Are There Any Conventional Obligations?

Abstract

There are reasons to believe that conventional obligations are impossible. Thus, it could be argued that for me to have an obligation to Φ in virtue of the fact that a convention so requires, it must be the case that I have a convention-independent obligation to do something else such that, given the existence of the convention, Φing is a way of doing just that. But, then, my obligation to Φ would not really be conventional at all. On closer inspection, so-called conventional obligations turn out to be no more than a specification of what our nonconventional obligations require given the circumstances. In this paper, I shall argue that contra to what this argument suggests, there can be genuinely conventional obligations. To do so, I develop a second-personal account of conventional obligations, according to which obligations are grounded by conventions in virtue of an explanation that does not follow the indicated pattern.

I. Introduction

Some people doubt whether there are any natural obligations, that is, whether there are any obligations that are not grounded in human conventions. However, I believe that, to the extent that there are any obligations at all, there is little room to doubt that at least some of them are natural. What should really puzzle us is how there can be any genuinely conventional obligations.

For suppose that I have an obligation to ϕ that obtains in virtue of the fact that a social convention requires that I ϕ. How could that be possible? The mere fact that a convention requires that I ϕ is not, in itself, a reason for me to ϕ: there is nothing intrinsically valuable in conforming to conventions as such. But then how can it be that I have an obligation to ϕ in virtue of the fact that a convention so requires? Well, surely, it must be the case that I have a convention-independent obligation to do something else (say, to ψ) such that, given the existence of the convention, ϕing is a way of doing just that (i.e., a way of ψing). But, then, my obligation to ϕ would not really be conventional at all (not in any interesting sense, at least). I have a “conventional” obligation to ϕ simply because I have a convention-independent obligation to ψ that, given the existence of the convention, requires that I ϕ. Thus, on closer inspection, so-called conventional obligations turn out to be no more than a specification of what our nonconventional obligations require given the circumstances.

In this paper, I shall argue that, contra what this argument suggests, there are genuinely conventional obligations. Here is how I will proceed. First, I shall articulate the puzzle concerning conventional obligations more precisely (Section II). Second, I shall briefly discuss different ways philosophers have tried to account for the way in which conventions give rise to obligations. Some of these accounts accept the abovementioned conclusion (i.e., that there are no genuinely conventional obligations), while others do not. I shall argue that both are unsuccessful. Although the former can, indeed, explain how conventions give rise to obligations, they do so in a way that does not fully or adequately capture the normative relevance of conventions. The latter, in contrast, appeal to more ambitious accounts of the mechanism by which conventions are supposed to generate obligations. If they were to work, that is, if obligations could be generated by conventions in the way they suggest, they would adequately explain the normativity of conventions. The problem is that they do not work (Sections III and IV). Third, I shall develop what I call a Second-Personal Account of Conventional Obligations. In a nutshell, the basic idea is that we are accountable for conforming to the convention in virtue of the fact that we hold each other accountable for so conforming. I shall progressively clarify and refine this very basic idea in a way that, I hope, will highlight why it is attractive. Importantly, according to this account, conventional obligations are not merely specifications of what our nonconventional obligations require, given the existence of the convention. They are, that is, genuinely conventional (Section V). Finally, I address some objections (Sections VI, VII, and VIII) and briefly conclude by reflecting on the nature of genuinely conventional obligations so understood (Section IX).

II. The Puzzle of Conventional Obligations

In the introduction, I sketched an argument that purports to challenge the possibility of genuinely conventional obligations. Let me now try to articulate it more precisely so that we can appreciate its force and make explicit its underlying assumptions. To do so, I will start by clarifying some of its key notions.

A. Obligations and the Reasons Only Assumption

Let us start with obligations. I am afraid that I will have to be somewhat dogmatic about this in this context, but bear with me.¹ Most of the things I shall say about conventional obligations could still be said, mutatis mutandi, by appealing to other (plausible) accounts of obligations. But there is one crucial exception: my own account of conventional obligations depends on understanding obligations in the way I explain next. This should come as no surprise. Any account of conventional obligations will depend on a more general account of what obligations are. Mine is no exception. But, although controversial, my account of obligations is not outlandish. And, even if you believe that my account is misleading or erroneous, it will still be worth discussing, I hope, how conventions give rise to obligations in the way I understand them.

So, what is it to have an obligation on my account? Roughly, an agent A has an obligation to ϕ if and only if A is accountable for ϕing, that is, if and only if it is appropriate for someone else to demand that A ϕ and to blame her if she fails to do so without justification or excuse. More precisely, A having an obligation to ϕ consists in it being the case that it is appropriate for someone else to demand that A ϕ. The fact that A is thus accountable for ϕing is a reason, what Stephen Darwall calls a second-personal reason, for her to ϕ.² And, in turn, if A fails to conform to what she is obliged to do without justification or excuse, then it is appropriate to blame her for having so failed.

What kind of facts can ground obligations thus understood? That is, what facts can make it appropriate that I demand that you ϕ? Here is an obvious suggestion. One way in which I could adequately justify my demand that you ϕ is by appealing to the fact that you have demand-independent reasons to ϕ. Of course, not just any reason will do. There are many situations in which all things considered you ought to ϕ, but in which, nonetheless, it would not be appropriate for me to demand that you do so. What special kind of reasons for you to ϕ, then, are such that they make it appropriate for me to demand that you ϕ? This is an awfully difficult question. Here I shall simply assume that obligations can be grounded in reasons as long as they have certain distinctive features (call them binding reasons).

Now, many people are skeptical about the possibility of obligations being grounded in any other way. And there is, indeed, something off in the idea that there can be facts that justify a demand in its own terms without counting in favor of the demanded action. For, how come you can appropriately demand that I behave in a certain way without there being any reasons for me to so behave? Call this the Reasons Only Assumption. If we accept it, it follows that if social conventions can change what we are obliged to do, they must do so by changing what we have reason to do.

B. Social Conventions

In order to discuss how conventions give rise to obligations, we need not only an account of obligations but also some working conception of conventions. I say that all we need is some working conception because, plausibly, our understanding of conventions will be responsive to our understanding of the distinctive mechanism by which they manage to give rise to conventional obligations, if there is one. The idea is that the essential features of conventions will be precisely those that are crucial in the distinctive mechanism by which they impose obligations. But still, to get the discussion off the ground, we need some working understanding of conventions that we might later subject to revision. In that spirit, I shall appeal to a broadly Hartian account of conventions. Among other virtues, Hart's account of conventions is quite in line with our pretheoretical understanding, and dovetails nicely with the way I understand obligations. Thus, I shall say that there is a social convention in society S that requires ϕing in circumstances C if and only if (i) most members of S regularly ϕ in circumstances C; (ii) most members of S hold each other accountable for ϕing in circumstances C (i.e., they demand each other to ϕ in C, and blame each other for failing to ϕ in C without excuse or justification); and (iii) most members of S believe that they are accountable to each other for ϕing in circumstances C, inter alia, because they so hold each other accountable.³ In a nutshell, there is a social convention that requires us to ϕ in C when we treat each other as if we had an obligation to ϕ in C in virtue of the fact that, among other things, we so treat each other.

C. Conventional Obligations

We have already seen what obligations consist in and what social conventions consist in. What about conventional obligations? Not every obligation to act as a convention requires is a conventional obligation. For an obligation to be conventional, it must be the case that it is explained, in one way or another, by the existence of a social convention.

But not every obligation that is explained by the existence of a social convention is a conventional obligation either. Suppose there is a social convention that requires us to return slaves to their (putative) owners. This convention might play a role in explaining why I have an obligation not to tell anyone where Joe, my slave friend, is. Roughly, I have an obligation to help him to escape the rule of his master. Given the existence of the convention, people will probably return him to his owner if I tell them where he is. Therefore, given the existence of the convention, I have an obligation not to tell people where Joe is. We can say that this is a convention-dependent obligation: it is explained, inter alia, by the existence of a convention. But it is not, however, a conventional obligation: its content has little to do with what the convention requires.

Taking into account these two considerations, we could say that, in a broad sense, conventional obligations are obligations to do what a social convention requires that obtain (in some way) in virtue of the existence of that very social convention.⁴ Conventional obligations thus understood are a somewhat perplexing phenomenon. How can the fact that people act as if they have an obligation to ϕ make it the case that they do? How can the pretense that there is an obligation make it the case that the obligation actually exists?

D. Triggering and the Standard Model

One way of trying to dispel the mystery is by arguing that, actually, social conventions matter normatively in the same way that just about any other nonnormative fact does.⁵ Indeed, social conventions can causally affect the nonnormative circumstances in normatively relevant ways, thereby triggering reasons and obligations that exist independently of the convention. Remember the slave example. In that case, the existence of the social convention makes it the case that people will probably return my friend Joe to his owner. And this fact, in conjunction with my convention-independent obligation to help him escape, explains why I have an obligation not to tell anyone where he is. I argued that this convention-dependent obligation is not a conventional one because it does not match what the convention requires me to do. But, of course, social conventions can also trigger obligations that do match what the convention requires.

Suppose there is a convention that requires us not to drive on the left-hand side of the road. This causally affects the circumstances so as to make it the case that if I drive on the left, I will be putting other people's lives at risk. But I have a convention-independent obligation not to put other people's lives at risk. Hence, I have an obligation not to drive on the left, an obligation that obtains because the social convention that requires us not drive on the left makes it the case that if I drive on the left I will be putting other people's lives at risk, thus triggering my convention-independent obligation not to do so.

Thus, the argument goes, there is nothing mysterious about conventional obligations. So-called conventional obligations are simply obligations to do what the convention requires triggered by that very convention. But, then, our “conventional” obligations (e.g., not to drive on the left) are simply what our convention-independent obligations (e.g., our obligation not to put other people's lives at risk) require given the way the corresponding social convention causally affects the nonnormative circumstances. Consequently, they are conventional only in a weak, deflated sense. There are no, as we might call them, genuinely conventional obligations (i.e., conventional obligations that are not simply a specification of what our nonconventional obligations require). This is, in a nutshell, the challenge I presented above. But, actually, the problem runs deeper. Let me explain.

Triggering explanations follow what Mark Schroeder calls the Standard Model of normative explanation, which can be summarized as follows:⁶

Standard Model (SM): The explanation that A has a reason to ϕ because of a fact F follows the Standard Model if and only if:

1. i. It appeals to a reason that A has to do a further action, ψ.

2. ii. (i) is a reason that A has not just because of F.

3. iii. F explains why A's doing ϕ is a way of doing ψ.

The idea is, thus, that F is explanatorily relevant because it makes it the case that by ϕing A will be performing another action, ψ, which she has independent reason to perform.

Let us reconsider our previous example in this light. The explanation appeals to a reason that I have to perform a further action, namely, not putting other people's lives at risk (i.e., it satisfies (i)). My reason not to put other people's lives at risk is a reason that I have independenly of the existence of the social convention (i.e., it satisfies (ii)). And, finally, the fact that there is social convention that requires me and others not to drive on the left (i.e., the triggering fact) explains why my refraining from driving on the left is a way of not putting other people's lives at risk (i.e., it satisfies (iii)). Furthermore, considering that my independent reason not to put other people's lives at risk is a binding reason (so that it grounds a corresponding obligation), and that conforming to my derivative reason not to drive on the left is necessary for me to conform to that preexisting binding reason, I also have a derived obligation not to drive on the left.

However, although all triggering explanations follow the Standard Model, not all explanations that follow the Standard Model are triggering explanations. In fact, we ought to distinguish two different ways in which the existence of a social convention that requires that A ϕ (i.e., our fact F in the schema) can make it the case that by ϕing A will be performing some other action, ψ, which she has independent reason to perform.

A first possibility is that the social convention causally changes the factual, nonnormative, circumstances in such a way that A's ϕing is now a way of ψing. This is what happens in our toy example. The social convention causally changes the factual circumstances so as to make it the case that by driving on the left I would be putting other people's lives at risk. This is what distinguishes triggering explanations as a distinctive instance of the Standard Model, namely, that the explanation of why F makes it the case that A's doing ϕ is a way of doing ψ is a causal one.

A second possibility is that A's conforming to the social convention is constitutive of A's ψing. Suppose, say, that A has a reason to honor the President and that following a social convention with certain features (say, one for which everyone understands its purpose to be that of establishing a formal way of honoring the President) is constitutive of honoring her. That is, to follow the convention just is a way of honoring the President. Therefore, if the convention requires A to ϕ, then by ϕing A will be honouring the President (i.e., doing something that she has independent reason to do). The role that social conventions play in explanations of this constitutive kind is somewhat more robust, given that the relation between conforming to the convention and conforming to the convention-independent reason is not as contingent as in triggering cases. However, all obligations that obtain because of a social convention in virtue of an explanation that follows the Standard Model, even of the constitutive variety, are conventional only in a relatively weak or diluted sense. Indeed, if the explanation of the normative relevance of social conventions follows the Standard Model, one can have a “conventional” obligation to ϕ only if one has a convention-independent obligation to ψ such that ϕing is a way of ψing in virtue of the existence of a convention that requires one to ϕ. But, then, they are not genuinely conventional.

E. The Challenge Summarized

We can now reconstruct the challenge to the possibility of genuinely conventional obligations as follows:

1. 1. A weakly conventional obligation is an obligation to act as a social convention requires that obtains in virtue of that very social convention.

2. 2. A genuinely conventional obligation is a weakly conventional obligation that obtains in virtue of an explanation that does not follow the Standard Model.

3. 3. An obligation to ϕ can be grounded only in reasons for ϕing. Therefore, social conventions can change what we are obliged to do only by changing what we have reasons to do (Reasons Only Assumption).

4. 4. But social conventions can change what we have reason to do only by virtue of an explanation that follows the Standard Model (Standard Model Assumption).

5. 5. Hence, social conventions can change what we are obliged to do only by virtue of an explanation that follows the Standard Model (from (3) and (4)).

6. 6. Therefore, though there are weakly conventional obligations, there are no genuinely conventional obligations (from (1), (2), and (5)).

In what follows, I shall argue that, contrary to what this argument suggests, there are genuinely conventional obligations. Thus, I will provide an explanation of how conventions impose obligations that does not follow the Standard Model. To do so, I shall reject premise (3), i.e., the Reasons Only Assumption. The idea is that social conventions are facts that, without themselves directly counting in favor of the actions they require, can (under certain conditions) make it appropriate for others to demand that the corresponding actions be performed. Thus, conventions can ground obligations without relying on preexisting reasons. This is enough to refute (5) and the conclusion (6). Furthermore, the obligations that are so grounded in the convention are, in turn, a reason to perform the corresponding obligatory actions so that, by rejecting (3), (5), and (6), we are naturally led to reject (4) as well.

Before discussing this second-personal approach to the normativity of conventions, it will be useful to briefly discuss alternative accounts of how conventions impose obligations, to contrast them with the one I shall defend. I will proceed as follows. First, I will argue against the view according to which conventional obligations can be reduced to obligations to do what the convention requires triggered by that very convention (Section III). Second, I shall consider nontriggering accounts of conventional obligations and argue that either they do not work or cannot adequately account for the normativity of conventions (Section IV).

III. The Insufficiency of Triggering Accounts of Conventional Obligations

Social conventions can, and do, causally affect factual circumstances in normatively relevant ways, thereby triggering reasons and obligations that exist independently of the convention. As I suggested, some philosophers argue that conventional obligations are no more than obligations thus triggered by conventions. Let us call this the Triggering View.

There are, indeed, more or less recurring ways in which social conventions trigger independent reasons. Two have been widely discussed in the literature. First, social conventions can trigger reasons by way of solving coordination problems (the traffic convention in Section II is a good example). This fact is at the heart of David Lewis's influential account.⁷ For our purposes, we can characterize coordination problems as follows. There are certain patterns of action that are valuable for all of us to follow, so that, for any agent A, if (almost) everyone else is following the pattern, then A has a reason to follow it as well. There is a coordination problem when there are two or more such valuable patterns that are mutually exclusive. Two patterns are mutually exclusive if and only if it is impossible or pointless for us to follow both. In such cases, while we all have reason to coordinate, none of us knows which pattern everybody else will follow. A coordination problem is solved if one of the patterns becomes salient so that everyone has reason to expect that (almost) everyone else will follow it. The idea is that social conventions can solve coordination problems by making one of the patterns salient. Given the existence of a social convention that enshrines one of the patterns, we all have reason to expect that everyone will continue to follow it, and, hence, each of us has a reason to follow it.

A second recurring way in which social conventions trigger convention-independent reasons and obligations is by giving rise to expectations concerning other people's actions. The existence of social conventions leads people to expect others to conform to them. Individuals rely on their convention-based expectations to plan their lives. Thus, agents will often adopt plans that will turn out to be detrimental for them unless others conform to the existent social conventions. In these circumstances, by failing to conform to the existent convention we will be harming those who have come to detrimentally rely on our so conforming. We have a reason/obligation not to harm others. Thus, by making it the case that people will detrimentally rely on others acting as they are conventionally required to, social conventions trigger a reason/obligation to conform to them.

However, there are reasons to believe that the Triggering View is an unduly narrow account of the normative significance of social conventions. Consider coordination. No doubt certain social conventions trigger reasons by way of solving coordination problems. However, there are also a number of binding social conventions that do not seem to solve any coordination problem. One way of bringing this point to the fore is by considering that a mere regularity of conformity to the relevant pattern of action would be sufficient to make it salient, even if people did not hold each other accountable for conforming. Thus, the accompanying practice of mutual accountability plays no role in explaining why our reason to conform to the social convention obtains. The order of explanation is, rather, the opposite. The fact that members of society S regularly conform to the relevant pattern of behavior triggers a reason for each of them to so conform, and it is because they recognize that each of them has such a reason to conform to the pattern that they hold each other accountable for doing so. This reveals the limits of coordination to account for the normative significance of conventions. Indeed, there are many patterns of actions such that if they were mere habits, they would not bring about a corresponding obligation to conform to them, but that, when accompanied with the relevant practice of mutual accountability, they do succeed in bringing it about. However, if coordination were the only way in which social conventions bring about obligations, that would not be possible; if a social convention triggers an obligation to conform to it in virtue of solving a coordination problem, then the corresponding habit would trigger the exact same obligation. Hence, if practices of mutual accountability can make the difference, then it must be the case that social conventions can bring about obligations in a way other than by solving coordination problems.

An analogous argument could be extended to the detrimental-reliance account. However, things are not so clear-cut here. Someone could argue that there is a relevant difference between mere habits and social conventions in their capacity to trigger our reason/obligation not to harm others. We do not have an obligation to perform just any action that others have detrimentally relied on us performing. The expectation must be, in some sense, reasonable or justified. And, the argument could go, though mere regularities might lead individuals to expect others to act as everyone else does, they are not enough to justify, in the required sense, that expectation. In contrast, continues the argument, if there is a social convention rather than a mere regularity, then the expectation of conformity it brings about is sufficiently justified or reasonable, so that, if people come to detrimentally rely on it, then others have a reason/obligation to act as expected. However, there are reasons to believe that detrimental reliance is not enough to explain the normative significance of conventions either. On this account, conventions trigger obligations in virtue of the (justified) expectations that they engender. But most social conventions are binding even if the obliged person explicitly announces that she will not conform to it. However, in such circumstances, despite the existence of the convention, others will not expect this person to conform to it or, if they do, their expectation would not be justified in the required sense. This shows that typically social conventions bring about obligations in a way other than by engendering justified expectations of conformity.

These brief commentaries are not meant to be a conclusive argument against the Triggering View. My point is simply to illustrate the kind of difficulties it faces. There are two ways in which one could argue more systematically against it. First, one could point to instances of conventional obligations that cannot be explained in terms of triggering. Thus, for example, David Owens convincingly argues that some conventional familial obligations cannot be explained in terms of triggering.⁸ Second, more directly, one could point to other ways in which social conventions bring about obligations. Here I shall pursue this second strategy.

IV. Nontriggering Accounts of Conventional Obligations

In very broad terms, there are three traditional models that attempt to explain how social conventions can ground obligations by a mechanism other than triggering. The first model appeals to the value or even “moral necessity” of the convention, the second tries to complement the first with a duty of fairness or reciprocation, while the third one appeals to the agents’ reasons for participating in the convention or playing their role in it.

A. The Principle of Established Practices

The first approach faces an obvious problem. The fact that it is valuable for the social convention to exist does not seem to constitute a reason to conform to it except in those rare circumstances in which failing to do so can undermine it. Thus, the fact that the social convention is valuable or morally necessary is a reason to bring it about, and a reason against doing anything that could lead to its disappearance (i.e., a reason to “support” the convention), but it is not a reason to conform to its requirements.

Consider, for example, Thomas Scanlon's Principle of Established Practices:

Principle of Established Practices: In situations in which (i) there is a need for some principle to govern a particular kind of activity, but (ii) there are a number of different principles that would do this in a way that no one could reasonably reject, (iii) if one of these principles is generally accepted in a given community, then (iv) it is wrong to violate it.⁹

This principle purports to explain “how the fact that a principle is generally but not unanimously accepted in a given society can make that principle binding on all.”¹⁰ However, I doubt that it succeeds. Some of the notions in the formulation of the principle require further clarification: What is it for an activity to be governed by a principle? And what does it mean for there to be a need that an activity be so governed? On one plausible understanding, an activity is governed by a principle when its participants regularly conform to it and hold each other accountable in terms of it, that is, when a social convention that enshrines the principle exists. And there is a need for some principle to govern an activity if and only if it would be morally valuable for the activity to be governed by a social convention that enshrines some principle so that individuals are morally required to bring about a social convention that does so. However, if we understand the principle in this way, then it is not clear why it would be wrong to fail to conform to existing social conventions where conditions (i)–(iii) are satisfied. The fact that there is a need for a social convention to regulate an activity does not seem to be a reason to act as it requires unless doing so would somehow threaten its existence. However, not every failure to act as a social convention requires threatens its existence.

But maybe I have misunderstood the principle. Maybe for a principle to govern an activity is not for its participants to hold each other accountable in terms of it. Rather, a principle governs an activity in the relevant sense if and only if its participants are accountable for conforming to it. On the first reading, that a principle governs a practice is a descriptive fact. On this second reading, in contrast, it is a normative one: a principle governs a practice just in case there is an obligation to act as it requires. Thus, on this reading, there is a need for some principle to govern an activity when it would be morally valuable for its participants to be accountable for conforming to it. The argument could be, thus, reconstructed as follows. It would be morally valuable for the participants in this activity to be accountable for conforming to some principle. But there are two or more inconsistent principles to which it would be morally valuable for them to be accountable to conform. Thus, given that there is more than one principle that it would be morally valuable for the participants of the activity to be obliged to conform to, social convention gets to determine which principle they are actually bound to.

This explanation relies on two main ideas. The first is that the fact that it would be morally valuable for us to be accountable for conforming to a principle can make it the case that we are so accountable. The second idea is that when there is more than one principle such that it would be morally valuable for us to be accountable for conforming to it, social convention determines which one binds us. Both are problematic, but here I will focus on the first one. We ought to reject the claim that the mere fact that it would be morally valuable for someone to be accountable for ϕing can make her so accountable. This is, indeed, an instance of the wrong kind of reasons problem.¹¹ The fact that it would be valuable, even morally valuable, for you to be accountable for ϕing is not a consideration on the basis of which I can come to the conclusion that you are accountable for ϕing. The fact that it would be valuable for it to be appropriate that I demand that you ϕ does not help to establish that my demand that you ϕ is appropriate. Nor does the fact that it would be valuable for you to be blameworthy for failing to ϕ help to establish that you are blameworthy for so failing. The fact that it would be morally valuable for you to be accountable for ϕing is a reason to bring it about that you are accountable, but it does not show that you are already accountable.

Owens has suggested yet another reading of the sense in which it is necessary that some principle P govern an activity. Thus, he claims that the need that makes adherence to the established practice obligatory is the “need for socially recognized bonds.”¹² In a nutshell, the idea is that we have what he calls “deontic interests,” so that “among the things good for us for their own sake [i.e., noninstrumentally] are phenomena like rights and obligations.”¹³ In turn, this sort of deontic value (Owens switches freely between talk about “value” and talk about “interests”) is a species of conventional value, that is, a value “whose realization essentially involves the recognition of a social convention.”¹⁴ The claim is that rights and obligations “must have a certain a social reality if they are to serve genuine human interest and make anyone's life go better.”¹⁵ Thus, there is a need that some principle P govern an activity if and only if: (i) it would be valuable for some P regulating the activity to be binding; and (ii) a binding P that regulates the activity would be valuable because it would serve our deontic interests, that is, interests that can only be satisfied by the existence of binding conventions. We are bound by conventions precisely because (and to the extent that) they fulfill such need, that is, because it is good for its own sake for us to be so bound. Hence, “only practices that make our lives go better by satisfying a genuine normative [deontic] interest ground obligation.”¹⁶

On Owens's rendering of the principle, the fact that it would be valuable for us to be accountable for conforming to a principle P is not enough to make it the case that we are so accountable. To ground our accountability, the value of our being accountable for conforming to P must be noninstrumental: our accountability must serve our deontic interests. However, the requirement that the value of accountability be noninstrumental does not help to answer the wrong kind of reasons objection. The fact that it would be noninstrumentally valuable for you to be accountable for ϕing does as little to help to establish that you are accountable for ϕing as the fact that it would be instrumentally valuable for you to be so. I cannot come to the conclusion that you are accountable on the basis of the consideration that it would be noninstrumentally valuable if you were so, that is, on the consideration that it would serve someone's (deontic) interests that you were so. The fact that it would be noninstrumentally valuable for you to be blameworthy for failing to ϕ does not make you blameworthy for so failing. Accountability cannot be thus grounded in its value, be it instrumental or noninstrumental.

B. The Principle of Fairness

Thus, the value of a social convention, or of its being binding, is not in itself capable of grounding an obligation to act as it requires. This led authors to try to search for some independent ground that could help explain why we have an obligation to conform to valuable social conventions. One famous account appeals to the Principle of Fairness, first articulated by Hart,¹⁷ and later further developed by Rawls and others. Rawls formulates the principle as follows:

Principle of Fairness: A person is required [i.e., has an obligation] to do his part as defined by the rules of an institution [or social convention] whenever (i) he has voluntarily accepted the benefits of the scheme or has taken advantage of the opportunities it offers to advance his interests, (ii) provided that this institution is just or fair.¹⁸

The intuitive idea is that:

[W]hen a number of persons engage in a mutually advantageous cooperative venture according to certain rules and thus voluntarily restrict their liberty, those who have submitted to these restrictions have a right to a similar acquiescence on the part of those who have benefited from their submission.¹⁹

There is a huge body of literature discussing the Principle of Fairness. Here I will limit myself to a few brief remarks. The underlying moral idea is that of reciprocity. The basic thought is that, under certain conditions, if someone deliberately does something good for us, then we have a reason/obligation to reciprocate. The suggestion is that this includes those cases in which others voluntarily subject their conduct to a social convention whose existence benefits us. No doubt we sometimes have a reason, and even an obligation, to reciprocate. But it is doubtful that we can adequately explain the bindingness of social conventions in these terms.

Notice that this is not a triggering explanation. Though it follows the Standard Model, the idea is that, in the relevant circumstances, doing what the just social convention whose benefits I have voluntarily accepted requires that I do is constitutive of reciprocating to those who have deliberately benefited me (in this case, by voluntarily submitting to the rules of the social convention in question). The claim is that there is no possible world in which I do what the just social convention whose benefits I have voluntarily accepted requires that I do and fail to reciprocate, and further, no possible world in which I reciprocate to those who have voluntarily submitted to the rules of the social convention without doing what the convention requires.

However, I doubt that these claims are true. Consider, first, the claim that by doing what the social convention requires I will be reciprocating to those who have benefited me by voluntarily submitting to its requirements. It seems clear that whether this is so depends on what the social convention requires me to do. For the required action to count as reciprocating to those who have benefited me, it must be something that benefits them in some way. However, this need not be the case. Suppose that a social convention forbids harmless trespassing. The existence of this convention benefits me. But it is not clear that I have a reason to refrain from trespassing in virtue of the fact that I have a reason to reciprocate to those who have similarly refrained. Indeed, it is hard to see how my refraining from engaging in a harmless act could count as reciprocating. Let us assume, further, that by refraining from trespassing I would be imposing a minor cost on one of those who have voluntarily submitted to the rules of the convention. Could I really say to her that, though I understand the burden my action imposed on her, I did it because I owed it to her and others who similarly submitted to the rules? That it was only fair for me to reciprocate what they have done for me? I am not denying that I have an obligation to refrain from trespassing in virtue of the convention. But if I do, it is not in virtue of the fact that by doing so I would be reciprocating.²⁰

Consider, now, the claim that there is no way in which I could reciprocate other than by doing what the convention requires. This seems unjustifiably restrictive. Suppose that in my neighborhood there is a social convention that requires everyone to help clean the public park the first Sunday of every month. I love the public park and use it every day. Furthermore, I voluntarily accept this benefit (I could easily avoid the park without cost). Thus, I have an obligation to somehow reciprocate for the efforts my neighbors make for me. And no doubt doing what the convention requires would be sufficient to reciprocate. However, I have better things to do on Sunday mornings, say, playing football with my old school friends. It is not that the costs (including the opportunity costs) of helping my neighbors to clean the park outweighs the benefits of a clean park. However, I would rather play football on Sunday mornings. But I do not want to wrong my neighbors by failing to reciprocate. Thus, I have decided to do something else for them. I am a well-known artist, so I thought I could use my talents to improve the neighborhood by making interventions here and there. Everyone loves them and greatly benefits from them, much more than they would if I were to clean the park instead. This is significantly costly to me, and I only do it as a way of reciprocating for what my neighbors do for me. However, some neighbors still resent me for not helping to clean the park. Many still demand that I do my part and blame me for not doing so. It might be that, indeed, I am accountable for doing what the convention requires. But it seems clear that the explanation cannot be that I am failing to reciprocate: I am doing a lot for the neighborhood, much more than anyone else, actually! Again, this is not to say that I do not have an obligation to clean the park. But the reason cannot be that by failing to clean the park I am failing to reciprocate—my artistic contribution is more than enough to reciprocate for what my neighbors do for me.

The precedent considerations show that, in the explanations that appeal to the Principle of Fairness, the fact that there is a social convention that requires that I ϕ plays no direct role in explaining my obligation to ϕ. Sometimes doing what the convention requires is a way of reciprocating, but not in virtue of the fact that the convention so requires. If conforming to the convention is a way of reciprocating, that is because, given the circumstances, it is a way of benefiting my neighbors. And even when conforming to the convention is a way of reciprocating, there might be other ways in which I could adequately discharge my obligation to reciprocate.

C. Reasons to Participate in the Convention

An alternative approach is to try to explain the bindingness of conventions by appealing to the reasons that individuals have to participate in a practice once it is established. The basic thought is that if I have reasons to participate in an activity that is defined by a set of conventional rules, then I also have reason to abide by those rules.

Andrei Marmor's account of the normativity of what he calls “constitutive conventions” follows this approach.²¹ A constitutive convention is a conventional²² system of rules that has three functions: (i) it defines/constitutes the practice; (ii) it regulates conduct within the practice; and (iii) it defines some of the values that are inherent to the practice and the kind of evaluative discourse that applies to it. Marmor argues that the normativity of constitutive conventions so understood is conditional, that is, “one has a reason to comply with such conventions only if one has a reason to participate in the practice, and undertake its basic commitments, which are constituted by the conventions.”²³ Thus, for the convention to be binding, there must be a convention-independent reason to participate in the practice. Constitutive conventions are merely auxiliary reasons that define what one needs to do to participate in the practice.²⁴ Marmor has in mind something like the following reasoning:

1. 1. I have a reason to play football.

2. 2. The system of rules SR defines what football is and how to play it. Thus, to play football I have to follow the rules that belong to SR.

3. 3. There is a rule R such that (i) R∈SR and (ii) according to R I have an obligation to ф.

4. 4. Therefore, I have an obligation to ф.

Again, this explanation follows the Standard Model, but is not a triggering explanation. The basic idea is that following the rules of football is constitutive of something I have convention-independent reason to do, namely, playing football.

However, there are several problems with this argument. One concerns the force of the reason in (1) vis-à-vis the force of the reason in (4). On this account, the force of the reasons that the constitutive convention provides (i.e., (4)) is not determined by the constitutive convention itself but by the force of the convention-independent reason to participate in the practice (i.e., (1)). Hence, it is invalid to conclude that I have an obligation to ф (say, refraining from touching the ball with my hands) from the fact that I have a reason to play football. All I can legitimately infer from (1)–(3) is that I have a reason to ф. However, this seems counterintuitive. When playing football, I normally have an obligation not to touch the ball with my hands, even if I had no obligation to play in the first place.

Furthermore, contrary to what the argument assumes, it is perfectly possible to participate in the practice without following all its rules. According to premise (2), in order to play football, I have to follow the rules of football. But this is simply false. If a football player commits a fault that does not mean that she is no longer playing football. On the contrary, one can only break a rule of football if one is paying football. Hence, if it is possible to break the rules of the game, it must be possible to play the game without conforming to all its rules. Thus, it is possible to conform to one's reason to play football without conforming to all its rules. It is true that unless I follow some rules of the game, then I will not be playing it. But for the argument to work, it is necessary for it to be the case that I cannot play the game without conforming to all its rules. And this seems clearly false.

V. Genuinely Conventional Obligations: A Second-Personal Account

Let us take stock of the argument so far. We are examining the question of how social conventions impose obligations and whether there are genuinely conventional obligations. One way in which conventions give rise to obligations is by triggering convention-independent reasons/obligations, but I argued that this was not enough to adequately explain their normative significance (Section III). Thus, we started looking for other ways in which social conventions can impose obligations. I examined various proposals in that direction and argued that they all fail for different reasons. Most of the examined accounts still follow the Standard Model. The idea is that doing what the convention requires is constitutive of doing something we have preexisting reason to do (e.g., supporting the convention, reciprocating to those who have benefited us by submitting to its rules, participating in the practice defined by the convention, etc.). Other explanations do not follow the Standard Model. The most important ones among the latter appeal to the value of there being a binding convention. But, as I have argued, the fact that it would be valuable for the convention to be binding cannot make it binding without further ado.

I want to explore a strategy of a different kind. The explanation I defend relies on the account of obligations I outlined in Section II according to which A has an obligation to ϕ if and only if it is appropriate for someone to demand that A ϕ. The basic idea is that the fact that I demand of you that you conform to a certain rule makes it the case that it can be appropriately demanded of me that I conform to that rule as well (provided there is a point or value in our being so accountable). Thus, contrary to the first family of explanations, what grounds our accountability to conform to the convention is not a reason to act as it requires. Rather, our accountability to conform to the convention is grounded in facts that are not themselves reasons to so conform (i.e., the fact that we hold each other accountable). Thus, the explanation does not follow the Standard Model. In this regard, the strategy pursued here resembles the second family of explanations that appeals to the value of accountability. However, in contrast to this second family of explanations, the account I defend points to other facts that, in conjunction with the value of our being accountable to conform to the convention, ground such accountability. My contention is that these other facts are the right kind of reasons to establish accountability, thus avoiding the wrong kind of reasons objection.

A. The Basic Mechanism

The account of conventional obligations I want to defend here starts from the fact that, for a social convention to exist, there must be a practice of mutual accountability. This feature of social conventions is crucial to understanding the way in which they ground obligations. The basic idea is that we are accountable for conforming to the convention in virtue of the fact that we hold each other accountable for so conforming.

The basic mechanism underlying genuinely conventional obligations is the following. A is accountable for conforming to the convention because she holds B accountable for so conforming, and B is accountable for conforming to the convention because she holds A accountable for so conforming. Thus, A and B are accountable to each other for conforming to the convention because they hold each other accountable for so conforming.

In order to make things more concrete, consider the following example. In the building where I live there is a convention requiring residents to refrain from hosting parties after 8:00 p.m. That is, we, the residents of the building, hold each other accountable for so acting. Thus, we normally demand of other residents that they refrain from hosting parties after 8:00 p.m., and reproach them when they fail to do so. For instance, last week I mentioned my neighbor Ann, from apartment 2A, that I was planning to celebrate my niece's birthday on Tuesday's night. She reminded me that it wasn't allowed to host parties of that kind after 8:00 p.m. in this building, and thus, she added, I should think of an alternative arrangement if I expected that it was going to last longer than that. Although politely, she was firmly demanding that I refrain from hosting a party after 8:00 p.m. I nodded grudgingly, saying that I would make sure everything was wrapped up before that. I said that I nodded grudgingly because I still remembered what happened about two months ago. Ann was celebrating her retirement with some friends and their meeting went on way after 8:00 p.m. I had to wake up early next morning and I couldn't sleep due to the noise. Around 11:00 p.m. I couldn't take it anymore. I went down the stairs, knocked loudly at her door, and, less than fully politely, I said that she should end the meeting right away, that she knew perfectly well that hosting parties after 8:00 p.m. wasn't allowed, and that her behavior was unacceptable. She apologized, saying that she hadn't realized, that time had flown by, etc. She and her friends left her place soon after and I finally managed to get some sleep.

As I said, exchanges of this kind are normal among the residents in my building. We normally hold each other accountable for refraining from hosting parties after 8:00 p.m. We do not do so because someone whose authority we recognize has ordered us to so refrain or because there is some preexisting reason to refrain from hosting parties after exactly 8:00 p.m. Rather, we say, it is a norm of the building. And what we mean by that is simply that it is something we hold each other accountable for in this building (more on this soon).

Now, what I would like to argue is that in so holding each other accountable for refraining from hosting parties after 8:00 p.m. we thereby actually make ourselves accountable for so refraining. That is, that in demanding that I refrain from hosting parties after 8:00 p.m., Ann is making herself accountable for so refraining; and that, likewise, in reproaching Ann for hosting a party after 8:00 p.m. and demanding that she stops, I am making myself accountable for so refraining as well. But why is that? Why is it that by holding Ann accountable for conforming to the building's convention, I thereby make myself accountable for conforming to it?

Well, by demanding of Ann that she refrains from hosting parties after 8:00 p.m., I convey that it is appropriate for me to so demand, that is, that she is accountable to me for doing so. But if she were accountable to me for conforming to the building's convention, then I would be accountable to her for conforming to it as well. We have, indeed, equal standing. I cannot regard Ann as being accountable to me without regarding myself as being accountable to her in the same way. The appeal here is to:

Principle of Equal Standing: There are no two persons, A and B, such that it is appropriate for A to demand that B ϕ in circumstances C without it being appropriate for B to demand that A ϕ in relevantly similar circumstances.

Thus, given the Principle of Equal Standing, by conveying that she is accountable to me, I am also treating myself as being accountable to her. Consequently, there is nothing that I can object to in her holding me accountable without undermining the standing I claim for holding her accountable—not even the fact that there is no other reason, apart from my holding her accountable, that appropriately justifies her holding me accountable. But if there is nothing I can object to in her holding me accountable, then it seems that it is not inappropriate for her to so hold me accountable. And if holding me accountable is not inappropriate, then it is appropriate (there are no instances of accountability that are neither appropriate nor inappropriate). And, by definition, if it is appropriate for her to hold me accountable, then I am accountable. Therefore, by holding her accountable for conforming to the convention I thereby make myself accountable for conforming to it, too. Thus:

Basic Mechanism (first draft): The fact that A and B hold each other accountable for conforming to a convention makes them accountable to each other for so conforming.

This is just the basic idea I want to begin with. It needs to be improved upon in several ways.

B. Holding Ourselves Accountable by Appealing to the Fact That We So Hold Ourselves

A first amendment is the following. Contrary to what the explanation above might suggest, it is not the case that every time I hold someone accountable for doing something, I thereby make myself accountable for doing that same thing in relevantly similar circumstances. Crucially, it matters to what facts I appeal in order to justify my demand that she so behaves. Consider the following vignette. My neighbor Ann owns a dog, Bernie, whom she adores. But when Bernie misbehaves, she is sometimes extremely harsh on him, inflicting on him what I regard to be unwarranted and gratuitous suffering. I once saw her beating him up right at the building's front door, and I grabbed her by the hand, demanding that she stop, which she did. But in doing so I did not make myself accountable to her for, say, not being cruel to animals. The explanation has to do with the kind of reasons in virtue of which I considered it to be appropriate for me to demand that she stop: roughly, that animal suffering, and Bernie's suffering in particular, is bad, and that she has a reason not to be cruel to him, a reason of a kind that adequately justifies me holding her accountable for acting as it requires. If I was right about those reasons, then I was already accountable to her for not being cruel to animals quite independently of me holding her accountable in the way that I did. And if I wasn't right, then she was not accountable to me, and I am not accountable to her either.

Thus, if to justify my demand that you ϕ I appeal to some convention-independent reasons for you to ϕ (that is, reasons for you to ϕ that are not grounded, in the way we are presently discussing, in the fact that we hold each other accountable for ϕing), then the fact that I demand that you ϕ does not thereby make me accountable for ϕing in relevantly similar circumstances. In such cases, whether I am accountable for ϕing in relevantly similar circumstances depends on whether the reasons I invoke are, indeed, good reasons (i.e., whether they actually count in favor of you ϕing), and on whether they are reasons of the right kind to ground accountability. If they are not, then you are not accountable, and neither am I. The appropriate response on your part is simply to deny that we are accountable in virtue of such supposed reasons, rather than holding me equally accountable when my time comes.

There is a further complication. The previous paragraph seems to suggest that if people hold each other accountable for ϕing in virtue of convention-independent reasons, then there cannot be a conventional obligation for them to ϕ. But this is too strong. There seem to be actions such that we have both conventional and nonconventional obligations to perform them, both obligations playing a role in our practices of mutual accountability (e.g., there could be a binding convention in the building that, say, prohibits certain forms of animal cruelty or something along those lines). Further, we have only said what accountability must not be based on; but we have not said anything about what people do appeal to in order to justify their demand and reproaches, if not reasons for doing the corresponding action. Let us start with the latter issue. The idea is that, when there is a convention, the members of the relevant group justify their demands and reproaches for ϕing in C by appealing, as an independent ground, to the fact that ϕing in C is something they hold each other accountable for doing. That is, they justify holding each other accountable by appealing to the fact that they so hold each other accountable (thus, on my example from previous section, in justifying our mutual demands and reproaches, both Ann and I appeal to the fact that there is a norm of the building that prohibits hosting parties after 8:00 p.m., which is no more than a shorthand for the fact that refraining from hosting parties after 8:00 p.m. is something we, the residents of the building, hold each other accountable for doing). They appeal, thus, to the very mechanism we are trying to understand.

Regarding the first issue, the qualification “as an independent ground” is all we need in order to address it. It is possible for the members of the relevant group to justify their demand and reproaches by appealing to different, independent, grounds. Thus, one can appeal to convention-independent reasons for performing the demanded action (e.g., that animal suffering is bad), and, as an independent ground, to the fact that it is an action the members of the group hold each other accountable for performing (e.g., to the fact that “there is a norm of the building” that prohibits public displays of animal cruelty in common areas). What matters is that the practice of mutual accountability be appealed to as an independent ground. Thus, we can reformulate the Basic Mechanism as follows:

Basic Mechanism (second draft): The following facts—(i) that A and B hold each other accountable for фing in C; and (ii) that when called upon to justify their mutual demands and reproaches, they appeal, as an independent ground, to the fact that this (фing in C) is something they hold each other accountable for doing—together make it the case that (iii) A and B are accountable to each other for фing in C.

C. Hume's Law

The account as it stands implies that the mere fact that there is a social convention that requires someone (who participates in the corresponding practice of mutual accountability) to ϕ makes it the case, without further ado, that she is obliged to ϕ. This, however, seems implausible. There are two concerns in this respect, a metaphysical-sounding one, which ultimately rests on a confusion, and a normative one, which is legitimate and should be addressed.

The metaphysical concern is that the Basic Mechanism runs afoul of Hume's Law, according to which normative statements cannot be deduced from descriptive or nonnormative statements alone. According to this objection, we can reconstruct the argument corresponding to the Basic Mechanism as follows:

1. 1. A and B hold each other accountable for фing in C.

2. 2. When called upon to justify their mutual demands and reproaches, A and B appeal, as an independent ground, to the fact that фing in C is something they hold each other accountable for doing.

3. 3. Therefore, A and B are accountable for фing in C, that is, they have an obligation to ф in C.

(1) and (2) are descriptive, nonnormative, statements. (3) is a normative statement. Hence, this argument violates Hume's Law. However, that is not an adequate reconstruction of the deductive argument we should associate with the Basic Mechanism. Rather, the argument should be reconstructed as follows:

1. 1. A and B hold each other accountable for фing in C.

2. 2. When called upon to justify their mutual demands and reproaches, they appeal, as an independent ground, to the fact that фing in C is something they hold each other accountable for doing.

3. 3. (1) and (2) constitute a reason of the right kind to justify A's demand that B ф in C and B's demand that A ф in C.

4. 4. Hence, there is a reason of the right kind to justify A's demand that B ф and B's demand that A ф (from (1)–(3)).

5. 5. By definition, if there is a reason of the right kind to justify A's demand that B ф in C and B's demand that A ф in C, then A and B are accountable for фing in C, that is, they have an obligation to ф in C.

6. 6. Therefore, A and B are accountable for фing in C, that is, they have an obligation to ф in C (from (4)–(5)).

This is a perfectly valid deductive argument. One could, of course, question the truth of (3). But not, I believe, on the ground that it amounts to a subterfuge that violates Hume's Law. (3) states that (1) and (2) ground (4). Is there anything objectionable about this? I do not believe so. The fact that by фing I would be killing ten persons is a reason for me to refrain from фing. The nonnormative fact (by фing I would be killing ten persons) grounds a normative one (I have a reason not to ф). Of course, the following argument would be invalid:

1. 1. By фing I would be killing ten persons.

2. 2. Therefore, I have a reason not to ф.

However, the following argument is perfectly valid:

1. 1. By фing I would be killing ten persons.

2. 2. (1) constitutes a reason not to ф.

3. 3. Therefore, I have a reason not to ф.

Is (2) a subterfuge that violates Hume's Law? Again, it does not seem to be. There is nothing metaphysically suspicious in general with nonnormative facts grounding normative facts. And, hence, there is no reason to believe that the Basic Mechanism violates Hume's Law in any relevant sense.

D. The Value of Accountability

The second concern is a straightforward normative one. The Basic Mechanism as described so far implies that, no matter what the convention requires, as long as we hold each other accountable on its terms, then we are bound by it. However, this seems to lead to implausible normative conclusions. Suppose we all hold each other accountable for returning slaves to their owners or for torturing youngsters who have indulged in minor sins. Does it follow that we have an obligation to return slaves to their owners or to torture youngsters? The Basic Mechanism, as stated, should be rejected. Some evaluative or normative criteria must be included in order to rule out the possibility of such hideous conventions being binding. Thus, the existence of a convention is not sufficient for us to be bound by it. It must also be the case that there is some value in our being so bound, that is, it must be valuable for us to be accountable to each other for acting as it requires.

Why is this so? How is it that, despite the Principle of Equal Standing, the fact that we hold each other accountable for torturing youngsters does not makes us accountable for doing so? Remember that the argument was that, given the Principle of Equal Standing, by conveying that you are accountable to me, I am also treating myself as being accountable to you. And, consequently, if I were to object to you holding me equally accountable, I would thereby be undermining the standing I am claiming to hold you accountable. However, if it would not be valuable for us to be accountable for doing what I am demanding that you do (e.g., because I am demanding that you torture a youngster), then, indeed, both of us ought to undermine the standing I am claiming to have rather than insist on both of us being equally accountable to each other. I cannot object to you holding me accountable without undermining my own standing to hold you accountable. Thus, once I have held you accountable for ϕing, whether I can reasonably object to you holding me equally accountable turns on whether there is any value in our being so accountable. Consequently, if I hold you accountable for doing something, then there is no reasonable objection I could make to you holding me equally accountable, provided there is value in both of us being so accountable. Thus:

Basic Mechanism (third draft): The following facts—(i) that A and B hold each other accountable for фing in C; (ii) that when called upon to justify their mutual demands and reproaches, they appeal, as an independent ground, to the fact that this (фing in C) is something they hold each other accountable for doing (and, if pressed, to the value of their being thus accountable); and (iii) that it is valuable for A and B to be accountable for фing in C—together make it the case that (iv) A and B are accountable to each other for фing in C.

Thus, the existence of a social convention is not sufficient for us to be accountable to each other for conforming to it. It must be the case that, in addition, it is valuable for us to be so accountable. Note, however, that the argument is not simply that we are accountable to each other for conforming to the convention because it would be valuable for us to be so. As I have argued, explanations that appeal only to the value of accountability run afoul of the wrong kind of reasons problem. In a nutshell, the problem is that the fact that it would be valuable for us to be accountable to each other is not a consideration on the basis of which I can come to the conclusion that we are so accountable. Here, however, our accountability is based not only on the value of us being accountable but also, crucially, in the fact that we so hold each other accountable.

To highlight this, let us compare the following two situations. Suppose, first, that without me ever having held her (or anyone else in the building) accountable for such a thing, my neighbor Ann demands that I maintain flowers on my balcony. She does not argue that there is a norm of the building that so requires (there isn't) or that there is some accountability-independent reason for me to maintain flowers on my balcony that justifies her demand that I do so. Rather, she says, I am accountable for maintaining flowers on my balcony because it is valuable for us to be so accountable to each other as residents of the building. I agree with her that it would indeed be valuable for us to be accountable to each other for maintaining flowers on our balconies. But I deny that the fact that it would be valuable for us to be so accountable is a reason of the right kind to justify that we are so accountable. Thus, I object that there is simply nothing that appropriately justifies holding me accountable for maintaining flowers on my balcony. Now suppose that, with time, a practice of mutual accountability does develop, wherein we demand of each other that we maintain flowers on the balcony, and reproach each other for failing to do so without justification or excuse. Now Ann, having noticed that my old flowers died, demands that I replace them, this time arguing that there is a norm of the building that requires that we maintain flowers on the balconies, that is, arguing that maintaining flowers is something we hold each other accountable for in this building. In having held Ann and others accountable for maintaining flowers on their balconies in the past, I've treated myself as being equally accountable to them. Thus, I cannot now reasonably object to her holding me equally accountable simply because there is nothing that appropriately justifies holding me accountable. Thus, the fact that I have held her accountable for maintaining flowers on the balcony is a consideration on the basis of which I can properly come to the conclusion that I am accountable to her for maintaining flowers on my balcony. Thus, it is the right kind of reason to justify my accountability. The only way in which I could reasonably object to her thus holding me accountable, despite having myself held her accountable in the past, is by arguing that there is nothing to be said in favor of this practice of mutual accountability, that we should not continue with it. The fact that we have equal standing to make demands and reproaches of each other should not condemn us to perpetuate a practice that lacks any value. This is why the fact that we hold each other accountable is not a sufficient condition for us to be accountable. However, as we saw, the fact that it would be valuable for us to be accountable is not sufficient either. Both the fact that we hold each other accountable for maintaining flowers on the balcony and the fact that it is valuable for us to be so accountable together explain why we are accountable to each other for doing so.

I have argued that my account is not vulnerable to the wrong kind of reasons objection because, according to it, our accountability to conform to the convention is grounded, crucially, in the fact that we hold each other accountable for so conforming, and not only on the value of us being so accountable. If I am right, then not every account that appeals to the value of accountability to explain the bindingness of conventions is undermined by the objection. But then, you might now wonder, why is it that the alternative accounts I examined in Section IV.A do fall prey to the objection? Granted, according to them, the value of us being accountable to conform to the convention plays a role in explaining why we are so accountable. But so does it on my account. And it is not as if those accounts appeal only to the value of accountability. After all, they are accounts of conventional obligations, and thus they would be nonstarters if the fact that a convention exists didn't also play some role in explaining the relevant obligation.²⁵ So what's the difference?

In a nutshell, the answer is that, on those other accounts, the value of accountability is still the sole direct ground of our being accountable to conform to the convention. This does not mean that other facts, including the existence of the convention, do not play some role in explaining our accountability. However, the explanatory role of these other facts, including perhaps the fact that we hold each other accountable, is auxiliary. This is most evident in the case of Owens's view. According to Owens's account, the fact that there is a convention that requires that we ϕ indeed plays a role in explaining why we actually have an obligation to ϕ. Thus, he claims that a rule is conventional in a group only if a sufficient proportion of the members of the group recognize it as binding.²⁶ And this social recognition, he argues, is crucial to explain why it is valuable for us to be bound by the rule: “In these cases, some form of recognition is essential to the value and so also to the bindingness of the obligation.”²⁷ Thus, for example, “the rules of Italian family life bind Italians because they are recognized in Italy; if no Italian acknowledged them, no one would be bound by them because no one could benefit from being so bound.”²⁸ Note, however, that the role of social recognition in this explanation is indirect. Social recognition helps explain why it is valuable for us to be bound. But it is the fact that it is valuable for us to be bound that, in turn, explains why we actually are bound. Thus, on Owen's account it is still the value of accountability that is supposed to do the explanatory heavy lifting of directly grounding the fact that we are accountable. And this is what the wrong kind of reasons objection suggests is impossible.

The case of Scanlon's Principle of Established Practices is somewhat more complicated. Here too the existence of the convention or, in Scanlon's terms, the general acceptance of a principle, plays a role in explaining why we are bound by that convention or principle. And the role it plays is not (or not only) indirect. However, its explanatory role is still auxiliary, simply enabling the fact that it would be morally valuable for us to be accountable for conforming to this principle to directly ground the fact that we are so accountable for conforming to it. In fact, the role of the convention seems to be limited to removing one particular kind of objection to the argument that I am bound by a principle in virtue of the fact that it would be valuable for me to so bound. Let us go back to my neighbor Ann's demand that I refrain from hosting my niece's birthday party after 8:00 p.m. Suppose that she were to argue that her demand is appropriate because it is valuable for us to be accountable to each other in this way. Here is one way in which I could object. I could concede that it would indeed be valuable for us to be so accountable. But it would also be equally valuable for us to be governed by a more party-tolerant principle, say, by one that requires that we refrain from hosting parties only after 10:00 p.m. rather than 8:00 p.m. These two different accountability arrangements, and many others, are evaluatively on a par.²⁹ So how come you, Ann, can demand that I conform to one rather than the other? Scanlon seems to envisage the following response on Ann's behalf: because, like it or not, the more demanding principle is the one that is generally accepted in this building. Thus, the existence of a convention that enshrines the principle, or its general acceptance, simply explains the following contrast, namely, why it is this valuable principle (no parties after 8:00 p.m.) rather than other conflicting principles that are evaluatively on a par with it (e.g., no parties after 10:00 p.m.) that gets to be binding. More precisely, the general acceptance of principle P explains why it is the fact that it is valuable for us to be bound by P that gets to ground the fact that we are bound by P, while the fact that it would be similarly valuable for us to be bound by some alternative conflicting principle does not get to ground any obligation. The explanatory role of the convention is that of a tiebreaker. Thus, it is still just the value of accountability that is supposed to be doing the explanatory heavy lifting of directly grounding the fact that we are accountable.

In contrast, on my account, the explanatory role of the fact that we hold each other accountable is not auxiliary to the value of accountability. Rather it is the fact that we hold each other accountable for doing something, and the fact that it is valuable for us to be so accountable, that together directly ground the fact that we are so accountable to each other. And that complex fact, unlike the value of accountability on its own, does constitute a reason of the right kind to come to the conclusion that we are so accountable to each other. Or so I have tried to argue.

VI. Beyond the Basic Mechanism

The account I have defended so far seems to imply that I cannot have a genuinely conventional obligation to ϕ unless I have held others accountable for ϕing. But this seems to be too strict. In fact, it rules out the possibility of certain kind of asymmetrical conventions being binding.³⁰ Roughly, a symmetrical convention is one that requires everyone to act in a certain way. An asymmetrical convention is one that requires different people occupying different roles (or in different circumstances) to act in different ways.³¹ Some asymmetrical conventions are such that the roles are relatively fixed so that it is atypical, if not impossible, for people to occupy more than one of the relevant roles either simultaneously or sequentially. Furthermore, some roles are what we might call outward-looking, that is, they are such that it is atypical for people occupying them to hold other people in that same role accountable for acting as the convention requires. Rather, people in outward-looking roles hold people in other roles accountable for doing their part without caring much about what people with whom they share a role do. The worry is that the Basic Mechanism, as described so far, seems to imply that people occupying outward-looking and relatively fixed roles cannot be bound by asymmetrical conventions.

Consider the following example. In Quiet Valley, houses are distributed in two areas, the High-Ground and the Low-Ground. Given the acoustics of the valley, the noises made in houses in the High-Ground can barely be heard by other houses in the High-Ground but are amplified in the Low-Ground and vice versa. With time, a convention developed that prohibits residents in the High-Ground from hosting parties during the night (from 8:00 p.m. to 8:00 a.m.) while it prohibits residents in the Low-Ground from hosting parties during the day (from 8:00 a.m. to 8:00 p.m.). In this way, each area fits people with different lifestyles. Most people own a house either in the High-Ground or the Low-Ground but not in both. Let us now consider the situation of Susan. Susan lives in the Low-Ground. She holds residents in the High-Ground accountable, demanding that they refrain from hosting parties during the night, and reproaching those who fail to do so. This, however, does not seem to make her accountable for doing what the convention requires of her as a resident of the Low-Ground, namely, refraining from hosting parties during the day. But Susan does not demand that her neighbors in the Low-Ground refrain from hosting parties during the day, nor does she reproach them for failing to do so. She couldn't be bothered. Thus, it seems that the convention does not bind Susan (i.e., that she has no obligation to act as the convention requires in virtue of the Basic Mechanism).³² But what is true of Susan is true of every typical resident in the Low-Ground, and, mutatis mutandis, of every typical resident in the High-Ground as well. Thus, it seems that, according to what I've said so far, this convention cannot be binding, not even on the assumption that it would be valuable for people in Quiet Valley to be accountable for acting as it requires. But this seems implausible. There is no reason to think that this convention cannot be binding just in virtue of being structured in the way that it is.

Admittedly, this is a somewhat silly and simplified example. But asymmetrical conventions of the relevant sort are not rare or peculiar. Consider, say, conventions that require employers to behave in a certain way, while requiring employees to behave in some other way, or conventions that require parents to behave in one way, while requiring children to behave in some other way, and so on.

The worry is, thus, that the account as described so far cannot account for the bindingness of asymmetrical conventions with relatively fixed and outward-looking roles. How could we amend it to allow for this possibility? An attractive suggestion is the following. Conventional norms often come in packages, that is, they constitute a system, so that by holding you accountable to conform to a conventional norm (that applies to you but not to me) I thereby make myself accountable to you for conforming to other conventional norms belonging to that same system (that apply to me but might not apply to you), provided it is valuable for us to be so accountable to each other. The difficulty here lies in providing an adequate account of the sort of systematic unity at stake. The account I am most sympathetic to at the moment claims that the relevant kind of unity is evaluative. Roughly, the idea is that two conventional norms constitute a single convention (or conventional system) just in case the value of you being accountable to me for conforming to one of them (that applies to you but not to me) depends on me being accountable to you for conforming to the other one (that does apply to me). Thus, by holding you accountable to act in a certain way I am treating myself as being accountable to you for acting in some other way you hold me accountable for acting because the value of you being accountable in the way that I hold you accountable depends on my being accountable to you in the way you hold me accountable.

That is a bit of a mouthful. But consider in this light the convention in Quiet Valley. The value of the convention in this case has to do, roughly, with normatively assuring residents a time of quietness while also allowing them some discretion for hosting parties. A binding convention imposing restrictions only on residents in the High-Ground or only on residents in the Low-Ground would have realized this value only partially, and defectively, being subject to objections that the one that actually developed is not. And, in contrast with a symmetrical convention requiring, say, everyone to refrain from hosting parties during the night (or during the day), the asymmetrical convention in Quiet Valley allows people certain leeway in determining what will be their assured time of quietness, and at what times they will enjoy the discretion to host parties. Thus, Low-Grounders being accountable to High-Grounders for refraining from hosting parties during the day contributes to realizing a value in a way that depends on High-Grounders being accountable to Low-Grounders for refraining from hosting parties during the night, and vice versa. This is the sense in which they constitute an evaluative unity. And this, in turn, is what explains why in holding High-Grounders accountable for refraining from hosting parties during the night, Susan is treating herself as being accountable to High-Grounders (and other Low-Grounders) for refraining from hosting parties during the day.

Although the account I've just sketched needs to be further refined, I think that something along those lines is a satisfactory way of extending the Basic Mechanism to explain the bindingness of asymmetrical conventions with fixed and outward-looking roles. Note, however, that, even so extended, the account implies that I cannot have a genuinely conventional obligation to ϕ unless I hold other people accountable for doing something. And some might worry that this is still too strict, quite independently of any worry about asymmetrical conventions. The charge here is not that the account fails to explain the bindingness of a certain kind of convention for individuals in standard or typical circumstances. Rather, the objection consists in pointing to somewhat nonstandard cases of individuals who, due to some peculiar circumstance, have not themselves held anyone accountable to conform to the convention (e.g., because the occasion to hold someone else accountable simply did not arise) but are still intuitively bound by it.

This worry, precisely due to its casuistic nature, is hard to answer adequately in general terms. In some cases, I would disclaim the intuition, try to explain it away, or, well, simply bite the bullet. However, there will surely be at least some cases in which I would like my account to be able to vindicate the intuition that the person in question is bound by the convention despite not having held anyone accountable. Here is a very sketchy suggestion of how we could further extend the account to cover such cases. The general idea is that other forms of second-personal engagement with the convention, or, more precisely, with the people whose conduct it regulates, could have the same normative effect, namely, making it appropriate that you demand that I act as the convention requires, provided it is valuable for us to be accountable to each other for acting as it requires.

To illustrate, consider the following example. Suppose there is a convention that confers Luisa, the owner of apartment 3A, the authority to regulate certain issues concerning common areas in the building,³³ while also imposing on her certain obligations as to how to exercise her authority. On the assumption that Luisa holds other residents in the building accountable for conforming to her directives, we could explain why she is bound to exercise her authority in accordance with the convention by arguing that the conventional norm that confers her authority and the ones that limit its exercise constitute an evaluative unity, so that in holding others accountable for conforming to the former she is treating herself as being accountable for conforming to the latter. But there is another way we could go. In fact, we could imagine that, for some reason (say, she has been living abroad since her authority was recognized, or some such), she has not held residents in the building accountable for conforming to her directives. Still, I would argue, she is bound by the convention. Simply by exercising her conventional authority, issuing orders that she claims to be binding in virtue of the fact that people in the building hold each other accountable for so conforming to her orders, she is treating herself as being accountable for conforming to the conventional norms that she is held accountable for conforming to, and that form an evaluative unity with the one that confers authority to her. Note that issuing an authoritative directive and holding other people accountable for conforming to it are two different actions. The claim here is that exercising conventional authority can have the same normative significance as holding other people accountable to conform to the orders issued in exercising it.

This, as I said, is just an illustration of the other forms of second-personal engagement I have in mind (the one I feel more comfortable with). But it is not the only one I would be ready to admit. I would probably also want to include something like a publicly displayed readiness to hold other people accountable, even if the occasion to actually do so has not arisen yet. However, note that these other forms of second-personal engagement have the relevant normative effect only against a background in which people normally actually do hold each other accountable for acting as the convention requires, that is, only against a background where the relevant convention actually exists as a matter of fact. In this sense, these various ways of extending the Basic Mechanism are parasitical on it.

Now, even thus extended, my account still implies that a person cannot have a genuinely conventional obligation unless she second-personally engages in some way with the people whose conduct is regulated by the convention, be it holding them accountable, publicly displaying a readiness to do so, purporting to make them accountable by issuing orders addressed to them, etc. In turn, this presupposes that the convention requires other people to behave in ways she could hold them accountable for acting. This is, I believe, exactly as it should be. I cannot be bound by a convention that I could not possibly hold other people accountable for conforming to. Thus, this rules out the possibility of people being bound by what we might call unilateral conventions, i.e., conventions that require me to act in certain ways without requiring other people to act in any way I could demand that they act.

A related implication is that young children who completely lack the ability to hold other people accountable (and of engaging second-personally in other relevant ways) simply cannot have genuinely conventional obligations. Some might find this troubling. I don't. In fact, I'm even attracted by the more general claim that very young children who lack the relevant second-personal competence cannot be bound at all, conventionally or otherwise. In any case, it should be borne in mind that children being unable to have genuinely conventional obligations is completely consistent with them having genuinely conventional rights. Furthermore, it is perfectly consistent with it being justified—even if not appropriate—to hold them accountable for conforming to the convention as part of the process of their socialization, that is, as a way of getting them to practically master the relevant second-personal repertoire, and eventually become bound by the convention. This, I believe, quite nicely reflects the way we actually treat children—we hold them accountable for conforming to conventions that apply to them, when we do, more as part of a pedagogical process than because we think of them as genuinely accountable.

The Basic Mechanism is, well, basic. In order to have a complete account of genuinely conventional obligations, we need to extend it in different ways. In this section I have discussed two ways in which we need to go beyond the Basic Mechanism. First, in order to account for the bindingness of asymmetrical conventions, we need to allow for the possibility of someone becoming accountable to do something others hold her accountable for doing in virtue of her holding them accountable for doing something else. Second, we need to allow for the possibility of someone becoming accountable for acting as the convention requires by forms of second-personal engagement other than actually holding other people accountable for conforming to the convention. These extensions are important and deserve to be discussed in further detail. But I hope it is clear that, still, the Basic Mechanism is more fundamental and, in a sense, the paradigmatic case.

VII. Avoiding Bootstrapping

I have argued that the existence of a practice of mutual accountability is crucial to explain the bindingness of conventions. However, this raises a further concern. How can binding conventions be rationally brought about? To be accountable to each other in virtue of the Basic Mechanism, it must be the case that we hold each other accountable. And if we are not yet accountable to each other, we cannot appropriately hold each other accountable. But if we cannot appropriately hold each other accountable, then what could justify holding each other accountable? It seems that conventions can only get started unjustifiably, by us holding each other accountable for something we are not yet accountable for doing.

To solve this puzzle, we must bear in mind the distinction between A's demand that B ϕ being appropriate or fitting and it being the case that, all things considered, A ought to demand that B ϕ. The latter implies that there are (conclusive) reasons for A to demand that B ϕ, but says nothing about the nature of those reasons. The former, in contrast, implies that there are reasons of a distinctive kind for A to demand that B ϕ, namely, reasons that make it appropriate for A to so demand, that is, reasons of the right kind to do so.

A convention can only get started if people begin to demand of each other that they do things they are not yet accountable for doing. This implies that they would be making inappropriate or unfitting demands. However, inappropriate or unfitting demands can be all things considered justified. Thus, the agents who perform the actions that help a convention get started need not be guilty of any sort of irrationality. The first, ungrounded, demands and reproaches can be perfectly justified precisely as a way of helping to build a valuable convention. Thus, they should be judged as invitations to build a valuable practice of mutual accountability.

VIII. Rejecting the Reasons Only Assumption

A distinctive feature of the account of genuinely conventional obligations I am defending is that it rejects what I called the Reasons Only Assumption. Conventional obligations are grounded in facts that are not reasons for the action we are thereby accountable for. The claim is that A can appropriately demand that B ϕ in virtue of the fact that B also demands that A ϕ and that it is valuable for them to be so accountable to each other. But the fact that B demands that A ϕ and that it would be valuable for A and B to be accountable for ϕing do not count in favor of A ϕing.

However, there is something off in the idea is that there can be facts that justify a demand in its own terms without counting in favor of the demanded action. How can that be? How come you can appropriately demand that I behave in a certain way without there being any reasons for me to so behave?

These rhetorical questions are not without force. But the explanation underlying the Basic Mechanism seems convincing. Thus, some argument would be needed to justify the requirement that a set of facts cannot be a reason of the right kind to justify a demand unless they count in favor of the demanded action. What could that argument be?

One possibility is to appeal to the nature of demands. The normal reason to demand that someone do something is to get her to do it. Of course, there might be other reasons to demand that someone do something (say, that I will a receive prize if I demand that you do something). But these are abnormal or deviant cases. The normal reason is to get the addressee to perform the demanded action. But for there to be a reason to bring about that the addressee does the demanded action, it must be the case that there is a reason for the addressee to do that action. Hence, for there to be a normal reason for A to demand that B ϕ there must a reason for B to ϕ. And what matters in order to determine whether a demand is appropriate or not are the normal reasons for demanding rather than those that are abnormal or deviant. Hence, a fact or set of facts cannot be a reason of the right kind to demand that B do something unless they include a reason for B to perform the demanded action.

This argument fails at the last step. We must distinguish between there being normal reasons for demanding that someone do something and there being reasons of the right kind that justify the demand in its own terms. Normally, one ought not demand that someone do something unless she has a nonoverridden reason to do it. However, this is not necessary for the demand to be appropriate. Whether the demand is appropriate turns on whether it would be appropriate for A to decide for B that she ϕ.³⁴ A, like B, ought not to decide that B ϕ unless she has an undefeated reason to ϕ. But whether it would be appropriate for A to decide for B that she ϕ is different from the question of whether A ought to so decide or not, if it were up to her. Thus, although for A to be (normally) justified in demanding that B ϕ it must be the case that B has a undefeated reason to ϕ, that implies nothing about whether the facts that make it appropriate for A to decide that B ϕ need to include a reason for B to ϕ. In other words, the normal reasons for demanding need not be among the reasons that make the demand appropriate or fitting.

A second argument could appeal to the nature of reproaches or blame instead. The fact that A is accountable for ϕing implies that it would be blameworthy for A to fail to ϕ without excuse or justification. But for A's failure to ϕ to be blameworthy, that is, for it to be fitting or appropriate to blame A for failing to ϕ, it must be the case that A had some reason to ϕ. In blaming A for failing to ϕ we are conveying, among other things, that it was wrong for A not to ϕ, that is, at the very least, that she had some reason to ϕ. Therefore, a fact or set of facts cannot be a reason of the right kind to ground A's accountability unless they include a reason for A to ϕ.

Again, this argument fails at the last step. It is true that the fact that A is accountable for ϕing implies that it would be blameworthy for A to fail to ϕ without excuse or justification, and that for an agent to be blameworthy for failing to ϕ, she must have some reason to ϕ. However, it does not follow that the facts that ground A's accountability must include a fact that counts in favor of A's ϕing. As Darwall has argued, I believe that the very fact that A is accountable for ϕing is a reason (i.e., a second-personal reason) for A to ϕ. In a nutshell, the idea is that this a matter of respect. I have a reason to respect you, and in order to do so, I have to act as it is appropriate for you to demand that I act. In turn, the fact that A has a second-personal reason to ϕ is a reason of the right kind to blame A for failing to ϕ without justification or excuse. If one accepts that accountability is itself a reason-giving fact, then there is no inconsistency in simultaneously claiming that accountability can be grounded in facts that are not reasons for performing the demanded action and that one cannot be accountable for ϕing unless one has a reason to ϕ.

The requirement that accountability must be grounded in facts that constitute a reason for performing the demanded action has some intuitive appeal. However, we can explain it away by revealing the confusions underlying the arguments that seem to support it. The claim has some plausibility because it seems to be implied by our belief that, normally, we ought not to demand that someone do something or blame her for not doing it, unless she has a reason to perform that action. However, it is not so implied.

IX. Conclusion

I have defended that there are genuinely conventional obligations. Genuinely conventional obligations are not any less “real” than nonconventional obligations. An agent A having a genuinely conventional obligation to ϕ consists in A's being accountable for ϕing and constitutes a second-personal reason for A to ϕ, just as any other obligation. What is distinctive about genuinely conventional obligations is the way they are grounded.

In the standard, nonconventional, case, I have an obligation to ϕ, that is, I am accountable for ϕing, in virtue of the fact that I have an accountability-independent reason to ϕ that appropriately justifies the demand that I ϕ. The fact that I am accountable for ϕing is, in turn, a further reason (a second-personal one) to ϕ. This second-personal reason to ф is, however, parasitic on my non-second-personal reasons to ϕ that ground my accountability for ϕing in the first place.

Genuinely conventional obligations are different. Their ground is not the existence of accountability-independent reasons that count in favor of the obligatory action and appropriately justify demanding that I do it. Rather, they are grounded in the existence of a practice of mutual accountability, provided there is value in the participants of that practice being so accountable to each other. These facts do not directly count in favor of doing those actions we hold each accountable for doing. They are facts that appropriately justify that others demand that we conform to the convention, without counting in favor of doing so. The fact that we are so accountable is, in turn, a reason to conform to the convention. But it is not one that is parasitical on any accountability-independent reason to so act.

This contrast between obligations that are grounded in accountability-independent reasons for doing the obligatory action and obligations that are grounded in other facts is a relevant one. I do not want to assume, however, that only genuinely conventional obligations are of this second kind. Thus, I do not want to draw the distinction in terms of conventional and nonconventional or natural obligations.

The relevant contrast can be adequately expressed, I believe, in terms of the distinction between moral and nonmoral obligations. Moral obligations are those grounded and supported by independent reasons for the obligatory action. Nonmoral obligations are those grounded in facts that are not reasons for the obligatory action. A subset of nonmoral obligations are genuinely conventional. Note that the claim that there are nonmoral obligations thus understood is controversial. A fortiori, the claim that conventional obligations are nonmoral is also controversial. Many authors insist that the explanation of obligations in terms of conventions must follow the Standard Model. I have tried to prove them wrong: there are nonmoral, genuinely conventional obligations.