Joint Ought

Abstract

Sometimes, what you ought to do depends on what other people will do: whether or not Dad ought to pick the kids up from school depends on whether Mom is going to do it. This seems obvious enough. Yet, it leads to a problem. Suppose that several agents each have two options: A or B. It would be best if they all did A, significantly worse if they all did B, and worst of all if some did A while others did B. As it happens, each is going to do B, regardless of what the others do. Given how everyone is else going to act, it seems as though each of these agents ought to choose B: since if they were to pick A, they would bring about the worst possible result. Nevertheless, there seems to be something wrong here: after all, they could all pick A and bring about the best result. But how could it be wrong for everyone to do what they ought to do?1 Cases like this pose an obvious problem for act-consequentialists. Act-consequentialism says that you are morally obligated to do whatever will have the best outcome. But in cases like the above, the theory implies that everyone ought to act in ways which will collectively bring about a suboptimal outcome. While it isn't strictly inconsistent, many consequentialists find this conclusion hard to swallow.2 Slice and Patch Go Golfing: Mr. Patient needs a life-saving operation from two surgeons, Ms. Slice and Mr. Patch. If left unattended, Patient will die, though not painfully. If Slice cuts and Patch stitches, then he will survive. But cutting without stitching would cause his death to be agonizing, as would stitching without cutting. As it happens, Slice and Patch will each go golfing, regardless of what the other does. This paper defends the following solution. I will argue that, as well as individual “oughts” which apply to individual agents, there are also joint “oughts” which apply irreducibly to pluralities of agents. Moreover, in cases like Slice and Patch, individual and joint “oughts” can pull in different directions: while Slice ought not to cut and Patch ought not to stitch, they jointly ought to operate on Patient. This joint requirement corresponds to Patient's moral claim to be saved, and the surgeons' failure to satisfy it (in part) accounts for the wrong he stands to suffer. A version of this view is presented in some old papers by Frank Jackson4 and Derek Parfit,5 and has recently been revived by Alexander Dietz.6 According to these authors, “we” together can be morally required to do something, even if none of “us” is required to do our parts. I develop this view in two ways. First, I argue that we should think of joint “oughts” not as “oughts” which are held by group agents, but rather as “oughts” which are held jointly by several agents.7 This may or may not be the view which Dietz, Jackson, and Parfit meant to defend. But, for reasons I will explain, I think it should be. Second, I address the question of who wrongs whom when a joint “ought” goes unmet. I propose that if Patient is wronged by Slice and Patch's failure to act as they jointly ought, then he is wronged by one or both of them individually, even though each surgeon acts as they individually ought. Whether Slice or Patch wrongs Patient, I suggest, turns on the motives with which they act. The paper proceeds as follows. In Section II, I critique four recent attempts to solve the Slice and Patch puzzle. Section III presents my joint-ought solution, while Section IV discusses the issue of wronging. In Section V, I consider an objection to my claim that Slice and Patch jointly ought to operate, while Slice ought not to cut and Patch ought not to stitch; this objection arises from considerations about instrumental normativity. The problem posed by cases like Slice and Patch has attracted a lot of recent attention. Before I present my solution, it will be helpful to consider some existing solutions, and the difficulties they face. Niko Kolodny8 distinguishes between two possible versions of Slice and Patch. In what he calls “good-motive” versions, both surgeons want Patient to be saved, and so would do their parts of the operation if they believed that the other was going to do theirs. However, each declines because they reasonably, though falsely, believe that the other would not do their part if they were to do their own (and they don't want Patient to suffer unnecessary pain). Kolodny claims, and I agree, that it's not intuitive to think that Patient is wronged in these versions of the case. Here, his death seems akin to a natural disaster: something bad but not morally objectionable. By contrast, in what Kolodny calls “bad-motive” versions, at least one of the surgeons would rather go golfing than save Patient's life, and so would not do their part even if they believed that the other was going to do theirs. It is these versions which seem to prompt the judgment that Patient has been wronged; and, unless otherwise specified, they will be my focus here. Distinguishing between good- and bad-motive versions of Slice and Patch suggests a simple solution to the puzzle: since Patient is wronged only in the bad-motive versions, it must be the bad motive of one or both of the surgeons which accounts for this wrong. As I'll explain in Section IV, I think this is partly right. But it is not the whole story. How do the surgeons' motives explain the wrong which Patient stands to suffer in bad-motive cases? Here's Kolodny's answer. When both surgeons decline to do their parts, they each do the right thing. But if, say, Slice would rather golf than save Patient's life, and so wouldn't cut even if she thought that Patch was going to stitch, then she exhibits a lack of due concern for Patient. Due concern requires Slice to value Patient's interests appropriately; and since she cares more about golfing than saving Patient, and would pick the former over the latter if she had to choose, her motivational psychology falls short of this standard. Insofar as Slice's behavior reflects this lack of due concern, she wrongs Patient, even though she does what she ought to do.9 10 However, this seems to mischaracterize the wrong which Patient stands to suffer. To illustrate, suppose that Slice and Patch both preferred to golf over saving Patient's life a year prior to his medical emergency, back when he was in perfect health. Kolodny's view says that Slice and Patch were wronging Patient back then, since they failed to show him due concern.11 Let's grant that this is true. Even so, this wrong seems importantly different to that which he suffers when the surgeons actually leave him to die. But Kolodny seems unable account for this: since, on his view, it is solely the surgeons' lack of due concern which accounts for the wrong in both cases. Kolodny has a possible reply. Just because an individual is wronged by the same behavior or attitude in different cases, it doesn't obviously follow that the severity of those wrongs is the same. A wrong could be more severe if the behavior/attitude in question leaves the victim worse off. For example, suppose that someone drives recklessly through your neighborhood. In Scenario 1, no one is hurt. In Scenario 2, the driver seriously injures you. On Kolodny's view, it's plausible that the driver wrongs you in both scenarios: since they act with the same disregard for your safety in both. But he needn't say that these wrongs are equally severe. Instead, he could say that, while the driver's disregard explains the wrong in both scenarios, the wrong is more severe in Scenario 2 because their disregard leaves you worse off than in Scenario 1. Returning to Slice and Patch, Kolodny might say something similar in response to the preceding paragraph's worry. In both the bad-motive version where he is left to die, and the version where he doesn't need an operation, Patient is wronged by the surgeon's lack of due concern. But, as in the reckless driving example, the wrong is more severe in the first version because the surgeons' disregard leaves him worse off.12 However, the moral difference between these two versions of Slice and Patch seems to be more significant than this proposal recognizes. It's not just that Patient stands to be worse off, and thus to be wronged more severely, in the original case where he needs the operation, than in the version where he doesn't. In the original, Patient seems to have a moral claim which he lacks in the second: namely, a claim to the life-saving operation. Intuitively, it is by having this claim ignored that he stands to be wronged. Yet Kolodny cannot say this. If someone has a moral claim to X, then some other/s has/have a corresponding moral requirement to provide X.13 Since Kolodny posits no moral requirement on the surgeons to deliver the operation, he cannot hold that Patient has a claim to it. Of course, Patient may have a further, related moral claim, both when he needs the operation and when he doesn't: roughly, a claim that others value his life appropriately, which would imply a claim on Slice and Patch to value saving his life more highly than playing golf. On Kolodny's view, it is by ignoring this second claim that the surgeons wrong Patient (both when he's sick and when he's healthy). I suggest, however, that this verdict does not match our intuitive understanding of the case; if the surgeons golf when they could instead have saved Patient's life, then Patient is wronged because he has a claim to be saved. This same problem also arises for other motive-based solutions to the Slice and Patch puzzle. For instance, take Donald Regan's “cooperative utilitarianism,” according to which one ought to “co-operate, with whoever else is co-operating, in the production of the best consequences possible given the behavior of non-co-operators.”14 A co-operator, for Regan, is someone who is motivated to coordinate their actions with those of others in order to produce the best available outcome. So, in a bad-motive variant in which both Slice and Patch fail to be co-operators, each surgeon acts just as cooperative utilitarianism says they ought: since they each produce the best possible consequence given the behavior of the non-cooperating other. Nonetheless, in failing to be co-operators, Slice and Patch fail to be as Regan's theory says they ought to be.15 In a similar vein, Felix Pinkert16 argues that virtue requires not only that one's actual actions produce the best possible results, but also that one's motivational psychology is such that one would act to produce the best consequences in nearby possible worlds. The problem with these views is that if Slice and Patch wrong Patient by failing to be motivated in the requisite way, then they wrong him in this way regardless of whether he needs the operation. As such, they fail to appreciate the distinctive way in which Patient is wronged when the surgeons leave him to die: viz., by having his claim to be saved ignored. Taken on their own, claims (1)–(3) are each insufficient to account for the intuitive moral judgments prompted by Slice and Patch. Neither (1) nor (2) can account for the intuition that Patient is wronged when Slice doesn't cut and Patch doesn't stitch: since here, neither conditional obligation gets triggered. Nor will (3) be able to account for this intuition. Since it is merely evaluative, all this claim implies is that things would have gone better if Slice and Patch had operated on Patient. Yet this seems insufficient to explain why Patient is wronged; it would be better if earthquakes didn't occur near densely populated areas, but that doesn't mean that anyone is wronged when they do. Nonetheless, Estlund argues that (1)–(3) can account for our intuitions if they are taken together. Combining (3) with (1) and (2), he claims, “ties” what would otherwise be a merely evaluative ought-sentence to deontic obligation, thus generating a hybrid of deliberative and evaluative ought-claims.21 An initial worry with Estlund's solution is that it is difficult to see how exactly his notion of plural requirement is supposed to account for the intuition that Patient is wronged when Slice and Patch leave him to die. The “tie” between plural requirement and deontic obligation holds only in the event that the plural requirement is satisfied: when Slice cuts and Patch stitches, (1)–(3) jointly imply both that they do what is best, and that they each act in accordance with at least one of their obligations. However, when Slice and Patch go off golfing and the plural requirement goes unmet, all that follows is that it would be better if the surgeons were to act differently. So, while the combination of (1)–(3) might do a better job of accounting for our intuitions than (3) alone in the case where Patient is saved, it's not clear that they do so in the case where he isn't. However, this initial worry overlooks the central point of Estlund's view. He characterizes plural requirement as a kind of non-agential moral requirement: that is, a moral requirement which can be violated without any agent acting contrary to an obligation.22 The core idea is that when Slice and Patch go off golfing, Patient is wronged because the plural requirement that Slice cuts and Patch stitches goes unmet. However, neither surgeon contravenes any of their obligations. The wrong which Patient suffers occurs despite the fact that everyone acts in accordance with their moral requirements: it is a wrong without a culprit. The idea of a non-agential moral requirement, of a wrong without a culprit, is intriguing. But it is also at odds with the moral framework I am working with. On this framework, if a person is wronged, then they have a moral claim which goes unmet. A moral claim corresponds to a moral requirement on others either to provide the claimholder with a good (if it is a claim to something), or to refrain from acting toward them in certain ways (if it is a claim against something). So, if someone is wronged, then there must be at least someone who has contravened a moral requirement: that is, a culprit. This framework is intuitive and can explain a wide array of moral phenomena.23 Giving it up would be a high price to pay for a solution to the Slice and Patch puzzle. Yet Estlund's solution requires that we scrap it. For Estlund, if Patient has a moral claim to be operated upon, it does not correspond to any moral requirement on any agent/s to deliver the operation; in Kolodny's words, it is a claim against no one.24 Estlund provides some independent reasons for parting ways with the moral framework assumed here. He offers two putative counterexamples to the claim that wrongs require culprits. However, I find neither persuasive. First, he cites certain views of distributive justice according to which “distributive patterns of social goods can be unjust whether or not there is anything anyone ought to do about it.”25 But this just begs the question. Estlund is simply pointing out that there are theories which imply that there can be wrongs without culprits. Yet this will not convince anyone who finds that implication implausible.26 As his second counterexample, Estlund claims that “there are familiar statements such as, ‘One of those two people ought to help’, where this is meant not to ascribe an obligation to any particular person, but only to a disjunction of persons (so to speak).”27 For instance, suppose that a swimmer gets into trouble off of a crowded beach; one of the sunbathers ought to jump in and save him, but it needn't be anyone in particular.28 Estlund claims that in examples like this it is not the case that any particular individual ought to help. Yet if no one were to help, then the swimmer would be wronged; in that scenario, he says, “[v]iolation takes place without a culprit.”29 To summarize, Estlund proposes that when Slice and Patch go off golfing, Patient is wronged even though no agent/s contravene/s any moral requirement. This requires us to scrap an intuitive and explanatorily powerful moral framework, according to which moral claims imply corresponding moral requirements on agents; and I suggest that Estlund gives us insufficient reason to do this. Christopher Woodard31 argues that there are two fundamentally distinct kinds of reasons for action. Act-based reasons count in favor of an action by virtue of the value of that action itself or its effects. Someone has an act-based reason to do something, on Woodard's view, just if it would be good (and it's possible) for them to do it. By contrast, pattern-based reasons32 favor an action by virtue of the value of a pattern of actions of which that action could be a part. In order for an agent to have a pattern-based reason to do something, it needn't be the case that it would be good for them to do it. Rather, it needs to be possible for that action to be part of a wider pattern of actions which would be valuable if enacted. Given that Slice and Patch are both going golfing, neither has an act-based reason to do their part of the operation; indeed, each has an act-based reason not to do their part. Nevertheless, the surgeons could still have pattern-based reasons to do their parts: while neither action would do any good on its own, the possible pattern comprised of Slice's cutting and Patch's stitching would produce a valuable outcome if it were enacted. Perhaps this can help to resolve the Slice and Patch puzzle. I've said that it's intuitive that Patient has a moral claim to receive the operation, and that we should expect this claim to imply some corresponding moral requirement/s on some agent/s to deliver the operation. Given Woodard's theory, we could say that Patient's claim corresponds to a pattern-based reason for each surgeon to do their part of the operation. Of course, we also want to say that, all things considered, they ought not to do their parts, and so the pattern-based reasons must be outweighed by the countervailing act-based reasons. So the resulting proposal would be this: while Slice and Patch each have most reason to not-cut and not-stitch, they wrong Patient by doing so, because they act contrary to their pattern-based reasons which are imposed on them by Patient's moral claim to the operation. However, this proposal faces several difficulties. First, on what is perhaps the most plausible version of Woodard's theory, Slice and Patch do not have pattern-based reasons to do their parts of the operation. For Woodard, whether or not this is the case depends on whether some version of what he calls the “willingness requirement” is correct. This says that in order for an agent a to have a pattern-based reason to play her part in some pattern P, there must be some chance that P would occur if a were to play her part in it.33 In earlier work, Woodard rejects all forms of the willingness requirement.34 However, in light of criticisms from Dietz,35 he has since endorsed a minimal version of it.36 But if the willingness requirement holds, then the surgeons will lack pattern-based reasons to do their parts of the operation: since if either were to do their part, the other would not do theirs, and so the valuable pattern (cutting-and-stitching) would not occur. Let's suppose, then, that the willingness requirement is false, as per Woodard's earlier theory: Slice and Patch each have pattern-based reasons to do their parts of the operation, despite the fact that no operation would occur if either one were to do their part. Yet even so, it is unclear why we should think that the surgeons stand to wrong Patient by contravening these reasons, when their doing so complies with their stronger act-based reasons not to do their parts. Granted, there are cases in which it can seem plausible that one person wrongs another by acting on stronger reasons.37 But such cases have a different structure to Slice and Patch. For instance, suppose that I have promised you, a sickly hospital patient, that I will smuggle you cigarettes. I may have most reason not to smuggle the cigarettes, since they will be very bad for you. But perhaps I also wrong you by not-smuggling, since I promised you that I would. If this is so, then, presumably, it is at least in part because my not-smuggling sets back an important interest of yours, viz., your interest in others respecting the promises they have made to you.38 However, if, say, Slice acts on her stronger act-based reason and declines to cut, then this does not set back any of Patient's interests; had she chosen instead to cut (in accordance with her pattern-based reason), Patient would still have died, just more painfully. How does Slice stand to wrong Patient by not-cutting when (a) this is what she has most reason to do, and (b) her doing so sets back none of his interests?39 Now suppose, finally, that this issue can be resolved: Patient has a moral claim on the surgeons to act on their (weaker) pattern-based reasons, and so is wronged by them if they contravene those reasons. Notice, however, that (assuming the willingness requirement is false) Slice and Patch not only have pattern-based reasons to cut and stitch. They also have pattern-based reasons not to cut, and not to stitch. While Slice's cutting is part of a valuable, possible pattern (cutting-and-stitching), it is also part of a seriously harmful, possible pattern (cutting-and-not-stitching); and the same goes for Patch's stitching. So, if the surgeons wrong Patient by failing to comply with pattern-based reasons, then, counterintuitively, they would wrong him by both doing their parts of the operation and saving his life; whether they do their parts or not, the surgeons fail to comply with some pattern-based reasons. There's one more possible solution I'd like to discuss. This solution adds an extra protagonist to the story: in addition to Slice, Patch, and Patient, we also have the group agent comprised of Slice and Patch (call it “Slice-and-Patch”). It is this fourth agent, the solution says, which bears the moral requirement corresponding to Patient's moral claim to the operation. That is, while neither Slice nor Patch ought to do their individual parts, Slice-and-Patch ought to undertake the operation. So, when Patient is left to die, it is not the behavior of the individual surgeons which wrongs him, but rather that of the group agent, Slice-and-Patch.40 This solution is attractively neat. But is it plausible that Slice and Patch comprise a group agent which can bear moral requirements? Maybe in some versions of the case. Christian List and Philip Pettit41 argue that a group of agents counts as an agent in its own right if it possesses both representational and motivational states, and is able to intervene in its environment on the basis of those states. In order to be ascribed representational/motivational states, they claim, a group needs to have in place some sort of organizational system which aggregates the states of individual group members. So, on the List-Pettit view, Slice and Patch will comprise a group agent only in situations in which they have some procedure which outputs group beliefs/desires as a function of their own individual beliefs/desires.42 However, the intuition that Patient stands to be wronged seems to persist in versions of the case in which Slice and Patch lack such an organizational procedure. For instance, suppose that the two surgeons have never met. Both of them are standing outside the hospital on a coffee break when Patient suddenly falls to the ground in front of them. The problem is common knowledge between Slice and Patch: he needs immediate surgery or else he'll die (though painlessly). Furthermore, it's also common knowledge that Slice is able to cut and that Patch is able to stitch. But, as it happens, both of them are going to stand there drinking their coffees, regardless of whether the other attends to Patient or not.43 Here, the List-Pettit view says that Slice and Patch do not constitute a group agent. Nonetheless, it still seems as though Patient is wronged when he is left to die. An advocate of the group-agent solution might respond by endorsing a more permissive theory of group agency.44 Such a theory would have to say that a concatenation of agents can comprise a group agent even if they lack any procedure for aggregating individual group-member beliefs/desires, and so lack any determinate basis on which they can be assigned group beliefs/desires. But this response seems a little ad hoc. What reason do we have to believe in such a permissive theory of group agency, beyond the fact that it helps the group-agent solution to avoid the above objection? An alternative response is to grant that Slice and Patch do not comprise a group agent in the coffee-break case, but to claim that they each bear a moral requirement to take steps to form one.45 So, when the surgeons stand around sipping their coffees, both fail to act as they ought, and thereby wrong Patient. However, this just defers the problem. Presumably, it is not true that Slice and Patch each unconditionally ought to take steps toward forming the group agent Slice-and-Patch: since it would be futile for only one of them to do so. But if Slice ought to take these steps only if Patch will, and Patch ought to take them only if Slice will, then neither of them will fail to act as they ought when they both decline to form a group agent. I've rejected four attempts to solve the Slice and Patch puzzle. This section introduces an alternative: the joint-ought solution. The central idea is that the requirement to operate on Patient is possessed not by any one thing, but rather jointly by Slice and Patch. It is this joint requirement which corresponds to Patient's moral claim to the operation. To elucidate this idea, consider some ways in which several things can possess a property. Sometimes, when we say that multiple objects possess a property, what we mean is that they each possess that property individually. For example, “these daffodils are yellow” is best understood as being synonymous with “this daffodil is yellow, and that daffodil is yellow, etc.” However, in other cases this is not what we mean. For instance, think of a shady glade in a forest. It's true that the trees shade the glade. But it's false that any particular tree shades the glade; for any given tree, it shades only a small section. So, there must be another way of understanding this sentence. One option is to posit a compound object as the referent of “the trees,” and to say that it is this object which shades the glade. But it's unnecessary to proliferate objects in this way. Instead, we can take seriously the grammatical plurality of “the trees:” we can suppose that this noun-phrase refers to several things, and that there is a single instance of the property shading-the-glade which is possessed by them. In other words, we can say that the trees jointly, or together, or collectively shade the glade.46 Similarly, I want to suggest that the property ought-to-φ47 can be jointly possessed by several agents.48 There are, I think, some fairly unproblematic examples of this. For instance, suppose that Teacher 1 and Teacher 2 share the marking for a class. It might be true that they ought to complete all of the marking before Friday. But it's false that either one of them ought to do this: Teacher 1 ought only to complete her share of the marking by Friday, and likewise for Teacher 2. This case is similar to that of the shady glade. The trees jointly shade the glade by virtue of the fact that each one of them shades some portion of it. Similarly, it seems that the teachers jointly ought to complete the marking before Friday by virtue of the fact that Teacher 1 ought to complete her portion by Friday, and Teacher 2 ought to complete his by Friday. However, I want to propose something bolder. I want to suggest that it could be true that several agents jointly ought to do something, even if each one of them ought not to do their parts. If this is true, then it enables the following solution to the Slice and Patch puzzle. We can grant that (given what Patch will do) Slice ought not to cut, and that (given what Slice will do) Patch ought not to stitch. Nevertheless, we can say that Slice and Patch jointly ought to operate. As such, while neither surgeon fails to do what they individually ought when they go off golfing, they fail to do what they jointly ought to do; and it is this joint failure which (in part: see Section IV) accounts for the wrong which Patient suffers. This solution assumes that Slice and Patch has a structure different from that of the marking example: since, on this view, it can't be that Slice and Patch jointly ought to operate by virtue of the fact that Slice ought to cut and Patch ought to stitch.49 Is this a problem? Not necessarily; not all instances of joint property possession share the structure of the shady glade and marking examples. For instance, think of a couple owning a house together. The two spouses jointly own the house. But this isn't so by virtue of Spouse 1 owning one half of the house, and Spouse 2 owning the other: their ownership, we might say, is brutely joint. On the view I'm putting forward, Slice and Patch's jointly possessing the property ought-to-operate is like this. Just as Spouse 1 and Spouse 2 jointly own their house without either one of them owning any particular bit of it, Slice and Patch jointly ought to operate on Patient despite the fact that neither of them ought to do their bit.50 I think this is the most plausible version of the view defended by Dietz,51 Jackson,52 and Parfit,53 according to which groups, “we” together, can be obligated to act. As formulated by them, the Dietz-Jackson-Parfit position is ambiguous between the group-agent and the joint-ought solutions. Both solutions claim that there is a sense in which the deliberative reading of “Slice and Patch ought to operate” is true.54 The difference, however, lies in who bears this requirement. The group-agen