Groups, Normativity and Disagreement

Abstract

We are members of many groups to which we ascribe the performance of intentional actions, and belonging to these groups seems to give rise, in many cases, to special normative relations. For instance, as a member of the university I have certain duties, as a member of the football team I have others, and yet I have other duties qua member of my law firm. But what is special about some groups that claims of the form “I am under a duty qua member of the group” seem adequate? This paper claims that the standard answer to this question faces two main difficulties. Firstly, most accounts appeal to one special normative notion (e.g. the idea of a joint commitment, or an agreement) to explain such relations, a notion such that, if instantiated, it gives rise to duties that are independent of the value of the joint action. But there are cases where participants think that they are under a duty qua members because the joint activity is valuable, and only because it is valuable. Secondly, most accounts seem unable to explain disagreements among participants about the content of their duties. The paper proposes a model of group action and of normative relations among participants that attempts to overcome both difficulties.

1 Introduction

We are members of many groups to which we ascribe the performance of intentional actions, and some of these groups have a particular characteristic: belonging to them seems to give rise to special normative relations, for members believe that they have certain duties qua members. Thus, my university is involved, primarily, in the activity of teaching. And my university would be a completely different entity if its professors thought that they are at liberty to perform activities such as teaching, that they had no duty qua members to teach their subjects. I shall label this sort of groups “groups which act with a normative unity” (GNU). Schools, armies, religious orders, banks and, in general, most institutions are GNUs. Despite our familiarity with GNUs, they seem to resist explanation.

First, consider the general strategy to explain members’ beliefs in duties qua members. Most accounts appeal to one special normative notion such that, if instantiated, it gives rise to duties that are independent of the value of the joint action. Thus, Gilbert (1996, p. 8, 2002a, pp. 73–74) claims that participants are “jointly committed” to doing something. Given that the notion of a joint commitment is a normative, irreducible notion, that would explain beliefs in duties qua members. Tuomela (1995, chapter 3) claims that the relevant notion is that of an agreement to do one’s part of a joint action. Bratman (1999, p. 126) and Kutz (2000, p. 85) make similar suggestions. The general strategy is, nevertheless, subject to counterexamples. For, however one conceives of joint commitments or agreements, they create duties (if they do) that are partially independent of the value of the action which participants are jointly committed to pursuing, or have agreed to pursue. And one can easily think of cases where this is not so. Consider two individuals working together to rescue an individual from drowning. As participants in the joint activity of saving somebody’s life, they may well think of themselves as members of this small, sporadic group. And they may well think that they are under a duty qua members of such a group because the joint activity (saving somebody’s life) is valuable, and only because it is valuable. This same belief may occur in larger institutional groups, from charities devoted to fighting global poverty to pro bono consulting agencies.

Secondly, the general strategy seems unable to explain why participants disagree about the content of their duties. Suppose that a group of pro bono lawyers has provided legal advice to individuals for many years, and that now a non-governmental organization requests legal advice. Half of the lawyers argue that they should disregard this request, for their only duty qua members is to provide advice to individuals. But half of them claim that their duty qua members is, and has always been, to provide advice to organizations too; it so happens that, so far, there has been no opportunity to do so. And each side grounds its views by claiming that this is what their collective practice, properly understood, really requires. This sort of situation appears to be fairly common. And it poses a challenge for the general strategy. For, if participants had in effect become jointly committed to doing something together (or reached an agreement to do something together), it seems that they should have had the same idea of the content of their joint commitment (or the agreement) and hence of their duties. But participants disagree about this.

In this chapter I propose an account of GNUs that attempts to overcome both difficulties. I begin by suggesting a general model of collective intentional action where no normative relations among members exist. This is a starting point to provide an analysis of GNUs (Sect. 1). I then focus on Gilbert’s account and criticize its faults. If my contentions are correct, it follows that any account appealing to only one normative notion (such as the idea of a joint commitment, or an agreement) is unsatisfactory (Sect. 2). I then propose a model of GNUs. According to this model, GNUs may take on different normative structures (Sect. 3). When the group is normatively structured by an agreement, as happens frequently but not necessarily, disagreement can be explained if one considers how the content of voluntary undertakings is determined (Sect. 4).

2 Collective Intentional Action without Normative Unity

Many groups which act are not groups whose members believe that they have duties qua members. Two or more individuals may be seriously coerced to work together on (what they believe is) a morally repugnant joint activity, or pursue the achievement of a joint goal that is (considered) morally indifferent and act together out of pure personal interest. Members may act motivated by convenience, ambition, fear, or for other reasons, without believing that they have (or indeed without having) any duties qua members. I shall label this sort of groups “groups which act without normative unity”. A good way to provide an account of GNUs is, I believe, to focus on groups which act without normative unity in the first place. For they seem to be more simple than GNUs.

One way of explaining groups which act without normative unity is to examine the relevant intentions. For, as with individual actions, it seems that there is no collective action if the action is not jointly intended.

An account that focuses on the relevant intentions of collective action is Christopher Kutz’s, and I use it as a starting point.¹ Kutz (2000, pp. 89, 94, 103–104) claims that there is a collective or joint intentional activity if, and only if, there is a set of individuals who are acting with participatory intentions, a participatory intention being an intention to do one’s part of a collective or joint act.

Notice that, despite its attractive simplicity, the key notion of a participatory intention seems to make the model uninformatively circular. For it employs the same idea (the notion of a collective act) that the analysis claims to elucidate. Naturally, whether there is a circularity problem is debatable. But to remove any doubts on the matter, Kutz’s model can be easily modified to avoid the appearance of circularity. Consider the following argument.

Suppose I conceive of a particular state of affairs to be brought about: that a house be painted. I conceive of certain actions as standing in a sort of instrumental relation to this state of affairs. Say, getting the brushes and the paint, painting the front first, the back next, and so on, such that, if these actions are performed, the state of affairs will likely be brought about. Of course, the state of affairs might be brought about otherwise, by performing other actions. But this is how I conceive of the matter now. In other words, I conceive of a state of affairs the bringing about of which involves performing these actions. Suppose I decide to perform these actions myself. I plan in advance what to do: the first day I will paint the front, the second day the back, etc. I am essentially dividing labour among my inner-selves. Something similar might happen if I decide to hire two painters to paint the house. I divide the labour among them, and assign tasks to each such that, if my plan is followed, the state of affairs (that the house be painted) will likely be brought about. And a similar situation may occur if the agents are you and me. Just as I can divide labour among my inner-selves, and between the two painters, you and I can divide labour among ourselves. If we do this, each of us will see his own actions, and the actions of the other, as standing in a sort of instrumental relation to the state of affairs. So we can say that some collective ends are just states of affairs, the bringing about of which is conceived of as involving the actions of two or more individuals. These actions can be conceived of as the parts each individual is to perform in order to bring about the state of affairs.

This definition of some collective ends does not employ any notion of collectivity. True, there is some kind of coordination among the relevant individuals. But this sort of coordination does not involve collectivity in any sense. The definition only relies on the simple idea of an activity that can be divided in parts. Or, to be more precise, it simply relies on the idea of states of affairs, the bringing about of which is conceived of as involving the performance of certain acts by several individuals, an idea with which we are very familiar. And with this idea of collective ends we can characterize at least some participatory intentions. Some of them can be characterized as intentions to perform certain acts that the agent conceives as, together with the actions of other agents, bringing about a state of affairs, the realization of which involves his doing certain things and their doing certain things. I find no uninformative circularity involved here.

Not all participatory intentions can be characterized thus, however. Some collective ends are states of affairs which are constituted by the performance of certain acts by different individuals. Suppose an assembly wants to honour a guest by offering a toast. Assume that there is a common conception of what counts as ‘an assembly offering a toast’. Say, it consists of each member of the assembly, when prompted by one of the members, facing the guest and raising his or her glass of wine for a couple of seconds as a way of showing respect. This counts as the assembly offering a toast. So it is a state of affairs, the bringing about of which is seen as constituted by the actions (and attitudes) of different individuals. Notice the difference between the first type of collective act and this one. Here the actions do not stand, in any plausible sense, in an instrumental relation to the state of affairs. Besides, the state of affairs is not achievable in ways other than the individuals performing the relevant actions and displaying the relevant attitudes. So their intentions to do their part of their giving a toast (a collective action) are just intentions to perform certain actions (coupled with certain attitudes) that, together with the actions (and attitudes) of the others, are seen as constitutive of the bringing about of a particular state of affairs. Again, I find no uninformative circularity involved here.

These two characterizations of participatory intentions seem to cover all cases. We can use this argument to propose a model of collective intentional action that, purportedly, does not face the problem of uninformative circularity. My provisional suggestion is this:

There is a collective intentional activity with no normative unity if, and only if, there is a set of individuals (defined extensionally or intensionally) such that:

• (a) Each conceives of a state of affairs, the bringing about of which involves, or is constituted by, the performance of certain actions (and the display of certain attitudes) by all members of the set,

• (b) their conceptions of this state of affairs overlap,

• (c) each intends to perform these actions (and displays the relevant attitudes), and each conceives of these actions (and attitudes) as related in the way described to the state of affairs,

• (d) and each executes his or her intention, such that the state of affairs mentioned in (b) is brought about.²

Some comments and clarifications are in place. Firstly, the idea of overlap should be understood as requiring that there must be a non-empty intersection of the states of affairs that each participant has in mind. This notion is necessary in order to capture the fact that, for there to be a collective intentional action, it must be the same joint enterprise in which agents intentionally participate. So, for instance, I may intend that we go together to a friend’s house for a quiet dinner, while you intend that we go there for a surprise party. While our going to the surprise party is not jointly intentional, our going to our friend’s house is.³

Secondly, the model is minimalistic.⁴ By this I mean that it should be interpreted as proposing necessary conditions for there to be any instance of joint intentional action with no normative unity, and also as providing sufficient conditions for there to be the simplest type of instance. However, this does not deny that the model has to be supplemented by adding further conditions if it is to capture cases which are, in an intuitive sense, more complex. For instance, it is difficult to explain what takes place in many groups unless one supposes that each participant knows what the other intends, and that all this is common knowledge. But, again, this situation seems not necessary, as the example of the two painters above shows. And, besides, it can be captured by adding the relevant clause (one that requires the relevant beliefs) to the model. This is possible due to its minimalistic character.

Many other clarifications could be made, but the foregoing considerations should suffice. My only aim was to provide a plausible model of collective intentional action with no normative unity that could remove any doubt about the circularity problem and the model appears to fulfill such purpose. Let us focus now on GNUs.

3 Normative Relations among Members of Groups

As previously mentioned, on many occasions participants regard themselves as under a duty, qua members of the group, to perform the actions conducive to the joint end. This need not be the case, as the examples above show. But this might be, and perhaps normally is, the case. Given that the model I have just proposed is minimalistic, it could be further developed to capture these cases. Before doing so, however, I shall consider Gilbert’s account. It is one of the most elaborated models in this respect, and examining it will provide guidance in suggesting an alternative.

3.1 Gilbert’s Account

Gilbert’s account can be stated thus: two or more people are involved in the collective action of J-ing if, and only if, they are jointly committed to accepting the goal of J-ing as a body and each one is acting in a way appropriate to the achievement of that goal in the light of the fact that each is subject to the joint commitment (1996, p. 8, 2002a, p. 68, 2002b, pp. 73–74).

Consider the general notion of being jointly committed to accepting the goal of J-ing. The basic idea is that agents join forces toward the achievement of a goal by committing themselves to each other in a particular way, i.e. by becoming jointly committed. The main properties of a joint commitment are these: (a) the parties become individually committed through the joint commitment simultaneously, and these individual commitments are interdependent; thus, if Jack and Sue are on a walk together, it is because they have become individually and interdependently committed to doing what is necessary for them to walk together; (b) relevant entitlements and obligations will be in place; so, if Jack inadvertently draws ahead, he would accept Sue’s criticism for violating the joint commitment; (c) normally, the joint commitment is not rescindable unilaterally—thus, if Jack wishes not to walk together anymore, he will seek for Sue’s approval (Gilbert 2002a, pp. 77–79, 90–91). Notice that feature (b) would explain normative relations among members of groups.

3.2 Criticism of Gilbert’s Account. An Alternative Solution

Gilbert’s model entails, inter alia, that all groups which act are groups whose members think of themselves as being under a duty qua members. For, in her view, there is a group only if members are jointly committed and, by definition of being “jointly committed”, this entails that they are under a duty to act accordingly (as members of the group). Her definition, insofar as it attempts to be an analysis of our ordinary concept of group intentional action, is, it seems to me, incorrect. There are clear cases where it seems undeniable that there is a group acting intentionally and where members do not conceive of themselves as under a duty qua members, such as the case of the painters above.

We may consider then whether Gilbert’s account is an adequate analysis of GNUs only, and focus on joint commitments, the key notion of her analysis. If individuals are jointly committed, the following takes place by definition (see properties (a–c) of joint commitments above): several agents are each individually committed in a particular way; each commitment is interlocked with the others (the individual commitments are interdependent, are arrived at simultaneously, and cannot be rescinded without the concurrence of all); this gives rise to duties to act in accordance with the commitments, and these duties are independent of the value of the actions which participants are individually committed to doing, and in fact they are independent of the value of the joint action itself.

This does not seem to capture all cases. Recall the case of the painters, but suppose now that the two individuals want the house to be painted because it is going to be a rest-home for elderly people. They do consider themselves under a duty qua members now, for the joint activity is seen by them as something valuable in relation to individuals other than themselves.⁵ Does Gilbert’s account capture this sort of case?

The individuals intend to perform the relevant actions, and hence they are committed to doing them. The commitments are interdependent in some sense, for they all concern actions which, taken together, are related in a special way to a state of affairs (that the house should be painted). But these commitments need not have been arrived at simultaneously. In other words, joining in to this set of interdependent commitments might have taken place in other ways. For instance, one of the painters might intend to perform the relevant actions first, in the hope that the other will join him. These commitments need not be non-rescindable without the concurrence of the other painter either. For example, the joint action might be taking place and one of the painters might simply change his mind as to the valuable character of the activity and opt out. He does not need the concurrence of all to do this in any sense. The conditions put forward by Gilbert seem, then, too demanding. More importantly, Gilbert requires that participants think that they are under a duty regardless of the value of the collective action and, as we saw, this is not the case. Participants think of themselves as under a duty precisely because the joint action is valuable in relation to others. Moreover, Gilbert thinks that participants are actually obligated. But it is easy to think of collective actions where individuals believe that the activity is valuable and are completely wrong in so believing.

In short, there are groups whose members think of themselves as under a duty qua members because they think of the joint activity as particularly valuable in relation to other individuals. In those groups, which I shall label “GNUs of type (I)”, Gilbert’s main conditions are not met. So the idea of joint commitment is not necessary to understand these groups.

It seems, in fact, that one can propose an account of GNUs of type (I) by expanding the model I suggested in the previous section. Consider this proposal:

There is an intentional activity of a GNU of type (I) if, and only if, there is a set of individuals (defined intensionally or extensionally) such that:

• (a) each of them conceives of a state of affairs the bringing about of which involves, or is constituted by, the performance of certain actions and the display of certain attitudes by all members of the set;

• (b) their conceptions of this state of affairs overlap;

• (c) each intends to perform the relevant actions (and displays the relevant attitudes), and conceives of these actions (and attitudes) as related in the way described to the state of affairs;

• (d) each executes his intentions and, as a result, the state of affairs mentioned in (b) is being achieved;

• (e) each believes that the previous conditions are satisfied, and that the state of affairs being brought about is valuable in relation to individuals other than themselves;

• (f) each thinks that applying to them is a normative consideration according to which everyone who is in a position of, together with others, bringing about a state of affairs that is valuable for individuals other than themselves, should do his part.⁶

This model is just an expanded version of the model deployed in the previous section. It only contains some additions, which are highlighted, namely clauses (e) and (f).⁷ They are introduced to explain why participants believe that they are under a duty qua members. And the expanded model simply claims that they so believe because they think that a normative consideration demands that they do their parts because of the valuable character of the activity. This normative consideration appears to be quite abstract. But it is a plausible normative consideration. And it explains why members might think of themselves as under a duty qua members, i.e. as individuals who belong to the group (as individuals described in terms of clauses (a) to (d)) and to whom the normative consideration is applicable. The model is, as far as I can see, adequate to capture cases of GNUs of type (I).

We still need, nevertheless, an account of groups where participants conceive of themselves as under a duty qua members even if they do not think of the activity as particularly valuable in relation to other individuals. I label them “GNUs of type (II)”.

Gilbert’s account seems to be an inadequate analysis of these groups as well. The main difficulty is that we do not know exactly what a joint commitment is.⁸ Gilbert refuses to break down the notion, and this makes it unclear. This concern may be defused—as Gilbert herself suggests—by arguing that a joint commitment is just a particular set of individual commitments. Thus, if I intend to do A, I am individually committed to doing A. A joint commitment would be, then, a set of individual commitments but with the particular properties mentioned above: they are interdependent, arrived at simultaneously, non-rescindable unilaterally, and give rise to duties. But these notions are still mysterious. For even if individual commitments could become interlocked in the way described, one might sensibly ask why duties arise out of that mesh. Just as individual commitments do not create duties (for instance, if I intend to do A, it does not follow that I have a duty to do A), a meshing set of individual commitments does not create duties either. Gilbert’s explanation of why duties arise out of joint commitments is that this is so analytically. That is, because it is part of the concept of being jointly committed.

This idea appears inadequate. Notice, first, that certain restrictions should apply. If one is seriously coerced, no obligation should appear. But Gilbert claims explicitly that, even if somebody is forcing another to become jointly committed by putting a gun to his head, he becomes obligated (1996, pp. 351–52). This sounds extremely odd, to say the least. Explanations of duties must involve a normative argument, an argument that brings in normative considerations, of what is good, valuable, worthwhile, etc. Gilbert’s account is problematic because it is not of the relevant form. According to her, duties arise of necessity.

Suppose, nevertheless, that we accept that joint commitments create duties by definition. The idea would still be problematic. For a theory is not supposed to introduce new theoretical constructs unnecessarily. And all the main features of the phenomenon that the concept of a joint commitment is supposed to capture can be captured in other terms. That is, by employing the idea of an agreement to do one’s part of a joint act. Consider the following proposal:

There is a GNU of type (II) if, and only if, there is a set of individuals, defined intensionally or extensionally, such that:

• (a) each conceives of a state of affairs the bringing about of which involves, or is constituted by, the performance of certain actions (and the display of certain attitudes) by all the members of the set; the relevant actions are the actions which each has agreed (explicitly or implicitly) to perform;

• (b) each has an overlapping conception of the state of affairs;

• (c) each intends to perform the relevant actions (and displays the relevant attitudes), and conceives of these actions (and attitudes) as related in the way described to the state of affairs;

• (d) each executes his intentions and, as a result, the state of affairs mentioned in (ii) is brought about; and

• (e) each thinks that the previous conditions are satisfied.

This model needs to be completed and elaborated. But the point is that all of the main features of the phenomenon that Gilbert attempts to capture seem to be captured by employing the idea of agreements, a notion with which we are familiar, and dispensing with the idea of joint commitments, which is a new theoretical construct.

In effect, agreements, under any plausible construal, are ways of voluntarily undertaking obligations. In this sense the relevant obligations are created by the parties. Accordingly, participants can conceive of themselves as under an obligation qua members of the group, i.e. qua individuals who have agreed. When agreements are reached, the parties think that they have become obligated regardless of whether the thing one has agreed to is particularly valuable. The obligations are also thought of as arrived at simultaneously (when one agrees, no party becomes obligated first), and normally the agreement is not rescindable unilaterally. And since participants intend to fulfill the agreement, they are committed to performing the relevant actions. Agreements, finally, create duties because there is a normative principle according to which agreements should be kept and, arguably, this principle is valid for certain normative reasons.⁹ It is not my intention to discuss those reasons, although my view is that the principle is valid insofar as it gives assurance that one’s part will be performed, regardless of whether one thinks that the relevant action is convenient for oneself or not. And assurance is a valuable thing.¹⁰ Nevertheless, whatever your view about the reasons why agreements might bind, the point is that the idea of agreements, and of their normativity, is familiar to us. The notion of joint commitments is not.

Gilbert is aware of the parallel between agreements and joint commitments. But she rejects the idea that joint commitments can be replaced by the idea of agreements because, in her view, agreements themselves are instances of joint commitments. They are instances of being “jointly committed to upholding a decision as a body” (1996, pp. 292–96). Her rejection, nevertheless, brings us back to all the problems I have mentioned.

In short, Gilbert’s account is problematic. There are GNUs of type (I) where her conditions are not met. I have, in fact, proposed a model to capture these cases. Moreover, there are GNUs of type (II) where her conditions are not met either, and where the idea of a joint commitment could be replaced easily by the idea of an agreement.

3.3 Refining the Provisional Model

The provisional model of GNUs of type (II) I have proposed should, nevertheless, be revised. Agreements are voluntary undertakings. But they are not the only kind of voluntary undertaking. Vows, for instance, are voluntary undertakings that exhibit certain important differences with agreements. Inter alia, vows need not involve several parties (and hence obligations need not be arrived at simultaneously), and they need not be rescindable with the concurrence of others. Some GNUs of type (II)—think of gangs, and certain religious orders—are groups where members have made a vow, instead of having agreed to perform their parts. Moreover, agreements and vows are but examples of voluntary undertakings. There are others, such as making oaths, consenting, or promising.

To capture all these possibilities we could create a model of GNUs that incorporates the idea of voluntary undertakings. But the model would still be unsatisfactory. The normative consideration according to which voluntary undertakings should be honoured is “content-independent”. That is, there is no direct connection between the action which, according to the normative consideration, one ought to perform, and the particular value of the action considered alone.¹¹ And groups may be structured by content-independent normative considerations other than the voluntary-undertakings principle. For instance, participants might be doing something together because a legitimate authority has issued an order to that effect. And the normative principle according to which one must obey, within certain limits, legitimate authorities is, arguably, a content-independent normative consideration too.

I shall not propose any particular account of these moral principles. My only point is that they are plausible, and that, when applicable, they may give different normative structures to GNUs of type (II). Moreover, there might be other content-independent normative considerations, and one should leave this possibility open. Perhaps the best way to proceed, then, is to propose a very general and abstract characterization of the activities of GNUs of type (II). My suggestion is that there is such a group if, and only if, the following conditions are met:

There is a GNU of type (II) if, and only if there is a set of individuals (defined extensionally or intensionally) such that:

• (a) each conceives of a state of affairs the bringing about of which involves, or is constituted by, the performance of certain actions (and the display of certain attitudes) by all members of the set; the relevant actions are the actions which, together with certain facts (e.g. the fact that they are the actions which they have voluntarily undertaken the obligation to perform, or have been ordered to perform by a particular authority), appear in the antecedent of a content-independent normative consideration;

• (b) their conceptions of this state of affairs overlap;

• (c) each intends to perform the relevant actions (and displays the relevant attitudes), and conceives of these actions (and attitudes) as related in the way described to the state of affairs;

• (d) each executes his intentions and, as a result, the state of affairs mentioned in (b) obtains;

• (e) each thinks that the previous conditions obtain, and that the content-independent normative consideration mentioned in (a) is in effect applicable to them.

I think that this model has sufficient descriptive coverage. Many aspects of it should be discussed. But I would rather leave them open. My only point in suggesting the model was to propose an account of the normative relations among members of groups and, as far as I can see, the models of GNUs of type (I) and type (II) capture all the possibilities. GNUs may take on different normative structures and, accordingly, appealing to only one normative notion, such as the notion of a joint commitment, is unsatisfactory.

I have not considered other accounts of GNUs. Tuomela, for instance, proposes a very sophisticated model based on the idea of an agreement,¹² and other theorists make similar suggestions. But if my contentions above are correct, any model which appeals to the notion of an agreement to explain all possible normative relations among members of groups is unsatisfactory as well.

4 Disagreement about Duties

The reason why theorists focus on the idea of an agreement, or on normative notions which are very close to it (such as the idea of a joint commitment), is because many GNUs are groups structured by agreements. One difficulty with that strategy, nevertheless, is that members often disagree about the content of their duties, as in the example of the pro bono lawyers mentioned above. It seems that a model based on the idea of an agreement could not explain this kind of dispute. For, it could be claimed, the parties should have the same ideas as to what they are committing themselves to in order to reach an agreement.

In this section I try to show that this objection is only apparent. It is based, I argue, on an incorrect assumption of how the content of agreements is determined.

4.1 The Content of Agreements

There are several views about the nature of agreements.¹³ Despite this variety, the issue of how the content of agreements is determined can be examined, I believe, without presupposing any particular view of what an agreement is and why it binds. This is the case if one concedes (as most theories of agreements in fact do) that there are agreeing practices. That is, if one concedes that, as a matter of fact, there are social rules according to which performing certain actions counts as agreeing to perform a particular action (say, A). One can concede, moreover, that there are special relationships (among friends, relatives, colleagues, etc.) constituted by rules according to which performing certain actions counts as agreeing to do A. In what follows, then, I assume that, whatever the nature of agreements, there are practices of agreeing and special relationships of the sort described, and I focus only on the question of how the content of agreements is established.

There are three general views in that respect: the subjective view, the objective view, and the mixed view.

According to the subjective view, for there to be an agreement to do A, the intentions of the parties must coincide.¹⁴ The difficulties of this view seem obvious, for there are plenty of counterexamples. Consider cases where one party makes an ambiguous offer. The individual intended to bind himself to do A, but becomes aware that the recipient will reasonably think that he intended to bind himself to do B because the context clearly supports that view. So he acknowledges that he agreed to do B, despite not having the intention to obligate himself to do B. The same applies to many other cases like blunders, mistakes as to the identity of the other party, errors about the nature of the action proposed, and so on. In many cases of this sort, an agreement has been reached and, contrary to the subjective view, the intentions do not coincide.

According to the objective view, whether one has agreed to do something depends on whether the parties have performed some actions that count as agreeing as defined by a practice of agreeing, regardless of whether the intentional states are present.¹⁵ This view is “objective” only in the sense that it is not subjective. And the approach is unsatisfactory for a simple reason: agreeing practices may require some intentions to be present.

According to the mixed view, some mental states are relevant while others are not. Endicott’s views are a good example of this approach. He claims that whether the parties have agreed to do A is determined by the meaning of the conduct by which the parties agreed as interpreted by a reasonable person. The only “subjective” aspect of agreement is that the parties must do intentionally what counts as entering into an agreement to do A. For instance, in Endicott’s view, if X reasonably thinks that she is signing an autograph (not a form of contract), then she has not agreed to anything, even if Y, a reasonable person, would interpret her conduct otherwise, e.g. because Z arranged things so that everything looked to Y as if X was signing a contract (2000, pp. 152–53, 157, 162–63).

This view is also subject to counterexamples. In some cases the “subjective” aspect that it requires may not be met. For instance, there might be (justified) agreeing practices which, while providing a remedy against Z for misleading X, stipulate that X has acquired an obligation by merely signing a form of contract, even if X reasonably thinks that she is signing an autograph, in order to enable third parties like Y to perform transactions rapidly and without bothering about X’s mental states. In other cases no “objective” aspect is required. For instance, Peter acts in a way that leads his intimate friend, John, to think that he has agreed to do A, where Peter’s doing A is something that both of them consider relatively unimportant. John thinks that Peter has agreed to do A because that is what a reasonable person would make of Peter’s conduct. John begins to act accordingly, and when Peter notices this, he promptly claims that he had no intention to bind himself. So John apologizes, and claims that he was wrong in thinking that Peter has agreed to do A. It seems clear that Peter has not agreed to anything, so the objective aspect that the view considers indispensable is absent.

One could attempt to provide more sophisticated arguments in favour of each of these views, but the result will always be unsatisfactory. It is clear that sometimes we adopt the “objective” view, sometimes the subjective view, and sometimes the mixed view.

This remains true because agreements normally take place within the framework of on-going relationships or agreeing practices that are thought to promote certain values. These relationships and practices may require that certain acts count as agreeing to do A. They may demand that the subjective view be adopted. That is the case of the friends, where agreeing requires the presence of all the relevant mental states because the relationship as such requires that one takes into special consideration what a friend intends. The relevant practices may require that the “objective” view be adopted, as in the case of the contract signed by mistake where a remedy exists for the misled person, such that one has agreed regardless of whether all the mental states are present. In other cases, the mixed view is appropriate, as in Endicott’s example.

Perhaps the best way of establishing when one has agreed to do A, where agreeing takes place within the framework of special relationships or agreeing practices, is in these terms: two individuals have agreed to do A, when agreeing takes place within the framework of special relationships or agreeing practices, when, and only when, the relevant practices, or the relevant relationships, require that their actions count as agreeing to do A. Whether they have agreed to do A in these contexts is, then, an objective question in the following sense: it depends on what the practices or relationships require, and what the practices or relationships require is something that is independent of what the parties to the alleged arrangement think in this respect. In fact, we can claim that, in these scenarios, the agreement to do A creates obligations to do A when, and only when, the relevant relationships or the agreeing practices are in effect valuable. That is, when a value is in effect promoted by the relevant relationship or practice requiring what they require.

This explains all cases, and shows that neither the “objective” view nor the subjective view nor the mixed view is correct.

4.2 Disagreement Reconsidered

Acknowledging the existence of these agreeing practices or special relationships helps explain the kind of disagreement which interests us. Notice that some of these practices have this form: two individuals have agreed to do A when their actions can reasonably be interpreted as if they intended to bind themselves to do A, regardless of whether this was their intention. A “reasonable interpretation” is an interpretation that assumes that the persons are reasonable in the context of the interaction. The contexts may vary but, typically, the interaction is nested within the framework of second-order practices which are taken to promote certain values, or within the framework of shared understandings, to the effect that certain values ought to be promoted. And these values are, normally, seen as objective, in the sense that what they amount to does not depend on what the parties think about that matter.

Consider agreements among tradesmen. There are agreeing-practices stipulating that two tradesmen have agreed to do A when they have performed certain actions that can reasonably be interpreted as if they intended to bind themselves to do A. A reasonable interpretation is one that assumes that the individuals are reasonable in the context of the interaction. And the context is normally nested within the framework of a second-order, broader commercial practice which is taken to promote values such as rapidity, security and fairness in profitable transactions. These values are normally seen as objective values. A reasonable tradesman is, then, a person who is acquainted with this second-order practice and, accordingly, he is a person who has a good grasp of how the relevant values bear on the matter. So what the agreeing practice demands is that the relevant actions be interpreted, assuming that the individuals who performed them are reasonable tradesmen, that is, individuals who have a good grasp of how the applicable values bear on the matter. To adapt the example mentioned before, if a tradesman signs a document that he had every reason to believe was a form of contract but negligently failed to acknowledge as such, everyone would understand that the first individual has bound himself to do what the document provides for, even if he did not intend to do so. Tradesmen would argue in favour of such a view by claiming that this is what the agreeing practice requires, that this is so because his actions can reasonably be interpreted as if he intended to bind himself to do what the document provides for. Other times, tradesmen disagree as to whether an agreement has been reached. Cases of mistakes, blunders, and ambiguities as to the thing agreed to are but examples. And when they disagree, they appeal to what they deem is the reasonable interpretation of the relevant actions, a disagreement that runs deep and is genuine because this depends on how the values in play (rapidity, security, and fairness), which are seen as objective and sometimes are in conflict, bear on the matter.

In short, when this sort of agreeing practices appear, participants may have agreed and still have a genuine disagreement about their duties.

Let us come back to the case of the pro bono lawyers. Assume that their providing legal advice (their collective action) is structured by an agreement, and that this agreement has been reached within the framework of an agreeing-practice of the sort considered, i.e. a practice according to which two or more individuals have agreed to do A when their actions can reasonably be interpreted as if they intended to bind themselves to do A. A reasonable interpretation is one that assumes that they are reasonable persons in the context. And the context of the interaction is, let us assume, a second-order shared understanding: free legal advice is necessary to promote fairness and equality in their local community.

It seems clear that there could be a genuine disagreement about whether they have agreed to provide advice to individuals only or to individuals and organizations. Which actions they have agreed to perform depends on what the agreeing-practice requires. Some of them think that it requires that the relevant actions count as having agreed to provide advice to individuals only. In their view, the relevant actions can reasonably be interpreted as if they intended to bind themselves to that. Put otherwise, they claim that the relevant actions can be seen, assuming that they were acting as reasonable individuals at the time the interaction took place (i.e. assuming that they had a good grasp of how the applicable values bore in the context), as if they intended to bind themselves to provide advice to individuals only. For, as they see it, fairness and equality require that free legal advice be provided only to persons without economic resources, as happens with individuals in their local community. And some of them might think that the practice requires that the relevant actions count as having agreed to provide advice to organizations too. For, in their view, fairness and equality require that free legal advice be provided to anyone who faces a legal problem regardless of its economic capacity. And they can significantly disagree about this matter, as it is an objective matter in the sense described.

The foregoing remarks are sufficient to show, I hope, that some normative relations among members of GNUs can be explained by the idea of an agreement to do one’s part of a joint act, and that this is compatible with the presence of disagreement about duties.

5 Conclusion

Normative relations among members of groups appear for a variety of reasons. However, there is a tendency in the literature to ignore this variety. That is, there is a tendency to appeal to only one particular normative notion that (allegedly) gives rise to duties that are independent of the value of the joint action. This ignores that members, in some cases, think that they are under a duty precisely because of the value of the joint action. It also ignores that, in other cases, they think that they are under a duty because content-independent normative considerations (principles not related to the value of the joint action, such as the principle that agreements should be kept, or that authorities should be obeyed) are deemed applicable. It is plausible to claim, nevertheless, that most groups are structured by agreements. And the fact that participants often disagree about the content of their duties should not be considered an objection. Normally, agreements take place within the framework of agreeing practices, and some of these practices leave room for this possibility: one might have agreed to do something even if not fully aware of what obligation one has acquired, as this is an objective issue over which participants may disagree.