Sereeta plans to vote for the Purple Party in the upcoming election. She believes that the Purple Party would form the best possible government, and that it alone is capable of solving her nation’s pressing problems. She knows, however, that very few people agree with her, and that the Purple Party candidate will not be elected in her constituency. Still, she takes herself to have sufficient reason to vote for him, and she does so.
Sereeta appears to take herself to have a group-based reason. These are reasons for individuals to act, of a special kind. They are reasons to participate in some pattern of action that the members of a group could perform, by playing one’s part in that pattern. The reason for playing one’s part comes from the goodness (or other favourable qualities) of the pattern, not the goodness of the part. Sereeta’s reason seems to be like this. She does not expect her vote to achieve anything. It will not make a difference to the election result, or encourage others to reflect on their views, or have any other good effect itself. Her reason is that the electorate could elect the Purple Party, and this would be best, and voting for the Purple Party candidate is her part in this optimal pattern. Though other members of the electorate will not play their parts in this pattern, she does not take this fact to extinguish her reason for playing her part in it.
According to philosophical orthodoxy, Sereeta must have made a mistake. Many philosophers simply deny the existence of group-based reasons altogether. Among these are Act Consequentialists. Act Consequentialists allow that there can be good reasons for playing one’s part in a good pattern of action, of course; but they insist that these reasons must be grounded in the difference one’s action would make. According to them, they cannot be grounded in the mere fact that the action is part of some good pattern. Other philosophers are willing to accept group-based reasons. However, they typically impose strict conditions on their existence. They insist that group-based reasons can exist only when (the agent believes that) others will play their parts in the pattern concerned, or (more strongly) only when the relevant group would form a group agent. Either of these proposals rules out the existence of group-based reasons in Sereeta’s case. So, if any of these philosophical views is correct, but Sereeta takes herself to have a group-based reason to vote Purple, she must have made a mistake.
Yet many people are in Sereeta’s position, or something relevantly like it. Many of us are not dissuaded from voting for our preferred candidate by the knowledge that this candidate has no chance of winning. Similarly, many engage in environmentally-friendly actions not because they expect these actions to achieve much or anything taken individually, but because each believes them to be her own part in a good possible pattern of action of environmental protection, performable by everyone. They are not always put off by others’ unwillingness to cooperate in producing this good pattern. Likewise, many parents think that there is a stronger reason not to have a third or fourth child than there is not to have a first or second child. This might be because they recognize that the world’s population would be controlled were no parents to have more than two children, and that not having more than two is their own part in this pattern. While it is always possible to imagine alternative explanations of these cases, their face value is that we sometimes take ourselves to have group-based reasons despite the uncooperativeness of the other members of the relevant group.
It appears that philosophical orthodoxy is out of joint with the reasons many people take themselves to have. I will argue that the orthodoxy is mistaken. The argument proceeds in admittedly broad terms: the strategy is to distinguish three conceptions of group-based reasons, and then to argue against two of them and in favour of a third. The orthodoxy rests on one or other of the rejected conceptions. Section 2 clarifies the concept of group-based reasons. Section 3 distinguishes the three conceptions. Section 4 presents some intuitive considerations bearing on our choice between the three conceptions. Section 5 attempts to undermine the intuitions that favour the orthodoxy. Section 6 discusses the real significance of cooperativeness, and Section 7 concludes.
2 Group-Based Reasons and Group Agency
GBR Agent A has a group-based reason to do X if and only if and because X is A’s part in some favoured and eligible possible pattern of action P performable by the members of some group G.
ABR Agent A has an act-based reason to do X if and only if and because X is favoured and eligible.
Both the concept of group-based reasons and the concept of act-based reasons make use of the idea that something can be favoured (e.g. good, right, courageous, just) or not, and of the idea that some action or pattern of action can be eligible or not. The concept of eligibility is needed to mark the fact that merely being favoured is not sufficient to generate a reason. For example, ‘curing cancer by clicking my fingers’ is favoured, since it would be of great value. But I have no reason to do it, because it is not something I could do. At minimum, an action or pattern of action is eligible (to be the object of a reason for action) only if it could be performed in the circumstances.
Group-based reasons have a more complex structure than act-based reasons. In the case of act-based reasons, the thing which is favoured is the thing for which there is a reason, and the fact that it is favoured provides the reason. In the case of group-based reasons, in contrast, the thing which is favoured (a pattern of action) is distinct from the thing for which there is a reason (an individual action, which is part of the pattern), and the fact that the pattern is favoured provides a reason (if it does) because the action is part of the pattern. So the parthood relationship between the action and the pattern mediates the reason. 2
Why should we think that group-based reasons exist? One reason is that it helps to make sense of people like Sereeta. Outside of philosophical discussion, many people appear to use the concept of group-based reasons, and to take group-based reasons to exist. They articulate ethical ideas by saying things like ‘I should do my part’, or ‘do your bit!’, and so on, in contexts where it appears that they are not pointing to act-based reasons for contributing to a pattern. This is, perhaps, the most important reason for philosophers to take seriously the idea of group-based reasons.
A second reason is more theoretical. As Regan (1980, p. 18–21), Bacharach (2006, p. 141–144), and others have argued, the concept of group-based reasons helps us to make sense of intuitions about reasons in simple games such as Hi-Lo (Gold and Sugden 2007, p. 283–285). In Hi-Lo, two players each face a choice between two options: Hi, and Lo. Both players agree that the outcomes are ranked as follows: Hi-Hi is best; Lo-Lo is next best; the other two combinations are worst. Intuitively, each player has more reason to choose Hi than to choose Lo; yet, in some circumstances, we cannot explain this in terms of act-based reasons. 3 Act-based reasons focus on the difference each individual action makes. Thus they tell each player to choose Hi if the other will choose Hi, but to choose Lo if the other will choose Lo. So, if the players end up converging on Lo, they have made no mistakes, and neither player had any reason to do anything else – even though both would have been better off had they done the opposite.
The core issue is that act-based reasons are based on ‘horizontal’ or ‘vertical’ comparisons between outcomes: comparisons of the difference Row can make (vertical comparisons) or of the difference Column can make (horizontal comparisons). But our intuition that there is more reason for each player to choose Hi than Lo concerns a diagonal comparison: between the outcome of Lo-Lo and the outcome of Hi-Hi. The concept of group-based reasons enables us to point to a reason that is based on this comparison. We can say that the pattern of action (Hi-Hi) is best, and that this gives each player a reason to play her part (Hi) in that pattern (Woodard 2007, p. 45–47).
This is by no means a decisive argument for the existence of group-based reasons. It is meant to show only that we should take the idea seriously, and investigate it further. Note that we can use the idea of group-based reasons in connection with an interest in motivating reasons, or normative reasons, or both. Here I shall focus on normative reasons, though I assume that, as with other normative reasons, people could be motivated by group-based reasons. If that is correct, then we could explain action using the concept of group-based reasons in those cases where people are indeed motivated by such reasons.
So far we have introduced the idea of group-based reasons, and contrasted it with the idea of act-based reasons. But of course the idea of group-based reasons makes use of the idea of a relevant group, G, the members of which ‘could perform’ a pattern of action, P. This raises the question of the relationship between group-based reasons and phenomena that we may loosely gather under the umbrella term ‘group agency’ (Tuomela 2013; Bratman 2014). For there to be a group-based reason, what must be true of group G and pattern P? Must G itself be an agent, and P an intentional action performed by this agent? Or will something weaker suffice? Note that we did not build any answer to these questions into our definition of group-based reasons. The definition did not take a stance on these issues, or even on related questions such as whether the pattern must serve the group’s values, or make sense from its point of view. It left those issues open. They are to be filled-in by reflection on the conditions under which the idea of group-based reasons makes most sense.
3. Three Conceptions of Group-Based Reasons
One way to approach these issues is to distinguish some different conceptions of group-based reasons and their relationship to group agency. Here we will distinguish three different conceptions. None of the three conceptions is intended to describe a full theory of group-based reasons. Instead, each is intended to articulate an underlying idea or background assumption about what needs to be true of the group for a pattern of action to be eligible – that is, to support a group-based reason.
SWR A has a group-based reason to play her part X in P only if the members of G would under that circumstance exhibit the behaviour and intentional states required for P to be an exercise of group agency. 4
According to the strong willingness requirement, there is no group-based reason for the agent to play her part in a pattern unless the conditions for the pattern’s being an exercise of group agency (of some specified kind) are either already satisfied or would be satisfied if she were to play her part. Others must be willing in a strong sense that requires them not merely to exhibit the requisite behaviour, but also to exhibit the requisite intentions, beliefs, or other intentional states. In a slogan: no group-based reasons without group agency. 5
WWR A has a group-based reason to play her part X in P only if the members of G would under that circumstance exhibit the behaviour required for P to be realized.
According to the weak willingness requirement there is no group-based reason to play one’s part in a pattern unless the pattern would thereby be realized (one’s own contribution need not complete it, but the pattern must be realized nonetheless). Others must be willing in the weak sense that they would exhibit the requisite behaviour, but not in the strong sense that they must have specific intentional states in addition. In a slogan: no group-based reasons unless the pattern would be realized (Regan 1980, p. 124; Hurley 1989, p. 146; Schwenkenbecher 2014, p. 70–71; Wringe 2016, p. 488).
Both the agency-first view and the behaviour-first view imply that there are no group-based reasons in uncooperative contexts such as Sereeta’s. Not enough of Sereeta’s fellow electors are minded to vote Purple, let alone to do so with the intentional states required for the possible act of electing a Purple government to constitute an exercise of group agency. In other cases, these two conceptions of group-based reasons will of course have divergent implications. Consider the pattern of action consisting of no-one’s breaking the motorway speed limit. The weak willingness requirement but not the strong willingness requirement is satisfied if the only reason that others are not breaking the limit is that their cars have mechanical problems.
WPR A has a group-based reason to play her part X in P only if the members of G could under that circumstance exhibit the behaviour required for P to be realized.
SPR A has a group-based reason to play her part X in P only if the members of G could under that circumstance exhibit the behaviour and intentional states required for P to be an exercise of group agency.
According to the weak powers requirement, there can be no group-based reason to play one’s part in a pattern of action that could not be realized by the individuals concerned. This amounts to no more than the minimum requirements of eligibility. In a slogan: no group-based reasons unless the pattern could be realized. According to the strong powers requirement, there can be no group-based reason to play one’s part in a pattern of action that could not be realized as an exercise of group agency. This is a slightly more demanding condition, since it requires that the members could act as a group agent. In a slogan: no group-based reasons without the possibility of group agency.
Each of these ways of developing the powers-first conception makes use of the concept of what the members of some group ‘could do’. It is important to emphasize that this is a matter of what the members could do, not a matter of what the group as an entity could do. For the weak powers requirement to be satisfied, it has to be true, of each member, that she could exhibit the behaviour required of her by the pattern in question. For the strong powers requirement to be satisfied, it has to be true, of each member, that she could exhibit the behaviour and intentional states required of her for the group to perform the pattern in question as an exercise of group agency. Neither of these conditions taken by itself requires that the group itself could act as an agent. Even if each member of the group is capable of playing her part, and exhibiting the right intentional states, it does not follow that the group itself is capable of this. 7
It is also worth emphasizing that the strong and weak willingness requirements, and the strong and weak powers requirements, each propose only necessary conditions for the existence of group-based reasons. Therefore each of them is compatible with the addition of further necessary conditions. Relatedly, none of the three ‘conceptions’ of group-based reasons that I have mentioned is a complete theory of group-based reasons. Each one is a partial theory, or describes a family of possible theories containing significant internal variety. Nevertheless, it is worthwhile focusing attention on the three conceptions, since each is a root idea with significant consequences. In particular, many contributors to debates in this area appear to take one or other of the agency-first or behaviour-first conceptions for granted, with the result that one or other version of the willingness requirement is also taken for granted.
The powers-first conception is consistent with the existence of group-based reasons in uncooperative contexts, while the other two are not. Which conception is most credible? In the remainder, I shall advance some intuitive and some theoretical considerations in favour of the powers-first conception.
Each of the three conceptions is open to an intuitive objection. Let us start with the easiest case to make – though I shall argue in §5 that the case ultimately fails. On the face of it, the powers-first conception of group-based reasons is open to a decisive objection: it appears to be reckless. It claims that there can be a group-based reason to play one’s part even when no other members of the group would realize their parts. While such unilateral participation may sometimes be harmless, on other occasions it could be deadly. One needs only think of unilaterally driving on the other side of the road, to take one obvious example. Moreover, this is not merely a matter of being reckless for another purpose (such as to win a game of Chicken). It is a matter of systematically ignoring facts about others’ behavioural responses to one’s actions, and so can be called systematic recklessness.
It is certainly true that we do not want our theories of reasons or rationality to imply that agents should act in a systematically reckless way. If they did that, these theories would offend our intuitive sense of prudence and ethics so grossly that they would surely fail, whatever their other merits. So let us take it as a constraint on conceptions of group-based reasons that they should not recommend systematic recklessness. The powers-first conception appears to violate this constraint.
Second, there is an intuitive case against the strongest conception, the agency-first view. This view appears to rule out, wrongly, opportunistic participation. Consider a case in which the strong willingness requirement is met: Paddy and Eddy are playing Hi-Lo; they both understand the game; they both recognize each other’s cooperativeness, and satisfy all of the conditions necessary for their actions together to constitute an exercise of group agency. They each choose Hi. Now change the facts, so that the weak willingness requirement but not the strong willingness requirement is satisfied: Paddy and Eddy* are playing Hi-Lo; they both understand the game; Eddy*, however, is an inept thwarter: he intends to thwart Paddy’s efforts to bring about a good outcome, but (although he understands the game) he makes reasoning errors that lead him, reliably and predictably, to behave in exactly the same way as Eddy; Paddy knows all of this. Intuitively, if we think that Paddy has a group-based reason to play Hi in the first case, we should think the same in the second case. 8 That is, the highly unusual explanation of Eddy*’s behaviour does not extinguish Paddy’s group-based reason, if he has one, for choosing Hi. The optimal pattern will still be realized if he chooses Hi, and Hi is his part in that pattern. 9 The agency-first conception’s focus on intentional states seems misplaced, insofar as the optimal pattern does not itself require them.
Third, there is an intuitive case, albeit harder to see, against both forms of the willingness requirement. To let the intuitions speak up we have to consider a case with two special features. First, to avoid worries about recklessness, it should be a case in which unilateral participation is costless. Second, to avoid confusing intuitions about group-based reasons with intuitions about act-based reasons, it should be a case in which unilateral participation is benefitless. That is, we should consider a case in which the act of participation costs nothing and, taken by itself, achieves nothing. Cases with these unusual features include ideal compensator cases. Consider, for example, Paddy’s situation when faced with Compo. Compo is a mirror agent: he acts in such a way as to nullify the effects of Paddy’s acts. Whatever Paddy does, Compo compensates perfectly for it. When they play Hi-Lo, Compo unfailingly plays Lo when Paddy plays Hi, and plays Hi when Paddy plays Lo.
In this situation, there is no cost and no benefit to either of Paddy’s options (Hi or Lo). Some will have the intuition, nonetheless, that he has more reason to choose Hi than Lo, or even specifically that he has a group-based reason to choose Hi.
Perhaps, though, you do not share this intuition. In that case, we can raise the stakes. Compensator cases can involve less anodyne options than Hi or Lo. Suppose, for example, that Paddy is considering whether to restrict his carbon emissions. Again he is beset by his mirror agent, Compo. Suppose that they could each reduce their emissions by up to 100 g, and that there is a threshold such that they would achieve some environmental benefit if and only if they together reduce their emissions by 200 g. But if Paddy reduces his emissions by 100 g, Compo will increase his emissions by 100 g. Nevertheless, we may think that Paddy has a group-based reason to reduce his emissions, because of the benefit they could achieve, and because this reduction is his part in that good possible pattern of action. Note, also, that Paddy’s situation is surprisingly realistic. For one thing, it approximates the situation of many market participants. If you do not consume that petrol, someone else will. 10
Further raising of stakes is possible, though it introduces additional complexity. Suppose that you are about to buy a lampshade when you discover that it is made from human skin. Another eager buyer is ready to step in. You may want to play your part in the possible pattern consisting of no-one’s buying such goods, even though this other buyer will not cooperate even in the weak sense in producing that pattern. Surely we will all think that there is a reason for you to refuse to buy, even though it will, on the face of it, achieve nothing (the market in such goods will not be undermined by your action). Admittedly, we might think that this case introduces distracting deontological considerations. On the other hand, we might entertain the hope of explaining those considerations by extending the idea of group-based reasons to such uncooperative contexts. After all, one natural way of articulating the ethical disgust one would feel in such a situation would be to say ‘I do not want any part in that!’ (Woodard 2013, p. 257–259).
Thus for each of the three conceptions there is at least one intuition that appears to speak against it. Table 1 summarises the situation.
Intuitive Responses to the Three Conceptions of Group-Based Reasons.
|Anti-recklessness||Pro-opportunistic participation||Pro-unilateral costless and benefitless participation|
5. Undermining the Willingness Requirement
Conception C is incompatible with intuition I.
Intuition I is true.
Therefore, conception C is false.
Arguments of this form are very often question-begging, in the sense that they rely on one or more premises that the opponent would not accept. A proponent of conception C may never endorse intuition I, or may cease to endorse it when she sees its incompatibility with C. For example, someone who endorses the agency-first or behaviour-first conceptions of group-based reasons may not accept that there is any group-based reason in the ideal compensator cases discussed in Section 4. 11 Thus, even if sound, arguments of this form will often fail to be dialectically effective.
Fortunately, we can go beyond these intuitive arguments. I shall argue in this section that when we reflect on the theoretical roles that the willingness requirement in either form may be thought to play, we will see that there is no good case for it.
The first and most obvious role for the willingness requirement is to guard against systematic recklessness. Tying the existence of group-based reasons to the other members’ willingness is one way to ensure what Bacharach called circumspection (1999, p. 118). It may seem as though one or other version of the willingness requirement is at least one of the features we should build in to a theory of group-based reasons in order to avoid recklessness. However, this appearance is deceptive.
Note as a preliminary point that if we are concerned to prevent recklessness, then the weak willingness requirement makes most sense. There is nothing reckless about opportunistic participation with people who have all the right behaviour but who lack the intentional states required for the pattern to be an exercise of group agency. If the pattern of behaviour is realized, joining in with it will not be reckless – even if it would have other defects. So, the strong willingness requirement would be overkill. I will return to the significance of this observation shortly.
The real problem is that the willingness requirement, in either version, is not a natural or elegant way to guard against recklessness. It may seem to be the only way to guard against it if we assume monism, which in this context is the claim that an agent can have act-based reasons, or group-based reasons, but not both at once. For then we have to find some device within the theory of group-based reasons that guards against recklessness, and the obvious way to do that is to adopt a version of the willingness requirement. But those who believe in group-based reasons have compelling independent reasons not to be monists in this sense. Anyone who believes that the parthood relation between an action and something good (a pattern of action) can provide a reason to perform the action, can hardly deny that a causal relation between an action and something good (an outcome) can provide a reason to perform the action. So, anyone who believes in group-based reasons should also believe in act-based reasons. Moreover, the normative significance of the causal properties of an action can hardly be expected to disappear just when the action also has one or more parthood properties with normative significance. Thus, believers in group-based reasons ought to be pluralists about kinds of reasons. 12
This point is quite independent of any analysis of intuitions about recklessness. But once we accept that group-based reasons may be accompanied by act-based reasons, we have no need to adopt the willingness requirement to guard against recklessness. In the cases where recklessness is a concern, acting on a group-based reason is likely or certain to have bad consequences. But where there are bad consequences, there will be an act-based reason not to act in that way. This reason is sufficient to explain our concern with recklessness. 13 Moreover, it gives the right sort of explanation. The problem with acts of reckless unilateral participation is that these individual actions have (or are likely to have) bad consequences. It is this feature of the situation that our intuitions about reasons are tracking. Thus, these intuitions are about act-based reasons. It is both artificial and unnecessary to say that there is, in addition, a constraint on the existence of group-based reasons having to do with willingness of others to cooperate. Moreover, doing so amounts to a limited re-introduction of the predictive stance towards others’ actions that, many have pointed out, is at odds with thinking of them as co-participants in some pattern performable by the group (Regan 1980, p. 143–144; Woodard 2007, p. 61–64; Roth 2014).
We avoid recklessness by striking an appropriate balance when considering the interaction of act-based and group-based reasons. Act-based reasons, not group-based reasons modulated by the willingness requirement, tell us not to drive on the other side of the road unilaterally. Thus the anti-recklessness intuition does not provide a compelling case against the powers-first conception of group-based reasons. It appears to do so only if we assume monism about kinds of reason. Since no believer in group-based reasons should accept monism, it is particularly inapt for recklessness-based arguments to be deployed in favour of the agency-first or behaviour-first conceptions.
However, perhaps the willingness requirement plays some other important theoretical role. One possible role is to exclude intuitively irrelevant groups or patterns. We might worry that, without some form of the willingness requirement, our theory of group-based reasons will fail to exclude some groups and patterns that, intuitively, should be excluded. To put the point starkly, if we are not concerned with willingness, what is to prevent a bizarre agglomeration of individuals counting as a ‘group’, or a bizarre agglomeration of actions counting as a ‘pattern’? When I am considering what I should do about carbon emissions, what is to stop me considering my part in the best that could be done by a group consisting of all the left-handed red-headed people in the world, plus me? Or from considering the best that we could together do on alternate Tuesdays?
First, note which version of the willingness requirement appears to help with this problem. The idea would be that checking for willingness might helpfully exclude groups or patterns that, intuitively, are irrelevant. But this means that it would have to be the strong willingness requirement playing this role. Presumably, it is not the case that all the left-handed red-headed people in the world are willing, in the strong sense, to cooperate with me. But they could be willing in the weak sense – especially if we allow the pattern to include intuitively irrelevant things, such as brushing their teeth. If behaviour is all that matters, a pattern involve their doing that and my reducing my emissions might well satisfy the weak willingness requirement.
This means that the version of the willingness requirement that is apt to avoid recklessness is different from the version that is apt to identify relevant groups and patterns. Neither the agency-first conception nor the behaviour-first conception of group-based reasons can claim both advantages. The agency-first conception engages in overkill regarding recklessness, thereby wrongly excluding opportunistic participation. The behaviour-first conception fails to explain why bizarre agglomerations are not relevant groups or patterns. This fact suggests that we cannot appeal to both kinds of argument in favour of any one version of the willingness requirement. Still, since we have rejected arguments for the willingness requirement based on recklessness, we may take the foregoing observation to motivate preference for the strong willingness requirement in particular, and hence to support the agency-first conception of group-based reasons. Though it does not offer a good explanation of concerns with recklessness, it may seem to offer the best explanation of the irrelevance of intuitively irrelevant groups or patterns.
Any assessment of this line of thought will trade on background assumptions about which groups and patterns really are irrelevant. In seeking to argue that concerns about irrelevance do not support any form of willingness requirement, I will adopt a two-stage approach. The first stage is to identify the resources for explaining irrelevance that are internal to the powers-first conception. My aim here is to show that these resources explain a surprising amount of irrelevance, and that they do so in a way that pleasingly mirrors the view most of us adopt about parallel questions concerning act-based reasons. But this will not satisfy everyone, since it implies that some groups or patterns are relevant that, some will think, just cannot be. Thus the second stage is to emphasize a point made earlier, that the three conceptions of group-based reasons identified here are not complete theories: they each describe a partial theory, or a family of similar theories. In particular, each is compatible with additional necessary conditions for the eligibility (and so relevance) of patterns or groups. Those who are not satisfied with the explanation of irrelevance offered at the first stage should consider resources for explaining it that are external to the powers-first conception, but nevertheless compatible with it.
First, then, consider what the powers-first conception can explain by itself. 14 An underlying idea of the powers-first conception is that we should think of the eligibility of patterns, in the case of group-based reasons, in broadly the same way that we think of the eligibility of actions, in the case of act-based reasons. In both cases we are considering what is required for some possible favoured item (a pattern of action, or a single action) to be ‘eligible’ – where that means that the fact that the item is favoured provides a reason for doing something (performing part of the pattern, or performing the single action). There is a degree of theoretical elegance in treating the answers to these two questions about eligibility in a somewhat uniform way, if that can be made to work.
With regard to act-based reasons to perform individual actions, most of us think that a necessary condition of eligibility is that the agent could perform the action in the circumstances. This is the act-based reason analogue of the powers requirement; the powers-first conception treats this requirement as a necessary condition of the eligibility of patterns for group-based reasons. Though this is, we might think, a very minimal requirement of eligibility, it does explain some irrelevance. As already noted, in the case of act-based reasons it explains why I do not have a reason to do the following: cure cancer by clicking my fingers. This logically possible action is certainly favoured, since it would be extremely beneficial. But, of course, the fact that it is favoured does not suffice for me to have a reason to do it. And the explanation for this seems to be that I could not do it; I lack the power to do it. Similarly, this requirement explains why I do not have a group-based reason to play my part in the following logically possible pattern of action: along with other members of Oxfam, eliminate global poverty. Though the members of Oxfam taken together have the power to achieve much, they do not (I am assuming) have the power to eliminate global poverty. So even though this possible pattern is favoured, this fact does not suffice for me to have a reason to play my part, since the relevant group lacks the power to eliminate poverty.
Thus, even the sparse resources strictly internal to the powers-first conception explain some of the irrelevance that we wish to explain. Still, we might not think that they explain enough. One source of such concern lies with irrelevant elaborations or extensions. This concern may seem particularly pressing with respect to group-based reasons. Suppose, for example, that you and I could fix the fence. We have the power to do this, since one could hold the posts while the other fixes the panels. However, so far as the powers requirement goes, it is also true that a group consisting of you, I, and Donald Trump could achieve this feat. One of you and I could hold the posts while the other fixes the panels, and Donald Trump could do something unrelated. Are there then two patterns that supply me with group-based reasons in this case: one performable by you and I, and the other performable by you, I, and Trump? Worse, are there as many eligible patterns and reasons as there are possible elaborations and extensions of the pattern consisting of you and I?
Though it is less obvious, a similar worry arises for act-based reasons to perform individual actions. These too can be elaborated and extended in irrelevant ways. Suppose that I have the power to keep my promise to return the book that you lent to me, by leaving it on your desk. I have an act-based reason to do this, because keeping my promise is a favoured possible action that I have the power to perform. But then, so is keeping my promise by leaving the book on your deskwhile whistling. So too is leaving the book on your desk and going to the cinema tonight. All of these are things that I could do and that would be favoured, since they are all ways of keeping my promise.
Again, it is attractive to deal with both sets of problems in a uniform way. In relation to act-based reasons to perform individual actions, we are likely to say that a criterion of practical relevance supplements the requirement that the agent could do the thing in question. We ignore practically irrelevant elaborations (whistling) or extensions (going to the cinema) of options, to avoid double-counting reasons. These additions are practically irrelevant in the sense that they do not contribute anything to the achievement of the aims or goals at hand. The very same criterion explains why I do not have a reason to fix my fence that is associated with a pattern containing Trump’s actions: his action is a practically irrelevant extension to any pattern of action in which the fence gets fixed. Whether we are considering act-based reasons or group-based reasons, the criterion of practical relevance operates against the background of some assumed set of relevant favouring considerations. This criterion seems to be very important in explaining our thoughts about act-based reasons, and it seems equally powerful in explaining why some patterns and groups are ineligible to support group-based reasons. It is either implicit in the powers-first conception, or else a very minor extension to that conception which gains support because it treats patterns and individual options in a uniform way.
Still, some will think that the following claims could both be true:
- Agent A is a member of some group G the members of which could perform some practically relevant pattern P.
- A has no normative reason (dependent on the facts reported in claim 1) for performing her part in P.
For example, suppose that one way for my fence to get fixed would be for my two neighbours, Jerry and Elaine, to come round and fix it for me. Strictly speaking, I have a relevant part to play in this pattern: I would need to open the door for them, and not get in their way. Thus there is a possible pattern, involving me, in which my fence gets fixed by Jerry and Elaine; this pattern is favoured and practically relevant, since it would result in the fence being fixed; and all agents involved could perform their parts in it. Yet, we might think, I do not have a group-based reason associated with this pattern to open my door and stand aside. And the reason for this is that Jerry and Elaine could not care less about my fence and are not about to come round.
One possible response on behalf of the powers-first conception is to accept the implication that there is a group-based reason in this case, and to try to swallow it. In fact, this is my own preferred response. There are at least three things to be said in favour of it. One is that, of course, it would be absurd for me to open the door and stand aside, given that Jerry and Elaine are not about to come round. But note that I have a powerful act-based reason not to do this: doing it would achieve nothing, given their unwillingness to participate in my fanciful fence-fixing scheme, and it would have opportunity costs. As with the analysis of recklessness, we can account for the significance of unwillingness by pointing to the agent’s act-based reasons. Second, there are some grounds for being wary of what Schroeder calls ‘negative reasons existentials’: that is, claims to the effect that, in some case, there is no reason to do something or other. As Schroeder points out, our intuitions about such claims are likely to be misled by the standing Gricean presumption that, when we discuss what reasons there are, we mention only the relatively weighty ones (Schroeder 2007, p. 92–97). 15 Finally, suppose that you observe me opening the door and standing aside, and you are puzzled about what I am doing. It would make my action somewhat more intelligible if you learned that I took myself to be playing my part in the pattern involving Jerry and Elaine. This is some evidence for the claim that there is a weak normative reason for me to do this – albeit a reason that is greatly outweighed, in the circumstances. It would make some kind of sense for me to behave in that way, even if not much.
For those who do not accept this response, it remains open to the advocate of the powers-first conception, of course, to cite additional necessary conditions for eligibility. Clearly this would be to appeal to resources external to the powers-first conception itself. But so long as the conditions do not include any form of the willingness requirement, the powers-first conception will remain distinct from the agency-first and behaviour-first conceptions. One possibility would be to adopt some perspectival view of normative reasons, such that what normative reasons an agent has depends in an important way on how things seem to her. Along these lines we might introduce some constraint on eligibility based on salience or framing. In the case involving my fence this may help to explain why the pattern involving Jerry and Elaine is ineligible, on the supposition that this pattern is not salient to me in the circumstances. Whether or not the appeal to salience could explain all of the intuitions about irrelevance that some may have would depend on psychological facts about what is salient to agents. 16
Other additional necessary conditions for eligibility are possible, of course. One interesting possibility is normative conditions on eligibility. Which patterns are eligible may be said to depend on what rights individuals or groups have, for example. Here I do not wish to commit to any of these possible additional necessary conditions. Instead, I wish merely to point out that we cannot properly motivate any form of the willingness requirement simply by noting the implications for eligibility of the sparse resources internal to the powers-first conception. To draw that conclusion would be to take for granted that the only additional necessary conditions worth considering are those expressed in one or other form of the willingness requirement. We should not take that for granted.
The willingness requirement does not earn its keep in a theory of group-based reasons. It is not the best way to avoid recklessness, and it is not at all obvious that it plays an essential role in determining the relevance of groups or patterns. Absent further argument to the contrary, we do not have sufficient theoretical grounds to adopt either form of it.
6. The Significance of Willingness
The powers-first conception denies that facts about willingness are relevant to the existence of group-based reasons. But of course it does not follow that such facts have no significance at all. The powers-first conception attributes other kinds of significance to them.
First consider the significance of facts about willingness in the weak sense. Recall that we are supposing that a theory of group-based reasons will be part of a pluralist theory of reasons according to which there are also act-based reasons. Whether others are willing to behave in a way necessary to realize the favoured pattern will obviously matter a great deal to the consequences of the agent’s act of participation, and so those facts will matter greatly to her act-based reasons. In particular, if they will not act in the required way she may well have very strong act-based reasons not to play her part in the best possible pattern. If so, they may well defeat her group-based reason to participate.
This means that facts about willingness are very likely to be significant to the relative strength of the agent’s group-based reasons, or to the likelihood that she should act on them. This perhaps offers an error theory of widespread attraction to the willingness requirement: in cases where others are unwilling, group-based reasons are likely to have low relative strength, and this may be mistaken as their non-existence. But it is important to remember that we noted some cases in which group-based reasons seem not to be defeated by unwillingness. In cases where the cost of participation is low or nil, the agent’s group-based reasons may determine what she ought to do or explain what she in fact does. This may be true for example of Sereeta, of people who reduce their carbon emissions despite knowing that others will not, and of people who refuse to participate in practices they deem to be unethical (such as trade in unethical goods).
The powers-first conception can also recognize the significance of facts about willingness in the strong sense. Whether others in the group have the requisite intentional states for the performance of the pattern to constitute an exercise of agency matters greatly. It matters for what exists and what happens. It matters also for what the group, considered as an agent, could do. It matters for what the group can be responsible for. And it may well bring into existence different kinds of reasons, that depend for their existence on the existence or capabilities of the group agent – for example, reasons of authority that depend on what the group can decide (Lawford-Smith 2015). Nothing I have said here is intended to deny or diminish the significance of these things. The point is only that these are distinct matters from the existence of group-based reasons as I have defined them.
The three conceptions of group-based reasons identified here exert a powerful influence over discussions of such reasons. If we fail to reflect on them we are likely to exclude certain theoretical possibilities without giving them sufficient thought. It would be far better to examine their relative merits and demerits explicitly.
This paper has aimed to initiate and contribute to that examination. I have argued that Sereeta does not make a mistake. Even though the possible pattern of action by which a Purple government would be elected will not be realized, she has a group-based reason to play her part in that pattern. Normative group-based reasons can exist in uncooperative contexts like hers, and as a result they can motivate people like Sereeta, and thus explain their actions. The ethical thought ‘I will do my bit even if others will not’ makes sense. Those conceptions of group-based reasons that imply that it does not make sense are, I have claimed, themselves mistaken.
If this is correct, it has many important implications. Since agents can be motivated by normative reasons, it enlarges the kinds of explanation of social action that we can give. Second, it enables us to derive conclusions about individuals’ reasons for action in cases where a problem requires collective action but others are unwilling to play their parts. These are amongst the most pressing ethical and political problems we face. Finally, it has implications for the study of phenomena associated with group agency. Whereas the agency-first conception implies a deep theoretical unity between concern with the conditions of group agency and concern with group-based reasons, the powers-first conception implies that these are distinct topics.
For very helpful discussion I am grateful to Michael Bratman, Stephanie Collins, Alex Dietz, Joseph Kisolo-Ssonko, Kirk Ludwig, Abe Roth, John Searle, Liam Shields, Joel Smith, Thomas Smith, Hillel Steiner, Raimo Tuomela, Paul Weirich, and members of audiences at the Collective Intentionality IX conference, Indiana University, Bloomington, September 2014 and at the University of Manchester, January 2015. I am also extremely grateful to two anonymous reviewers for this journal.
Bacharach, Michael (1999): “Interactive Team Reasoning: A Contribution to the Theory of Co-operation”. In: Research in Economics 53, p. 117–147.
Bacharach, Michael (2006): Beyond Individual Choice. Teams and Frames in Game Theory. Natalie Gold and Robert Sugden (Eds.). Princeton: Princeton University Press.
Bratman, Michael E. (2014): Shared Agency. A Planning Theory of Acting Together. New York: Oxford University Press.
Fischbacher, Urs and Simon Gächter, and Ernst Fehr (2001): “Are People Conditionally Cooperative? Evidence from a Public Goods Experiment”. In: Economic Letters 71, p. 397–404.
Gold, Natalie and Robert Sugden (2007): “Theories of Team Agency”. In: Peter, Fabienne and Hans Bernhard Schmid (Eds.): Rationality and Commitment. New York: Oxford University Press, p. 280–312.
Hurley, Susan L. (1989): Natural Reasons. Personality and Polity. New York: Oxford University Press.
Regan, Donald (1980): Utilitarianism and Co-operation. Oxford: Clarendon Press.
Schroeder, Mark (2007): Slaves of the Passions. Oxford: Oxford University Press.
Schwenkenbecher, Anne (2014): “Joint Moral Duties”. In: Midwest Studies in Philosophy 38, p. 59–74.
Tuomela, Raimo (2013): Social Ontology. Collective Intentionality and Group Agents. New York: Oxford University Press.
Woodard, Christopher (2007): Reasons, Patterns, and Cooperation. New York: Routledge.
Wringe, Bill (2016): “Collective Obligations: Their Existence, Their Explanatory Power, and Their Supervenience on the Obligations of Individuals”. In: European Journal of Philosophy 24, p. 472–497.
I aim to remain neutral on issues about whether reasons are facts or mental states, and on issues about whether the relevant conditions for having a group-based reason involve facts obtaining or instead believing that those facts obtain. In all cases discussed here assume that the agent has only true beliefs.
We can explain it if at least one player has grounds for predicting the other’s choice other than the assumption that s/he will act rationally (if the ground is merely that s/he will act rationally, the reasoning will be circular since each player’s reasons are assumed to be the same in Hi-Lo).
This is the strongest version of the strong willingness requirement, since it requires that the behaviour and intentional states will occur; weaker versions require only that there is some chance they will occur. For example, some versions might claim that a greater than 0.5 probability of the cooperative states is required, while others will require a higher or lower probability. Note that this is not the same dimension of ‘strength’ that separates the strong willingness requirement from the weak willingness requirement, discussed below. The weak willingness requirement also comes in stronger and weaker forms according to the probability of the cooperative states required. For simplicity, and because these issues are orthogonal to the ones I will be discussing, I will state and discuss both the strong and weak versions of the willingness requirement assuming that the cooperative states are required with a probability of 1. I am grateful to John Searle and to an anonymous reviewer for discussion here.
This conception of group-based reasons is exemplified by Gold and Sugden (2007). Tuomela (2013, p. 106) also seems to commit to it, when he claims that ‘member-level group reasons […] are collectively accepted we-mode reasons conceptually based on the group agent’s reason for action’ – on the assumption that the existence of the group agent’s reason depends on the existence of the group agent. The agency-first conception is also encouraged by the common practice of discussing group-based reasons under the rubric of the ‘forward-looking responsibility of groups’, since one then naturally assumes that group-based reasons exist only if groups are agents (and so can be responsible).
That is, unless the pattern itself requires specific intentional states. ‘Universal lying’ requires the members of the group to exhibit the intentional states necessary for lying, for example.
Thus the weak and strong powers requirements (and, in fact, the weak willingness requirement) could be met in cases where the group is what Lawford-Smith (2015, p. 230–231) has called an ‘uncoordinated group’. According to Lawford-Smith, such groups cannot have obligations as collectives because they cannot act as collective agents; for an opposing view, see Pinkert (2014). However, Lawford-Smith is discussing the obligations and abilities of the group itself, while the present topic is whether individual members of groups have reasons based on some pattern of action performable by the members of the group. One complication is that some patterns of action might themselves necessarily involve the exercise of group agency: for example, the pattern ‘the committee resolves to change its constitution’ arguably requires this. In respect of patterns like this, even the weak powers requirement requires that the members of the committee could play their parts in this exercise of agency (this is akin to the caveat noted in n. 6 about some patterns requiring certain intentional states). But note that (according to the powers-first conception) this requirement on the individual members stems from the nature of these specific patterns, not from a general requirement on group-based reasons. A further complication is that, in some cases, the members could have the requisite power only if the group itself already has the power to act as an agent: again, this might be true of the committee case. Not all cases could have this second feature, however, if individuals can create new groups with the power to act as agents. I am very grateful to an anonymous reviewer for discussion here.
Note that the claim here is conditional: intuitively, we should not think that whether Paddy has a group-based reason to play Hi depends on whether he is facing Eddy or Eddy*, since both opponents exhibit exactly the same behaviour in all circumstances, just with different motivations. We might, in addition, think that Paddy has act-based reasons of various kinds (see n. 9, below): as I will shortly explain, anyone who believes that there can be group-based reasons should accept that they can co-exist with act-based reasons in a single case. I am grateful to an anonymous reviewer for pressing me on this point.
One complication here is that, given these facts, Paddy will have a straightforward act-based reason to choose Hi. The intuitive case is thus weakened by the possibility that we are mistaken in thinking that he also has a group-based reason for choosing Hi. However, the ideal compensator cases discussed below also speak against the agency-first conception, and they do so in a way that is not subject to this risk.
It is not just market interactions that approximate Paddy’s situation. There is experimental evidence that some contributors to public goods reduce their contributions conditionally on others’ contributions increasing: see Fischbacher et al. (2001, p. 401). I am grateful to an anonymous reviewer for drawing this to my attention.
I am grateful to an anonymous reviewer for pressing me on this.
This point applies also to the motivational significance of group-based reasons: that is, we should expect it to be possible that subjects are motivated both by act-based and by group-based reasons on a single occasion. For example, it seems quite possible that Sereeta is motivated by the group-based reason associated with playing her part in the pattern consisting of the election of a Purple government, and by the act-based reason that by voting Purple she will make a small difference to encouraging others to do likewise in future elections. I am grateful to an anonymous reviewer for pointing this out.
If our concern is only that the theory should account for a pro tanto reason not to participate when doing so is reckless, then pluralism is sufficient. Perhaps we think that the theory should also be able to vindicate the judgement that in such cases the agent ought not to participate, all things considered. If so, we can add that the theory should allow act-based reasons to defeat group-based reasons in these cases.
I will discuss the powers-first conception on the assumption that it adopts the weak rather than the strong powers requirement. This makes the challenge as hard as possible, since the more demanding strong powers requirement offers additional grounds for explaining the ineligibility of intuitively irrelevant possible groups and patterns.
He writes: ‘using negative existential intuitions about reasons to evaluate ethical theories is a pervasive mistake in moral philosophy, suspect on the same methodological grounds wherever it arises’ (Schroeder 2007, p. 96).
I am very grateful to an anonymous reviewer for pressing me to say more about irrelevant patterns and groups, and for making the specific suggestion about salience. For one application of ideas of framing and salience to group-based reasons, see Bacharach (2006). I am also grateful to Michael Bratman for discussion of the possibility of additional necessary conditions on eligibility.