Obligation and Reasons

Before I begin my argument, I want to make two preliminary points concerning the relation between moral obligations and normative reasons for acting, both moral reasons and reason, all things considered. The first concerns the relation between moral obligations and moral reasons.

There is a close conceptual connection between the ideas of normative reason (whether “reason” is used as noncount noun (“there is reason”) or as a count noun (“there are reasons”)) and oughts of different kinds. First, “reason” as a noncount noun. What we ought to believe, for example, is what there is reason for us to believe; what we ought to desire is what there is reason for us to desire; what we ought to do is what there is reason for us to do, and so on. Roughly, it would seem, for any agent, S, attitude, A, and circumstance C:

Normative reasons (“reason” as a count noun) are pro tanto ought-making or noncount-reason-making facts. A reason to do something, for example, is a fact that makes it true that there is reason, or that one ought pro tanto to do that thing, that one ought, that is, other things being equal. If, when taking all the reasons that bear on the case into account, the reasons in favor of S’s doing A in C outweigh those in favor of S’s not doing A, then A is what there is reason for S to do in C, or what S ought to do, all things considered.

Because this is true quite generally, and because it is often natural to express deontic moral claims like S is morally obligated (or has a moral duty) to do A in C, by saying S morally ought to do A in C, it can be tempting to try to understand these deontic moral claims in terms of the balance of moral reasons. I have argued, however, that this is a mistake (Darwall, Reference Darwall2006, Reference Darwall, Lord and Maguire2016, Reference Darwall2019). We should distinguish two different senses of the moral ought.

Note first that for there to be a clear moral reasons sense of the moral “ought,” we must be able to distinguish between moral and non-moral reasons for acting (Darwall, Reference Darwall2019). Even if we can do that, however, there are good reasons for thinking that moral obligation claims cannot be understood in terms of the balance of moral reasons.Footnote ⁶ We can imagine two people agreeing that moral reasons most highly favor a given action, but also coherently disagreeing about whether that action is morally obligatory or would be wrong to omit. Imagine, for example, a disagreement between an act consequentialist and someone like Scheffler in The Rejection of Consequentialism (Scheffler, Reference Scheffler1982). Both reject agent-relative deontological constraints, and both could agree that morality always most recommends the optimific act.Footnote ⁷ But they disagree about whether especially burdensome costs to the agent can render an action that would otherwise be morally obligatory morally permissible. The act consequentialist thinks that non-optimific acts are always morally wrong, whereas the Schefflerian holds that they are sometimes not wrong when they are covered by an agent-relative prerogative or permission that defeats the claim of moral wrong when the personal costs are sufficiently high in relation to the impartial benefits.Footnote ⁸ (Think, e.g., of a rescue case where the risks to the rescuer are great enough to render rescue not obligatory but not so great to make it morally prohibited.)

For such a disagreement to be so much as possible, the parties must distinguish the distinctively deontic concept of moral obligation (“morally ought” in the fully deontic sense) from the “moral ought” of moral reasons. I argue that this difference is rooted in the fact that moral obligation is conceptually connected to accountability in a way that claims about moral reasons are not (at least, not on their face) (Darwall, Reference Darwall2006, Reference Darwall, Lord and Maguire2016).Footnote ⁹ What we are morally obligated to do are things that it would be blameworthy for us to omit without excuse. In effect, I claim, the disagreement just considered concerns whether failing to do what is optimific always requires an excuse in order to avoid being blameworthy. According to someone who defends agent-relative permissions or prerogatives, sufficiently large agent cost can justify, and not just excuse, failures to assume them since it would be unreasonable for agents to be required to bear them. They make an action that would otherwise be wrong morally permissible.

What someone is morally obligated to do thus differs from what that person morally ought to do in the moral reasons sense. Note that I have not been distinguishing in this section between relational and nonrelational obligations. It should be evident that both of these fully deontic notions differ from claims about the balance of moral reasons. This then gives us three distinct notions: moral ought in the moral reasons sense, relational obligation, and nonrelational moral obligation.

The second preliminary point concerns the relation between moral obligations and normative reasons for acting all things considered. Moral reasons are sometimes said to have a special overriding force that trumps, or perhaps even “silences,” other reasons. This is not, as I see it, however, true of moral reasons in general, but only of moral obligation, in particular. It is only when the relevant moral considerations combine to create an all things considered moral obligation that there is necessarily reason, all things considered (including normative reasons for acting of all kinds), to comply with it.

What guarantees the link between moral obligations and normative reasons, I argue, has to do with moral obligation’s conceptual connection to accountability (specifically, to blameworthiness or culpability) and with conditions of justifiable blame (Darwall, Reference Darwall, Lord and Maguire2016). The idea is that it is a presupposition of blame that there was not sufficient reason for the person one blames to do what one blames him for doing.Footnote ¹⁰ There is a kind of incoherence in simultaneously holding the attitude of blame toward someone for his action and simultaneously believing that he had sufficient reason for doing what one blames him for. Since this is so, blame cannot be justified unless there was conclusive reason for the agent not so to have acted. But if, as I argue, an action is morally obligatory if, and only if, it is a kind of action that would be blameworthy were it performed without excuse, then it cannot be morally obligatory without there being reason to blame (unexcused) noncompliance, and therefore without there being sufficient reason to comply.Footnote ¹¹

There are two important consequences for our purposes. First, the kind of moral obligation that guarantees all things considered reason for compliance must itself be an all-things-considered moral obligation. Clearly there can be pro tanto moral obligations to do something without there being normative reason, all things considered, to do it, if only because there may be conflicting pro tanto obligations that make some other action morally obligatory, all things considered. In the stock case, I may have promised to meet you for lunch but then find myself the only person who can prevent some terrible disaster. If I can do the latter at little cost, I will arguably be morally obligated to do it, with the latter pro tanto moral obligation outweighing the one created by my promise. It is only all things considered moral obligations that guarantee all things considered normative reason.

The second point is that the kind of moral obligation that brings conclusive normative reasons for compliance is therefore nonrelational moral obligation and not relational obligation. When we speak of obligations as either pro tanto or all things considered, it is the concept of nonrelational obligation that we have in mind (even if the pro tanto obligation derives from a relational obligation).

There may be a possible use for the pro tanto/all things considered distinction within any obligor/obligee pair. We can perhaps sensibly ask what someone owes to someone else all things considered, weighing up relational obligations within that pair. But there is not any sensible relational way of weighing up relational obligations across obligees. Even if we stipulate that there is no nonrelational moral obligation that is not grounded in relational obligations, it will nonetheless remain true that both the pro tanto and the all things considered moral obligations relational obligations ground must themselves be nonrelational moral obligations.

Moral Obligation and Justification

To approach these points from a different direction, let us see how far we can get with the concept of relational obligation alone without any notion of nonrelational moral obligation. And again, let us assume that there are no ungrounded nonrelational moral obligations, that all the most basic obligations are relational. What can we and what can we not do if our only concept of moral obligation is relational?

Suppose that I have made two conflicting promises, one to A and one to B, which I cannot both keep. By promising to A, I have created a relational obligation to A, and by promising to B, I have created a relational obligation to B. I have now to decide what to do, which relational obligation to keep. I can keep my promise to A, breaking my promise to B, or I can keep my promise to B, breaking my promise to A. (Of course, I might not keep either promise, thereby violating my obligations to both A and B.)

There is broad agreement that some kind of correlativity holds between relational obligations and claim rights (Hohfeld, Reference Hohfeld and Cook1923), or, at least, between relational obligations and claims of some sort (Wallace manuscript). By creating my obligation to A by a promise, I give A a right or a claim to my doing what I promised. And ditto with B. So now I have two relational obligations, which I cannot both keep, and therefore, two rights or justified claims that I cannot both honor.

Following Feinberg, I take it that to have a right or claim is to have authority or standing as the right or claim holder to address or press the right or claim in a variety of ways (Feinberg, Reference Feinberg1980). I call this standing individual authority, since it is one that the obligee has as the very individual to whom the obligor is obligated (the individual having the claim) (Darwall, Reference Darwall and Shafer-Landau2012). Only the obligee has the individual authority to do a variety of things, including waiving the claim or right (if it is alienable) and holding the obligor personally accountable if it is violated, for example, by seeking compensation or, perhaps, by forgiving the obligor. Only the right holder or perhaps those who are authorized to act on her behalf have this authority. We might call these facts concerning my obligations to A and B, following Thompson, relational normative facts.

What I want now to consider is how relational normative facts bear on two different kinds of questions. The first is the agent’s deliberative question of what they should do given their relational obligations. Here, we can distinguish three different questions of this kind: (i) What do I have most reason to do? (ii) What do I have most moral reason to do? and (iii) What am I morally obligated to do, all things considered? The second kind of question I want to consider is an “observer’s” question, what are fitting attitudes in response to the agent’s conclusion regarding questions of the first kind and her subsequent action.Footnote ¹² We will focus on the first question in this section and the second in the next.

In asking what I should do given my obligations to A and B, respectively, I am asking a question that the relational normative facts taken by themselves cannot adequately answer. In order to decide which promise I should keep, it seems, I need to ask which obligation is weightier. Weight is not itself, however, a relational notion. Although one obligation is owed to A and the other to B, the weight of the first is not itself owed to A; neither is that of the second owed to B. If one is weightier than the other, then it is weightier nonrelationally.

But what kind of weight is involved here? If by the weight of a relational obligation we mean its pro tanto obligatory weight, that is, how much weight it has in determining what one is morally obligated to do, all things considered, then we will need to invoke the notion of nonrelational moral obligation. As we noted before, neither the idea of pro tanto obligation nor that of obligation, all things considered, is a relational notion. They both concern nonrelational obligation, in this case, which relational obligation I am, both relational obligations considered, nonrelationally morally obligated to comply with.

Notice, by the way, that we cannot avoid making use of the concept of nonrelational moral obligation by accepting some version of incommensurability or moral dilemmas. If I will do wrong whether I keep my obligation to A or my obligation to B, that just means that in either case I will violate my all things considered nonrelational moral obligation. That I will violate my obligation to someone, whichever promise I keep, is itself what sets the deliberative problem, not a deliberative response to it.

Indeed, the very concepts of right and wrong are themselves “monadic” rather than relational notions. Wronging is a relational concept, as is doing right or wrong by someone. But that is different from doing wrong or acting rightly simpliciter. These conceptually implicate nonrelational moral obligation. Necessarily, an action is wrong, pro tanto or all things considered, if, and only if, it violates a pro tanto or all things considered nonrelational moral obligation. And necessarily, an action is morally permissible (“all right”) if it does not violate one’s nonrelational moral duty.

However, we can ask which of my obligations to A or B is weightier in senses that do not necessarily involve the notion of nonrelational moral obligation. Suppose I ask which relational obligation I have most reason, all things considered, to comply with. Clearly, we do not need the notion of nonrelational moral obligation to raise this question, and we might not need the notion of nonrelational moral obligation to answer it. Suppose my promise to A concerned a relatively trivial matter whereas my promise to B involved something of greater importance. Considering just these two relational obligations, I might come to the view that I have most reason to keep my promise to B rather than my promise to A, all things so far considered. But from the fact that my obligation to B gives me a weightier reason than my obligation to A, nothing follows about what I have reason to do, all things, that is, really all things, considered.

In the last section, I discussed the claim, for which I have argued elsewhere, that all things considered nonrelational moral obligations entail conclusive reasons for acting (Darwall, Reference Darwall, Lord and Maguire2016). If, all things considered, I am nonrelationally morally obligated to keep my promise to B (rather than keeping my promise to A, or anything else I might do), then that is what I have most reason to do. I do not have sufficient reason to do anything else. That must be so, again, because keeping my promise to B could not be morally obligatory all things considered unless it would be blameworthy to fail to keep it (without excuse), and blame for some action can be fitting only if there is not sufficient reason not to do it.

Nothing like this can follow from the fact that in failing to keep my promise to B, I am violating a relational obligation to B, even one that I have a weightier reason to comply with than my obligation to A. It is obviously consistent with my obligation to B giving me a weightier reason than my obligation to A, that I have, all things considered, reason to keep my promise to A (or to do something else entirely). So far as the concepts involved go, there might be weighty nonmoral reasons, say, reasons of self-interest, that outweigh the reasons created by or involved in my relational obligation to B.Footnote ¹³

Suppose, however, that we restrict ourselves to moral reasons. And suppose we take the view that my promise to B creates a weightier moral reason than my promise to A. We can do this without invoking the notion of nonrelational moral obligation. And let us stipulate further that there are no other moral reasons that favor complying with either relational obligation. We could then think that discharging my obligation to B rather than my obligation to A is what I have most moral reason to do. Where would that leave us?

We still would not be in a position to conclude that keeping my obligation to B is what I have most moral reason to do, even all moral reasons considered. To see why, consider a kind of case described by Frances Kamm (Reference Kamm1985). We can stipulate that my obligations to A and B are the only relational obligations in the case and, again, that there are no ungrounded nonrelational moral obligations. Consistently with this, there may nonetheless be great goods that I am in a position to, but not obligated to, bring about, but which I should bring about in the moral reasons sense. To adapt Kamm’s case, suppose that as I am on my way to keeping my promise to B, I come upon an accident victim whose kidneys have been crushed and who can survive only if I give them one of my kidneys (Kamm, Reference Kamm1985, pp.119–121). She argues, plausibly, that although it would be morally better for me to donate my kidney, I am not (nonrelationally) morally obligated to; that would be supererogatory.

What lesson should we draw? We have been assuming that all moral obligations are relational and that we can somehow assign weights to the moral reasons for complying with them in order to determine which we have most moral reason to comply with when they conflict. This we can do without invoking the notion of nonrelational moral obligation. When, however, we consider cases such as Kamm describes, we encounter a problem. Even if the assigned weights determine which relational obligation I have most reason to comply with, they will not determine what I have most moral reason to do, all things considered, since there may be more moral reason to do something else that does not discharge any relational obligation.

To be sure, as Kamm analyzes things (correctly, in my view), something similar will be true if we parse the case in terms of nonrelational obligation. Though I have a weightier pro tanto obligation to keep my promise to B than to keep my promise to A, I have more moral reason not to keep either and donate my kidney, thereby violating both pro tanto obligations. The crucial difference is that we need not say in such a case that I violate (nonrelational) moral obligation, all things considered. The great good of preserving life is obviously a sufficient justification for not keeping my promise to B, so that all things considered, donating the kidney is not wrong.

This response is available owing to the conceptual connection between nonrelational moral obligation and culpability, which is not in play with relational obligation. Blameworthiness claims can be defeated both by excuses and by justifications. An excuse is an extenuating factor that shows that while what the agent did was wrong, she should not be blamed or that blame should be mitigated. A justification, by contrast, defeats the charge of wrongdoing. It shows, as in the instant case, that although not keeping my promise to B is wrong—violates my nonrelational moral obligation, pro tanto—it is not, all things considered.

The connection to culpability also shows why even though donating my kidney is not wrong, neither would it be wrong for me not to donate my kidney. I am under no all things considered nonrelational moral obligation to do so. I would have such a moral obligation only if failing to keep it would be blameworthy (lacking excuse). But since, as Strawson points out, reactive attitudes like blame implicitly make a demand, such an action would be blameworthy only if we could reasonably make such a demand. Kamm supposes, and I agree, that we cannot reasonably demand kidney donation in such a case; hence, we cannot blame the (clear-eyed) refusal to donate. Therefore, refusing to donate cannot be wrong, that is, contrary to (nonrelational) moral obligation, all things considered.

Without the notion of nonrelational moral obligation, we cannot say these things. We can of course say that I have more moral reason to donate my kidney despite my relational obligations to A and B. But is that a feature or a bug? A’s and B’s claims to having promises made to them kept would presumably be unchanged by the existence of the opportunity to donate a kidney. Of course, we would be unlikely to say in such a case that either A or B had been wronged if I donated my kidney, but it is hard to see how that could be explained without having to assume that donating my kidney would not be wrong (that is, would violate no all things considered moral (nonrelational) obligation). Donating the kidney would do no wrong; therefore, do no wrong to either of them.

Suppose we were to try to account for this by saying that, although A’s and B’s underlying claims are unchanged, once the opportunity for kidney donation arises, neither A nor B is any longer in a position to hold me to my promise. So I do not wrong either by donating my kidney and failing to keep my promise. Again, it is hard to see how we could justify this conclusion without assuming that donating would not be wrong (nonrelationally), and therefore would wrong neither A nor B. Moreover, we want to be able to say that were I to refuse to donate my kidney, this would not wrong the accident victim either. We can easily say this if we help ourselves to the idea that failing to donate would not be nonrelationally wrong and so would not wrong the accident victim.

Suppose, however, that what would be required to help the accident victim is something much less costly. Maybe the victim is in the situation of Singer’s famous case of a drowning child in a shallow pool, and I can save her by wading in and ruining my shoes (Singer, Reference Singer1972). We might say in this case that I would be obligated to save the child and that I owe this to the child. Failing to wade in would not just be wrong nonrelationally; it would wrong the child. This seems right. But how, now, are we to explain why it would wrong the child though failing to come to donate a kidney would not wrong the accident victim. Obviously, moral reasons considered independently of moral obligation cannot explain it, since there is more moral reason to help both. I don’t see any way of explaining this that does not already assume that failing to help the drowning child would be wrong nonrelationally whereas refusing to donate a kidney would not be.

Blame, Objection, and Complaint

We have been focusing on the conclusory deliberative role of the concept of nonrelational moral obligation as opposed to relational obligation. All things considered, nonrelational moral obligation brings both conclusive moral reason and conclusive reason, all things considered, for compliance in its wake in a way that relational obligations cannot. This is so, I have been arguing, because nonrelational moral obligation has a conceptual tie to blameworthiness that relational obligation does not have. Since blame presupposes that its target action was unjustified, it can be justified itself only if this is so.

In this final section, I want to consider further ways in which blame contrasts with responses that are conceptually implicated in the violation of relational obligations. As Jay Wallace points out, “the notion that someone in particular has been wronged by [an] action is conceptually connected to the idea that the wronged party has special ground for complaint” (Wallace, Reference Wallace2007, p. 29). The same point can be put in the “language of authority,” he says, “by speaking of the aggrieved party’s privileged authority to complain or object when relational obligations to them have been violated” (Wallace, Reference Wallace2007, p. 29). Wallace and I are agreed that obligees’ distinctive standing to complain or object is intrinsic to the concept of relational obligation.

As I see it, this standing is part of the distinctive individual authority obligees have as the very individual to whom the obligor is obligated. It contrasts with the representative authority that is conceptually implicated in nonrelational moral obligation. Actions that violate nonrelational moral obligations are blameworthy, lacking excuse, and blame is an attitude that is fitting not just from the standpoint of obligees, if there are any, but from anyone’s perspective, including the obligor, as when she has the attitude of blame toward herself in feeling guilt.

To see this difference between blame and objection and complaint, note that while self-blame is a thoroughly familiar phenomenon—indeed it is the attitude that we call violators to have toward themselves when we blame them—there is no intelligible notion of self-objection or self-complaint. Objection and complaint are not attitudes but communicative acts that essentially involve explicit address to another person. They are exercisings of obligees’ normative power to object and complain.

An objection is a move in a proceeding, whether legal, intellectual, or some other kind of interpersonal transaction.Footnote ¹⁴ When an attorney raises an objection that is “in order” in a court of law, the judge must then consider and either sustain or overrule it. When a philosopher’s claims or arguments elicit objections at a philosophical talk, they must attempt to respond to them. And so on. Similarly, complaining is something one does to someone else. (We do put forward objections to our own claims and arguments, but that is different from the kind of objection that an obligee might make to our having made them if say, we were obligated not to make them because this amounted to plagiarizing the obligees. The latter is not the kind of objection one could make to oneself.)

To be sure, we do also speak of blaming as an action. And it may well be that not everyone has the very same standing to engage in blaming actions or other expressions and that victims have greater standing. But if an action is blameworthy, then the attitude of blame is no less fitting a response from someone reading of it in the newspapers than it is from the victim or from the offender himself.

There is an attitude closely akin to blame that is tied to the victim’s perspective, namely resentment. Whereas blame is the fitting response to unexcused wrongdoing, resentment is the fitting response to having been wronged oneself. When we judge whether resentment is fitting, we assess how it would be fitting to feel in the victim’s shoes. This differs from blame. We assess culpability not from the perspective of any particular individual like the victim, but from the impartial standpoint of a representative person or member of the moral community.

Blame is an attitude that presupposes our mutual accountability as equal persons or members of the moral community. We blame someone from the same point of view we expect her to take up in acknowledging her wrongdoing and blaming herself. Guilt is the attitude that reciprocates blame both in accepting blame’s content (unexcused wrongdoing) and in beginning to take responsibility for the wrongful act, as blame implicitly calls on its object to do (Darwall, Reference Darwall, Roughley and Schramme2018).Footnote ¹⁵

In the last section, I argued that we need the concept of nonrelational moral obligation from the deliberative standpoint since without it we cannot think about our moral obligations, all things considered, in a way that enables us to come to deliberative conclusions. In this section, I have been arguing that without the concept of nonrelational moral obligation, we cannot hold one another answerable as equal, mutually accountable moral persons. Ultimately, these different points come from the same root. It is (nonrelational) moral obligation’s link to blameworthiness that grounds both its connection to deliberative conclusiveness and its tie to mutual accountability.Footnote ¹⁶