Back in 1982, when I started studying with Samuel Scheffler, a major subject of debate was whether – and, if so, exactly why and how – the existence of agent-relative reasons for action justifies making fundamental changes to consequentialist moral theory. To be sure, some philosophers insisted, on metanormative grounds of the kind Thomas Nagel had earlier explored in The Possibility of Altruism, that the question is effectively moot because agent-relative reasons are not really distinct from agent-neutral ones. However, most philosophers, including Scheffler and (pretty quickly) Nagel himself, allowed that agent-relative reasons can diverge from agent-neutral ones in several important ways but wondered on metaethical grounds whether this justifies making fundamental changes to consequentialist moral theory.Footnote 1 As consequentialists typically acknowledged, it might justify making changes to their axiological views about the good that is to be maximized, but does it also justify making changes to their deontological views about the priority of maximizing the good?
How dramatically the tables have turned in the years since! Now the question tends to be whether – and, if so, exactly why and how – agent-neutral reasons can have a significant role to play in the determination of morality’s requirements. To be sure, some philosophers again insist, on metanormative grounds, that the question will prove moot, this time because they are persuaded that fully agent-neutral reasons could exist only if an implausibly substantive form of metanormative realism were true. But even when they allow that agent-neutral reasons exist, many philosophers now wonder whether such reasons can have a significant role to play in the determination of morality’s requirements, often concluding, in keeping with Thomas Scanlon’s version of contractualism, that, at its core, anyway, morality’s requirements are determined by what people owe one another, where what people owe one another is in turn determined by how contending principles would impact various other sorts of reasons that people standardly possess.Footnote 2
I myself don’t have much sympathy for the metanormative concerns. As I explain in the first section of this paper, I am perfectly happy to assume that a substantive form of metanormative realism is true. What interests me is a set of questions that arise after this has been assumed. Does this assumption in fact strengthen the case for concluding that agent-neutral reasons exist? Where does it leave the case for concluding that reasons of various other sorts exist? What does it suggest about the relation that agent-neutral reasons are likely to bear to reasons of other sorts? Are they likely to be comparable in importance or strength?Footnote 3 Even if not, could they still be somehow more basic? It is crucial that we be clear about these matters if we are to make any headway on the concerns contractualists inspired by Scanlon have raised about the role agent-neutral reasons might play in the determination of morality’s requirements.
These are concerns for which I do have considerable sympathy. Like many contractualists, however, I also feel deeply conflicted. Because I believe there is a sense in which agent-neutral reasons will be more basic, I believe we must give them a central role to play in moral theory – a role helping to determine what morality requires of people in their dealings with one another, and not merely in other contexts that might understandably be regarded as more peripheral. The problem, however, is that it is not easy to see how we can give agent-neutral reasons such a role to play in moral theory without jeopardizing the distinctive account contractualists give of what people owe one another. My suggestion in the second section of this paper will be that we look for help on this score to Scheffler’s recent discussions of the important ways in which future generations can have a value for people that is basic.Footnote 4
1. The Metanormative Case for Agent-Neutral Reasons
As I understand it, the agent-neutral/agent-relative distinction applies exclusively to reasons that have their source in facts about the goals that agents should be promoting. As we shall see in just a moment, a related distinction can be drawn among reasons having a different source, but I draw that distinction using different terms.Footnote 5 Thus suppose that an accident has left a friend of mine in serious pain and that I have a pro tanto reason to help rid her of it. My reason is agent-neutral if it has its source in the fact that ridding everyone of pain is a goal that everyone should be promoting; it is agent-relative if it has its source in the fact that ridding my friends of pain is a goal that I should be promoting (or, perhaps more plausibly, since reasons are surely universal, in the fact that ridding their friends of pain is a goal that everyone should be promoting).
Are there actually any reasons for action that are neutral in this teleological sense? Christine Korsgaard has advanced a two-pronged argument suggesting that there may not be, claiming, first, that only someone who embraced an implausibly substantive form of metanormative realism could seriously suppose that there are any goals that every agent should be promoting, and, second, that normative neutrality is better understood in terms of the possibility that some ways of treating people are ones to which every agent should be adhering.Footnote 6 Thus, for example, while she may or may not agree that every agent should treat my friend as having a pro tanto claim on their help, Korsgaard undoubtedly would insist that every agent should treat every other agent as an end-in-itself and not as a mere means of accomplishing things, as I would be treating a child if I were to twist its arm to get its grandmother to give my friend the help she needs.
Since we are reserving the terms ‘agent-neutral’ and ‘agent-relative’ to draw a distinction exclusively among reasons that have their source in facts about the goals that agents should be promoting, let us introduce the terms ‘relation-neutral’ and ‘relation-relative’ to distinguish among reasons that have their source in facts about the ways of treating people to which agents should be adhering. Thus my reason to treat my friend as having a pro tanto claim on my help will be relation-neutral if it has its source in the fact that she is a person and that every agent should treat other people as having pro tanto claims on their help; it will be relation-relative if it has its source in the fact that she is my friend and that every agent should treat their friends as having pro tanto claims on their help. (Of course, to fully distinguish these reasons from their agent-neutral and agent-relative counterparts, we would have to explain what is involved in treating people as having pro tanto claims on one’s help, showing that it cannot be correctly understood in terms of any goal that one should be promoting.)
Again, however, we should ask if any reasons for action actually are neutral in this sense. Korsgaard maintains that some must be on the grounds that reasons must be public in nature, by which she means that they must be possible objects of discussion and debate among people, not the exclusive preserve of what Wittgenstein would have dismissed as some putatively merely private language.Footnote 7 Why does she think it follows from this fact that some reasons must be relation-neutral? Why does she think people couldn’t discuss and debate reasons if they were all relation-relative? I think we can see what Korsgaard is getting at here by recalling that Wittgenstein’s case against the possibility of merely private languages turns on a claim about the determination of meaning and content – to wit, that neither meaningful talk nor contentful thought would be possible for people who could not establish suitable practices governing the use of their words and concepts through their interactions with one another. Given this, I take Korsgaard’s contention to be that such interactions couldn’t establish practices governing thought and talk about relation-relative reasons without also establishing practices governing thought and talk about relation-neutral reasons.
Suppose that we were trying to determine when friends may correctly be regarded as having special claims on one another. Could our interactions establish a practice governing this question without at the same time establishing practices governing various adjacent questions? One would think, at a minimum, that they would also have to establish practices governing the question when people related to one another in various different sorts of ways may correctly be regarded as having special claims on one another, not to mention practices governing the question when all these variously related groups of people may correctly be regarded as having claims on third parties not to intervene or even to help in their attempts to discharge their special obligations. Of course, these questions still concern only relation-relative issues, but it’s hard to see how matters could stop here. I take Korsgaard’s contention to be that we could not establish practices governing any of these relation-relative issues without also establishing practices governing the question when people quite generally may correctly be regarded as having claims on one another.
More argument is evidently needed to get from this contention to the conclusion that some reasons really are relation-neutral. In Korsgaard’s case, given her thoroughgoing commitment to metanormative constructivism, we need to be told when normative truths are constructible. In particular, we need to be told how the existence of practices governing thought and talk about relation-neutral reasons guarantees that truths concerning relation-neutral reasons could in principle be constructed and command the assent of all rational agents. However, as I am not myself a thoroughgoing metanormative constructivist, I think Wittgenstein’s insight here should be developed rather differently. In particular, I think the interactions that Wittgenstein speaks of should often be regarded as triangulations on independently existing properties. Once we have basic contents and meanings at our disposal, we can use them to introduce new topics of discussion, but it cannot possibly be construction all the way down. As I have argued elsewhere, I think this holds of normative contents and meanings as well as non-normative ones, and that it commits us to a substantive metanormative realism.Footnote 8
As we saw earlier with Korsgaard, the idea of an independently existing normative property strikes many people as intolerably queer, so worries on that score would have to be overcome before we could confidently take Wittgenstein’s thought in this direction.Footnote 9 As there is no time for that here, however, let us turn our attention immediately to the following questions. If there are substantive normative properties on which people triangulate when learning to think and talk about reasons for action, what sorts of reasons are they able to underwrite and what are the relations among those reasons likely to be? Is there anything keeping substantive normative properties from underwriting reasons of the agent-relative, relation-relative or relation-neutral sorts? If not, how are reasons of the agent-neutral sort likely to be related to reasons of these other sorts? It may be wondered whether there is anything illuminating to be said in response to the last of these questions, but I submit that two points can plausibly be made, both of which will prove to be important going forward.
First, though, some brief remarks on the question whether substantive normative properties can underwrite reasons other than agent-neutral ones. Some may be inclined to argue that substantive normative properties can only underwrite reasons that are extremely simple in nature, involving nothing more complex than goals that should be promoted, and promoted regardless of any particularities of history or circumstance. But it’s difficult to see why this should be so. Why couldn’t particularities of history and circumstance make a difference? It’s a fact that I share histories of special sorts with those people who have come to be my friends. So why couldn’t that fact help make it the case that I have special reasons to promote their well-being?Footnote 10 And why couldn’t the fact that I have such special reasons influence how other people have reason to treat me, not simply by influencing what my well-being consists in, and so by influencing how it might best be promoted, but also by helping make it important that I always be treated in relation-neutrally and relation-relatively mandated ways?
Turning now to the question how agent-neutral reasons are likely to be related to the others, the first point to note is that, being simpler, they are likely also to be more basic, in that the existence of agent-neutral reasons will help explain the existence of the other reasons, whereas the existence of the other reasons won’t help explain the existence of agent-neutral reasons. Surely, for example, the special sorts of histories that I share with my friends did not create, out of nothing, the special agent-relative reasons that I now have to promote their well-being. On the contrary, it seems more plausible to suppose that there already were agent-neutral reasons to promote their well-being and that our shared histories have amplified their importance for me. Similarly, it seems plausible to suppose that other people’s reasons to treat me in relation-neutrally and relation-relatively mandated ways are to be explained, not just by the fact that I’m an agent with reasons all my own, but also by the ways in which other people’s reasons to promote my well-being prove to be insufficiently responsive to this fact about me.
The second point to note is that the most important of these other sorts of reasons seem likely to be very much more important than the agent-neutral reasons upon whose existence they depend. In the case of agent-relative reasons, the individual histories and circumstances giving rise to them are often centrally important to making people and their lives the particular people and lives that they are. It is only to be expected, therefore, that these reasons will often come to have an importance for agents very much greater than that of the agent-neutral reasons out of which they develop. For their part, the ways of treating other people that relation-neutral and relation-relative reasons mandate are often mandated because they exhibit respect for this particularity of people and their lives, so it is again only to be expected that these reasons will often have an importance for agents very much greater than that of reasons simply to promote other people’s well-being. Given how insufficiently responsive such reasons are to people’s agential nature, they are likely to be very much less important than the relation-neutral and relation-relative reasons that fill that need.
I realize that I am painting quickly here, and with a broad brush, but I hope that a plausible picture of reasons is coming into view.Footnote 11 It is a picture that at least partially vindicates both consequentialists and contractualists, while at the same time challenging each camp to make a serious concession. Consequentialists may welcome the claim that agent-neutral reasons are the most basic, but their treatment of other sorts of reasons is extremely hard to justify. If the importance of other sorts of reasons often is so much greater, should they not be making changes to their deontological (and not merely their axiological) views? Contractualists, for their part, may welcome this claim about the importance of other sorts of reasons, but the claim about agent-neutral reasons must surely give them pause. If such reasons are more basic than reasons of other sorts, should they not have a central (and not just a peripheral) role to play in moral theory?
Some consequentialists do concede one part of what I am asking of them, maintaining that morality requires agents to perform actions promising to satisfy the conjoined set of their agent-neutral and agent-relative reasons as completely as possible.Footnote 12 But that still leaves relation-neutral and relation-relative reasons out of account, and indeed it is hard to see how strong reasons of those sorts could be incorporated into a deontological view that remains fundamentally consequentialist in character. When it comes to contractualists, however, the challenge may not be so daunting, especially considering that I am only asking them to find a way to acknowledge that agent-neutral reasons are more basic than reasons of the other three sorts, while acknowledging myself that reasons of the other three sorts are frequently much stronger. It is hard to see how contractualists could give agent-neutral reasons equal billing, but the more limited concession that I am seeking may prove to be manageable.
2. The Metaethical Case against an Individualist Restriction
Scanlon neatly encapsulated many features of contractualist morality when he wrote that ‘An act is wrong if its performance under the circumstances would be disallowed by any system of rules for the general regulation of behavior which no one could reasonably reject as a basis for informed, unforced general agreement’ (Reference Scanlon, Sen and Williams1982, p. 110). The question is whether some systems of rules can reasonably be rejected on the ground that they do not take satisfactory account of agent-neutral reasons, or whether reasonable rejection of a system of rules must always be grounded in facts about the impact it would have on the personal concerns of representative individuals, whether that be on their opportunities of advancing their own agent-relative goals or on their prospects of being treated in relation-neutrally and relation-relatively mandated ways. Of course, if there were no agent-neutral reasons, this question would be moot. But, assuming that agent-neutral reasons do exist, how are we to answer it?
Scanlon originally suggested that agent-neutral reasons may be relevant to contractualist deliberation concerning certain peripheral cases but not to contractualist deliberation concerning more central cases.Footnote 13 Thus he allowed that agent-neutral reasons to preserve sites of great natural beauty may be relevant to contractualist deliberation concerning people’s treatment of such sites, but squarely denied that agent-neutral reasons to promote human or agential well-being are ever relevant to contractualist deliberation concerning people’s treatment of one another. Unfortunately, however, the metaethical rationale for this ‘individualist restriction’ was never made terribly clear. Why should some agent-neutral reasons be relevant to contractualist deliberation but others not? It’s true that trustees charged with representing the environment in contractualist deliberation would have nothing but agent-neutral reasons to preserve the environment to invoke in its defense. But why should the fact that people can invoke other reasons in their own defense preclude them from invoking agent-neutral reasons to promote well-being as well?
Perhaps the best answer to this question is the one that we have from Jay Wallace (Reference Wallace2019), according to which contractualist morality, at its core, is a nexus of directed duties individuals owe one another, violation of which would wrong specific individuals by failing to treat their concerns with equal and appropriate respect, justifying them in feeling a kind of resentment third parties would not be justified in feeling. But the question is why we can’t grant that contractualist morality is fundamentally relational in this sense while allowing that the content even of directed duties can sometimes be influenced by agent-neutral reasons. Indeed, Scanlon himself has come around to a view of this sort in recent years, still maintaining that objections to proposed systems of rules must be couched in terms of personal reasons, but now allowing that the force of these objections can sometimes be influenced by agent-neutral reasons, and not just by comparisons of the importance of the personal reasons lying behind them.Footnote 14 Again, if agent-neutral reasons did not exist, proposals like this would not be worth considering. However, given that agent-neutral reasons do exist, shouldn’t contractualists be taking such proposals more seriously?
Wallace does take Scanlon’s proposal seriously but worries that it would ‘attenuate’ the relational character of contractualist morality (Reference Wallace2019, p. 188). He suggests that this would be so even if the influence of agent-neutral reasons could be limited to cases where the personal reasons at stake on each side of the question are equivalent. Thus imagine a stock case where we have to choose between saving one and saving five; and suppose that each of these individuals has the same personal reasons to want to be saved. If agent-neutral reasons to promote human or agential well-being were allowed to tip the balance, our duty would be to save the five. But would it be specifically directed at them? If we went ahead and saved the one, would we treat the five’s concerns with less respect? Would they be justified in resenting our action in ways third parties would not be? Their concerns would be frustrated in a way the concerns of third parties would not be. But what Wallace wants to know is how, by frustrating their concerns, we would be failing to treat them with equal and appropriate respect, since the one’s concerns are just as important.
To make matters worse, Scanlon clearly does not intend to limit the influence of agent-neutral reasons to such cases. Thus consider a choice of the sort Johann Frick discusses, between saving one person from certain death and reducing by more than one the number of their colleagues who are statistically certain to suffer the same fate in the future.Footnote 15 Scanlon allows that agent-neutral reasons could tip the balance, making it our duty to look out for the colleagues, even though they each stand to gain nothing more than a small reduction in an already small risk of dying (Reference Scanlon, Stepanians and Frauchiger2021, p. 37). But would the resulting duty to look out for the colleagues be specifically directed at them in the way Wallace advocates? Would the colleagues be justified in feeling resentment at its violation of a sort third parties would not be? As before, its violation would frustrate concerns of theirs in a way it would not frustrate concerns of third parties. But their lesser importance makes it even harder to see how these concerns would be denied equal and appropriate respect.
As I read him, therefore, Wallace is challenging Scanlon to explain how agent-neutral reasons can influence what it takes to treat people’s personal concerns with equal and appropriate respect. I’m not sure how Scanlon himself would reply to this.Footnote 16 In the short time we have left, however, I want to suggest that a reply to this challenge may be found in the facts we have assembled about the relation that agent-neutral reasons bear to reasons of other sorts. The thought is that, if reasons of other sorts depend for their existence on the existence of agent-neutral reasons, we may sometimes fail to treat people’s personal concerns with equal and appropriate respect by failing to take the agent-neutral reasons out of which they arose into proper account. If we don’t look out for the colleagues, for example, we may fail to treat their personal concerns to minimize their risk of death with equal and appropriate respect by not taking proper account of the agent-neutral reasons to minimize loss of human life out of which they arose.
The thought here is not that we will fail to treat the colleagues’ concerns with equal and appropriate respect even if we increase by only very marginally more than one the number of them who are statistically certain to die. And it is certainly not that taking proper account of the agent-neutral reasons lying behind personal concerns requires us to acknowledge that aggregated concerns about enduring minor burdens, say, could win out over a concern about facing certain death.Footnote 17 The thought here is rather that the increase in the number of colleagues who are statistically certain to die could be so great that our duty would be to look out for them instead of looking out for the one, and that, if that were the case, and we went ahead and looked out for the one all the same, we would not just be frustrating the colleagues’ concerns but actually failing to treat them with equal and appropriate respect, focusing too much on their comparative lack of importance and not enough on their shared origins in agent-neutral reasons.
Now that it’s clear where I’m going with the idea that agent-neutral reasons are more basic than the others, some readers may want to go back and consider again the very sketchy arguments that I offered for that claim. Indeed, they may now want to challenge the very idea that some sorts of reasons are more basic than others, understood in the sense that the existence of some sorts of reasons is needed to explain the existence of others. If the facts about human nature suffice to explain the existence of agent-neutral reasons to promote human well-being, why can’t the existence of agent-relative, relation-neutral, and relation-relative reasons also be explained directly by the facts?Footnote 18 Why can’t the facts about our history suffice to explain my agent-relative reasons to promote a friend’s well-being? Similarly, why can’t the facts about my agential nature and about the relations in which I stand to other people suffice to explain the reasons that other people have to treat me in relation-neutrally and relation-relatively mandated ways?
My response to this challenge turns in part on the fact that I take normative properties also to be causal properties and so expect them to exhibit a certain amount of structure. It would be quite remarkable, to my mind, if it were to turn out not just that there are different sorts of reasons but also that none are more basic than others. But I of course acknowledge that this metaphysical conviction is not widely shared and needs defending, and that, even if it can be successfully defended, it will not take us very far. Even if it’s to be expected that some sorts of reasons will be more basic than others, it remains to be seen whether agent-neutral reasons really are more basic than all the rest. It remains to be established that agent-relative reasons develop out of agent-neutral ones as facts about people’s histories and circumstances amplify the importance of certain of those agent-neutral reasons for them, and that relation-neutral and relation-relative reasons emerge only because agent-neutral and agent-relative reasons to promote human well-being prove to be insufficiently responsive to the facts about people’s agential nature.
How might such claims be established? I think we would do well to follow the example that Scheffler has set in his discussions of the importance that future generations have for people. Scheffler’s immediate concern in those discussions is with the importance that people place on the fact that the activities they prize and the traditions they cherish will continue long after they die. So his primary focus is on what people do value, not on what they should value, and on their agent- and relation-relative valuings, not on their agent- and relation-neutral ones. Still, Scheffler makes an extremely strong case for concluding that people would not prize the activities and cherish the traditions they do if they did not also value their continuation long into the future. And that is strong evidence for concluding that the existence of people’s reasons to pursue such activities and honour such traditions depends on the existence of other, less obvious, agent- and relation-relative reasons. Of course, that’s not yet to show that it depends on the existence of agent-neutral reasons, but it raises the real possibility that further ‘excavation’ of people’s valuings will show that it does.Footnote 19
Consider again the colleagues in Frick’s example. Suppose that they understand perfectly well that, while the one person is being saved at the cost of a large increase in the overall number of their lives that will be lost in the future, the increase in their individual risk of dying will not be very great. Initially, they might be persuaded by the argument that, being comparatively so much smaller, their personal concerns do not constitute reasonable grounds on which to base an objection. But could they not be brought to see that, by allowing that they count only comparatively and not collectively, the decision to save the one person fails to exhibit equal and appropriate respect for their personal concerns because it fails to take proper account of their shared origins in agent-neutral reasons? And would that not lead them to resent the choice that was made, and to feel they are both perfectly and uniquely justified in resenting it?
If such questions end up getting answered in the ways I suspect they will, we will have strong grounds for concluding that contractualists should jettison the individualist restriction. So at least we have an idea of the form our argument should take, even though there is a lot of heavy lifting that remains to be done. What is less clear to me is whether we also have further vindication of my assumption that normative properties are not just substantively real but also straightforwardly causal. I appealed to that assumption to make sense of the idea that the existence of some sorts of reasons might depend on the existence of others, but I didn’t show that this is the only way to do that, nor did I show that this is the best way to do that. Could realists of a non-naturalist bent also make sense of these dependence relations? That’s a very interesting question that I’ll have to leave for another time.Footnote 20
Open access
