Famine, Affluence, and Mortality - David Bonin - A Priori - University of Canterbury - New Zealand

Famine, Affluence, and Mortality

David Boonin, University of Colorado

In 1972, Peter Singer published an article entitled ‘Famine, Affluence, and Morality’.1 In it, he argued that it is morally wrong for those who can afford to contribute money to famine relief efforts to decline to do so. Singer, of course, is a well-known utilitarian. And if we assume that utilitarianism is true, it is not at all difficult to arrive at the conclusion that it is wrong for us not to contribute money to famine relief efforts. But what made Singer’s argument so powerful, and his article so influential, is that it did not seem to depend in any way on the presumption of a consequentialist framework. Instead, the argument for the wrongness of failing to contribute to such efforts rested on a quite modest moral principle that most people – consequentialists and non-consequentialists alike — would find difficult to deny: if it is within your power to prevent something very bad from happening to someone at no significant cost to yourself or anyone else, then it would be wrong for you not to do so. And Singer brought out the nearly universal appeal of this assumption by means of a now famous example: a child is drowning in a shallow pond and you can save his life, but if you do so, your clothes will be ruined. Virtually everyone will agree that it would be wrong for you not to provide assistance in this case. But, as Singer went on to argue, it is surprisingly difficult to identify any morally relevant difference between providing assistance in the case of the shallow pond, and providing it in the case of famine relief. If we cannot identify such a difference, then it seems that we must accept Singer’s conclusion regardless of whether or not we accept Singer’s utilitarianism.

Singer’s argument has more recently been developed and defended by Peter Unger in his important and challenging book, Living High and Letting Die. Unger presents the position as a puzzle rather than an argument, what he calls ‘a puzzle about our behavior toward people in great need’.2 And instead of a child drowing in a pond, Unger appeals to a case involving an injured bird-watcher whom you could drive to the hospital at some cost to yourself (24-5). But in appealing to the fact that most people agree that it would be wrong not to provide assistance in the hypothetical example and that most people believe that it is not wrong not to contribute money to famine relief organizations, it is clear that Unger is essentially defending the same argument that Singer originated.3 Since Unger’s treatment of the issue is more recent and more fully developed than Singer’s, I will frame my discussion of it in this paper in terms of Unger’s example and Unger’s framework. I will refer to the position in what follows as Unger’s puzzle, and will refer to my response to it as my solution to Unger’s puzzle. But it should be clear that if my position succeeds as a solution to Unger’s puzzle, then it will succeed equally well as a rebuttal to Singer’s argument.

Unger’s puzzle turns on two cases, which he refers to as The Vintage Sedan and The Envelope (I make some changes to the cases in what follows in order to render them more readily comparable).4 In the case of the Vintage Sedan, you can save the life of an injured bird-watcher by driving him to the hospital, but if you do so, he will bleed on the upholstery of your car, causing $100 worth of damage. In the case of the Envelope, you receive a letter from UNICEF correctly informing you that if you send them $100, one fewer child will die prematurely than will otherwise die. Most people respond to the two cases very differently. They believe that it would be immoral for you not to provide assistance in the first case but that it would not be immoral for you not to provide assistance in the second case. And their response generates a puzzle because, at least on the face of it, it is not clear that there is a morally relevant difference between the two cases that would justify this asymmetric judgment.

Two kinds of solution might be proposed in response to this puzzle. One kind of solution is what Unger calls a ‘Preservationist’ one. A Preservationist solution involves identifying a morally relevant difference between the two cases that would justify our accepting our intuitive responses to both of them. A Preservationist solution, in other words, is a solution that shows that the appearance of inconsistency in our responses to the two cases is misleading, and that our divergent responses can, in fact, be reconciled. The other kind of solution is what Unger calls a ‘Liberationist’ one. A Liberationist solution involves maintaining that our responses to the two cases cannot be reconciled, and then demonstrating that while our response to one of the cases is a reliable indicator of our most basic moral values, our response to the other case provides, for one reason or another, a significantly distorted reflection of those values. A Liberationist solution, in short, is one that urges us ultimately to endorse our response to one of the cases and to reject our response to the other.

The solution to the puzzle that Unger himself endorses is a Liberationist one. In particular, Unger argues that there is no morally relevant difference between the two cases, that our response to the Vintage Sedan case accurately reflects our most basic moral values, and that our response to the Envelope case is most reasonably attributed to a variety of distortional tendencies in our thinking. The result is that, on Unger’s account, declining to give money to organizations like UNICEF is, in fact, just as wrong as declining to take the bird-watcher to the hospital in the Vintage Sedan case. This conclusion strikes most readers as extremely difficult to accept. But if no successful Preservationist solution to the puzzle is forthcoming, we may well have no choice but to accept it. I will argue in this paper, however, that a successful Preservationist solution to Unger’s puzzle is available.

In order for a Preservationist solution to succeed, we must identify a difference between the Envelope case and the Vintage Sedan case and show that the difference is a morally relevant one. In his treatment of the Preservationist approach to the puzzle, Unger considers and rejects a number of such differences that might be appealed to. In arguing against each of them, Unger appeals to two distinct kinds of strategy: he argues that our moral common sense denies that the difference being appealed to is a morally relevant one, and he claims that this judgment is confirmed by appealing to our intuitive responses to suitably devised variant cases. In response to the observation that there is a difference between the two cases in terms of physical proximity, for example, Unger argues that our moral common sense tells us that differences in mere physical distance are not themselves morally relevant, and he argues that this judgment is confirmed by our intuitive responses to further cases: most people will still think that your behavior in Vintage Sedan is immoral even if the bird-watcher is ten miles away from you when you learn of his plight, and most will still think that your behavior in Envelope is not immoral even if the needy children are just a few feet away from you, say on the other side of the wall that surrounds the resort you are visiting in Haiti (33-4).

I believe that Unger’s use of these two strategies is successful in responding to the various differences between Vintage Sedan and Envelope that he considers in his book. But I also believe that there is a further difference between the two cases that he overlooks. This further difference has to do with the nature of the causal chain that runs from your act of providing aid to the final result that your aid helps to produce. I will refer to it as the distinction between aid that is ‘uniquely directed’ to a specific beneficiary and aid that is not. I will begin, in section I, by explaining what I mean by ‘uniquely directed’ aid and by justifying the claim that the Vintage Sedan case involves uniquely directed aid while the Envelope case does not. In sections II and III, I will then follow Unger’s two principal strategies as a means of testing the claim that the difference between uniquely directed aid and non-uniquely directed aid is a morally relevant one. In section II, I will argue that when the Envelope case is modified to make it like the Vintage Sedan case with respect to unique directedness, our intuition about whether or not it is wrong to refrain from giving aid changes. This provides evidence that the difference between aid that is uniquely directed and aid that is not does, in fact, make a moral difference. In section III, I will argue that when the common sense basis of our intuitive response to the Vintage Sedan case is properly understood, it can be seen that our moral common sense supports the idea that the difference is itself a morally relevant one. In section IV, I will consider and respond to four objections that might be raised against my solution to Unger’s puzzle. And in section V, I will conclude by briefly acknowledging several respects in which the position taken in this paper is a somewhat limited one and by explaining why it is significant nonetheless.

I

By saying that P’s aid is ‘uniquely directed’ to Q, I mean that whether or not Q suffers a serious loss depends on whether or not P provides aid and that there is no one else whose prospects for suffering a serious loss will be effected by P’s decision. I will focus here on cases in which Q is a single individual, but what I say below would apply equally to cases where Q refers to a set of individuals (e.g., a group of five people that only P can save by lowering a lifeboat big enough to hold them all, where P’s decision about whether or not to lower the lifeboat will not significanly effect the life prospects of anyone else). More precisely:

P’s act A provides aid that is uniquely directed at Q iff all of the following obtain:

  1. if P does A, then Q will not suffer a serious loss
  2. if P does not do A, then Q will suffer a serious loss
  3. there is no one other than Q whose prospects for suffering a serious loss will be determined by P’s doing or not doing A

To make the application of this property to Vintage Sedan and Envelope more conspicuous, it may be useful to begin by modifying Unger’s original stories so that the cost to you in providing assistance and the magnitude of the benefit to the needy person in receiving it is the same in both cases. So let us now suppose that it would only cost you $100 to repair the damage caused to your car by saving the bird-watcher in the Vintage Sedan case, that this would save his life and not merely his leg, and that you must send $100 to UNICEF in order to be assured that one fewer child will die prematurely as result of your assistance. For the remainder of the paper, this is what I will mean by the Vintage Sedan and Envelope cases.

Having said this, we can start by confirming that in the case of Vintage Sedan, the aid you would provide would be uniquely directed in the sense defined above. If you give the bird-watcher a ride, he will not die. If you do not give the bird-watcher a ride, he will die. And there is no one else whose life depends on whether or not you give the bird-watcher a ride (if giving the bird-watcher a ride meant that a nearby photographer whose life you would otherwise save would die instead, after all, then it would no longer be at all clear that it would be wrong for you not to give the bird-watcher a ride in the first place).

What about the aid that you could provide in the case of Envelope? Here things are different. It is not the case that your decision will effect one and only one person and that the effect will be that this one person either lives or dies depending on your choice. Rather, as Unger himself acknowledges in a related but distinct context, ‘on one end of a causal chain, there are many donors contributing together and, on the other, there are all the people saved by the large effort they together support. The more support given, the more folks saved, and that’s all she wrote’ (48-9). The reason that your act of sending money to UNICEF does not provide uniquely directed aid is simple. When you send UNICEF some money, it does not treat the money as a discrete sum to be used for a discrete purchase to be given to a discrete individual. Rather, UNICEF adds the money to the other money that it already has. Having more money rather than less, in turn, has an effect not just on how it spends the particular amount of money it received from you, but also on how it spends some of the money it received from other people. And this, in turn, means that when you send UNICEF enough money to ensure that one fewer child dies prematurely, there is no particular child at whom your aid is uniquely directed.

This feature of the complex dynamics of famine relief can be brought out more clearly by attending to the following, particularly vivid, example, but the feature of the example that prevents your aid from being uniquely directed in this case will appear in any instance in which money is sent to a large famine relief organization.

The Two Villages: next week, UNICEF will deliver a large number of packages of food and supplies to one or the other of two large villages within driving distance of its distribution point. Because the villages are in opposite directions from the distribution point, and because of its limited supply of fuel, UNICEF can only deliver supplies to one village or the other before the children in question will die. There are 1,000 children starving in Village A and 1,001 children starving in Village B. UNICEF has one truck and 1,001 packages available at the distribution point. Because the drive to Village B is a bit longer, UNICEF currently has enough fuel to deliver the supplies to Village A but not enough to deliver them to Village B. If you do not send $100 to UNICEF, the packages will be delivered to Village A, and the 1,000 children there will be saved. If you do send $100 to UNICEF, the money will be spent on more fuel, the packages will instead be delivered to Village B, and the 1,001 children there will be saved.

In the case of The Two Villages, if you send UNICEF $100, one fewer child will die prematurely. But, as this case makes very clear, this is not because there is some one particular child who will be the only one effected by your decision and who will live or die depending on what you do. Rather, your decision about whether or not to send the money to UNICEF will have an effect on a large number of children. 1,000 children in Village A will be spared premature death while 1,001 children in Village B will die prematurely if you do not send UNICEF the money, and 1,001 children in Village B will be spared a premature death while 1,000 children in Village A will die prematurely if you do send UNICEF the money. In this case, then, although it is true that your sending money to UNICEF will mean that one more child will be spared, it should be clear that there is no one particular child that your aid would be uniquely directed to in the sense defined above. And while this is particularly clear in this case because of the particular impact your act will have on UNICEF’s decision about how to spend its money, it should also be clear, upon reflection, that the same sort of thing will happen in any realistic case involving such organizations. Given that such organizations do not treat each donation as a discrete quantum of money to be used to purchase a discrete set of goods to provide to a discrete individual, no donation made to them will be uniquely directed in the sense defined above.

II

I have argued thus far that there is a difference between the Envelope case and the Vintage Sedan case. In both cases, it will cost you $100 to provide aid and if you do provide aid one fewer person will die prematurely, but in the Vintage Sedan case, your aid would be uniquely directed to a specific beneficiary, whereas in the Envelope case it would not be. The question now is whether or not this difference is a morally relevant one.

One way to answer this question, following Unger’s own technique, is to modify the Envelope case so that if you send the money to UNICEF, your aid will be uniquely directed to a specific beneficiary, just as it is in the case of the Vintage Sedan. In the original Envelope case, our moral common sense response is that it is not immoral to refrain from sending the money. If we have the same response to a variant case in which the aid would be uniquely directed, this would count as evidence that the difference between aid that is uniquely directed and aid that is not does not make a moral difference. If, on the other hand, our response changes, so that we think it would be immoral not to send money in the variant case, this would count as evidence that the difference between aid that is uniquely directed and aid that is not does make a moral difference. So consider the following case:

e-velope5: you are checking out various vacation-planning web sites when a computer error mistakenly connects you to a web site sponsored by UNICEF. While you are reading the organization’s materials, a new bulletin is posted to the site. After reading the bulletin, you correctly believe that there is a particular child who has just contracted a fatal illness and who will die soon unless someone who is currently logged onto the site sends in $100 to pay for the medicine she needs by clicking on an icon at the bottom of the screen, which will authorize a $100 withdrawal from the user’s bank account. As you are considering whether or not to send in the needed money, a second bulletin appears. After reading this second bulletin, you correctly believe that the server UNICEF is using is experiencing problems and that no one else will be able to log on to the site until it is too late to save this particular child. Finally, you look at a window at the bottom of the screen, after which you correctly believe that you are the only person who is currently logged onto the site. You correctly believe, in short, that this particular child will die if you do not click on the ‘send $100’ icon and that this child will live if you do click on the ‘send $100’ icon. You decide not to click the icon and turn off your computer, and, instead of living for many more years, this particular child dies.

In order to use the e-velope case to vindicate my solution to Unger’s puzzle, two claims must be sustained: that the aid you would provide by sending money to UNICEF in the e-velope case would be uniquely directed toward a specific beneficiary, and that our moral common sense says that it would be immoral for you not to send the money. The first claim can be established quite easily. If you click on the icon, this particular child will live. If you do not click on the icon, this particular child will die. And there is no other person whose prospects for suffering a serious loss will be determined by whether or not you click on the icon (if, for example, in order to click on the icon, you would have to stay at your computer rather than save the life of a child who was choking to death across the hall from you, the case would be relevantly different). Your aid in e-velope, therefore, unlike in Envelope, would be uniquely directed at a specific beneficiary.

The second claim is a claim about what our intuitions are. It cannot be established as decisively. But I can at least say the following. First, my own intuition is that your behavior in the e-velope case is seriously wrong. It is true, of course, that it is not your fault that you stumbled onto the UNICEF site just when this child was in need of assistance, and that it is not your fault that there was no one else on the site to assist her when you stumbled onto it. But it is equally true that it is not your fault that you stumbled upon the bird watcher just when he was in need of assistance, and that it is not your fault that there was no one else at the scene to assist him when you stumbled upon him. Just as this fact does nothing to diminish your obligation to assist him in the Vintage Sedan case, it does nothing to diminish your obligation to assist her in the e-velope case. My intuition, then, is that your behavior is wrong in e-velope, but not wrong in Envelope. Since the only salient difference between the two cases is that your aid would be uniquely directed in e-velope not in Envelope, my intuitive reaction to the cases supports the claim that there is a morally relevant difference between aid that is uniquely directed and aid that is not.

This vindication of the claim that the difference involving uniquely directed aid is a morally relevant one, of course, is based upon my own intuitive reaction to the e-velope case. But I strongly suspect that most readers will share this intution. In an informal poll that I conducted among approximately 50 colleagues and students, a very clear majority responded that your behavior in e-velope is wrong and, indeed, just as wrong as it is in cases like Vintage Sedan. In addition, virtually all of the handful of people who believed that your behavior was less wrong or not at all wrong in the e-velope case said that their reason for holding this was that they believed there was a substantial chance that the web site appeal was a scam. If they were absolutely convinced that the web site was legitimate, or if they were equally uncertain as to whether or not the bird-watcher was faking his injury in order to rip them off, then they, too, agreed that your behavior in e-velope was just as wrong as it is in Vintage Sedan. When other factors were successfully held constant, that is, virtually everyone agreed that your behavior in e-velope is just as wrong as in Vintage Sedan. Since virtually everyone agrees that your behavior is not wrong in Envelope and since, again, the only salient difference between e-velope and Envelope concerns the unique directedness of the aid, this suggests that virtually everyone does, in fact, treat the difference between aid that is uniquely directed and aid that is not as a morally relevant one.

Finally, and in some sense most importantly, it is worth noting that it is extremely difficult to believe that Unger himself could endorse a response to the e-velope case that is different from mine and that of virtually everyone I have talked to about it. Unger clearly endorses the intuition that your behavior in Vintage Sedan is wrong, after all, and the differences between Vintage Sedan and e-velope (in one case the victim is physically near, in the other case the victim is physically distant, etc.) are differences that Unger himself finds to be irrelevant to our intuitions. Given this, it is difficult to see how Unger can avoid conceding that, so far as our intuitive responses to cases is concerned, e-velope is like Vintage Sedan and unlike Envelope. And if I am correct about this, then one of the two strategies that Unger employs when evaluating solutions to his puzzle proves to vindicate the solution that I am proposing. It would be wrong not to provide aid in Vintage Sedan because your aid would be uniquely directed in that case, and it would be wrong not to provide aid in e-velope for the same reason. But since your aid would not be uniquely directed in Envelope, there is no inconsistency between the moral common sense view that it is wrong not to help in Vintage Sedan and the moral common sense view that it is not wrong not to help in Envelope. Our initial reactions to the two cases prove to be perfectly consistent with each other, and the Preservationist response to the puzzle is therefore vindicated.

III

The technique involving our intuitive reaction to suitably devised variant cases provides one way to evaluate the claim that there is a morally relevant difference between aid that would be uniquely focused and aid that would not be. I have argued that this technique supports my solution to Unger’s puzzle. If the e-velope, but not the Envelope, case really does involve uniquely directed aid, and if the e-velope, but not the Envelope, case really does elicit from most people the moral common sense response that it would be wrong not to provide aid, then our moral common sense really does treat the difference between aid that is uniquely directed and aid that is not as a morally relevant one. A second approach to evaluating the solution I am proposing takes place at a more theoretical level and tries to explain why the difference between aid that is uniquely directed and aid that is not should be morally relevant from the point of view of our moral common sense. What precisely is it about unique directedness that makes it relevant to our moral judgments? I am not prepared here to offer a fully-developed answer to this question. But I do want to offer something of a sketch that should help to explain why the kind of theoretical considerations that best cohere with our common sense moral views would treat the distinction between uniquely directed aid and non-uniquely directed aid as a morally relevant one. I do not claim to provide a decisive reason to believe that our basic moral values endorse the relevance of the distinction, but I hope to say enough to put the burden on those who would claim that they do not.

We can start by noting that there are two different kinds of reason that our moral common sense might be appealing to in telling us that it would be wrong for us not to save the bird-watcher in the Vintage Sedan case. One kind of reason is impersonal. On this kind of account, the reason that it would be wrong for you not to save the bird-watcher has to do with the fact that the resulting state of affairs if you do not save him is considerably worse than is the resulting state of affairs if you do save him, where the different states of affairs are assessed in a manner that takes into account the interests of all equally. The other kind of reason is personal. On this kind of account, the reason that it would be wrong for you not to save the bird-watcher has to do with the fact that the bird-watcher has a legitimate claim against you in particular. He has a right to your assistance, and if you decline to assist him, then you violate this right and this is why it would be wrong for you not to assist him.

My first claim is that our moral common sense coheres better with the second kind of explanation than with the first. This claim, too, can be tested by considering our intuitive reaction to a suitably devised variant case. The case involves changing things so that, from an impersonal point of view, the consequences of your saving the bird-watcher’s life are the same as the consequences of your declining to do so. So suppose that you have one hundred one dollar bills spread out across the back seat of your sedan, and that you are on your way to the local UNICEF office, where you intend to donate the money, thus ensuring that one fewer child dies prematurely. When you encounter the bleeding bird-watcher, you realize that if you give him a ride, he will ruin the dollar bills, making them worthless to UNICEF or to anyone else. In this version of the Vintage Sedan case, you have two choices: pick up the bird-watcher or take the money to UNICEF. Whichever choice you make, the results will be the same from an impersonal point of view: you will be one hundred dollars poorer than you were this morning, and one fewer person will die prematurely than would have died had you stayed in bed. In the original version of the Vintage Sedan case, our moral common sense clearly tells us that it would be wrong for you not to save the bird watcher. If our moral common sense reaction to that case is explained by the fact that saving the bird watcher produces much better consequences than not saving him, when the consequences are assesssed from an impersonal point of view, then our moral common sense reaction to this modified version of the Vintage Sedan case should be one of complete indifference. Since, in this modified version of the case, the consequences impersonally measured are the same either way, we should respond by thinking that it is just as good to leave the bird-watcher to die so that you can take the hundred dollars to UNICEF as it is to ruin the hundred dollars so that you can take the bird-watcher to the hospital (and if we are concerned about any extra suffering that the bird-watcher might endure from watching you drive off without him, we can stipulate that he is unconscious when you come across him and will die before regaining concsciousness if you do not drive him to the hospital). Indeed, if the impersonal account is the correct explanation of our reaction to the original Vintage Sedan case, then we should think that even in that case, it would be perfectly fine for you to leave the bird-watcher to die, provided that you then sent $100 to UNICEF that you were not otherwise planning to send (indeed, further still, we should think that it would be positively better for you to leave the bird-watcher to die and bring the money to UNICEF, since doing so would also spare you the temporary grief and inconvenience of having your upholstery damaged). But surely our common sense moral reaction to these cases is not at all like this. We surely think that you should save the bird watcher you have come across even if this means not giving $100 to UNICEF that you would otherwise have given. And this strongly suggests that the correct explanation of our response to the original Vintage Sedan case is based on the claim that the bird-watcher has a right to your assistance, regardless of whether saving him will make the world a better place from an impersonal point of view.6

Let us now suppose that I am correct about this. Our intuitive belief that it is wrong not to save the bird-watcher in the Vintage Sedan case is better explained by an appeal to personal rather than impersonal reasons. The next question is what kind of personal reason best coheres with our moral common sense reaction. Our moral common sense clearly dictates that we save the bird-watcher in the Vintage Sedan case, but it also clearly does not suggest that it is always wrong for us not to assist other people who are in need. It does not, for example, indicate that it would be wrong for you not to prevent someone from suffering a minor inconvenience, nor does it suggest that you must prevent someone from losing both of their arms if you would have to lose one of your arms in the process. The best explanation of this seems to be that the rights-based reasoning behind our intuitive reaction to the Vintage Sedan case is something like this: B has a right to A’s assistance if A can prevent something very bad from happening to B without incurring any significant cost to himself or imposing any significant cost on anyone else.7

Let us next suppose that I am correct about this as well: the best explanation of why it would be wrong for you not to save the bird-watcher in the Vintage Sedan case is that the bird-watcher has a right to your assistance and that he has this right in virtue of the fact that you can prevent something very bad from happening to him at a relatively trivial cost to yourself (and to anyone else). And now consider what this explanation would imply about the following case:

Fragile Sedan: you are driving past a bird-watcher who is walking back to his car. He has one mile to go. You know that there is a four in ten million chance that he will suffer a heart attack walking back to his car and that, if he does have a heart attack, no one else will come by to take him to a hospital in time to save his life. You also know that giving him a ride back to his car will reduce his risk of suffering a fatal heart attack to three in ten million. Because your car is so fragile, however, you also know that giving him a ride will cause $100 worth of damage to your car.

In the Fragile Sedan case, you can reduce the bird-watcher’s chance of death by one in ten million. Being exposed to a risk of death that is one in ten millionth greater than one would otherwise be exposed to is not a very bad thing. And so by giving a ride to the bird-watcher in the Fragile Sedan case, you would not be preventing a very bad thing from happening to him. On the assumption that you must save the bird-watcher in the Vintage Sedan case because people have a right to your assistance when you can prevent something very bad from happening to them at no significant cost to yourself, then, it turns out that the best explanation of why you must give the bird-watcher a ride in that case does not imply that you must give him a ride in the Fragile Sedan case. This result, moreover, coheres well with our common sense reactions to other cases, including cases in which we have special obligations of care to the people in question. Suppose, for example, the you are sending your child off to camp for the summer and are looking into two different means of transportation he could use: you can send him by train or by bus. Suppose, in addition, that you happen to learn that the chances that he will die in a fatal transportation accident are one in ten million greater if he takes the train, but that the train is also $100 cheaper. I doubt that anyone would think that it would be wrong for you to nonetheless send him on the train. Indeed, people make such calculations about their own lives all the time.

In cases where you could incur $100 worth of harm in the act of decreasing someone’s chance of death by one in ten million, then, you do not treat the person immorally if you decline to do the act in question. But now consider the fact that, if this is so, then it is so no matter how many people there are whose risk of death your single act would decrease by this amount. This can be made clear by considering the following, admittedly quite fanciful case:

Enormous Sedan: you are driving past ten million bird-watchers who are all walking back to their cars. Every one of them has one mile to go. You know that, for every single one of these ten million bird-watchers, there is a four in ten million chance that that particular bird watcher will suffer a heart attack walking back to his car and that, if he does, no one else will come by to take him to a hospital in time to save his life. You also know that, for every single one of these ten million bird-watchers, giving him a ride back to his car will reduce his risk of suffering a premature death from a fatal heart attack to three in ten million. Because the process by which a heart attack is triggered is indeterministic, there is no particular bird-watcher about whom it is true that he will die if you do not pick him up and live if you do. The facts are simply that if you do not let everyone in, four will die, and if you do let everyone in, three will die and, as Unger puts it, that’s all she wrote. Because your car is so enormous, you could, in fact, put all ten million bird watchers in your car and give them all a ride. You also know that doing so will cause $100 worth of damage to your car.

In the Enormous Sedan case, the cost to you of putting everyone in your car is $100 and the expected benefit is that one fewer person will die prematurely (three will die rather than four). So, from the impersonal point of view, the Enormous Sedan case is exactly like the Vintage Sedan case. In both cases, the choice is between a state of affairs in which you do not incur $100 dollars worth of damage to your car and a state of affairs in which you do incur $100 dollars worth of damage to your car and one fewer person dies prematurely. If the degree to which the latter state of affairs is impersonally better than the former is what makes it wrong for you not to provide assistance in the Vintage Sedan case, then it will also suffice to make it wrong for you not to provide assistance in the Enormous Sedan case.

But suppose again that I have been correct in maintaining that the better explanation of our common sense reaction to the Vintage Sedan case is the one that appeals to personal rather than impersonal reasons, the account that appeals not to the goodness of the states of affairs that you can produce but rather to the rights to assistance that others can legitimately claim against you. On this account, the Enormous Sedan case turns out to be fundamentally different from the Vintage Sedan case. In the Vintage Sedan case, there is a particular, determinate individual who will have something very bad happen to him if and only if you do not give him a ride in your car. And since you can give him a ride at the cost to you of only $100 and no significant cost to anyone else, it follows from the rights-based principle proposed above that this individual has a right to your assistance.

But in the Enormous Sedan case, things are crucially different. In that case, the facts are simply that if you put everyone in the car, three of them will die prematurely and if you don’t, then four of them will die prematurely. And given this, there no individual who can legitimately claim that he will die if and only if you do not give him a ride in your car and that doing so will not impose any significant costs on anyone else, and so there is no individual who can legitimately claim that he has a right to your assistance. It is true, of course, that in the Enormous Sedan case there are ten million people each of whom can legitimately claim that he will be exposed to a slightly higher risk of death if and only if you do not give him a ride in your car. But, as we have already seen, being exposed to a risk of death that is one in ten million greater than one would otherwise be exposed to is not in itself a very bad thing. If you decline to let any of the ten million bird-watchers into your enormous car, then, you will not be violating the rights of any one of them. And since, on the personal account, the reason it would be wrong for you not to give the single bird-watcher a ride in the Vintage Sedan case was that doing so would violate his right to your assistance, it follows that the reason for thinking it wrong not to provide assistance in that case does not imply that it would be wrong for you not to provide assistance in the Enormous Sedan case.

Finally, let us now suppose that I have been correct about all of this: the best account of our common sense moral reaction to the Vintage Sedan case appeals to the claim that the bird-watcher in that case has a right to your assistance, and that the best account of this right does not entail that it would be wrong for you not to provide assistance in the case of the Enormous Sedan. If all of this is so, we can now see precisely why the difference between Vintage Sedan and Envelope in terms of unique directedness is morally relevant from the point of view of common sense morality. In the case where UNICEF asks you for $100 to reduce by one the number of children who die prematurely, the situation is relevantly like it is in the case of the Enormous Sedan: there are ten million children who face an extremely high probability of premature death. If you do not send $100 to UNICEF, the risk they face will be precisely the same as it would be had UNICEF not contacted you in the first place. If you do send in the $100, then of those ten million children, the number who die prematurely will be smaller by one. On the assumption that the aid you provide would not be uniquely directed, however, this is not because your act will have an impact on one and only one child who will live if you send the money and die if you do not. Rather, it is simply the case that the total number who die will be smaller by one if you do send the money than if you do not. Since there is no determinate individual about whom it is true that your act would prevent something very bad from happening to them without having a significant impact on anyone else, there is no individual whose rights will be violated if you do not send the money to UNICEF. In addition, although it is true that every one of the ten million children will be exposed to a slighly higher risk of death if you do not send the money, being exposed to a risk of death that is higher by only one in ten million is not in itself a very bad thing and so, again, none of the children in question have a right that you send your money to UNICEF. But on the account that has been defended here, the best explanation of why it would be wrong for you not to help the bird-watcher in the Vintage Sedan case is precisely that the bird-watcher does, in fact, have a right to your assistance. And it follows from this that the best account of why it would be wrong not to help in the Vintage Sedan case does not, in fact, imply that it would be wrong for you not to send the money to UNICEF in the Envelope case. The apparent inconsistency between our common sense moral intuitions in the two cases, therefore, turns out again not to be an inconsistency at all. In the end, our different reactions to the Vintage Sedan and Envelope cases prove instead to be a reflection of the fact that our intuition in the Vintage Sedan case arises from a basic moral belief about the right to assistance that others can legitimately claim against us and the fact that this right to assistance cannot be directly transferred from cases involving uniquely directed aid to cases that do not involve uniquely directed aid. The comparison between e-velope and Envelope demonstrates that whether or not aid is uniquely directed does, in fact, have an impact on our intuitive reactions. The general lesson about rights extracted from the cases of the Fragile Sedan and the Enormous Sedan helps to explain why it is appropriate that it should have this impact.

IV

There are, not surprisingly, a number of objections that can be raised againt my solution to Unger’s puzzle. In this section, I will identify what strike me as the most conspicuous of these and attempt to show that they can be overcome.

The first objection arises from the fact that there seem to be cases in which the aid you could provide would fail to be uniquely directed in precisely the same way that it fails to be uniquely directed in my Enormous Sedan case but in which most people would agree that it would be wrong for you not to provide it. So consider the case of

Russian Roulette: a bird-watcher-hating villian has devised a special kind of gun with which to play Russian Roulette. The gun has ten chambers and can be spun in a truly random, indeterministic manner. He has trapped one hundred bird-watchers and is about to expose each of them, one by one, to a nine in ten chance of instant death by putting bullets in nine of the ten chambers before firing at each of them. You can prevent the one hundred bird watchers from being exposed to this risk by letting them all into your very large sedan. Because the bird-watcher-hating villian has boobytrapped your sedan, however, you know that doing so will expose each of them to a risk of death by being fired at by a second set of such guns, each of which has a bullet in only one of its ten chambers. None of the guns are pointed at the driver, so there will be no risk of premature death for you, but the firing of the guns, if you let them aboard, will cause $100 worth of damage to your sedan.

In the case of Russian Roulette, there is no particular bird-watcher who can legitimately claim that he will die if and only if you do not let him into your sedan. If you do not give a particular bird-watcher a ride, it is quite likely that he will die, but he might not. And if you do give a particular bird-watcher a ride, it is much less likely that he will die, but he still might. So all we can say is that 90 bird-watchers will die if you do not let everyone into your sedan, that 10 bird-watchers will die if you do let everyone on board and that, as Unger puts it in the context of the Envelope case, that’s all she wrote. This does, indeed, mean that the aid you would provide in the case of Russian Roulette would not be uniquely directed, and I do, in fact, agree that it would be wrong for you not to let the bird-watchers on board in this case. Does this mean that there is a problem with my solution to Unger’s puzzle?

It does not. The first reason for this is a simple one. My solution does not depend on the claim that unique directedness is the only thing that can make it wrong not to do an act that would provide aid. If yesterday you promised that you would sent a check to UNICEF today, for example, then this fact could also make it wrong for you not to send a check to UNICEF today, and its doing so would be perfectly consistent with my solution to Unger’s puzzle. My solution depends only on the claim that the unique directedness of the aid that one could give is sufficient (but not necessary) to make it wrong not to give the aid. Only this claim is needed because Unger’s puzzle asks how it could be wrong not to help in the Vintage Sedan case but not wrong to not help in the Envelope case, and this claim answers this question. The claim answers the question by showing that there is a sufficient reason for helping in the Vintage Sedan case that is not present in the Envelope case. And since this claim is not undermined by our intuitive reaction to the further case of Russian Roulette, this further case can pose no problem for my solution to Unger’s puzzle.

This first response to the objection based on cases such as Russian Roulette may seem to some to ring a bit hollow. If there can be some cases in which it would be wrong not to provide aid that would fail to be uniquely directed, after all, then why couldn’t the Envelope case turn out to be one of them? Even if I am correct in maintaining that Unger’s Vintage Sedan case fails to show that it is wrong not to provide aid in the Envelope case, that is, maybe the wrongness of not giving aid in the Envelope case could still be established by instead appealing to a case such as Russian Roulette. If this turns out to be the case, then even if Unger’s particular version of the puzzle turns out to be dissolved, a second and equally troubling puzzle will simply take its place.

But this further worry is unfounded. For if the account based on unique directedness does provide the best explanation of why it would be wrong not to provide aid in the Vintage Sedan case, then the best explanation of why it would be wrong not to provide aid in the Russian Roulette case will simply be based on a broadened version of that account. In the Vintage Sedan case, that is, letting the bird watcher into your car reduces his risk of imminent death from one to zero. In the Russian Roulette case, by contrast, letting any particular bird watcher into your car reduces his risk of imminent death from nine in ten to one in ten. So to modify the argument based on unique directedness in order to account for the wrongness of not providing aid in the case of Russian Roulette, it will be sufficient simply to relax the first two conditions required in order for aid to count as uniquely directed. Rather than insisting that if P does A, then Q will not suffer a serious loss and that if P does not do A, then Q will suffer a serious loss, that is, we will simply require that there be a sufficiently great difference between the probability that Q will suffer the loss if P does A and the probability that Q will suffer the loss if P does not do A. If there is a sufficiently great difference in the probabilities, then this broadened version of the account based on unique directedness will entail that it would be wrong not to provide the requisite assistance, even if the difference between giving aid and not is not the difference between certain death and certain survival.

This broadened version of the unique directedness position produces the intuitively correct results in both the case of the Vintage Sedan and the case of Russian Roulette. And the revision it makes in order to accomplish this is not ad hoc. Indeed, even in the original Vintage Sedan case, the broader version is presumably what is really doing the work in generating our intuition. Even if you were not literally certain that the bird-watcher would die if you did not give him a ride to the hospital or that he would survive if you did, that is, what generates the belief that it would be wrong not to give him a ride is the fact that your giving him the ride will be uniquely directed to him (there is no chance that it will benefit anyone else) and that there is a sufficiently great probability that your doing so will turn out to make the difference for him between life and death.

And while this broadened version of the unique directedness view accounts for the wrongess of declining to provide aid in the Vintage Sedan and Russian Roulette cases, it fails to entail that it would be wrong to provide aid in the Envelope case, just as the original version of the unique directedness account failed to entail this. For the broadened version of the unique directedness account depends on the claim that there is a sufficiently great difference in probabilities, and in the Envelope case, there is no such difference. As we already saw in the discussion of the Enormous Sedan case, for example, the difference between, say, a four in ten million chance of imminent death and a three in ten million chance of imminent death is not sufficiently great to generate an obligation to spend $100 to prevent it. If we changed the Russian Roulette case so that the gun outside your sedan had four bullets in ten million chabers and the gun inside it had three bullets in ten million chambers, therefore, then it would no longer be wrong for you to decline to incur $100 in damages to your sedan by letting the bird-watchers in. And if I was correct in maintaining that the Envelope case was relevantly like the Enormous Sedan case in the context of the unmodified version of the unique directedness account, then it should be clear that the Envelope case will be like this version of the Russian Roulette case in the context of the modified version. There is no particular child about whom it is true that your sending $100 to UNICEF would provide uniquely directed aid to them even on this broader account. No child can legitimately claim that the probability that he will avoid immanent death will be significantly reduced if you send the money. And so there is no reason to think that the wrongness of not providing aid in the (first version of the) Russian Roulette case will provide support for the claim that it would be wrong not to provide aid in the Envelope case.

A second objection maintains that my solution to Unger’s puzzle has a further problematic implication. If I am correct that the reason it is not wrong not to send money in the Envelope case is that the aid you would provide by doing so would not be uniquely directed, the objection claims, then groups such as UNICEF could simply change their strategies in order to ensure that any aid you provide would be uniquely directed to its beneficiary. Suppose, for example, that there are ten million children at risk of imminent death from starvation and ten million potential donors such as yourself who could easily afford to donate $100. Suppose further that all of the famine relief agencies and organizations in the world pooled their resources and sent each of the ten million potential donors the letter described in the following example:

Uniquely Directed Envelope: you receive a letter from a famine relief consortium correctly informing you of the following: there is a particular child in a particular town in a particular country who will die in a matter of weeks if she does not receive assistance. If you send the consortium a check for $100, the consortium will ensure that this particular child recevies the requisite aid and will survive. If anyone else in the world other than you sends $100 to the consortium, the consortium will ensure that someone other than this particular little girl will receive the aid that the money pays for. And so if you do not send $100 to the consortium, this particular child will die.

It seems clear that your aid would be uniquely directed in the case of the Uniquely Directed Envelope, and that my solution would therefore entail that it would be wrong for you not to send money in this case. But if this is so, then it would seem that organizations like UNICEF could render your aid obligatory simply by changing the way that they distribute their assistance.

The claim that my solution entails that it would be wrong not to send money in the case of the Uniquely Directed Envelope is correct. But the claim that this poses a problem for my solution is mistaken. My solution is aimed at a particular puzzle. The puzzle arises from the fact that we have different intuitions in the case of Vintage Sedan and Envelope. A solution to this puzzle must explain why we are justified in responding differently to the two cases. It must explain, in particular, why it is not wrong not to send money in the case of Envelope. But a solution to the puzzle need not also justify the much stronger claim that it is never wrong to decline to send money to famine relief organizations. And, indeed, this further claim is almost certainly false. As noted earlier, for example, declining to send money could be wrong because it would involve breaking a promise you had previously made. My solution to Unger’s puzzle simply maintains that in the kind of Envelope case that Unger describes, the kind of case that represents the way that famine relief actually operates, the fact that your aid would not be uniquely directed explains why it would not be wrong for you to decline to send money. That is all that my solution has to maintain, and the fact that it would be wrong not to send money in the case of the Uniquely Directed Envelope does nothing to challenge this claim.

Indeed, rather than posing a problem for my solution, I am inclined to think that the case of the Uniquely Directed Envelope simply provides further support for it. For it seems clear to me that it would, in fact, be wrong not to send money in the case of the Uniquely Directed Envelope, just as it would be wrong not to send money in the case of e-velope. And since the distinction between aid that is uniquely directed and aid that is not would explain why it is wrong not to send aid in Uniquely Directed Envelope (and e-velope) but not wrong not to send aid in Envelope, the case of the Uniquely Directed Envelop provides further confirmation that the distinction between the two kinds of aid is morally relevant. And since my solution to Unger’s puzzle rests precisely on the claim that this distinction is a morally relevant one, the case of the Uniquely Directed Envelope turns out to support my solution rather than to oppose it.

Despite all of this, however, it might be thought that this implication of my solution is still problematic for two further reasons. First, although the implication does not directly threaten my claim to have identified a morally relevant distinction between Envelope and Vintage Sedan, it does suggest that whether or not I am obligated to donate money for famine relief depends on contingent facts about how famine relief organizations operate. This feature of the implication might be taken to be sufficiently puzzling to warrant rejecting my solution. But this worry seems unwarranted. In the first place, there should be nothing puzzling about the general idea that whether or not I am obligated to donate money to a particular organization should depend on contingent facts about how it operates. If an organization is going to simply flush my money down the toilet, for example, then that seems clearly to be morally relevant. Second, and more specifically, there should, at least upon reflection, be nothing puzzling about the claim that you can, in fact, in relevantly similar circumstances, be manipulated into being obligated to sacrifice $100 by a suitably resourceful manipulator. After all, if I am correct that you are obligated to incur such a cost in Vintage Sedan but not in Enormous Sedan, then a suitably resourceful manipulator could manipulate you into being obligated to incur a $100 loss simply by arranging to have you find yourself in the position involved in the former case rather than the latter. In this instance, as we have already seen, there are two cases which are similar from an impersonal point of view (in each case, the expected result of sacrificing $100 is that one fewer person will die), but which are dissimilar from a personal point of view (in the former case, there is someone who has a right to your assistance, in the latter case, there is no one who has a right to your assistance). There is nothing puzzling about the claim that you can be obligated to help in one of the cases but not in the other since there is nothing puzzling about the claim that whether or not you are obligated to help can depend on whether or not there is someone who has a right to your assistance. Since my claim is that, upon reflection, we should see the difference between Envelope and Uniquely Directed Envelope in just the same way that we see the difference between Vintage Sedan and Enormous Sedan, there should, at least upon reflection, be nothing puzzling about the claim that organizations like UNICEF could, in fact, manipulate things so that you would be obligated to give them money (by arranging to send you the Uniquely Directed Envelope) where you would otherwise not be obligated (if they instead sent you the regular Envelope). This result, of course, might still strike us as a surprising one. But the fact that a claim has surprising implications is not in itself any reason to reject it.

A second worry that might be raised about my simply accepting this implication of my solution arises from a claim about the source of our resistance to Unger’s argument. The claim is that our resistance to Unger’s argument is grounded in a resistance to the argument’s implication that morality can be so demanding of us. We believe that morality can demand something of us, on this account, but we do not believe that it can demand too much. If all we have to do is help the isolated bird-watcher or two that we might come across during the course of our lives, morality will not be too demanding. But if we would have to help a starving child whenever our assistance would be uniquely directed, then, because there are so many starving children, morality could be made to require too much of us. If the famine relief consortium sent me a new letter with a uniquely directed envelope singling out a new child every week, for example, then I would have to spend $100 every week on famine relief. And intuitively, this critic maintains, that is just too much.

There are a few reasons to reject this further objection to my simply accepting that my solution entails that it would be wrong not to send money in the case of the Uniquely Directed Envelope. First, it is not at all clear that it would be possible, as a practical matter, for famine relief agencies to divide their labor up in the way that the Uniquely Directed Envelope requires. Second, it is not at all clear that it would be morally permissible for them to do so, even if it were possible. To enure that my aid would be uniquely directed, after all, they would have to actively prevent anyone else from helping the young child whose life would be saved by my sending them money and would have to actively prevent me from helping any of the young children whose lives would be saved by other people sending them money. But more importantly, there is simply no reason to assume that morality could never demand so much of us. Indeed, as long as you agree that it would be wrong not to help the bird-watcher in the Vintage Sedan case, it is difficult to see how you could avoid agreeing that, at least in certain extreme circumstances, morality could, in fact, demand as much of you as might be demanded of you in certain extreme circumstances if my solution to Unger’s puzzle is accepted.

The worry about my solution, after all, is that if it commits me to agreeing that it would be wrong for you not to send one check to the famine relief consortium if you received one Uniquely Directed Envelope from it, then it will also commit me to agreeing that it would be wrong for you not to send many more such checks if you received many more Uniquely Directed Envelopes. It will commit me, that is, to the claim that morality can be very demanding. But if you agree that it would be wrong not to save the one bird-watcher in the single Vintage Sedan case, then this will equally commit you to the claim that morality can be very demanding in just the same way. If the world were such that every week you had to drive right by a bird-watcher who would die unless you incurred a $100 loss in saving him, for example, then you would have to keep saving bird-watchers and keep losing $100. If our resistance to Unger’s argument arises from a belief that morality could never, under any circumstances, demand a great deal of us, then I accept that my solution to Unger’s puzzle may prove unable to accommodate this belief.8 But, if it is unable to do so, it seems far more reasonable to view this as a problem with the belief rather than as a problem with my position.

My solution has another implication that might seem to be a cause for worry. Setting aside uniquely directed envelopes and returning to the way that famine relief operates in the world as it is, my solution implies that if there are a great number of starving children, then you don’t have to help any of them, but if there are a much smaller number of starving children, then you do. My solution has this implication for the following reason: suppose that there were only five starving children in the world. You receive a letter from UNICEF correctly informing you that if you send them $100, then one fewer of the these five will die prematurely. In this case, even if there is no one child toward whom your aid would be uniquely directed if you were to send it, it would still be the case that, for each of those five children, the difference between their chances of survival if you do send the money and if you don’t is quite substantial. So for the same reason that it would be wrong not to help in the Russian Roulette case, it would be wrong not to help in the Envelope case if there were a very small number of starving children in the world. But this means that my solution implies that whether or not you must send money in the case of the (non-uniquely-directed) Envelope depends on how many starving children there are.

I believe that this is, indeed, an implication of my solution. And I accept that it is a surprising one. But, again, the fact that a claim has a surprising implication is not, in itself, any reason to reject the claim. We must have some reason to believe that the implication is mistaken. And I do not see any reason to think that it is. Indeed, if we look at other structurally similar cases, we should see that our intuitions commonly accept the idea that the numbers can matter in just way that my solution makes them matter in the case of non-uniquely-directed famine relief. Consider, for example, the following two cases:

Mine Shaft: five miners are trapped as a result of an accident in a mine shaft. They will be stuck there for another hour before rescuers can reach them. There is a large rock near the entrance to the shaft that is about to roll into the shaft. If it does roll in, one more miner will die in the accident than will otherwise die. You can use your car to push the rock away from the shaft, but doing so will cause $100 worth of damage to your car.

Factory Emissions: there is a factory that produces emissions that are damaging to the ozone layer. There is a large rock near the entrance to the factory that is about to roll into a container. If it does roll into it, the container will release more emissions than would otherwise be released. The amount of extra damage to the ozone layer that will result will make it the case that one more person somewhere in the world will die prematurely of skin cancer than would otherwise die prematurely. You can use your car to push the rock away from the shaft, but doing so will cause $100 worth of damage to your car.

It seems clear that it would be wrong not to push the rock away in the case of the Mine Shaft. But it is not at all clear that it would be wrong not to push it away in the case of the Factory Emissions. In this case, it is the difference in numbers that makes the difference, and so it is no problem for my solution if it entails that the numbers would make the difference in just the same way in the case of famine relief.

A final objection to my solution concerns the numbers that I have been working with in developing and deploying it. My version of the Envelope case, in particular, says that it will cost you $100 to ensure that one fewer child will die prematurely. But, according to Unger, $100 is actually enough to ensure that the number of children who die prematurely will be reduced by over thirty. A more accurate version of the Envelope, this objection points out, would therefore be one in which you only have to send $3 to ensure that one fewer child dies prematurely. And, it might then be thought, even if you would not be obligated to send $100 to provide life-saving aid that will not be uniquely directed, it doesn’t follow that you wouldn’t be obligated to send a mere $3 to provide such aid. Once my solution is correctly calibrated to correspond to the basic facts about the real costs of famine relief, therefore, this objection maintains that the solution’s foundation is undermined.

Even if we suppose that it only costs $3 to reduce by one the number of children who die prematurely, however, the argument underlying my solution will remain intact. Suppose, after all, that in deciding which means of transportation to use in sending your child to camp, you found that there was a one in ten millionth difference in the fatality rates between the bus and the train. I doubt that anyone would think you acted wrongly in deciding to go with the cheaper fare, even if this only saved you three dollars. This shows that even your own child does not have the right that you reduce by one in ten million his chances of immanent death even if it will only cost you three dollars to do so. But surely if your own child does not have the right to demand this of you, then other children do not either. Even if it only takes the act of sending $3 to UNICEF to reduce each child’s chance of immanent death by a tiny fraction, therefore, it will still be the case that none of the children at risk of immanent death have the right that you send the money. And so, even on this understanding of the costs involved in saving lives, it will turn out that the reason that it would be wrong not to provide assistance in the case of Vintage Sedan will not, in fact, show that it would be wrong not to provide assistance in the Envelope case. The stark difference in our intuitive response to the two cases will remain consistent. Unger’s puzzle will remain solved.

V

At the risk of stating the obvious, I would like to conclude by emphasizing that in this paper I have not been defending the claim that it would be postively immoral for you to give money to famine relief in Unger’s Envelope case. Nor have I been denying that giving money in such a case is morally better than not doing so. Indeed, I have not even been denying Unger’s normative conclusion that it would be positively wrong for you not to give aid in such a case. There are many grounds that might be appealed to in defending the existence of a postive duty on your part to give more money to famine relief efforts than you presently give, and I have not been insisting that no arguments appealing to any such grounds are cogent.

Instead, I have been making a more limited claim: that the particular argument that Unger provides for believing that your declining to give such aid would be wrong, the argument that rests on the claim that this judgment is in conflict with our judgment in the Vintage Sedan case, should be rejected. Unger’s argument should be rejected because there is a morally relevant difference between cases in which your assistance would be uniquely directed and those in which it would not. The claim that this difference is a morally relevant one is supported both at the particular level of our reactions to specific and suitably devised variant cases (the change from Unger’s Envelope case to my e-velope case makes a difference to our moral intuitions, and my e-velope cases involves uniquely directed aid while his Envelope case does not) and at the general level of the moral principles that best account for our particular intuitions (they are best accounted for by a principle on which the bird-watcher in the Vintage Sedan case has a right to your assistance because your aid would be uniquely directed toward him whereas none of the starving children that UNICEF provides assistance to have a right that you send money to UNICEF because your doing so would not provide uniquely directed aid to any of them). My conclusion is therefore, in one important respect, a narrowly focused one: it is the conclusion that Unger’s particular defense of the immorality of declining to give money to famine relief is unsuccessful. But since Unger’s argument is now widely taken to be the most philosophically sophisticated defense of the claim that your declining to give such aid would be wrong,9 this conclusion, although limited, is important nonetheless.10

Notes

  1. Philosophy and Public Affairs, Vol. 1, No. 3 (Spring 1972), pp. 229-43.
  2. Unger, Living High and Letting Die: Our Illusion of Innocence (Oxford: Oxford University Press, 1996), p. 24. All page references in the text are to this book. Brad Hooker, among others, has referred to Unger’s treatment of the issue raised by this puzzle as the ‘centerpiece’ of Unger’s book, and Unger himself has referred to this characterization of his work as an accurate one (Brad Hooker, ‘Sacrificing for the Good of Strangers — Repeatedly’, Philosophy and Phenomenological Research, Vol. LIX, No. 1, March 1999, p. 177; Unger, ‘Replies,’ Philosophy and Phenomenological Research, Vol. LIX, No. 1, March 1999, p. 203).
  3. Unger acknowledges as much (24).
  4. In Unger’s version of the cases, it will cost you more money to save the injured bird-watcher and he will lose only his leg, and not his life, if you fail to act, and it would cost you less money to ensure that one fewer child dies of famine.
  5. I would like to thank Graham Oddie for his help in constructing this example.
  6. A further case provides some additional support for this claim: suppose we stipulate that if (and only if) you decide not to save the bird-watcher’s life, you will go home and conceive a new person whose life will contain just as much happiness as the life of the bird-watcher would contain if you instead decided to save him. On this version of the story, too, the overall state of affairs that results from not saving the bird-watcher is just as good as the overall state of affairs that results from saving him. Our moral common sense surely indicates that it would still be wrong for you not to save the bird-watcher in this variant case. The appeal to impersonal reasons cannot explain why this would be, since the resulting state of affairs will be impersonally just as good if you decline to save the bird-watcher as it will be if you save him. But the appeal to personal reasons can easily explain our moral common sense reaction to this case. Whatever right the bird-watcher has to your assistance, after all, he surely has it regardless of whether or not you will later conceive some other person who will be just as happy as he would have been had you saved him.
  7. Although he puts things in terms of the wrongness of not helping rather than in terms of violating a right to assistance, for example, the principle that Singer appeals to in his famous 1971 article on famine relief employs just these limits: it is wrong to act when doing so would prevent something very bad from happening without costing anything significant.
  8. I say ‘may’ prove unable to do so because one might argue that in the case where you know that you will continue to drive past dying bird-watchers, it might be argued that after you have incurred a significant enough total cost, you need not save any more. If so, then by parity of reasoning the same would presumably be true in the case of famine relief, in which case my solution would not have to entail that you would have to keep sending money for famine relief.
  9. E.g., ‘Unger’s discussion . . . must now be the basic text for study of the issues involved’ (Hooker, ‘Sacrificing the Good’, p. 177).
  10. I would like to thank Graham Oddie, Michael Tooley, Mike Huemer, and members of the Faculty Research Group in Ethical Theory at the University of Calgary for their helpful comments on earlier versions of this paper.
  • A Priori
    College of Arts

    University of Canterbury
    Private Bag 4800, Christchurch
    New Zealand
  • Follow us
    FacebookYoutubetwitterLinked In