Abstract

Theron Pummer's impressive new book The Rules of Rescue seeks to defend effective altruism without taking on the controversial moral theoretical commitments. Through an exploration of the framework of requiring reasons and permitting reasons that is the backbone of his argument, this article raises some doubts about how successful Pummer's strategy of avoidance can be.

Arguments from a Shallow Pond

As Peter Singer famously pointed out, “if I am walking past a shallow pond and see a child drowning in it, I ought to wade in and pull the child out. This will mean getting my clothes muddy, but this is insignificant, while the death of the child would presumably be a very bad thing.”1

One way to explain this judgment is as the application of a consequentialist principle that says I am morally required to bring about the best outcome, impartially judged. That would be stronger than the principle that Singer took his example to illustrate. Singer's principle says that “if it is in our power to prevent something bad from happening, without thereby sacrificing anything of comparable moral importance, we ought, morally, to do it.”2 But the consequentialist principle would certainly license the conclusions Singer wanted to draw concerning the extent of our obligations to distant people threatened by severe poverty.

The consequentialist principle would also license a conclusion that Singer did not at the time draw, but which some of his contemporary philosophical inheritors do infer. It would license the conclusion that I am morally required to be an effective altruist. Effective altruists think that we should (at least to a significant degree) benefit others as much as possible with our resources, for example, by directing our charitable donations to organizations that boast the highest ratios of good done per amount donated. Since anything less than the most efficient conversion of the resources at my disposal into goodness fails to bring about the best outcome, impartially judged, the consequentialist principle implies effective altruism as a moral requirement.

Although the consequentialist principle thus offers a simple argument for effective altruism, it is controversial. Certainly its intuitive power is far surpassed by that of the example it is supposed to explain. (So is the intuitive power of the principle Singer himself identified.) Insofar as they are interested in winning people over to the effective altruist cause, then, effective altruists might want to appeal to the example but skip the principles in defense of their view.

Can this be done? The Singer-style conclusion that our obligations to distant people in dire poverty are as strong as our obligations to those nearby does look temptingly within reach via reflection on the example alone. It just needs to be shown that the situation we are in vis-à-vis the distant needy is morally analogous to the shallow pond, so that we are subject to analogous requirements to save the lives, if we can, of those whose lives are threatened by severe poverty. Indeed, this is more or less how Peter Unger proceeded in Living High and Letting Die—though he went on to infer controversial moral principles anyway.3

So effective altruists might reasonably hope that the same method can be put to work in defending their view. In the most sustained philosophical defense of effective altruism that I know of, that is precisely what Theron Pummer tries to do.4 He presents shallow-pond-like examples, confident that we will agree that agents in the examples have a “requiring reason” to rescue people in danger. And then he tries to show that this judgment doesn't change even when the examples are varied by introducing factors that might be thought to explain why our obligations to distant people threatened by severe poverty differ from our obligations to the child in the shallow pond.

The notion of a requiring reason is just the innocuous-seeming notion of a fact that tends to give rise to a moral requirement, and no appeal to consequentialism or any other normative moral theory is made. As Pummer suggests, the advantage of this approach is that it allows him to defend morally significant conclusions without taking on controversial theoretical commitments.5

Pummer's Argument

In Pummer's telling, then, effective altruism can travel light, theoretically speaking. But can it really? In what follows, I shall raise some doubts.

I will start by setting out the argument in more detail. Pummer's analysis of the examples that fuel his argument employs a framework of requiring reasons and permitting reasons.6 As I have just mentioned, a requiring reason to Φ is a reason that contributes toward making Φ-ing morally required. Unsurprisingly, requiring reasons tend to produce requirements. But they don't always produce requirements because sometimes they are opposed by permitting reasons. A permitting reason to Φ is a consideration that contributes to making Φ-ing morally permissible by weakening or countering requiring reasons to choose an alternative to Φ-ing.

Thus, the requiring reason not to push a person to the ground tends to generate a requirement not to do that. But in some circumstances, it may be defeated by permitting reasons such as the reason to save one's own life from some deadly threat, or the reason to prevent someone else from being killed by a boulder—itself a requiring reason that also functions, in such circumstances, as a permitting reason.7 In such circumstances, the agent has a permission to push the person to the ground despite the presence of the requiring reason not to.

Pummer's idea is that avertable threats to people's lives constitute requiring reasons for those in a position to avert the threats to help them, reasons that generate moral requirements except where they are defeated by permitting reasons. This does seem a very plausible analysis of what is going on in the shallow-pond example, and in other examples of rescuing nearby strangers in peril that Pummer uses to illustrate his argument. It explains the moral requirement to wade into the shallow pond and save the child's life as the result of a requiring reason to save the child, together with the absence of any sufficiently strong permitting reason not to. And it does not seem to depend upon any obviously controversial commitments in moral theory, such as the consequentialist principle.

Against the background of this framework, Pummer goes on to defend a significant moral requirement of effective altruism. His argument has three key premises. The first is what he calls the ubiquity of requiring reasons, which is a result of the dire poverty of millions of people around the world. The life of the child in the shallow pond is worth no more or less than the lives of the people in life-threatening poverty around the world, so their plight gives rise to requiring reasons just as the child's does, insofar as we can help them. Moreover—this is the second premise—these requiring reasons aggregate, so that the requiring reasons to save each member of one group of people together defeat the requiring reasons to save each member of a smaller group of different people. The third premise is that there aren't permitting reasons not to act in accordance with the ubiquitous requiring reasons to help those in need that are strong enough to preclude a significant requirement of effective altruism.

Pummer defends the ubiquity thesis by considering factors that might be invoked to explain why opportunities for lifesaving charitable giving don't give rise to requiring reasons in the way that opportunities to wade in and save the lives of children in shallow ponds do. These factors include the physical distance from the agent of the people who need help, the vividness to the agent of their plight, and whether or not the agent is the only person in a position to effect the rescue. Pummer argues that the factors are irrelevant, deploying Unger's method of devising examples that differ only in respect of such factors and observing that intuitive judgments about the examples find them to be morally indistinguishable.

One such example is as follows:

While out for a hike in a country you have never been to before, you see on the digital map you are using a red dot indicating that there is an emergency taking place at distance X from you. You tap the dot for a brief description of the situation. It turns out that, on the other side of a tall brick wall, a boulder is hurtling toward stranger A, who is stuck in the boulder's path. You cannot reach A yourself but realize that you are able to save A's life using your phone. For $Y you can tap a button on the screen of your phone that causes a large bulldozer to move in front of the boulder, saving A without doing any damage to the bulldozer or anything else. Many other people can similarly help A, but you are certain none of them will.8

It makes no difference, Pummer contends, whether X = 10 feet or X = 1,000 miles. If you are required to pay $Y to save A in the one case, then you are required to do the same in the other.9 Similarly, it makes no difference whether anyone else is in a position to help A,10 whether tapping the dot opens “a live video of A screaming in terror while trying to escape from the boulder” (a detail that you find it difficult to put out of your mind),11 or whether A's plight is the result of injustice.12

The only factors that Pummer thinks can make a moral difference are uncertainty (the strength of a requiring reason to save someone is discounted by the probability that one's efforts will not save them) and the way in which the risks of failure of one's lifesaving effort among a given number of people are distributed so as to produce a given probability of saving some number of them for the relevant amount of resources.

Of course, we might think there are permitting reasons that explain why we are not on the hook for charitable giving as we are for wading into the shallow pond. But, as I noted, the third premise of Pummer's argument is that there aren't strong enough permitting reasons to preclude a significant requirement of effective altruism. He identifies permitting reasons of cost, autonomy, and the special connections we have to particular people and causes as possible defeaters of the requiring reasons we have to help others. Pummer accepts that these may be strengthened on individual occasions by the lifetime costs that would be incurred if, on each of those occasions, the permitting reasons were too weak to defeat the requiring reasons. But he argues that they are not so strong as to defeat the ubiquitous requiring reasons in a significant range of real-world opportunities to donate resources or volunteer time and energy to lifesaving causes.

Despite the permitting reasons, then, “a significant proportion of us are required to spend at least 10 percent of our time and money helping others, and if we do the minimum required by allocating exactly 10 percent to helping others, we must allocate it in the most cost-effective way available.”13

Ways of Making a Moral Difference

Let the “donation situation” be a situation in which an agent can give to charity or volunteer or otherwise engage to any degree in “the effective altruist project of using time, money, and other resources to help others the most,”14 where any such use of time, money, and other resources will prevent severe harms befalling others. What Pummer is trying to show, of course, is that the donation situation is analogous to cases like the shallow pond, in that by default it is wrong not to help, thanks to the presence of requiring reasons to do so.

It hardly needs to be pointed out that people don't behave as if these cases are morally analogous. If they did, effective altruists wouldn't need to make their arguments. There are two main ways in which people might try to justify their ineffectively altruistic or altogether non-altruistic behavior. They might deny that requiring reasons obtain in the donation situation. (Charitable giving is supererogatory!) Or they might accept that they obtain, but think there are good permitting reasons not to comply with them. (You can't expect me to sacrifice everything in order to help the poor!)

Pummer's first and third premises correspond to these two kinds of attempted justification. Although the second premise (that requiring reasons aggregate) is crucial for the vindication of effective altruism, rather than a version that is more sanguine about inefficient helping, I want to focus on the first premise—the ubiquity of requiring reasons. That premise sounds quite theory-laden, and it will be part of what I want to say that when we scrutinize the notion of a requiring reason more closely, we find some support for doubting Pummer's conclusions. But Pummer defends the premise without any appeal to theory, and my initial doubts about it can be raised in terms that tread with a theoretically light step to match Pummer's own.

As we have seen, Pummer's strategy for defending the ubiquity of requiring reasons involves controlling for the factors he thinks people might cite as disanalogies between shallow-pond-type examples (represented in Pummer's argument by the case of the hiker with a small value for X, a live video stream, no other potential saviors, etc.) and the donation situation. If there are requiring reasons in shallow-pond-type situations, and Pummer is right that none of the factors he considers makes a morally relevant difference, then there are requiring reasons in the donation situation too.

Such strategies always run the risk that someone will come up with a factor you hadn't thought of. But I share Pummer's confidence that no one such factor of the sort he focuses on—empirically tractable things like proximity and vividness—is going to be the difference-maker that explains and justifies divergent reactions to the shallow pond and the donation situation.

I do think, however, that such factors can help to explain moral differences of the sort Pummer aims to deny. That is not to say that their presence reliably implies a moral difference. Nor, when they help to explain a moral difference, is that because they are in themselves morally significant. When they help to explain a moral difference, to my mind, it is rather because they make it reasonable to suppose that the agent has registered or engaged with the plight of those in need in a certain way. And that is what's morally significant in this context. People who engage in this way but don't help look as if they can be guilty of the sort of wrongdoing that I would be guilty of if I left the child in the shallow pond to drown. But in the absence of the relevant sort of engagement, it is much less clear that they can be charged with that sort of wrongdoing.

If that is right, it is unsurprising that I can be guilty of shallow-pond-like wrongdoing even when none of the factors that Pummer considers are present. For those factors are not necessary for the relevant sort of engagement. There may be that sort of engagement even though the person in danger is physically distant and their plight is not vivid to the agent, who is only one of many potential rescuers, and the danger is not the result of injustice.

Imagine, for example, that as George is wandering around New York, he calls a wrong number and ends up talking to Roman, a stranger who answers from his home in Massachusetts. In the course of their call, Roman suffers a ruptured aneurysm and rapidly loses consciousness. Insofar as George's failing to help in this situation—for example, by calling an ambulance—seems to be a shallow-pond-like wrong, that is, I submit, because we take him to be engaged in the relevant way with Roman despite the distance, the lack of vividness, and so on. The same goes for the analogous versions of Pummer's hiker examples.

At the same time, though, since most (perhaps all) of the factors do not seem sufficient for the relevant kind of engagement, it also would not be surprising to find that even in their presence, a person who fails to help can be innocent of shallow-pond-like wrongdoing. And, as it seems to me, that is what we do find. A person in danger can be physically near, you could be the unique potential rescuer, and the danger could be a result of injustice, and yet your failure to help may not be wrongful in the way it is in the case of the shallow pond. Cycling on your daily commute, for instance, you don't register the danger that a nearby 5-year-old is in, even though you would be in a position to notice and avert it if you were on the lookout for it. The child has slipped unnoticed through the faulty gate of a nearby school—the result of years of unjust underfunding—and is about to fall into the notoriously dangerous river that runs next to the cycle path. It seems to me that your failure to save the child is not a shallow-pond-like wrong. It is not a shallow-pond-like wrong even if you yourself have for many years highlighted the danger of children escaping from the school to authorities, telling them that it was inevitable that a child would be killed sooner or later.

It is because I think the relevant sort of engagement is the morally significant thing, then, that I share Pummer's confidence that no single empirically tractable factor will turn out to be the difference-maker that justifies divergent reactions to the shallow pond and the donation situation. No such factor will be the unique determinant of whether requiring reasons to help obtain. Perhaps some such factors are sufficient for the requisite engagement and so sufficient for a moral difference. Of those Pummer considers, the most likely contenders are vividness to the agent and what he calls “personal encounter of the sort had when you see, hear, or communicate with another person.”15 But no such factor seems likely to be necessary for the requisite engagement, even if, in any given situation, it forms part of the reason to think that the requisite sort of engagement obtains.

However, it is also because I think that engagement is the morally significant thing that I don't think Pummer's arguments show that the donation situation is morally analogous to the shallow pond—that if there is requiring reason to help in one, then there is requiring reason to help in the other. Of course, the typical donation situation differs in many respects from the case of the child too (it is what Pummer calls an “unclean case”).16 But the donation situation may just as well be like the case of the child as like the shallow pond in respect of the degree to which agents are engaged with the plight of those whom they might save. I would venture to say that it usually is.

What Are Requiring Reasons?

An obvious response to what I have said is that the absence of the kind of engagement I have appealed to can itself be wrongful, so it can't be necessary for the wrongfulness of a failure to help. I will get to that below. First, though, I want to address a different concern, which is that the notion of engagement I have invoked is too vague and too psychologically contingent to bear much argumentative weight.

The relevant notion of engagement can be made a bit more substantial and connected more closely with matters of moral significance, I think, via reflection on the framework of requiring and permitting reasons. Pummer himself is content to endorse that framework as a satisfying explanation of intuitive judgments about a range of examples of lifesaving rescue, that at the same time avoids problems with more consequentialist approaches—and to leave it at that.17 But the framework invites questions. What, we might ask, makes requiring reasons requiring reasons? Why, that is, when they win out in moral reasoning, when there is no permitting reason sufficient to defeat them, do we have moral requirements rather than reasons that it may be in some sense unreasonable or even morally unattractive but not morally wrong to defy? How do permitting reasons work? I will suggest that answering these questions will tell us something about why we are so confident in our judgments about the shallow pond—and why engagement might matter.

To learn what gives requiring reasons their requiring nature, we need to say something about moral requirements themselves, the conclusions that requiring reasons contribute to establishing. One thing that seems clear is that when a person acts contrary to a moral requirement, they act in a way that does not respond to others as their value gives them sufficient reason to respond. But we can fail to respond as others’ value gives us sufficient reason to respond without doing anything contrary to moral requirement, as we do when we merely don't love a person, or don't do something supererogatory for others, or simply don't tend to think about others much at all.18 So appeal to the value of others isn't enough by itself to explain what is distinctive about action contrary to moral requirement.

Instead, we need to distinguish something like the morally mandatory minimum response to that value from other responses that it also warrants. One way of making headway here is by thinking about our reactions to moral wrongs, and what they seem to focus on. We take acts that manifest a failure to bear toward another the minimum morally necessary concern distinctively to justify the negative “reactive attitudes” of indignation, resentment, and hurt feelings that are partly constitutive of blame, and the most familiar indicator of the presence of an interpersonal moral requirement.19 These attitudes express recognition of, protest against, and reassertion of status in the face of a distinctive kind of depreciatory valuation of others that falling below the mandatory minimum involves. In the situations that we are concerned with—situations exemplified by the shallow pond, in which others have fundamental interests (e.g., in survival) at stake in what the agent does—this depreciatory valuation amounts to what we might think of as a failure of basic recognition or solidarity. Call it “recognition failure.”

There are, of course, many competing accounts of what exactly such recognition and solidarity call for, and so of what counts as recognition failure. But whatever the right account is, we can say that an undefeated requiring reason in a shallow-pond-like situation will be such that unexcused failure to comply with it implies recognition failure. Accordingly, it makes sense to think of requiring reasons of the sort that obtain in the case of the shallow pond as reasons whose presence establishes a presumption of recognition failure if they are not complied with.

There are permitting reasons because the presumption can be overturned, thanks to the possibility of circumstances in which acting contrary to the reasons does not, after all, imply the depreciatory valuation that would typically be implied. For example, that an act will harm someone establishes the presumption in favor of not performing it as a condition of due recognition and solidarity. But if the harm is done in order to prevent a much greater and irremediable harm, the presumption may be defeated. The agent who does the harm can point to the much greater harm prevented as evidence that what they did implied no depreciatory valuation of the person harmed, as it would have had they done the harm merely in order to prevent their tea from getting cold, say. There was no recognition failure after all.

That requiring reasons work like this shows both in the fact that we suspend blame of agents who have done harm where they can give a justification for their action that defeats the presumption of recognition failure, and in the fact that we will not normally suspend blame unless they can provide such a justification. Meanwhile, if a person is causally responsible for harm but we don't expect them to provide a justifying explanation for it as a condition of forestalling blame, then that shows that we don't think they were subject to a requiring reason in the first place, as when the conductor brings in the whole orchestra at the start of Siegfried's funeral march and gives you a heart attack.

As Pummer notes,20 only certain kinds of considerations seem capable of counting as permitting reasons. That makes sense: a consideration can count as a permitting reason, according to the account I am developing, only if it can contribute to justifications that defeat the presumption of recognition failure. Considerations of minor experiential pleasure, say, or aesthetic considerations cannot in themselves do this. In the practical reasoning that either avoids or defeats the presumption of recognition failure, then, such considerations seem incapable of opposing requiring reasons. That is true even if, taken in isolation, they might seem to weigh very heavily. For example, even if the aggregated weight of the reasons to kill a person provided by the minor experiential pleasure of very many people would be very great if they counted, appealing to them cannot defeat the presumption of recognition failure by one who kills.21 The relevant kind of recognition of and solidarity with others seems to involve seeing it as distinctively inappropriate to take considerations of minor experiential pleasure to count at all in favor of killing them. Requiring reasons operate in this way as “exclusionary reasons”: a requiring reason to Φ not only favors a requirement to Φ, but it also prevents certain other considerations from bearing on the question of whether it is permissible to Φ.22

The connections this account draws between requiring reasons, the presumption of recognition failure that gives standing to hold to account via the reactive attitudes, and the possibility and limits of permitting reasons as defeaters of that presumption accord well with some of the other distinctive features of the framework of requiring and permitting reasons that Pummer observes too. One such feature is the way in which permitting reasons are tied to requiring reasons. As Pummer points out, permitting reasons to Φ don't function by adding up to an overall permission to Φ regardless of whether or not there are requiring reasons not to Φ. They function only to contribute to defeating requiring reasons not to Φ.23 The account explains this. In the absence of a requiring reason not to Φ, we don't expect a person to need a permitting reason to Φ, because Φ-ing does not justify the presumption of recognition failure that a permitting reason would be needed to defeat in the first place.

Another, related feature is that opportunities to prevent significant harm seem much more capable of generating requiring reasons than opportunities to provide “pure benefits” such as gains for people who are already very well-off.24 Again, the account supports this. Failure to provide pure benefits doesn't ordinarily justify any presumption of recognition failure. If someone is doing fine, then my failure to respond to the reason that is constituted by the advantage to them of me providing them with a pure benefit is not normally a presumptive recognition failure, even if the benefit is quite big. But if someone is at risk of significant harm, my failure to respond by Φ-ing to the fact that Φ-ing will save them from the harm seems much more likely to justify the presumption of recognition failure—most obviously, as we will see in a moment, if the significant harm in question will be done by me.

Engagement and Requiring Reasons

Above, I suggested that the donation situation may be less like the case of the shallow pond, in which not helping is (all other things equal) uncontroversially a wrong, and more like the case of the child and the school, which (I think) it is not, or at least not a wrong of the same type. And I proposed that a difference in the agent's engagement with the plight of the victim justifies that doubt.

The account of requiring and permitting reasons I have just set out gives us a way to flesh out that proposal. It tells us that requiring reasons to Φ of the sort that feature in the shallow pond obtain only where the agent's not Φ-ing can reasonably be presumed to imply recognition failure. And the question whether the agent has registered and engaged with a person's plight seems relevant to establishing that presumption.

To see this, start by considering requiring reasons not to engage in direct physical attacks on other people. We can reasonably assume that—at least in relatively well-ordered, peaceful societies—people normally have plenty of courses of action open to them that don't involve any such attacks and yet are compatible with ongoing engagement with the projects, commitments, and relationships that give meaning to their own lives. So if they attack us, that is normally going to be a matter of their going out of their way to do the harm involved, acting on reasons that are not especially important from the point of view of what really matters for them, and so are surely excluded by the reason not to attack—or, worse yet, a matter of a commitment or project that itself implies recognition failure, a kind of defection from morality altogether.

Moreover, insofar as it is an unprovoked, direct, physical attack, undertaken in circumstances involving the sort of encounter that involves eye contact, an attack can hardly be understood except as proceeding from a valuation of the victim as apprehended in their full particularity in the agent's experience. That is, it proceeds from a valuation that is made or persists in circumstances of the fullest possible engagement with the victim.25 There is almost no way, then, for this to be anything but an instance of recognition failure—indeed, not just a failure, but an active denial of recognition and solidarity. It is the fact that this sort of attack takes place in circumstances of the fullest sort of engagement that licenses the greatest confidence here.

Now, not helping someone can have some of these features as well. Suppose, for instance, that you can, simply by saying “Look out!” prevent someone with whom you are chatting on the street from being killed right in front of you by an obvious danger that you have had plenty of time to see taking shape and that represents no threat to you at all. Given that we could reasonably assume that giving the warning is compatible with your ongoing engagement with the projects, commitments, and relationships that give meaning to your life, your not saying anything may be presumed to imply the same sort of recognition failure that a physical attack does. Once again, the circumstances make it clear that your response to the victim is formed or sustained in response to them as apprehended in their full particularity in your experience. So, once again, the presumption of recognition failure seems particularly clearly justified.26

I take it that as we are expected to think about the example of shallow pond, it, too, has these features. The child is drowning right in front of me; the pond is obviously shallow, and it is clear to me that I can effect the rescue very easily and quickly; the muddy trousers can be cleaned or replaced. If I don't rescue the child, then, the presumption that this reflects recognition failure seems amply justified.

Other examples of failing to help, though, do not have features that so obviously justify the presumption. The point is not just that, in these other examples, permitting reasons obtain that defeat the presumption. If I have severe aquagenic urticaria (a rash caused by contact with water) or, for some reason, keeping my trousers clean is my only way to save my sister's life, the requiring reason to save the child in the shallow pond may be defeated, since failing to save the child in that case may turn out not to imply the kind of valuation that amounts to recognition failure.

But the requiring reasons do obtain in such a case insofar as there is a justifiable presumption of recognition failure that is overturned in the circumstances. And there does seem to be one: others would surely take themselves to have standing to hold me to account, corresponding to a responsibility on my part to explain myself. This is what we would expect, since, as in the standard telling of the story of the shallow pond, my engagement with the plight of the child as concretely presented to me in experience can be assumed.

However, as we get further away from the circumstances of engagement that characterize the paradigm of direct, physical attack, the grounds for taking the relevant type of requiring reasons to be present fade along with confidence in the presumption of the requisite sort of recognition failure, the standing to hold the agent to account, and the agent's responsibility to justify herself.

One way of moving away from that paradigm is by reducing the degree to which the person in a position to be helped is at any risk of harm. This is what is going on in the case of pure benefits, which may nevertheless be cases of full engagement. But another way is to reduce the degree of engagement itself. The point is that when the degree of engagement is small enough, it may not be plausible to regard the agent's not helping as presumptively implying recognition failure at all. This is what I think is going on in the case of the escaped child I gave earlier. It is not that I have a permitting reason not to act on the requiring reason to prevent the child from coming to harm. I am subject to no requiring reason to prevent the harm in the first place. If someone asks why I didn't help, I might say “I didn't see.” But that is not citing a permitting reason. It is explaining why I don't need one.

Requiring Reasons in the Donation Situation

For the purposes of determining what this account of shallow-pond-type requiring reasons tells us about the donation situation, then, the question is: How far removed is the donation situation from the paradigm of direct, physical attack and the near necessity of recognition failure that it implies? To what extent can failure to help in the donation situation be regarded as a matter of depreciatory valuation in response to engagement with the plight of those in severe poverty? On the account I have given of requiring reasons, if there are to be requiring reasons to help others of the sort that are present in the case of the shallow pond, then the answers to these questions will have to be “Not far” and “To a significant extent,” respectively.

Note that the examples presented as analogous to the case of the shallow pond in the course of Pummer's argument against the relevance of distance, vividness, and so on encourage the presumption of recognition failure by virtue of the fact that they present us with agents who do appear to have engaged with the plight of those they are in a position to save. They do this partly by virtue of the fact that they give us full knowledge of the relevant situations. We feel as if the protagonists may be blamed for not helping because we feel confident that the only possible explanation of their actions in the circumstances must be a depreciatory response to the victims, just as we are in the case of the shallow pond.

But donation situations are not all like that. We certainly do not often know to what degree other people have registered and engaged with the plight of those who would be helped by their charitable donations or volunteering.27 We may not be sure even in our own case.

It might be supposed that the donation situation for those who are aware of the extent and severity of global poverty—as Pummer's audience will be—cannot be like the case of the escaped child, where the presumption of recognition failure doesn't obtain precisely in virtue of a lack of awareness of the plight of this particular child itself. So ordinary moral thinking will validate the requiring reasons to help others that are needed to vindicate Pummer's defense of effective altruism after all.28 But I think things aren't so clear as this makes them seem. The awareness of global poverty that Pummer's readers can be expected to have is like the awareness of the inevitability of an accident, in the case of the child, that you have been communicating for years to the authorities. It is rare that the donation situation involves awareness of the particular plight of particular individuals, a shallow-pond-style depreciatory valuation of whom could therefore be inferred from the failure to help them.

Negligence and Theory

I should now address the obvious response I set aside above to the proposal that engagement matters for determining whether not helping is a shallow-pond-like wrong. That response is that the absence of the kind of engagement I appealed to can itself be wrongful, so it can't be necessary for the wrongfulness of a failure to help. In just the same way that I tried to substantiate my own proposal via an account of requiring reasons as justifying presumptions of recognition failure, moreover, so someone giving this response might point out that failing to engage with the plight of a person in need can itself constitute recognition failure. So the absence of engagement is not sufficient for the absence of requiring reasons, and questions of engagement are not therefore relevant to the ubiquity of requiring reasons.

In one way, this seems right to me. Clearly, the absence of the sort of engagement I have in mind can be a kind of negligence proceeding from a depreciatory valuation, as when an abusive parent does not even think to check on the welfare of their child in the bath upstairs. But it matters for such cases that there is a normal and conventional expectation of both engagement and the sort of conduct that would normally flow from that engagement in an appropriately caring person. We expect parents as such to be attentive to what is going on with their (young) children, to be engaged with their lives and on the lookout for risks to them. This kind of example does not show that opportunities to help in general suffice for requiring reasons. To get that result, we would need a broader account of negligence and the proper limits of responsibility for strangers analogous to parental responsibility for children. But such an account could not, without begging the question, simply assume that requiring reasons obtain whenever there is need and the opportunity to meet it.

It is tempting to insist that inattention to the plight of the world's most needy must be a culpable failure to recognize the requiring reasons given by that plight, just as the abusive parent's inattention to his child is. Surely, we might think, the plight gives rise to a requiring reason whether or not anyone actually engages with it. The unacceptable alternative is to think that it matters morally only if those in a position to help happen to be interested.

But if I am right, this insistence cannot be defended on the basis of ordinary intuitions indicating the irrelevance of distance, vividness, and so on—not, at any rate, insofar as it is supposed to vindicate the charge that failing to be an ineffective altruist is to any degree a shallow-pond-like wrong. As I have suggested, these intuitions don't by themselves support the charge of shallow-pond-like wrongdoing in the donation situation. Nor can the insistence be defended on the basis that severe need, or severe need coupled with an agent's opportunity to help, obviously suffices for requiring reasons that tend to make not helping a shallow-pond-like wrong. As long as the account I have sketched of requiring reasons and their connection to recognition failure is not obviously mistaken, the sufficiency of need (or need + opportunity to help) for requiring reasons is itself not obvious.29

All of this is consistent with the possibility that not helping in the donation situation, not being an effective altruist, is a wrong unlike the wrong of not helping the child in the shallow pond. And, of course, my account of requiring reasons and the connection to recognition failure may be mistaken. I have certainly not offered anything close to a full defense. But, either way, there is no prospect of defending effective altruism without packing some theoretically heavier bags than Pummer's argument brings. Excursions into deeper moral theory are unavoidable.

Notes

I am grateful to Theron Pummer and the participants in the Winter 2023 online workshop on effective altruism for their helpful comments and suggestions. I owe special thanks to Jonas Hertel for his detailed comments on an earlier version of this paper.

4.

Pummer, Rules of Rescue.

5.

Pummer, Rules of Rescue, 4.

8.

Lightly adapted from Pumer (Rules of Rescue, 107–08).

13.

Pummer, Rules of Rescue, 205

17.

Pummer, Rules of Rescue, 46.

18.

Compare R. Jay Wallace's challenge to Joseph Raz's account of moral reasons in “The Rightness of Acts and the Goodness of Lives.” 

19.

The concept of a reactive attitude is, of course, due to P. F. Strawson. See his “Freedom and Resentment.” 

20.

Pummer, Rules of Rescue, 21.

21.

This is, of course, a way of putting a familiar point. For a seminal presentation, see T. M. Scanlon (What We Owe, 234–36).

22.

The term “exclusionary reason” is due to Joseph Raz (see Raz, Practical Reason, 35–48, 178–99), but the idea does work in a range of moral philosophical theories. See, for example, Barbara Herman (Moral Habitat, 104–07); Scanlon (What We Owe 50–55, 156–57; McDowell (“Are Moral Requirements”).

24.

Pummer, Rules of Rescue, 21.

25.

Notwithstanding Murdochian thoughts about the impossibility of such a valuation if the engagement involves seeing the victim as they really are. The kind of engagement I have in mind doesn't presuppose seeing the victim in that way.

26.

Perhaps this is one source of the intuition about the equivalence of doing and allowing that James Rachels appeals to in his influential paper on euthanasia. See Rachels (“Active and Passive”).

27.

Or to what degree their not doing so reflects commitments and projects that themselves imply depreciatory assignments of weight to the reasons given by that plight. I shall say more about this below.

28.

Thanks to Jonas Hertel for pressing me on this point.

29.

I am sure that to many it will seem just obviously a mistake at the level of moral theory to think that whether a requiring reason obtains or not could depend on the brute psychological fact that the agent has registered and engaged in the right way with the need that must surely be the basic ground of the reason. The impulse to think they cannot stems, I think, from metaphysics of requiring reasons that takes them to be identified and assigned their reason-giving force in the first instance from an impersonal or God's-eye perspective. Against the background of that metaphysics, it is hard to see how it could rationally be supposed that any need with the force of a requiring reason in the impartial perspective nevertheless doesn't stand as a requiring reason for any particular agent, even if justifications or excuses may be offered on behalf of an agent who fails to recognize or act upon it. But the metaphysics is contentious: there are both familiar objections to it and philosophically reasonable alternatives to it. (See, for example, Williams, “Point of View”; Raz, Value, Respect). So however obvious it may seem to some, it cannot be taken for granted.

References

Gert, Joshua. “
The Distinction between Justifying and Requiring: Nothing to Fear
.” In
Weighing Reasons
, edited by Errol Lord and Barry Maguire,
157
72
.
Oxford, UK
:
Oxford University Press
,
2016
.
Gert, Joshua. “
‘Requiring and Justifying’: Two Dimensions of Normative Strength
.”
Erkenntnis
59
(
2003
):
5
36
.
Herman, Barbara.
The Moral Habitat
.
Oxford, UK
:
Oxford University Press
,
2021
.
McDowell, John. “
Are Moral Requirements Hypothetical Imperatives?
” In
Mind, Value, and Reality, 77–94
.
Cambridge, MA
:
Harvard University Press
,
1998
.
Pummer, Theron.
The Rules of Rescue
.
Oxford, UK
:
Oxford University Press
,
2023
.
Rachels, James. “
Active and Passive Euthanasia
.”
New England Journal of Medicine
292
(
1975
):
78
80
.
Raz, Joseph.
Practical Reason and Norms
. 2nd ed.
Oxford, UK
:
Oxford University Press
,
1990
.
Raz, Joseph. “
The Rightness of Acts and the Goodness of Lives
.” In
Reason and Value: Themes from the Moral Philosophy of Joseph Raz
, edited by R. Jay Wallace, Philip Pettit, Samuel Scheffler, and Michael Smith, 385–411.
Oxford, UK
:
Oxford University Press
,
2004
.
Raz, Joseph.
Value, Respect, and Attachment
.
Cambridge, UK
:
Cambridge University Press
,
2001
.
Singer, Peter. “
Famine, Affluence, and Morality
.”
Philosophy & Public Affairs
1
, no.
3
(
1972
):
229
43
.
Scanlon, T. M.
What We Owe To Each Other
.
Cambridge, MA
:
Harvard University Press
,
1998
.
Strawson, P. F.
Freedom and Resentment
.” In
Free Will
, edited by Gary Watson,
72
93
. 2nd ed.
Oxford, UK
:
Oxford University Press 2003
.
Unger, Peter.
Living High and Letting Die: Our Illusion of Innocence
.
Oxford, UK
:
Oxford University Press
,
1996
.
Williams, Bernard. “
The Point of View of the Universe: Sidgwick and the Ambitions of Ethics
.” In
Making Sense of Humanity and Other Philosophical Papers, 153–71
.
Cambridge, UK
:
Cambridge University Press
,
1995
.
Freely available online through the Public Affairs Quarterly open access option.