1. Introduction
There is a longstanding debate about the extent to which social justice permits individuals to benefit from their natural talents. On the one hand, it may seem that individuals have a right to the fruits of their labor. On the other hand, superior abilities are already rewarded with personal enjoyment and public admiration of their exercise “without adding to these a superior share of the world’s goods” (Mill Reference Mill and Robson1969, 254). In the early twentieth century, British socialists recognized that it might be necessary to offer incentives for those with scarce talents to put them to good use, but they tended to view this “rent of ability” as an unfortunate compromise (Jackson Reference Jackson2007, 72–79). Liberal egalitarians have not been so clear about the moral status of incentives. If inequalities benefit the worst off, Rawls’s difference principle seems to count them as just. Yet critics argued that individuals who accepted the rationale for this principle would not expect extraordinary wages for ordinary work simply because they were blessed with superior productive ability (Grey Reference Grey1973; Narveson Reference Narveson1976; Cohen Reference Cohen1992). According to G. A. Cohen, people’s economic choices ought to be guided by an egalitarian ethos (Reference Cohen2008, 2, 17, 123). Even if we all shared the same egalitarian conception of justice, however, we might be uncertain about the extent of compliance with duties that are not legally enforceable. Moreover, real societies are home to a diversity of conceptions of justice, many of them not egalitarian. Are egalitarian norms still binding, or might lack of assurance about the compliance of others justify what would otherwise be objectionably self-interested choices?
The idea that some duties of justice are subject to a compliance condition figures prominently in Andrew Williams’s publicity defense of the basic structure objection to the egalitarian critique of productive incentives (Reference Williams1998, Reference Williams2008). Following Rawls, Williams argued that principles of justice apply differently to public rules than to conduct directly. In order for rules to count as public, it must be clear what they require of us and whether others are complying. The rules of Cohen’s egalitarian ethos allegedly can’t count as public because verification of compliance would require too much information about individual circumstance.
Critics have tended to interpret the publicity defense in an uncharitably strict manner, assuming, for example, that in order to vindicate the basic structure objection we must claim that all duties of justice are conditional on confidence that everyone is complying. We need only claim that some duties are conditional on a sufficient level of compliance, however, to cast doubt on Cohen’s claim that inequality-generating incentives are unjust. The main problem with the publicity argument is not that it has obviously unacceptable implications, but that its rationale is underdeveloped. Verifiability of whether others are complying is necessary in order to have positive assurance that they are, but what justifies making compliance of others a condition for duties to be binding in the first place? The fact that general compliance with duties of justice would be a good thing doesn’t show that these duties are conditional on general compliance. If people are motivated by reciprocity, conceptions of justice that are too unilateral may be unstable. The stability rationale simply shows that true justice may be infeasible, however, not that duties of justice truly are conditional. What’s needed is an ethical rather than a merely psychological interpretation of reciprocity. The purpose of this paper is to explain the structure of a reciprocity condition on some requirements of justice, and to make the justification for such a condition clearer than it has been in the literature.
The fact that some of our duties of justice are conditional on reciprocity is a consequence of the relational or “second-personal” nature of social justice. The reciprocity condition is fundamentally bilateral, but may take on a proportional form due to uncertainty or because conduct cannot be targeted to specific individuals. Not all duties are conditional on reciprocity, because some duties are owed to third parties, as well as to their primary targets. Our basic moral duties have this multilateral structure. Duties of egalitarian distributive justice are conditional, however. Critics of incentives may admit that some duties are conditional on reciprocity but insist that exploitation, like racism, is wrong regardless of whether others are engaging in it. Some forms of exploitation are unconditionally wrong, but not the specific form of exploitation at stake in Cohen’s critique of incentives.
2. The publicity defence of the basic structure objection to the egalitarian critique of incentives
The core idea of the difference principle is that if social inequalities benefit the worst off they can be justified to the worst off, rendering these inequalities compatible with mutual respect in a society of equals.Footnote 1 The principle was always framed in terms of the social positions created by basic institutions, however (Rawls Reference Rawls1957, 653–54). If against the backdrop of just institutions the choices you and I make lead you to have more than me, that inequality need not benefit me, as per the idea of “pure procedural justice” (Rawls Reference Rawls1999, 76–77). The theory’s institutional focus makes the difference principle friendlier to common intuitions about responsibility (Van Parijs Reference Van Parijs and Freeman2003, 216), but raises a question about the justice of incentives.
Because no one deserves their “greater natural capacity,” no one should get a greater share of the social product “merely because they are more gifted.” Those who have inherited scarce talents may only gain “to cover the costs of training and education and for using their endowments in ways that help the less fortunate as well” (Rawls Reference Rawls1999, 87). Where higher income offsets costs of training, there is no net inequality that needs to be justified. The ethically problematic inequalities are those that persist after subtracting whatever special burdens are attached to particular social roles, inequalities resulting from “incentives”Footnote 2 that elicit and direct productive effort.
Early on, critics questioned whether such incentives were truly justifiable to the worst off. Rawls seemed to be assuming that those with scarce talents were morally justified in withholding their efforts from the community unless rewarded with superior pay, which socialists would see as extortion (Grey Reference Grey1973, 322–23). Inequalities were only necessary to raise the worst off given that the talented were out for themselves (Narveson Reference Narveson1976, 12). Drawing these criticisms together, Cohen distinguished “strict” and “lax” versions of the difference principle. The strict version endorses only those inequalities that are truly necessary to raise the worst off, apart from human choice, whereas the lax version also endorses inequalities that are necessary because “talented producers operate as self-interested market maximizers” (Cohen Reference Cohen2008, 69). All things considered, we might be justified in accepting inequalities made necessary by selfishness, but in doing so we would be giving in to injustice (121–23). The argument for incentives from the original position is not one that the better off could offer to the worse off as a justification for these inequalities because it is the decisions of the better off that are responsible for making it the case that unequal rewards are necessary to elicit productive effort. The kidnapper can’t argue that the parents should pay, even though it’s true, because the fact to which the argument appeals is one the kidnapper causes to obtain. The incentive argument fails the “interpersonal test,” and thus undermines “justificatory community” (40–48).
In a fully just society, Cohen argued, people’s everyday economic decisions would reflect an “egalitarian ethos” (Reference Cohen2008, 22). People with greater productive capacity wouldn’t bargain so hard for higher pay, and would be willing to work reasonably hard even under high marginal taxation (17). They would also make some sacrifices in career choice, accepting more productive work they enjoy less, for the sake of benefitting the worse off (70, 182). Yet some incentives would still count as just, mainly because individuals have, to some reasonable degree, an “agent-centered prerogative” to pursue their own interests (61).
Some have questioned whether Cohen’s ethos should be referred to as “egalitarian” since it attributes no intrinsic value to economic equality (Quong Reference Quong2010, 327–28; Tomlin Reference Tomlin2010, 237–40). Cohen was explicit that those who accept the ethos are committed to maximizing the position of the worst off (subject to the limits of the personal prerogative), not to leveling down, making his ethos prioritarian rather than intrinsically egalitarian (Cohen Reference Cohen2008, 31). He argued that the ethos was egalitarian in its conclusions rather than its premises. Those who accept the ethos believe that justice permits little inequality because in a society of appropriately motivated individuals inequality is rarely necessary to benefit the worst off (34). However, the correct interpretation of the ethos depends on how one interprets the difference principle—in particular whether it permits or mandates inequalities that benefit or do not harm the worst off (29n6). On the “benefit” versions of the principle, morally arbitrary inequalities are intrinsically bad at least to this extent: that if there are no such benefits, we stick with equality.
One response to the egalitarian critique was to argue that the egalitarian ethos undermines individual liberty, but since the ethos is not legally enforceable, this objection may not be decisive (Williams Reference Williams1998, 228). An alternate response was to argue that justice doesn’t apply to personal choices, but only to basic social institutions. Rawls said that principles of justice apply not “only” to the basic structure of society, but “primarily” (Reference Rawls1999, 53, 6), and he discussed a number of noninstitutional duties for individuals, such as the duty not to be cruel, the duties of mutual respect and mutual aid, and the creative aspect of the duty of justice (98, 297). However, in Political Liberalism, he did say that the difference principle applied to policies and institutions rather than to individual decisions (Reference Rawls1996, 283), which was the basis for Cohen’s charge that Rawls permits “unrestrained market maximizing” (Cohen Reference Cohen2008, 10).
Cohen claimed that the basic structure objection faces a dilemma. If the basic structure only includes coercive rules, principles of justice will ignore social norms that are not enforced by the state but that have profound impacts on people’s lives, while if the basic structure includes norms not codified in law, principles of justice will apply to personal choices too, because that is where such norms reside (Reference Cohen2008, 124–29). One might deny the existence of this dilemma on the ground that the question of how limited or extensive coercive institutions should be must be answered by appeal to principles of justice (Neufeld and Van Schoelandt Reference Neufeld2014, 88), and that the extent of the influence of informal norms will depend on what rules are coercively enforced. However, any view committed to basic liberties must leave a substantial range of conduct to be regulated by ethics rather than law, raising the question of whether merely ethical norms can count as requirements of justice, and whether they are binding even in the absence of compliance on the part of others.
Andrew Williams sought to evade Cohen’s dilemma by arguing that the rules of the basic structure need not be coercive but must be public, and that the norms of Cohen’s egalitarian ethos could not achieve the necessary publicity. Williams pointed out that Rawls understood an “institution” to be a public system of rules in the sense that with due diligence citizens could attain common knowledge of what the rules require and “the extent to which individuals conform with those requirements” (Williams Reference Williams1998, 233; citing Rawls Reference Rawls1999, 55). Norms do not qualify as public if they are “so informationally demanding that individuals are incapable of mutually verifying the status of their conduct” (Williams Reference Williams1998, 234). To know whether someone with high income has lived up to the ethos, we would have to determine whether their additional income offsets special labor burdens or falls within their prerogative for the pursuit of personal interests. Yet we will, in general, have limited information about each other’s circumstances and productive potential, generating significant uncertainty about the level of compliance with egalitarian economic norms (241).
Williams argued that verifiability of compliance was essential to a valuable ideal of social unity that Rawls termed “well-ordered social cooperation.” In a well-ordered society, it is common knowledge that we all accept the same principles of justice, that our institutions satisfy these principles, and that we all generally comply with these rules. “[E]veryone willingly complies with the conception because, having witnessed others’ readiness to act justly, they have internalized its requirements” (Reference Williams1998, 244). Cohen summarizes Williams’s position by saying that individuals can justifiably refuse sacrifices putatively demanded by justice “if they lack assurance that others, too, are sacrificing” (Reference Williams2008, 22, 347). Thus, verifiability of whether others are complying is needed in order to have assurance that they are.
Critics have responded to the publicity defense of the basic structure objection in three main ways. First, they have sought to disable the publicity defense as a response to the critique of incentives by claiming that an egalitarian ethos can be suitably public. Second, they have rejected the publicity defense on grounds that it has implausible implications. Finally, they have rejected the publicity defense by claiming that the demand for verifiability lacks a compelling rationale.
2.a Disability
It would be unreasonable to demand certainty about the compliance status of each individual as a condition for duties of justice to be binding. Yet, if we only demand a reasonable degree of confidence about the overall level of compliance in the population, some egalitarian economic norms might pass the test (Cohen Reference Cohen2008, 347; Lippert-Rasmussen Reference Lippert-Rasmussen2008, 41–42). However, the fact that we can have a rough sense of what proportion of individuals comply with egalitarian norms doesn’t undermine the publicity defense of the basic structure objection if verifiability matters for the sake of assurance. It’s clear that few people today live up to the egalitarian ethos, even though it’s hard to say just what the proportion is.
Cohen claims that we can sometimes have confidence of general compliance, citing the case of Britain during WWII, when most people were evidently doing their bit (Reference Cohen2008, 353). War is a temporary, exceptional condition, however. Also, if some are being conscripted to die for their country, it is not surprising that others voluntarily pitch in. They are not doing so unilaterally, but as form of reciprocity for the sacrifice imposed on others. It is a leap to imagine that most people would forswear “market maximizing” on an ongoing basis. That’s not to say that merely ethical economic norms could never achieve high levels of compliance. Evident restraint in wage negotiation by an occupation with above-average income might prompt other occupations to follow suit (Casal Reference Casal2015, 824). We don’t need to argue that the difference principle never applies directly to personal choices, however, in order to argue that it generally won’t apply to wage negotiation and career choice and, thus, that incentive payments will typically count as just.
2.b Implausibility
Critics have pointed out that the demand for publicity undercuts other aspects of the theory. Because beliefs are not directly observable, it is hard to know whether others accept the same principles (Shiffrin Reference Shiffrin2010, 121n21). Its also hard to be sure that our institutions maximize the position of the worst off (Lippert-Rasmussen Reference Lippert-Rasmussen2008, 47). In order to raise doubts about the critique of incentives, however, we need not demand verifiability of common acceptance of principles, nor verifiability of institutional realization of principles. We need only demand assurance of compliance with the rules in place. Uncertainty about the justice of our current tax policies is unfortunate, but so long as most people are paying their taxes I will pay mine (even if I think I could get away with cheating). Most people are not choosing their career so as to maximize the position of the worst off, however, so I won’t make this sacrifice either, not without some assurance that the people for whom I would be making that sacrifice would do the same for me were our positions reversed.
Critics have also pointed out that some genuine duties are vague, making it hard to verify compliance. The natural duty of justice requires that we comply with existing just institutions but also that we “further just arrangements not yet established, at least when this can be done without too much cost to ourselves” (Rawls Reference Rawls1999, 99; emphasis added). Vagueness about how much is too much means that the creative part of the duty of justice fails the test of publicity (Cohen Reference Cohen2008, 358). This aspect of the duty is not conditional, however, making verifiability irrelevant. It is because the creative part of the duty is unilateral that it is subject to a demandingness constraint. The duty to comply with approximately just institutions is conditional, which is why it is not subject to any similar cost limitation.
Williams had argued that norms about the sharing of household labor could pass the test of publicity because the requirements of domestic justice “can be stated quite clearly,” and “any serious failure to conform … is readily apparent to the victims of injustice” (Reference Williams1998, 242). Critics responded that this injustice may not be apparent to others. Although the wife may be confident her husband is shirking, their dinner guests may not be able tell whether the backache that serves as his excuse for not washing up is genuine (Cohen Reference Cohen2008, 359). However, it is irrelevant to my duty to share my family’s labor burden fairly whether other husbands are doing the same. In the case Cohen describes, both husband and wife are employed but social norms permit the husband to do less domestic labor, leading to an unequal total division of labor. In this situation, my primary duty is to my wife, not to society at large. The condition on my duty to share domestic burdens is not that most other husbands do so too, but that my wife would do so, were our positions reversed.Footnote 3
Finally, critics have pointed out that some duties of justice are binding whether or not others are complying, making assurance irrelevant. Racism and domestic abuse are wrong regardless of their prevalence (Cohen Reference Cohen2008, 348; Casal Reference Casal2015, 823). However, we need not claim that all moral duties are conditional on assurance of compliance in order to raise doubts about unilateral egalitarianism.
2.c Arbitrariness
Some critics have found the rationale for verifiability obscure (Casal Reference Casal2015, 823–24; Lippert-Rasmussen Reference Lippert-Rasmussen2008, 46). I think verifiability is needed for assurance, but what is the rationale for assurance? What makes it reasonable to insist on compliance on the part of others as a condition for me to be willing to comply with (some) duties of justice?
Williams’s appeal to well-orderedness pointed in two different directions: the intrinsic value of a certain kind of social unity and the instrumental value of stabilizing just cooperation. Neither provides a fully adequate rationale for making duties conditional on compliance on the part of others. We may grant that it would realize “civic friendship” (Rawls Reference Rawls1999, 5) for it to be common knowledge that everyone generally complies with the rules of institutions that realize principles of justice we all accept. The value of this kind of moral community is not a reason for thinking that our duties of justice are conditional on general compliance, however—not unless making them conditional is necessary in order to secure general compliance, but why should that be? Rawls’s argument about stability supplies an answer, but of the wrong kind.
For better and for worse, evolution has given us a noninstrumental tendency to respond in kind (Rawls Reference Rawls1999, 433). Psychologists and experimental economists call this “strong reciprocity” in order to distinguish it from strategic, instrumental forms of behavior (Ostrom Reference Ostrom2000; Fehr, Fischbacher, and Gacther Reference Fehr, Fischbacher and Gacther2002; Gintis et al. Reference Gintis, Bowles, Boyd and Fehr2005). I treat you well because you treated me well in the past, not because I hope to receive some benefit from you in the future. Or I treat you badly, even at some cost to my own interests, because you treated me badly.Footnote 4 Motivation by reciprocity means that we are to some extent conditional rather than unilateral cooperators, willing not to cheat even when we can get away with doing so, but only if we think others generally comply with the rules (Rawls Reference Rawls1999, 236–38, 296). This psychology is a stabilizing force in one respect. When it is evident that others comply with rules that I take to be approximately just, I will tend to interpret their compliance as a benefit I ought to reciprocate. When willingness to cooperate is conditional, however, difficulty verifying compliance poses a threat. Even if each of us prefers not to cheat when others are complying, I might be uncertain about whether others are willing to comply and, so, not comply for this reason. Even if I believe that others do prefer to comply, I might be uncertain about whether they know that I believe this, and so not comply for that reason, and so on. This is the assurance problem, as Rawls adopted it from Sen (Sen Reference Sen1967, 112–13; cited by Rawls Reference Rawls1999, 237).
The concern with stability provides a rationale for making duties conditional on compliance on the part of others. Given the facts of human psychology, a conception of justice that is too unilateral might be unstable. It would at best be futile and at worst counterproductive to try to implement a conception of justice that was overly unilateral; we might end up with something worse than we could have had if we had set our sights lower. However, the claim that compliance conditions are a stabilizing device would elicit the objection that we ought not define justice down, so to speak, but insist on a sharp distinction between what is just and what is feasible.
Although the tendency to meet kindness with kindness is attractive, the unwillingness to aid or share fairly without assurance of reciprocity may seem too cramped a view of our duties to others (White Reference White2003, 65).Footnote 5 Reciprocity can involve repayment of harms as well as benefits—tit-for-tat rather than gratitude. If one thinks that true justice has no retributive element, tailoring principles to fit with reciprocity may seem to involve diluting true justice for the sake of stable quasi-justice. Admittedly, it would not make sense to implement truer principles now if that meant getting less justice in the long term. Yet principles play an evaluative as well as a practical role, influencing the attitudes we have toward one another and, thus, our relationships, even when we can’t presently act on these assessments. If people are more selfish or vengeful than they should be, it might be important for us to recognize that these aspects of human psychology are regrettable—regrettable from the point of view of justice—even if not presently changeable (Tomlin Reference Tomlin2010, 243). What’s needed is an interpretation of reciprocity as an ethical value rather than merely a feasibility constraint resulting from facts of human psychology.
It might be objected that we need not appeal to assurance to resist the egalitarian critique of incentives since mutual accountability will do the trick. Following Stephen Darwall, Elizabeth Anderson argues that on a relational or “second-personal” conception of social justice, claims of justice are demands that we make on each other based on standards to which we hold each other accountable (Anderson Reference Anderson2010, 4; citing Darwall Reference Darwall2006). Mutual accountability provides a rationale for the publicity condition on duties of justice, a rationale that is distinct from that of assurance (Anderson Reference Anderson2010, 20n25). Reasonable persons would not accept standards of justice that “left them in the dark about others’ compliance” because such standards would make it impossible to hold people to account (20). However, while it may be true that we couldn’t have a duty to 
$$ \phi $$
if 
$$ \phi $$
-ing could never be verified by anyone, we can have duties that are difficult to verify. My duty not to litter does not evaporate when I am alone in the wilderness. Also, egalitarian economic norms are not so unverifiable as to make mutual criticism impossible. It will be difficult for me to know whether strangers are complying with an egalitarian ethos, but I may be confident about whether friends and colleagues are living up to its demands. If I suspect others are not, I may criticize types of behavior I believe to be common even if I can’t identify specific delinquents. To resist the critique of incentives, we need to appeal to mutual assurance and reciprocity rather than simply mutual accountability.
3. Reciprocity conditions and the second-personal point of view
3.a The idea of a reciprocity condition
Turning from psychology to ethics, reciprocity is often understood as involving a duty to return benefits (Gouldner Reference Gouldner1960), or the virtue of having a tendency to do so (Becker Reference Becker1986, 74). However, reciprocity can also be a limit or condition on other duties—general duties not grounded in past receipt of benefits.
Duties can be conditional on compliance on the part of others for a number of reasons. Noncompliance on the part of others can affect the costs of compliance for me and, so, bring the duty into conflict with a prior right, as in the case of Hobbes’s state of nature, wherein keeping an agreement first puts my life at risk.Footnote 6 Noncompliance can also influence the effects my performance has on others. There is no point in me trying to queue if no one else is, not unless my initiating the queue will trigger a cascade of further compliance (Cohen Reference Cohen2008, 350–51). In these cases, what matters is not reciprocity per se (between duty holder and targetFootnote 7), but the general level of compliance, in so far as it affects the consequences of performance, and makes my compliance too costly or futile. In contrast, reciprocity conditions arise from the fact that duties are owed to other people, and are constitutive of relationships. In some circumstances, it may be an appropriate expression of our equal status to insist the other party be willing to comply with a duty as a condition for my being willing to comply. Reciprocity conditions arise naturally within a relational conception of social justice, I believe, but since neither Anderson nor Darwall make this point, I will spend some time drawing out the connection.
Darwall defines the second-personal standpoint as the stance one takes when one makes claims on someone else’s conduct, either explicitly—as in the case of speech acts such as “demanding, reproaching, apologizing”—or implicitly—as in the case of reactive attitudes such as resentment (Darwall Reference Darwall2006, 3). Such claims give others “second-personal reasons,” which are reasons that are “grounded in (de jure) authority relations” (4). Resentment is not just a negative attitude towards a norm-violator, but a form of address (72), e.g., “How dare you do that to me!” Such claims presuppose the existence of a norm, but also that I have standing to insist that you abide by this norm (14). I may, in addition, have standing to waive your duty, or to waive compensation for breach of duty, but even where my right is inalienable, I will have standing to demand and accept or reject your apology if you violate my right. The counterpart to this standing must be a capacity on your part to recognize and be moved by claims of authority. It is one of the “normative felicity conditions” of second-personal claims that the addressee have this kind of moral competence. I can’t resent the branch that falls on my head, and my resentment at some wrong you do me will be undermined by realizing that you were acting under extreme duress (4, 68, 77–78).
Jiwei Ci points out that resentment presupposes reciprocity as well as norm-violation (Reference Ci2006, 16). I can’t resent you for cutting in front of me in the queue if remember I did the same thing to you yesterday. I can dislike it, but not feel resentful, not without some self-deception (“It was an emergency for me, but you’re just being selfish”). I can’t coherently say, “How dare you do to me exactly what I did to you!” My normative power to object to nonperformance is cancelled by lack of reciprocity. It seems, then, that reciprocity is also a felicity condition of second-personal claims, at least in some cases. It’s true that hypocrisy can arise with respect to all moral norms, not simply justice (May Reference May2008, 449). Hypocrisy has a special significance, however, for relational or second-personal conceptions of justice. What is distinctive about such conceptions is their recognition that morality gives rise to relations of differentiated or particularized interpersonal authority. Rather than everyone having the same standing everyone else has to object to any infraction, individuals have specific degrees of control over particular aspects of the conduct of specific others. Lack of reciprocity can undermine such claims of authority when they are understood as claims made between persons.
Darwall argues that recognizing the second-personal nature of moral obligations is essential to vindicating morality’s “categorical” character (Reference Darwall2006, 27). That may be so, but the reactive attitudes that serve as a model of morality’s second personality are fundamentally conditional; whether I feel resentful or grateful toward you depends on your conduct towards me. Darwall reviews the evidence from psychology suggesting that we have the capacity to be guided by norms (171–78). This evidence also shows motivation by reciprocity, however, including the conditionality of norm compliance (Ostrom Reference Ostrom2000). Darwall responds to the objection that reactive attitudes are “reactionary” by arguing that resentment need not involve retaliation or revenge (Reference Darwall2006, 82–84). However, between unilateral compliance and taking pleasure in retribution lies the possibility of conditional compliance.
Ci argues that justice is in general a conditional virtue, in the sense that “universality of performance” is a condition for one’s own duties of justice to be binding (Reference Ci2006, 24). Ci justifies this conditionality on the ground that it protects one’s ordinary, nonrelational interests. His view is therefore vulnerable to the objection that it waters down justice for the sake of feasibility. Ci underemphasizes what to my mind is the most important theme from Strawson, which is that reactive attitudes have a relational character. If social justice is fundamentally a kind of relationship, then it might involve reciprocity conditions apart from the need to limit justice so as to make it compatible with self-interest.
Ci’s account of justice’s conditionality is also too broad, in two ways. First, not all duties of justice can be conditional on reciprocity. I have a duty not to kill you, except in self-defense, even if you would kill me if given the chance; I am not permitted to kill villains for sport or profit. Ci might respond by distinguishing duties of justice from duties of morality. Yet some duties of justice must be unconditional too. I have a duty to vote for just tax law even if others are voting for whatever laws most benefit themselves. The duty to help bring about justice would be self-defeating if it were conditional on everyone else making similar efforts. It must be unilateral, though therefore also subject to a demandingness constraint.
Second, the conditionality of duties of justice on reciprocity must itself be conditional on a sufficient level of background justice. The model I have in mind is Stuart White’s defense of a duty of social contribution, which he understands as a requirement of reciprocity. Ideally, income support ought to be conditional on willingness (not ability) to contribute to society. However, this contributive duty only applies if other social institutions have achieved a minimum level of justice (White Reference White2003, 86–91; White Reference White2017, 184). Reciprocity conditions are themselves conditional. We can explain this conditionality by the fact that we have multiple duties to each other, and that noncompliance on your part with duty 1 may be motivated by having suffered the effects of previous noncompliance on my part with duties 2, 3, and 4. It may be wrong of me to insist on you being willing to fulfill duty 1 with respect to me as condition on my fulfilling duty 1 with respect to you, if in the past I have neglected duties 2, 3, and 4 to you. We can’t insist on reciprocity piecemeal considering only one duty in abstraction from all other duties, but must take into account the broader context of our social relationships.
Finally, there is an important question about the form of justice’s (partial) conditionality. Ci assumes that the reciprocity condition is proportional; a given duty is binding only if enough of my fellows are complying. In contrast, I have assumed that reciprocity is fundamentally bilateral. My duty to share total family labor fairly doesn’t depend on whether other husbands do the same (while as discussed earlier the couple’s duty to resist sexist gender norms is unilateral but cost limited). Other cases are undoubtedly proportional, however. I’m not going to obey laws I disagree with if most others won’t obey laws they disagree with. So we need to understand the relation between the different forms of conditionality.
3.b Rationale for reciprocity conditions
Recognizing the second-personal nature of moral obligation makes it intelligible that some duties should be conditional on reciprocity in a way that would be incomprehensible from a purely third-person perspective. According to utilitarianism, for example, my basic moral duty is simply to bring about more well-being. How I go about doing that will vary depending on how others behave, but the duty itself does not because it is not directed to particular persons, and so can’t be conditional on their attitudes or conduct. In contrast, if we think of duties interpersonally, then it is at least possible that some duties are conditional on reciprocity.
The core idea is that insisting on reciprocity may be an appropriate expression of our equal status. How could I owe it to you to do something for you, if you would never do the same for me were our circumstances reversed? I could if I saw myself as your subordinate. If I see myself as your equal, however, there will have to be further considerations that justify unilateral compliance (such as the fact that I owe it to others to do something for you—a possibility to which I will return below). The point of distributive justice, at least in its egalitarian elements, is to make manifest our recognition of each other as equals (Anderson Reference Anderson1999; Scheffler Reference Scheffler2003). On such a view, it is natural to think that the demands of justice are attenuated where assurance of compliance on the part of others is lacking. Merely unilateral compliance with duties of distributive justice will not be successful, and could even be counterproductive. To continue to share fairly with you when you refuse to do likewise with me could be to treat myself simply as means for satisfying your needs. As White argues, “scaling down one’s efforts on the non-reciprocator’s behalf can then be understood as an assertion of one’s own dignity (‘I am not a doormat’),” and indirectly the dignity of others, not simply a “brute fact” of psychology (White Reference White2003, 65–66).
Reciprocity in this sense is not simply a proxy for demandingness. Whether or not lack of compliance on the part of other duty holders makes a duty more demanding, lack of compliance on the part of the duty’s target(s) attenuates the burdens it can impose. On this view, the essential difference between voluntary and mandatory redistributive schemes is not that voluntary schemes are psychologically more burdensome (Nagel Reference Nagel1975, 145-46; Shiffrin Reference Shiffrin2010, 120), but that voluntary schemes can’t be as burdensome where lack of legal enforcement undermines assurance of compliance.
Thus, there is a defeasible egalitarian rationale for a reciprocity condition. This rationale is not contingent on the effects my compliance is expected to have on myself or others. Instead, the condition follows from the directedness of our duties and the role they play in constituting relationships between persons. Relational egalitarianism cannot be unilateral egalitarianism, though it needs a unilateral component to put us on the path to justice. The reciprocity condition emerges from justice itself rather than from facts of human psychology that may limit its realization.
3.c Form of reciprocity conditions
In the simplest case, A’s duty to B to 
$$ \phi $$
is binding unless B—not C, D, E, etc.—fails a reciprocity condition. The primary form of the reciprocity condition is thus bilateral rather than proportional, meaning that whether the duty is binding depends only on the conduct of the person to whom the duty is owed as opposed to being binding if some threshold proportion of the relevant population is complying. The reciprocity condition is about what B would do if B were in the conditions in which the relevant duty applies independent of the probability that B will, in fact, be in these conditions. The condition is about willingness to comply with the duty, not the ability to provide or withhold benefits. Because of sickness or disability, B may never be in a position to owe assistance to anyone, so A may not stand to lose anything by not providing assistance to B, but A’s duty to B would still be binding as long as A didn’t have sufficient reason to doubt that B would fulfill the same duty if it were to apply.
There may be considerable uncertainty about how B would behave in the circumstances in which the duty applied, if B has never previously been in these circumstances. The stingiest (Hobbesian) view would be that any plausible doubt about B’s willingness to comply voids A’s duty to B, to 
$$ \phi $$
. More generously, the reciprocity condition might be that only strong evidence of B’s unwillingness to comply with the duty to 
$$ \phi $$
is sufficient to void A’s duty (perhaps including evidence of past noncompliance). The rule should presumably be to presume good faith until demonstrated otherwise. It might be that B wouldn’t comply first, but would comply second. In this case, A shouldn’t use the fact that B wouldn’t comply first as an excuse not to comply because if A does comply first, then B will reciprocate (should the relevant circumstances arise).Footnote 8
Some difficult issues arise in relation to disagreement. Suppose A believes in the duty to 
$$ \phi $$
, but B does not; is A’s duty to 
$$ \phi $$
with respect to B still binding? If A disagrees with the duty to 
$$ \psi $$
, a duty B takes to be genuine, A and B may recognize each other’s willingness to comply with different duties, and so consider the reciprocity condition fulfilled. Complications arise when one view recognizes more duties than another (as in the case of libertarians versus social democrats, perhaps), and when there are more than just two views, making it hard to discern the relevant symmetry. Such problems are normally solved by recognizing a particular set of rights and duties in law.
I have said that the primary form of the reciprocity condition is bilateral rather than proportional, but this distinction can be blurred by uncertainty and the difficulty of targeting conduct to particular persons.
Uncertainty: Suppose a particular duty is conditional on the other party being willing to comply. If you and I are well acquainted, I may have a lot of information about how you have behaved in the past, and how you are likely to behave under various possible future circumstances. If we are strangers, however, I may lack information about your past conduct. In this case, I may take the general level of compliance in the population as a predictor of your conduct in the counterfactual scenario in which our positions are reversed. A proportional condition can thus serve as a proxy for an underlying bilateral condition when the relationship is not close.
Multiple beneficiaries: In other cases we owe duties to many people, but can’t target our conduct towards reciprocators, as in the case of contribution to public goods. If rates of contribution are high, then the importance of my duty to the large majority of compliers may outweigh the moral cost of a few people benefitting from my compliance without themselves being likewise willing to comply. If rates of compliance are low, however, it may be reasonable for me to refuse to be taken advantage of by many for the sake of fulfilling my obligations to a few.
Large-scale cooperation is unstable because uncertainty about compliance on the part of others can motivate noncompliance, which then undermines compliance on the part of others. How generous or stingy our decision-rule for uncertainty is will determine how quickly assurance breaks down. We shouldn’t underestimate the power of social norms to constrain behavior. Elinor Ostrom and others have found solid evidence of communities’ ability to regulate use of common pool resources (goods that are rivalrous but nonexcludable, such as fish or grazing land) without state enforcement. Such schemes depend on having relatively clear, monitorable rules, and are vulnerable to factors such as rapid migration, which can disrupt understanding of norms and trust of compliance on the part of others (Ostrom Reference Ostrom2000, 153). I’m not going to sanction a norm violator unless I’m confident that bystanders will agree with me that there is a norm, that it was violated, and that sanction is appropriate. The solution to the challenges of diversity for large-scale cooperation is to transform norms into laws, and to create centralized mechanisms for monitoring and enforcement.
With the transition from ethical to legal duties, the structure of the reciprocity condition changes. If I have a legal duty to 
$$ \phi $$
with respect to you under criminal law, I have that duty regardless of whether you would comply with the same duty with respect to me. For example, although I have a right to defend myself from attack, the fact that you attacked me in the past does not give me the right to assault you in the future once the initial threat is past (as one might think it would, in a state of nature). While I may take back what you stole from me, I can’t then go on to take the rest of your property. Legal duties concerning theft and assault have a multilateral character in that they are always owed to everyone, not just their primary targets. These duties are therefore not bilaterally conditional on reciprocity. At the same time, people won’t follow laws they disagree with unless most others are likewise willing to follow laws they disagree with. Thus the duty to obey the law is subject to a proportional compliance condition even where the underlying moral duties are bilaterally conditional on reciprocity.
3.d Limits of Reciprocity
The case of legal duties suggests that we can adopt a relational or second-personal conception of justice without concluding that all duties are conditional on reciprocity, because some duties may be owed to third parties as well as their primary target. We can grant, for example, that there is a fundamental attitude of respect that one must hold with respect to every moral agent regardless of their conduct, and which grounds certain basic rights that are not forfeitable by misconduct. I have a duty to everyone not to torture the torturer, even though the torturer could not himself object. Procedural rights have the same multilateral character; misconduct does not lead a person to forfeit their right to have their guilt or innocence determined by a fair procedure (Miller Reference Miller, Sakurai and Usami2014, 30). Everyone needs to be confident that their guilt or innocence will be determined by a reliable procedure, and general recognition of reliability is impossible if procedural rights can be forfeited. Free-speech rights have a similar structure. I can’t object to violations of my right to free speech if I refuse to respect the same right in others, but third parties may be able to object because they have an interest in hearing what I have to say. In these cases of multilateral directionality, lack of reciprocity with respect to the duty’s main target will not void the duty.
Consider also the duty to rescue those in imminent danger of dying. This duty applies even with respect to egoists who would never rescue others, one might argue, because third parties have an interest in having confidence that others would come to their rescue. If only those of good character had a claim to be rescued, this confidence would be undermined by lack of information about individual character. It’s less obvious that this reasoning extends to the case of the egoist who is in need but not in imminent danger, but perhaps there, too, we should presume good faith in face of uncertainty. Must we keep coming to this person’s aid if they never make any effort to help themself and, hence, are never able to share in the burden of helping those who can’t help themselves? Must we keep helping those who simply refuse to help others even after their character becomes clear? We may want to say that the right to assistance cannot be waived or forfeited, no more than can the right to be rescued, because third parties have an interest in avoiding participation in relations of domination. My failure to help myself or others would not void your duty to assist me, in this case, because even though I couldn’t object to not receiving assistance, others could object, on the grounds that it is difficult to avoid relating to me in a dominating manner if I fall below some threshold of access to resources. Thus, with respect to the duty of rescue, and perhaps even the duty of assistance, we can explain why we have a duty to benefit nonreciprocatorsFootnote 9 at significant cost to ourselves, a duty owed to third parties who are reciprocators. There is no such explanation, I think, for the duty to keep on benefiting the worse off beyond the point of deprivation and domination, nor the general duty to share burdens and benefits fairly. Or, at least, there is no such explanation that is consistent with adopting a relational account of social justice.
4. Racism and exploitation
Critics of productive incentives might concede that some duties of justice are conditional on reciprocity, but not the duty to avoid engaging in exploitation. Cohen insists that justice forbids racism and exploitation even when they are rife (Reference Cohen2008, 348), but says little in defense of this parallel between racism and exploitation. Racism is insulting and harmful, one might think, whereas at least some exploitation simply involves not benefitting others as much as one might.
We can grant that some kinds of exploitation are wrong regardless of any assurance of reciprocity. Consider the case of The Port of Caledonia and the Anna (Zwolinski Reference Zwolinski2007, 705). The tugboat’s offer of rescue to the stranded vessel at the price of 1,000 pounds was mutually beneficial, compared to no offer at all, but still wrong because the tug had a duty to rescue at no cost. The tug had no right to threaten not to rescue unless paid. This looks like a case in which the duty not to engage in exploitation is unconditional; the tugboat’s conduct can’t be justified by doubts about whether others would fulfill their duty of rescue at no cost. That is presumably because the underlying duty to rescue is not conditional. A has a duty to rescue B from imminent danger even if B would not do likewise. A therefore can’t charge for performance of this duty even if B would attempt to charge A were circumstances reversed.
Cases of sweatshop labor involving profitable multinationals in developing countries also provide examples of situations in which the duty not to exploit is unconditional. Consider the case reported by Arnold and Bowie of a pregnant female worker threatened with termination if she sought medical assistance (Arnold and Bowie Reference Arnold and Bowie2003, 230–31). Imagine that all workers were told up front that requests for medical assistance during work hours would be grounds for dismissal. If workers’ other employment options were bad enough, it might be in their best interest to accept this condition. It is wrong, however, to demand that employees waive their right to seek medical assistance as a condition of employment regardless of what employees would do were circumstances reversed. The duty to give access to emergency medical treatment is unconditional. If you fall to the ground with cardiac arrest, I have a duty to dial 911 at no charge whether or not you would do the same for me. It would be wrong of me to demand payment for performance of this basic duty of rescue regardless of what you would do in my place. And it is similarly wrong to demand that you waive your right of rescue as a condition of employment. The right of rescue is neither forfeitable by misconduct nor alienable by choice.
In order to undermine Cohen’s critique of productive incentives, however, we need not claim that all forms of exploitation are subject to a reciprocity condition. The central case at issue in Cohen’s discussion of incentives is that of people who have been lucky in the genetic and social lottery, and so have developed talents that happen to be scarce, which allows them to secure higher than average wages for averagely burdensome work. In accepting such wages and refusing to contribute the excess to the worst off, these people are refusing to share the benefits of cooperation fairly. What Cohen needs to argue is that this specific form of exploitation is unconditionally wrong. He needs to argue that I have a duty to share fairly with others even where I lack assurance that the beneficiaries of my fair sharing would do the same for me were our circumstances reversed. It is true that when A and B have a duty to 
$$ \phi $$
with respect to C but A fails to comply the resulting unfairness between A and B doesn’t void B’s duty to C (Cohen Reference Cohen2008, 349). However, the worry about reciprocity is not about A and B’s position in relation to each other but about C’s conduct with respect to A or B, were C in their shoes.
Seana Shiffrin has offered an argument that would explain why the relevant form of exploitation is, like racism, unconditionally wrong. Shiffrin does not suppose that individuals have a duty to do whatever will maximize the position of the worst off. Instead, she claims that in the course of pursuing their various legitimate personal goals, individuals should never treat as a reason for claiming advantages the fact that they possess relatively scarce innate talents (Shiffrin Reference Shiffrin2010, 126, 141). Shiffrin draws an analogy with race. I need not devote my life to combating racism. Nonetheless, if I suspect that my prospective employer is racist in my favor, I may not try to “leverage” this racism by negotiating for a higher salary (127). Taking advantage of the scarcity of my innate talents for personal gain is no different, Shiffrin claims. Shiffrin acknowledges that there is a difference between “actively leveraging” morally arbitrary factors and unknowingly accepting a benefit whose causal history is compromised by the influence of such factors (136). But where the facts are reasonably clear, it is wrong to take advantage of morally arbitrary factors both with respect to race and with respect to talent.
Race is not analogous to talent, however, both in terms of the sources of the morally arbitrary advantage and the effects of my making use of it. Advantage due to race is the result of social preferences that are inherently unjust. Advantage due to scarcity of innate talent also depends on the shape of social preferences, but these preferences are not in themselves objectionable. There is nothing inherently unjust in preferring more to less, or in preferring the same product at a lower price to a higher price, or in wanting to listen to great music or watch excellent athletes. It is not inherently unjust to want to watch Lionel Messi play soccer, rather than me. Furthermore, when I, a white person, take advantage of racism in the marketplace for my own gain, I participate in and help sustain a system of expectations and assumptions that relegates nonwhites to the status of second-class citizens. In contrast, when I let my employment decisions be influenced by market incentives, I don’t relegate those not favored by possession of scarce talents to second-class status.
Letting people respond to price signals helps allocate resources efficiently, and so potentially makes everyone (including the worst off) better off. If people don’t respond to financial incentives, then talents will be misallocated (e.g., a suboptimal ratio of gardeners to doctors) leading to less prosperity and, depending on the structure of surrounding institutions, less prosperity for the worst off. Shiffrin suggests that the problem can be mitigated if justice requires contribution, or if motivation by justice is supplemented by beneficence (Shiffrin Reference Shiffrin2010, 130). Duty and beneficence will only solve this problem, however, if people can figure out what personal choices will provide maximum social benefit. Given the vast number of possible allocations of resources across productive activities, the diversity of individual preferences, and the dispersion of this information in the population, there is no simple way to construct an index of social urgency (Moon Reference Moon2015, 46–47). Letting prices fluctuate with supply and demand solves this problem because each decision by an agent serves as a signal to others about what skills are in short supply relative to people’s wants (Hayek Reference Hayek1945; Heath Reference Heath2018). If people were to make personal choices based on their own assessments of social needs, their choices would no longer provide signals about the information to which they alone have access, which is information about their own preferences, capabilities, and resources. In the process of communicating this information, however, prices also incidentally reward those who happen to have been born with relatively scarce abilities. Market prices don’t distinguish those who were lucky to be born with a talent that is currently scarce from those who have worked hard to cultivate this ability. Prices reflect supply relative to demand, and are justified on the forward-looking basis that they provide incentives for adjustment in face of changing tastes and technology, not on the backward-looking basis that they reflect merit or desert (Hayek Reference Hayek1960, 85–102).
Joseph Carens proposed a way to use the market for the sake of prosperity without sacrificing fairness. In an ideally just society, moral incentives would be sufficient for individuals to maximize their pretax income, which the state would then tax and redistribute Carens (Reference Carens1981). However, a similar problem of assurance arises with respect to work effort and occupational decisions in this context (Williams Reference Williams1998, 239). How do I know that others are trying to boost their pretax income when they have no personal incentive to do so, but must be acting purely on the basis of social duty? Even in a society well-ordered by an egalitarian conception of justice this would be hard to verify, but any real society will be home to a diversity of conceptions of justice, many not egalitarian. That I happen to have scarce talents is not by itself a reason that justifies me getting a greater share of the social product. But I cannot be expected to make unilateral sacrifices (e.g., to donate to the worst off the amount by which my salary exceeds my reservation wage or, under Carens’s set up, to maximize my pretax income knowing others aren’t doing likewise).
The appeal to reciprocity helps explain why some forms of conduct that make inequality-generating incentives necessary pass Cohen’s test of interpersonal justification (Cohen Reference Cohen2008, 42–43). When “Worse Off” asks “Well Off” what makes inequalities necessary, Well Off might respond as follows: “I am willing to make the sacrifices you call for (working longer hours instead of going for a run or watching my kid’s soccer game; getting another degree instead of spending more time in my garden; taking a job that pays more but that I like less), but only as long as I have some assurance that you would do the same for me, were our positions reversed. I don’t have such assurance. In fact, given that many people are not egalitarians (even among the worst off), I know that many people would not reciprocate.” If we accept that duties of distributive justice are subject to a reciprocity condition and that the society in question is not so deeply unjust in other ways as to render this requirement void, this response licenses the conduct in question, thus passing Cohen’s interpersonal test.
It may be objected that the reciprocity condition gives too much leeway to selfish behavior. Suppose university professors got a big raise by threatening to go on strike last year; should pilots consider themselves entitled to do the same (Casal Reference Casal2016, 4)? The mechanics and baggage handlers and cleaning staff would do the same to us if they were in our shoes, the pilots may think. Thus it seems that the reciprocity defense licenses rent seeking on the part of the privileged.
One response would be to invoke the conditionality of reciprocity conditions, and to deny that the society in question meets the standard of minimum justice. This response would void all demands for reciprocity between better and worse off until sufficient background justice was attained, but would provide no reason for applying the reciprocity condition differently to different kinds of choices in the same society. I think that there is a difference between a privileged individual accepting a market wage and privileged collectivity bargaining for above-market wages. I’m not going to refuse part of my salary just because I’d be happy to continue working as a professor for less, because I have no assurance others would do the same for me. But I would vote against going on strike for higher wages even if other highly paid workers were using their clout to secure financial gains in this way. What justifies this asymmetry?
One factor is that the public interest rationale for individuals being able to respond to price signals won’t apply to collective action to achieve above-market rewards (particularly for occupations that have legal restrictions on entrance). A group’s going on strike also promises to have a greater effect on a society’s prospects for justice than does a set of individuals independently deciding to work less. A strike is a public event on the part of a collective agent. There may be a realistic prospect of establishing a social norm that the advantaged ought not strike for even higher pay. If so, not going on strike could be counted as part of the natural duty of justice, even if donating the portion of one’s salary that exceeds one’s reservation wage normally cannot.
5. Conclusion
I have tried to explain why the appeal to publicity in defense of the basic structure objection is not arbitrary. Publicity is needed for assurance, which is a matter of reciprocity, not just stability. Not all duties of justice are conditional on reciprocity, but some are—in particular duties of distributive justice, if we think of justice in relational terms. The analogy with racism does not hold.
In response to relational criticisms, advocates of luck egalitarianism have sometimes claimed to accept that relationships are fundamental. From this perspective, it’s not clear what the root disagreement is (Tan Reference Tan2008, 686; Kymlicka Reference Kymlicka2006, 27). This paper suggests an answer. My central idea has been that a relational approach can justify a reciprocity condition on some duties of distributive justice and, so, partially vindicate the basic structure objection. The fact that such an approach generates these conditions might be taken as a strike against it, however, and a reason for reaffirming a more unilateral view.
Acknowledgements.
I presented earlier versions of this paper to Queen’s Political Philosophy Reading Group, Montreal’s Groupe Interuniversitaire de Recherche en Philosophie Politique, and the Annual Meeting of the Canadian Philosophical Association. Thanks very much to the participants in these events for the valuable comments and criticism I received.
Andrew Lister is associate professor of political studies at Queen’s University (Kingston, Ontario). His work focuses on democracy, public reason and distributive justice, and has been published in journals such as Politics, Philosophy & Economics, the Critical Review of International Social and Political Philosophy, and the Journal of Political Philosophy.
