut off at its foundation.

Velleman is certainly right to insist that a person can be glad that he was conceived while believing that it was wrong for him to have been conceived. A person who was conceived in an act of rape, for example, could be glad to be alive while recognizing that the act of conceiving him was morally wrong. And, more generally, a person can be glad that an act was performed while believing that it was wrong for the act to have been performed. If I wake up and discover that you performed life-saving surgery on me last night, for example, I will be glad that you did so. If I discover that the surgery involved your killing innocent people and removing their organs in order to save me, I will think that your act was morally wrong but I could still feel glad that I was the person who was saved by your wrongful behavior given that someone was saved by your wrongful behavior. Velleman is therefore right to say that the fact that a child would be glad to have been conceived does not entail that the child would excuse his mother from her moral obligations. And so the mere fact that the fourteen-year-old girl’s child will be glad to be born if she conceives him does not mean that he would waive whatever right her act of conceiving him would otherwise violate.

But there is an important difference between the rape and surgery cases and the case of the fourteen-year-old girl. In the former cases, the act in question is wrong for reasons that are independent of their effect on the person whose existence depends on the act. It is wrong to rape a person in order to conceive a person and wrong to kill people in order to save a person regardless of whether the person who ends up existing as a result of the wrongful act turns out to be glad that the act happened, and it is wrong because the act wrongs people other than person who now exists as a result of the wrongful act. This is why it can be coherent for the person who now exists as a result of the act in those cases to be glad that the act happened while agreeing that it was wrong that it happened. The act was wrong because it wronged other people. But the fourteen-year-old girl’s act of conceiving her child did not wrong any other people. The rights-based solution claims that her act was nonetheless wrong because it wronged her child and that it wronged her child by violating her child’s rights. Since her act did not wrong anyone else, on this account, it will be wrong only if it violates her child’s rights, and it will violate his rights only if he does not or would not waive them. But if, once he is alive, the child will be glad that his mother conceived him, then the only reason he could have for not waiving the right in question would be if he thought that doing so would amount to tacitly approving of behavior that was immoral for other reasons. This can be so in the rape and surgery cases because those acts are wrong for other reasons. But if the only reason that conceiving the child would be wrong would be because it would violate his rights, and if he will be glad that he was conceived, then he will have no moral or prudential reason not to waive the right in question. Waiving his right will not, as Velleman puts it, excuse his mother from her procreative obligations, because waiving his right will make it the case that her act violated no such (p.267) obligations. Her only obligation in this case was not to violate any unwaived rights and by waiving whatever right her act would otherwise violate he will ensure that she satisfies her obligation. Since he has no moral or prudential reason not to waive his right, he will waive it. And since he will waive it, her act will not be wrong. The fact that a person is glad that someone did a particular act cannot be enough to prevent the act from being wrong as a general matter because it cannot be enough to prevent the act from being wrong in cases where it wrongs other people. But the fact that a person is glad that someone did a particular act can be enough to prevent the act from being wrong in cases where there would be no reason for the act to be wrong other than that person’s not wanting it to have occurred and where the person will therefore happily waive the right in question as a result. In the end, then, Velleman’s second objection to the rights waiver argument is also unsuccessful.

A final objection to the rights waiver argument is suggested, though only tentatively endorsed, by Parfit himself. Referring to a case in which if Jane conceives now rather than later, her child will be born with a life-shortening condition, Parfit writes that if he were the child that Jane conceived with the life-shortening condition, “I would not regret that my mother caused me to exist. And I would deny that her act was wrong because of what it did to me. If I was told that it was wrong, because it caused me to exist with a right that cannot be fulfilled, I would waive this right.” But he then adds: “though I would waive this right, I cannot be certain that, in all such cases, this is what such a child would do. If Jane’s child does not waive his right, an appeal to this right may perhaps provide some objection to her choice.”20

In order to see whether this possibility points to a genuine problem with the rights waiver argument, it may help to begin by assigning specific probabilities to the case. So let’s suppose that Jane knows that if she conceives a child now, there is a one in twenty chance that the child will grow up to resent her decision and to refuse to waive his right not to have been conceived in such a condition. And now consider the following case: you are a doctor and I am brought unconscious into your emergency room. You can save my life only by performing surgery on me, but even if you do, I will be permanently paralyzed below the waist and significantly disfigured. You must decide what to do before I wake up and thus before I can have an opportunity to tell you whether I waive my right not to have you cut into my body. Based on your experience in similar cases, you are confident that I would want you to save my life, but you also recognize that there is a one in twenty chance that I would rather die than live on in such a condition. If that turns out to be the case, then when I wake up, I will resent your decision and refuse to waive my right not to have you cut into my body.

It is clear that it would not be morally wrong for you to operate on me in this case. The mere fact that there is a modest chance that I will later refuse to waive my right not to be operated on is therefore not enough to make operating on me wrong. But given (p.268) that this is so, we must say the same thing about the case of Jane. The mere fact that there is a modest chance that her child will later refuse to waive his right not to have been conceived is not enough to make conceiving him wrong. In order for the rights waiver argument to succeed, then, it is not necessary to insist with certainty that the fourteen-year-old girl’s child will later waive his rights. It is enough to maintain that it is highly likely that he will. And there is no good reason to doubt this assumption.

Appendix F: Rejecting Parfit’s Defense of the No Difference View (6.1.3)

Wanda is planning to conceive a child tonight. Her doctor tells her that as things now stand, the child she conceives tonight will be incurably blind. But if she takes a handful of tiny pills this afternoon, then when she conceives that child tonight, the child will not be blind. The pills are easy to take, have no side effects, and will be paid for by her health insurance. And since taking the pills this afternoon will have no effect on the timing of the conception that will occur tonight, she will conceive the same child tonight regardless of whether she takes the pills this afternoon. So either that child will be incurably blind or that very same child will be sighted. Like Wilma, Wanda decides that having to deal with a bunch of pills is a bit too much of a bother. So she tosses the pills away, goes home, and conceives Patty who is born incurably blind. Like Wilma, Wanda ends up conceiving a blind child rather than a sighted child when she could easily have conceived a sighted child rather than a blind child. But unlike Wilma, Wanda was going to conceive the same child regardless of which choice she made. Wanda made Patty worse off than Patty would have been if Wanda had taken the pills, that is, but Wilma did not make Pebbles worse off than Pebbles would have been if Wilma had taken the pills. Wanda’s act harmed someone. Wilma’s act did not. Does this difference between the two cases make a moral difference?

One view maintains that this difference does not make a moral difference. This is what Parfit calls the “No Difference View.”21 On the No Difference View, Wilma’s act is just as morally wrong as Wanda’s act is, and is morally wrong for just the same reason. On a second view, what Parfit calls the “Two-Tier View,” the difference between the two cases does make a moral difference. Parfit puts the second view this way: “Though we always have reasons not to cause future lives to be less worth living, these reasons would be weaker if, because these lives would be lived by different people, these acts would not be worse for any of these people.”22 On this second view, Wanda’s reasons for not causing her future child to be blind Patty rather than sighted Patty are stronger than Wilma’s reasons for not causing her future child to be blind Pebbles rather than sighted Rocks. On the first view, Wilma’s act is just as morally wrong as Wanda’s act. (p.269) On the second view, Wilma’s act is less morally wrong than Wanda’s act or perhaps not even morally wrong at all.

Parfit argues that we should accept the No Difference View. His argument for this claim is indirect. Rather than providing reasons that support the view, he attempts to establish that denying the view has unacceptable implications. Two cases, in particular, serve as the basis for this reductio ad absurdum argument. The first case involves our choosing between two medical programs.23 Program A would test millions of pregnant women for Disease A and cure the women who tested positive for it before their illness could cause their children to be born with a life-shortening condition. Program B would test millions of women for Disease B before they became pregnant and cure the women who tested positive for it before their illness could cause them to conceive a child who would be born with a similar life-shortening condition. The cure for Disease B would cause the women to postpone conception by two months. The women who currently have Disease A, then, will give birth to the very same children regardless of whether Program A cures their disease before they give birth. The women who currently have Disease B, though, will conceive and give birth to one set of children if Program B cures them before they conceive and will conceive and give birth to a different set of children if Program B does not cure them before they conceive. Program A and Program B would affect the same number of women and would prevent the same number of children from being born with a similar life-shortening condition.

One way to picture the difference between Program A and Program B is to look at a representative sample of the children who would be born under either program. Ann and Arnold are currently being gestated in mothers infected with Disease A. Ann and Arnold will thus be born in a few months regardless of whether their mothers are tested and cured by Program A. If their mothers are cured, they will be spared the life-shortening condition. Beth and Beverly are currently infected with Disease B and plan to become pregnant later this year. If Program B does not test and cure them, Beth will soon conceive Bob and Beverly will soon conceive Becky. Bob and Becky will both have the life-shortening condition. If Program B does test and cure Beth and Beverly, thus delaying their acts of conception by two months, they will instead conceive Bill and Barbara who will be spared the life-shortening condition. We can represent the number of years that each child would live under the two programs as follows:

Program A:

Ann 70

Arnold 70

Bob 50

Becky 50

Program B:

Ann 50

Arnold 50

Bill 70

Barbara 70

And we can simply assume that for every additional case like Ann and Arnold there would be an additional case like Bob or Bill and Becky or Barbara.

(p.270) Parfit finds that when he considers this sort of case, the two programs seem to be morally on a par. Neither would be better or worse than the other. The No Difference View entails that this judgment is correct. This is because the No Difference View says that the fact that Ann and Arnold will be better off under Program A and worse off under Program B while there is no one who will be better off under Program B and worse off under Program A does not matter morally. But the Two-Tier View entails that Program A is better than Program B. This is because the Two-Tier View says that our reason for adding twenty years of life to a particular person like Ann or Arnold is stronger than our reason for adding twenty years of life that will simply be lived by some person or other by having someone like Bill or Barbara live seventy years rather than having someone like Bob or Becky live fifty years. The argument that Parfit grounds in this first case, in short, is that the two programs are morally on a par, that the No Difference View accounts for this and that denying the No Difference View is incompatible with this.

Let’s suppose that Parfit is correct that Program A and Program B are morally on a par.24 He is certainly correct that the No Difference View entails that this judgment is true and that the Two-Tier View entails that it is false. But the Two-Tier View is not the only alternative to the No Difference View. The exclusive version of the Moderate Principle that I defend in section 6.1 is another alternative. Unlike the No Difference View, the exclusive version of the Moderate Principle entails that there is a moral difference between the cases of Wilma and Wanda. And unlike the Two-Tier View, the exclusive version of the Moderate Principle does not entail that there is a moral difference between Program A and Program B. The exclusive version of the Moderate Principle, therefore, defeats Parfit’s argument from the claim that the two medical programs are morally on a par to the conclusion that we should accept the No Difference View.

Consider first what the exclusive version of the Moderate Principle entails about the cases of Wanda and Wilma. Assuming that the total amounts of well-being that Wanda and her child would enjoy on each of the two options available to Wanda are the same as the total amounts of well-being that Wilma and her child would enjoy on each of the two options available to Wilma, we can appeal to the numbers that I appealed to in Chapter 6 and represent the two cases as follows:

(p.271)

Wilma

Don’t take pills first

Take pills first

Wilma: 40,000

Wilma: 39,999

Pebbles: 56,000

Rocks: 80,000

Wanda

Don’t take pills first

Take pills first

Wanda: 40,000

Wanda: 39,999

Patty: 56,000

Patty: 80,000

As I showed in section 6.1.1, the exclusive version of the Moderate Principle does not entail that Wilma’s act of conceiving Pebbles is morally wrong. Conceiving Pebbles does not make things worse for Wilma and Pebbles than they would otherwise be where the interests of Wilma and Pebbles are counted equally. What about the case of Wanda?

The exclusive version of the Moderate Principle tells us to focus our attention on the interests of all and only those people who actually exist given that Wanda decided not to take the pills before conceiving Patty. These people are Wanda, Patty, and everyone else in the world. Taking the interests of each of those people into account equally, the principle directs us to ask: would the interests of this particular group of people have been better served if Wanda had instead taken the pills before conceiving Patty? Since the interests of everyone else who exists are not affected by Wanda’s choice, we can answer this question by simply focusing on the interests of Wanda and Patty, and by asking whether their interests would have been better served if Wanda had taken the pills before conceiving Patty. If the answer to this question is no, then the principle generates no objection to Wanda’s choice. If the answer is yes, then the principle entails that Wanda’s act was morally wrong.

The answer to the question is yes. When Wanda conceives Patty without first taking the pills, Wanda enjoys 40,000 units of well-being and Patty enjoys 56,000 units of well-being, for a total of 96,000 units of well-being. If Wanda had instead taken the pills before conceiving Patty, then Wanda would have enjoyed 39,999 units of well-being and Patty would have enjoyed 80,000 units of well-being, for a total of 119,999 units of well-being. 119,999 units of well-being is greater than 96,000 units of well-being. The interests of Wanda and Patty combined, that is, where the interests of each are counted equally, would have been better served if Wanda had taken the pills before conceiving Patty. And so while the exclusive version of the Moderate Principle does not entail that Wilma’s act of conceiving blind Pebbles rather than sighted Rocks is morally wrong, it does entail that Wanda’s act of conceiving blind Patty rather than sighted Patty is morally wrong. The exclusive version of the Moderate Principle, that is, rejects the No Difference View.

Let’s now turn to the two medical programs. If we choose Program A, Ann and Arnold will live for seventy years, Bob and Becky will live for fifty years, and Bill and Barbara will never exist. If we choose Program B, Ann and Arnold will live for fifty years, Bob and Becky will never exist, and Bill and Barbara will exist and will live for seventy years. Parfit’s argument maintains that we must endorse the No Difference View because doing so is necessary in order to account for the judgment that the two programs are morally on a par. But the exclusive version of the Moderate Principle enables us to reject the No Difference View without rejecting this judgment.

Suppose, for example, that we choose Program B. If we choose Program B, then Ann and Arnold are each made worse off than they would otherwise have been. Since no one would be made worse off than they would otherwise be if we chose Program A, (p.272) the Two-Tier View entails that choosing Program A is morally better than choosing Program B. But the exclusive version of the Moderate Principle directs us to ask an importantly different question. It tells us to focus our attention on the interests of all and only those people who actually exist given that we chose Program B. These people are Ann, Arnold, Bill, Barbara, and everyone else in the world. Taking the interests of each of those people into account equally, the principle directs us to ask: would the interests of this particular group of people have been better served if we had instead chosen Program A? Since the interests of everyone else who exists are not affected by our choice, we can answer this question by simply focusing on the interests of Ann, Arnold, Bill, and Barbara, and by asking whether their interests would have been better served if we had instead chosen Program A. If the answer to this question is no, then the principle generates no objection to our choice. If the answer is yes, then the principle entails that our choice was morally wrong.

The answer to the question is no. When we choose Program B, Ann and Arnold each enjoy fifty years of life and Bill and Barbara each enjoy seventy years of life, for a total of 240 years of life. If we had instead chosen Program A, then Ann and Arnold would have each enjoyed seventy years of life, but Bill and Barbara would never have existed and so would have enjoyed no years of life at all. The total number of years of life enjoyed by Ann, Arnold, Bill, and Barbara would then have been only 140 years. 140 years is not greater than 240 years. And so the exclusive version of the Moderate Principle generates no objection to our choosing Program B. By the same reasoning, the principle also generates no objection to our choosing Program A. If we choose Program A, Ann and Arnold will live for seventy years and Bob and Becky will live for fifty years. In order to evaluate this choice, the exclusive version of the Moderate Principle therefore requires us to focus our attention on the interests of Ann, Arnold, Bob, and Becky, and to ask whether their interests would have been better served if we had instead chosen Program B. And the answer to this question is also no. When we choose Program A, Ann, Arnold, Bob, and Becky enjoy a total of 240 years of life. If we had instead chosen Program B, Ann and Arnold would each have enjoyed fifty years of life, and Bob and Becky would never have existed and so would have enjoyed no years of life at all. The total number of years of life enjoyed by Ann, Arnold, Bob, and Becky would then have been only one hundred years. And one hundred years is not greater than 240 years. So the exclusive version of the Moderate Principle generates no objection to our choosing Program A either.

The exclusive version of the Moderate Principle, then, unlike the Two-Tier View, does not entail that there is a moral difference between Program A and Program B. Parfit’s argument maintains that if there is no moral difference between Program A and Program B, then we should accept the No Difference View. But this is a mistake. If there is no moral difference between Program A and Program B, we can account for this by appealing to the exclusive version of the Moderate Principle. And if we accept the exclusive version of the Moderate Principle, then we must reject the No Difference (p.273) View. The claim that there is no moral difference between Program A and Program B, then, fails to support the No Difference View.

Parfit has more recently appealed to a second case in arguing for the No Difference View in On What Matters. This is what Parfit refers to simply as Case Six:25

We choose A

Adam lives for 70 years

Bernard lives for 40 years

———

———

We choose B

———

Bernard lives for 90 years

Charles lives for 10 years

———

We choose C

———

———

Charles lives for 50 years

David lives for 20 years

In this case, Parfit argues that even the most defensible version of the Two-Tier View entails that we ought to choose C.26 This is because while A would be worse for Bernard than B and B would be worse for Charles than C, C would be worse for no one than A or than B. But if we were to choose C rather than A, we could cause two people to live for only fifty years and twenty years when we could instead have caused two entirely different people to live for seventy years and forty years. It is virtually impossible to believe that we ought to choose C rather than A, far harder to believe than to believe that the two medical programs are not really morally on a par. And the No Difference View easily accounts for the claim that we ought to choose A rather than C. On the No Difference View, choosing C rather than A would be morally on a par with choosing that two people live for only fifty years and twenty years rather than that those very same two people live for seventy years and forty years. And that would clearly be wrong. For this reason, Case Six provides a potentially much stronger foundation for Parfit’s argument. If denying the No Difference View means believing that C really is morally better than A, then we should almost certainly accept the No Difference View.

But denying the No Difference View does not mean believing that C is morally better than A. Accepting the Two-Tier View means believing that C is morally better than A, but as I have already noted, denying the No Difference View does not require us to accept the Two-Tier View. We can instead account for the fact that C is not morally better than A by appealing to the exclusive version of the Moderate Principle. And the exclusive version of the Moderate Principle entails that we should reject the No (p.274) Difference View, not that we should accept it. We can therefore deny the No Difference View without having to insist that C is morally better than A.

To see what the exclusive version of the Moderate Principle entails about Case Six, it may help to start by supposing that we have chosen A and that Adam is now beginning his seventy-year life and Bernard is now beginning his forty-year life. Have we done something morally wrong by selecting A? The exclusive version of the Moderate Principle directs us to answer this question by focusing our attention on the interests of all and only those people who actually exist given that we chose A. These people are Adam, Bernard, and everyone else in the world. Taking the interests of each of those people into account equally, the exclusive version directs us to ask: would the interests of this particular group of people have been better served if we had instead selected B or C? On the assumption that the interests of everyone else who exists are not affected by our choice, we can answer this question by simply focusing on the interests of Adam and Bernard, and by asking whether their interests would have been better served if we had instead selected B or C. If the answer to this question is no, then the principle generates no objection to our choosing A. If the answer is yes, then the principle entails that our choice was morally wrong.

The answer to the question is no. When we choose A, Adam enjoys seventy years of life and Bernard enjoys forty years of life, for a total of 110 years of life between them. If we had instead selected B, then Bernard would have contributed ninety years of life to the total, but Adam would not have existed and so would not have contributed any years of life at all. The total number of years lived by Adam and Bernard would then have been ninety. Ninety years is not more than 110 years, and so selecting B instead of A would not have been better for Adam and Bernard where their interests are treated equally. If we had instead selected C, moreover, neither Adam nor Bernard would have existed and so neither would have contributed any years of life to the total lived between them. The total number of years lived by Adam and Bernard would have been zero. And zero years is also not more than 110 years. So the exclusive version of the Moderate Principle does not entail that it would be morally wrong for us to choose A.

For the same reasons that applied in the case of the two medical programs, of course, the principle also does not entail that it would be morally wrong for us to choose B or C. Choosing B will not make things worse for Bernard and Charles than they would be if we chose A or C, and choosing C will not make things worse for Charles and David than they would be if we chose A or B. The exclusive version of the Moderate Principle, in short, does not entail that any of the available choices would be morally wrong. But this is not a problem for the claim that the principle successfully undermines Parfit’s argument. Parfit’s argument maintains that we should accept the No Difference View so that we can avoid the conclusion that C would be better than A. And we can avoid that conclusion without accepting the No Difference View by accepting the exclusive version of the Moderate Principle regardless of what that principle implies about whether any of the choices would be morally wrong. Parfit concludes his discussion of the cases I have discussed here by saying that “these various cases show, I believe, that (p.275) we should reject the Two-Tier View and accept the No Difference View.”27 But even if they do show that we should reject the Two-Tier View, they do not show that we should accept the No Difference View. They provide no reason to reject the exclusive version of the Moderate Principle, and if we accept the exclusive version of the Moderate Principle, then we must reject the No Difference View.

Two objections might be raised against my response to Parfit’s argument for the No Difference View.28 The first objection concedes that my response may succeed as a response to the argument as Parfit states it, but maintains that Parfit’s argument can easily be modified to overcome the response. The second objection maintains that my response does not succeed even as a response to the argument as Parfit states it. The first objection focuses on Case Six. It acknowledges that the exclusive version of the Moderate Principle does not entail we ought to choose C rather than A, and thus acknowledges that we can reject the claim that C is better than A without accepting the No Difference View. But the objection maintains that Case Six can still be used to undermine the exclusive version of the Moderate Principle because the principle has a different unacceptable implication in that case. This is the implication that it would not be morally wrong for us to choose C rather than A. That claim, too, is likely to strike most people as highly implausible.

(p.276) This objection should be rejected because it is mistaken in claiming that the exclusive version of the Moderate Principle entails that it would not be morally wrong for us to choose C rather than A. The principle does not entail that choosing C rather than A would be morally wrong, but this is not the same as entailing that choosing C rather than A would not be morally wrong. Choosing C rather than A might still be morally wrong for some other reason. For example, choosing C rather than A might be wrong because it violates an optimizing principle on which it is wrong to make the suboptimal choice in different person cases. This optimizing principle is consistent with the exclusive version of the Moderate Principle. The Moderate Principle simply says that it is wrong to make one choice over another at least in cases where a certain set of conditions are satisfied. It leaves it open that there might be other conditions that can also make a choice wrong. And the optimizing principle is also consistent with rejecting the No Difference View. Nothing about the optimizing principle can be used to reject the claim that Wanda’s act is morally worse than Wilma’s act. And so the claim that it would be wrong to choose C rather than A in Parfit’s Case Six cannot be used to ground a modified version of his defense of the No Difference View.

The second objection claims that my response rests on a misinterpretation of Parfit’s argument. I have been characterizing Parfit’s argument as maintaining that if the two medical programs are morally on a par or if it is not the case that we ought to choose C in Case Six, then we should accept the No Difference View. As I pointed out, Parfit concludes his discussion of these cases in On What Matters by saying that “these various cases show, I believe, that we should reject the Two-Tier View and accept the No Difference View.” But, this objection points out, Parfit also prefaces his discussion of the cases by identifying and rejecting what he calls the “Narrow Person-Affecting View”: “one of two outcomes cannot be worse, nor can one of two acts be wrong, if this outcome or act would be worse for no one.”29 And it is only after setting the Narrow Person-Affecting View to the side that he says that “there are now two possibilities”: the No Difference View and the Two-Tier View.30 What Parfit’s argument therefore really maintains, according to the second objection, is that if the two medical programs are morally on a par (or if it is not the case that we ought to choose C in Case Six), and if we reject the Narrow Person-Affecting View, then we should accept the No Difference View. And, the objection maintains, my response cannot undermine this version of Parfit’s argument because my response appeals to the exclusive version of the Moderate Principle and the exclusive version of the Moderate Principle is simply a version of the Narrow Person-Affecting View. If Parfit’s argument is based on the assumption that the Narrow Person-Affecting View is false and if my response to Parfit’s argument depends on the assumption that the Narrow Person-Affecting View is true, then my response to Parfit’s argument, once that argument is properly understood, turns out to simply beg the question.

(p.277) The principle that I have appealed to in responding to Parfit’s argument is in some respects very close to the Narrow Person-Affecting View. It is therefore understandable that my response might seem to beg the question against Parfit’s argument. But the principle that I have appealed to is nonetheless distinct from the Narrow Person-Affecting View. And so, in the end, this second objection to my response to Parfit’s argument must also be rejected. The Narrow Person-Affecting View says that if one of two acts would not be worse for anyone, then the act cannot be wrong. Making things worse for someone, on this view, is a necessary condition in order for an act to be wrong. If one of two acts would not be worse for anyone, then the exclusive version of the Moderate Principle will generate no moral objection to it. Making things worse for someone, in other words, is necessary in order for an act to violate the exclusive version of the Moderate Principle. But the Moderate Principle does not maintain that there are no other moral principles. Nor does it maintain that there are no other moral principles on which an act might be wrong despite that fact that it is worse for no one. The Moderate Principle, for example, is consistent with the claim that an act can be wrong because it violates someone’s rights even if the act does not make things worse for anyone. This is one reason that one can accept the exclusive version of the Moderate Principle without accepting the Narrow Person-Affecting View.

Second, and more importantly, the Narrow Person-Affecting View claims that if one of two outcomes would be worse for no one, then it cannot be the worse outcome. The outcome in which Wilma conceives Pebbles rather than Rocks, for example, is not worse for Pebbles or for anyone else, and so the Narrow Person-Affecting View entails that the outcome is not worse when Wilma conceives Pebbles than it would have been if she had instead conceived Rocks. But the exclusive version of the Moderate Principle says nothing at all about which outcomes are better or worse. Accepting the principle therefore entails nothing about which outcomes are better or worse than others. One can accept the principle while agreeing that it would be better for Wilma to conceive Rocks. One would simply have to deny that the fact that it would be better for Wilma to conceive Rocks means that it would be positively immoral for her to conceive Pebbles. This is a second reason that one can accept the exclusive version of the Moderate Principle without accepting the Narrow Person-Affecting View. And since one can accept the principle without accepting the view, the fact that my response to Parfit’s argument appeals to the principle does not mean that my response requires us to accept the view.

There is one final interpretive wrinkle. Immediately after referring to the Narrow Person-Affecting View, Parfit writes:

Most of us would rightly reject this view. We would believe that (D) it would be in itself worse if some of the lives that will later be lived will be less worth living, and that (E) we have reasons not to act in ways that would have such effects, and if these effects would predictably be very bad, and we could avoid them at little cost to ourselves, such acts would be wrong.31

(p.278) And it is immediately after this passage that he says that we are now left with two possibilities: the No Difference View and the Two-Tier View. Parfit’s argument for the No Difference View might therefore be understood as saying not just that if the two medical programs are morally on a par (or if it is not the case that we ought to choose C in Case Six), and if we reject the Narrow Person-Affecting View, then we should accept the No Difference View, but rather that if the two medical programs are morally on a par (or if it is not the case that we ought to choose C in Case Six), and if we accept (D) and (E), then we should accept the No Difference View. These are not simply two ways of stating the same argument. One can reject the Narrow Person-Affecting View without accepting both (D) and (E). Accepting (D) and the first part of (E), for example, would be sufficient for rejecting the Narrow Person-Affecting View.

If Parfit’s argument presupposes not only that we reject the Narrow Person-Affecting View but also that we accept (D) and (E), then my response to the argument might seem to be subject to the second objection after all. The second part of (E) maintains that it would be wrong to produce an outcome that would be much worse than the outcome that would otherwise have been produced even if the outcome was worse for no one, and this claim might seem to be incompatible with the principle that I have appealed to in responding to Parfit’s argument. But the exclusive version of the Moderate Principle is not incompatible with the conditional claim made by (E). The Moderate Principle does not say that acts of the sort identified in the second part of (E) are not wrong. It simply says that they do not violate the Moderate Principle. We could therefore reject the Narrow Person-Affecting View and accept both (D) and (E) and Parfit’s claims about the two medical programs and Case Six while still appealing to the exclusive version of the Moderate Principle as a reason for rejecting the No Difference View. Wanda’s act would still have a wrong-making property that Wilma’s act lacked. The No Difference View would still be false.

Notes:

(1) This is a modified version of an example taken from Parfit (1984: 165–6).

(2) On some accounts of negation, there might prove to be an important difference between the claim “it is not the case that Pebbles is worse off existing than not existing” and the claim “Pebbles is not worse off existing than not existing.” If that is correct, then it does not follow from the claim that the former claim is coherent that the latter claim is as well. And this would mean that the coherence of the former claim would not be enough to establish the coherence of P1’s claim that Wilma’s act has not made Pebbles worse off than she would otherwise have been. But even if this is so, it poses no problem for my response to the incoherence objection. We could simply revise P1 to read: “It is not the case that Wilma’s act makes Pebbles worse off than she would otherwise have been” and revise P2 to read: “If A’s act harms B, then it is the case that A’s act makes B worse off than B would otherwise have been.” An objection to my response to the incoherence objection that depends on this distinction would therefore fail to be sufficiently robust to ground a satisfactory solution to the non-identity problem.

(3) This appendix is a modified version of Boonin (2012).

(4) Benatar (2006: 31 n. 23).

(5) Benatar (2006: 34).

(6) Benatar (2006: 34–5).

(7) Benatar (2006: 35).

(8) It may also be worth noting that if the example of regret really does pose a problem for my RSP, then a parallel example poses the same problem for Benatar’s RAP. If the Cursed Couple does not conceive, then there is no one for whom they can feel pleased about their decision. If the fact that there is no one for whom the Blessed Couple can feel regret if they do not conceive means that conceiving the Blessed Child would not have made things better for him, then the fact that there is no one for whom the Cursed Couple can feel pleased if they do not conceive would mean that conceiving the Cursed Child would not have made things worse for him. But RAP and RSP both agree that conceiving the Cursed Child would make things worse for him (McMahan (2009: 63) makes a similar point).

(9) Benatar (2006: 35).

(10) Benatar (2006: 32).

(11) Harman (2004: 100).

(12) Harman (2009: 140).

(13) Harman (2009: 147–8).

(14) Reiman (2007: 87–8). See also Jecker (1987: 231–2).

(15) Reiman (2007: 88). The same objection is also pressed by Hudson (1986: 100–1).

(16) Velleman (2008: 277 n. 12).

(17) Strictly speaking, we might say that there is a distinction between positively fulfilling a right and simply not violating a right. We might then say that the girl can ensure that she does not violate the child’s right in the negative sense by not conceiving when she is fourteen but that she cannot actually fulfill the right in the negative sense in this way because a person’s right to something can’t be fulfilled if they don’t exist. If this is so, then my claim that Velleman’s argument depends on construing the right in the positive sense is mistaken. It could also be construed in the negative sense and my first objection to Velleman’s argument only applies to the positive sense. But even if we accept this distinction, it cannot vindicate Velleman’s first argument. This is because if we agree with Velleman that it makes sense to retain a right when there is a chance of appealing to the right as a way of having it fulfilled, we will also think that it makes sense to retain a right when there is a chance of appealing to it as a way of simply preventing it from being violated. The fact that the negative sense of the right is one that the girl can at least avoid violating by not conceiving when she is fourteen will therefore still be enough to show that Velleman’s argument depends on the positive sense of the right.

(18) “[S]‌ince I am glad that you act as you do, with respect to you I waive this right” (1984: 364). See also Parfit (1984: 373).

(19) Velleman (2008: 277).

(20) Parfit (1984: 375).

(21) E.g. Parfit (1984: 367; 2011: vol. 2, 219).

(22) Parfit (2011: vol. 2, 219).

(23) Parfit (1984: 367; 2011: vol. 2, 221). There are some differences between the two versions. The version I give here follows the 2011 version.

(24) Though see Justin Weinberg (2013) for a powerful attempt to explain away this common intuition by identifying several potentially distorting features of the way that Parfit presents the case. See also Hope and McMillan (2012: 25–6).

(25) Parfit (2011: vol. 2, 228).

(26) Parfit (2011: vol. 2, 228–9). Strictly speaking, there are weighted versions of the Two-Tier View that can avoid the problem in this particular case, but the numbers in the case can be revised to generate the same basic problem for any version of the Two-Tier View. Since the argument I develop in the text concedes that Case Six (or at least a case very much like it) provides a decisive reason to reject the Two-Tier View, the details are not relevant to the discussion. I therefore omit them.

(27) Parfit (2011: vol. 2, 231).

(28) A third objection may also be worth noting. A critic might point out that in responding to the first objection to P5, I did not simply provide reasons to favor the exclusive version of the Moderate Principle over the inclusive version. In section 6.1.3.1, I also gave one reason to reject both versions of the principle, the reason based on the case of Fred. If it is not wrong for Fred to save blind Billy rather than sighted Timmy when saving blind Billy is a bit more convenient for him and he cannot save both, then even the exclusive version of the Moderate Principle should be rejected. And if the exclusive version of the Moderate Principle is false, then it cannot be used to undermine Parfit’s argument for the No Difference View.

I do, in fact, think that the case of Fred provides a good reason to reject the Moderate Principle on either formulation. So I do, in fact, think that even the exclusive version of the Moderate Principle should be rejected. But the reason that I find to be a good reason for rejecting the exclusive version of the Moderate Principle is not available to a defender of the No Difference View. This is because the No Difference View entails that the case of Fred is morally no different from what I will call the case of Frank: Frank walks by a pond in which Bobby is drowning. Bobby is sighted and there are two different ways in which Frank could save Bobby. One way would be a bit more convenient for Frank, but it would cause Bobby to become incurably blind. The other way would be a bit less convenient for Frank, and it would enable him to rescue Bobby intact.

It is clear that it would be wrong for Frank to choose the option in which Bobby survives and is incurably blind rather than the slightly less convenient option in which Bobby survives and is sighted. The No Difference View entails that the case of Fred is morally on a par with the case of Frank. This means that if the No Difference View is true, then it would be wrong for Fred to choose the option in which Billy survives and is incurably blind rather than the slightly less convenient option in which Timmy survives and is sighted. Since a proponent of the No Difference View must therefore insist that it really is wrong for Fred to save blind Billy rather than sighted Timmy, the proponent of the No Difference View cannot endorse my objection to the exclusive version of the Moderate Principle. If my objection to the exclusive version of the Moderate Principle is correct, then, we should reject the No Difference View because it is clear that it would be wrong for Frank to save Bobby in a manner that causes him to become incurably blind. And if my objection to the exclusive version of the Moderate Principle is not correct, then we can use the principle to undermine Parfit’s argument in defense of the No Difference View. So either we have no reason to accept the No Difference View or we have a positive reason to reject the No Difference View.

(29) Parfit (2011: vol. 2, 219).

(30) Parfit (2011: vol. 2, 219).

(31) Parfit (2011: vol. 2, 219).