A. The Argument from Intrinsic Duties

Consider the following argument by Raz:Footnote ⁴

I'll show now that this argument fails as an argument against Hohfeld, since it is not valid once we restrict our attention to claim-rights (rather than rights) and directed duties (rather than duties).

In support of 1-P1, Raz claims that there are duties to express respect for value, e.g., the value of art. Such a duty would be derived from the value of art. Yet, Raz suggests, there are no rights correlative to such duties: art does not have any rights, and the duty to respect art does not obviously derive from anyone's right. But the immediate problem for this as an argument against correlativity is that the duty to respect art is not owed to anyone, certainly insofar as it is a duty not derived from anyone's right. Since this is not a directed duty, it lends support to the argument only insofar as 1-P1's mention of “duties” does not include directed duties. Substitute “duties” with “directed duties” and there is no longer good evidence for 1-P1.

Requiring that the argument refer to “directed duties” rather than “duties” is clarifying rather than question begging. It would be question begging against Raz if Hohfeld equivocated on whether it is duties or directed duties that are correlative to rights. But (this is the clarification) the Hohfeldian claim I am defending is that directed duties are correlative to rights. And this is not the tautologous claim that duties correlative to rights are correlative to rights. So Hohfeld is not itself question begging, since it must be defended against the evidence that there are directed duties that are not correlative to rights.

Raz presents another bit of evidence for the argument: that there are duties of friendship, i.e., duties that are constitutive of friendship. Perhaps an example of these is the duty to listen to a friend vent about a bad day. It does not sound strange to say that you owe it to your friend to listen to them, i.e., that the duty is directed. Yet, Raz suggests, your friend does not exactly have a right against you to listen to them. So perhaps this provides a more pressing counterexample to correlativity.

But it's not at all clear that 1-P2 is in fact satisfied. Part of the strangeness of saying that your friend has a right that you listen may come from a tendency to associate rights with enforceability. Certainly your friend should not force you to listen, or use force in some other way to ensure you satisfy your duty. That would be so heavy-handed as to destroy any basis for the friendship, and appears in conflict with the supposed constitutive aspect of the duty. Of course, the argument need not insist that the duty is constitutive, just that it is a duty that arises with friendship. More to the point, once we clarify that the right in dispute is a claim-right, one that confers standing to claim the direct object of the right, then it seems altogether more ordinary to think that your friend does in fact have such a right. If your attention is caught between your friend and an inane TV advert, it seems quite ordinary to say that your friend is the one who gets to say which you should listen to. Many everyday tiffs between friends (“Hey, I'm talking to you!”) presuppose just this sort of standing.

Raz may have something more subtle in mind, namely that the duty to listen to your friend really does constitute your relationship as a friendship, but that your friend's standing to claim your attention and complain about its absence does not. The idea here might be that friendship is a normative phenomenon involving a special kind of ought, the ought of friendship, and recognizing these oughts makes friendship the kind of relationship it is. If one does not recognize this kind of ought, then one is not properly in a friendship. But the friend's standing to complain is not constitutive of friendship in quite the same way, so the constitutive duty of friendship is not correlative to a right that is similarly constitutive. Now, I suspect that insofar as there are norms constitutive of friendship, they are the relational kind that import the idea of standing to claim. But setting my suspicion aside, if the interpretation of Raz I am proposing here is right, he should be understood as saying that the constitutive duties of friendship are not directed. This would disqualify the argument as an argument against Hohfeld. So what we do not yet have is an argument that there are duties constitutive of friendship that are directed, and thus import the idea of standing, and yet cannot be understood as correlative to rights in the sense described in the previous paragraph.

The next three arguments are more clearly targeted at directed duties in particular, and are more pressing for correlativity. They each proceed by arguing that for some intuitive example of an entitlement, the relevant correlative right or duty would be implausible. But each counterexample can be rejected by making sure that we describe the correlative right or duty at the same level of abstraction as the intuitive incident with which we begin. For each counterexample we will need to acknowledge a dimension of abstraction along which entitlements may vary in their description. By the end, we'll see that entitlements can exhibit these various forms of abstraction in their subject, direct object, and indirect object. The first two dimensions of abstraction are not novel in the literature, though I will employ distinctive terminology in order to keep each apart.

B. The Argument from Dynamic Rights

The second argument is also due to Raz, who notes that

The form of the argument is as follows:

In favor of 2-P1, Raz thinks that there is a right to security that generates lots of particular duties (such as a duty not to assault, a duty not to rape, etc.). However, in favor of 2-P2, none of these particular duties is correlative to the right to security. And that appears to be a problem for correlativity, which we imagine to be a one-to-one relation. Here we have one right, but many duties.

As Kramer notes, this argument suffers from ignoring the degrees of particularity of the relevant rights and duties. That is, Hohfeldian Correlativity concerns

Kramer's first thought is that entitlements vary along a dimension of generality versus particularity. A claim-right can be more or less particular in the sense that it includes more or fewer actions in the scope of its direct object; for example, S's right against T to security requires that T abstain in a variety of ways that guarantee S's security, whereas S's right against T to not be assaulted requires that T only abstain in those ways that guarantee S's not being assaulted. But the same is true of directed duties. T's duty not to assault S requires that T abstain from doing those things that amount to assaulting T, whereas T's duty to guarantee S's security requires that T abstain from doing those things that undermine S's security.

The further thought is that Hohfeld should be precisified as the claim that rights and duties of the same generality (or particularity) are correlative. Now there seems to be no barrier to saying that there is a duty to guarantee S's security correlative with S's right to security. That is consistent with thinking the right to security does generate more particular duties, such as the duty not to assault; but here again there is no barrier to thinking that there is a more particular correlative right not to be assaulted.

The reasons we might not think in these terms immediately are practical. The terminology of “right” has been useful for various movements that advocate for protections against the state and other individuals. This advocacy naturally frames the claimed protection in general terms, corresponding to the ambition of the advocates—it is usually left to policymakers who do not lean on the language of rights to refine the claimed protection. Similarly, the soaring language of a Bill of Rights will tend to invoke protections of a very general kind. It will also seek to state the rights as pithily as possible, and this will mean using an abstract noun (e.g., “security”) as the direct object of the claimed right. But an abstract noun is likely to cover a wide range of particular actions. Meanwhile, a countervailing pressure arises in our talk of duties, which turns our attention toward thoughts about what we and others must do. When we think in this way, we tend to want to know exactly what the duty-bearer is supposed to do. Furthermore, syntax does not allow us to put an abstract object in the direct object place of a statement of duty.

So there is rhetorical and syntactic pressure to frame rights in general terms and duties in particular terms, and this will make examples like Raz's seem familiar. But there is no conceptual reason we are limited to using an abstract object in the direct object place of a statement of rights; and indeed it would serve conceptual clarity if we did not do so in our philosophical discussion of rights. That is, the reasons for framing rights generally are rhetorical, and so we should discount our tendency to do so when doing rights theory. So there is no barrier due to rights theory or normative theory that prevents us from articulating rights and duties with the same degree of generality or particularity, and finding them correlative.

C. Specified vs. Unspecified Entitlements

A different problem for Hohfeld is exemplified by a second example that Raz uses to support 2-P2 of the Argument from Dynamic Rights:

Raz's point seems to be that the right to political participation is abstract not in the sense that it does generate a variety of more particular duties, but in the sense that it could generate more particular duties depending on the social context. To use my terminology, the right is not just general rather than particular, but unspecified rather than specified. Specification involves interpreting a general right in light of the social context in order to produce one particular duty (rather than many). But this means that we cannot defend Hohfeld by producing a similarly particular right that is the duty's correlative, since the unspecified right has its own existence. The problem for Hohfeld is that this unspecified right seems to lack a correlative.

MacCormick also points to this sort of problem when he discusses Scotland's Succession Act of 1964:

MacCormick argues that the right of the survivors to the intestate estate lacks a correlative duty since the right vests in them before the appointment of an executor. At that point, nobody has a duty to pass the estate on to the children, since there is no administrator and the intestate is deceased. So there can be no correlative duty. MacCormick also points out that the children's pre-appointment right is a genuine right with ramifications, since “one who has a right in a solvent estate is preferred to other candidates when it comes to being confirmed as executor.”Footnote ⁸

The response to this problem rests on my earlier distinction between specified and unspecified entitlements, which has also been remarked upon by Kramer.Footnote ⁹ Add to this a further precisification of Hohfeld: that only rights and duties of the same level of specificity and generality are correlative. Raz and MacCormick's apparent counterexamples can now be understood as evidence for the claim that the precisified version of Hohfeld isn't true of an important set of claim-rights, i.e., the unspecified ones. What completes the response to their objection is the claim that there are correlative unspecified duties. But does that make sense?

Consider first Raz's case. Correlative to the right to political participation is a duty to do what secures (or enables) political participation. Perhaps this is specified in the twentieth century by a duty to respond to a request under the Freedom of Information Act, and in the twenty-first century by a duty to allow access to the President's social media feed, though neither of these duties could make sense in ages and places that lack the apparatus of social media feeds and FOIA requests. Still, we might abstract from these particular duties and point to a duty to do whatever secures political participation, i.e., a duty that is unspecified in its direct object.

Here again we can point to an error theory to explain away the apparent failure of correlativity. Duties concern what particular people are to do, and so we are practically inclined to talk of duties with specified direct objects. But we often refer to a right in order to describe a sphere of benefit that is secured for a person, or a sphere of activity that is protected against interference. So here we are much more comfortable describing the right in terms that leave the direct object of the right unspecified. But these practical inclinations do not imply any conceptual constraint on describing a right with an unspecified direct object as correlative to a duty with a similarly unspecified direct object.

The question that remains is whether we can extend this response to MacCormick's example, which concerns a right that is unspecified in its indirect object rather than its direct object. I turn to this concern next.

D. The Non-Instantiation Argument

In MacCormick's case, we observe that the survivor child's right needn't be entirely unspecified insofar as it consists in a negative and in rem right against everyone that the child inherit. But if it goes beyond this to include a positive and in personam right that someone take steps to distribute the estate with respect to the child's share, then it is unspecified in a further way: it is unspecified in its indirect object, i.e., the person against whom the child has a right that the estate be properly distributed. After appointment the person with this responsibility will be the particular person who has been appointed executor, whoever that is. This last sentence shows us how to say who has the responsibility before appointment: whoever will be appointed as executor. The wh-word here is an unspecified indirect object of the right, and it can take the subject place in a correlative duty: whoever will be appointed executor has the duty to distribute the estate with respect to the surviving child's share.

MacCormick's counterexample thus differs from Raz's in that it presents a right with an unspecified indirect object, whereas Raz's right has an unspecified direct object. MacCormick's case points to a more general Non-Instantiation Argument:

Note that there is a correlative version of this argument on which there is some right held against someone under an indefinite description, and that MacCormick's apparent counterexample is an instance of this correlative argument. But I frame the Non-Instantiation Argument in terms of directed duties owed to indefinitely described indirect objects because it can be used to generate further apparent counterexamples to Hohfeld, such as:

Wrongful Life is an instance of the Non-Instantiation Argument only if S's duty is in fact directed. But if one agrees that T lacks a right because T was not around at the time of S's decision, then one may resist the thought that S's duty is owed to the child or anyone else. This way of resisting the Non-Instantiation Argument is suggested by Kamm:

Kamm's claim relies on the thought that the duty to C is based on characteristics of D. Yet it is unclear whether it makes sense to think that a duty owed to C can be justified in a way that makes no reference to characteristics of C.

Whether Kamm's solution works or not for the case of D, the mother's duty in Wrongful Life does not seem to be of the same sort as D's duty. Perhaps the mother has reason to use birth control because of the impersonal value of refraining from increasing the amount of suffering in the world, and this sort of reason does not underwrite a directed duty. But she also has reason to use birth control because of the suffering she would impose on her future child, whoever that might be. That sort of reason, which is unlike the reason for D's duty, does underwrite a directed duty, since it accompanies the special standing of the child (whoever she might be) to blame and claim moral repair from the mother. Here again our syntactic apparatus of wh-words seems more than capable of bearing the weight of this thought, and it makes perfect sense to say that the mother's duty to use birth control is owed to whoever would otherwise be born. Moreover, the wh-word makes syntactic and semantic sense in the subject place of a correlative right: whoever would otherwise be born has a right that the mother use birth control. If this locution is at all awkward it is only because we do not ordinarily need to be so precise in articulating the normative relation, and not because we are contorting our language.

Now consider the following objection: if child T were never born, then T could not enjoy the benefits of having the right, so T's having a right is pointless, so we should accept that S has a duty without a correlative right. This objection mistakes the proposal I have made. On my view, if T were not born, then T would never have a right. The unspecified person whoever would otherwise be born is the right-holder, and that unspecified person would enjoy the benefits of having the right (were they born), such as having standing to blame and claim repair.

Here's a stronger form of the objection: if no child were ever born, then no child could enjoy the benefits of having the right, so the unspecified right is pointless, so we should accept that S has a duty without a correlative right. But the inference from nobody enjoying the benefits of the right to the right being pointless is bad. By this I don't mean to resist the thought that the point of a right is that it benefits someone. Rather, I want to resist the thought that the point of a right is that it benefits some specified person. The unspecified right, i.e., the right held by whoever would otherwise be born does the work of focusing S's mind on why she should take her duty seriously. It does this in the same way that S's directed duty does—by pointing to someone to whom S will be accountable if she violates the duty. We may think that the language of rights is awkward here because of its reference to an unspecified person, but the same would be true for using the language of directed duties.Footnote ¹¹ So the thought that there is a claim-right here stands or falls together with the thought that there is a directed duty—which is all that we need for Hohfeldian correlativity.

The difficulty with the proposal, insofar as there is one, is metaphysical rather than conceptual. Does it really make sense to think that a merely counterfactually existing person, i.e., someone who would have been born were the mother to violate her duty can (a) have a right, and (b) be referred to in the justification of the right? This may appear to rest on a controversial modal realism. But whether a right-holder exists, as counterpart or as some other kind of entity, is really beside the point. An entity's existence is needed neither for its having some trait (Santa Claus has a beard), nor for its usefulness in justification (we lock our doors to protect ourselves from thieves who never come). What matters for correlativity is not whether the right-holder exists, but whether we can attribute and make sense of the right. In fact, we do so routinely in altogether more ordinary contexts. We might sensibly say, for example, that everyone has a right to a clean environment, that this right extends to those who are not yet born, and that this right is the basis for apportioning part of the current budget to ensuring that we leave a clean environment behind. I don't mean to deny that there is something puzzling about right-holders who can only be specified counterfactually. But my guiding assumption is that our normative practice should determine how we think about this metaphysical puzzle and not the other way around.

To be fair, the statement that our budget should be determined by the rights of those not yet born might also raise normative challenges. In particular, it might raise the non-identity problem insofar as balancing rights depends on their justification, and the justification for the right of future people to a clean environment rests on a comparison of the harms that would be inflicted by various courses of action. But we should not jettison unspecified rights just because they lead us downstream toward a type of comparative thinking that we in any case encounter elsewhere in normative deliberation. We should be concerned only if the non-instantiation problem turns out to be another guise for the non-identity problem, since then positing unspecified rights would in itself involve us in an apparent contradiction.

But the non-instantiation problem for rights is not the non-identity problem. The non-identity problem arises because we (apparently illicitly) rely in justification on a comparison of the outcomes for two different sets of people—those who would exist if we do the right thing, and those who would exist if we don't do the right thing—but we do not include the disvalue of nonexistence for the second set in the comparison. The non-instantiation problem arises because our deliberation is guided by the rights of people who would not exist if we do the right thing. But since rights act as side constraints there is no illicit comparison between the value of states of affairs in which the right-holders do and don't exist, and no disregard for the disvalue of existence. We are simply guided by the thought that if we did the wrong thing, then we would violate the rights of those who would in that case exist.

That we are not here dealing with the non-identity problem is illustrated by another instance of the Non-Instantiation Argument:

This case is a challenge for Hohfeld because it appears that X has a duty to refrain from injuring anyone who crosses the border, but Y lacks a correlative right insofar as he does not cross the border. And since we are dealing with a case involving positive law, we are free to stipulate that Y lacks rights in advance of crossing the border. The solution I have been suggesting is that it is not Y who has the correlative right, but whoever crosses the border, i.e., an unspecified person. Now the value of this illustration of my solution is that it shows why we need to talk about unspecified right-holders even in cases in which the people specifiable as right-holders do exist—and their existence means that there is no place for the non-identity problem.Footnote ¹²

The case is also helpful because it helps to bring some alternative attempts to rescue correlativity into focus:

I think it is odd that the constituted rights strategy would have to say that X has no duty in circumstances where nobody has crossed the border—for we can reasonably stipulate that X's reasons not to set a trap have deontic force. But the constituted rights strategist could say X potentially has duties, or even that he has potential duties (and, correlatively, Y has potential rights). But heading in this direction, the difference between abstraction and the constituted rights strategy becomes faint. Perhaps we should concede that the two strategies are notational variants. That would be so if the indefinite assertion whoever is F, is G can be translated as $( {\forall i} ) Fi\to Gi$. For an instance of this schema is: whoever crosses the border has a right against harm can be translated as for all i who could cross the border, if i crossed the border then i has a right.Footnote ¹³ (Similar translations could be given for properties of the relevant rights and duties, and second-order normative incidents such as powers.) The difference, if there is one, is supposed to be one of ontological commitment. Above, I denied that we need to commit to the existence of an unspecified right-holder in order to pursue the strategy of abstraction. But the strategy does commit to the existence of a right in advance of anyone crossing the border, one that is ascribed to an unspecified right-holder; while the constituted rights strategy seems to commit only to the conditional existence of the right.

Which should we prefer? My suggestion at the beginning of the previous paragraph was that we want to be able to take into account the properties of duties when describing the particular way in which X's deliberation should be shaped by consideration of the harms his behavior would cause to whoever crosses the border. For example, the fact that X owes someone particular a duty should shape his thoughts about whether he would owe justification or apology and repair to someone in particular. The conditional rights strategist can say that X's thought about his potential duty is enough to do this work. But at this point it is not clear that the idea of a potential duty is any less ontologically committed than the idea of a duty with an unspecified indirect object. For just as an unspecified duty has an unspecified indirect object, a potential duty has a potential indirect object. I will assume then that these strategies are equal in costs and benefits, and stick to the strategy of abstraction because of its usefulness for understanding the problems raised in Section III.

E. Indefinite vs. Definite Entitlements

Entitlements (both rights and duties) that are unspecified in their subject or indirect object are sufficiently distinctive that they deserve their own terminology—call these “indefinite” rather than “definite” entitlements. When characterized objectively (i.e., independent of anyone's epistemic perspective), a definite entitlement takes names or definite descriptions as arguments in the subject and indirect object place, e.g., “I owe it to the university to pay my library fine,” or “Minh has a right of privacy against Nick.” An indefinite entitlement, when characterized objectively, takes indefinite descriptions in these argument places, e.g., “the first person across the line has a right to the trophy” or “I have a legal duty to anyone who crosses the border not to injure them.” I'll contrast this distinction with the related de re/de dicto distinction, and then explain why it is necessary to highlight the idea of an indefinite entitlement.

A. Appropriate Nonconformity

The word “inappropriately” does a lot of work in the above formulation. But this sort of qualification is necessary, for there are clearly justifiable forms of nonconformity with a right. Consider the following example, presented by Joel Feinberg:

It is plausible that your behavior infringes the rights of the owner of the cabin, in the sense that it does not conform with what the rights require. One might think that the owner's property rights do not extend to circumstances like these—that they grant control over the property to an extent circumscribed by clauses such as “except when someone's life is at stake.” But Feinberg resists this thought by insisting that you would later have a duty to compensate the owner for the loss of furniture. Nonetheless he thinks that this is a case in which you have justifiably infringed the owner's rights. This makes sense if we distinguish violating a right from merely infringing it.Footnote ²³ We can then say that violating a right—and, correlatively, wronging the right-holder—involves inappropriately (perhaps: impermissibly) infringing the right.

In an as yet unpublished manuscript, Johann Frick argues that wronging involves infringing a right in a way that causes a moral injury to the right-holder that impairs the agent's relationship with the right-holder.Footnote ²⁴ The argument rests on the following case:

Frick judges that in this case you act permissibly by pushing the man, but that you nonetheless wrong him. This threatens correlativity if we maintain that violating a right requires permissibly infringing it, and that correlativity means that wronging requires violating a right. But all I have committed to in defending correlativity is that it establishes a necessary connection between rights and wrongings such that wronging requires inappropriate infringement in some sense. And in Threshold Trolley the fact that you have wronged the large man entails that you have inappropriately infringed a correlative right of his.

What is potentially more troubling for Bentham is Frick's thought that there are two different “faces of morality”: an agent-focused question whether the agent acted permissibly, and a patient-focused question whether the agent injured the patient in a relationship-impairing manner.Footnote ²⁵ This would be troubling insofar as rights and wrongings appear on different sides of this divide, since that would threaten to sever Bentham's necessary connection between rights and wronging. But in fact Frick seems to allow both rights and wrongings to appear on the same side of the divide, since he defines wronging in the following way:

This definition suggests that the patient-focused language of wronging is to be understood in terms of whether the patient had a right and exercised it—a suggestion that contrasts with the view that rights present guideposts for agent deliberation. Perhaps there is a way to define wronging so as to maintain our judgments about cases like Threshold Trolley while also attributing our concepts of wronging and right to different faces of morality. But this is not Frick's argument, and I don't see how to provide it.

B. Cornell Cases

A prominent critic of Bentham is Nico Cornell, who in a provocative article sets out three kinds of cases that seem to involve someone being wronged without their rights being violated. The argument is based not just on counterexamples, but on an affinity with pluralism in moral explanation:

Cornell uses each of the examples to press a proliferation complaint, i.e., each case involves a wronging, but if we posit a right to explain the wrong we end up multiplying rights unnaturally.Footnote ²⁸

Consider a first example, Overheard Lie:

Cornell's judgment about this case is that you have been wronged by your coworker, but that he did not infringe any right of yours. The latter claim is supported by a judgment that it would not just be overly demanding to insist that you have a right, but that doing so would result in an unnatural proliferation of rights. That is because anyone who overheard your coworker would be in the same position, and would have a similar right. To emphasize how unnatural this might sound, consider what the right would have to be. It is true that you have not been lied to, so it is not a right against being lied to that has been infringed. You are wronged, instead, because you have suffered harm by relying on defective information disseminated as part of lie. And that does seem wrong, since one who lies should take into account the further effects that disseminating the information contained in the lie might produce. But the correlative right would appear to consist in a right against harm produced by relying on disinformation disseminated as part of a lie. Now it seems demanding to insist that everyone has such a right. After all, anyone could enter the workplace at the time your coworker tells the lie, and so overhear and rely on the disinformation it contains. So it seems that everyone must have a right against the coworker that he not spread disinformation, if you do. But, Cornell suggests, it is unnatural to think that everyone has a right against your coworker.

I find that my own judgment about the case wavers between two thoughts. On the one hand, it does seem that there is a case to be made for your coworker doing something wrong, even though I'm not sure that the lie itself need be wrong. But your coworker may very well have a duty to ensure that she not spread disinformation to someone who might reasonably rely on it, and who she foresees would be in a position to do so. So it seems to me no accident that Cornell sets the case up as he does, as involving two coworkers between whom there may well be a relationship of relative trust such that one reasonably relies on the words of another. On the other hand, it does seem implausible to think that everyone—even everyone who might enter the workplace—has a right against your coworker that they tell the truth. So if there is no plausible way to limit the scope of the coworker's duty, then we will have to admit that correlativity involves an unnatural enlargement of the number of rights that lie against the coworker.

Cornell's proliferation complaint is grounded in two more basic concerns, and these help to explain why we should avoid positing a right on the part of everyone who might overhear the coworker.Footnote ³⁰ There is first of all an allocative concern, i.e., that a proliferation of rights undermines their allocative function. The concern assumes that a function of rights is to entitle the right-holder to normative control over some sphere of activity of the correlative duty-bearer, e.g., if A has a right that B speak the truth, then A has normative control over what B says in the sense that it is up to A whether B does wrong by speaking the truth. But if there are multiple right-holders that have a right with the same direct object with respect to a duty-bearer, then this undermines their ability to control the duty-bearer's action. The second basic concern is a guidance concern, i.e., that a proliferation of rights undermines their action-guiding function. The concern assumes that a function of rights is to provide the correlative duty-bearer with guidance in moral deliberation, e.g., if A has a right that B speak the truth then the right guides B in deliberating about whether to speak the truth or not.

Supposing a right on the part of everyone who might overhear the coworker in Overheard Lie runs into the allocative concern because it seems to allow that very many people have control over what the coworker says. Perhaps it's not troubling insofar as we think most people would want the coworker to speak the truth. But this needn't be the case. Perhaps other coworkers within earshot would be happy to waive their rights because they enjoy the prank. Or perhaps the coworker's prank involves saying something false that would strike certain hearers as obviously implausible, but not others. The likelihood of disagreement as we increase the number of right-holders undermines the allocative function of each right-holder's right.

This last point raises the worry about the guidance function of rights. Who should the coworker have in mind when deciding whether to play the prank on the customer or not? The problem here is not strictly one that arises from the possibility of disagreement among the many people who might come to overhear him. For even if there is no disagreement, the fact that there are very many right-holders makes it very difficult for the coworker to determine that there is in fact no disagreement, and therefore to decide what weight to put on possible disagreement in his deliberation. In short, the problem is that the wide availability of a right does not so much guide the duty-bearer's deliberation as confuse it.

In case Overheard Lie does not seem a particularly compelling example of this dynamic, consider two cases that Cornell uses to emphasize the role of the allocative and guidance concerns.

First, Lottery:

Second, Fireworks:

In Lottery, it appears that you have been wronged by X, but that you have no right against X that he not take the money. Cornell works through several theories as to why you might have a right against X. Property cannot be the basis of the right since ownership of the prize has not vested in you at the point at which X absconds with it. Reliance cannot be the basis of the right since you can hardly rely on winning the money at the point of X's theft. A special relationship with Donor cannot be the basis of the right, since whatever special relationship you have with Donor is shared by all the other lottery entrants who end up losing.

Perhaps the most compelling theory on offer is that you are the intended beneficiary of the prize money. In a typical third party beneficiary case, such as where X promises Y to give something to Z, Z is intended as the beneficiary of the promise and may even be thought to have a right against X (though perhaps one that is different in kind and weight to Y's right). If there is such a right, it is because X knows very well that Z is to benefit and can be guided in deliberation by this consideration. For example, if X promises to give food for Z's benefit, it should be food that Z can actually eat rather than something that irritates Z's known allergies. But Cornell finds the idea that you are intended as the beneficiary in Lottery in this way implausible. As he puts it, Donor “did not really intend you as his beneficiary as much as he intended whoever it was whose name got pulled out of the lottery.”Footnote ³³

We can again frame the problem in terms of the allocative and guidance concerns, regardless of the proposed basis of the right. Suppose that at the moment of the theft, a persuasive Representative appears on the scene. Representative is powerless to stop X or alert anyone else, but can reason with X. In particular, Representative has the power to persuasively represent the interests of each of the lottery entrants to X, and can on that basis persuade X to part with the lottery money because of how it will eliminate the pressing needs of the winner and help them to accomplish great things. But Representative can only represent one lottery entrant at a time. Of course Representative could work her way through the entire list, but in each case X might want to know why he should care about that entrant, given that she is very likely not the winner. The intuitive problem here is that there is no one entrant yet who has been allocated control over X's activity, and so there is no one entrant's situation that should guide X's deliberation about what to do. The natural response is that X's deliberation should be guided by the interests of all the entrants, but now we are faced with the problem of multiplicity. There seem to be too many people who have an equal interest in the money, and the balance is not yet tipped in favor of any one of them on the basis that the lottery has been decided in their favor.

The guidance concern plays an especially important role in Fireworks. Here Cornell's judgment is that you wrong either child if you do not save them, and that each child has an equal right to be saved. But in addition—and this is where wrongings and rights come apart—the boy's parents would be wronged if you don't save him. But it cannot be the case, according to Cornell, that the parents have a right that the boy be saved, for this would tip the balance in favor of saving the boy, and so undermine the intuitive judgment that each child has an equal right to be saved. This can be put in terms of the allocative constraint in the sense that the parents’ rights and the girl's right come into conflict when determining what you should do. But it is even more vividly a concern about guidance, for if we posit rights on the part of the parents then they seem to play a role in your deliberation that distorts how you should really think about the case, that is, as one in which you are torn between the equal importance of saving the girl and the boy.

While I have been trying to lend credence to Cornell's judgments as much as possible, I have already been suggesting a way of framing the relevant rights that avoids the proliferation complaint. In Overheard Lie it is not so much that everyone, or everyone who comes into earshot, has a right that the coworker not spread harmful disinformation, but rather that whoever might reasonably foreseeably rely on the disinformation has such a right. That should lessen our worry that there are guidance and allocative concerns since there are not a large number of people who have the right, but simply one indefinite person. Deliberation may of course struggle in light of the fact that this person and their interests are unspecified, but this is a reason for thinking in a generic way about the concerns of someone who might overhear, rather than being put to the task of resolving conflicts between the interests of all those who might overhear. And in Lottery it is not the case that you, or any one of the other lottery entrants, has a right to the prize money at the point at which X absconds with it, but rather that whoever ends up winning has that right. That means that the right can play an allocative role, in the sense that the money is allocated to whoever fits the indefinite description whoever ends up winning.

Can we say something similar about Fireworks? Here we do not face a problem defining who the primary right-holders are. Both the boy and the girl have a right that you not have set the fireworks without warning, and each has a right (possibly derived from the first) that you do what you can to rescue them. The question is whether there are further rights held by the boy's parents that serve to tip the balance in favor of saving him. Cornell's thought is that there could not be, since intuitively you should treat them as equals when deciding whom to save. But equally intuitively, you wrong the boy's parents if he dies because of the accident. We make the case easy for ourselves if we stipulate that you do not know which of the children has parents. In that case, for each child you wrong whoever loves the child (supposing that the loving relationship is the ground of the wrong—if it is something else, we can reformulate the previous wh-clause in the relevant way). This indefinite right-holder stands in the same place in your deliberation, regardless of who fills it. It just so turns out that, if you kill the girl, nobody satisfies the wh-clause.

How should we think about the harder case—the case in which you know that the boy has loving parents and the girl has none? Here the epistemic formulation of the previous paragraph should be understood as an attempt to articulate the considerations of generality that really bear on contractualist deliberation. In thinking about whether to undertake some risky course of action, what weighs on deliberation, apart from the complaints of those who are directly hurt by the action, are the complaints of whoever stands in the appropriate loving relation to those who are directly hurt. These complaints are held by indefinite people in the case of both boy and girl, so that their positions in deliberation are equal—even though ex post it turns out that there are two people who satisfy the indefinite description in the boy's case, and none in the girl's.

The question is why we should think deliberation should take this general form in a case like this in which you know exactly who stands to be harmed. The main reason, I think, is egalitarian: the generality of this way of thinking is demanded of moral deliberation precisely because it is a way of maintaining the equal standing of the potential victims of your action. For certainly it could have been the case that the girl had more loving parents than the boy does, and it seems unfair to treat the girl worse just because of an accident of her history. Doesn't this mean that we don't really think that the parents are right-holders, since we are not giving them a determinative place in deliberation? No—it is a familiar feature of rights that they sometimes stand in relations of lexical priority, and it should be unsurprising that the parents’ rights are of a lower priority than those of the children. What should be more surprising is Cornell's assumption that the parents’ rights should play a tiebreaking role, for this threatens to collapse our thinking about rights into a kind of aggregative thinking that is incompatible with the priority of rights.

What we are doing here should be familiar from the defense of Hohfeld: using the strategy of abstraction to show that correlativity can be preserved once we calibrate rights and wrongings by formulating them with the requisite degree of generality. Doing so requires in the case of Cornell's cases requires an extra move. While rights and directed duties are correlative normative instances at a particular moment in time, rights and wrongings are often formulated at different moments: a right can be ascribed in advance of it being violated. A wronging is typically ascribed only after violation of a correlative right. Cornell's cases involve ascription of wrongings in definite terms, which seems to require a correlative definite right, running into the problems he mentions. But we could just as well formulate them in indefinite terms. In Overheard Lie, you are the one who is wronged, but really it is in virtue of fitting the indefinite description whoever was reasonably foreseeably wronged by the lie. And so on for the other cases.

What I am proposing is that each of Cornell's examples can be squared with Bentham once we acknowledge that they involve indefinite entitlements. That is, each case involves a right on the part of an unspecified person. Given our practical bias against thinking in terms of unspecified persons, it is fairly easy to go along with Cornell's thought that there either is nobody who could be the right-bearer, in a case like Lottery, or that there would have to be a large number of people, indeed too many people, who would have the right in cases like Overheard Lie and Fireworks. But that thought can and should be resisted. It can be resisted because in each case the unspecified right is available to explain the grounds of the wrong, and it does not result in an unnatural proliferation of rights. And Cornell's judgments should be resisted not just because they disturb the pleasing conceptual elegance of Bentham, but because they leave us without a way to explain why there is not just a wrong but a directed wronging in each of the cases.

My judgments about each of Cornell's three cases are shared by Jay Wallace in his discussion of Cornell's view.Footnote ³⁴ As he points out, in each case “the details that would make it plausible that a third party is wronged through the agent's action would also provide a basis for an antecedent claim or right that is assignable to that individual.”Footnote ³⁵ Yet Wallace also allows that there may be cases in which Cornell is correct that there are wrongings without correlative rights—though he does not specify which cases these are. Perhaps this is because Wallace's aim is to defend the idea that all interpersonal morality can be understood within the relational framework, so that it is more important to him that we not deny that some duties are directed than that we save correlativity (since, as we saw in the previous section, Hohfeldian correlativity may come at the cost of denying that duties of friendship are correlative).

In conceding the possibility of some Cornell cases, Wallace suggests a way to understand them within a relational framework, though he explicitly withholds judgment about whether it is correct. Cornell himself thinks that wronging requires that someone's right has been violated, though not a right of the one who has been wronged. Wallace understands this structure in terms of the idea of a secondary claim: someone who is wronged by a right violation has a secondary claim not to be harmed by the rights violation of another. This, he further suggests, can be understood in contractualist terms. One who stands to be harmed by an action that is morally wrong has personal reasons to complain about the action. This complaint, if it can be the basis for reasonably rejecting the action, is a secondary claim in virtue of which the complainant would be wronged by the action. But the agent of the action cannot counterpose their own reasons for acting against such a complaint, since the action is already (as stipulated) morally wrong. So the secondary claim stands and is the basis of a wronging.Footnote ³⁶

Why should we not think that the secondary claim is itself a right? Wallace points to two ways in which secondary claims differ from rights: (i) they do not serve to indicate to whom the wrongdoer is ex ante accountable, since they are not ex ante identifiable; (ii) they do not serve to indicate what the claim-holder is in a position to ex ante demand from the wrongdoer, since it is unclear whether a particular claim-holder would in fact be harmed by the agent's wrong. To illustrate these points, consider again Overheard Lie. It is, presumably, unclear in advance that you will need to cross the South Bridge, so that at the time that your coworker tells the lie, it is unclear first of all that he is accountable to you in terms of a secondary claim. And it is unclear that you could demand that your coworker not tell the lie, since we cannot know in advance that you will be harmed by overhearing it.

Notice that if these points about Overheard Lie are correct, they tell against not only your having a right, but also against your having a secondary claim against your coworker's lying in public. For the secondary claim is grounded in your complaint against the harm to you that is caused by the overheard lie, but we cannot know in advance of the lie that you will in fact be harmed by it. I've put this in epistemic terms—whether we know you will be harmed—though I doubt this is the correct way to understand the contractualist formula. But the epistemic formulation is the most convenient way to put what is really a consideration about generality. When thinking about whether your coworker may lie or not in public, we should consider the generic reasons that ground complaints against the lie. This means considering not whether you have personal reasons to complain about being harmed, but whether someone in the relevant position, i.e., someone who is reasonably foreseeably harmed by the lie, has reason to complain. The indefinite person whoever overhears the lie and is reasonably foreseeably harmed by it does have such a reason. But now it seems that their complaint is no longer secondary in the way that Wallace has in mind. In particular, it does not differ from a right in failing to identify to whom the wrongdoer is accountable ex ante, and what the claim-holder may ex ante demand. The wrongdoer, your colleague, is accountable to whoever overhears the lie and is reasonably foreseeably harmed by it, and that person may demand that he not tell the lie. So we should simply say that this indefinite person is the holder of an indefinite right. This line of thinking seems to me replicable in all cases with the structure of Cornell's cases, making Wallace's concession too quick.