The Duties of Non-Agential Groups: Some Comments on Stephanie Collins’ Group Duties

1 Introduction

Stephanie Collins’ recent book Group Duties is a superb contribution to our understanding of the moral responsibilities of groups. One of its most significant virtues is that it gives an account of collective obligation which allows that notion to play a central role in analytic normative political philosophy – a domain from which, over the last 50 years, it has been largely and mysteriously absent. The views she puts forward are distinctive, well-motivated and clearly and engagingly presented. As will become clear, I disagree with a great deal of what she says. Despite this, I think it important to signal that I take her work to be of the foremost quality: it should set the standard for future work in this area.

For reasons of space I shall focus on one central issue: the question of whether non-agential groups can be duty-bearers. Collins thinks they cannot. I think they can. The first third of Group Duties focusses on precisely this issue. In chapter 2, Collins sets out and responds to a number of arguments for thinking that groups can have duties, and argues that though they may have some merit none of them are compelling enough to defeat an initial intuition that only agents can have duties. This leads, she thinks, to a stand-off. In chapter 3, she then attempts to resolve this stand-off, by presenting a new argument for the claim that only groups with decision procedures can have duties.

I suspect the arguments of chapter 3 are less compelling than Collins thinks. If I am right, this returns us to a stand-off on the question of whether non-agential groups can have duties. I am also less convinced than she is by some of the arguments of chapter 2. But for reasons of space, I shall concentrate on the argument that is central to chapter 3.

2 Preliminary Concerns

We might think that before addressing the question of whether non-agential groups can be duty bearers, there’s a prior question that we need to discuss: namely ‘what is a group agent’. Collins doesn’t tackle this question fully until chapter 6 of Group Duties. Her arguments in chapters 2 and 3 don’t, therefore, seem to depend on the details of the answers she gives to it. However, they do depend on something that she takes to be a necessary condition of being a group agent: namely the possession of a decision procedure (Collins 2019, pp. 11–14)

The notion of a decision-procedure is important for at least two reasons. First, it plays a significant role in determining which groups are group agents. So we can’t understand what her view does and does not rule out without knowing what counts as a decision procedure. Secondly, Collins characterizes duties as being potential inputs into decisions: something that can’t be an input into decisions can’t be a duty. This characterization of duties plays a role in some of the arguments she puts forward in chapter 2, when arguing against the potential explanatory power of some kinds of claim about duties falling on non-agential groups. Duties that can’t play a role in decisions, she suggests, don’t explain anything at all (Collins 2019, pp. 29–32).

Collins emphasizes that decision procedures can be both formal (e.g. voting) and informal (e.g. conversation-based consensus among a group of friends deciding where to have lunch.). So, her view that group agents are the only kinds of groups that can have duties shouldn’t be taken as committing her to the view that the only kinds of groups that can have duties are highly-structured groups (Collins 2019, pp. 13–4).

You might think that a decision procedure is something which takes the agent from its own beliefs and desires to the formation of intentions – either intentions prior to action, or intentions in action. We can see that is not what Collins means by a decision procedure by considering two things which Collins regards as paradigmatic examples of, respectively, formal and informal decision procedures: voting and conversationally generated consensus (Collins 2019, pp. 13–15). There is no obvious reason to think that groups that have procedures of this sort must have anything worth describing as group beliefs or desires. Even if they do, there’s no reason to think that these group beliefs and desires will be causally relevant to the group’s decision.^[1]

For Collins, a group has a decision procedure provided i) each of the members of the group is committed to abiding by the procedure’s results and ii) systematically has influence over the results in such a way that iii) the enactment of those results requires actions on the part of the members, which are properly understood as attributable to the collective. Collins also holds that ‘the beliefs and preferences that the procedure takes as inputs and the way the procedure processes those inputs’ must be ‘operationally distinct from the inputs and processes that any member uses when deciding for themselves’, where ‘operational distinctness’ defined in the following way: a group’s procedure is operationally distinct from that of members insofar as it can take in distinct inputs from those that individual members take into account; its method for processing those considerations is different from the procedures of individual members, and the decisions it produces are not the straightforward conjunction of individual decisions.

This conception of a decision-procedure is intended not to be so wide as to include every way in which actions might emerge from the beliefs, desires, and interactions of a group. We can see this by considering two different kinds of groups, which do not constitute groups agents which Collins calls coalitions and combinations. Coalitions are groups whose members are responsive to one another’s actions, while combinations are groups whose members are not so responsive.

On Collins’ account neither coalitions nor combinations have decision procedures (Collins 2019, pp. 16–21). Consequently, they have no duties in their own right. Nevertheless, their members do have duties, and their duties differ in interesting and significant ways. Coalition members have duties to ‘we-reason’ to advance shared goals, whereas combination members have – at best – duties to co-ordinate themselves as collective agents.

These features of Collins’ position entail that a coalition whose members are each committed to a policy of we-reasoning does not constitute a group agent. Consequently, its members’ shared commitment to we-reasoning does not (by itself) count as a decision procedure. Still less does a group, each of whose members does what seems best by their own lights while presuming that every other member will follow the same policy, count as a group agent. A fortiori such a group’s members’ shared policy of doing what seems best by their own lights also does not count as a decision procedure.

How does Collins’ account rule these out? The key consideration is operational distinctness: in particular, the condition that the decisions the procedure produces are not the straightforward conjunction of individual decisions. But this suggests that many groups which come close to being coalitions and combinations will count as group agents. If a group contains a single member who is willing to subordinate their personal decision to what they take to be majority opinion (or even only to unanimous opposition), the operational distinctness condition will be met. If so, we will have something which, at least as I understand Collins’ position, has a group decision procedure and constitutes a group agent. If this is correct, then her official characterization of group agents may mean that her claim that non-agential groups cannot be duty-bearers may not rule out as much as she wishes it to.

One further slightly counter-intuitive consequence of Collins’ account is that even where there are group agents, some of the individuals who we might expect to count as members of an agential group will not actually be members of that group. In particular, any individual who is not committed to abiding by a procedure’s results will not be a member of the group agent. One might think that this will make it unduly easy to avoid any duties that fall on one in virtue of one’s membership of a group agent: one could presumably avoid such duties by ceasing to be a member of the group agent – that is to say, by ceasing to be committed to abiding by the results of the group’s decision procedure. One might also think that one way for an individual to cease to be committed to the results of a decision procedure would be for them simply to refuse to perform tasks which are assigned to them as a result of that decision procedure. This is worrying: a duty which ceases to be a duty if one refuses to perform it isn’t much of a duty – its normative force seems too evanescent to count for much.

Collins might respond that someone who repudiates a commitment to abiding by a group’s decision procedure will have done something wrong in doing so. That may be true, though it won’t be much comfort to an individual who is left in the lurch by a group member’s failure to carry out a task which a decision procedure assigned to them. More helpful, perhaps, is the suggestion that the now disaffiliated individual might acquire an obligation to rejoin the group from which they have disaffiliated themselves.

Collins’ account will, I think, entail the existence of a duty of this sort when the refusing member’s disaffiliation from the group causes the group agent to collapse into a combination, for she holds that members of a combination of this sort may have an obligation to constitute a group agent. But it’s less clear that anything of this sort will be true if the group agent is less fragile. In order to guard against this possibility, we’d need to be able to argue that there are circumstances, in which an individual has an obligation to affiliate themselves to an existing group agent. It would be an interesting and, I think, non-trivial, though perhaps not hopeless task, to formulate an account of the sort that Collins would need to deal with this kind of case.

3 Collins’ Master Argument

Here’s an argument that is central to the case that Collins makes in chapter 3 for thinking that duties cannot fall on non-agential groups (Collins 2019, pp. 85ff).

•  P4) In order to fulfill a duty, you must make a decision.

•  P5) To be a duty-bearer, one must have the ability to fulfill at least some duties.

•  C) Non-agential groups cannot make decisions.

•  C) follows from P4 and P5 provided one thinks, as Collins does, that groups that can make decisions are groups with decision procedures, for she holds, as we have seen, that such groups are group agents. I’ll call this Collins’ Master Argument.

It’s worth emphasizing two things about this argument. First, fulfillment is a technical term for Collins. Not every way of doing what one has a duty to do counts as fulfilling one’s duty. One fulfills a duty on Collins view only when one does what one does for the right reason. Secondly, Collins isn’t saying that to bear a duty to F you must be able to fulfill a duty to F. You may have duties which you are constitutionally incapable of fulfilling. Consider here, for example, my duty to speak to time in a colloquium: P4 does not entail that if I am constitutionally incapable of speaking to time except in the face of physical threats from the colloquium’s moderator, I have no such duty (This example also helps illustrate one can allow for the existence of duties that an individual cannot fulfill, without repudiating the principle that ought implies can.).

The notion of fulfillment might seem familiar: Collins suggests, rightly, that it has Kantian resonances (Collins 2019, pp. 87–8). These become clear from her claim that in fulfilling a duty I acquire moral worth. However, we should note that not every way of dependably and non-flukily doing what a duty requires counts as fulfilling a duty. One way of seeing this is to notice that although I can be coerced into doing what a duty requires, I can’t be coerced into fulfilling a duty. If I’m coerced, then I’m not acting for the right kind of reason (and I don’t acquire moral worth). There may also be ways in which I can dependably and non-flukily do what a duty requires without making a decision. Perhaps if my environment is structured correctly, I’ll do what duty requires without giving it any thought: that’s an idea that’s central to recent discussions of the idea of ‘nudging’. Again, I won’t be fulfilling a duty, though I will be doing what the duty required.

It’s going to be helpful to have a word for ‘doing what a duty requires’. Quite often, when someone owes us a duty we don’t mind too much whether the duty bearer fulfills their duty or merely performs it. And even when we do care about how the duty is performed, it may be enough that the duty is performed in a way that is non-fluky and relatively reliable, across a reasonably wide range of circumstances in which the duty applies. Collins says something that bears on this when she says that the kinds of duties that are correlative to human rights have to be claimable (Collins 2019, p. 40). A right is only claimable, she says, if we can coerce someone into doing what the duty requires. If the only thing we cared about where duties are concerned was whether they are fulfilled, then claimability wouldn’t matter. For, as I’ve already observed, you can’t coerce someone into fulfilling a duty. The best you can hope for is to compel performance.

We might have doubts about P4: many virtue ethicists will want to say that agents who have been habituated in the right way will be capable of acting for the right reasons without reflecting on their decision.^[2] Much will depend here on how much we take the notion of decision to involve. If Collins’ account is to be plausible, we will probably need to be able to say that where individual agents are concerned, it is possible to make a decision without having anything that we would find it natural to identify as a decision procedure. If so, this is a fairly striking disanalogy between individual and group agents.

However, for the purposes of this paper I shall focus on P2. Collins’ defense of this premiss is fairly brief: she argues that i) the ability to fulfill duties distinguishes the average adult human from dogs, tables and young children, and ii) the presence of this distinction is the best explanation of why (we think) average adult humans (AAHs) have duties while dogs, tables and young children (DTCs) do not (p. 93).

We might be suspicious of the idea that there is any one feature which distinguishes duty-bearers from non-duty bearers. For the sake of argument, let’s suppose there is. There are, of course, any number of distinctions between DTCs and AAHs. For example, AAHs are capable of following someone else’s presentation of means-end reasoning. Tables and dogs are not; nor are sufficiently young children. So, which of the differences between AAHs one the one hand, and tables dogs and children on the other, are relevant to explaining why creatures in the first category can have duties and the other not?

One answer is Collins’. Another is that AAHs are capable of performing duties in ways that are non-fluky and relatively reliable across a range of circumstances whereas dogs, tables and sufficiently young children are not. If that’s right, we don’t have an argument for P5: we only have an argument for the weaker claim that you can only have a duty if you are capable of performing or fulfilling some duties in ways that are non-fluky and reliable across a range of circumstances.

We might try to rescue the argument by weakening P5 to ‘In order perform a duty in a non-fluky manner, you must make a decision.’ However, consideration of what is sometimes called ‘nudging’ suggests that it’s sometimes possible to perform a duty in a non-fluky manner without making a decision. For example, some countries make registration as an organ donor card opt-out rather than opt-in. As a result, I can register without making a decision to register. If we have a duty to be registered, an opt-out policy makes it possible to perform this duty – though not fulfill it – without deciding to. So, the weakened version of P5 is false.

So far, I’ve argued that P2 is under-supported. But we might think there’s worse news for Collins. Consider two candidate explanations of the fact that AAHs have duties while DTCs do not.

A. The pattern of duty possession and lacking is explained by the fact that AAHs can fulfill duties while DTCs cannot.

B. The pattern of duty possession and lacking is explained by the fact that AAHs can perform duties in a non-fluky and relatively reliable manner while DTCs cannot.

We can distinguish between these candidate explanations by considering examples of beings that can perform duties, but can’t fulfill them. Our judgments about these cases will give us grounds for deciding between A and B.

Some people think that psychopaths can’t fulfill duties: they can be coerced or manipulated into doing what morality would require of them, but they can’t do it for the right reasons. Suppose that’s right. If Collins is right, psychopaths can’t have duties. Here are two reasons for thinking that they can. If psychopaths can’t have duties, they can’t breach duties. If so, reactions which would be warranted by an individual breaching a duty can’t be warranted by the actions of psychopaths. This seems implausible. Furthermore, we think it’s permissible to coerce psychopaths in some ways, but not in others. Why? Well, one natural answer would be that a necessary (but not sufficient) condition for the permissibility of coercing someone to do F is that they have a duty to do F. If psychopaths can’t have duties, that can’t be right. If psychopaths do have duties it can be, but Collins is in trouble.

Some people have thought that some kinds of collective agent, such as business corporations, might be systematically incapable of responding to moral considerations. Whether or not this is true, we can certainly postulate collectives which have decision procedures which exclude moral considerations. On Collins’ view, that makes them incapable of having duties. However, collectives of this sort have a capacity to set their ends. Kant takes this to be definitive of humanity. We might think this has significant implications for the question of how we might treat them: in particular, it may be impermissible to treat them as a mere means (Cf. Wringe 2014). May such (collective) agents permissibly be coerced? That may depend on whether they have duties. If so, Collins’ position has unwelcome implications.

4 Reconsidering the Phenomenology

If Collins Master Argument fails, then we are back to a stand-off on the question of whether non-agential groups can have duties. In concluding, I now want to address some of the points that Collins makes about some of the phenomenological considerations which I and others have taken to support the case for thinking that non-agential groups can have duties. If I’m right that Collins’ points here establish less than she thinks, I think the advantage lies with the view that non-agential groups can have duties

Elsewhere (Wringe 2016), I have discussed scenarios in which two individuals who are disinclined to co-operate are faced with a morally urgent task that can only be performed in concert. In such situations, I suggest, the situation may present itself to participants as one in which ‘we ought to do something’. I have also argued that, other things being equal, we should take this phenomenology at face value.

Collins makes two points, neither of which seem conclusive. First, she suggests that we might understand the phenomenological judgment as involving a distributive we-predication, rather than a non-distributive one. It is a judgment that each of us should do something, rather than that we should do something together. It’s worth noting that Collins presents her discussion of this example in third-personal terms rather than first-personal ones. This seems relevant to assessing the plausibility of her position. It’s easy to see how we might understand another person’s utterance ‘we ought to do something’ along the lines Collins suggests, it’s harder to imagine how someone could be mistaken in this way about whether their own judgment should be taken as involving distributive or non-distributive predication.

Furthermore, we can describe situations, where the only judgment we can plausibly imagine someone making is non-distributive. Consider a case involving three agents, in which it is clear that any two of whom can carry out the morally urgent task. In a situation like this a judgment expressed in the words ‘we ought to do something’ would simply be false if understood distributively: ex hypothesi, it would not be true that each of the individuals should do something.

Collins holds that in the kinds of situation I envisage the two individuals ought to form a group agent. Once they have done so, that group agent will have a number of duties. In the light of this, she suggests that the phenomenology I appeal to may be seen as anticipating a duty, which will come into existence only once the group agent is formed. Whatever the merits of this suggestion, it’s difficult to see how the account can be extended to cover backwards-looking phenomenology.

Suppose that, as a matter of fact, we fail to co-operate to carry out the morally important task. Reflecting on the situation later, the right thought to have might be ‘we really messed up there’. It may not be true that I messed up: if you were unco-operative, there might have been nothing more I could have done. So we can’t account for what happens here by appeal to a distributive judgment.

Nor does it seem helpful to appeal to the idea that if we had formed a group agent, that group agent would have had some duties that weren’t carried out. It’s hard to see why duties that would have fallen on an agent if that agent had existed should be an occasion for regret – let alone justifiable regret on my part. I might as well regret the non-intervention of Superman. I might of course, regret my failure to do what I could to form a group agent. But this, I think, locates the regret in the wrong place: what I should regret is a failure to solve the problem, not a failure to co-operate. But that perhaps is a matter that merits further discussion on another occasion.