INTRODUCTION
The legal theorist Scott Shapiro puts it well: “Legal philosophers never tire of saying that the law is a social practice. But what precisely does this dreary bit of jurisprudential boilerplate mean?”1
Scott Shapiro, Legal Practice and Massively Shared Agency (unpublished manuscript), at 1.
For a representative sample of the Lewisean approach, see Gerald Postema, Coordination and Convention at the Foundations of Law, 11 J. LEGAL STUD. 185 (1982); Jules Coleman, Negative and Positive Positivism, 11 J. LEGAL STUD. 139 (1982); Gerald Postema, CONVENTIONS AT THE FOUNDATIONS OF LAW, IN 1 THE NEW PALGRAVE DICTIONARY OF ECONOMICS AND THE LAW 465–472 (P. Newman ed., 1998); Jules Coleman, INCORPORATIONISM, CONVENTIONALITY AND THE PRACTICAL DIFFERENCE THESIS, IN HART'S POSTSCRIPT 99–148 (Jules Coleman ed., 2001); and ANDREI MARMOR, POSITIVE LAW AND OBJECTIVE VALUES (2001), esp. ch. 1. See also Eerik Lagerspetz, THE OPPOSITE OF MIRRORS: AN ESSAY ON THE CONVENTIONALIST THEORY OF INSTITUTIONS (1995), ch. 7. Postema, who was the first to use a Lewisean framework to explain how rules of recognition are social rules, presumes that the function of rules of recognition is the resolution of recurrent coordination problems. But one need not adopt this view to be a conventionalist about the law. This is important because Leslie Green has provided a very strong argument against taking fundamental legal rules—rules of recognition—to be Lewisean conventions whose function is to solve a recurrent coordination problem. See Leslie Green, Positivism and Conventionalism, 12 CAN. J. LAW & JURISPRUDENCE 35 (1999). Marmor, id., argues that one can be a conventionalist without endorsing the view that the aim of the rule of recognition is resolution of recurrent coordination problems. Lewis's account of conventions is found in DAVID LEWIS, CONVENTION: A PHILOSOPHICAL STUDY (1969). For a representative sample of the Bratmanian approach, see Christopher Kutz, The Judicial Community, 11 PHIL. ISSUES 442 (2001); SCOTT SHAPIRO, Laws, Plans, and Practical Reason, 8 LEGAL THEORY 387 (2002); and JULES COLEMAN, PRACTICE OF PRINCIPLE (2001). For Bratman's views, see MICHAEL BRATMAN, FACES OF INTENTION (1999), chs. 5–8.
But there is an ambiguity here. For what are legal philosophers talking about when they say that the law is a social practice? Since the publication of H.L.A. Hart's The Concept of Law, contemporary positivists are often just talking about the rule of recognition.3
H.L.A. HART, THE CONCEPT OF LAW (Raz and Bullock, eds., 2d ed., 1994).
See RONALD DWORKIN, Model of Rules I, in TAKING RIGHTS SERIOUSLY 14 (1977); and Model of Rules II, in id., 46. See also RONALD DWORKIN, LAW'S EMPIRE (1986). Postema and Coleman, in their groundbreaking articles (Postema, CONVENTION AND COORDINATION AT THE FOUNDATION OF LAW, supra note 2; and Coleman, Negative and Positive Positivism, supra note 2) that established this approach to analyzing the way in which the rule of recognition is a social rule, were responding to some of Dworkin's objections to Hart. There have been too many epicycles of this debate to cite all the relevant literature. I do not mean to suggest that Razian objections to Hart have not played a role in shaping how people have thought about the rule of recognition as a social practice. But by a wide margin, Dworkin's objections and the objections of those who are sympathetic to Dworkin's position—whatever the merit of these objections—have been the primary source of the hurdles philosophers of law have had to clear in their reflections about how to conceive of the social practice of the rule of recognition.
With one or two exceptions, those who have attempted to develop a theory of how the rule of recognition is a Lewisean convention or a Bratmanian shared activity have abandoned this project.5
The most important exception is MARMOR, supra, note 2, who defends a modified Lewisean conventionalist account of the rule of recognition.
See Shapiro, Law, Plans, and Practical Reason, supra note 2, in which Shapiro argues for a Bratmanian theory of legal authority; and Shapiro, Legal Practice as Massively Shared Agency, supra note 1, in which Shapiro argues for a novel theory of legal institutions as social practices. In COLEMAN, PRACTICE OF PRINCIPLE, supra note 2, Coleman's suggestion that we might analyze the rule of recognition as a Bratmanian shared activity turns out to be a suggestion that we analyze legal systems as Bratmanian social practices (I say more about this below). Gerald Postema now focuses his attention on the overall practice of the law and not just the rule of recognition. See Gerald Postema, Law's Melody, 7 ASSOCIATIONS 227 (2003); and Melody and Law's Mindfulness of Time, 17 RATIO JURIS 203 (2004). See also Kutz, supra note 2.
When exploring this issue, one must be careful to distinguish claims about the social practices at the foundations of law based upon purported conceptual truths about law (e.g., Raz argues that a conceptual truth about the law is that it claims authority and so the social practices at the foundation of law must be ones that somehow allow for such a claim) and claims about social practices at the foundations of law based upon claims about law that are not conceptual truths (e.g., there is often disagreement among legal officials about the criteria of legal validity and so the social practices at the foundation of law must not be fatally disrupted by such disagreement).7
Conversations with Jules Coleman have helped me to see the wide-ranging significance of this point.
Shelly Kagan first used this term in a Yale Law School seminar in the Spring 2006 semester when discussing Shapiro, Legal Practice and Massively Shared Agency, supra note 1.
Two other important members are Margaret Gilbert's and Raimo Tuomela's theories of social practices. Gilbert's main work is MARGARET GILBERT, ON SOCIAL FACTS (1989). She discusses Hart at length in GILBERT, SOCIALITY AND RESPONSIBILITY (2000), ch. 5. A relevant sample of Tuomela's view is Raimo Tuomela & Kaarlo Miller, We-Intentions, 53 PHIL. STUD. 367 (1988). Neither Gilbert's nor Tuomela's views have been incorporated into Anglo-American jurisprudence to the degree to which Lewis's and Bratman's have been.
To be clear, my argument does not rest on the claim either that the law in some community is identical to a social practice or that the law is identical to some set of legal institutions.
My conclusion may be reminiscent of Dworkin's criticisms of Hartian positivism (and his criticisms of positivism in general). In brief, Dworkin argues that there is far too much disagreement in the practice of law for the foundations of law to be conventional social norms. My line of argument, although focused on disagreement, is nonetheless distinct from Dworkin's and therefore does not suffer from some of the well-known problems from which his suffer.11
Dworkin's misreading of Hart in Dworkin, Model of Rules II, supra note 4, is well known. For a recent and relevant criticism of the semantic sting argument, see Kenneth Einar Himma, Ambiguously Stung: Dworkin's Semantic Sting Reconfigured, 8 LEGAL THEORY 145 (2002).
Before diving into the body of the discussion, I shall introduce an important caveat. Nothing I say in this paper is meant to be an objection to Lewis's theory of conventions or Bratman's theory of shared activities. My only goal is to raise worries about analyzing the social practices at the foundations of law using the Lewisean or Bratmanian frameworks. Neither Lewis nor Bratman ever argue that all social practices should be analyzed according to their respective models. In Lewis's case, he is interested entirely in conventions understood as solutions to iterated coordination problems and he makes clear that he does not take every social practice to be a solution to an iterated coordination problem.12
See LEWIS, supra note 2, ch. 1.
Although he suggests that his analysis can be extended to cover large-scale social practices when he writes in Shared Cooperative Activity in BRATMAN, supra note 2, at 94:
Such shared cooperative activities can involve large numbers of participating agents and can take place within a complex institutional framework—consider the activities of a symphony orchestra following its conductor. But to keep things simple I will focus here on shared cooperative activities that involve only a pair of participating agents and are not the activities of complex institutions with structures of authority.
See also I Intend that We J in BRATMAN, supra note 2, at 144, in which Bratman asks whether his account of shared agency picks out just one species of shared agency from a broader genus of shared agency. He says nothing of institutions or large-scale social practices.
The paper proceeds as follows. In Section II, I give a formal account of what we mean by the claim that the law is a social practice. In Section III, drawing on Joseph Raz's work, I give a very brief and preliminary formal explanation of what people in legal systems do. Sections IV and V are the heart of the paper. In these sections, I give an overview of the Bratmanian theory of shared activity, I give an example of its use as a framework for an analysis of legal institutions, I identify five defining features of the Bratmanian framework, and finally, I explain why the Bratmanian analysis of shared activities is a poor framework for analyzing modern legal systems.
SOCIAL PRACTICES
When we say that the law is a social practice, we are making two claims. First, we are saying that the law is to be understood in terms of something that people do. That is what we mean when we say that the law is a social practice. Second, we are saying that the law is to be understood in terms of something that people do together. That is what we mean when we say that the law is a social practice. So what we need to explain when we say that the law is a social practice is what it is that people in legal systems do together and how it is that they do it together.
In a trivial way, everyone who is alive, insofar as they are doing something, is doing something together, namely being alive. But that just reflects an ambiguity in the term “together.” Sometimes (but not all the time) we say people are doing something together if we use the same term to describe what it is that they are doing and they are doing that thing contemporaneously. So everyone in New York City who is at this moment riding the subway is riding the subway together. This is very loose usage and applies just as well to animals as it does to persons: cows graze in a pasture together, and alligators sun on the riverbank together. The stricter sense in which we use the term “together” to describe people's activity is when there is some kind of systematic unity to the activity, for example as when friends share a meal together or musicians play music in a band or orchestra together.14
Some of these examples are from Postema, Law's Melody, supra note 6, at 227.
When philosophers try to explain social practices, most of their energy is spent on trying to explain what is going on when we do things together in the stricter sense I describe above—the sense in which there is a systematic unity of the activity. It is fair to say, then, that the philosophy of social practices is primarily an inquiry into the systematic unity of social practices. This is what Bratman seeks to explain when he gives an analysis of joint intentions and shared cooperative activity.15
For example, Bratman writes in Shared Intention in BRATMAN, supra note 2, at 110: “Supposing, for example, that you and I have a shared intention to paint the house together, I want to know in what that shared intention consists.” Bratman aims to explicate what makes the shared intention shared, i.e., what gives the systematic unity characteristic of sharing to a collection of intentions held by individual persons.
On the other hand, what it is that we do together (as opposed to what it is to do something together) when we do things together is not really in its own right a subject of philosophical analysis. For there are innumerably many and heterogeneous things we can do together, from painting a house to participating in a massive legal system. The only philosophically interesting generalizations we can make about the class of things we can do together will be made when working out what it is to do something together. Nonetheless, it is important when talking about a specific institution to spell out what people are doing together in that institution so that we have an idea of where to look for the systematic unity.16
For example, suppose we are trying to explain the systematic unity of the institution of baseball teams (as a class of social practice). It is important to specify what the activities of baseball teams are when explaining the systematic unity of baseball teams. For a baseball team may also regularly have an NCAA March Madness basketball pool. But this is not part of the social practice that is part of the social practice constituting baseball teams, because regularly engaging in basketball pools is not a standard thing people do when they are on baseball teams together (even though basketball pools may also be a social practice). Citing participation in the basketball pool as part of a general account of the institution of a baseball team would therefore be an error (even if it might be relevant for a more specific historian's account of the institution that is some particular baseball team).
So when we say that the foundations of law are social practices, we are saying something like the following: there are foundational components of the law that should be understood as systematically unified activities. Perhaps the rule of recognition is a legal entity that should be understood as a social practice. Certainly, legislatures, courts and administrative agencies such as police departments and prisons are, for the most part, social practices and so are systematically unified activities. We have no single term that comfortably refers to all and only these components of the law that are social practices. Nonetheless, I shall use the term “legal institution” to refer to those features of the law that the positivist takes to be susceptible to analysis as a social practice. This is an imperfect solution, because there are some legal institutions that are not best understood as social practices and there are features of the law that may be analyzable as social practices but that are not best understood as institutions. But it will do for my purposes. An analysis of legal institutions as social practices therefore involves giving an account of both the activities constitutive of the institution and the systematic unity of those constitutive activities.
WHAT PEOPLE DO IN LEGAL INSTITUTIONS
Drawing heavily on Joseph Raz's formal discussion of institutionalized systems of norms, I stipulate here a simple, general account of what people do in legal institutions.17
See, esp., JOSEPH RAZ, PRACTICAL REASON AND NORMS (2d ed. 1990), at 123–148. See also Raz's criticisms of Kelsen in RAZ, THE CONCEPT OF A LEGAL SYSTEM (2d. ed., 1980), at 95–109; and his discussion of the relation of law and state in RAZ, THE AUTHORITY OF LAW (1979), at 97–102. For the law claiming legitimate authority, see RAZ, AUTHORITY OF LAW, ch. 1.
Legal institutions are, broadly speaking, relatively stable integrated social practices that make possible the creation and/or application (i.e., the production, interpretation, and/or enforcement) by third-party formal agents of rules governing a bounded domain of agents not entirely coextensive with the domain of third-party formal agents. Thus institutions are stable social practices that (a) have some kind of internal, self-sustaining structure that ensures stability and integration; (b) provide guidance to those governed by the institution by the production, propagation, and/or enforcement of rules; (c) include formal agents whose practices sustain the institution and apply the rules of the institution.18
This view is designed to rule out as institutions informal conventional norms such as conversational norms and governance by consensus of the sort Quaker communities seek to achieve. On conversational norms, see ERVING GOFFMANN, RELATIONS IN PUBLIC (1980).
Following Joseph Raz, I leave aside for another day spelling out the identifying characteristics of officials. In particular, the most difficult thing to spell out is what it is for an official to be a formal official as opposed to an informal official. This is a very difficult question to answer and one that I need not settle now, given the purposes of this paper (although I have something to say about this below). See Raz, PRACTICAL REASON, supra note 17, at 133. I later include as officials of legal institutions practicing private attorneys.
Once again, I follow Raz here. See id.
I take primary rules not only to be simple rules directed at individual agents but also rules that set priorities and targets, allocate resources, and stipulate long-term plans governing those who engage in some practice. Here I am following Edward Rubin in his discussion of administrative agencies. See, e.g., Edward Rubin, Law and Legislation in the Administrative State 89 COLUM. L. REV. 369 (1989); and Edward Rubin, It's Time to Make the Administrative Procedure Act Administrative, 89 CORNELL L. REV. 96 (2002).
RAZ, AUTHORITY OF LAW, supra note 17, at 132–137.
We can summarize, in dirty simplicity, the activity that is characteristic of legal institutions as follows: officials create and apply primary rules. So when a positivist seeks to explain how legal institutions are social practices, he seeks to explain how officials create and apply laws together. Or, to put it in more complex terms: the positivist seeks to explain how it is that a collection of very many and very different people can create and apply laws in a manner that displays the systematic unity necessary for those activities of creating and applying laws to constitute the social practice of a legal institution. This is where Bratman's theory of joint intentional and shared cooperative activities comes in.
THE BRATMANIAN FRAMEWORK
There is more than one theory of social practices. But some very important Anglo-American legal philosophers have appealed to either the Lewisean theory of conventions or the Bratmanian theory of shared activity as preferred theories of social practices.23
See references at note 2.
Overview
Bratman identifies the following two conditions that must be met by agents in order for their activities to have the kind of systematic unity that a shared activity has:
- Mutual responsiveness: … each participating agent attempts to be responsive to the intentions and actions of the other, knowing that the other is attempting to be similarly responsive. Each seeks to guide his behavior with an eye to the behavior of the other, knowing that the other seeks to do likewise.
- Commitment to joint activity: … the participants each have an appropriate commitment (though perhaps for different reasons) to the joint activity, and their mutual responsiveness is in the pursuit of this commitment.24
See Bratman, Shared Cooperative Activity, in BRATMAN, supra note 2, at 94–95.
If several persons' activity displays these features, then they are engaged in what Bratman calls “joint intentional action.”25
Bratman names actions that meet these two conditions jointly intentional actions (JIAs) in id. at 104. Some of the caveats that apply to SCAs, though, surely apply to JIAs, such as, for example, the caveat that rules out what Bratman calls “pre-packaged cooperation” in id. at 106.
This is a weak form of shared activity that does not rise to the level of cooperation. To see why, imagine two line cooks working in a professional kitchen. Each wants to see the other fail in front of the chef so that she can get a promotion, but neither can succeed on her own, because cooking dinner in a professional kitchen requires at least two people working together. So each has to play her part as long as the other plays her part (lest the chef sees one failing to play her part and then fires her on the spot), but neither will help the other out if the other fails to do her part. The two competitive line cooks in this example are not being cooperative, because in cooperative activity, which Bratman calls “shared cooperative activity,” there is a commitment to mutual support. Bratman defines the condition in this way:
- Commitment to mutual support: … each agent is committed to supporting the efforts of the other to play her role in the joint activity. If I believe that you need my help to find your note [if singing a duet together] (or your paintbrush [if painting a house together]) I am prepared to provide such help; and you are similarly prepared to support me in my role. These commitments to each other put us in a position to perform the joint activity successfully even if we each need help in certain ways.26
Id.
When the first two—and better yet, all three—of these conditions are met with respect to some action, then the action will have the systematic unity of a social practice.
So how is it that at least two, if not all three, of these conditions can be met with respect to some action, J? Bratman explains how these three conditions can be met by developing a theory of shared intentions. His view is that it is in virtue of agents sharing intentions that an activity in which they are engaged is a shared activity. So Bratman explains the systematic unity of shared activity by appeal to shared intentions. Shared intentions, which are a state of affairs in which individuals' intentional states are interrelated in a certain way, have three roles. First, they coordinate individual intention action so as to achieve the aim of the shared intention. Second, shared intentions regulate our “subplans” so that they do not conflict (or, in Bratman's terms, so that they mesh). Third, shared intentions are the backdrop against which bargaining about how to achieve the shared end proceeds.
What exactly are these shared intentions that make an activity a shared activity? Bratman is very careful to analyze shared intentions to J by way of meshing subplans in what he calls a “cooperatively neutral” manner. The reasons he does this are first, that Bratman rejects the possibility of a superagent intending to J; and second, that Bratman notes that it would be circular to explain shared agency by appeal simply to agents intending to cooperate. Instead, Bratman explains what a shared intention to J is in terms of each agent intending that all relevant agents achieve the same end (J-ing) by way of each acting in accordance with her own subplan that meshes with the subplans of the relevant other agents.27
Bratman says a great deal more about this in Shared Intention and I Intend that We J, in BRATMAN, supra note 2. None of this is relevant to the objections I raise, although I will likely remind some readers of the objections Bratman addresses in these two articles.
What makes shared intentions shared is that they are coreferential and interreferential: the intentions reference the same end (and so are coreferential) and the intentions reference each other (and so are interreferential).28
This is what sets Bratman's shared intentions apart from mere we-intentions (as Searle calls them). We-intentions can be a single person's intention that we J without the other people in the extension of the “we” having that intention as well. Bratman's shared intentions are knit together by their co- and interreferentiality. On we-intentions, see JOHN SEARLE, Collective Intentions and Actions, in INTENTIONS IN COMMUNICATION (Cohen, Morgan, & Pollack eds., 1990), at 401–415.
In Shared Intention, in BRATMAN, supra note 2, at 123, Bratman writes: “[Shared intention] is a state of affairs that consists primarily in attitudes (none of which are themselves the shared intention) of the participants and interrelations between those attitudes.”
Bratman writes, in id. at 112: “Our shared intention, then, performs at least three interrelated jobs: It helps coordinate our intentional actions; it helps coordinate our planning; and it can structure relevant bargaining.”
Here is Bratman's formalized analysis of the attitudes essential to a shared activity (SCA):
- (1)(a)(i) I intend that we J.
- (1)(a)(ii) I intend that we J in accordance with and because of meshing subplans of (1)(a)(i) and (1)(b)(i).
- (1)(b)(i) You intend that we J.
- (1)(b)(ii) You intend that we J in accordance with and because of meshing subplans of (1)(a)(i) and (1)(b)(i).
- (1)(c) The intentions in (1)(a) and in (1)(b) are not coerced by the other participant.
- (1)(d) The intentions in (1)(a) and (1)(b) are minimally cooperatively stable.31
This is the condition ensuring commitment to mutual support.
- (2) It is common knowledge between us that (1).32
Bratman, Shared Cooperative Activity, in BRATMAN, supra note 2, 105.
Bratman, acknowledging that the systematicity of the intentions is what provides the systematic unity of the shared activity, writes:
It is the web of intentions cited in (1) that ensures the commitments to the joint activity characteristic of SCA.33
Id. at 102 (emphasis added). The point Bratman is making in this brief passage is not what ensures commitment to some action but what ensures commitment to a particular kind of action, namely, a shared activity. What ensures commitment is a matter of the reasons each agent takes herself to have for performing the shared activity, and this is not something with which Bratman is concerned (or could possibly have much to say).
To complete the overview of Bratman's position, let me state his definition of shared cooperative activity (in id. at 106):
For a cooperatively neutral J, our J-ing is an SCA if:
(A) we J;
(B) we have the attitudes specified in (1) and (2); and
(C) leads us to (A) by way of mutual responsiveness (in the pursuit of our J-ing) of intention and in action.
Conditions (A) and (B) are not relevant for my purposes because (A) adds only a success condition and (B) is already contained in the text accompanying footnote 32. (C) is of some interest because it makes explicit the causal connection between each agent's beliefs and intentions and her J-ing.
Consider a variation on Searle's example in which people sitting on the grass in the park suddenly get caught in a downpour and so all jump up and run for shelter. In the variation, each intends to run for shelter, each sees every other person running for shelter, and each correctly believes that every other person has the intention of running for shelter (and that every person believes that every other person has the intention of running for shelter). So the common-knowledge requirement is easily met, but this is quite clearly not a shared activity. See Searle, supra note 28.
Legal Institutions as Bratmanian Shared Activities
I will now explore one concrete example of a Bratmanian analysis of legal institutions in order to illustrate how Bratman's theory is used to analyze the way in which legal institution are social practices. The example is Jules Coleman's account of legal institutions as Bratmanian shared cooperative activities.36
In COLEMAN, PRACTICE OF PRINCIPLE, supra note 2, at 95–102. Coleman has since abandoned this account, but not on the grounds I present here. I am avoiding Shapiro's account of legal authority because it is too involved to spell out and I need only a sketch to serve as an example.
Here is the most complete statement by Coleman of his view that the rule of recognition is a shared cooperative activity:
Judges coordinate their behavior with one another through, for example, practices of precedent, which are ways in which they are responsive to the intentions of one another. The intention of an appellate court is that its decisions be binding on lower courts. Lower-court judges typically respond to these intentions by treating higher-court judges' decisions as constraints on their own behavior. The best explanation of judges' responsiveness to one another is their commitment to the goal of making possible the existence of a durable legal practice (though judges may have different reasons for thinking that a durable, sustained legal practice is desirable). Abiding by a practice of precedent is one way in which each judge helps the other do his part in fulfilling the aims of a legal practice.37
Id. at 97.
Coleman also describes the aim of the officials as the creation and sustenance of law:
The practice of officials necessary to create and sustain law is a more general form of social coordination [than a Nash equilibrium solution to a game of partial conflict], a form that is otherwise familiar to us. Bratman has a plausible and attractive account of such practices and of their possibility conditions. (Id.)
So we have established who is engaged in the shared activity and what the shared activity is. What about the interlocking intentions? Let us begin with the coreferring intentions to J. Coleman carefully follows Bratman here when he points out that judges must explicitly share the intention to J together, which in this case must be an explicit intention to apply primary rules together: “… when [judges] participate jointly in SCA, [they] must share an intention that converges on a common goal—even if [their] reasons or motives for doing so are importantly different.”39
Id. I have replaced the first-person plural in the original text with the term “judges” and the third-person plural because I take Coleman to be giving an analysis of how judges engage in shared cooperative activity.
But merely having coreferential intentions to J together is not sufficient; the agents must also have interreferential intentions, in particular, intentions to J together by way of meshing subplans. For example, if you and I intend to wash dishes together, we must have meshing subplans so that we can wash dishes together as opposed to the activity of washing dishes merely at the same time and in the same place. For example, suppose I have the subplan of soaping and rinsing and you have the subplan of drying (as opposed to both of us having the subplan of drying the dishes and neither having the subplan of soaping and rinsing the dishes). In order for us to wash dishes together, each of us must know of the other's subplan that is part of our larger shared plan to wash dishes together. That is, each of us must have intentions that successfully refer, somehow, to the other's intentions to wash the dishes. How, then, do the judges successfully form these sorts of intentions? According to Coleman, judges' subplans mesh when their decisions in the cases they hear and the published opinions explaining these decisions mesh in the appropriate manner. Judges' subplans are therefore expressed through the making of decisions and the production and publication of opinions, which in turn become precedent.
But, as noted above, it is not enough for shared activity that subplans mesh but that they mesh because of the general intention to J together (the judges' decisions might end up meshing only accidentally). For example, if we are going to wash dishes together, then it is not enough if it merely luckily turns out to be the case that I soap and rinse the dishes and then you dry them. Rather, I must intend to soap and rinse the dishes in accordance with your intention that we wash the dishes together and in accordance with your subplan to dry the dishes; and you must intend to dry the dishes in accordance with my intention that we wash dishes together and in accordance with my subplan to soap and rinse the dishes. Similarly, the decisions produced by judges must be, in some sense, components of the overall plan to apply primary rules together. So each judge issues a decision (i.e., intends to do her part in the shared activity of applying laws) in accordance with all other judges' intentions that all judges apply laws and in accordance with all other judges' decisions. The meshing of judicial decisions with precedent is what it is for the subplans of the agents who are officials in a legal institution to mesh.
Finally, Coleman stresses how the judges' shared intention to apply rules provides a backdrop against which bargaining about how to achieve this shared end proceeds. This is how Coleman proposes to explain how there can be disputes about the content of the rule of recognition without such disputes undermining the shared activity that constitutes the legal institution.40
See id. at 97–98.
Thus Coleman employs the Bratmanian theory of shared activities to explain how legal institutions are social practices. He argues that judges have a certain kind of shared intention that knits together their activities into a shared action. As a result, their activities as judges have a systematic unity, and it is on the basis of this that the legal institution of which the judges are a part is a social practice.
In this section I summarize one way in which the Bratmanian framework has been used to explicate the way in which legal institutions are social practices. In the next section, I identify five core features that I believe this kind of analysis possesses. I then argue in the final section that analyses of social practices that are hypercommittal, such as Bratman's analysis of shared activity, which are used as frameworks for explaining how legal institutions are social practices in the way in which Coleman has done, are inadequate for the task.
Analysis
I focus on five features of Bratman's view of shared activities.41
I think that Bratman's analysis of shared intention has these five features because his analysis of shared intention is meant to mirror his analysis of an individual's intention. Bratman writes: “Just as an individual's intention helps to ‘organize and unify her individual agency over time,’ shared intention helps ‘to organize and to unify our intentional agency.’” Shared Intention, in BRATMAN, supra note 2, at 112.
By modeling shared agency on individual agency, then, Bratman presumes a level of semantic and epistemic transparency that one finds in individuals but which may not be the norm in larger groups.
Lewis's theory of conventions is the most well known and has among the strongest common-knowledge requirements. See LEWIS, supra note 2, at 52–68. But for comments on the demands of the common-knowledge requirement, see note 35.
Before leaping in, I must note an important distinction between beliefs and intentions. Beliefs and intentions are distinct from one another primarily because they have different directions-of-fit (briefly put, beliefs aim at fitting the world as it is, whereas intentions aim at making the world fit to the content of the intention). If the direction-of-fit of a propositional attitude is a significant factor with regard to the content of the propositional attitude, those who wish to explain systematic unity of social practices in terms of the inter- and coreferentiality of the content of intentions owe us a more robust semantics of intentions and, in particular, a sketch of a theory of how intentions refer in the manner that facilitates co- and interreferentiality.43
The semantics of intentions that Bratman offers us is primarily a set of possibility conditions restricting the domain of possible contents of intentions. See I Intend that We J in BRATMAN, supra note 2, at 142–161. See also DONALD DAVIDSON, Intending, in ESSAYS ON ACTIONS AND EVENTS (1980), at 83–102.
The first characteristic feature of a Bratmanian analysis of social practices is conceptual agreement of the participating agents, by which I mean both extensional and intensional agreement of the agents' relevant beliefs and intentions.44
A related issue is discussed in much greater depth in Jules Coleman & Ori Simchen, “Law,” 9 LEGAL THEORY 1 (2003).
For the J-ing that concerns us happens in the actual world and not in some merely possible world.
Another way to say this is to say that our intentions are not mediated entirely by the action we intend to perform but are in fact more directly related in virtue of the fact that they refer to each other. This makes sense because it is not entirely clear how our intentions can be connected solely in virtue of their object being an action that has not yet realized. Bratman writes: “In SCA each agent intends that the group perform the joint action in accordance with and because of meshing subplans of each participating agent's intention that the group so act.” Shared Cooperative Activity, in BRATMAN, supra note 2, at 100.
So this is just another way to put the old Fregean point that within de dicto contexts there cannot be substitution salva veritate by merely coreferring terms.
Arguably, intensional equivalence is more important, because Bratman puts more weight on responsiveness to the intentions of others than on responsiveness to the actions of others.
For example, suppose you say to me while we are standing on a basketball court and you are holding a basketball, “Let's play some ball.” I respond, “Okay, let's play some ball.” Now, it turns out that by “play some ball” you mean “practice one-on-one drives to the basket” and I mean “play pickup basketball.” So although at first blush our intentions mesh—we both have the intention that we play some ball—upon closer inspection, it looks like you actually have the intention that we practice one-on-one drives to the basket and I actually have the intention that we play pickup basketball.49
It is worth noting here that we may never discover that we had different intentions because, as there are only two of us, practicing one-on-one drives and playing pickup basketball end up getting realized in more or less the same way.
In fact, they are not even extensionally equivalent. For each of us is playing a different game because the rules governing each of our behaviors are different. You are playing by the rules governing practicing one-on-one drives to the basket, and I am playing by the rules governing pickup basketball. So each of us is doing something different from the other even though it looks as though we are playing the same game. This is why I introduce the neologism “implementation conditions” instead of extension equivalence. I thank George Bealer for very illuminating discussions on this point.
There is an even deeper problem faced by any account that is based upon the coreferentiality intentions of the agents. Let an intention be a relation, I, between an agent, a, and an action, r: aIr. Because intentions are future-directed, r cannot be a past or existing action. One cannot intend to do what one is already doing or what one has already done (although one can intend to continue doing what one is already doing). This has the following consequence: intentions cannot, strictly speaking, corefer. A solution is to say that r is an act-type, and so intentions corefer by referring to the same act-type. This solution, though, drags us into a new nest of problems in philosophy of language and metaphysics because act-types are presumably abstract objects.
Now this problem of conceptual disagreement can easily be rectified through limited communication. But this just makes clear that full conceptual agreement requires some sort of commitment to conceptual agreement. Parties who seek to share agency in the Bratmanian fashion must be committed both to establishing conceptual agreement and then, once it is established, to sustaining that conceptual agreement. Absent such commitment, if the joint activity extends over a long enough period of time, there is a strong possibility that the parties engaged in the shared activity may slowly come to have differing understandings of what they are up to. This, in turn, can lead to a case in which the parties merely appear to share intentions, as in the playing of pickup basketball/practicing one-on-one drives case.
This suggests an additional problem. For the commitment to sustain conceptual agreement looks like another shared activity. This threatens to build into Bratman's account of shared activity an infinite regress. Someone might respond that this agreement need be neither fixed nor maintained through the intentional action of the cooperating agents. This would avoid the regress. This seems perfectly fine—it is more or less what happens in normal language use. One might further argue that the threat of deviation from conceptual agreement once it has been fixed is so limited that commitment is hardly necessary. For one might argue that the threat of “semantic drift” is so minimal that parties can lack commitment without conceptual divergence emerging. This seems a reasonable position to adopt, as well.52
It is worth noting that the Lewisean analysis of convention as found in LEWIS, supra note 2, is not an analysis of the unintended adoption of a convention. In fact, he has rather robust common-knowledge requirements, to which many objections have been lodged. See, esp., Tyler Burge, On Knowledge and Convention, 84 PHIL. REV. 249 (1975). Some of my objections to a Bratmanian analysis of legal institutions are the distant products of reflections inspired by Burge's objections to the Lewisean common-knowledge requirement.
The third important feature is what I shall call epistemic agreement. Bratman writes:
In SCA I will see each of the cooperators, including me, as participating, intentional agents. If this obliges me to include the efficacy of your intention and subplans in the content of my relevant intention, then it also obliges me to include the efficacy of my own intention and subplans in this content.53
Shared Cooperative Activity, in BRATMAN, supra note 2, at 100 (emphasis removed from “including me”).
See Shared Intention, in id. at 121ff.
- I intend that we paint.
- You intend that we paint.
- My intention is known to you, and yours is known to me.
- The persistence of (1a) depends on my continued knowledge of (1b): if I did not know that (1b) I would not intend that we paint.
- The persistence of (1b) depends on your continued knowledge of (1a): If you did not know that (1a) you would not intend that we paint.
- We will paint but only if (1a) and (1b).
- (1)–(4) are common knowledge between us.55
I Intend That We J, in id. at 153 (emphasis added).
So what is required for shared intention and shared action is not only that there is conceptual agreement with respect to concepts deployed with respect to the activity to be shared but that the agents have more or less correct beliefs about each other's subplans and intentions.56
Bratman notes, in id. at n. 14, that he has excluded mention of meshing subplans for ease of exposition. Bratman writes:
I ignore here the idea that in shared intentions we each intend that the activity proceed by way of meshing subplans of each of our intentions. This idea is central to my overall view, but can be safely put to one side in a discussion of the … objections [being dealt with in this article].
it seems reasonable to suppose that in shared intention the fact that each has the relevant attitudes is itself out in the open, is public.57
Shared Intention, in id. at 117.
As in the case of conceptual agreement, there also needs to be commitment to epistemic agreement. Consider conditions (3a) and (3b) above: the persistence of shared intention depends upon continued true belief by the relevant agents about each other's relevant intentions. This is especially significant because intentions and subplans can change in response to developments exogenous to the shared activity. For example, we could be painting a house together, and I feel what I take to be drops of rain. So I stop painting and begin to collect the brushes and paint. It turns out that what I felt was an errant sprinkler from next door. As a result, you keep painting while I am cleaning up (because you did not feel the errant sprinkler's spray). The systematic interconnection between our subplans and intentions has been broken. This can be addressed if you ask me what I am doing and why I am doing it and then I sincerely answer you. This would allow us to achieve epistemic agreement again. But it requires a commitment on both our parts to do so. An alternative way to keep me painting would be for you to threaten to beat me up unless I start painting again, but this would no longer be a shared intention (even if it might get the house painted).
Finally, it is true that in many cases epistemic agreement may be easy to achieve and that sustaining it may not be very difficult. In the painting case, as well as in the other small-scale cases Bratman considers, it is almost costless to ask someone a question and for that person to answer sincerely. But I argue below both that epistemic agreement is difficult to achieve in some large-scale contexts and that the commitment to epistemic agreement is difficult to sustain in large-scale contexts.
The fifth feature of Bratman's view is strong practical commitment to the shared activity, namely a commitment by each party to engage in the activity with the other parties and (if a shared cooperative activity) to being mutually supportive in the activity. It is not enough merely to have shared beliefs about each other's intentions and subplans; parties must also be practically committed to the shared activity and the subplans. Jon might believe that Phil intends to make a pizza with Jon, and Phil might believe that Jon intends to make a pizza with Phil. Furthermore, they might both intend that their subplans mesh so that they can make pizza together. Phil, though, might decide, just as they are about to begin the pizza-making, that he could be doing something better alone, for example, playing the guitar, and so announces to Jon (thereby maintaining epistemic agreement) that he intends to abandon the shared pizza-making endeavor. So Phil needs to have a strong practical commitment to making pizza together with Jon—strong enough that it will not dissipate when opportunities for other desirable activities come up.58
The source of this practical commitment cannot be coercion by the other agent involved in the cooperative activity. It remains an open question how much outside coercion can generate this practical commitment. For example, if Grant points a loaded pistol at Jon and Phil and says “Make a pizza together or die,” could Jon and Phil form the relevant joint intention and then perform the SCA of making pizza together (they will be mutually supportive, lest Grant shoots each of them)? I am not certain where Bratman would stand on this issue.
The requirement for strong practical commitment might seem extraneous. But that is because in small-scale coordinated activities of brief duration, such as making dinner together, driving to New York together (unless one is leaving from California), painting a house together, and playing a game together, practical commitment is cheap.59
Bratman writes, in Shared Intention, in id. at 114, that “the nonreconsideration of one's prior intentions will typically be the default.” This is probably true in almost all short-term shared activities. But it is another matter altogether in long-term activities. In such cases, we can no longer presume that the nonreconsideration of prior intentions is the default. It becomes a matter of trust. See Matthew Noah Smith, Trust and Social Norms (in progress).
In sum, the five key features of Bratmanian shared activity are the following: conceptual agreement, commitment to conceptual agreement, epistemic agreement, commitment to epistemic agreement, and strong practical commitment. Bratmanian shared activity, then, is hypercommittal: parties must seek agreement and be committed to sustaining it. In the next section I argue that Bratmanian shared activities are too hypercommittal to be an adequate model of the social practices that constitute legal institutions. To show that this is the case, I argue that each of the five characteristics outlined in this section is not likely to be realized in legal institutions.
ARE LEGAL INSTITUTIONS SHARED ACTIVITIES?
I begin first by reminding the reader that this section is not meant to be read as an attack on Bratman's theory of shared activity or shared intentions. I have no firm opinion about that theory. This section is, instead, the completion of an argument against the use of that theory as a framework for analyzing how the legal institutions at the foundations of law are social practices.
Second, let us recall an important feature of modern legal institutions. The agents whose activities constitute modern legal institutions are not only legislators and judges. In particular, modern legal institutions include many administrative agencies that have the authority to issue regulations that we have no reason not to take to be law (even if they are not legislation). The staffs of these agencies apply the regulations produced by these agencies, including settling questions of the content of the rules and settling disputes between parties about the application of the rules. There may be the option of appeal to the judiciary, but the existence of this option is not evidence that the administrative rules and the application of these rules should not be understood as law and the application of law, respectively. So officials in a modern legal institution are not only judges and legislators but also bureaucrats. Legal philosophers who wish to restrict legal institutions only to legislators producing legislation or judges applying legislation or judge-made common law bear the responsibility of argument here. It is worth noting, after all, that in The Concept of Law H.L.A. Hart never specifies who must adopt the internal point of view with respect to the rule of recognition, only that certain officials must if the rule of recognition is to exist in some population. So any theory of how legal institutions at the foundations of law are social practices must reckon with the fact that modern legal institutions are sprawling and involve individuals with heterogeneous educations and backgrounds. With this in mind, the problems facing an account of how legal institutions are hypercommittal social practices begin to emerge into something like stark relief. For such accounts presume that the officials of legal institutions at the foundations of law are hypercommitted to the activity of their respective institution. In this section, I show how this is implausible with respect to all five criteria of hypercommitment outlined above.
Turning first to conceptual agreement, we can note that in a legal institution, officials, together, are creating and/or applying laws governing private individuals. So the question is: Is there conceptual agreement among officials about the concept law and is there a commitment to conceptual agreement about the concept law?
Dworkin's criticisms of Hart in “Model of Rules II” and his “semantic sting” argument are supposed to show that there is always a fair amount of conceptual disagreement within an individual legal institution about the concept law.60
See DWORKIN, Model of Social Rules II, supra note 4; and DWORKIN, LAW'S EMPIRE, supra note 4. Actually in Model of Rules II Dworkin argues that there is disagreement about the grounds of law and then in LAW'S EMPIRE he argues that there is disagreement about the concept law itself.
JEREMY WALDRON, THE RIGHT TO PRIVATE PROPERTY (1989), at 51–52. The Gallie article is W.B. Gallie, Essentially Contested Concepts, 56 PROC. ARISTOTELIAN SOC'Y 167 (1956).
On the incidents of private property, see A.M. Honoré, Ownership, in OXFORD ESSAYS IN JURISPRUDENCE (A.G. Guest ed., 1961), at 160.
The way that Waldron spells out essentially contested concepts allows him to retain agreement as a background to disagreement. He does this by mobilizing the well-worn concept/conception distinction. There is a concept of private property about which there is agreement; the disagreement is about which competing conception of property is best. See WALDRON, supra note 61. I believe Waldron's account of the concept and conception of private property is incoherent. But even if it is not, it remains unclear whether the concept/conception distinction can be deployed to resolve the problem faced by the concept of law.
There are some additional concerns. Recent work by Joshua Knobe has revealed what has become called the “Knobe effect.”64
For an overview, see Joshua Knobe, The Concept of Intentional Action: A Case Study in the Uses of Folk Psychology, PHIL. STUD. (forthcoming).
See Shaun Nichols & Joseph Ulatowski, Intuitions and Individual Differences: The Knobe Effect Revisited, MIND AND LANGUAGE (forthcoming).
Knobe suggests that it is rather widespread. See Knobe, supra note 64.
It is not at all news that there is not univocality about the concept law.67
For more, see Coleman & Simchen, supra note 44.
The problem becomes even worse if either Dworkin or Waldron is correct. For if law is an essentially contested concept, then one cannot appeal to an unintentionally adopted convention that fully fixes the content of the concept law. Rather, the convention is that this concept is essentially contested and so is not fully shared. And because it is an essentially contested concept, there will be a great deal of pressure—in the form of political contestation—pushing officials away from conceptual agreement. Overcoming this would require substantial shared commitment to conceptual agreement. This threatens a Bratmanian account with a regress of iterated shared activities seeking to secure conceptual agreement.
Could hierarchical structures resolve this problem? It is unlikely. Within a large-scale institution, hierarchical structures can ensure at best only local agreement (e.g., a police officer learns from his commander what is and is not law). This is because the “semantic guidance”68
Apologies to Scott Shapiro, whose concepts of motivational and epistemic guidance I have just bastardized. See Scott Shapiro, On Hart's Way Out, 4 LEGAL THEORY 469 (1998).
One might reject the conceptual semantics I have employed, or one might take the requirement of conceptual agreement to be too strong, or one might reject my argument that this strong requirement will not be met in a legal institution.69
This last tactic is especially plausible if one restricts the members of the class of officials to only those with the appropriate degree from accredited American law schools, but I have already indicated how this tactic is multiply problematic.
Epistemic agreement, recall, is necessary because, in order for an agent to include the efficacy of another's intention and subplans in the content of her relevant intention, she needs to have a reasonably accurate belief about what the content of the other agent's intention and subplans is. Even if the other agent has not yet worked out his subplans and the two need to bargain over the subplans at a later date, at that later date each would still need to have reasonably accurate beliefs about the subplans the other is considering adopting.
In a large-scale legal institution, even if there is conceptual agreement among officials about the concept law, there may remain disagreement among officials about what it is that they are doing together. As I mention above, most large-scale legal institutions are composed of a large and heterogeneous group of officials. These officials may not have the same education, political commitments, and level of identification with the institution within which they work. Thus it is likely that there will be a nonnegligible diversity in the beliefs among officials about what it is that they are doing. Officials may disagree about what the J of the legal institution is, what everyone else takes that J to be, and what each other's subplans are. The possibility of this disagreement cannot be glibly ignored: any Bratmanian account of legal institutions must give us sense of the source of epistemic agreement.
There is an even more serious problem faced by a Bratmanian analysis of legal institutions. It is unlikely that all officials will know, much less have beliefs about, who all the other officials are. For example, suppose that unbeknownst to one group of officials, another group of participants, with whom the first group never directly works but whom the first group has met on several occasions, drops out of the legal institution (say that the group's department loses funding). The members of the first group of officials can intend that they J with those who dropped out but they cannot successfully J with them. (It does not matter what the members of the second group intend because they are no longer part of the institution.) Suppose officials in the first group never knew about the second group in the first place. The problem becomes even worse for the Bratmanian account. In this case the members of the two groups could not have even intended that they J together from the get-go, much less have actually shared an activity.70
This is surely one reason why Bratman insists on cases of shared intention with usually only two people.
One way to get around this would be to specify the extension of the “we” in “we will J” by the definite description “the people who will be J-ing with me” or a name “the Justice Department.” But this may be just the kind of question-begging Bratman wants to avoid by characterizing J-ing in a cooperatively neutral way. Additionally, some philosophers of language have argued that there would simply be a failure to intend to J with anyone in particular if one replaces the “we” in “we will J” with a reference-fixing definite description or a proper name.71
See, e.g., David Kaplan, Quantifying In, 19 SYNTHESE 178 (1968–1969), esp. at 201ff. Kaplan restricts the condition in which such substitution is legitimate to conditions in which the agent with the propositional attitude can supply a sufficiently vivid name:
The notion of a vivid name is intended to go to the purely internal aspects of individuation. Consider typical cases in which we would be likely to say that Ralph is acquainted with X. Then look only at the conglomeration of images, names, and partial descriptions which Ralph employs to bring X before his mind. Such a conglomeration, when suitably arranged and regimented, is what I call a vivid name. (Id. at 201.)
It is consistent with what I have argued so far that without sharing beliefs about the J of the institution, officials have shared beliefs both about some local shared activity and about each others' subplans for the completion of that activity. For example, lawyers in the Justice Department might share beliefs about each other's intentions to write a brief together and might share beliefs about each other's subplans. This will allow the lawyers to engage in the shared cooperative activity of writing a brief.72
See also Keith Donnellan, The Contingent A Priori and Rigid Designators, in CONTEMPORARY PERSPECTIVES IN THE PHILOSOPHY OF LANGUAGE (P. French, H. Uehling, & H. Wettstein eds., 1979). Or it could be a joint intentional action—it does not matter here because in a JIA the only things missing are the parties' commitments to mutual support.
The same problem that arose for the commitment to conceptual agreement arises for the commitment to epistemic agreement, namely the problem of separate and possibly dueling cantons of epistemic agreement with respect to the activity that constitutes the social practice of the institution. I will not recapitulate that discussion.
Finally, we can turn to the requirement for strong practical commitment to the shared activity that is characteristic of the institution.73
The discussion in this paragraph is much indebted to Scott Shapiro's discussion of alienation in Shapiro, supra note 1.
This will be especially clear if we imagine a case in which an official of an institution is offered by another institution a job that has better wages. This official may immediately leave her current institution and take up employment at the better-paying one. Or consider a case in which a costless opportunity for failing to perform the required activity comes up. The official motivated only by threats of sanctions or promise of wages will take that opportunity and not act. I see no way in which one could correctly claim that in these cases there is a strong practical commitment to the J-ing of the institution.
It is important that alienation from the institution is not a matter of the reasons for action that alienated agents take themselves to have. For the issue here is not the reasons for which someone forms a practical commitment but instead what it is to which the agent has a practical commitment. For the practical commitment of an agent, and not the reasons the agent takes herself to have, determines whether she contributes to the shared activity. In most ordinary cases, the same set of reasons warrants commitment to many different activities.75
What a set of reasons do is to rule out commitments. But only some reasons rule out all but one commitment (and usually these are cases of authoritative reasons, i.e., reasons that are both peremptory and content-independent).
Furthermore, there is psychological evidence that people do not commit themselves to actions for reasons that prior to commitment they take themselves to have. Instead, they generate reasons post hoc to justify their commitment. See, e.g., Jonathan Haidt, The Emotional Dog and Its Rational Tail 108 PSYCHOL. REV. 814 (2001). If shared activity is possible only in cases of explicit deliberative agency in which the agent reflects on all her reasons and then, based upon a careful consideration of all of them, identifies what it is to which the reasons recommend she ought to be committed, and she so commits herself, then shared activity will be quite a rare phenomenon. For a nice list of several forms of agency, from deliberative agency to automatistic agency, along with brief discussions about how rare fully deliberative agency is and the significance for responsibility attribution, see Neil Levy & Tim Bayne, Doing without Deliberation: Automatism, Automaticity and Moral Accountability, 16 INT'L REV. PSYCHIATRY (2004), at 209–215.
It must be the case that many officials of a legal institution can suffer alienation without the legal institution ceasing to be a social practice. This seems especially the case because there can be many sources of this alienation, not least of which are beliefs that the institution is corrupt, beliefs that one is underappreciated, desires just to make it through until retirement, a desire to have the social capital that goes along with being an official of the legal institution, or just plain boredom. I suspect that alienation from legal institutions is far more common than its opposite, the happier identification with legal institutions. Any theory of how legal institutions are social practices that would fail to account for how legal institutions in which alienation is rampant are social practices is a weak theory.
In the section above, I identify five characteristic features of members of a significant class of theories of shared activity and claim that these requirements make these popular theories of shared activity hypercommittal. In this section, I argue that the activities of officials in modern legal institutions are likely not to display at least one, if not every one, of these five characteristics.77
A natural objection to this argument is that I am not giving a charitable reading of the Bratmanian position by taking it to require such high levels of conceptual univocality and such explicit beliefs and intentions. I grant that there may be an alternative reading of the Bratmanian position that does not require the tokening of explicit beliefs or intentions. On this reading, one takes Bratman's view to be purely dispositional. This would help, because dispositions are not propositional attitudes in the way that intentions are and so are not subject to the problems I highlight in this section. But I fail to see any reason to read Bratman as defending such a dispositionalist position. Bratman explicitly argues that systematicity of shared activity consists in the way in which propositional attitudes refer to one another and corefer to joint actions. It seems to me that to retreat to a dispositionalist reading of Bratman, Gilbert, Lewis, and the others of their ilk would amount to abandoning their positions without good reason. Hypercommittal shared activity is a real phenomenon, and Bratman and company have provided an analysis of it. Why junk their stated views because they do not successfully model large-scale, temporally extended social practices such as legal institutions?
CONCLUSION
I argue that there are some distinctive constraints on explaining how the foundations of law are social practices. In particular, I argue that legal positivists ought not to represent the foundations of law as constituted by hypercommittal social practices. This constraint on theorizing about the foundations of law is not a conceptual truth about law but is instead based upon unobjectionable observations about law in contemporary society. Furthermore, my criticisms are not based upon any conceptual claims about law that are at odds with the conceptual claims to which leading positivist theories of law are committed. So my criticisms are consistent with leading positivist accounts of the law, and my conclusions therefore apply to positivist accounts of law and legal authority.
My arguments, because they focus on disagreement, might appear to be in line with Dworkinian attacks on positivism. But, unlike Dworkin, I do not believe that we should abandon legal positivism. For the mysteries of how the law is a social practice are no different from the mysteries of how anything in our world is a social practice. In this sense, the problems I claim are faced by positivists are not unique; they are difficulties faced by all, from philosophers to sociologists, who seek to explain how marriage, etiquette, language, racism, major league baseball, the state, and so on are, to some extent, social practices. So my arguments should be no more than cold comfort for the critic of positivism.
