1. Introduction: From political to ethical cosmopolitanism
When scholars discuss Kant’s cosmopolitanism, they typically focus on his political cosmopolitanism, described in the Doctrine of Right and Perpetual Peace, and summarized as the “rational idea of a peaceful, even if not friendly, thoroughgoing community of all nations on the earth that come into relations affecting one another” (6:352). The ideal form for this community would require nations to “give up their savage (lawless) freedom, accommodating themselves to public coercive laws, and so form an (always growing) state of nations (civitas gentium) that would finally encompass all the nations of the earth” (8:357). For Kant, political cosmopolitanism is morally necessary. Kant insists that “states, considered in external relation to one another, are (like lawless savages) by nature in a nonrightful condition” and even if “no state is wronged by another in this condition (insofar as neither wants anything better), this condition is still wrong in the highest degree, and states . . . are under obligation to leave it” (6:344).
The moral necessity of this pacific league is clearest in Kant’s Doctrine of Right, where his account of the need to form such a cosmopolitan “league of nations” (6:344) follows from his more general account of the need to form political communities at all. The need to leave the state of nature and enter into a civil condition arises, for Kant, from the fact that a right relation among people depends upon limiting the freedom of each such that “the freedom of each can coexist with everyone’s freedom in accordance with a universal law” (6:230). This is not simply an abstract requirement. It requires instantiation in concrete rights, especially property rights, which receive authorization from “a will that is omnilateral . . . . [f]or only in accordance with this principle of the will is it possible for the free choice of each to accord with the freedom of all” (6:263). As long as issues of right are settled unilaterally (or even multilaterally), it is not truly “possible for there to be any right” (6:263). Thus before human beings leave a state of nature and submit to laws within a state, all rights are merely “provisional,” and the same applies to a state of nature among nations. As Kant explains,
Since a state of nature among nations, like a state of nature among individual human beings, is a condition that one ought to leave in order to enter a lawful condition, before this happens any rights of nations, and anything external that is mine or yours . . ., are merely provisional. Only in a universal association of states . . . can rights come to hold conclusively. (6:350).
Kant’s political cosmopolitanism, described in the Doctrine of Right and Perpetual Peace, thus involves the increasing interdependence of states that is necessary to ensure that people’s rights are grounded in a truly “omnilateral” will.
Grounding political cosmopolitanism on the same basis as that of rights more generally has important implications for the nature of that cosmopolitanism. First, political cosmopolitanism is morally necessary to provide a legitimate basis of rights. Second, because political cosmopolitanism “is not an . . . ethical principle but a principle having to do with rights” (6:352), it does not relate to internal ends, either of states or individuals within them. As Kant emphasizes, “the concept of right . . . has to do . . . only with the external . . . relation of one person [or state] to another” such that “no account at all is taken of the . . . end that each has in mind” (6:230). Thus political cosmopolitanism neither depends on nor gives rise to good wills. Rather, it simply ensures that in their actions, human beings (whether as individuals or as states) act in ways that allow maximum freedom of action for others. The first implication, that cosmopolitanism is morally necessary, implies that political cosmopolitanism would be required even for morally perfect angels. The second, that it deals only with external actions, makes this cosmopolitanism possible even for “a nation of devils” (8:366).
In his Anthropology from a Pragmatic Point of View and Religion within the Boundaries of Mere Reason, however, Kant discusses something that might be called an ethical cosmopolitanism: “an enduring and ever expanding society, solely designed for the preservation of morality” (6:94). In contrast to political cosmopolitanism, which seeks merely to ensure that the actions of states are in accordance with cosmopolitan right, ethical cosmopolitanism promotes the cultivation of virtue. Unlike political cosmopolitanism, then, this cosmopolitanism will be unnecessary for angels, who are already virtuous, and impossible for devils, who lack any basis for becoming virtuous. But like political right, and to an even greater extent, this ethical community must be truly cosmopolitan, encompassing all human beings: “a multitude of human beings united in that [moral] purpose cannot yet be called an ethical community as such but only a particular society that strives after the consensus of all human beings . . . in order to establish an absolute ethical whole” (6:96).
Kant discusses ethical community in the greatest detail in his Religion within the Boundaries of Mere Reason, a work within which Kant recasts traditional Christian theological beliefs in terms of his own philosophy. The Christian doctrines that form the backbone of the book are the doctrines of human sinfulness and God’s mercy, and Kant’s ethical cosmopolitanism cannot be understood without reference to these central doctrines. But this dependence upon fundamentally Christian concepts taints the cosmopolitan character of Kant’s ethical project here. In particular, this dependence raises the question of whether an ethical ideal founded on specific theological claims can ever meet the requirements for a truly universal cosmopolitanism.
In this paper, I propose two readings of Kant’s cosmopolitanism that offer two different solutions to the problem of reconciling religious commitment with cosmopolitan ideals. The first, a “liberal” reading of Kant’s cosmopolitanism insists that theological commitments are not a necessary part of the content of cosmopolitan ethical community. A liberal need not rule out any religious beliefs per se, but only a necessary role for them in ethical communities. The second, a “religious” reading of Kant, sees a role for limited “rational theology” – including belief in God’s existence and mercy – as a necessary part of ethical cosmopolitanism. Each reading has advantages, and in my conclusion I suggest that Kant’s Religion within the Boundaries of Mere Reason can be seen as attempt to reconcile them, though one that ultimately favors the religious reading. Before turning to these interpretations, however, I must first lay out in more detail why ethical cosmopolitanism is necessary, and what problems in particular these readings of Kant are designed to solve.
2. The need for ethical cosmopolitanism
For Kant, ethical cosmopolitanism is necessary because of human beings’ propensity to evil and the role that social life plays in cultivating human evil. The notion of a propensity to evil comes from Kant’s recasting of the Christian doctrine of universal sinfulness in terms of his account of “radical evil,” and it is this doctrine that provides the reason for the importance of ethical cosmopolitanism. The locus classicus for this doctrine within the Christian tradition is the Apostle Paul’s claim in Romans that “There is no one who is righteous, not even one,” a passage that Kant explicitly connects with his own theory of radical evil:
[I]f nowhere is a virtue which no level of temptation can overthrow, if whether good or evil wins us over only depends on which bids the most and affords the promptest pay-off, then what the Apostle says might indeed hold true of human beings universally, “There is no distinction here, they are all under sin – there is none righteous (in the spirit of the law), no, not one.” (6:39)
That all are “under sin” is precisely the claim that Kant defends in this section of his Religion, a section entitled “the human being is by nature evil” (6:32).
Kant not only defends the existence of universal and radical evil; he philosophically describes the nature of that evil. In general, radical evil takes the form of a subordination of the moral law to maxims of self-love. When a person allows him or herself to choose in such a way that the moral law can be overridden by sensuous incentives, that person is evil. And this overriding need not occur all the time:
The statement, “The human being is evil,” cannot mean anything else than that he is conscious of the moral law and yet has incorporated into his maxim the (occasional) deviation from it. (6:32)
Kant is, in this sense, a “rigorist” about morality – even the occasional deviation from complete prioritization of the moral law to one’s sensuous interests constitutes “evil.”
For Kant, however, radical evil goes beyond mere occasional deviation from the law. Human evil also involves the cultivation of a propensity to evil, a subjective disposition that makes one more likely to act on evil maxims in the future. Radical evil is not simply a matter of choosing maxims that are themselves contrary to the moral law. It involves acting on maxims that constitute or bring about a propensity to and persistence in evil. One who is evil – and all humans are – chooses evil now and also seeks to ensure that evil will continue in one’s choices throughout life.
For Kant, however, the cultivation of this propensity to evil is not a solitary endeavor. Human beings live in societies, and our social lives provide fertile breeding grounds for evil. As Kant explains,
If [a human being] searches for the causes and the circumstances that draw him into this danger [i.e., assault of the evil principle] and keep him there, he can easily convince himself that they do not come his way from his own raw nature . . . but rather from the human beings to whom he stands in relation or association. (6:93)
The primary means by which we cultivate our own worst tendencies are social. Competition, resentment, and a myriad of diverse desires arise only in social contexts, and the “passions,” which for Kant are among the greatest hindrances to self-mastery and virtue, “assail his nature, which on its own is undemanding, as soon as he is among human beings” (6:94). Radical evil manifests itself not merely in individual wrongdoing, but in a social climate that fosters vice.
This, then, brings us to the need for ethical cosmopolitanism. For Kant, the propensity to evil is not merely an unfortunate fact about human beings; it sets an important moral task. As Kant explains,
[A human being] remains . . . exposed to the assaults of the evil principle; and, to assert his freedom, which is constantly under attack, he must henceforth remain forever armed for battle. (6:93, cf. 6:408)
Human beings cannot rest easy in ethical life. We cannot simply face each decision with the firm conviction to follow the categorical imperative. Instead, we must recognize our own propensity to evil and actively arm ourselves against its influence. As Kant explains in his criticism of stoic ethics,
They drew the moral laws directly from reason . . . and so was everything quite correctly apportioned . . . subjectively, with respect to the incentive – provided that one attributes to the human being an uncorrupted will . . .. The mistake of those philosophers, however, lay in just this last presupposition. For . . . we must rather start by dislodging from its possession the evil which has already taken up possession [in the will]. (6:58n)
Were human beings morally pure, ethical cosmopolitanism would be unnecessary, and every reader of the Groundwork would become (or better, would already be) a morally perfect follower of the categorical imperative. But because we actively promote evil in ourselves, we can become good only by combating this evil tendency. And because our propensity to evil is cultivated by our social relationships in particular, one can struggle against corrupting influences only through the moral reform of society.
The dominion of the good principle is not otherwise attainable . . . than through the setting up and the diffusion of a society in accordance with, and for the sake of, laws of virtue. (6:94)
Thus ethical cosmopolitanism is the morally necessary consequence of our own radical evil. Without a social struggle for more improvement, human beings will lay down arms in a battle in which we are required to be ever vigilant.
3. The first problem with ethical cosmopolitanism: radical evil
Radical evil is crucial for Kant’s argument that ethical cosmopolitanism is needed, but it raises an apparent problem for Kant because of two further commitments: (1) One can rationally hope to be fundamentally morally good and (2) There is no middle ground between moral good and evil since the moral law insists on unswerving obedience (rigorism). The first of these commitments lies at the heart of Kant’s whole practical philosophy, which is based on seeking the conditions of the possibility of morality. If it is impossible to be morally good – if there is no hope for us – then Kant’s philosophy is practically inert. The second commitment follows (as noted above) from the specific nature of Kant’s morality, and in particular from the categorical and universal nature of the moral law. John Hare nicely summarizes the challenge:
If we want to keep morality as demanding as Kant says it is, and if we want to concede what Kant says about our natural propensity not to live by it, and if we want at the same time to reject these traditional Christian doctrines [of grace], then we will have to find some substitute for them. (Hare 1996: 37)
Hare does not include (1) in this description, although denying (1) is a way to avoid Christian doctrines or their substitutes. One can simply admit the rigor of morality and our propensity not to live by it, but claim that we have freely fallen short and now are morally evil without any further hope of reform. This would be an option, but it is not one that Kant is willing to accept.
The doctrine of radical evil seems to challenge Kant’s commitments to moral hope and moral rigorism because radical evil, at least as Kant describes it, seems to be inextirpable.
This evil is . . . not to be extirpated through human forces, for this could happen only through good maxims – something that cannot take place if the subjective supreme ground of all maxims is presupposed to be corrupted. (6:37, cf. 6:45)
There are two reasons, for Kant, that radical evil is inextirpable. First, given that one has subordinated the moral law to sensuous inclinations, it can never be the case that one completely prioritizes morality over inclination. One’s overall moral status depends on one’s life as a whole (6: 22-25). To be morally good, one cannot ever compromise morality. But if one has already compromised morality, even if one always does one’s duty from now on, one is nonetheless a person who, given the right circumstances (which may include temporal conditions), violates the moral law. That is to say, one is nonetheless evil. In Kant’s terms,
however steadfastly a human being may have persevered in such a [good] disposition in a life conduct conformable to it, he nevertheless started from evil, and this is a debt which is impossible for him to wipe out. (6:72)
The evil from which we have started provides a reason for the inextirpability of evil overall.
The second reason that evil is inextirpable is that one’s evil is not merely past evil deeds, but includes a propensity to evil. The fundamental maxim governing one’s life is a commitment to prefer inclination over morality. And this fundamental maxim provides no ground for its own overturning. One will not reject evil because the basis of one’s decisions is evil. And although obstacles to choosing rightly do not erode one’s responsibility for one’s evil, they seem to undermine the real possibility of moral reform.
Given rigorism (commitment 2), the inextirpability of radical evil leaves one without hope of being morally good. But Kant still insists that what is required by morality is that one do what is right, which includes improving the basis of one’s choices. As Kant explains, “In spite of the fall, the command that we ought to become better human beings still resounds unabated in our souls; consequently, we must also be capable of it” (6:45). Kant is unwilling to give up his commitment to moral hope. But if one ought to become better, one must still be capable of moral reform. And that means that radical evil must be eradicable after all.
One approach that might seem promising, given Kant’s commitment to ethical cosmopolitanism, is to turn to society to overcome the radical evil that is found in each individual. In his Anthropology, Kant suggests something like this:
No matter how great his animalistic inclination may be to abandon himself passively to the enticements of ease and comfort . . ., he is still destined to make himself worthy of humanity by actively struggling with the obstacles that cling to him because of the crudity of his nature. Man must, therefore, be educated to the good. (6:324-5, emphasis added)
But Kant immediately realizes that the turn to education, or to the social, will not avoid the problem posed by radical evil. Instead, it only makes that problem more acute, because “he who is to educate him is again a human who still finds himself in the crudity of nature. This human, now, is expected to bring about what he himself is still in need of” (6:325). This challenge arises in any attempt of the human race to educate itself, but especially in the context of moral education:
Since good men, who must themselves have been trained for it, are required for moral education, and since there is probably not one among them who has no innate or acquired depravity himself, the problem of moral education for our species remains unsolved. (6:327)
A society of radically evil individuals is a seedbed for further evil, and thus Kant requires an ethical community to promote the social struggle against evil that marks our only hope for moral progress. But this ethical community itself depends upon having already overcome the very problem that it is supposed to solve, the problem of the propensity to evil.
4. Solving the problem of radical evil with divine mercy.
Kant solves the problem of radical
evil by invoking the doctrines of providence and divine mercy. In his Religion,
he explains that “Some supernatural cooperation is also needed to [a person]
becoming good or better” (6:44). And in his Anthropology, he says, “it is only from
To avoid despair, one must believe that there is some way to be morally good, but practical reason will not condone any lenience in the interpretation of the moral law. Thus there must be some inconceivable supplement for our failings, some supplement that reason does not fully specify.
It may be impossible for an individual to overcome the propensity to evil without some assistance, but God can mercifully provide for such transformation. From God, one can hope that one will be capable of overcoming one’s own radical evil. But Kant makes clear that the revolution made possible by divine mercy does not absolve one of the responsibility to actively promote one’s own virtue. Even with the cooperation of providence and divine mercy, the most that one can expect is “an endless progress toward the complete conformity” of human wills with the moral law (5:122), a “battle . . . against the attacks of the evil principle” (6:93) and a constant “struggle” (6: 78) of one’s efforts to be good against one’s tendencies for evil.
Likewise, it may be impossible for
the human species to overcome the social corruption that fosters the propensity
to evil. Kant goes so far as to say that
“even with the good will of each individual, because of the lack of a principle
which unites them, [human beings] deviate . . . from the common goal of
goodness as though they were instruments of evil” for one another (6:97). The corrupting effects of society are strong
and self-reinforcing, and “the problem of moral education for our species
remains unsolved” (6:327), at least at a
human level. At the level of
providence, one can hope not only for God’s merciful intervention facilitating
individual transformation, but also for the providential establishment of an
ethical community: “To found a moral people of God [i.e., a cosmopolitan
ethical community] is, therefore, a work whose execution cannot be hoped for
from human beings but only from God himself” (6:100). Again, Kant does not take God’s role to
absolve human beings of moral responsibility: “human beings are not permitted
on this account to remain idle in the undertaking and let
5. The second problem with ethical cosmopolitanism: the role of religion
Invoking providence and divine mercy helps alleviate the apparent conflict between Kant’s commitments to moral hope, rigorism, and the reality of radical evil. Human beings are radically evil, but we still have hope of satisfying the demands of morality through God’s mercy. However, the invocation of divine mercy raises a potential problem with Kant’s account because of the nature of Kantian ethical cosmopolitanism and the sources of his account of divine mercy. In particular, Kant insists that ethical cosmopolitanism does not depend upon divine revelation nor upon specific features of any revealed religion, but divine mercy is a core doctrine of Christianity – especially German Protestantism – and Kant’s explanation of it is permeated with references to Christian Scriptures.
Kant was well aware of the importance of the doctrine of divine mercy within the Christian tradition, especially within German Lutheranism and the Pietist forms of that Lutheranism with which he grew up. Gordon Michalson describes his doctrine of radical evil as “a Kantian adaptation of the Lutheran simul justus et peccator” (Michalson 1990:117). Similarly, Robert Adams points out how Kant is “like his Lutheran forebears” (Adams 1998:xviii) and suggests that “his thought about good and evil in human nature is deeply attuned to the dynamics of the Lutheran piety in which he was raised” (Adams 1998:xv). And John Hare argues that Kant “takes the four central items of the traditional Christian faith (Creation, Fall, Redemption, and Second Coming) and the translation of these dictates the structure of the rest of [Religion within the Boundaries of Mere Reason]” (Hare 2004, cf. Hare 1996). The sense that Kant is drawing on a specifically Christian (even Lutheran) doctrine of divine mercy is heightened by Kant’s extensive use of Christian Scriptures relevant to his account of divine mercy. In Religion within the Boundaries of Mere Reason, there are over 100 references to Christian Scriptures, making up over half of the references in the text. The close connection between German Lutheranism and Kant’s doctrine of divine mercy need not imply a dependence upon specifically Christian authorities or arguments, but there is at least an apparent tension between the theological commitments that underlie Kant’s hope for ethical cosmopolitanism and his account of the nature of that cosmopolitanism.
Kant allows ethical cosmopolitanism to involve religion in a sense. As he explains, “an ethical community is conceivable only as a people under divine commands, i.e. as a people of God” and “the idea of a people of God cannot be realized (by human organization) except in the form of a church” (6:99, 100). But Kant is explicit that true cosmopolitanism depends on freeing moral religion from “revelation” and anything tied to “historical” or “ecclesiastical” faith (cf. e.g. 6:106, 109). For Kant, the need to free religion from anything historical or revealed is a necessary consequence of linking religion to cosmopolitan ethical community. For an ethical community to be truly cosmopolitan, it must not depend on anything that cannot “be convincingly communicated to everyone” (6:103). Kant insists that any “religion” in an ethical cosmopolitanism must be purified of contingent details of different “faiths” that cover the globe (6:107-8).
But this requirement of universality for the religion that underlies ethical community brings with it an apparent problem in the context of Kant’s discussion of divine mercy. On the one hand, Kant’s account of divine mercy suggests that it must be an essential part of any ethical community. Without divine mercy, one apparently must sacrifice either the hope of moral progress, the rigor of the moral law, or the recognition of one’s own radical evil. Giving up the first would make ethical cosmopolitanism hopeless, dissolving it in moral despair. Giving up the second would fail to be truly ethical cosmopolitanism and would cultivate a sort of moral laxity rather than true virtue. And giving up the third, in addition to being intellectually dishonest given “the multitude of woeful examples that the experience of human deeds parades before us” (6:33), would make the pursuit of ethical cosmopolitanism unnecessary, since ethical cosmopolitanism is needed only to combat our own evil.
But on the other hand, the doctrine of divine mercy seems to be a specifically Christian doctrine, and Kant defends it in specifically Christian terms. As we have seen, Kant explicitly refers to the apostle Paul and to the Christian Scriptures in explaining the reality of radical evil, even connecting it with Paul’s claim that all are “under sin” (6:39, cf. Romans 3:9). And Kant draws his solution to this problem, at least in part, from Paul’s further claim that “since all have sinned and fall short of the glory of God, they are justified by his grace.” Kant’s discussion of the mechanics of divine mercy is permeated throughout with Christian concepts and jargon. Kant discusses, for example, how the “son of God . . . bears as vicarious substitute the debt of sin for him, and also for all who believe . . . in him; as savior, he satisfies the highest justice through suffering and death” (6:74). Thus whereas Kant’s doctrine of divine mercy seems to be an essential part of his moral religion such that any cosmopolitanism ethical community must include this doctrine as an essential part, it also seems to be an idea deeply rooted in a particular faith, tied to specific revelation, and thus must be precluded from the context of any truly cosmopolitan ethical community. And hence, we have an apparent paradox.
6. Solving the problem: some preliminaries
As clear as it may seem, this apparent paradox is not a real one. There is nothing strictly inconsistent with saying that divine mercy is necessary for ethical cosmopolitanism but that a doctrine of divine mercy has no place in that cosmopolitanism. If there is a God who is the Creator and Sustainer of the universe, then presumably the existence of God is necessary in order for human beings to exist. But even those who believe that God is Creator and Sustainer of the universe do not take this to imply that one must believe in God in order to exist. Similarly, it may be that divine mercy is necessary in order for ethical cosmopolitanism to be possible, but that does not necessarily mean that members of a cosmopolitan ethical community must believe in divine mercy.
Moreover, Kant’s prohibitions on including contingent details of faith in ethical religion are directed against specifically historical aspects of faith. And while divine mercy as a solution to the problem of human evil is a doctrine tied to a specific religious tradition, and while Kant draws heavily on historical features of Christianity in his explanations of divine mercy, he also interprets those features in ways that are less historically contingent. Thus with respect to the role of the “Son of God” as “savior” and “vicarious sacrifice,” Kant interprets this Christian doctrine in a way that does not require any specific belief in the historical person of Jesus Christ as the Son of God, but only in some Son of God whose sacrifice atones for one’s past sins.
Still, however, Kant emphasizes the importance of divine mercy in giving a philosophical defense of the reasonableness of pursuing ethical community, but he explains grace in specifically Christian terms. Even without his Christian rhetoric, one might think that grace as a divine response to human sinfulness is a doctrine just too tied to historical Christianity to have a place in any truly cosmopolitan ethical community. And this at least raises the question of what role this doctrine of divine mercy has in the context of ethical community. The two ways of resolving the problem I have just mentioned, in fact, suggest two different roles for divine mercy. The first resolution – that divine mercy is necessary but not belief in it – would imply that those seeking to promote ethical community need not specifically seek to convince others of the importance of divine mercy, since such a belief is inessential to what ethical community involves. We might call this the ‘liberal reading’ of Kant’s cosmopolitanism, since it does not require any specific religious beliefs. The second resolution – that divine mercy can be cleansed of its specifically historical features – suggests that one can (and perhaps even should) promote this non-historical doctrine of mercy as a part of ethical cosmopolitanism. Thus on this ‘religious reading’ of Kant, religious beliefs are necessary for cosmopolitanism, though not contingent religious beliefs.
It is important to note here, of course, that liberal cosmopolitanism is consistent with a great deal of religiosity, and a religious cosmopolitanism is consistent with a great deal of liberalism. With respect to the former point, one can be deeply religious at a personal and even corporate level while still holding a “liberal” view towards ethical community, as long as one believes that belief in the doctrine of divine mercy is not necessary for a flourishing and cosmopolitan community. With respect to the latter, one can hold to a religious cosmopolitanism and still be quite liberal with regard to the diversity of ways of understanding divine mercy. To hold a “religious” stance towards ethical community is simply to believe that the belief in divine mercy is necessary for ethical community. And of course, neither view about ethical cosmopolitanism – religious or liberal – commits one to any particular view about political cosmopolitanism. In particular, the religious view that certain core theological doctrines are required for the flourishing of ethical community need not imply that these doctrines (or any others) are required for political community.
7. The liberal reading: Belief in divine mercy is not an essential part of ethical cosmopolitanism
As we have seen, the doctrines of providence and divine mercy play a role in Kant’s philosophical defense of the possibility of ethical cosmopolitanism, but on the liberal reading of this cosmopolitanism, theological commitment to divine mercy is not part of the content of cosmopolitan ethical community. That is, one need not actually believe in divine mercy to be fully a part of an ethical community, and one need not promote this belief as part of promoting “an ever expanding society . . . designed for the preservation of morality” (6:94).
Allen Wood defends something like this liberal reading in his account of Kant’s religion. He argues that “I can have religion . . . even if I am an agnostic,” as long as I believe that it is possible that God exists and “so long as my awareness of duty is enlivened with the thought that if there is a God, then my duties are God’s commands” (Wood 1992: 406). Wood does not specifically apply this liberal approach to divine mercy, but we can easily reconstruct what he could say. For Wood, the key relevant desideratum in articulating the nature of a true moral religion is that “Kant does not want to find moral fault with anyone whose religious beliefs fall within the range of opinion that is compatible with the theoretical evidence” (405). And just as Wood’s “hopeful agnostic” does not actually believe in God but believes that God’s existence is possible and bases moral hope on that, the agnostic could refrain from actually believing in divine mercy, so long as she allowed that divine mercy is possible. Kant’s language sometimes even suggests that this is what he has in mind, as when he insists that “Reason does not contest the possibility . . . . of these objects” and describes a “faith . . . with respect to the possibility of this something” (6: 52, emphasis added).
We might even go further than Wood. As long as a person is willing to commit to the pursuit of virtue, they can be a member of an ethical community, whether they have any beliefs about divine mercy at all. As long as one does not both see divine mercy as necessary for virtue and deny its possibility, one can be a member of an ethical community. Philosophers, such as Kant, may need to posit some sort of divine mercy to make sense of how ethical cosmopolitanism could be possible, but ordinary people need not have any beliefs about this at all, as long as they are willing actively to promote ethical community.
For the content of ethical cosmopolitanism, this liberal reading has an important advantage. It suggests that the belief in divine mercy, which (in Kant at least) is closely connected with Christianity, is an inessential component of the moral religion about which one seeks “the consensus of all human beings” (6:96). Eliminating a need to reach consensus about the nature of divine mercy is particularly important given the conflicts to which that doctrine has given rise both within Christianity itself and in interactions between Christians and others. These conflicts could pose problems for any proponent of ethical cosmopolitanism that sought to impose doctrines of mercy informed by Christianity on a world of Jews, Muslims, Hindus, Buddhists, agnostics, and others. The more one can eliminate doctrines of faith from ethical community, the easier it will be to promote a truly cosmopolitan religion.
Moreover, Kant points out that with respect to the doctrine of divine mercy in particular, belief in this doctrine can have a dangerously counterproductive effect on the cultivation of virtue. He explains that this doctrine is “very risky” because “what is to be accredited to us as morally good conduct must take place . . . through the use of our own powers” (6:191) and thus that
It can . . . be asked whether this deduction of the idea of a justification of a human being who is indeed guilty . . . has any practical use at all, and what such use could be. It is hard to see what positive use can be made of it for religion and for the conduct of life. (6:76)
What worries Kant here is the possibility that the belief in divine mercy will lead people to think that certain rituals or formulae can gain mercy in the absence of a life well lived. If people rest hope on divine mercy as opposed to a good life, then the doctrine of divine mercy will actually inhibit ethical cosmopolitanism (the purpose of which is, remember, to promote virtue), rather than promote it. Thus Kant emphasizes vehemently that “Apart from a good life-conduct, anything which the human being supposes that he can do to become well-pleasing to God is mere religious delusion” (6:170, Kant’s emphasis).
On the liberal reading, then, divine mercy is not an essential part of ethical cosmopolitanism. One can be a member of a cosmopolitan ethical community without believing in divine mercy, and one can promote that community without promoting belief in divine mercy. This reading has two important advantages. First, it fits well with Kant’s claims about the universal communicability of “pure” religion. Divine mercy seems not to be easily and “convincingly communicated to everyone,” so leaving it out of one’s cosmopolitan religion seems required. Second, belief in divine mercy can have a tendency to make people morally complacent, hoping that God will simply take care of them or that certain rituals will gain them divine favor. Purifying ethical community of a doctrine with these effects would seem to promote the cultivation of virtue that it is the purpose of ethical community to foster.
8. The religious reading: Belief in divine mercy is an essential part of ethical cosmopolitanism
Like a liberal reading of ethical cosmopolitanism, the religious reading sees no role for merely historical aspects of faith, but a religious reading of Kant sees a role for limited rational theology – including belief in God’s existence and mercy – as a necessary part of ethical cosmopolitanism. Thus to be a fully participating member of an ethical community one would need to believe in God’s mercy, and to promote ethical cosmopolitanism, one would need to promote this belief. Since joining ethical communities and promoting such cosmopolitanism are morally required, belief in God’s mercy would thus be required. Despite the apparent problems with this view discussed in the previous section, there are two important (and related) reasons to advocate a religious reading.
The first is that Kant seems to think that the belief in divine mercy is rationally required by anyone seriously pursuing personal virtue and ethical cosmopolitanism. Shortly after the passage quoted above in which Kant claims that “It is hard to see what positive use can be made of it for religion and for the conduct of life,” he adds,
The investigation is only an answer to a speculative question, but one that cannot therefore be passed over in silence, since reason could then be accused of being absolutely incapable of reconciling the human being’s hope of absolution from his guilt with divine justice, and this accusation might be disadvantageous to reason in many respects, most of all morally. (6:76, emphasis added)
Despite the apparent dangers of the doctrine of divine mercy, and thus Kant’s resistance to giving it too much weight practically, it is a doctrine that is necessary in order to resolve otherwise impossible conflicts within reason. And these conflicts, though speculative, are not merely speculative. They are morally “disadvantageous.” In this respect, the belief in divine mercy is like belief in the practical postulates that Kant discusses in his Critique of Practical Reason. Just as “it is morally necessary to believe in God” because “our reason finds [the possibility of the highest good] thinkable only on the presupposition of a supreme intelligence” (5:125-6), so it is similarly “morally necessary” to believe in divine mercy, because only divine mercy can reconcile moral hope with a rigorous morality.
Unlike the postulates, the need for divine mercy does not follow simply from the fact that one has a duty to obey the moral law, nor even from the fact that this duty implies that “one ought to strive to promote the highest good” (5:125). One’s moral obligations combined with radical evil are what make divine mercy necessary. But although Kant’s the theory of radical evil is related to the Christian doctrine of sin, Kant insists that “only common morality is needed to understand the essentials of this text [Religion within the Boundaries of Mere Reason]” (6:14). And Kant’s overall defense of the reality of radical evil, although it uses Christian terms and references Christian Scriptures, is not based on Christianity. Rather, he sketches an empirical proof of radical evil based on “the multitude of woeful examples that the experience of human deeds parades before us” (6:32-33). As he explains, “according to the cognition we have of the human being through experience . . ., we may presuppose evil as subjectively necessary in every human being” and “the existence of this propensity to evil in human nature can be established through experiential demonstrations” (6:32, 6:35, cf. 6:20). The essence of this demonstration is that people’s actions in the world provide irrefutable evidence that they have corrupt maxims, and thus that they are evil (6:24, cf. Frierson 2003: 104-8). Kant claims that the presence of radical evil is something that “everyone can decide for himself” (6:39), but anyone honestly surveying their own life will find ample evidence that one lacks a wholehearted commitment to always obey the moral law. In giving this empirical argument for the universality of radical evil, Kant highlights those aspects of the Christian religion that can be found “within the limits of reason (understood as including empirical judgments) alone.” In that sense, the quasi-postulate of divine mercy is grounded in an argument that can “be convincingly communicated to everyone,” and thus it is both acceptable within an ethical cosmopolitanism and (like the postulates) rationally required (though not proven speculatively) for those who genuinely seek virtue.
The second reason that divine mercy may need to be part of any ethical cosmopolitanism is the psychological correlate of the first. It is not only irrational but also psychologically difficult, if not impossible, to promote ethical cosmopolitanism (or even one’s own virtue) without believing in divine mercy. The psychological character of Kant’s concern here can be understood by a further analogy with the postulates of practical reason. In the Critique of Judgment, Kant revisits these postulates, but in a way that puts them in a more psychological and less epistemic light. He discusses “a righteous man (like Spinoza) who takes himself to be firmly convinced that there is no God” (5:452) and he asks what kind of life such a man will live. Kant explains that this righteous man will pursue the moral law from pure motives, but “his effort is limited” because nature does not, as far as he can tell, cooperate with him. There are two possible results of this, Kant says. One is that such a Spinoza will eventually “give up as impossible” his morally righteous ends, regardless of how “well-intentioned” he is. The only way to avoid this effect on one’s disposition is to “assume the existence of the moral author of the world, i.e. of God” (5:452-3). Unlike the derivation of the rational need to believe in God, Kant’s point here is psychological. Theoretically, human beings could be morally good without believing in God, but (according to Kant) atheism will “weaken the respect” for morality and “damage the moral disposition.” One should believe in God in part because that belief will function as an aid to cultivating a moral disposition, whereas not believing will weaken one’s commitment to morality.
Kant’s psychological claim about the effects of belief in God can be extended to his treatment of God’s mercy and divine providence. The arguments for the important of grace are rooted in a philosophical problem, but also in a psychological one. Without believing in divine mercy, one cannot reconcile the absolute demands of the moral law and the fact of one’s own radical evil with a continuing commitment to the moral law. And while it may theoretically be possible to maintain a long-term commitment to the moral law while believing that one has no hope of meeting its demands, this is probably impossible psychologically. Without a belief in divine mercy, the only way (psychologically) to avoid weakening one’s commitment to the moral law is to lessen the demands of the moral law or to deceive oneself about the severity of one’s own evil. Without a belief in divine mercy, the only options that are psychologically open to human beings seem to be moral despair, moral compromise, or hypocritical self-satisfaction.
If this psychological claim is true (and it is certainly an open question whether or not it is true), it provides important reasons for endorsing a version of the religious interpretation of Kant’s ethical cosmopolitanism. The purpose of ethical cosmopolitanism is to cultivate virtue in oneself and others, and the need for it emerges from human beings’ social nature and the way that radical evil takes advantage of that social nature. But our social nature is not the only aspect of our nature relevant to radical evil, nor to our efforts to correct it. We are also beings whose commitment to morality is affected by our beliefs, and in particular our religious beliefs. Thus an ethical community must actively promote those beliefs that are necessary (or even helpful) to the cultivation of virtue.
In the last two sections, we have seen good reasons to advocate both a liberal reading of Kant’s cosmopolitanism and a more religious one. The strength of the liberal account is its commitment to the principle of “the consensus of all human beings” (6:96) that must underlie any truly cosmopolitan ethical community. However, a liberal cosmopolitanism risks embarking on a failing project if it seeks to promote virtue without attending to the role that the belief in divine mercy plays in counteracting the moral despair that can come with recognition of the reality of radical evil. In this respect, the strength of the religious reading is its awareness of the rational and psychological importance of the doctrine of divine mercy in cultivating virtue in human beings. However useful this doctrine is, however, it is useful for a cosmopolitan ethical community only if it can be communicated in a way that can be universally accepted.
Here, however, Kant’s account of ethical community in the Religion can help provide support for a nuanced version of the religious reading. Although Kant appeals to Christian Scriptures more than to any other source in his Religion, he consistently interprets those Scriptures “within the boundaries of mere reason.” Given Kant’s recognition of the importance of the doctrine of divine mercy, this text can be read as a first attempt at articulating that doctrine in ways that are not specifically Christian. Even the appeals to Scripture can be read not as historically rooted justifications of his claims but as models for how revelation should be reinterpreted to make it fit rational religion. Thus Kant’s accounts of radical evil, providence, and divine mercy provide universally acceptable versions of deeply religious claims. He gives what would otherwise seem to be historical doctrines of faith a cosmopolitan interpretation, and thereby helps, if only in a small way, to promote that ethical cosmopolitanism which will bring about a universal community of virtue, a “kingdom of God on earth.”
 Because this ideal is impractical, Kant allows for a different structure of cosmopolitan right, a “league that averts war, endures, and always expands” (8:357). As Kant explains,
This league does not look to acquiring any power of a state but only to preserving and securing the freedom of a state itself and of other states in league with it, but without there being any need for them to subject themselves to public laws and coercion under them . . .. This could provide a focal point of federative union for other states, to attach themselves to it and so to secure a condition of freedom of states conformably with the idea of the right of nations; and by further alliances of this kind, it would gradually extend further and further. (8:356)
Despite the concession to practicality that requires a “league” or “federation” of states rather than a single overarching state, Kant’s general point is constant. Individuals in a state of nature must give up their lawless freedom to enter into a civil condition (a state), and nations must give up lawless freedom vis a vis one another to enter a cosmopolitan whole.
 As Kant explains in Towards Perpetual Peace,
reason, from the throne of the highest morally legislative power . . . makes a condition of peace, which cannot be instituted or assured without a pact of nations among themselves, a direct duty, so there must be a league of a special kind, . . . a pacific league (foedus pacificum). (8:356)
 Devils, as Kant understands them, are hopelessly committed to evil. They could not benefit from ethical communities, the purpose of which is to foster one’s commitment to good, nor could they form ethical communities, which depend on people ingenuously seeking to promote goodness in themselves and others.
 Calling this second reading a “religious” reading is potentially misleading, since a liberal interpretation of Kant can be just as religious as a religious one. One could adhere to a liberal version of Kantian ethical cosmopolitanism while still being quite religious personally, and even allowing for the possibility of religious belief within particular ethical communities. One simply denies that any religious beliefs (including one’s own) are necessary for ethical cosmopolitanism. In that sense, liberal cosmopolitanism need not be secular – i.e. nonreligious – cosmopolitanism. It is also important to point out that one could hold what I am calling a religious view of ethical cosmopolitanism while remaining a liberal about political cosmopolitanism. That is, one could see religious belief as essential to the promotion of an ethical community but not essential to the establishment of a stable political order.
 This universal sinfulness is not the same as “original sin” as traditionally understood within Christianity. Although the two doctrines are connected within Christianity, Kant is careful to distinguish his claim that human beings are radically evil, and even evil “by nature,” from any doctrine that would ascribe this evil to the choice of another person. For more, see Quinn 1984.
 Paul himself develops this doctrine in the light of earlier Jewish sources. This passage in particular is taken from the Psalms (cf. Psalm 14:2-3). The doctrine becomes a central doctrine within the Christian tradition. For some influential examples, see Augustine, Aquinas’s Summa Theologica I-II QQ 74-89, John Calvin’s Institutes of the Christian Religion Books II and III, and Martin Luther’s The Bondage of the Will.
 Kant says, for instance,
If [a person] took [incentives of his sensuous nature] into his maxim as of themselves sufficient for the determination of his power of choice, without minding the moral law . . ., he would then become morally evil. (6:36)
Throughout this discussion, it is important to remember that for Kant self-love, or “incentives of his sensuous nature,” can include motives that are normally considered altruistic. The point is simply that one does what is in one’s sensuous interests, either in that it immediately satisfies some desire that one happens to have or that it is conducive to the satisfaction of desires at some later date. Thus someone who acts from a good-hearted desire to see others flourish is acting from “self-love” in this Kantian sense.
 This is a doctrine that Kant defends with reference to his moral philosophy. As Kant explains, “a human being can [not] be morally good in some parts, and at the same time evil in others. For if he is good in one part, he has incorporated the moral law into his maxim. And were he, therefore, to be evil in some other part, since the moral law of compliance with duty is in general a single one and universal, the maxim relating to it would be universal and yet particular at the same time, which is contradictory” (6:25). The argument here starts with Kant’s account of the moral law, as “single and universal.” The moral law precisely commands that we will in such a way that we do not make our particular circumstances determinative for whether an action is right or not. But if we will differently at different times or in different contexts, this shows that we have never really acted on a categorical imperative that commands that particular details not determine our actions. We have acted rightly sometimes, but even then, our underlying maxim allowed for deviation, as is evident by actual deviation in a different context.
This rigorist doctrine thus fits well with Kant’s overall moral philosophy, but it is also an important example of religion within the boundaries of mere reason. The book of James in the Christian New Testament insists that “whoever keeps the whole law and yet stumbles at just one point is guilty of breaking all of it” (James 2:10). Kant here provides a philosophical argument to support a fundamentally Christian doctrine of rigorism, and Robert Adams connects this with “one of the points at which his thought about good and evil in human nature is deeply attuned to the dynamics of the Lutheran piety in which he was reared.” (Adams 1998: xv).
 For an extended defense of the claim that human evil involves the cultivation of a propensity to evil, see Frierson 2003: 108-13.
 For more detail on this, see Anderson-Gold 2001, Frierson 2003, and Wood 1991 and 1999.
 For Kant, “passions” (Leidenschaften) do not exhaust the range of what we would today call “emotions,” and Kant does not think that emotions in general are threats to self-mastery. For more on the relationship between passions and emotions more generally, cf. Sorenson 2002 and Borges 2004.
 John Hare nicely summarizes the challenge:
If we want to keep morality as demanding as Kant says it is, and if we want to concede what Kant says about our natural propensity not to live by it, and if we want at the same time to reject these traditional Christian doctrines [of grace], then we will have to find some substitute for them. (Hare 1996: 37)
Hare does not include (1) in this description, although denying (1) is a way to avoid Christian doctrines or their substitutes. One can simply admit the rigor of morality and our propensity not to live by it, but claim that we have freely fallen short and now are morally evil without any further hope of reform. This would be an option, but it is not one that Kant is willing to accept. For another description of the problem that closely follows the discussion in this paper, cf. Frierson 2003.
 In this respect, Kant’s treatment of ethical cosmopolitanism is quite different than his treatment of political cosmopolitanism. In the political case, Kant claims that nature can make use of our unsocial sociability to bring about a just political order. Because political cosmopolitanism deals only with actions and not with motives, we can be led to it by considerations that are not specifically moral. But in the case of ethical cosmopolitanism, the nature of the change is more deeply ethical, and thus this ethical community cannot be coerced (6:96) but can only come about through genuine effort towards moral reform.
 See too 6:45, 6:61, and 7:328, where Kant makes clearer the connection between our duty to morally improve and our need for divine grace. To understand the specific roles that grace plays, it is important to remember the two very different problems posed by radical evil (see footnote 13). First, the stain of radical evil cannot be removed regardless of the extent to which one improves in the future since one has done wrong in the past. Second, radical evil seems to undermine the possibility of transformation by deliberately hindering one’s own moral development, promoting a “propensity to evil.” Kant addresses the first problem with a conception of atoning grace, whereby one is justified before God, “after the fact,” so to speak. The second problem is addressed by a conception of sanctifying grace according to which God actually facilitates moral transformation. For more on these different senses of grace, see Adams 1998, Frierson 2003:114-22, Mariña 1997, Michaelson 1990, and Quinn 1986, 1984, and 1990. Kant distinguishes them throughout the Religion (see e.g. 6: 143) and in the Conflict of the Faculties (see 7: 43-44).
 For passages that more clearly connect this providence with divine mercy in particular, cf. Religion within the Boundaries of Mere Reason, 6:44; and The Conflict of the Faculties, 7:43-4.
 Similarly, in the Conflict of the Faculties, Kant repeats this point:
If worst comes to worst reason is entitled [befugt] to adopt on faith a supernatural supplement to fill what is lacking for his justification (though [reason is] not to specify in what this consists) . . .. We need not be able to understand and state exactly what the means of this replenishment is (for in the final analysis this is transcendent and, despite all that God Himself might tell us about it, inconceivable to us); even to lay claim to this knowledge would, in fact, be presumptuous. (7:43-44, my emphasis)
Like the Conflict of the Faculties, the lectures on religion both before and after the Religion emphasize the inscrutability of atoning grace and give little to no further details. The Religion was published in parts during 1792-93. The Conflict of the Faculties was written, at least in part, by December of 1793 (see letter to Kiesewetter, Ak. 11: 456), but it was not published until 1798. In the Religion, by contrast, although Kant does point out that grace is ultimately inscrutable, he offers some details about the way in which atoning grace works. See Quinn 1986 for a discussion of these details.
 There are numerous problems with Kant’s account of divine mercy that I do not discuss in this paper. For the purposes of this paper, I assume that invoking divine mercy solves the first problem. My primary interest here is on the second problem. I discuss some of the problems with mercy as a resolution to the first problem in Frierson 2003: 114-22. See too Adams 1998, Mariña 1997, Michaelson 1990, and Quinn 1986.
 The full context of the passage quoted here shows that particularly clearly:
The only faith that can found a universal church is pure religious faith, for it is a plain rational faith which can be convincingly communicated to everyone, whereas a historical faith, merely based on facts [or on revelation], can extend its influence no further than the tidings relevant to a judgment on its credibility can reach (6:103).
At the same time, however, Kant does not require that all religion be freed from these historical details immediately. “Due to a peculiar weakness of human nature, pure faith can never be relied on as much as it deserves, that is, enough to found a Church on it alone” (6:103) and thus “because of the natural need of all human beings to demand for even the highest concepts and grounds of reason something that the senses can hold on to . . . some historical ecclesiastical faith or other, usually already at hand, must be used” (6:109). Nonetheless, the goal of cosmopolitan ethical community is to increasingly purge faiths of their impure elements, or at least to marginalize those contingent features as merely vehicles for universal religion.
Robert Adams (Adams 1998: xxxi) rightly leaves “to the reader” the issue of “whether Kant believes that an ethical community that would dispense with all commitment to historically conditioned doctrines and practices is a real historical possibility or . . . an ideal to be approximated.” For my purposes, it is unnecessary to resolve this issue, since on either reading, it is the ideal that should govern the pursuit of cosmopolitan ethics (that is what it means, after all, for it to be an “ideal to be approximated”).
 Romans 3:23-4. Although Kant does not explicitly quote this passage in the relevant context, the doctrine of divine mercy expressed in it underlies his whole account.
 Kant even interprets this son of God to be one’s own self, in a sense. For details, cf. Quinn 1986.
 Kant is not, of course, primarily interested in outlining membership requirements for being a part of ethical community. Anyone sincerely committed to the promotion of virtue can be a part of such community, and even those not seriously committed can be a part in the sense that an ethical community should welcome them as a way of helping cultivate their commitment to virtue. Nonetheless, the question here is whether Kant thinks that one truly can be sincerely committed to promoting virtue without certain religious beliefs, that is, whether these beliefs should be made part of the program by which ethical communities seek to promote virtue. I thank an anonymous reviewer for Faith and Philosophy for forcing me to clarify this point.
 The fact that Kant says “possibility or actuality” here could mean that the actuality of the objects is no less contested than their possibility, or it could mean (as Wood may read it) that not contesting the possibility of the objects might be sufficient for practical life, even without any commitment to their actuality, but that a belief in their actuality is permitted as well.
 These passages and others have led Robert Adams to claim that “There is no place in the Kantian scheme of things for prevenient grace – that is, for divine assistance that precedes our first turning toward the good” (Adams 1998: xxi). For views that are more open to the possibility of prevenient grace, see Mariña 1997 and Frierson 2003.
 Immediately after this passage, Kant adds, “we are not thereby denying that . . . there might yet be something in the mysteries of the supreme wisdom which only God can do” (6:171). Thus Kant’s worry here is not about belief in divine mercy altogether, but only a concern about the practical effects that mistaken beliefs about divine mercy might have. Cf. 6:117-8.
 As Kant insists in his discussion of the postulates of practical reason, this “moral necessity to assume the existence of God” or of divine mercy should not be confused with a “duty to assume the existence of anything” (5:125). Working out the nature of the requirement to believe in divine mercy would involve distinguishing between a strict “duty to believe” and a requirement that is both rationally and psychologically connected to one’s duties. Fully working out this distinction is beyond the scope of the current paper, for which I need only the concept of a “moral necessity” to believe, rather than a duty in a strict sense.
 Kant describes this conflict clearly at 6:117, in the second prong of his “antinomy” of divine mercy.
 In Wood 1992: 403, Allen Wood makes a similar comparison, though to a different purpose.
 The nature of this rational necessity can be spelled out in more detail through further analogy with the postulates of practical reason. Kant’s discussion of the sort of assent that is warranted here is provided in the Critique of Practical Reason (5:110-48). I will not here arbitrate between the different ways of interpreting his claims there, although this religious interpretation of Kant’s cosmopolitanism depends at least upon not reading this assent as merely a matter of acting “as if” these claims are true.
 Even in the Critique of Practical Reason, some aspects of Kant’s discussion (e.g. 5:143) are more psychological. For the purpose of this paper, I merely want to highlight this psychological aspect of the postulates. It is not my intention to make any claim about shifts between the second and third Critiques.
 Kant explicitly discusses the proper way to interpret Scripture in both Religion within the Boundaries of Mere Reason and The Conflict of the Faculties. His own use of Scripture could be helpfully studied as a model of the general method he lays out in those discussions.
would like to thank the organizers of the Conference on “200 Years after Kant”
(November 2004) and the government of the Islamic Republic of Iran for inviting
me to participate in this conference and for making it possible for me to visit