I. That 'Sort of Case'

Forty odd years ago Ronald Dworkin drew a distinction between two kinds of moral practices. The first he called concurrent morality and the second he called conventional morality.¹ He drew the distinction in the course of criticizing Hart's 'social rule' theory, according to which moral obligations are generated by social practices. The example Hart had used throughout the Concept of Law was the social rule requiring men to bear their head in church.² Dworkin's initial criticism was that not all moral obligations arise in that way. He argued that people believe moral obligations exist, and do so plausibly, even in the absence of a social practice to that effect;³ morality does not have to be practiced to be obligatory. Dworkin concluded that Hart had mistaken the part for the whole and that his social rule theory must be weakened to apply only in 'one sort of case', namely, 'when the community is by-and-large agreed that some such duty does exist', and follows the same rules.

That 'sort of case' is of direct relevance to legal philosophy. Law is a social practice of collective action involving officials engaging with rules. If at least one type of obligation arises in virtue of social practices in which individuals accept and follow the same rules, then legal obligation -at least that of officials, if not of citizens- could be explained as falling within that type. Dworkin objected however that Hart had failed to draw an important distinction within 'that sort of case'; this is the distinction between practices of concurrent and practices of conventional morality, both of which involve individuals agreeing upon and following the same rules. Hart's theory, Dworkin argued, can only explain practices of conventional morality, whereas legal practice is an instance of concurrent morality. But his main criticism against Hart's 'social rule' theory cut deeper; even if the theory were to be confined to practices of conventional morality, it was -he argued- an inadequate account. Dworkin raised in effect a two-fold charge against the social rule theory: it does not apply to practices of concurrent morality and, where it does apply (i. e. conventional practices), it is inadequate and needs modifying.

The rest of the story is well known to legal philosophers. In the posthumously published postscript, Hart accepted Dworkin's initial criticism and weakened his theory as applicable only to 'conventional social rules'. But he resisted Dworkin's main objection. He insisted that the social rule theory can account for conventional practices without any modifications. And he insisted further that conventional social rules include the rules practiced by legal officials (what he called the Rule of Recognition). Legal positivists were since divided between those who sought to defend Hart's turn to conventionalism⁴ and those who repudiated it.⁵ Dworkin, on the other hand, went on to argue that conventionalism fails as a general theory of law.⁶

In this paper I want to revisit the distinction between concurrent and conventional morality and its bearing on law. I shall begin by arguing that there are in fact two different formulations of the distinction and that Dworkin equivocates between them. But my interest is not merely exegetical. On one version of the distinction, the one I think is most plausible, the subsequent debate in jurisprudence, on whether law is conventional, turns out to have been based on a misunderstanding of Dworkin's challenge to Hart. Moreover, and perhaps more surprisingly, the reading I propose makes the conventionality of law more central to Dworkin's own interpretive theory than it was previously thought.

II. The Background

The first part of Model of Rules II (hereinafter MOR II) Dworkin tells us, seeks to challenge a thesis that 'belongs to legal as well as moral philosophy', a thesis that he attributes to Hart. The thesis is this:

It argues, in its strongest form, that no rights or duties of any sort can exist except by virtue of a uniform social practice of recognizing these rights and duties. If that is so, and if law is, as I suppose, a matter of rights and duties and not simply of the discretion of officials, then there must be a commonly recognized test for law in the form of a uniform social practice, and my argument must be wrong.⁷

Dworkin's interest in MOR II is in explaining judicial obligation as an instance of genuine moral obligation. His starting point is that talk of obligation in law is to be taken at face value and to be explained in moral terms. This moralized starting point is not uncontroversial of course; it can be contrasted with a debunking approach, according to which judicial 'obligation' is to be explained as someone's belief or claim that law morally obligates, rather than as a genuine moral obligation. Be that as it may, Dworkin attributes the same moralized starting point to Hart; that is why he claims that Hart's 'social rule' theory belongs to moral and not just legal philosophy. And he takes Hart to be advancing a general thesis about moral obligation, namely that no moral obligation can exist, unless there is some uniform social practice recognizing it.

So the dialectic of Dworkin's argument in MOR II is this: if Hart's theory were to fail as an account of moral obligation, then it would also fail as an account of legal obligation. There would be no need to examine whether law meets the conditions of the 'social rule' theory, if such conditions could not possibly generate moral obligations. Of course, if the 'social rule' theory did provide a correct account of moral obligation then the question would be whether the practice of law meets the conditions for that obligation to obtain.

We can bracket for the moment the issue of whether Dworkin was right to read Hart's theory in this moralized way. It is not the only possible interpretation⁸ of pp. 55-58 of the Concept of Law, where the social rule theory is presented, and it has been subsequently challenged by legal positivists. The interpretation seems to beg the question against legal positivism, which denies that legal obligation is necessarily moral obligation. Be that as it may, what interest me here is the dialectic of Dworkin's argument: Hart's theory of law stands and falls together with his social rule theory qua a theory of moral obligation.

III. Conventional Morality: the Orthodox Reading

So what's wrong with Hart's social rule theory, qua a theory of moral obligation? Dworkin's initial objection need not detain us much, as its validity is relatively uncontroversial. Not all moral obligations are practiced-dependent. For example, the obligation not to torture depends neither on people refraining from torturing, nor on them agreeing that it is wrong. So if Hart's social rule theory is to succeed, it must be limited to a sub-set of moral obligations.⁹ Having established that not all moral obligations are practice-dependent, Dworkin moves on to introduce the distinction between concurrent and conventional morality. He defines the distinction as follows:

A community displays a concurrent morality when its members are agreed in asserting the same, or much the same, normative rule, but they do not count the fact of that agreement as an essential part of their grounds for asserting that rule. It displays a conventional morality when they do. If the churchgoers believe that each man has a duty to take off his hat in church, but would not have such a duty but for some social practice to that general effect, then this is a case of conventional morality. If they also believe that each man has a duty not to lie, and would have this even if most other men did, then this would be a case of concurrent morality. ¹⁰

Dworkin's objection is that Hart's theory fails to distinguish between the two practices, because the conditions of his social rule theory are met in both cases. In both cases there is convergence of behavior and an attitude of acceptance towards it. Yet Dworkin thinks there is a difference in kind between the two and that Hart's theory should be weakened as to apply only to cases of conventional morality. But at the beginning of the ensuing discussion, the argument for why Hart's theory should be weakened to exclude concurrent morality is unclear. In discussing the social rule not to lie, which he takes to be a paradigm case of concurrent morality, Dworkin writes:

It would distort the claim that the community made, when they spoke of a duty not to lie, to suppose them to be appealing to that social rule, or to suppose that they count its existence necessary to their claim. On the contrary, since this is a case of concurrent morality, the fact is that they do not. So the social rule theory must be confined to conventional morality.¹¹

The upshot of the argument seems to be that, though there is a social rule not to lie, we should not suppose people to appeal to that social rule when they speak of a duty not to lie. We should not suppose that the social rule is the reason why people claim that it is wrong to lie. Doing so distorts their claim. But why shouldn't we so suppose? What if they do in fact appeal to the social rule? Shouldn't we just check?

One way to read Dworkin's argument is that, as an empirical matter, people do not cite the social rule against lying when they assert the duty not to lie. Put differently, their motivating reasons for not lying, do not include the fact that there is a social practice of not lying. Motivating reasons are the considerations in the light of which people act.¹² If asked, they would reply that lying is wrong, regardless of whether there is a social rule against it. By contrast, when people are asked about why wearing a hat in church is wrong, they do -as a matter of empirical fact- appeal to the social rule as their motivating reason. This way of reading Dworkin's argument makes the distinction between concurrent and conventional morality turn on a distinction within the motivating reasons of those who engage in a social practice. Indeed, Dworkin presents it as such. In his definition, quoted above, the difference between the two is cast in terms of what members of the community 'count as an essential part of their grounds' for asserting duties and in terms of what duties churchgoers believe they have.

Let us call the Motivating Reasons reading of the distinction between conventional and concurrent practices MR for short. MR is further encouraged when Dworkin moves on to discuss conventionalism in law. He notes that 'it may be that at least some part of what judges believe they must do represents concurrent rather than conventional morality' (my emphasis), but that 'it is at least plausible to suppose that this is not so'. The idea seems to be that whether legal practice is conventional or concurrent depends on whether judges, as a matter of empirical fact, count some general judicial practice as an essential part of their grounds for any claim about their judicial duties.

The MR reading of Dworkin's distinction is also the one followed in the subsequent literature. Julie Dickson writes:

In order for the rule of recognition to be a conventional rule, the fact that there is a common official practice of recognizing certain things as constituting valid law must form part of the reasons why each official accepts the rule of recognition and treats it as binding.¹³

Here, like in Dworkin's discussion above, what is singled out are the considerations in the light of which practitioners (in this case legal officials) accept duties and comply with them. For Dickson, law is conventional if the social practice forms part of the reasons why officials accept an obligation. She moves on to remark that Hart perceived the case of conventional morality as a 'lifeline' thrown to him by Dworkin, and that Hart took that lifeline in the Postscript. Some passages in the Postscript suggest that Hart too understood the distinction under the MR reading. He writes:

Certainly the rule of recognition is treated in my book as resting on a conventional form of judicial consensus. That it does so rest seems quite clear at least in English and American law for surely an English judge's reason for treating Parliament's legislation (or an American judge's reason for treating the Constitution) as a source of law having supremacy over other sources includes the fact that his judicial colleagues concur in this as their predecessors have done.¹⁴

Reference to particular jurisdictions, with which Hart was familiar, shows clearly that Hart took the question of the conventionality of law to turn on empirical facts about the motivating reasons of officials. Following the MR reading, Dickson argues that in the first edition of the Concept of Law, Hart did not attempt to provide an explanation of the motivating reasons why officials accept and follow certain criteria of validity. But her argument goes further: Hart did not need to offer such an explanation; he could remain agnostic about the motivating reasons of officials, compatibly with his social rule theory. She argues that for the purposes of Hart's theory, all that is relevant is that legal officials accept and practice the same criteria of validity. In other words, Dickson disputes that Dworkin's distinction between conventional and concurrent morality, understood as a distinction within motivating reasons, is relevant to explaining judicial obligation. This is why, on her view, Hart jumped too quickly and 'jumped in the wrong direction'.

IV. Conventional Morality: the Alternative Reading

We have reasons to be skeptical about the MR reading of the distinction between concurrent and conventional morality. To begin with, it makes the debate between Hart and Dworkin turn on a factual point: do judges, as a matter of empirical fact, accept and apply the same rules (partly) because other judges do the same? Or do they do so for other reasons? But surely the answer to such questions requires legal sociology, not legal philosophy. It is not something that can be settled on a philosopher's armchair. The MR is even more puzzling when we locate it in the dialectic of Dworkin's argument in MOR II. Recall that Dworkin's aim is to challenge Hart's social rule theory as theory of moral obligation. Why would a distinction between the motivating reasons of practitioners challenge such a theory? If Hart's social rule theory correctly identified the conditions for a moral obligation to arise, then it would not matter that, as a matter of empirical fact, it captured practices of both concurrent and conventional morality. Dworkin would only make good on his aim of challenging Hart's theory if he were to show that it misfires in a moral sense: it postulates practice-dependent obligations where none exists, and it fails to identify practice-dependent obligations when they do exist.

My view is that Dworkin did not intend the MR reading, despite the formulations quoted above. These formulations no doubt en-couraged the MR reading on the part of legal positivists and started a prolonged -and in my view unproductive- discussion on why officials accept the criteria of validity that they do. Be that as it may, I want to propose an alternative reading, which I think is plausible in its own terms, whether or not it is supported by the text of MOR II. The alternative reading builds on the distinction, missed as often as it is observed, between motivating and normative reasons. Normative reasons are considerations that count in favor of an action;¹⁵ motivating reasons are considerations in the light of which someone acts.¹⁶ It is one thing to ask why people claim that lying is wrong; it is a different thing to ask what counts in favor of telling the truth.

So here is my suggestion: we should read Dworkin's distinction between practices of conventional and practices of concurrent morality, as a distinction within normative reasons. In both cases there is a genuine moral reason (or obligation) to take the action that is being practiced within a social group. The difference between concurrent and conventional practices lies in the grounds for that moral reason (or obligation). In the case of conventional morality, the fact that there is a common practice of φ-ing counts in favor of φ-ing; it is one of the grounds for the reason (or obligation) to φ. By contrast, in the case of concurrent morality, the fact that there is a common practice of φ-ing is not one of the grounds for the obligation to φ; the obligation would persist, even in the absence of a common practice. In other words, obligations within conventional practices are practice-dependent, whereas obligations within concurrent practices are practice-independent.

Here is then this second reading, paraphrasing Dworkin's original definition, which I shall call NR for short:

(NR) A community displays a concurrent morality when its members are agreed in asserting the same, or much the same, normative rule, but the fact of that agreement is not an essential part of the normative reasons for being bound by that rule. It displays a conventional morality when it is.

Applied to Dworkin's examples, the NR reading says that people's normative reason for telling the truth does not include the fact that others tell the truth and expect everyone to do the same. By contrast people's normative reasons for removing their hat in church does include the fact that others do the same and expect everyone to do so. The duty to tell the truth persists in a world full of liars. But the duty to remove one's hat vanishes in a world in which churchgoers have some other way to show respect, say by removing their shoes.

It is worth discussing two objections that can be raised against the NR reading of the distinction between practices of conventional and practices concurrent morality. First, it might be objected that our concept of convention picks out a type of motivating reasons and cannot, without distortion, be used to pick out a type of normative reasons. This is because conventions are typically thought of as an instance of a distinctive kind of motivating considerations, ones whose content includes the belief that the existence of a common practice provides normative reasons.¹⁷ I have addressed this objection at greater length elsewhere.¹⁸ It suffices here to say that our concept of convention is hardly confined to motivating reasons. Many paradigmatic cases of conventions (such as driving, queuing, greeting) are cases in which normative reasons obtain, and the grounds of these reasons include the fact that there is a common practice. A normative, or moralized, conception of conventions is hardly ruled out by the very meaning of the word 'convention'. But the more important point is that motivating considerations themselves can be grounds for normative reasons. What grounds the moral reason to follow convention, is not only the common practice of others, but also the considerations (e. g. expectations) in the light of which they act. This moralized approach to conventions provides a deeper explanation in my view of the central place motivating considerations have in our concept of convention.

The second objection to the NR reading is a friendly one, and it goes as follows: the MR is simply a proxy for the NR reading.¹⁹ The point of asking what people's motivating reasons are for taking part in some social practice, is in order to test whether their normative reasons are practice-dependent. Suppose we want to test whether people's normative reasons to remove their hat in church is grounded on the fact of a common practice to that effect. If, as a matter of fact, everybody removes their hat motivated by the existence of a common practice, then this provides a good indication that their normative reasons are practice-dependent. But if at least some churchgoers do not appeal to common practice, or if they appeal to different reasons (all however concurring that hats ought to be removed), then it is conceivable that their duty is practice-independent. Likewise, suppose we want to test whether people's normative reasons for telling the truth is practice-dependent. If, as matter of fact, nobody appeals to that common practice when they assert a duty not to lie, then this provides a good indication that their reasons are practice-independent. But if at least some do appeal to the common practice then this shows that it is conceivable that the duty not to lie is practice-dependent.

This second objection offers a more nuanced reading of Dworkin's argument in MOR II. It says that the point of the distinction between conventional and concurrent morality is not to classify social practices in terms of people's motivating reasons. Rather, the point of looking at what people take as their motivating reasons is two-fold: first to show that people who participate in social practices take the distinction between practice-dependent and practice-independent obligations to be of general significance; and second, to serve as a test for whether obligations within particular practices can conceivably be practice-dependent or practice-independent. More specifi-cally, the presence of concurrent reasons within a practice shows that it is conceivable that the obligations it creates are practice-independent. The latter point is of particular importance to Dworkin, as one of his aims is to dispel the assumption that there cannot be disagreement about what obligations the law imposes. If it shown that at least some judges do not count what other judges do amongst their reasons for deciding a case, then it is conceivable that judicial obligation is practice-independent, at least in the sense that it goes beyond what practitioners themselves think.

I have little quarrel with this second objection, as it is ultimately premised on the NR reading, using the MR reasing as its proxy. It is not so much a critique of the NR reading, as it is a recasting of Dworkin's argument. Whether or not the text of MOR II can bear this exegesis should not concern us here. But in my view the recast argument still suffers from the following defect: it fails properly to acknowledge the primacy of the NR reading over the MR reading. This becomes clear in cases where normative and motivating reasons are not aligned. In these cases, using the MR reading as a proxy for the NR reading misfires; as a result, the same social practice will come out as concurrent on one reading and conventional one the other. Consider the following examples.

Suppose that in some community people refrain from lying and accept that one ought not to, motivated solely by the fact that others do the same. They have not reflected on the morality of truth-telling and they refrain from lying just because that is the done thing in that community. This would be a case of conventional morality under MR, but not under NR. Objectively speaking, the moral reasons for telling the truth, do not include the fact that others do the same. There is a moral obligation not to lie, even if most people lie and even if most people think lying is morally fine. Or suppose that in some community most people drive on the right-hand side, and expect others to do the same, solely because the newly enacted traffic code says so and they believe that the law must be obeyed regardless of what others do. This would be a case of conventional morality under NR, but not under MR. The moral reason for driving on the right in that community would include the fact that others drive on the right and expect everyone to do the same. The further motivating considerations in the light of which they do so, and expect others to do the same, would be irrelevant. From the standpoint of real normative reasons, one has a moral duty to drive on the same side as that which others drive and expect everyone else to drive. This is because of moral principles to do with harm prevention. But whether common practice is the consideration in the light of which drivers expect others to drive on the right side, is neither here nor there. It commonly is, but it needn't.

There is another aspect of the NR reading which Dworkin's emphasis on motivating reasons obscures. Dworkin often contrasts conventional practices, in which people are motivated by common practice, with practices in which people are motivated by moral or substantive principles. He says for instance that if lawyers think a particular proposition is true by convention then 'they will not think they need a substantive reason for accepting it'.²⁰ But this contrast is false under the NR reading. On the NR reading, conventional reasons are no less in need of substantive or moral justification than non-conventional reasons. For instance one needs to say, by way of justification, whether the moral duty to shake people's hand is based on the principle of respect, or the principle of avoid offending reasonable expectations, or what have you. Moreover, practitioners too may have substantive moral views about the conventional reasons that apply to them: they may have read moral philosophy and take the principle of respect to be their substantive reason for accepting that they ought to shake people's hand. This would not undermine the character of their practice as conventional, understood in a normative sense. The epithet 'conventional' does not mean the absence of moral justification -be it in the minds of acting agents or morally speaking- not least because it is meant to capture a subset of genuine moral obligations.

In sum, the NR reading is self-standing and detached from the MR reading. It invites to ask the following question about each and every practice that meets Hart's social rule theory ('that sort of case'): is the moral duty that obtains therein, practice-dependent or practice-independent? This is not a question in social psychology, as the MR reading has led people to think. It is a question in moral philosophy. This is I think how we should read Dworkin's argument in MOR II.

V. Is The Social Rule Theory an Adequate Account of Conventional Morality?

Understood as a distinction within normative reasons, the distinction between concurrent and conventional morality poses a different challenge to Hart's thesis. Whereas Hart was right to point out that social practices harbor normative reasons (i. e. real reasons), the account of social rules he gave was over-inclusive. It captured practices (like the practice of not lying) where the normative reasons for going along are not -objectively speaking- grounded on the fact that others do so. Within the social practice of truth-telling for example, the fact that there is convergence of behavior, coupled by an attitude of acceptance of truth telling and shunning lying, is not a constitutive element of the reason for telling the truth. That is how we should understand Dworkin's point that 'it would distort the claim that members of the community made, when they spoke of a duty not to lie, to suppose them to be appealing to that social rule, or to suppose that they count its existence necessary to that claim'. The point is not that doing so distorts the community's motivating reasons. It may or it may not. The point is that it distorts the correct normative explanation of the moral reasons for not lying. By contrast, in the church example, the fact that churchgoers remove their hat and expect others to act accordingly, is a constitutive part of their normative reason to do so. Absent this fact, they would not have such reason.

Yet there is another, perhaps more important, sense in which Hart's social rule theory is over-inclusive, even if it is weakened so as to apply to conventional practices only. It captures practices that, though they meet Hart's conditions, are morally evil. Consider for instance Mafia practices in which people accept that they have a duty to commit murder and extortion and they are motivated solely by the fact that there is some such common practice. We would want to deny here that Mafiosi have a moral reason (or duty) to commit these impermissible acts. ²¹ Dworkin touches on this when he says that 'it is generally recognized, even as a feature of conventional morality, that practices that are pointless, or inconsistent in principle with other requirements of morality, do not impose duties'.²² So Hart's 'social rule' must be doubly weakened: first, to exclude social practices in which the normative reasons are practice-independent; and second, to exclude practices, which are conventional under the MR reading, yet generate no normative reasons because they are morally evil. Doubly weakened, Hart's 'social rule' theory captures a subset of genuine moral obligations, those whose ground include the fact that there is a common practice, as well as attitudes (such as acceptance or expectations), that people have towards it. Call these moral obligations conventional.

Doubly-weakened, Hart's 'social rule' captures a subset of moral obligation. And Dworkin far from disputes that conventional obligations constitute such a subset. Recall that his main objection is to do with whether Hart's theory gives an adequate explanation of why conventions impose obligations, taking for granted that they do. So we can now reformulate his main objection against Hart in the light of the NR reading: the social-rule theory gives an inadequate account of conventional moral obligations.

Consider now Dworkin's reformulated objection. Why is it that Hart's theory fails to explain adequately why churchgoers have a duty to bear their head in church? The answer lies in the connection Hart postulates between rules and obligations. It is useful here to note that both Hart and Dworkin are talking about rules understood in the context of social practices, and not about rules understood as verbal formulations, found in textual enactments or pronouncements. On Hart's view, rules in the former sense are constituted by empirical facts about the social practice in question, facts to do with the actions and mental states of practitioners. In other words, Hartian rules are non-moralized elements, which are defined by reference to social facts, and whose existence does not depend on whether the social practice in question generates moral obligation. When we judge the practice to create a moral obligation, then we are simply endorsing the social rule that exists therein. The moral obligation gets its content from the social rule, descriptively identified.

Dworkin thinks that Hart's account misconceives the connection between rules and the moral obligation that conventional practices generate. It does so in at least two ways. First, because it takes the content of the obligation to depend entirely on the content of the social rule that is being practised. It makes the social rule, understood as a narrow set of social facts about the practice and people's attitudes towards it, both the threshold and the limit of the moral obligations that the practice generates. Dworkin writes:

The social rule theory fails because it insists that a practice must somehow have the same content as the rule that individuals assert in its name. But if we suppose simply that a practice may justify a rule, then while the rule so justified may have the same content as the practice, it may not; it may fall short of, or go beyond it.²³

It follows that according to the 'social rule' theory, unprecedented or novel cases, about which there are no social facts (e. g. generally accepted attitudes), cannot -by definition- be governed by the existing social rule. In other words, social rules -understood in a non-moralized way- cannot be uncertain or be subject to non-factual disagreement.²⁴Dworkin says for example, that if there is no general agreement that male babies ought to bear their head in church, then according to the social rule theory, it is neither true nor false that they ought to. The upshot then of the 'social rule' theory is that practitioners cannot have obligations that go beyond the finite set of social facts that have obtained within it. This is exactly what Dworkin thinks is a mistake: people often assert practice-dependent duties whose content goes beyond what is generally accepted in the practice as a matter of social fact. They might for instance insist that male babies too have a duty to remove their bonnets in church, even though there is no general agreement to that effect. His point is that it seems perfectly intelligible to assert practice-dependent obligations that go beyond the social rule being practiced. And in so far as the social rule theory is meant to identify all practice-dependent obligations, then it is inadequate because it is under-inclusive. There is often more to conventional obligations than the obligation to do what most people do and expect others to do.

The second way in which the 'social rule' theory fails as an account of conventional practices on Dworkins view is this. Consider again novel or unprecedented cases in which people assert obligations that go beyond social facts to do with what most people do and expect others to do. One proposition they might assert is that practice-dependent duties still obtain in those cases. A further proposition they might assert is that the existing rules of the practice already govern these cases. Dworkin thinks that the latter assertion is no less intelligible than the former. He says for example that churchgoers might claim that the rule that men should bear their head applies equally to male baby bonnets,²⁵even though no such case has arisen before and even though the group does not share any views about it. Dworkin's argument, as I understand it, is that the intelligibility of such assertions suggests that the concept of a rule is itself a moralized one. It suggests that what the rules of a social practice are depends entirely on the question of what moral obligations the practice as a whole generates, and not on social facts, identified independently of that question. In other words, judgments about the obligations that social practices generate precede judgments about the rules of these practices. By contrast, on Hart's analysis it is the other way around: judgments about what rules exist within a social practice precede judgments about what obligations the practice imposes. On Hart's view, the normative judgment that men have a duty to bear their head is an endorsement of a rule whose content has been identified in a non-normative way. If Dworkin is right, Hart's analysis gets wrong the relation between judgments about the obligations that the practice imposes and statements of its rules. Rules track obligations, not the other way around.²⁶

Now let us assume that Dworkin is right that rules are moralized elements in the way just described. It is still possible on Dworkin's view that morality sometimes makes the social facts that obtain within certain practices the limit of the obligations that they generate. In these cases, we should not say that it is one thing whether a social rule exists and another thing whether the rule ought to be endorsed or followed. This would get the order of justification wrong: the social rule exists because it ought to be followed. Here is how Dworkin puts it:

It is true that normative judgments often assume a social practice as an essential part of the case for the judgment; this is the hallmark, as I said, of conventional morality. But the social rule theory misconceives the connection. It believes that the social practice constitutes a rule which the normative judgment accepts; in fact the social practice helps to justify a rule which the normative judgment states.²⁷

Dworkin's point is that social rules exist if, and only if, facts about the practice help to justify a moral duty to do what others in the practice do, and expect one to do, and no more. To say that a social rule -understood as a set of social facts- exists, is to have assumed not only the normative significance of these social facts but also that they exhaust the moral obligations that the practice generates. This follows from Dworkin's premise that the content of social rules is sensitive to normative judgments about what moral obligations the practice as a whole generates. And if that premise is sound, then it makes no sense to ask what reason there is to follow an existing social rule. By singling out certain social facts as constituting a rule of the practice, one has already assumed the normative significance of those facts and asserted that there is a reason to follow that rule. To wit: saying that there is a reason to follow a social rule is pleonastic; saying that there is no reason to follow a social rule is oxymoronic.

In sum, Dworkin's argument in MOR II, as I understand it, is that an adequate account of the normativity of 'those sort of cases' must meet the following five conditions. First, it must exclude social practices in which there is convergence of behavior and attitudes of acceptance, but in which the obligations that obtain are not practice-dependent (the no concurrent practices condition). Second, it must show that a substantive moral principle governs the social practice, such that the fact of common practice and related attitudes (e. g. expectations) are part of the grounds of the obligations that the social practice generates (the conventionality condition). Third, it must exclude social practices in which people are motivated by common practice to commit morally impermissible acts (the no threshold condition). Fourth, it must accept that practice-dependent obligations may not be exhausted by social facts to do with common practice and existing attitudes of acceptance towards it (the no limit condition). Fifth, it must make the content of the rules of the practice sensitive to judgments about the moral obligations that the practice, as a whole, generates (the moralized rule condition).

VI. 'The Positivist May be Right'

Many will no doubt question the conditions Dworkin imposes on an adequate account of social practices, summarized at the end of the previous section. But this question can be bracketed for a moment. For there is another question that has so far been left open by Dworkin's argument in MOR II: is law a conventional moral practice, assuming this time an adequate account of social practices?

We should begin by noting that this question is very different from Dickson's question of whether the Rule of Recognition is a conventional rule.²⁸ If Dworkin is right, and given that the Rule of Recognition is a social rule, then Dickson's question is tautological; it contains its own answer. For according to the moralized rule condition, statements about the existence of social rules are normative judgments about the obligations that the practice as a whole generates. The question for Dworkin is not whether there are reasons to follow the Rule of Recognition; the question is whether the common practice and attitudes of legal officials exhaust the grounds of the moral obligations that law imposes on them. And that latter question is for Dworkin the same as asking whether a Rule of Recognition exists. To repeat the upshot of Dworkin argument in the context of law: saying that officials have a reason to follow the Rule of Recognition is pleonastic; saying that there is no reason to follow the Rule of Recognition is oxymoronic. If rules track obligations, asserting rules amounts to asserting obligations.

So when Dickson asks whether the Rule of Recognition is conventional, she is not asking Dworkin's question. And it is in my view no accident that she follows the MR reading of conventional practices. If I am right, the MR reading is not the one Dworkin put as a challenge to Hart. What matters ultimately for Dworkin's challenge to Hart is not whether legal officials are motivated to act in accordance with what other officials do, but whether they have normative reasons so to act. It seems that much of the jurisprudence literature on the Hart-Dworkin debate (including some of Hart's own remarks in the Post-script) may have been a case of talking past each other.

Be that as it may, is law conventional in Dworkin's own terms? Here, I find Dworkin's own subsequent writing puzzling. At the fi-nal paragraph of the relevant section in MOR II, Dworkin leaves the question of the conventionality of law, as he understands it, open. He writes:

It may be that judicial duty is a case of conventional morality. It does not follow that some social rule states the limit, or even the threshold of judicial duty. When judges cite the rule that they must follow the legislature, for example, they may be appealing to a normative rule that some social practice justifies, and they may disagree about the precise content of that normative rule in a way that does not represent merely a disagreement about the facts of other judges' behavior. The positivist may be right, but he must make out his case without the short-cut that the social rule tries to provide.²⁹

But what is interesting about this passage is that Dworkin makes two points that in later work he seems to abandon. The first is that the conventionality of a social practice, understood in the NR reading, does not entail that the social facts of the practice exhaust the obligations that it imposes. He assumes in other words that conventionalism does not entail rejecting the no limit condition: the conventionalist needn't argue that novel cases are necessarily outside the scope of the obligations that the conventional practice imposes, nor that non-factual disagreement about the requirements of the practice is by default impossible. In other words, Dworkin does not equate Hart's 'social rule' thesis with conventionalism; he has argued the former is inadequate, whereas the latter is not. The second point Dworkin makes is that he takes the positivist's main claim to be that law is conventional, and not that judicial obligation is exhausted by the social facts of legal practice. He equates legal positivism with conventionalism, rather than with Hart's social rule thesis. This is why he thinks that the positivist 'may be right'. In other words, legal positivism does not entail rejecting the no limit condition either.

So the aftertaste of MOR II is constructive towards both legal positivism and legal conventionalism: positivism can succeed in a claim that law is conventional, so long as it understands this claim under the NR reading, and so long as it observes the no limit condition. But this aftertaste gets washed away quite abruptly by the waves of Dworkin's attack against conventionalism in Law's Empire. There, Dworkin equates positivism with conventionalism, but he defines the latter in a different way than he did in the MOR II: conventionalism is the view that the requirements of the practice have the same content as the rule that most practitioners assert in its name. That is, he equates conventionalism with the 'social rule' thesis, the one he has shown to be inadequate in MOR II. In chapter 4 of Law's Empire, Dworkin construes conventionalism as consisting in two claims, one positive and one negative. The positive claim is that conventions establish the threshold of judicial obligation: judges must respect the established legal conventions of their community. The negative claim is that conventions establish the limit of judicial obligation: there is no law apart from the law drawn from convention. The result of defining conventionalism in this way, is that legal conventionalism becomes implausible from the get-go. It is saddled with the all the problems stemming from rejecting the no limit condition. For instance the conventionalist cannot, by definition, offer an explanation of theoretical disagreement in law, since he denies that law obligates when there is no acceptance. And legal positivism gets in turn saddled with the same problems, as it too is committed to the rejection of the no limit condition. The constructive spirit that the MOR II displayed towards legal positivism is gone.

There are further differences between the MOR II and Law's Empire. The distinction between concurrent and conventional morality also appears in chapter 4 of Law's Empire. But Dworkin renders it exclusively under the MR formulation. He asks us to assume that there is consensus amongst lawyers and judges in Britain that if a statute is duly enacted by Parliament then what it says is valid law. This assumed consensus, he says, has two possible explanations:

Perhaps lawyers and judges accept that proposition as true by convention, which means true just because everyone else accepts it, the way chess players all accept that a king can move only one square at a time. Or perhaps lawyers and judges all accept the proposition as obviously true though not true by convention: perhaps the consensus is a consensus of independent conviction, the way we all accept that it is wrong to torture babies or to convict people we know are innocent.³⁰

The passage follows the MR reading and a strong one at that: a practice is conventional when practitioners accept a normative proposition 'just because everyone else accepts it' (my emphasis) and not partly because everyone else accepts it. Later glosses on the distinction reinforce the MR reading. Dworkin remarks that if lawyers think a particular proposition is true by convention then 'they will not think they need any substantive reason for accepting it' (emphasis in the original). Later on he contrasts accepting the authority of the US Constitution as matter of convention and accepting it as 'the upshot of sound political theory'. These formulations aim to support Dworkin's general claim that law is an interpretive concept. Lawyers disagree even when all the facts are known, while insisting that there is a right answer about what the law requires or permits. But as I noted earlier, these formulations are orthogonal to the distinction between conventional and concurrent morality, under the NR reading of MOR II. Judges may have a sound political theory according to which the actions and attitudes of other officials matter normatively to what they ought to do. Or they may just be applying criteria of validity just because other officials do it, whilst -unbeknownst to them- they are justified in so doing under the best political theory.

VII. Conclusion: Are We All Conventionalists Now?

So one might be left wondering why the more plausible version of conventionalism with which the MOR II ends, is not tested in LE. But it might be objected that this is just a matter of pure terminology. It might be argued that not only does Dworkin test legal conventionalism, but that he actually endorses it. After all, Dworkin's own theory emphasizes that past judicial practice matters for determining the content of legal obligations. He argues that past practice helps to justify normative reasons in virtue of the value of integrity, and that this value requires a degree of fidelity to the past. According to law as integrity we have to appeal at some point to the acts and beliefs of other legal actors (those we made decisions and those then followed them), in order to justify our current legal obligations. The past practice of governmental institutions, both courts and legislatures, changes what obligations we now have as citizens.³¹ Moreover law as integrity meets the no limit condition: one of its central claims is that the content of legal obligation is not exhausted by the historical record of past political decisions, but is sensitive to the moral principles that can be used to justify those decisions.

So we might as well say that Dworkin is a conventionalist in the sense that the MOR II leaves open. And the idea that non-positivists can be conventionalists should not at all surprise us. In an elaborate defense of normative conventionalism, Dimitris Kyritsis has argued that legal conventionalism sits comfortably within an interpretivist or natural law framework and that it is not exclusive to legal positivism. He puts the matter as follows:

What distinguishes natural lawyers from legal conventionalists does not lie in the natural lawyers' putative indifference to social facts -including the fact of a practice being adopted and practiced- but rather in their respective explanations of how social facts such as political decisions or social practices are connected to legal rights and duties.³²

I think Kyritsis is right to emphasize this point and right to highlight the common ground between legal positivists and their opponents. And maybe we should regret the fact that Dworkin did not write Law's Empire as a defense of conventionalism, but chose instead to write it as a critique of conventionalism. Had he done so, his approach would perhaps have been less polemical towards legal positivism. It may also have prevented a significant amount of the two camps talking past one another. Instead, Dworkin changed tack in Law's Empire and turned the strengths of a revised version of conventionalism into an argument in favor law as integrity and against legal positivism.

Be that as it may, the more important issue is to be clear as to what it means to say, as Hart did, that certain social practices are conventional. On the terms of MOR II, it means that the said practices generate moral obligations that are grounded in social facts to do with people's convergent behavior and their shared attitudes. And on the terms of MOR II, it also means that conventional moral obligations are not exhausted by these social facts. I have argued in this paper that the question of whether law is conventional must be interpreted and tested in those terms, and not in terms of what legal officials' motivating reasons are. It should be no objection against this interpretation that both positivists and non-positivists may turn out to be conventionalists, albeit of a differing kind. What should matter is whether conventionalism -properly understood- reveals important truths about the nature of legal obligation.