Compensated Altruism and Moral Autonomy

Abstract

It is sometimes morally permissible not to help others even when doing so is overall better for you. For example, you are not morally required to take a career in medicine over a career in music, even if the former is both better for others and better for you. I argue that the permissibility of not helping in a range of cases of “compensated altruism” is explained by the existence of autonomy-based considerations. I sketch a view according to which you can have autonomy-based permissions to choose between alternatives when these alternatives differ in terms of the valuable features they instantiate. Along the way, I argue that considerations of moral autonomy do not support rejecting the plausible view that we each constantly face reasons with morally requiring strength to help (distant) strangers.

Prudence-based permitting reasons and compensated altruism

Let us here take a “requiring reason” to be equivalent to a reason with morally requiring strength. A requiring reason to do A thus serves to make A morally required. If there is a requiring reason to do A and no countervailing consideration is present, A is morally required. Reasons can conflict. There is a requiring reason not to bruise my arm. There is also a requiring reason to save a stranger’s life. The second reason is significantly stronger than the first. Suppose you can save a stranger’s life only if you bruise my arm. That is what you have most requiring reason overall to do. Absent further considerations, that is what you are morally required to do; it is morally impermissible to do anything else.

What other considerations might there be? Permitting reasons. Let us here take a “permitting reason” to be equivalent to a consideration with purely morally permitting strength. ¹ A permitting reason to do A thus contributes toward making A morally permissible without also contributing toward making A morally required. As I interpret them, permitting reasons contribute toward permissibility by serving to prevent the balance of requiring reasons from making an act morally impermissible. If there is more requiring reason overall to do B than there is to do A, then, absent further considerations, A is morally impermissible. But if there is a sufficiently strong permitting reason to do A, it can prevent the fact that there is more requiring reason overall to do B from making A morally impermissible.

Suppose that, by pressing a button, you can prevent a stranger from being crushed to death by a boulder. Pressing the button would have no downside whatsoever. There is most requiring reason overall to press the button and no permitting reason not to do so. You are morally required to press. Next, suppose that the only way to save the stranger’s life is by sacrificing your arms. While there is still the same requiring reason to save the stranger as before, now there is a permitting reason not to save the stranger. In this case, the permitting reason seems sufficiently strong. So, it is morally permissible not to save the stranger, but not morally required not to do so; in this case, helping is optional. You are morally permitted to do it and you are morally permitted not to do it. The permitting reason at work here is prudence-based: it is grounded in the fact that losing your arms would be worse for you overall. ²

It is doubtful that all permitting reasons are prudence-based. To see this, consider three cases of compensated altruism, that is, cases in which helping others is better for you overall.

Career Choice: You have two career options: spend decades of your life working in medicine or spend them as a musician. Each career has valuable features, but they are quite different. You would not enjoy either sort of work more than the other, but you would make much more money in medicine. You would be better off overall working in medicine. You would also save many lives if you took this career option, whereas you would save none if you became a musician.

Social Entrepreneurship: You can either stick with your career in music or spend two years of your life setting up a company that drastically improves the delivery of medicine in developing countries. Suppose it distributes medicine much better than nongovernmental organizations (NGOs) or governments could. You would really enjoy remaining in music. While you would not enjoy setting up the company, you would make a lot of money and you would be able to return to music afterward. You would be better off overall if you spent two years setting up the company. You would also save many lives if you did this, whereas you would save none if you simply stuck with music.

Upgraded Arms: A collapsing building is about to crush a stranger to death. You can save them only by placing your arms beneath some rubble. This would destroy your arms and hands, but you would immediately receive a new pair of robotic arms and hands. Your new arms and hands would not only be arthritis-free but also much stronger, more efficient, and more dexterous than normal. Sacrificing your arms would be better for you overall. ³

In each of these three cases, your altruism is compensated. Helping others is better for you overall. In a sense, helping is a win-win strategy. Nonetheless, in each case, helping involves sacrificing something. In Upgraded Arms, you give up your natural arms and hands. In Social Entrepreneurship, you spend two years doing something you do not enjoy. In Career Choice, you forgo the valuable features of one possible career for the very different features of another. In each case, it seems to me morally permissible not to help, even though helping would be better for you overall. ⁴ My aim is to uncover what would make this true. This will involve identifying permitting reasons that are not prudence-based, that is, not grounded in the fact that helping would be worse for you overall. ⁵

To be clear, my focus is on whether you are morally required to help in cases of compensated altruism such as Career Choice, Social Entrepreneurship, and Upgraded Arms. I take it that even if it is prudent for you to help in each of these cases, that alone would not entail that you are morally required to help. Even if you have morally requiring reasons to do what is better for yourself, in addition to morally requiring reasons to help others, in each case it seems you still have a sufficiently strong morally permitting reason not to help. For simplicity, I will assume you do not have a reason with morally requiring strength to do what is better for yourself, but my main arguments do not rely on this assumption.

Possible explanations: Uncertainty, difficulty, burdens, projects, and preferences

I will offer a novel account of what makes it morally permissible not to help in cases of compensated altruism such as Career Choice, Social Entrepreneurship, and Upgraded Arms. Before doing that, I will here consider some alternative explanations.

Uncertainty. It may seem morally permissible not to help when you are uncertain about whether your sacrifices for others will be met with compensating benefits. But in the three cases of compensated altruism mentioned above, we are to assume you are certain that you will receive adequately compensating benefits, so that you will be better off overall. Moreover, even when you are uncertain about whether your sacrifices for others will be met with compensating benefits, it may be clear that making these sacrifices is better for you overall in expectation. For example, suppose that in Social Entrepreneurship there is an 80 percent chance that setting up the company would make you $100 million and a 20 percent chance that it would make you only marginally more than remaining in music. Even if risk-averse prudence favors setting up the company, it still seems morally permissible not to. ⁶

Difficulty. It may seem morally permissible not to help when it would be psychologically difficult to help, as in cases like Social Entrepreneurship or Upgraded Arms. ⁷ But even if psychological difficulty affects permitting reasons (and not merely excuses), it does not provide a full explanation. A simple way to see this is to consider versions of cases such as Social Entrepreneurship or Upgraded Arms in which you accidentally ingest a drug that makes it psychologically easy for you to undergo the burdens of helping. It would still seem morally permissible not to help.

Burdens. Perhaps you have a permitting reason not to undergo certain burdens of helping. By “burdens” I mean things that are intrinsically bad for you (or noncomparative harms). ⁸ It can be overall better for you to suffer burdens for the sake of avoiding greater burdens or gaining compensating benefits. For example, it is overall better for you to suffer a bruised arm if that is the only way to save your life. Perhaps in cases of compensated altruism, the burdens of helping are great enough to make it morally permissible not to help, even though helping is overall better for you and even if it is psychologically easy for you to undergo the burdens in question. But even if burdens affect permitting reasons, they do not provide a full explanation. A simple way to see this is to suppose that in cases such as Career Choice, Social Entrepreneurship, and Upgraded Arms, the burdens of helping are so minimal that they would fail to generate significant permitting reasons. Assuming that your choice still makes a significant difference to how your life unfolds, it still seems morally permissible not to help.

Projects. Another possible explanation is that you have a permitting reason not to sacrifice your personal projects, independently of whether doing so would involve incurring costs or burdens. Personal projects are aims or activities that are importantly constitutive of your identity and that give meaning to your life. ⁹ For example, even if spending two years setting up the company in Social Entrepreneurship is in no significant way intrinsically bad for you, it could nonetheless seriously impede your personal projects. Perhaps this is why it is morally permissible not to help in cases such as Social Entrepreneurship. But again, this would not provide a full explanation. Even if helping does not impede your personal projects, it would still seem morally permissible not to help.

Preferences. Perhaps preferences—including fleeting interests, wants, or desires—can generate sufficiently strong permitting reasons not to help in key cases of compensated altruism. Perhaps, in Upgraded Arms, it is enough that you prefer not to give up your natural arms and hands. Perhaps, in Career Choice, it is enough that you presently prefer a career in music over a career in medicine. No further justification is required. But it cannot be this simple. Suppose that, by pressing a button, you can prevent a stranger from being crushed to death by a boulder. Pressing the button would have no downside whatsoever, apart from the fact that you presently prefer not to press the button. It seems clear that you are still morally required to aid the stranger.

In this section, I considered five possible explanations of what makes it morally permissible not to help in cases of compensated altruism such as Career Choice, Social Entrepreneurship, and Upgraded Arms. I have argued that these explanations are inadequate. At best, they provide partial explanations of some of the cases. I argue that there is another kind of nonprudence-based permitting reason that provides a deeper and more unified explanation of the permissibility of not helping in these cases of compensated altruism.

Moral autonomy and the ubiquity of requiring reasons

My proposed account of what makes it morally permissible not to help in cases of compensated altruism is that you enjoy autonomy-based permitting reasons not to help. Roughly, these are permitting reasons to choose between different ways your life might unfold when these alternatives differ in relevant respects. Below, I will sketch a view of these permitting reasons. In this section and the next, though, I will build up to my view by considering existing ideas about moral autonomy or freedom from morality’s reach.

In general, proponents of moral autonomy are concerned with moral requirements that encroach on too many choices and aspects of your life; they hold that there must be a suitably wide range of cases in which helping others is morally optional. Proponents of moral autonomy can differ along at least two axes. First, they can differ in terms of the ubiquity of requiring reasons to help others. Some hold the Ubiquity View, according to which requiring reasons to help are ubiquitous, given that opportunities to help are ubiquitous, while others hold the Sparsity View, according to which there is a wide range of cases in which there are no requiring reasons to help. Second, proponents of moral autonomy can differ in terms of grounding. Some hold the Rights View, according to which moral autonomy is grounded in property rights, while others hold the Value View, according to which moral autonomy is grounded in the value of nonmoral choice. These two axes can be mixed and matched, giving rise to four possible views of moral autonomy: ubiquity/rights, sparsity/rights, ubiquity/value, and sparsity/value.

In this section, I consider the “ubiquity” axis. Why should we accept the Sparsity View? Some might argue that cases of compensated altruism support the Sparsity View. The argument is that because there is no prudence-based permitting reason not to help in the three cases of compensated altruism mentioned above, and it is morally permissible not to help in these cases, it must be that there is no requiring reason to help in these cases. This argument is unsound. It incorrectly assumes all permitting reasons are prudence-based. I argue that there are nonprudence-based permitting reasons not to help. Nonetheless, there are independent arguments for the Sparsity View. If some such argument is successful, one could argue that it is morally permissible not to help in cases of compensated altruism because there is no requiring reason to help in the first place. Let us consider the Sparsity View and the independent support it might enjoy.

According to an extreme version of the Sparsity View, there are no requiring reasons to help strangers at all, not even in nearby emergency rescue cases. This view is implausible. There is a requiring reason to press a button to save a nearby stranger from being crushed to death by a boulder. This is true regardless of whether pressing the button is accompanied by weighty considerations of prudence or autonomy. Even if pressing the button would cause you to suffer months of pain or otherwise dramatically affect how your life would unfold, the requiring reason to save the stranger would remain.

Of course, defenders of the Sparsity View need not adopt such an extreme view. In fact, they typically allow that there are requiring reasons to help individual strangers in nearby emergency rescue cases but deny that there are requiring reasons to help individual strangers beyond these sorts of cases. They may hold that there are nonetheless imperfect duties to contribute to aiding distant strangers enough of the time or requirements to do your fair share of a collective aid effort. ¹⁰ Assuming that nearby emergency rescue cases are rare enough, this would enable them to establish a suitably wide range of cases in which there are no requiring reasons to help. And it is true that cases of compensated altruism such as Career Choice and Social Entrepreneurship involve helping at a distance. These cases differ from nearby emergency rescues, such as Upgraded Arms, in several ways: the strangers are far away, you cannot see them, and many others can help. But for all that, there remain strong requiring reasons to aid distant strangers. Suppose someone is drowning thousands of miles away. While you cannot see them, you have excellent evidence that you can save them by pressing a button that would instantly put a life buoy within their reach. Many others can help, including the governments who bear primary moral and legal responsibility for those imperiled on their territories, but you are confident they will not help, even if you try to get them to. It would seem you are morally required to press the button, even if doing so involves incurring a significant cost. So, there is a strong requiring reason to help. Cases such as Career Choice and Social Entrepreneurship differ from nearby emergency rescues in several additional ways, some of which may be morally relevant. But even when all these differences are considered, I submit there remains a strong requiring reason to help. There are strong requiring reasons to help distant strangers by donating to aid agencies, volunteering, taking up certain careers, or engaging in social entrepreneurship.

Defenders of the Sparsity View may argue that we cannot secure a suitably wide zone of moral optionality unless we have a correspondingly wide zone without requiring reasons. Because strong requiring reasons to help distant strangers are ubiquitous, you would be under constant pressure to help and it is difficult to see how you will not end up being more or less constantly morally required to help. More slowly, here is the worry. At more or less any given moment throughout your life, you can do more to aid distant strangers in need. You can always pause what you are doing and give a little more to charity or you can pause and raise a little more money, so that you can later donate it to charity. Because you face a ubiquity of opportunities to help distant strangers, and you have strong requiring reasons to help distant strangers, you face a ubiquity of strong requiring reasons to help. Whenever you face a requiring reason, you are morally required to act in accord with this reason, in the absence of any other sufficiently strong considerations. If there is a ubiquity of requiring reasons to aid others, then there will have to be a comparable ubiquity of sufficiently strong countervailing considerations to achieve a wide enough zone of moral optionality. But it might seem that sufficiently strong countervailing considerations are too sparse. Right now, you can stop reading this essay and instead go online to raise more money for effective aid agencies. What (requiring or permitting) reason is present right now that could morally justify you in not acting in accord with the strong requiring reason to raise money for aid agencies? You may presently have a permitting reason of some strength to ignore the present requiring reason to help, but presumably that permitting reason is quite weak—we are only talking about giving up a little bit of your time. Because there is a strong requiring reason to raise money for aid agencies right now, it would appear you are morally required to do this, right now. And what is true right now is also true a little bit later, and a little bit later, and so on. Sufficiently strong permitting reasons are all too rare. Since most of the time there will not be a sufficiently strong countervailing consideration, most of the time you will be morally required to help. That is overly demanding as there is not a wide enough zone of moral optionality. According to defenders of the Sparsity View, the problem was letting in all the requiring reasons to help distant strangers.

I reject this argument’s premise that sufficiently strong permitting reasons are too sparse. The sacrifices for others you (reasonably expect you) could or will make over the course of your life can amplify your permitting reasons not to make sacrifices now. Even if you could save a stranger’s life at the relatively modest cost of sacrificing an hour, it can turn out that such lifetime facts amplify the permitting reason not to sacrifice an hour enough to make it morally permissible not to save the stranger’s life. There are different views about how exactly this amplification might work. For example, we can disagree about whether it is lifetime facts about what you could do or instead lifetime facts about what you will do that amplify present permitting reasons. Consider a miser who in fact will not help anywhere close to enough over the course of his life. Does the fact that he nonetheless could help enough over the course of his life amplify his permitting reason not to help now? For present purposes, I can remain neutral about this. ¹¹

Some such amplification of permitting reasons prevents our lives from being saturated with moral requirements to incur small cost after small cost to help others; clearly an hour is itself a small cost, but hours can quickly add up to weeks, months, or years. The amplification of permitting reasons by lifetime facts explains how it is the case that, even though we are under constant moral pressure to help, we need not be constantly morally required to help. The ubiquity of requiring reasons is matched by a ubiquity of sufficiently strong permitting reasons, enabling a wide enough zone of moral optionality.

According to another argument for the Sparsity View, it is problematic that we are constantly under strong moral pressure to help others, even if we regularly have sufficiently strong permitting reasons not to help. But it is unclear what is problematic about constantly being under strong moral pressure per se. It would be problematic if you had to constantly entertain and balance reasons to help others, call to mind justifications for not acting in accord with these reasons, and offer these justifications to others. ¹² My view is that permitting reasons not to help include permitting reasons not to think about helping. You do not have to entertain the never-ending requiring reasons to help or constantly call to mind your justifications not to help. Likewise, it would be problematic if you had to constantly apologize to all those you permissibly fail to help. But having a requiring reason to help someone does not entail a requiring reason to apologize to them when you permissibly fail to help. And even if it did, we could invoke permitting reasons at this level, too. To the extent that it is problematically demanding to have to constantly apologize for failing to help, it is plausible that we typically have sufficiently strong permitting reasons not to apologize. Permitting reasons could also plausibly serve to make it the case that it is regularly appropriate not to apologize for or feel sorry about each failure to help.

Arguments for the Sparsity View come up short. By contrast, there is a straightforward and powerful argument for the Ubiquity View. There is a significant requiring reason to press a button that would save a distant stranger’s life by putting a life buoy within their reach. Now consider a long sequence of life-saving opportunities, each just like this one. For any given opportunity, it seems there is a requiring reason to help, independently of the others; intuitively, each person you can help is a separate source of moral pressure on you. ¹³

In addition, it is worth observing that, even if the Sparsity View is correct, so that there are no requiring reasons to help in cases such as Career Choice and Social Entrepreneurship, it remains plausible that there are requiring reasons to help in some relevantly similar fictional cases. Consider a hypothetical world in which you have only one opportunity over the course of your whole life to help: a one-off opportunity to save the life of a distant stranger by donating just $1 to a charity. It seems plausible that you are morally required to make this donation, and hence that there is a requiring reason to do so. This suggests that if your one-off opportunity to help were of the sort found in Career Choice or Social Entrepreneurship, you would have a requiring reason to so help. But in these more fictional cases of compensated altruism, it still seems that helping is morally optional. This is evidence of autonomy-based permitting reasons.

If, instead, the Ubiquity View is correct, there remains a methodological benefit of turning to hypothetical worlds in which you only ever have one opportunity to help. Namely, doing so enables us to bracket complications involving the amplification of permitting reasons by lifetime facts. Consider a case such as Career Choice in which the opportunity to help found in this case constitutes the only opportunity to help you will ever have. It still seems that helping is morally optional. This suggests that an unamplified autonomy-based permitting reason explains the moral permissibility of not helping.

The grounds and extent of moral autonomy

Proponents of moral autonomy hold that there must be a suitably wide range of cases in which helping others is morally optional. What is the ground of such moral autonomy? It would seem odd to posit a wide range of moral optionality as a brute moral fact. If it is a fact, it presumably holds in virtue of something deeper.

Defenders of the Rights View hold that moral autonomy is grounded in property rights. More specifically, they claim that you enjoy moral autonomy as a by-product of having moral property rights over your body and various external resources (henceforth “your things”). ¹⁴ The idea is that your having genuine moral property rights over your body and things is incompatible with being morally required to help too often, and you would be morally required to help too often if you were morally required to help whenever this involves no setback of your well-being, projects, or preferences. If you were so frequently morally required to use your body and things for the sake of others, you would not have moral property rights to them in a meaningful sense. Moral property rights, the thought goes, should robustly protect us from “interference” from morality as well as from interference from flesh-and-blood agents.

By contrast, defenders of the Value View hold that moral autonomy is grounded in the value of nonmoral choice. More specifically, they claim that there is value in being able to choose between alternatives without being morally required to choose either way. ¹⁵ Some suggest that being significantly morally free in this way is prudentially good for us. ¹⁶ A related but distinct idea is that such autonomy is entailed by a kind of unconditionally valuable status that persons have. ¹⁷

Defenders of the Sparsity View might appeal to either the Rights View or the Value View. Appealing to the Rights View, they could argue that you do not have meaningful moral property rights to your body and things if you are too frequently under moral pressure to help, even if that pressure—in the form of requiring reasons—falls short of moral requirement. Similarly, appealing to the Value View, they could claim that the value of being able to choose between alternatives depends on not being under constant moral pressure to choose either way. However, as I argued above, it is unclear what is problematic about being under constant moral pressure per se, once this pressure is separated not only from requirements to help, but also from requirements to entertain and balance reasons to help. Facing a ubiquity of requiring reasons to help does not itself undermine meaningful property rights to your body and things, nor does it undermine the value of being able to choose between alternatives. Rather than lend support to the Sparsity View, it is more plausible that the Rights View or Value View would ground moral autonomy via permitting reasons (in a way that is compatible with the Ubiquity View).

I will not defend either view about the grounds of moral autonomy. Instead, I will primarily focus on the extent of moral autonomy. Which choices are protected by autonomy-based permitting reasons? What factors determine the strengths of these permitting reasons?

For starters, defenders of autonomy-based permitting reasons can deny that there are any such permitting reasons to inflict serious harm on innocent nonconsenting people. There is no value in being able to choose to inflict serious harm without “interference” from morality, nor is this choice protected by rights. By itself, this restriction on autonomy-based permitting reasons is minimal. We could take a more general view according to which there are no autonomy-based permitting reasons to act in ways that violate or even merely contravene the rights of others. Then, the extent of autonomy-based permitting reasons would in part be a function of the extent of others’ rights. For example, whether and when there are autonomy-based permitting reasons not to help others will depend on whether these others have rights to your help. ¹⁸

While such a rights-based restriction on moral autonomy strikes me as plausible, it does not provide the whole story. In particular, it is not plausible that any failure to help is morally permissible, just in case it involves no rights contravention. Even if a stranger lacks a right to be saved, it would be morally impermissible to let them die if there were no downside to saving them. An absence of conflicts with others’ rights may be necessary for autonomy-based permitting reasons, but it does not seem sufficient. At least, it will not always be sufficient for sufficiently strong autonomy-based permitting reasons.

According to one view, you must have a wide zone of moral autonomy over the course of your life, so, in the face of a ubiquity of requiring reasons to help, you must enjoy sufficiently strong autonomy-based permitting reasons not to help in correspondingly many cases. This view seems to get things the wrong way around. Instead, various details of your life determine the strength of your autonomy-based permitting reasons, which in turn play a role in determining the extent to which you enjoy moral autonomy not to help. At least, it is not plausible that having some fixed portion of moral autonomy is itself something we all must possess for the sake of rights or value, regardless of the details of our lives.

Consider a variant of Upgraded Arms in which you could press a button that would instantly and painlessly sacrifice your arms to save the stranger’s life and immediately regenerate new, exactly similar arms. Suppose you are indifferent between having your original arms and having these new ones. In this variant of the case, it seems to me you would be morally required to sacrifice your original arms to save the stranger’s life. You would not have a sufficiently strong autonomy-based permitting reason not to help. Your alternatives do not differ enough in terms of the valuable features they instantiate.

Similarly, consider a variant of Career Choice. Suppose you could take a career in which you spend your working hours wearing red-tinted glasses or take an otherwise exactly similar career in which you spend your working hours wearing green-tinted glasses, and you will always be indifferent between red and green. The only other difference between these careers is that you would save many lives if you took up the green-tinted career and save none if you took up the red-tinted career. Yes, decades of your time are at stake, but it seems implausible that you would in such a case have any significant permitting reason to choose the red-tinted career. Again, your alternatives would not differ enough in terms of their valuable features.

It does not particularly matter whether these are one-off or repeated opportunities to help. Even if you were to face a long sequence of opportunities to help different strangers, each relevantly like the variant of Upgraded Arms or the variant of Career Choice, it seems you would be morally required to help on each occasion. If these constituted your only opportunities to help, you would have no moral autonomy (not to help) over the course of your life. It is therefore implausible that there is some fixed portion of moral autonomy that any life must possess, regardless of the details. ¹⁹

Autonomy-based permitting reasons: The feature choice view

I have suggested that various details of your life determine the strength of your autonomy-based permitting reasons. In this section, I will sketch a view of how some such details do this. I propose the Feature Choice View, according to which, at least in the absence of conflict with others’ rights, you have an autonomy-based permitting reason to choose between alternatives if and only if these alternatives differ in terms of the valuable features they instantiate.

Alternatives often differ in terms of valuable features, forcing you to decide which features to go for and which to neglect. This is true both of big choices like what career to pursue, where to live, and whether and whom to marry, as well as of small choices like whether to order the soup or the salad, what outfit to wear, and what route to take to work. ²⁰ A career in music may not pay as well as a career in medicine, but it may present more opportunities for artistic creativity. So, the choice between the music career and the medicine career is in part a big choice between different sorts of values: creativity and wealth. A soup might be tastier but less healthy than a salad, presenting you with a small choice between the different values of taste and health.

I here understand valuable features simply as features that it is in some sense appropriate to value, promote, honor, or admire. They can be features of lives, activities, objects, or states of affairs. They can be morally valuable, but they need not be. As Susan Wolf famously illustrates in her discussion of moral saints, there are many nonmorally valuable things that one could reasonably like and pursue, such as those I have already mentioned: creativity, taste, wealth, and health. ²¹ At the same time, the moral badness of certain things, such as sadistic pleasure, may prevent them from having nonmoral value.

According to the Feature Choice View, you have an autonomy-based permitting reason to choose between alternatives if and only if these alternatives differ in terms of the valuable features they instantiate. I take it that choosing an alternative involves some kind of responsiveness to its valuable features rather than picking it whimsically. ²² What is essential is that you are choosing an alternative that instantiates valuable features other alternatives lack, not that you are choosing an alternative that is overall best or even on a par with alternatives. ²³

I also propose the following addition to the Feature Choice View. The autonomy-based permitting reason to perform an alternative is stronger (i) the more different the valuable features instantiated by this alternative are compared to those instantiated by other alternatives and (ii) the more you favor valuable features instantiated by this alternative over those instantiated by other alternatives. For reasons that will become clear soon enough, it seems there is an asymmetry between these factors. While you can have a strong autonomy-based permitting reason when (ii) is minimal if (i) is sufficiently big, you cannot have any significant autonomy-based permitting reason when (i) is minimal, no matter how big (ii) is. (For an “elitist” spin on my view, we could add that permitting reasons are magnified in accord with the objective value of the features instantiated, so that there is a stronger permitting reason to choose poetry over medicine than there is to choose pushpin over medicine.)

Factor (i) will tend to be bigger in the context of big choices (what career to pursue, whether and whom to marry, and so on) and smaller in the context of small choices (whether to order the soup or the salad, what route to take to work, and so on). I take it this factor is sensitive to both quality and quantity. There is a big qualitative difference between the values of medicine and music, and choosing between a career in medicine and a career in music involves a big quantitative difference (decades of your life). By contrast, choosing between spending one hour learning physiology and one hour learning violin involves a small quantitative difference, so factor (i) remains relatively small in the context of this choice of how to spend an hour despite the big qualitative difference. I will not take a stand on how to measure qualitative and quantitative differences between values. Intuitively, (i) is big in the context of a choice between a career in medicine and a career in music, but small in the context of a choice between a career as a violinist and a career as a violist, other things being equal.

Factor (ii)—how strongly you favor valuable features—can range from “minimal” (responding to these features by choosing an alternative at least partly on the basis that it instantiates these features) to “mild” (taking something of a fleeting interest in these features), “moderate” (caring substantially and persistently about promoting or protecting these features), or “maximal” (taking on the promotion or protection of these features as a central life project). You will have a stronger autonomy-based permitting reason to choose the music career over the medicine career, the more you care about or get behind the values that would be instantiated were you to opt for the music career.

The Feature Choice View can deliver intuitive results in our three cases of compensated altruism. All these cases involve alternatives that differ dramatically in terms of valuable features. So, in each of these three cases, at least as long as you minimally favor some valuable features instantiated by one alternative over those instantiated by another, you have a strong autonomy-based permitting reason to choose the alternative that instantiates these features. That is, you can have a strong autonomy-based permitting reason when (ii) is minimal if (i) is sufficiently big. In each case, the permitting reason in question seems to me strong enough to make it morally permissible not to help, even though that means failing to save lives.

What’s more, the Feature Choice View avoids the implication that it would be morally permissible to refuse to muddy your shoes to save a stranger’s life, when this is your only opportunity to help over the course of your life. ²⁴ In easy rescue cases like these, the value instantiated by not helping is minimal: it is basically just the value of not having mud on your shoes or of not having to buy another pair (we can suppose these shoes do not have any special sentimental value). So even if you strongly favor saving your shoes from mud, there will not be much value at stake to be magnified by this favoring, effectively setting a low limit on the strength of the autonomy-based permitting reason not to help. More generally, you cannot have a significant autonomy-based permitting reason when (i) is minimal, no matter how big (ii) is. The permitting reason here certainly will not be strong enough to make it morally permissible not to act in accord with the strong requiring reason to save someone’s life.

I have suggested that you have a sufficiently strong autonomy-based permitting reason in each of the three key cases of compensated altruism, insofar as they involve alternatives that differ dramatically in terms of valuable features and you at least minimally favor one set of valuable features over the other (choosing at least partly based on these features). However, it seems to me that the Feature Choice View should be taken to say that you cannot generate sufficiently strong autonomy-based permitting reasons out of such minimal favoring when the values in question accrue to a stranger. To show this, I will offer a pair of cases in which you have no opportunities to help over the course of your whole life other than those mentioned. ²⁵

Your Interview: You are driving to an interview that would kick off a career in music. If you miss the interview, you will take on a career in medicine instead. The values instantiated by either alternative differ dramatically, but the career in medicine would be better for you overall. Nonetheless, you mildly prefer music to medicine. Here is the catch: you will be able to save a life by donating just $1 to charity only if you slow down and miss your music interview. You are aware of all the relevant details of the case.

Stranger’s Interview: You are driving a bus with one stranger on it, who you allow to board for free. You learn they are on their way to an interview that would kick off their career in music. If they miss the interview, they will take on a career in medicine instead. The values instantiated by either alternative differ dramatically, but the career in medicine would be better for the stranger overall. Nonetheless, the stranger mildly prefers music to medicine, and because you know this you also mildly prefer music over medicine for them. Here is the catch: you will be able to save a life by donating just $1 to charity only if you slow down, causing the stranger to miss their music interview. You are aware of all the relevant details of the case.

It seems to me that you have a sufficiently strong autonomy-based permitting reason not to donate in Your Interview, but not in Stranger’s Interview. I do not think the fact that you minimally (or even mildly!) favor music over medicine in Stranger’s Interview can ground a sufficiently strong autonomy-based permitting reason to forgo saving a distant stranger’s life by donating. Minimally favoring some valuable features over others does not generate strong autonomy-based permitting reasons when these valuable features are instantiated in a stranger’s life. The key cases of compensated altruism are different in that the valuable features you at least minimally favor accrue to you. Cases of minimal favoring reveal an agent-relative aspect of autonomy-based permitting reasons: shaping your own life is especially significant. ²⁶ Perhaps this is not too surprising—after all, autonomy concerns self-determination. Of course, this is not to deny that you can care about and take a strong interest in shaping the lives of others. It is plausible that you can have strong autonomy-based permitting reasons to choose some valuable features over others, even when these features are instantiated in someone else’s life, when you more than minimally favor the features in question, at least if the features you favor overlap enough with the features this person favors. ²⁷

Given that the Feature Choice View says you cannot generate strong autonomy-based permitting reasons out of minimal favoring when the values in question accrue to a stranger, it similarly avoids implausible implications about charitable giving. Consider a version of Stranger’s Interview in which you could either donate enough to help a stranger (who only mildly prefers music over medicine) get to their music interview or instead donate an equal amount to charity, saving a different stranger’s life. The Feature Choice View does not imply it is permissible to help the stranger get to their music interview. More generally, the view is compatible with the claim that, in a significant range of cases, it is morally impermissible to give to less cost-effective charities rather than more cost-effective charities. ²⁸

In addition to cases of compensated altruism, autonomy-based permitting reasons can explain what are sometimes called “agent-sacrificing permissions.” ²⁹ Agent-sacrificing permissions are moral permissions to prevent lesser losses to strangers at the expense of greater losses to yourself. Now recall that I said I will assume you do not have a requiring reason to do what is better for yourself. If that assumption is correct, then agent-sacrificing permissions follow automatically. But if there is a morally requiring reason to do what is better for yourself, then it can turn out that there is more morally requiring reason overall to avoid greater losses to yourself than to prevent lesser losses to strangers. Even so, autonomy-based permitting reasons could nonetheless make it morally permissible to prevent lesser losses to strangers at the expense of greater losses to yourself. At least, they could do so in cases in which the self-sacrificial alternative instantiates valuable features that you at least minimally favor. (If there is a morally requiring reason to do what is better for yourself, then I take it that it would be morally impermissible to self-sacrifice on a whim, as you would not even minimally favor this alternative.)

Finally, it is worth noting that whether an autonomy-based permitting reason is sufficiently strong in a given case of compensated altruism may depend on how much better off you are for helping. In Social Entrepreneurship, you give up two years of music to do something you do not enjoy. This involves forgoing significant valuable features you significantly favor, so that you are morally permitted not to help, even though the wealth you would accrue would make it the case that you would be better off overall if you did help. It could for all that turn out that you would be morally required to help if setting up the company made you much better off overall, for example, if it not only made you rich but also much happier and healthier and led to many deep friendships. In other words, even when there is no sufficiently strong prudence-based permitting reason not to help, considerations of prudence might still interact with whether a given autonomy-based permitting reason is sufficiently strong. On the other hand, some autonomy-based permitting reasons may be so strong that they remain standing largely independently of prudential considerations. Think of the autonomy-based permitting reasons grounded by central life projects, anchored around objectively valuable activities. Suppose music is your sole passion, and that setting up the company in Social Entrepreneurship would prevent you from ever returning to music. Now I find it plausible that you are morally permitted not to start up the company, no matter how much better off overall you would be for doing so.

Summing up

Let me sum up. In the key cases of compensated altruism I have discussed here, you are not morally required to help others even though doing so is overall better for you. I have suggested that, while you lack a prudence-based permitting reason not to help, you nonetheless have a sufficiently strong autonomy-based permitting reason not to help. Other possible explanations of the permissibility of not helping in these cases are either implausible or do not go far enough. I have sketched the Feature Choice View, according to which you have an autonomy-based permitting reason to choose between alternatives if and only if these alternatives differ in terms of the valuable features they instantiate. Along the way, I have argued that considerations of moral autonomy do not support rejecting the plausible view that strong requiring reasons to help (distant) strangers are ubiquitous.