Moral Obligation: Relational or Second-Personal?

1. Introduction

The Problem of Obligation is the challenge to explain the features that distinguish moral obligations from other normative phenomena in a way that vindicates their place within modern moral philosophy. In recent years, two superficially very similar accounts have been put forward to solve this problem: the Second-Personal Account and the Relational Account. Both accounts purport to explain moral obligations as part of a system of norms which are constituted by moral relations that hold between persons as such. However, the subtle differences between the ways in which the two accounts conceive of these relations lead to vastly different conclusions about the place of moral obligations in practical reason and moral psychology.

According to the Second-Personal Account, put forward by Stephen Darwall, obligations implicate a distinct class of practical reasons, “second-personal reasons”, whose “validity depends on presupposed authority and accountability relations between persons and, therefore, on the possibility of the reason’s being addressed person-to-person” (2006: 8, italics altered). In other words, obligations are grounded in what persons can legitimately demand of one another.

According to the Relational Account, which is independently put forward by both R. J. Wallace and Ariel Zylberman, moral obligations are fundamentally relational because they are “constitutively connected to claims that others have against us, just insofar as they are persons” (Wallace 2019: 1; also see Wallace 2013; Zylberman 2017; 2021). In this picture, obligations are invariably directed, that is, they are obligations to someone.

While both accounts hold that moral obligations are in some sense based on persons’ claims or demands, the Relational Account denies that these claims are in turn based in any practical authority, such as the authority to make or address these claims (e.g., Wallace 2007: 29; Zylberman 2017: 935). Instead, it regards claims as primitive normative phenomena which are not grounded in anything more fundamental (e.g., Wallace 2019: 16; Zylberman 2021).

Wallace and Zylberman contend that the Second-Personal Account’s solution to the Problem of Obligation is unsatisfactory (Wallace 2007; 2019; Zylberman 2017; 2021). They argue that the Second-Personal Account is committed to the claim that we have an obligation to ϕ only if, and because, someone demands that we ϕ. This claim is implausible, however, since the existence of many obligations (e.g., the obligation not to enslave others) does not seem to depend on whether others demand that we comply with them. To the contrary, such demands seem to be warranted by the prior and independent existence of obligations. Following Zylberman, we may refer to this issue as the Problem of Antecedence (2017: 925).

Since the Relational Account’s proposed solution to the Problem of Obligation avoids the Problem of Antecedence, Wallace and Zylberman conclude that the Second-Personal Account is dialectically inferior to the Relational Account. In other words, they conclude that theorists who are attracted to the Second-Personal Account due to its promise to solve the Problem of Obligation should be even more attracted to the Relational Account.

In this paper, I defend the Second-Personal Account against the charge of dialectical inferiority, by arguing for two main claims:

1. The Relational Account does not solve the Problem of Obligation.

2. The Second-Personal Account does not fall prey to the Problem of Antecedence.

I start by introducing the Problem of Obligation, characterising it in a way that Darwall and defenders of the Relational Account (henceforth ‘Relationalists’) can agree on (Section 2). Subsequently, I explain how the Second-Personal Account purports to solve the Problem of Obligation by appeal to the notion of second-personal reasons (Section 3).

Next, I develop my argument for claim (1): that the Relational Account does not solve the Problem of Obligation. This argument proceeds in three steps. First, I outline the Relational Account and clarify that, in order to solve the Problem of Obligation, Relationalists need to disentangle the relational character of moral obligations from their obligatory character, and then explain the latter in terms of the former (Section 4). Second, I note that a promising way for Relationalists to do so is by appealing to practical relations between persons without which moral obligations are unintelligible, but I argue that the Relational Account does not seem able to accommodate such a close link between practical relations and moral obligations (Section 5). Third, I contend that the Relational Account also does not offer a way of solving the Problem of Obligation beyond the appeal to practical relations (Section 6).

Subsequently, I turn to my argument for claim (2). I start by showing that the accusation that the Second-Personal Account falls prey to the Problem of Antecedence rests on a misunderstanding of that account. This account does not depict obligations as dependent upon what persons happen to demand of one another, but on what they can legitimately demand, which is in turn dependent upon the constitutive standards of the practice of making demands of one another (Section 7). I then consider the potential objection that this response undermines the Second-Personal Account’s ability to solve the Problem of Obligation. According to this objection, by depicting moral obligations as grounded in independent and prior moral standards, the Second-Personal Account fails to set obligations apart from other normative phenomena (Section 8). I respond to this objection by pointing out that the constitutive standards of making demands are not independent and prior moral standards in a troubling sense because they are internal to that practice (Section 9). I conclude that the Second-Personal Account is not dialectically inferior but superior to the Relational Account (Section 10).

My arguments in this paper do not purport to establish the overall plausibility or implausibility of the Second-Personal Account or the Relational Account. There are many substantive objections to either account that I bracket here. All I try to show is that, for theorists who are troubled by the Problem of Obligation, the Second-Personal Account is still more promising than the Relational Account.

2. The Problem of Obligation

The Problem of Obligation can be seen as a variation on the challenge that G. E. M. Anscombe introduced in her seminal paper “Modern Moral Philosophy” (1958).¹ Roughly, Anscombe argued that the idea that morality obligates us suggests that it takes the form of laws, and laws are characteristically issued by some authority. While God might serve as the relevant authority within theistic ethics, it is not clear who might play that role in a secular moral theory. Accordingly, Anscombe advised that, unless we are prepared to accept a divine command theory, we should abandon the notion of moral obligation, and the associated “law conception of ethics”, altogether (1958: 5).

To solve the Problem of Obligation, then, one needs to explain the features of obligations that set them apart from other normative phenomena in a way that vindicates their place in our conception of morality. This requires that obligations are rendered intelligible against the background of our modern convictions (Darwall 2006: 29–30; Wallace 2019: 22–23).

To get a better grasp of the problem, we need to take a closer look at the distinctive features of obligation. The features listed by Darwall do not entirely coincide with those named by Relationalists. However, there are three key features of obligation that both accounts are trying to explain. These features are sufficient to frame a shared dialectic.

The relevant features can be illustrated with an example by Darwall (2006: 5–10). Suppose one person—let’s call him Arnold—is standing on top of the foot of another person—let’s call her Bella. In this situation, Arnold might acknowledge the fact that by standing on Bella’s foot he is making Bella worse-off than she would otherwise be, thus bringing about a sub-optimal state of the world. Following Darwall, we might refer to this fact as an “outcome-based reason” (2006: 8). Darwall and his Relationalist critics agree that Arnold’s outcome-based reason does not amount to an obligation to withdraw his foot (Wallace 2019: 39–41; Zylberman 2017: 922). In particular, they hold that there are at least three important differences between outcome-based reasons and obligations.

The first important difference between obligations and outcome-based reasons is that the latter are best interpreted as reasons in what we might call the standard sense. That is, they are considerations that recommend, or speak in favour of, some course of action.² By contrast, obligations present themselves as requirements that make demands on our agency.

This difference between reasons and obligations manifests itself in our practical deliberation. Arnold’s outcome-based reason does not purport to determine, by itself, whether he ought to withdraw his foot or not; for it does not purport to rule out the existence of reasons not to withdraw his foot, which might be more weighty. By contrast, if Arnold is under an obligation to withdraw his foot, then this obligation will present withdrawing his foot as what he ought to do, whatever else is true. In this way, obligations purport to conclude deliberation; they purport to rule out that further considerations could make a difference as to what we ought to do.³ This is not to say that obligations are absolute in the sense that they can never be outweighed. On the contrary, obligations seem to be weighted, pro tanto considerations. Nevertheless, obligations purport to conclude deliberation from the start. In Wallace’s words, obligations “enter the deliberative field in a distinctive normative key” (2013: 164). We may capture this distinctive normative key by saying that obligations purport to give us “conclusive reason”, or that they function as “presumptive constraints” on our deliberation (Darwall 2006: 26 and Wallace 2019: 27, respectively).

The difference between obligations and reasons in the standard sense also manifests itself in the degree of “deliberative discretion” that they allow (Wallace 2019: 49). While it may conceivably be ‘up to us’ to decide how we respond to our reasons, we have no similar liberty in responding to our obligations. This is also suggested by the conceptual possibility of supererogation, that is, acts that are good but not required. This possibility implies that there is a conceptual difference between doing what we are obligated to do on the one hand and acting on the balance of reasons on the other hand (Darwall 2013a: 5; Wallace 2019: 41). To act on the balance of reasons may be good, admirable, desirable or valuable, but “[w]hen we are morally obligated, we are not morally free to act otherwise” (Darwall 2006: 27).

The second important difference between outcome-based reasons and obligations is that the former are agent-neutral (Darwall 2006: 6; Wallace 2019: 38). That is, since Arnold’s outcome-based reason stems entirely from the relative value of states of the world, it equally serves as a reason for anyone who, like Arnold, is able to affect which state of the world comes about. For example, if a bystander is in a position to remove Arnold’s foot from on top of Bella’s, he has the same reason to do so as Arnold. Connectedly, if Arnold could bring about an even better state of the world by continuing to stand on Bella’s foot (suppose that, somehow, this diminishes the overall number of acts of this type being performed in the world) outcome-based reasoning would favour his doing so. By contrast, obligations are agent-relative: if Arnold is under an obligation to remove his foot, then this requires him in particular to remove his foot (regardless of how this affects the overall number of feet being trampled upon in the world).

The third important difference between outcome-based reasons and obligations is that obligations are intrinsically linked to the agent’s accountability (Darwall 2006: ch. 5; Wallace 2019: ch. 3). Insofar as Arnold is obligated to withdraw his foot, Bella can legitimately hold him accountable for failing to do so. This is not yet to say that Bella is justified in punishing or otherwise sanctioning Arnold for his failure to comply. But Arnold’s obligation, by its very nature, entails that his failure to comply, lacking excuse, warrants Bella’s adoption of certain Strawsonian “reactive attitudes”, such as blame or resentment, towards him (Darwall 2006: 17; Wallace 2019: 71; see Strawson 1962). This suggests that moral obligations, by their very nature, are standards compliance with which we can legitimately expect or demand from one another (Darwall 2006: 17; Wallace 2019: 70; Zylberman 2018: 740). For by adopting reactive attitudes towards an agent, we hold them to these standards, thereby addressing purportedly legitimate expectations or demands to them (Darwall 2006: 17; Wallace 2019: 70; Zylberman 2014: 170). As Darwall puts it, these reactive attitudes come with “an implicit RSVP” in the sense that they “aim to draw [the agent] into an exchange that will constitute their being held accountable” (2006: 145). By virtue of their intrinsic connection to reactive attitudes, then, moral obligations structure a practice of person-to-person address.

By contrast, Arnold’s outcome-based reason to move his foot, by itself, does not warrant the adoption of reactive attitudes towards him. That he would bring about a superior state of the world by moving his foot does not entail that he is accountable to anyone. Indeed, the only standing that Bella can invoke by ‘addressing’ this reason to Arnold has a purely third-personal, epistemic character (Darwall 2006: 6; Zylberman 2017: 924). It implies her ability to advise or counsel Arnold on matters of values or reasons, but not the standing to hold him to a standard by demanding that he comply with it. In fact, it seems that the most natural way to accommodate accountability practices within outcome-based reasoning is by evaluating the norms that regulate the adoption of reactive attitudes in terms of their propensity to bring about good or bad states of affairs. However, as Darwall and his Relationalist critics agree, the value or desirability of the outcome of holding an agent accountable cannot capture, let alone justify, our accountability practices because they are reasons of the wrong kind for adopting reactive attitudes (Darwall 2006: 15; Wallace 2019: 77; also see Strawson 1962).

It is important to note that the link between obligation and accountability is not just another aspect of the distinction between obligations and reasons in the standard sense. We cannot account for Arnold’s obligation to remove his foot from on top of Bella’s by replacing his outcome-based reason with a normative requirement. There are several ways to spell out the idea of such a requirement. To name only two, the requirement might arise from a primitive normative fact or express a constitutive standard of agency (see Ross 1930/2002 and Korsgaard 2009, respectively). In the first case, the only standing that Bella can invoke by ‘addressing’ this requirement to Arnold is the third-personal authority to advise or counsel Arnold about the world. As we saw above, this kind of authority is not linked to accountability, in the sense that is relevant to moral obligation. If, on the other hand, the requirement is a constitutive standard of agency, then the only standing that she can invoke is the first-personal authority to point out that Arnold is contradicting himself, by violating a standard that he is committed to. By violating such a standard, Arnold would be making a mistake akin to a logical blunder or a miscalculation (Darwall 2006: 26–27; 2013c: 63–64). But while Arnold would thereby ‘go wrong’ in some sense, he would not ‘commit a wrong’ in the sense entailed by the violation of an obligation (also see Wallace 2019: 76–77; Zylberman 2017: 924).

In this section, I outlined three features of moral obligations that distinguish them from other normative notions. First, obligations take the form of requirements that function as presumptive constraints, purporting to give us conclusive reasons for or against some action. Second, obligations are agent-relative rather than agent-neutral in that they bind the agent to conduct herself in a certain way. Third, obligations are intimately linked to accountability by expressing standards that, by their very nature, we can demand or expect each other to comply with. As a result, an obligation, if violated without excuse, intrinsically warrants the adoption of certain reactive attitudes towards the agent. To solve the Problem of Obligation, then, one needs to account for these three features. Since this characterization of the Problem of Obligation is shared between the Second-Personal and the Relational Account, the relative dialectical position of the two accounts can be evaluated in terms of their ability to solve this problem.

3. The Second-Personal Account

Darwall’s solution to the Problem of Obligation takes its cue from the link between obligation and accountability that I mentioned at the end of the previous section. Darwall claims that the only way to explain this link is to hold that obligations are grounded in relations of authority and accountability. That is, to return to our example, Arnold’s obligation to move his foot is not just intrinsically linked to his accountability for doing so. Rather, Arnold is obligated to move his foot because he can legitimately be held accountable if he does not do so. More precisely, Darwall argues that the authority that we invoke when we hold Arnold accountable, for example, by blaming him, is an irreducibly second-personal, practical kind of authority (2006: 11–15). This authority is irreducibly second-personal in that the reasons it addresses are not valid independently of this authority. It is practical in that it “creates a distinctive reason for compliance (a second-personal reason)” (2006: 11). Thus, Darwall’s claim that obligations are grounded in relations of authority and accountability amounts to the claim that obligations implicate second-personal reasons.

Darwall’s argument for this claim states that any kind of non-second-personal reason would be the wrong kind of reason to hold Arnold accountable. In other words, the fact that (as we saw above) outcome-based reasons and normative requirements are reasons of the wrong kind to hold Arnold accountable is not an isolated phenomenon. It is a consequence of the very concept of moral obligation. This concept, argues Darwall, forms part of an “interdefinable circle” of “irreducibly second-personal concepts”, which includes the concepts of legitimate demand and second-personal authority (2006: 11–12). Not only do these concepts imply each other, but it is also impossible to establish the applicability of any of these concepts to a particular case without invoking any of the others. In Darwall’s words, “there is no way to break into this circle from outside it” (2006: 12).

Darwall illustrates the notion of an irreducibly second-personal, practical authority with a further example (2006: 12–13). Suppose that a sergeant orders her platoon to fall in. While the platoon may recognise several reasons for falling in that hold independently of the sergeant’s command, it is natural to think that the sergeant’s command generates an additional, distinct type of reason to do so. This reason is grounded in the sergeant’s practical authority over her troops (as opposed to her epistemic authority to point out independently existing reasons). Darwall essentially proposes that we analyse the case of Arnold and Bella—and, in fact, all cases of moral obligation—on the model of the case of the platoon and the sergeant.

Having explained the close link between obligation and accountability, by depicting the former as grounded in the latter, Darwall invokes this link to explain why obligations function as presumptive constraints. Darwall observes that when we blame or resent an agent for acting in a certain way, we presuppose that she had conclusive reason not to act this way. Indeed, he argues that “[j]ust as it is unintelligible to assert something but deny that one believes it, so also does it make no sense to blame someone for doing something and then add that he had, nonetheless, sufficient reason to do it, all things considered” (Darwall 2006: 28). Now, if an agent’s obligation to ϕ is grounded in someone’s authority to hold the agent accountable for failing to ϕ without excuse by adopting reactive attitudes towards her, and such reactive attitudes, on pain of unintelligibility, presuppose that the agent had conclusive reason to ϕ, then it is no wonder that obligations purport to provide conclusive reasons.

The Second-Personal Account also explains why obligations leave less deliberative discretion than reasons in the standard sense. On this account, obligations always implicate legitimate demands. Failing to comply with our obligations is therefore something for which we must account. We might say that, once I am obligated to ϕ, the decision of whether or not to ϕ is no longer simply mine to make. As Darwall puts it, “I no longer simply answer to myself” (2010: 275). Thus, if Arnold has an obligation to remove his foot from on top of Bella’s, he answers to Bella for any failure to comply with this obligation.

Connectedly, the Second-Personal Account also explains the agent-relativity of obligations. When Arnold is obligated to remove his foot from on top of Bella’s, on this account, then he is accountable for his removal of his foot. We might say that the obligation implicitly addresses Arnold, not simply as someone who is in a position to bring about the removal of his foot, but as someone who occupies a particular position in the network of accountability relations (Darwall 2006: 7). Accordingly, he cannot discharge his obligation in any other way than by removing his foot. Similarly, his obligation cannot be discharged by someone else’s removing his foot. In neither case, Arnold is living up to his place in the accountability relation that grounds his obligation. This can be seen from the fact that this accountability relation remains in place—and usually generates new demands, for instance a demand that Arnold explain, apologise or compensate for his conduct—if Arnold does not remove his foot.

In this section, I outlined the Second-Personal Account’s proposed solution to the Problem of Obligation. I return to this account in Sections 7, 8 and 9. In particular, I take on the question of whether its solution is satisfactory by discussing the Problem of Antecedence. In the meantime, I introduce the Relational Account and argue that it fails to provide a solution to the Problem of Obligation.

4. The Relational Account

Relationalists hold that obligations are relational in that they are invariably directed. On the Relational Account, fulfilment of our obligations is owed to a person such that, by failing to fulfil them, we do not merely commit a wrong but wrong that person. The flipside of our obligations are claims that persons hold against us. Obligations are therefore bipolar in that they connect two persons, as opposing poles, in a “normative nexus”: X’s obligation to ϕ is always to some person, Y, who in turn has a claim against X that X ϕ (Wallace 2019: 1; also see Thompson 2004).

While Relationalists thus hold that obligations are based on claims, they do not regard these claims as dependent upon persons’ authority to address these claims. On the contrary, whatever authority we have to address these claims, by adopting reactive attitudes to those who neglect them, derives from the fact that we have these claims.

How does the Relational Account purport to solve the Problem of Obligation? Let us start with the fact that moral obligations present themselves as providing conclusive reasons for action, and thus presume to constrain our deliberation. The Relational Account seems to provide a natural explanation for this feature of obligations. In particular, as Wallace points out, when we owe an obligation to someone, the act required by that obligation appropriately becomes “a presumptively fixed point in [our] ongoing planning about [our] activities” (2019: 50). Thus, for example, if I promise to give you a call tomorrow, and thus acquire a directed obligation that is owed to you, the act of giving you a call enters my deliberation, not as a mere option, but as something that I hold fixed while I plan the rest of tomorrow’s activities around it. In other words, the decision to call you tomorrow presents itself as irrevocable—all relevant considerations have already been taken into account and deliberation has concluded. Consequently, if all obligations are directed to someone, as the Relational Account claims, this seems to explain why they appear in the guise of presumptive constraints or conclusive reasons.

The Relational Account also seems to offer a natural explanation of the agent-relativity of obligations. As Michael Thompson notes, when we owe something to another person, we are bound to this person in a “peculiar nexus” and “the rest of the world is, at least to a certain extent, closed out” (2004: 334). For example, my promise to give you a call tomorrow narrows my deliberative field so that certain considerations are excluded. One consideration that is excluded in this way is whether, by violating my promise, I could bring about a state of affairs in which, overall, more people keep their promises. By acting on this consideration, I would not be acting in the way that the obligation calls for. After all, this obligation constitutively involves your claim against me. Accordingly, I would not do it justice by breaking my promise so that other promisors keep their promises to other promisees. In a similar vein, if a third party somehow tricked me into keeping my promise to you (say, by replacing all the numbers stored in my phone with yours), she would not thereby discharge my obligation. Since this obligation forms part of a normative nexus between you and me, it is not directly concerned with the third party’s actions. Hence, if obligations are fundamentally relational, we should not be surprised by their agent-relativity.

Finally, the Relational Account also seems to offer a straightforward explanation of the intrinsic demandability of obligations. When others owe us a certain kind of treatment, and we hold a corresponding claim against them, a natural response to their failure to give us this treatment is to hold them accountable. For example, if I break my promise to call you, it would be fitting for you to hold me accountable by blaming or resenting me. According to Wallace, this response is intrinsically warranted by my breaking the promise because you naturally take a “special interest” in whether your claims against me are met (2019: 97). Wallace concludes that relational obligations are “suited by their nature to structure relations of accountability” (2019: 67).

Thus, it appears that the Relational Account provides a natural explanation for all three distinctive features of obligation. However, before we can reach a verdict on the Relational Account’s ability to solve the Problem of Obligation, some clarification is in order. The Relational Account, as I just characterised it, is ambiguous. At first glance, the account might be read as purporting to explain the phenomenon of obligation in terms of the phenomenon of relational obligation. Indeed, it seemingly purports to explain the distinctive features of obligation by appealing to paradigm cases of obligations that are owed to someone, such as promissory obligations (e.g., Wallace 2019: 5–9, 105). However, if this was the Relational Account’s strategy, then all this account could show is that some obligations are relational—namely, those that are directed—and that these obligations share the three distinctive features of obligation. It could not show that, by turning to these obligations, we can illuminate the phenomenon of obligation in general.

The problem is not just that it is doubtful whether all obligations are directed.⁴ Nor am I here making the plausible point that, even if all actual obligations were directed, non-directed obligations would remain a conceptual possibility.⁵ Instead, what I want to point out is that, even if we assume that, by conceptual necessity, all obligations are directed, it is not clear that the Relational Account solves the Problem of Obligation. For if the Relational Account is read as proposing to explain obligation in terms of relational obligation, it might be objected that the explanans merely instantiates—and, thus, fails to illuminate—the explanandum. The worry is that relational obligations appear to illuminate the distinctive features of obligation although in fact they merely exemplify these features, precisely because they are obligations.

This worry is particularly troubling in light of the fact that, while many theorists believe in directed obligations, almost none claim that these obligations illuminate the phenomenon of obligation as such. I submit that the reason is not just that most theorists take non-directed obligations to be at least conceptually possible. Instead, unless and until an argument to the contrary is provided, they assume that the phenomenon of directed obligation is explanatorily downstream from, or at best on a par with, the phenomenon of obligation.

In order to be in a position to assess the Relational Account’s ability to solve the Problem of Obligation, then, we should not interpret it as purporting to explain obligation in terms of relational obligation. Instead, we should interpret it as purporting to explain the obligatory character of directed obligations in terms of their relational character. This means that, if the Relational Account is to solve the Problem of Obligation, it must provide a way to disentangle the relational character of directed obligations from their obligatory character, and then explain the latter in terms of the former. In the next section, I argue that the Relational Account fails to do this.

5. Claims and Practical Relations

How might we spell out the relational character of directed obligations without falling back on their obligatory character? Zylberman’s statement of the Relational Account suggests a promising proposal: obligations are relational in that they are not “intelligible independently of our practical relations to others” (2017: 926). Thus, my promissory obligation to call you tomorrow, by virtue of being directed to you, is relational in that it is not intelligible independently of my practical relation to you. Since Wallace does not appeal to a link between obligations and practical relations, Zylberman will be my main interlocutor in this section. I discuss Wallace’s views in some more detail in the next section.

The supposedly practical nature of the interpersonal relations that Zylberman takes to be linked to obligations seems crucial here. What does it amount to? Zylberman does not provide a definition, but the following passage illustrates the idea (2017: 939; also see Thompson 2004: 340):

My right to not be tortured is not a reminder, a moral sign-post, as it were, of your non-relational duties. I am not the occasion of your moral fall. I am your victim. My standing is irreducibly practical because it is my practical standing before you as an equal bearer of rights.

The idea seems to be that the relevant relations are practical in that we cannot understand our obligations without thereby relating to some concrete other as a ‘you’. This is the sense in which, when we violate an obligation, there is someone who stands before us as the victim of our action. We might say that the relevant relations are practical in that they are not mere relations, but relatings (also see Darwall 2006: 44). They are unlike the type of relation of which, say, being smarter than is a token. While this type of relation relates objects to one another, there is no non-trivial sense in which these objects themselves relate to one another. That is, the role that the relata play is a passive rather than an active one. Accordingly, being smarter than is what we might call a theoretical rather than a practical relation.

Can we say more about the difference between theoretical and practical relations? What is the non-trivial sense in which the relata of practical relations themselves actively relate to one another? The first thing to note is that the relata of a practical relation are at least disposed to have attitudes about one another. That is, they are prone to be the objects of each other’s intentional states. But that is not all. The relata of practical relations are disposed to have attitudes about one another that exhibit a special kind of intentionality. In Adrian Haddock’s words, they exhibit “a neglected manner or mode of de re intentionality: a manner of relational directedness that is not merely directedness at a particular object, but directedness to (a fellow) subject” (2014: 345).

Let’s unpack this. First, to say that the relevant mode of intentionality is de re is to say that its content is not completely specified by a proposition. Instead, the object of the mental state is singled out partly by something more akin to a demonstrative element: ‘I am obligated to this person.’ Second, the manner of de re intentionality involved in practical relations is distinctive in that it singles out its object, not as a mere object, but as (another) subject (see Haddock 2014: 346–47). If two persons, X and Y, relate to one another, they represent each other, not merely as objects or features of the external world, but as fellow beings with intentional states. More specifically, they represent each other in the mode in which they each represent themselves. Now, Y’s thought of herself is first-person thought: Y thinks of herself as an ‘I’. In other words, its intentionality is not only de re but de se. But if X’s thought of Y must represent Y in the same way as Y’s thought of herself, it follows that X’s thought of Y must exhibit de se intentionality as well. In other words, X must think of Y as an ‘I’ or, more precisely, as another ‘I’. This means that X must think of Y as a ‘You’: ‘I owe this obligation to you.’ Thus, X’s thought must be a distinctive kind of first-person thought which we might call second-person thought. To sum up, the relata of practical relations are disposed to represent themselves and each other as fellow subjects. In Thompson’s words, each of them is disposed to view “herself as related to others, and as other to others” (2004: 337, italics altered). Or, as Darwall puts it, they occupy “the intersubjective stance of someone relating to others as, and reciprocally recognizing his status as, one among others” (2006: 102, italics altered).

On the present proposal, then, the Relational Account purports to solve the Problem of Obligation by arguing that X’s obligation to ϕ, by virtue of being directed to a concrete other, Y, who in turn has a claim against X that X ϕ, implicates a practical relation between X and Y. That is, X cannot understand her obligation to ϕ without representing Y as a fellow subject, a ‘You’. This seems promising. If directed obligations implicate practical relations in this way, we have an explanation why they exhibit the three features of obligation that give rise to the Problem of Obligation. First, if obligations direct our attention to a ‘You’, and thus trigger a distinct mode of intentionality, then perhaps it is not surprising that they present themselves as presumptively fixed points in our deliberation.⁶ In particular, since this mode of intentionality is distinct from the modes by which we relate to mere objects or features of the world, it is not surprising that obligations are presumptive constraints which cannot be neglected in the way in which reasons based on the value of objects or states of affairs can. Second, the link between obligations and the ‘I’-‘You’ mode of intentionality also seems to explain their agent-relativity, by explaining why “the rest of the world is [. . .] closed out” when we face an obligation (Thompson 2004: 334). Third, the notion that obligations relate us to one another as fellow subjects promises to explain their intrinsic link to accountability practices: they naturally facilitate the person-to-person form of address implicit in the adoption of reactive attitudes.

Thus, the notion of a practical relation promises to facilitate the explication of the relationality of directed obligations that the Relational Account needs in order to explain the obligatory character of such obligations (also see Eilan 2014; Haase 2014a; 2014b; Lavin 2014; Thompson 2004). However, I now want argue, the Relational Account cannot accommodate the notion of a practical relation. The problem arises from the way the Relational Account characterises the claims on which obligations are based.

To see this, consider how the Relational Account might try to sustain the notion that obligations are unintelligible independently of our practical relations to others. Undoubtedly, the Relational Account will appeal to its assertion that all obligations are directed, that is, they are to someone, and thus constitutively connected to someone’s claim. But how does the link between obligations and claims support the link between obligations and practical relations? In other words, what is it about claims that makes it impossible to understand them (and, thus, their corresponding obligations) without representing their holders as fellow subjects? The challenge is that, if claims are to facilitate practical relations in the relevant way, they must not be mere objects or properties. For otherwise understanding them will not require the distinctive, second-person de se mode of intentionality which is central to practical relations.

I suspect that, for many of us, it is intuitively plausible that there is an intimate link between claims and practical relations. I further suspect that the reason for this intuitive plausibility is that many of us think of claims along the lines of Joel Feinberg’s account (1970). Feinberg links directed obligations to a sense of ‘claim’ that derives its significance from the activity of claiming. He argues that the “characteristic use” of directed obligations “is to be claimed, demanded, affirmed, insisted upon” (1970: 252). According to Feinberg, then, the claims that correspond to these obligations are not to be thought of as “things” (1970: 253). Instead, the “verbal” form of ‘claim’ has “primacy” over the “nominative” from: “having a claim consists in being in a position to claim” (1970: 253, italics altered). What is more, Feinberg argues that to claim what one is owed is “to ‘stand up like men,’ to look others in the eye, and to feel in some fundamental way the equal of anyone” (1970: 252). Less metaphorically, when we claim what we are owed, we relate to others, representing them as fellow subjects.

If Feinberg’s account of claims is correct, it is indeed impossible to understand the obligations that correspond to claims without representing others as a ‘You’. On this account, to conceive of someone as having a claim is to conceive of them as being in a position to relate to us as fellow subjects. And by conceiving of them in this way we are, in turn, relating to them as fellow subjects. However, Feinberg’s account is not available to Relationalists. As we saw in the previous section, the Relational Account departs from the Second-Personal Account by insisting that whatever standing we have to address claims derives from the prior and independent existence of these claims. Thus, the Relational Account regards being able to claim as separate and explanatorily downstream from having a claim. This is incompatible with Feinberg’s view that being in a position to claim is what having a claim consists in.

Perhaps it is not necessary to appeal to the activity of claiming in order to establish the sought-after connection between claims and practical relations. Perhaps there is another mode of interaction that could connect persons to each other so that they relate to each other in the distinctive second-person mode of intentionality. However, the way the Relational Account depicts claims seems to rule out any such appeal to a mode of interaction. The reason is that the Relational Account depicts claims as part of a normative reality which is independent of, and prior to, any mode of interaction. This picture of claims not only prevents the Relational Account from appealing to some mode of interaction to establish a link between claims and practical relations, but also pushes the Relational Account to a conception of the moral relations between persons as decidedly theoretical rather than practical relations.

To see this, consider Zylberman’s statement that the Relational Account “represents the moral ought as a property internal to actions” (2021: 410–11):

Moral necessity attaches to certain actions as such, then, because original claims demarcate a domain of actions as required, prohibited, or permitted. This necessity is [. . .] generated by the fact that moral agents act and interact in ways that are both consistent and inconsistent with reciprocal relations of respect and recognition.

The notion that actions are right or wrong depending on whether they are consistent or inconsistent with the moral relations between persons strongly suggests that these relations form part of a normative reality that our actions purport to reflect or express. In a similar vein, Wallace states that our obligations “derive from the fact that we inhabit an extensive notional community together with other beings who are ‘equally real’ (in Thomas Nagel’s striking formulation), and whose interests are no less significant than ours” (2019: 37, citing Nagel 1978: 14). Thus, according to Wallace, by complying with our obligations “we are acknowledging our equal standing within such a moral community” (2019: 38, italics added).

The Relational Account thus seems committed to a conception of moral action that we may call the Expressive Conception.⁷ Tamar Schapiro sums up this conception as follows: “[m]orality, [. . .] is a concern for the expressive aspect of action, its capacity to make claims about the value of others (and perhaps also of ourselves)” (2001: 96). In other words, an action is morally right or wrong by virtue of what it says about the value or other “objective properties” of persons and things (2001: 96).

There is a sense in which these objective properties relate us to persons and things in a certain way. As Schapiro puts it, action “involves a reflective awareness of ourselves in relation to others” in that it “expresses a conception of ‘where we stand’ in relation to the other constituents of the world, conceived as a realm of status relations” (2001: 97). Nevertheless, since moral action merely aims to reflect or conform to this realm of status relations, and these relations arise from the properties of persons and things, it does not relate us to others practically, that is, as one subject among others. Instead, it “relates us as observers to the world” (2001: 97).

One might object that there is a sense in which the relevant status relations are practical. After all, they are relations of normative status, and thus concerned with what is to be done. But here the label ‘practical’ merely describes the content of the relation, not its nature as a relation.⁸ Like the relation of being smarter than, the relevant relations are theoretical rather than practical relations. The mode of intentionality in which agents represent these relations and their place within them is not distinct from that in which they represent the rest of the world. Indeed, Zylberman himself suggests that a claim is to be understood as a “relational property” on the model of such properties as being the teacher of (2021: 408). The latter strikes me as a paradigm example of what I have been calling a theoretical relation.

On the Expressive Conception, persons, in their capacity as fellow subjects, do not play any essential role in the recognition of our obligations. Therefore, contrary to what Zylberman says, the notion that violating an obligation relates us to another person as the “victim” of our action, as opposed to the mere “occasion of [our] moral fall”, is not congenial to the Relational Account (Zylberman 2017: 939). Instead, the Relational Account seems to fit much more closely with the view expressed in the following passage from Frances Kamm (1992: 385):

[We should] see the constraint as a way of expressing the view that persons’ characteristics make them each inviolable in a certain way. An agent’s own victim is special only in that he is the one whose constraining right the agent comes up against. The agent’s own act is special only in that it makes him come up against the constraining right.

In this view, the victim of a wrong, considered as a fellow subject, is backgrounded in moral deliberation. What is foregrounded are the claims that constitute the standing of the victim. Kamm notes that if the agent addressed the potential victim of her action, the truthful thing to say would be: “it’s nothing personal, there’s nothing between us, but you’re really something” (1992: 386).

Now, one might argue that the Relational Account goes beyond the Expressive Conception endorsed by Kamm in insisting that the standing of persons is inseparably tied to directed norms (see Wallace 2019: 178; Zylberman 2021: 412). Indeed, Zylberman writes (2017: 938):

[W]hat makes my standing practical is not that I actually make a claim against you, but rather that I have a standing in relation to you, the standing of an equal. And that this standing [. . .] is constituted by claims against others and by duties others owe one.

Here, Zylberman suggests that what connects claims to practical relations is that they are constitutive of an original relational standing of persons (also see Zylberman 2021: 409). More specifically, he follows Thompson in holding that the standing of being a person is fundamentally relational (Thompson 2004: 353, cited in Zylberman 2017: 939):

The judgement X is a person [. . .] is essentially a ‘de-relativization’ of the prior bipolar judgement X is a person in relation to Y. Similarly, X is a sister is a de-relativization of X is a sister of Y. ‘Recognizing someone as a person’ is registering her as a person in relation to yourself; it is the appropriation of such a proposition in the first person.

Zylberman’s suggestion is that claims are linked to practical relations, not by virtue of their relation to the activity of claiming, but by virtue of being constitutive of personhood itself, which cannot be understood independently of practical relations to others.

Whether this suggestion is helpful depends on what the Relational Account can tell us about the link between personhood and practical relations. More specifically, the Relational Account needs to link personhood to practical relations without relying on either the notion of obligation (which it is trying to explain) or some mode of interaction (whose explanatory significance it denies). Here, it is important to note that many accounts of the moral standing of personhood do not characterise it as fundamentally relational. For example, many accounts identify the standing of personhood with some intrinsic value that all persons instantiate, such as rational agency or freedom. As Zylberman notes, nothing in these accounts indicates a connection between personhood and practical relations (e.g., 2017; 2021; also see Lavin 2014: 283). One can understand and be guided by the intrinsic value within persons without ever relating to them as persons. In this respect, the value of persons is no different than that of, say, a work of art.

So, how does the Relational Account connect personhood to practical relations? Zylberman seems to think that what links personhood to practical relations is precisely the fact that this standing is constituted by claims which are linked to practical relations (2017: 940; 2021: 411–12). However, the link between personhood and practical relations was supposed to explain the link between claims and practical relations. So, the latter cannot in turn be invoked to explain the link between personhood and practical relations.

The more fundamental problem with Zylberman’s suggestion is that, since on the Expressive Conception to recognise claims is to recognise a normative reality which is independent of any mode of interaction between persons, the Relational Account seems committed to the possibility of representing these claims in the same mode of intentionality in which we represent the rest of the world.⁹ Maintaining that these claims constitute a fundamentally directed standing of persons is not helpful because it seems to simply expand this (non-second-personally representable) normative reality to include directed norms, and thus to sever the otherwise plausible link between such norms and practical relations.

I therefore contend that the Relational Account cannot solve the Problem of Obligation by appealing to the idea that claims, and thus obligations, cannot be understood independently of practical relations to others. The problem is not only that the Relational Account cannot appeal to the activity of claiming, or some other mode of interaction, to establish the sought-after link between claims and practical relations, but also that the Relational Account adopts a conception of the claims that make up the standing of persons according to which they are decidedly not linked to practical relations in the relevant way. In the next section, I argue that the Relational Account also does not provide an illumination of the relationality of directed obligations that might account for their obligatoriness that does not proceed in in terms of practical relations.

6. Beyond Practical Relations

According to the Relational Account, obligations are relational in that they are constitutively linked to claims against others. Yet, I argued, the Relational Account adopts a conception of moral action, the Expressive Conception, according to which there is no essential link between recognising these claims and relating to others practically, that is, as a ‘You’.

One might respond that this does not warrant the conclusion that the Relational Account cannot solve the Problem of Obligation. After all, the Relational Account might spell out the relationality of directed obligations without invoking the notion of practical relations at all. Indeed, Wallace seems to reject the invocation of practical relations (2019: 120). However, as I argue in this section, his remarks on relationality do not provide a solution to the Problem of Obligation either.

It is not obvious how exactly Wallace proposes to illuminate relationality, and thus solve the Problem of Obligation. Two different proposals might be reconstructed from his remarks. The first proposal denies the need for any illumination of relationality beyond that which is provided by the Expressive Conception of moral action. The second proposal is that we can learn something about relationality by seeing how claims, and thus obligations, are justified by appeal to individual interests. I discuss these proposals in turn.

Wallace characterises directed obligations as “self-standing” in that they are “not grounded in any antecedent relation in which agents stand to those who have claims against them” (2019: 146). Against approaches that link directed obligations to persons’ disposition to form a certain kind of intentional attitude, such as the attitudes involved in practical relations, he argues that beings who lack the capacity to form such attitudes can still be owed a certain treatment (2019: 120–21). Wallace might thus be read as denying the need for an account of relationality beyond the Expressive Conception of moral action. On this reading, the relationality of directed obligations is exhausted by theoretical relations between persons.

However, on the Expressive Conception, the claims that correspond to moral obligations are features, or ‘relational properties’, that persons just have. Consequently, the Expressive Conception does not explain the features that distinguish moral obligations from other normative phenomena. First, if fulfilling moral obligations is a matter of reflecting normative reality, it is not clear why obligations function as presumptive constraints. For, arguably, acting on outcome-based reasons is a matter of reflecting an aspect of normative reality as well, namely the value of states of affairs. Second, it is not clear why, on the Expressive Conception, obligations are agent-relative rather than agent-neutral. One might argue that the reason is that obligations reflect our place within normative reality. However, we might say that our outcome-based reasons reflect our place within normative reality as well, namely our ability to manipulate outcomes. Finally, the Expressive Conception cannot explain why the unexcused violation of an obligation intrinsically warrants the adoption of reactive attitudes such as blame or resentment towards the agent. While failure accurately to reflect normative reality might be an instance of ‘going wrong’, it does not as such amount to ‘doing wrong’ or ‘committing a wrong’. Thus, while others may ‘address’ the agent in the sense of pointing him towards the true nature of normative reality, they cannot hold him accountable for failing to get it right. Consequently, the only authority that they might invoke is the epistemic, third-personal authority of a wise councillor or advisor.

It does not help to insist that the Relational Account departs from other versions of the Expressive Conception by depicting moral obligations as directed. After all, as I argued in Section 4, in order to solve the Problem of Obligation, the Relational Account must disentangle the relationality of directed obligations from their obligatoriness in such a way that the former explains the latter. The Expressive Conception, even if supplemented with directed norms, does not offer any way of doing so. It merely asserts that the parts of the normative reality that have the distinctive features of obligatoriness also have the feature of directedness. Therefore, it cannot solve the Problem of Obligation.

The closest Wallace comes to providing something resembling a positive account of relationality is when he discusses how persons’ claims might be justified by their interests.¹⁰ Wallace emphatically warns that claims cannot be reduced to interests (2019: 147). And he seems right to do so; for if claims were reduced to interests, they would seem to correspond to outcome-based reasons rather than to moral obligations (see Section 2). Nevertheless, Wallace argues, “there is [. . .] reason to think that the interests of individuals are relevant to questions about their claims” (2019: 147). In particular, “a duty is directed to another party only if the considerations that go into establishing the duty center around that party, and it is personal interests of the putative claimholder that will be prominent within such a person-involving justification” (2019: 163).

However, I now argue, the way in which Wallace takes claims to be justified by interests does not illuminate the relationality of directed obligations. The problem is that, since claims are not to be reduced to interests, the procedure of moral reasoning that takes us from interests to claims must involve an independent relational element. Consequently, the relational nature of claims cannot be illuminated by their connection to interests. To see this, consider the two models of how individual interests justify claims that Wallace suggests.

The first model is intuitionist. On this model, “there is a route that will get us from the personal interests of people potentially affected by our actions to the assignment of moral claims”, but “there is nothing illuminating to be said, in general terms, about how moral justification of this kind will proceed in all cases to which it might pertain” (Wallace 2019: 178). Instead, claims are established by an exercise of moral judgment, presumably of a quasi-perceptual nature. Wallace states that what distinguishes this model “from other forms of intuitionism in ethics is that the conclusions that are grasped through unsystematizable intuition are [. . .] specifically about what we owe it to each other morally to do” (2019: 178, italics added).

The problem with this model, when it is invoked to solve the Problem of Obligations, is that it effectively denies the very possibility of disentangling relationality from obligatoriness in such a way that the former could explain the latter. Since claims are not reduced to interests (and, in any case, not all interests give rise to claims), the relationality of claims cannot be due to their basis in interests. Instead, the intuitive judgment that takes us from interests to claims must employ an independent relational element. But since this judgment is unsystematisable, its relational element cannot be spelled out in a way that would solve the Problem of Obligation.

Wallace’s second model of how claims might be justified by individual interests takes its cue from T.M. Scanlon’s contractualism (Wallace 2019: 179–89). Contractualism states, in short, that an act is wrong if it is “disallowed by any principles that no one could reasonably reject” (Scanlon 1998: 10). In order to determine which principles can reasonably be rejected, for each principle, we need to identify the person whose interests would be most negatively affected by its adoption, and then compare the strength of the resulting objections. A principle disallowing an act cannot reasonably be rejected if the strongest objection to it is weaker than the strongest objections to any principle that allows the act.

Wallace suggests that if a person’s interests make it reasonable for her to reject any principle that would permit us to ϕ, then we plausibly owe it to that person not to ϕ. He concludes that contractualism can be understood as a “further theoretical specification” of relational morality (2019: 185):

To be something that we owe it to others to do, just insofar as they are persons who have corresponding claims against us, is on [Scanlon’s] account to be required by principles for the general regulation of behaviour that nobody could reasonably reject.

Unlike the intuitionist model, then, the contractualist model systematises the moral reasoning that takes us from claims to interests. Consequently, one might hope that it illuminates the relational element in moral reasoning that is required, in addition to interests, to generate relational outputs.

The problem with this suggestion is that, if the outputs of contractualist reasoning are to be irreducibly relational, the interests that furnish its inputs must be transformed into something relational before they enter it. In particular, individuals’ objections to principles that set back their interests must be understood as putative claims or, as Wallace puts it, as expressing “a specific kind of grievance or complaint” (2019: 181). If they are not understood in this way, it is unclear how contractualist reasoning could yield irreducibly relational claims; for, then, this kind of reasoning would simply process interests in a certain way.¹¹ Thus, Wallace states that the “[p]ersonal interests of affected parties enter such deliberation [. . .] as potential grounds for moral claims” (2019: 180, italics added). Similarly, he writes that “the elements in the relational conception of moral requirements [. . .] seem to fall out of the contractualist way of thinking about their derivation” (2019: 181, italics added).

In other words, claims result from contractualist reasoning only if it is interpreted as reasoning about putative claims. Wallace acknowledges this when he criticises Scanlon for not consistently interpreting contractualist reasoning in this way (2019: 184–89; see Scanlon 2013: 403).¹² But then contractualism, by itself, offers no more illumination of relationality than intuitionism. Each model yields claims only if it is injected with an independent relational element, which it therefore cannot explain. Indeed, this is how it must be if the relationality of claims is not to be reduced to non-relational aspects of these models of reasoning, such as the interests that furnish their inputs.¹³

Let us take stock. In Section 4, I argued that the Relational Account should not be read as explaining obligation in terms of relational obligation. Instead, the Relational Account needs to provide an account of the relationality of these obligations which explains, and hence does not rely on, the obligatoriness of these obligations. In Section 5, I argued that the Relational Account cannot accommodate the promising suggestion that obligations are relational in that they cannot be understood independently of our practical relations to others. In this section, I argued that the Relational Account does not provide a solution to the Problem of Obligation that proceeds independently of the notion of practical relations either.

7. The Problem of Antecedence

If my argument so far is successful, I have already shown that, for theorists who are troubled by the Problem of Obligation, the Relational Account does not constitute a viable alternative to the Second-Personal Account. Even if the latter falls prey to the Problem of Antecedence by virtue of the way in which it solves the Problem of Obligation, it has the decisive advantage that at least it provides a solution to the Problem of Obligation. In this section, I argue that, in addition, the Second-Personal Account does not fall prey to the Problem of Antecedence.

Let me start by explaining the Problem of Antecedence. As we saw in Section 3, the Second-Personal Account purports to solve the Problem of Obligation by depicting obligations as grounded in relations of authority and accountability. In this picture, Arnold’s obligation to remove his foot from on top of Bella’s is not independent from Bella’s authority to demand that he do so. To the contrary, Bella’s authority to demand that Arnold remove his foot “addresses a reason that would not exist but for her authority to address it” (Darwall 2006: 13). This might seem to support a voluntarist reading of the Second-Personal Account, according to which moral obligations are contingent upon the will of actual persons.

It is this voluntarist reading of the Second-Personal Account that prompts Relationalists to object that this account faces the Problem of Antecedence (Wallace 2007: 27; 2019: 34 and 96–97; Zylberman 2017: 926). Zylberman writes (2017: 926):

[T]he voluntarist model of norms is particularly vulnerable to the Problem of Antecedence. This is because at least some moral obligations appear to be independent of the actual claims (implicit or explicit) that members of a moral community make on one another. [. . .] Surely, your obligations not to intentionally harm others, steal from them or lie to them obtain independently from and prior to any actual demands of others. The voluntarist model gets the order of explanation backwards.

The objection states that, on the Second-Personal Account, the existence of a moral obligation to ϕ depends on someone’s actually demanding—explicitly or implicitly—that we ϕ. Intuitively, however, it seems that many moral obligations—say, the obligation not to enslave—are in place regardless of whether anyone demands that we comply with them (also see Wallace 2007: 26–27). Indeed, it seems that the prior existence of the obligation not to enslave is what warrants the demand not to enslave in the first place.¹⁴ The objection can be understood as the claim that, while the Second-Personal Account might provide a plausible analysis of the case of the sergeant ordering her platoon, this analysis cannot be extended to moral obligations in general.

In reply, I want to argue that the Second-Personal Account should not be read as a voluntarist approach in the sense suggested here. The Second-Personal Account depicts obligations as grounded, not in the actual demands that persons make of one another, but in their authority to make demands or, equivalently, in their legitimate demands. And the authority to make demands is not entirely discretionary. In other words, a person cannot legitimately demand whatever comes to her mind.

But what determines whether something can legitimately be demanded? Here, it is important that Darwall’s notion of an irreducibly second-personal, practical kind of authority is linked to a Kantian element in his Second-Personal Account (2006: chs. 9–12). Darwall argues that “second-personal address”, the act of addressing second-personal reasons by invoking this kind of authority, is subject to “normative felicity conditions” (2006: 5). These are conditions that must be in place if an instance of second-personal address is to succeed at giving the addressee a second-personal reason. In other words, they are conditions of the possibility of second-personal address (Darwall 2006: 242). Accordingly, any demand, if it is to be intelligible as an instance of second-personal address, must implicitly presuppose that these conditions are in place.

What do the normative felicity conditions of second-personal address consist in? Darwall argues that when we address an agent second-personally, demanding that she ϕ, we must implicitly presuppose that she can internalise this demand and address it to herself. Thus, when Bella holds Arnold accountable for stepping on her foot, she implicitly presupposes that Arnold can hold himself accountable for doing so. To see this, it is important to distinguish holding an agent accountable from merely “goading” her (Darwall 2006: 49–52). When we hold someone accountable, we do not simply try to make her act a certain way. Instead, we try to get her to acknowledge, freely and rationally, the legitimacy of our demand. Now, the authority upon which Bella’s demand is based is irreducibly second-personal, which means that it bottoms out in the authority “to demand respect for this very authority” (Darwall 2006: 14). It follows that Bella’s attempt to hold Arnold accountable implicitly aims at Arnold’s removing his foot from respect for Bella’s authority alone. Accordingly, Bella’s act of address presupposes that Arnold can hold himself accountable for complying with it, by addressing himself second-personally. In Darwall’s terminology, Bella’s “second-personal authority” implies Arnold’s “second-personal competence” (2006: 269–76).

Thus, Bella’s demand on Arnold implicitly presupposes that Bella is in turn accountable to Arnold. In particular, Arnold has the authority to demand that Bella’s demand be justifiable to him. Hence, at the most fundamental level, the relation of authority and accountability between Arnold and Bella is symmetrical. Putting this point in Darwall’s terminology, Arnold’s “second-personal competence” (which follows from his accountability to Bella) implies Arnold’s “second-personal authority” (which in turn implies Bella’s accountability to him). Hence, if Bella’s demand on Arnold is to be intelligible, then even prior to addressing it Bella must relate to Arnold in a certain way—in a justifying, as opposed to a goading manner (Darwall 2006: 272).

Thus, by virtue of viewing ourselves and others as participants in second-personal address, we implicitly regard all persons as equal members of a moral community who are mutually accountable for respecting each other’s second-personal standing (Darwall 2006: 269–76). That is, we regard all persons as obligated to act in ways that they can justify to one another from within the “second-person standpoint”, which is shared between all persons and represents them symmetrically in their capacity as participants in second-personal address, and thus as equal members in a moral community (Darwall 2006: 35). In this way, according to Darwall, the necessary presuppositions of second-personal address vindicate a contractualist moral requirement, according to which “an act is wrong if the act would be disallowed by a principle no one could reasonably reject our holding one another accountable for complying with” (2006: 301).

The important upshot is that, according to the Second-Personal Account, a demand that is not compatible with the basic contractualist principle “misfires” (Darwall 2006: 54). That is, the demand fails to address a genuine second-personal reason, and thus a moral obligation. Thus, even the sergeant’s authority over her platoon, insofar as it is a genuinely second-personal, legitimate type of authority, has its limits. She cannot generate second-personal reasons for her platoon to do whatever she wants them to do. Even her asymmetric and discretionary authority is dependent on an institutional framework that must, fundamentally, be justified to all persons as equal members of the moral community.

Conversely, there are certain demands that are “in force”, independently of whether anyone actually addresses them, “if no one could reasonably reject principles that would warrant them” (Darwall 2007: 65). In short, it is not true that, on the Second-Personal Account, an agent’s obligation to ϕ depends on someone’s actually demanding that she ϕ. Instead, the obligation might be grounded in a demand that is in force, regardless of whether anyone actually addresses it, because no one could reasonably reject all principles that would warrant such a demand.¹⁵ Arguably, for example, the obligation not to enslave others is grounded in such a demand.

8. The Problem of Obligation Revisited

I have argued that the Second-Personal Account does not fall prey to the Problem of Antecedence because it does not subscribe to a voluntarist conception according to which obligations are grounded in the actual demands of persons. Instead, the Second-Personal Account depicts obligations as grounded in the authority to demand, which is determined by the normative felicity conditions of second-personal address and not always discretionary.

However, this gives rise to another objection. In particular, it may seem that the way in which the Second-Personal Account avoids the Problem of Antecedence undermines this account’s distinctiveness. What is more, it may seem to undermine this account’s ability to solve the Problem of Obligation. Recall that the Second-Personal Account’s chief contribution to the philosophical debate about moral obligations is the notion of a second-personal reason. Second-personal reasons are distinctive in that they are grounded in the authority to demand compliance with these reasons. Connectedly, this authority is irreducibly second-personal and practical, that is, it is the authority to address reasons that would not exist but for this authority. Now, the objection has it that, if what we can demand of one another is determined by the normative felicity conditions of second-personal address, and thus by the contractualist moral principle, then second-personal reasons are not grounded in the authority to demand after all; instead, they are grounded in this, seemingly more fundamental, moral principle (see Fan 2014: 79; Heuer n.d.; Lavin 2014: 284–87; Zylberman 2017: 928–33). Accordingly, the authority to demand is not irreducibly practical and second-personal; rather, it is the first- or third-personal authority to point out the verdicts of a prior and independent moral principle. And, as I noted repeatedly, the latter kind of authority cannot solve the Problem of Obligation.

The Second-Personal Account may seem to face a dilemma, then. It can give up the claim that obligations are grounded in an irreducibly second-personal, practical authority, but then it will not be able to solve the Problem of Obligation. Alternatively, it can endorse a voluntarist conception according to which obligations are grounded in the actual demands of concrete individuals, but then it will fall prey to the Problem of Antecedence.

9. The Practice of Second-Personal Address

I argue that the Second-Personal Account does not face the above-mentioned dilemma between solving the Problem of Obligation and avoiding the Problem of Antecedence. In particular, I argue that the Second-Personal Account’s invocation of the normative felicity conditions of second-personal address does not undermine its claim that obligations are grounded in an irreducibly second-personal, practical authority. My argument rests on the notion that demands, as characterised by the Second-Personal Account, should be understood as moves in a practice.

As John Rawls notes, the rules of practices are “logically prior” to the actions to which they apply (1955: 25). That is, the rules of a practice do not merely regulate but also constitute the actions that fall under them, by enabling us to describe them as actions of the relevant types in the first place. As Rawls puts it, “[w]e may think of the rules of a practice as defining offices, moves, and offenses” (1955: 25). To see this, consider the game of chess. The rules of chess do not merely determine when, how, and by whom the various moves of the game may be performed. Instead, they also determine when, how, and by whom these moves can be performed. In other words, they determine under what circumstances an agent can be described as performing the move of, say, castling.

It follows from Rawls’s observation that outside the context created by the rules of a practice, it is impossible to perform the moves of that practice. For example, if the rules of chess did not exist, then none of our motions would be describable as instances of the move of castling. In fact, the pieces and board with which we might perform the relevant motions would not even be describable as chess pieces and a chess board, respectively.

Connectedly, in order to count as performing the moves of a practice, one needs to follow the rules of that practice at least to some extent. We can distinguish at least two different ways in which one might fail to follow the rules of a practice. The first way is by violating the rules. Since the rules do not merely regulate but also constitute the moves of the practice, one will fail to perform any of these moves if one violates the rules.

The second way of failing to follow the rules of a practice is by conforming to the rules, but not because that’s what the rules require. In this case, one ‘goes through the motions’ for some practice-independent reason. For example, my external acts may conform to the rules of chess because I want to appear to be playing chess (suppose I am a spy who is trying to eavesdrop on a conversation at the neighbouring table). Schapiro plausibly argues that, in this kind of case, I cannot be regarded as playing chess, and thus as making chess moves, in the fullest sense (2001: 102). The idea is that making a move in a practice, in the fullest sense, requires that one sees oneself as making this move. In other words, it requires that one intends one’s act under a certain description. In Schapiro’s words, “making a move in a practice is a way of participating in it, a way of engaging in it as a form of activity” (2001: 102, italics added). In the case of chess, it requires that one acts as a chess player, rather than just like one. But to self-consciously perform a move in a practice, and thus to intend one’s act under a description that is made available by the rules of that practice, is to act from those rules. Thus, Schapiro concludes, “to fully participate in a practice is to identify oneself and others in terms of the concepts defined by the practice rules, and to regulate oneself in terms of its principles” (2001: 102, italics added).

As I said above, I want to suggest that the Second-Personal Account characterises demands as moves in a practice. In a way that mirrors the relationship between the rules of chess and the move of castling, the normative felicity conditions of second-personal address do not merely regulate but constitute demands. However, this does not mean that second-personal address is a historical practice, that is, a practice whose rules are contingent on the needs and conventions of particular communities at particular times. Instead, it is what we might call a transcendental practice, that is, a practice whose rules could not be otherwise because they are the necessary conditions of its possibility.

This suggestion enables us to maintain both that second-personal reasons are grounded in an irreducibly second-personal, practical authority and that some demands fail to generate obligations while others are in force independently of whether they are actually addressed. The reason is that the normative felicity conditions of second-personal address, understood as practice rules, are internal to the activity of making demands. As we saw above, to make a move in a practice is to make a move that follows the rules of the practice. Under the present suggestion, then, to make a demand is to make a demand that follows the normative felicity conditions of second-personal address. These conditions do not regulate an independently existing activity of demanding, by determining when, how, and by whom demands may be made. Instead, they constitute the activity of demanding by determining when, how, and by whom demands can be made—and thus what demands are.

Thus, the fact that the normative felicity conditions of second-personal address regulate the authority to make demands does not mean that second-personal reasons are grounded in something other than this authority. For the normative felicity conditions of second-personal address are not external to the authority to demand. One cannot use this authority in a way that violates these conditions because they are the conditions of the possibility of using it. Accordingly, when we address a second-personal reason by making a demand we do not point to anything beyond the authority to do so. We address a reason that would not exist but for the authority to address it. This authority is therefore irreducibly second-personal and practical.

In a sense, my suggestion implies that we can distinguish two senses of ‘demand’. There is a pre-theoretical, wide sense of ‘demand’ which is independent of the normative felicity conditions of second-personal address. But the Second-Personal Account invokes a second, irreducibly second-personal sense of ‘demand’ which is spelled out by the normative felicity conditions of second-personal address. Acts of address that are demands in the former but not in the latter sense do not count as uses of second-personal authority. They either misfire or they do not even purport to address a second-personal reason in the first place.

One might object that, while my suggestion accommodates demands that fail to address genuine second-personal reasons because they violate the normative felicity conditions of second-personal address, it does not meet the challenge posed by demands that are in force irrespectively of whether anyone actually addresses them, such as the demand not to enslave others. It might appear that these demands indicate that obligations must be grounded in something other than the authority to demand.

Here, the suggestion that demands are moves in a practice once again proves helpful. As noted above, Rawls observes that the rules of practices constitute not only moves but also offices (1955: 25). As noted above, one cannot fully be regarded as playing chess unless one acts from the rules of chess. This captures a condition for acting as a holder of the office of player as much as it captures a condition for participating in the game. I want to argue that, analogously, the normative felicity conditions of second-personal address are conditions for acting as a holder of the office of person as much as they are conditions for making a demand. Indeed, Darwall regards person as a second-personal concept, and thus as part of the same circle of interdefinable concepts as legitimate demand (2006: 80). In this view, person simply is the concept of a participant in second-personal address, that is, someone who is capable of addressing and acknowledging demands.

Now, the normative felicity conditions of second-personal address are such that some demands cannot be made while others are ‘in force’ independently of whether anyone addresses them. I argue that these are demands that holders of the office of person, as such, cannot and necessarily do make, respectively. Thus, there are demands that we cannot fail to make, just insofar as we are persons. Indeed, Darwall says that demands that are ‘in force’ by virtue of the normative felicity conditions of second-personal address themselves are ones that we “must presuppose [are] made of everyone by the moral community or representative persons as such” (2013b: 37). Crucially, this means that obligations which do not depend on whether anyone actually addresses them are nevertheless grounded in demands, particularly in demands that we count as making just by virtue of participating in the practice of second-personal address.

Thus, the Second-Personal Account does not face a dilemma between abandoning its solution to the Problem of Obligation and falling prey to the Problem of Antecedence. It can maintain that some demands (in the pre-theoretical, wider sense) do not give rise to obligations while other demands (in the irreducibly second-personal, narrow sense) are in force independently of whether we actually address them without abandoning the claim that obligations are grounded in second-personal authority. Before I conclude, however, I want to address one more worry. One might worry that the Second-Personal Account, as I characterise it, abandons its original impetus, namely the attractive idea that “[t]he authorities to whom modern moral consciousness appeals are simply you and I” (Watson 2007: 38). In particular, it may appear that to the extent to which some obligations are in force independently of whether we address them, the actual address of demands that people comply with these obligations is redundant.

In response, I want to highlight another implication of the suggestion that demands are moves in a practice. Since, as I explained above, full participation in a practice requires acting from the rules of the practice, it also serves to uphold these rules, and thus the practice. Schapiro writes (2001: 105):

To uphold rules is to sustain the normative structure they impose on the world. By making choices on the basis of the practice rules, an agent, along with other rule-abiding participants, contributes to making the context in which she acts such that it is properly described in terms of concepts and principles of the practice.

Schapiro contends that the move-constituting nature of practice rules not only means that, when we fully participate in a practice, we act from its rules; it also means that, by acting from these rules, we contribute to a context in which our acts can be described as moves in the practice in the first place. Thus, castling requires acting from the rules of chess, and acting from the rules of chess, in turn, contributes to the possibility of describing acts as ‘castling’ in the first place. This means that the significance of moves in a practice does not only consist in what they accomplish within the practice; in addition, these moves have the significance of stabilising or reinforcing the rules of the practice. In this way, by performing the move of castling, one does not only affect the position of one’s pieces on the chess board, but one also upholds the practice of chess.

Thus, if demands are moves in the practice of second-personal address, an agent who makes a demand by following the normative felicity conditions of this practice is thereby upholding these conditions and this practice. This means that, although by demanding that A not enslave others, B does not create a new obligation on the part of A,¹⁶ B does thereby contribute to the continued possibility of applying second-personal concepts, such as moral obligation, legitimate demand, and second-personal authority, to the world. Thus, although B’s demand does not affect which obligations anyone has, it ensures, together with others’ similar demands, that people have the obligations they have. It follows that demands which do not merely conform to, but also follow the normative felicity conditions of second-personal address are never redundant. More specifically, they do not merely serve to point out independently existing obligations, but also contribute to the continued existence of these obligations by supporting the normative context that enables us to describe the world in these terms in the first place. As Darwall puts it, by making such demands, “we add our voice to or second, as it were, a demand that we must presuppose is made of everyone by the moral community or representative persons as such” (2013b: 37, italics added). By extension, we might say, we add our voice to or second the very practice of second-personal address. On my interpretation of the Second-Personal Account, then, while the normative felicity conditions of second-personal address are in a sense prior to demands, these conditions would not apply to anything—and, consequently, there would be no obligations—if no one ever made any demands.

10. Conclusion

The Relational Account purports to solve the Problem of Obligation by stating that obligations, by virtue of their inherent directedness, are implicated in a network of irreducibly relational norms. However, the Relational Account fails to illuminate the relationality of these norms in a way that is independent from, and thus could serve as an explanation for, their obligatoriness. On the contrary, the view of the moral relations between persons that is most congenial to the Relational Account is incapable of accounting for the three features of obligation that give rise to the Problem of Obligation. By contrast, the Second-Personal Account draws a connection between obligations and the practice of second-personal address that allows it simultaneously to solve the Problem of Obligation and to avoid the implausible implication that obligations depend on what people happen to demand of one another. More specifically, the Second-Personal Account regards many obligations as grounded in the demands that we count as making just insofar as we participate in the practice of second-personal address. If we decide to address these demands to one another after all, we uphold this practice, reinforcing the concepts that it imposes on the world, including the concept of obligation.