1 1 CONVENTIONALISM AND NORMATIVITY TORBEN SPAAK We have seen (in Section 3) that Hart objects to Austin s command theory of law, that it cannot account for the normativity of law, and that what is missing in Austin s theory is the idea of a rule. We have also seen that, on Hart s analysis, the foundation of the normativity of law is to be found in the rule of recognition, which is a judicial customary rule, that is, a social rule. As Hart sees it, the rule of recognition qualifies as a social rule because it satisfies the criteria in his general theory of social rules, sometimes referred to as the practice theory of rules (see, e.g., Raz 1990, 50-8). Under this theory, there is a social rule, R, in a community, S, if, and only if, the members of S (i) display a certain pattern of behavior that corresponds to R (the external aspect) and also (ii) hold that they ought to act in this particular way (the internal aspect). And since it is clear from Hart s account (1961, 112) that legal officials must view the rule of recognition as a common standard of correct judicial decision, and not as something which each judge obeys merely for his part only, we may with Hart (1994, 255) describe the rule of recognition as a conventional rule. As we have seen (in Section 2), Hart appears to operate with the conception of social normativity, according to which a standard is a social norm if and in so far as it is socially upheld as a binding standard and the society involved exerts pressure on people to whom the standard applies to conform to it. But we have also said that the conception or grade of normativity that is most interesting in this context is not social, but justified (or authoritative) normativity. However, some authors, who appear to have something like justified normativity in mind, have argued that if we can only refine Hart s practice theory of rules, we may be
2 2 able to account for the normativity of law. Gerald Postema (1982, 167), for example, proposes to shift the focus of the [conventionalist] doctrine away from the regularities of behavior and attitude which Hart believes constitute the rule of recognition to the strategic context, the context of practical reasoning, in which such regularities take on normative significance. On his analysis, a regularity, R, in the behavior of persons in a population, P, in a recurring situation, S, is a convention if, and only if, in any instance of S (1) it is common knowledge in P (a) that there is in P general conformity to R; (b) that most members of P expect most other members of P to conform to R; (c) that almost every member of P prefers that any individual conform rather than not conform to some regularity of behavior in S, given general conformity to that regularity; and (d) that almost every member of P prefers general conformity to some regularity rather than general non-conformity; and (2) that part of the reason why most members of P conform to R in S is that (1a)-(1d) obtains. 1 As Postema sees it (ibid., 182), a person who fails to do his part in such a conventionally defined joint effort acts unfairly, and this means that at least in some cases he may have an obligation to conform to the convention. Postema maintains (ibid., 194), that coordination is fundamental to law and that no legal system is conceivable without substantial coordination elements at its foundation. To defend his claim, he introduces a distinction between (i) first-, (ii) second-, and (iii) third-level coordination in law (ibid., ) and argues (ibid., 183) that there are significant secondand third-level coordination problems in a community living under law... [that it] is legitimate to describe social interaction (and the structure of practical reasoning) within the 1 One may well wonder whether (1c) adds anything important to (1d). If the members of the community prefer general conformity to some regularity, R, to general non-conformity, as (1d) has it, must they not also prefer, on pain of contradiction, that any individual conform rather than not conform to some regularity of behavior in S, given general conformity to that regularity, as (1c) has it?
3 3 context of law as a set of complex and persisting coordination problems. His idea, then, is that legal thinking typically aims at coordinating the actions of officials and lay persons. As he puts it (ibid., 187), law-identifying, law-applying, and law-interpreting activities of both officials and lay persons essentially involve a complex form of social interaction having the structure of a coordination problem or, rather, of an interrelated, continuous series or overlapping network of coordination problems. Having pointed out (i) that law aims to regulate the behavior of agents that respond to reasons, (ii) that law is a common public standard for behavior, (iii) that, generally speaking, the use of rules presupposes a fixed background context of interpretation and application, and (iv) that the authoritative law-identifying, law-interpreting, and law-applying activities of judges tend to shape the practical import of legal rules in a special way, Postema concludes that the understanding of law and legal requirements by officials and citizens are interdependent to an important degree in the sense that the understanding of the law on the part of the citizens depends on the expectations and understanding of the law on the part of the legal officials, and, conversely, that the understanding of law on the part of the officials depends on the understanding and expectations on the part of the citizens and, indeed, that all the elements of strategic interaction can be found in the relation between citizens and judges: Law can direct action to its ends only if its rules are integrated into the practical reasoning of those subject to the rules, and communication is an interactive process in which the understanding of each party depends on the expectations and understanding of the other. Authoritative interpretation and application of the law can to some extent take the lead there is, therefore, a significant asymmetry in the strategic relations between subjects and officials but it cannot depart substantially from the background practice of interpretation and remain an intelligible enterprise (or at least not one with the distinctive characteristics of law). (Ibid )
4 4 He then argues that we have reason to view the rule of recognition as precisely such a conventional rule aiming at coordinating the behavior of officials and citizens that may give rise to an obligation (or a duty): On this view, conventional judicial duties are genuine duties, not just forms of behavior which people may believe to be obligatory. Because they rest on an important (albeit limited) concern for fairness, they belong to a species of moral duty. (Ibid., Footnote omitted.) The important thing about the rule of recognition, on this analysis, Postema points out in conclusion, is not the underlying reasons for accepting it, but the fact of acceptance itself. The normative force of the rule of recognition, he explains,... depends simply on the fact that the rule succeeds in the task of coordinating law-identifying and law-applying activities of officials and lay persons. It provides a common way of acting on a large number of occasions, where often it is more important that there be a common way of acting than that it be the right, just, or otherwise best way. Judges need not believe, nor need it be true, that the existing rule of recognition is the ideally best such rule. They may believe, and have good arguments to show, that some other rule would be better, more just, or the like. But as long as the existing rule succeeds, and alternative rules cannot promise success, and as long as achieving coordination is at least minimally desirable, the existing rule will continue to generate (prima facie) obligations on the part of officials to conform to it. (Ibid., ) I find plausible Postema s claim that the rule of recognition is a coordination convention in the somewhat loose sense explained. It seems right to me to hold that the understanding of the law on the part of the citizens depends on the expectations and understanding of the law on the part of judges and other legal officials, and that the understanding of the law on the part of judges and other legal officials depends on the understanding of the law on the part of the citizens. I am not, however, convinced by Postema s claim that this means that law is essentially normative in the sense of justified normativity, that is, in the sense that judges and
5 5 citizens alike have a moral obligation (or duty) to obey the law at the most, it might mean that judges and other legal officials have such an obligation. 2 Indeed, I am not convinced that benefitting from a cooperative enterprise yields an obligation even in the best of circumstances (on this, see Nozick 1974, 90-5; Simmons, 1979, chap. 5). In any case, I doubt whether almost all citizens take part voluntarily in such an enterprise. Even anarchists who refuse to vote may to some extent be parties to such a coordinative enterprise, in the sense that they will have certain expectations and a certain understanding of the law of the land, which will depend to some extent on the expectations and understanding of the law on the part of judges and other legal officials. And it seems farfetched to maintain that they acquire an obligation to obey the law by virtue of being more or less forced to predict the lawinterpreting and law-applying activities of judges and other legal officials. Andrei Marmor (2010), who speaks about the rules of recognition in the plural, shares Hart s (and Postema s) view that these rules are conventional rules, and he also appears to share Kelsen s and Olivecrona s and Postema s view of what normativity is, namely something like justified normativity. He maintains, however, that the rules of recognition are best understood as constitutive as distinguished from coordinative conventions, and that this means that the normativity of law can only be conditional upon a wish on the part of the subjects of law to play the legal game. Marmor lays down the following three individually necessary and jointly sufficient conditions for the existence of a conventional rule: 2 To be sure, Postema mentions only officials in the quotation above. His general discussion of coordination at the foundation of law suggests, however, that he thinks that anyone who takes part in a cooperative enterprise acquires an obligation to obey the rules involved in the cooperation.
6 6 (i) There is a group of people, P, whose members normally follow the rules of recognition, RR, in circumstances, C. (ii) There is a set of reasons, A, for the members of P to follow RR. (iii) There is at least one other potential set of rules, SR, such that if the members of P had actually followed SR instead of RR in C, then A would have been sufficient for the members of P to do so. We see that, on Marmor s analysis, a rule, R 1, is a conventional rule if, and only if, (i) the members of a group of people follow R 1, (ii) they have reasons, A, to follow R 1, and (iii) A would also have been sufficient for them to follow another, similar rule, R 2. Condition (i) makes it clear that the members of the group must normally, usually, or on the whole, follow the rule. This simply means that not every member must on every occasion follow the rule. Condition (ii), as I understand it, means that the members of the group do not have to follow the rule because of the reasons, A, or, indeed, be aware of A at all. As Raz (1990, 178) might put it, while they must conform to the rule, they need not comply with it. And condition (iii) makes it clear that conventions are arbitrary, in the sense that a different rule than the one at hand might have fulfilled the function equally well. What is important is that there is a rule, not its specific content. Marmor notes (2010, 150) that some authors have objected to the view that the rules of recognition are conventional, that it is difficult to see how a mere convention could give rise to an obligation. He cites an article by Leslie Green (1996, 1697), in which the author says that the view (Hart s view) that the rule of recognition is a mere convention continues to sit uneasily with any notion of obligation. And he (Marmor) takes the gist of Green s
7 7 objection to concern the above-mentioned arbitrariness of conventions (ibid., 151), if the rules of recognition are arbitrary in the requisite sense, how can we explain the fact that they are supposed to obligate judges and other legal officials to follow them? His answer to this question suggests that he considers the objection to be well founded, arguing as he does that we have to make a distinction between the legal obligation to follow the rules of recognition and the moral obligation, if there is one, to follow these rules: The rules of recognition, like the rules of chess, determine what the practice is. They constitute the rules of the game, so to speak. Like other constitutive rules, they have a dual function: they both determine what constitutes the practice, and prescribe modes of conduct within it. The legal obligation to follow the rules of recognition is just like the chess players obligation, say, to move the bishop diagonally. Both are prescribed by the rules of the game. What such rules cannot prescribe, however, is an ought about playing the game to begin with. As I noted elsewhere, the normativity of constitutive conventions is always conditional. Conventional practices create reasons for action only if the relevant agent has a reason to participate in the practice to begin with. And that is true of the law as well. If there is an ought to play the game, so to speak, then this ought cannot be expected to come from the rules of recognition. The obligation to play by the rules, that is, to follow the law, if there is one, must come from moral and political considerations. The reasons to obey the law cannot be derived from the norms that determine what the law is. (Ibid., 152. Footnote omitted.) Marmor s view that the rules of recognition are constitutive conventions makes good sense to me, though I am not convinced that this means that they cannot also and, indeed, primarily fulfil a coordinative function. Marmor offers three distinct arguments in support of the claim that the rules of recognition, conceived as constitutive conventions, cannot also be coordination conventions. Specifically, he objects (ibid., 153-4) that the view that the rules of recognition are coordination conventions (i) misses the point that the rules of recognition constitute what law is, (ii) is not easy to reconcile with the political importance of these rules,
8 8 and (iii) that it blurs the distinction between the question of what law is and the question of what counts as law in a particular legal order. I am not, however, convinced by his argumentation. Let me explain why. Marmor argues, on the first count, that coordinative conventions do not usually have a constitutive function, that we cannot even begin to coordinate the actions of judges if we do not already see them as judges, that it is the rule of recognition that makes them judges, and that this means that that the rule of recognition cannot be a coordination convention (ibid., 153). But this argument is hardly air-tight. To begin with, although coordination conventions do not usually fulfil a constitutive function, this does not mean that they never fulfil a constitutive function, still less that they cannot fulfil such a function. Although Marmor is right to say that the much-cited problem of which side of the road to drive on clearly precedes and is independent of the convention that solves it, this hardly shows that there cannot be other conventions that fulfil a constitutive function by virtue of fulfilling a coordination function. Moreover, the idea that it is the rules of recognition that make a person a judge ( constitute a judge ) is problematic, given that, on Hart s analysis, it is the behavior of judges (together with a certain attitude, known as the internal point of view) that make up or constitute the rules of recognition. More specifically, in order to have the rules of recognition, we need to have judges, in order to have judges, we need to have rules of adjudication, and in order to have rules of adjudication, we need to have the rule of recognition. As far as I can see, we have here a vicious circle (on this, see MacCormick 1981, 108-9), and this in turn means that it cannot be the case, on Hart s theory, that it is the rule of recognition that turns a person into a judge. Hence if Marmor wishes to argue that the rules of recognition are constitutive conventions, he needs to explain just how the rule of recognition constitutes judges.
9 9 Marmor argues on the second count that since the rules of recognition are often politically important and it is difficult to see how coordination conventions which are necessarily arbitrary can be politically important, the rules of recognition cannot be coordination conventions. And he rightly points out that it matters to us who makes the law (politicians, judges, private parties, etc.) and infers that this means that the question of who makes the law cannot be a matter of coordination. But while it seems reasonable to assume that matters of pure coordination cannot be politically important there is, as I have said, the possibility that the rules of recognition may be both constitutive and coordinative in the sense explained above. Marmor argues, on the third count, that the view under consideration makes it unclear just how the rules of recognition relate to the concept of law. He maintains, as we have seen, that it is typically the case that we have independently of the relevant convention the concept of the activity that the convention is meant to coordinate, and that this follows from the point of having a coordination convention in the first place, namely that there is an activity that needs coordination (that is, a coordination problem that needs to be solved). As he puts it (ibid., 154), without the social conventions that constitute ways of making law and recognising it as such, it is difficult to imagine what kind of concept of law we could possibly have. But this particular argument appears to be aimed only at the view that it is in the nature of law to fulfil a coordinating function and not at the view that the rules of recognition are coordination conventions or at the view that the rules of recognition are both coordination conventions and constitutive conventions. For these are different ideas. One may certainly hold that the rules of recognition are coordination conventions without having to maintain that the basic function (or purpose) of law is to coordinate human behaviour. Moreover, just as a duty-imposing norm can also be a competence norm (or, if you will, a power-conferring norm), on the grounds that it confers competence on the members of a class of persons, A, by
10 10 virtue of imposing a duty on the members of another class of persons, B, to act in accordance with, say, the norms issued or the decisions made by the members of A (on this, see Spaak 1994, 171-3; 2003a, 94-7), a coordination convention can also be a constitutive convention, on the grounds that it imposes a duty on the members of B (assuming, of course, that it can impose a duty) to act in accordance with, say, the norms issued or the decisions made by the members of A. Let us, however, suppose, for the sake of argument, that Marmor is right that the rules of recognition are constitutive conventions and nothing more. My objection will then be that Marmor s claim that the normativity of law is, and can only be, conditional upon the existence of valid moral reasons to play the game of law is not a very novel claim. For it seems to be in substance, if not in form, more or less the same as Kelsen s view put forward by Kelsen already in 1925 (Kelsen 1993, 251) and repeated in his later books (see, e.g., Kelsen 1960; 1992; 1999) that the normativity of law can only be conditional upon the presupposition of the basic norm, or, if you prefer, seen from a point of view of someone who adopts the basic norm. Marmor s introduction of the idea of a constitutive convention is, to be sure, an important contribution to the debate, but it does not add much to the substance of Kelsen s claim as regards the central question of whether the normativity of law is conditional or non-conditional. Interestingly, Marmor objects that not much is explained by the presupposition of the basic norm (2010, 146): Instead of telling us something about the foundations of the basic norm, Kelsen simply invites us to stop asking. But this objection misses the mark. First, why would a presupposition need a foundation? Since Kelsen makes it clear that the theory of the basic norm provides no more than a conditional foundation of the normativity of law, why accuse him of failing to provide a non-conditional foundation? Secondly, as we have seen, whereas Kelsen appears to be offering a belief-based, not a validity-based, explanation of
11 11 legal normativity (see Section 7, Marmor clearly has in mind a validity-based explanation. And, as I have said, while it is difficult to see how a presupposition on the part of, say, judges or legal scholars could explain the normativity of law, conceived as an essential property of law, it is not difficult to see how such a presupposition (even if implicit) could explain the use of normative language on the part of those who make the presupposition. Kelsen may thus have succeeded in answering his own question, even if he did not succeed in answering Marmor s question. I agree, however, with (what I take to be) Marmor s view, that Marmor s question is more interesting. One might also be tempted to object that Marmor s second existence-condition for conventional rules, which does not require that the members of the relevant group of persons be aware of the set of reasons, A, for the members of the group to follow the rule, that the connection between the beliefs or attitudes of the people in question and the relevant pattern of behavior will be rather weak. As far as I can tell, the usual assumption in the literature on conventions is instead that the members of the relevant group of people must have a belief or an expectation about or an attitude to something, such as a pattern of behavior or some other type of entity. Thus Hart (1961; 1994), Lyons (1969, chaps. 1-2), Postema (1982, 176), and Lagerspetz (1995, chap. 1) all take this view. On Lagerspetz s account (1995, chap. 1), for example, a social rule, R, exists if the members of the relevant group of people (1a) believe that R exists and (1b) believe that the others in the group believe that R exists, and (2) act accordingly, that is, speak of R as existing and, if occasion arises, treat R as existing, at least partly because they have the beliefs (1a) and (1b). 3 3 Note that the account lays down sufficient, but not necessary, conditions for the existence of conventional facts. The reason, Lagerspetz explains (1995, 19), is that there may be some rules that are not generally followed by large parts of the population, perhaps because they are not known to the population, but which nevertheless may be said to exist in the sense that they are part of a system of rules that satisfies the conditions laid down in the account.
12 12 Marmor does not explain why he chooses to formulate condition (ii) the way he does. But one may assume that he has in mind objections to David Lewis s account of convention, such as the one put forward by Tyler Burge in his article On Knowledge and Convention (1971). Burge s central objection (ibid., 254-5) to Lewis s account according to which there is common knowledge in the relevant group of people that the relevant conditions hold is that it takes too little note of the extent of the unconscious element in many conventions and the possibility of their being essentially misconceived or misvalued. Burge objects, in particular, that the members of the community need not know that there are alternatives to the convention in question. He considers the case of an isolated community in which the inhabitants speak a certain language and have no idea that there are other languages, and that they might equally well have spoken one of those other languages, since they believe perhaps for religious reasons that their own language is the only language there is and can be (ibid., 250). Nevertheless, Burge says, we would not hesitate to say that their language is conventional. I shall, however, have to leave it an open question whether Burge s objection is well founded.