1. Introduction: The puzzle of the cold sage, and previous commentary

Kant’s theory of friendship plays a crucial role in defending his ethics against the objection that it is too emotionally detached to capture the value of feelings and particular relationships. This objection often originates in response to a famous passage in the Groundwork on sympathy which distresses many readers, where Kant describes a philanthropist whose benevolent actions only have “genuine moral worth” when his “sympathy with the fate of others” has been “extinguished” by “grief,” and he acts “simply from duty” (G 4:398).Footnote ¹ Kant says that while sympathy is “amiable” and a disposition to be encouraged, it is “on the same footing with other inclinations,” and we can have a “far higher worth than what a mere good-natured temperament” confers “even if we are cold [kalt] and indifferent [gleichgültig] to the sufferings of others” (G 4:398). Kant does not suggest that we should avoid sympathy—his point is rather that in cases like that of the grieving philanthropist, “the worth of character comes out” (G 4:398–99). Nonetheless, his emphasis on the claim that dutiful yet cold and indifferent agents can have a far higher worth than sympathetic agents has prompted philosophers dating back to his contemporary Schiller to object that Kant’s moral theory is too emotionally detached to properly value sympathy.Footnote ²

However, other parts of Kant’s corpus provide material for a defense against this objection. Kant entitles the conclusion of the Doctrine of Virtue “On the most intimate union of love with respect in friendship.” He proceeds to tell us that

Remarks like this also appear in a number of Kant’s lectures. According to Collins’s Ethics notes, to “possess such a friend, of whom I know that his disposition is upright and kindly, neither malicious nor false … [t]his is the whole purpose of man, which allows him to enjoy his existence” (CE 27:427). On the basis of such passages, Christine Korsgaard writes that “[t]o become friends is to create a neighborhood where the Kingdom of Ends is real” (Reference Korsgaard1996, 194). We might also say that the way friendship correlates valuable dispositions of friends toward one another with their shared enjoyment of life makes friendship a fragment of the highest good which is accessible within experience. At least since Paton (Reference Paton1956), and increasingly in recent years,Footnote ³ commentators have directed proponents of the detachment objection to Kant’s theory of friendship, helping to diminish the objection’s influence.

However, the motif of cold indifference appears again in passages on friendship, in connection with the Kant’s depictions of the stoic sage or wise man. The sage is not a “finite holy [being] (who could never be tempted to violate duty),” but rather a human being who has “autocracy of practical reason,” that is, mastery of “one’s inclinations when they rebel against the law745.” The sage serves as “an ideal (to which one must continually approximate)” (MM 6:383). The earliest passage is in Herder’s Ethics notes (1762–64):

This passage does not explicitly address friendship, but Herder’s notes go on to address friendship immediately afterwards (HE 27:54), suggesting an endorsement of indifference toward friends we cannot help too.

The second passage is in Collins’ Ethics notes (1784):

Here too we have no direct reference to friendship, but once again a discussion of friendship immediately follows (CE 27:422). So here too Kant can seem to hold that if I cannot help a suffering friend, I need not experience any sympathetic pain.

The most famous remarks on this theme appear in the Doctrine of Virtue (1787):

This passage can also seem to assert the view that if I cannot help my suffering friend, I need not suffer in sympathy. Though I am in the company of readers who find this view distressing, I do not wish to claim that it is obviously false. Utilitarians, for example, could endorse it. However, if it is in fact the view we should attribute to Kant, then Kant’s theory of friendship offers only a partial defense against the detachment objection. This is likely to become increasingly damaging to the reception of Kant’s ethics as contemporary ethics makes the value of care more central.Footnote ⁴ I take it to be essential to our intuitive understanding of care that it hurts us when someone we care about suffers, and that we can only banish that suffering if we stop caring. If Kant’s advice is that we may stop caring when we cannot help a suffering friend, then the detachment objection may be insurmountable. This paper offers a new argument for thinking that this is not Kant’s advice.

Commentators have already made substantial progress in this endeavor. My approach builds in particular on a line of thought developed in the work of Marcia Baron and Lara Denis. Baron argues that Kant’s central theme in the passages about friendship and the sage is not to advocate absence of sympathetic feeling, but rather the ability to voluntarily regulate it, emphasizing Kant’s distinction between (a) sensitivity [Empfindsamkeit], which Kant describes as “a faculty and a power which either permits or prevents both the state of pleasure as well as displeasure from entering the mind, and … possesses choice,” and (b) sentimentality [Empfindelei], which Kant describes as a “weakness by which we can be affected, even against our will, by sympathy [Theilnehmung] for others’ condition” (AP 7: 235–36), a weakness Baron describes as “self-inflicted” (Baron Reference Baron1995, 196). She thinks Kant nonetheless attributes to the sage the ability to “turn off” sympathetic pain “if it does no good” (216), and that Kant approves of the sage’s exercise of it when she cannot help. Baron thinks that if Kant is to properly value care, then Kant’s ethics must be corrected so that it values sympathetic suffering even when we cannot help, and so that it advocates sensitive moderation of sympathy in such cases rather than cessation of sympathy (221).

Denis argues that fine-grained analysis of Kant’s texts shows that Kant already holds the views that Baron thinks Kant must adopt and so feels that no correction is required. She notes that Kant says the sage exemplifies a virtue he calls apathy [Apathie], such that the “wise man [Weise] must never be in a state of affect [Affect], not even in that of compassion [Mitleids] with the misfortune of his best friend” (Kant AP 7:253; Denis Reference Denis2000, 51–52). Apathie does not mean “lack of feeling [Fühllosigkeit]” but rather “absence of affects [Affectlosigkeit]” (Kant MM 6:408; Denis Reference Denis2000, 51–52). Affect is a technical term in Kant’s moral psychology which refers to feeling which we allow to develop without moderation by reason, which can be sudden, overwhelming, and agency-undermining. Affect can prevent us from helping effectively even when we discover means to do so (MM 6:407; AP 7:253; FA 25:589), and can dispose us to violations of the law when it conflicts with duty, as in Barbara Herman’s example of a sympathetic onlooker who helps a struggling thief move his loot (Herman Reference Barbara1993, 4–5; also see RR 6:30 and FA 25:611). Allowing ourselves to be afflicted by affect makes us sentimental rather than sensitive. Denis proceeds to show that Kant’s treatment of the concept of cold-bloodedness incorporates a similar distinction. Kant says that frigidity [Kaltsinnigkeit] is a “want of love” and a “lack of the feeling whereby the state of others affects us,” while “cold-bloodedness [Kaltblütigkeit] is a want of affect [Affekts] in love, and that “cold-bloodedness [Kaltblütigkeit] of love^t provides regularity and order” (Kant CE 27:420; Denis Reference Denis2000, 53). Denis argues that this shows that an apathetic, cold sage does have sympathy for suffering friends she cannot help—she simply ensures that it does not prompt affect (53–55).

I think Denis’s response to Baron is correct, and the account presented here is an extension of her response into the intentional teleology of the duty of beneficence, that is, the duty to take others’ ends as one’s own. Denis reads Kant as holding that it may be psychologically impossible for actual human beings to completely suppress sympathy when they cannot help once they have become sympathetically engaged with their friends in the way they have a duty to strive to be. She argues that this gives us grounds to worry that actual human beings who are able to entirely suppress sympathy for their purported friends have never really become sympathetically engaged with them in this way, and that they may never really have adopted their purported friends’ ends as their own (64–66). This paper extends this thought by arguing that sympathy is necessary for adopting others’ ends, and that it is necessary not only as a matter of empirical psychology, but also as a matter of practical rationality. When we properly individuate those ends of others which are valuable only because they have been set by rational agents (such as cessation of one’s pain), ends which I call merely permissible ends (MPEs), we see that we can only adopt them through sympathy. This implies that sages as well as ordinary human beings must sympathize to adopt others’ MPEs. This in turn implies that ceasing to sympathize with a friend entails ceasing to adopt the friend’s ends, and thereby forsaking the friend, which we ought not do. The claim that sympathy is practically necessary for adopting others’ ends may seem to imply that the will must be impure and heteronomous because it may be thought that even if apathy allows us to moderate sympathy, sympathy cannot arise voluntarily in the way that respect for law can.Footnote ⁵ However, Kant has a detailed distinction between rational sympathy and natural sympathy which shows this concern to be misplaced.

2. The distinction between rational and natural sympathy

This section explains the distinction between rational and natural sympathy, which is central to this paper’s account of friendship and the sage.Footnote ⁶ Kant’s most famous comments on sympathy are the Groundwork remarks discussed above, where Kant claims that people motivated by respect for law but “cold [kalt] and indifferent [gleichgültig] to the sufferings of others” can have a far higher worth than sympathetic people (G 4:398). The nuanced distinctions Denis highlights imply that this claim requires careful interpretation. To be cold is to be apathetic rather than indifferent—to love in an orderly way, not with agency-undermining affect. Denis does not discuss HE 27:54 (quoted earlier), where Kant claims that Gleichgültigkeit can be a “moral quality” and a “good trait,” but her interpretation encourages attention to how it functions when it serves as a good trait: it gives the love inspired by sympathy the “right degree^t [rechten Grad]” (HE 27:54). Thus we can suppose we have a capacity to be indifferent (perhaps fully possessed only by sages) which allows us to dampen sympathy to zero, but its moral employment is to regulate sympathy rather than to eliminate it. This capacity to regulate sympathy through dampening is an aspect of a broader capacity which I call rational sympathy, which also includes the capacity to voluntarily prompt sympathy.

The Doctrine of Virtue’s account of sympathy supports the view that we have this capacity. It appears both immediately before and immediately after the passages on the sage quoted above. In it, Kant writes that “Sympathetic Feeling Is Generally a Duty [Theilnehmende Empfindung ist uberhaupt Pflicht]” (MM 6:456), and goes on to specify that “it is a duty [Pflicht] to actively sympathize [thätige Theilnehmung] in [others’] fate” (MM 6:457). If it is a duty to sympathize, then sympathy cannot be merely “on the same footing with other inclinations,” as we saw Kant claim in the Groundwork (G 4:398). It must be a capacity which we can voluntarily exercise, guided by practical reason.

The contrast between the Groundwork and the Metaphysics of Morals can seem so stark that commentators sometimes argue that Mary Gregor’s translation of the Metaphysics of Morals (the translation used here) is wrong. Some have pointed out that Theilnehmung can be translated as “participation” as well as “sympathy,” and have argued that this means that what the duty of theilnehmende Empfindung demands of us does not involve sympathetic feelings.Footnote ⁷ But there is decisive support for Gregor’s translation in the Friedländer Anthropology lecture notes, where Kant asks us to “consider sympathetic feeling [sympathetische Gefühl],” and says that

If we couple this passage with Kant’s remark that “Empfindung is also called [genannt] Gefühl” (CA 25:51), then we have Kant’s own resolution of this controversy: he tells us that theilnehmende Empfindung means the same as sympathetische Gefühl. This should make us confident that the Metaphysics of Morals tells us we have a duty to sympathize which involves sympathetic feeling.

If we look at Kant’s corpus more broadly, it becomes clear that the Groundwork elides a distinction between two ways of sympathizing which can involve qualitatively identical sympathetic joys and pains but are differently oriented to practical reason: one is a voluntary sympathy guided by practical reason, which I refer to as “rational sympathy,” and the other is a sympathy of self-inflicted passivity, which I refer to as “natural sympathy.”Footnote ⁹ Kant draws this distinction in five places over a period of twenty years. I label terms for rational sympathy “(a)” and terms for natural sympathy “(b).” MM 6:456 distinguishes (a) “humanitas practica,” the “capacity and the will to share in others’ feelings [Gefühle mitzuteilen],” which is “free,” and based on “practical reason,” and (b) “humanitas aesthetica”; “the receptivity, given by nature itself, to the feeling of joy and sadness in common with others,” which “can be called communicable … like receptivity to warmth or infectious^t [ansteckender] diseases … since it spreads naturally.” FA 25:607–11 distinguishes (a) “reason’s sympathy [Antheil der Vernunft]” and (b) “physical sympathy [physicalischen Sympathie]” based on “animality.” AP 7:235 distinguishes (a) “sensitivity” [Empfindsamkeit] and (b) “sentimentality” [Empfindelei], as we saw Baron note earlier. The same distinction appears at 25:1320–21 in Mrongovius’s Anthropology notes (1784–85). Sensitivity “possesses choice” and “permits or prevents both the state of pleasure as well as displeasure from entering the mind” and gives us the “fine feeling^t [feines Gefühl]” necessary to “judge [others’] sensation [Empfindung],” while sentimentality “is a weakness by which we can be affected, even against our will, by sympathy [Theilnehmung] for others’ condition” (AP 7:235). VE 27:677–78 distinguishes (a) “moral” [moralisches] sympathy and (b) “instinctual” [instinctmäßiges] sympathy. Though these discussions differ in emphasis, they all support a distinction between (a) a sympathy which is voluntary and guided by reason, and (b) a passive sympathy which disposes us to affect.

Kant understands sympathy as an activity of the imagination (MM 6:321n, 6:457; AP 7:179, 7:238), and thus the difference between rational and natural sympathy is a difference between ways this activity can proceed: a voluntary, reason-guided way versus a passive way. Imagination is a fundamental power in Kant’s theory of mind, one of two “parts” of sensibility, the other of which is “sense” (AP 7:153). Kant thinks we have significant control over the imagination, which we can choose to exercise or not. Mrongovius’s Anthropology lecture notes tell us that if we do not direct the imagination, “[t]he imagination directs itself according to the inclinations” (MA 25:1258, also see FA 25:515).Footnote ¹⁰ Kant says that the sympathetic imagination puts us “in the other’s place” (FA 25:575, also see MM 6:321n; HE 27:58, 27:65; FA 25:575, 25:607, for similar language).Footnote ¹¹ It is this imaginative activity which enables rational sympathy:

Putting ourselves in others’ places is necessary but not sufficient for prompting sympathetic feelings. Kant distinguishes two ways of putting ourselves in others’ places: (i) a “logical [logisch],” “heuristic [hevristisch]” way of adopting others’ perspective which helps us “arrive more easily at certain things^t” such as understanding the reasoning of other philosophers (he gives “a follower of Crusius” as an example), which does not involve sympathetic feelings, and (ii) a “moral [moralisch]” way of adopting another’s perspective which yields “true sympathy [wahrhaftigen Sympathie], where we really feel [fühlen] ourselves to be in his place” (HE 27:58). In natural sympathy, we passively adopt others’ perspectives, while in rational sympathy, we adopt others’ perspectives in a voluntary way regulated by the duty of sympathy.Footnote ¹²

The claim that we can voluntarily prompt sympathy conflicts with much of the commentary on Kant’s theory of feeling. Commentators often read Kant as holding that we can do things to cultivate sympathetic feeling over the long term, but not voluntarily prompt it in particular cases. Such commentators often emphasize Kant’s remark that we cannot will ourselves to have feelings of love: “Love is a matter of feeling, not of willing, and I cannot love because I will to” (MM 6:401).Footnote ¹³ The thought that “we have no direct control over our feelings,” as Korsgaard puts it (Reference Korsgaard1996, 182), seems accurate not only to Kant’s ethics but also to the everyday phenomenology of feeling. The view advanced in this paper may seem incompatible with this thought. But the view advanced here holds that the locus of volitional control in rational sympathy is in imagining, not in feeling. The voluntary actions which occur in rational sympathy are the mental actions involved in imaginatively stepping into the first-person perspective of the other and furnishing it with imagined sensible content.Footnote ¹⁴ Sympathetic feelings arise in response to these imaginings—we do not will the feelings to spontaneously arise. In this way, rational sympathy fits the description of moral feeling in Theory and Practice: it is “not the cause but the effect of the determination of the will” (TP 8:283). Further, Kant makes a connection between feeling and practice which provides a significant scope for volition in feeling. According to the Doctrine of Virtue, “Beneficence is a duty. If someone practices (ausübt) it often … he eventually comes actually to love the person he has helped” (MM 6:402). The Collins’s Ethics lectures put the point more concisely: “if I love others from obligation … by practice [Uebung] it becomes love from inclination (CE 27:419; also see MM 6:402). This connection makes it worth repeating Kant’s claim quoted above that “[t]o take a point of view is a skill [Geschicklichkeit] which one can acquire by practice [sich durch Uebung erwerben kann]” (FA 25:475). If we have the ability to acquire a skill through practice, then it is deontically consistent for reason to require us to acquire it, because there is a “can” to match the “ought.” When we have acquired a skill, we can do the thing we are skilled in doing with volitional control. Kant’s view is that rational sympathy is such a skill.Footnote ¹⁵

3. Friendship as an ideal of sharing all our ends and all our feelings

This section argues that the ideal of friendship includes maximal sharing of feelings and maximal sharing of ends, and then poses a question about the nature of the correlation between these maxima to set the stage for the argument that sympathy is necessary to take others’ merely permissible ends as our own. The distinction between natural and rational sympathy is not always relevant in this section and is thus referenced only where necessary.

It is a characteristic mark of Kantian ideals that they contain maxima (OP 21:30; also see FA 25:609, ML2 28:555, CE 27:247). Analysis of ideals in terms of the maxima they contain is therefore essential for understanding them. At MM 6:469, Kant says we must strive for friendship “as a maximum of good disposition [Gesinnung] toward each other” (MM 6:469). Vigilantius’s Ethics notes make clear that that this involves a maximum of shared feelings: Kant says that in the ideal of friendship, “each mutually sympathizes^t [teilnehmen] with every situation of the other, as if it were encountered by himself” (VE 27:677; my boldface). This maximum is probably not well-understood as a maximum of intensity of feeling. I should feel as if I encountered my friend’s situations myself, and even if I should have feelings of the same intensity as hers, she presumably rarely experiences feelings of maximal intensity. The maximum appears instead to be one of extension, since it extends to all my friend’s situations.

Kant emphasizes caution in implementing this aspect of the ideal in actual friendships. He thinks we are driven to friendship not only by duty but also by a need to “reveal [ourselves] to others,” but friends should aim at “complete confidence … in revealing their secret judgments and feelings to each other” only insofar as such revelations are consistent with mutual respect and prudence (MM 6:469–71). Friendships in the actual world are fragile because of asymmetries of attachment: if my friend concludes that I love her more than she loves me, she may lose respect for me, so I must constrain my sympathetic attachment. Further, if I share feelings that reflect badly upon me—for example, an unreasonable resentment for a third party—then I may not only lose her respect, I may also place myself in jeopardy if our friendship ends and she uses my feelings against me.Footnote ¹⁶ In this vein, Kant comments that “[e]ven to our best friend, we must not discover ourselves as we naturally are and know ourselves to be, for that would be a nasty business” (CE 27:427). But it is clear that by “best friend” here he means the best friend he thinks we can hope to have in the world as it is, not the best friend of the ideal. Both my need and my duty compel me to search for friends with whom I can closely embody the ideal: “pure sincerity in friendship can be no less required of everyone even if up to now there may never have been a sincere friend” (G 4:408). Kant may be too skeptical in his dim view of the prospects for actual friendships that closely conform to the ideal, but even if he is correct, this view is not in conflict with the claim that the ideal of friendship is an ideal of sharing all feelings.

We should also read Kant as holding that the ideal of friendship includes a maximum of adopting all one’s friend’s (permissible) ends. We have a universal imperfect duty to take others’ ends as our own (G 4:430; MM 6:450), but Kant acknowledges that “one human being is closer to me than another” (MM 6:451), and that our friends are our closest others, and this makes it intuitive to think that we have a special duty to adopt our friends’ ends.Footnote ¹⁷ It is natural to think that striving for a “maximum of good disposition toward each other”(MM 6:470) requires us to adopt more of our friends’ ends as we become closer, striving toward adoption of all their permissible ends. Kant says this in fairly clear detail with respect to friends’ needs [Bedürfnisse]. In Collins’s Ethics notes, Kant says that “[t]he friendship of need is that whereby the participants may entrust each other with a reciprocal concern in regard to their needs in life,” and that “we must presuppose” this kind of friendship “in every friendship.”Footnote ¹⁸ He says that “I must … have confidence in each of my true friends, that he would be able and willing to look after my affairs, and promote my interests”—“my friend is … ready to aid me in any difficulty” (CE 27:424–26; my boldface). According to Vigilantius’s Ethics notes, “[f]riends … undertake to support one another in their needs with all their powers and means” (VE 27:684; my boldface). It might seem that our needs are only a subset of our ends, while happiness includes all our ends, but in fact Kant understands happiness as a need (2C 5:110; also see MM 6:393). So these passages provide strong support for the thesis that we ought to adopt all our friend’s permissible ends as our own. Here too the sort of maximum we find is one of extension.

Kant sees this aspect of the ideal of friendship as constrained by actuality for reasons like those for constraining sympathy. Power and wealth are often asymmetrical between friends, so accepting “a favor from the other” can lead favor-givers to take on the status of benefactors, and can lead beneficiaries to lose respect in the eyes of the benefactors and in the eyes of the beneficiaries themselves (MM 6:471). Even in this discussion, however, Kant acknowledges “the help that each may count on from the other in case of need” (MM 6:471). But actual constraints on maximal end-taking do not conflict with the claim that the ideal includes maximal end-taking. Thus, the most plausible view to attribute to Kant is as follows. I must strive to adopt all my friend’s ends, but it is rational for me to prioritize our shared end of preserving our friendship since this my friend’s only end in which I am essentially involved. If my friend can achieve her other ends without my help, and my help would threaten our friendship, I do not help her achieve her other ends. On the other hand, if she cannot achieve her other ends without my help and those ends are more important to her than our friendship, then I help, despite the threat. Kant’s point about the hazards of favors is thereby made compatible with his view that we ought to adopt all our friends’ permissible ends as our own.Footnote ¹⁹

When we see correlated maxima, as we do between sharing ends and sharing feelings in the ideal of friendship, it is helpful to inquire about the nature of the correlation. Is Kant making a normative claim that we should share more feelings as we share more ends, or claiming that, as a matter of fact, we actually do? If the latter, is the correlation (1): a correlation with exceptions, such that we will sometimes find cases of sharing ends without sharing feelings?Footnote ²⁰ Or is the correlation (2): the consequence of laws of empirical psychology which govern all actual human beings and make it psychologically impossible for actual people to adopt others’ ends without sharing their feelings?Footnote ²¹ Or could the correlation be (3): a conceptual necessity of Kant’s practical philosophy, so that looking for people who adopt others’ ends without sympathizing turns out to be a complicated analogue of looking for married bachelors? If (1), then actual human beings could adopt suffering friends’ ends without suffering in sympathy. If (2), then perhaps we can imagine that a sage, as an unactualized ideal with perfect rational control of her feelings, could accomplish this feat, even though actual human beings cannot. If (3), then neither actual human beings nor sages can adopt suffering friends’ ends without suffering in sympathy. It will be argued that (3) is the right way to understand the correlation, at least with respect to those ends of others which are valuable only because they have been set by rational agents, that is, their merely permissible ends (MPEs).

4. Others’ ends and rational sympathy

This section argues that rational sympathy is necessary to voluntarily adopt others’ MPEs as our own.Footnote ²² Though Kant does not discuss this point in detail, one’s ends are individuated in terms of one’s own concepts of one’s ends. Kant implicitly acknowledges this when he states that “an end is the object of a concept insofar as the latter is regarded as the cause of the former” (3C 5:220). Concepts individuating ends cannot plausibly be supposed to belong to anyone other than the agents with the ends. This implies that others’ ends are individuated by their own concepts of their ends. If there were no motivationally significant variances among agents in their concepts of their ends, then this point might be trivial. But there are motivationally significant variances, and they create two puzzles we must resolve to understand how we might take others’ ends as our own, one having to do with marks of the first person in their concepts of their ends, and another having to do with marks of law in those concepts.

First, consider marks of the first person. Our concepts of our ends often (perhaps always) include a first-person indexical. For example, my end of repaying a debt is typically individuated in terms of the concept repaying my debt. There is no motivational puzzle about having an end which is individuated by a concept containing a first-person indexical when the end is unproblematically my own and the first-person indexical is unproblematically indexed to me. But my friend is likely to have the end of stopping her pain in terms of the concept stopping my pain. Since her end is individuated by her concept, I can only truly adopt her end if I do so in terms of her concept. But in doing so, I must keep the indexical indexed to her or I form a desire to stop my pain rather than hers, and I fail in my goal of adopting her end. I can succeed if I step into an imaginary version of her first-person perspective. It is not clear how else I might succeed. As we have seen, rational sympathy gives us the capacity to voluntarily step into such a perspective.

It may be objected that the inference to a need for rational sympathy is too quick—as we have seen, Kant thinks we can step into others’ perspectives in a merely logical, heuristic way which does not involve sympathetic feeling, and it might seem that such unsympathetic perspective-taking would suffice for acquiring others’ concepts of their ends, and using those concepts to set their ends as our own. The reason that unsympathetic perspective-taking does not suffice has to do with the second motivational puzzle, which has to do with marks of law included in concepts which individuate ends. Consider again the concept of repaying my debt. If I am practically rational, this concept includes a mark of obligation since this is something I have a perfect duty to do. If I am motivated by respect for law, there is no puzzle about how I can have an end individuated by this concept—this concept engages with my motivation because of feelings made up of pleasures and pains which I necessarily have as a rational agent (G 4:460–61; 2C 5:9n; MM 6:211–12). Next consider the concept of caffeinating myself. If this is a concept of an end for me, then insofar as I am a rational agent, the only mark of law it can contain for me is permissibility because caffeination is a constituent of my happiness which is contingent from the perspective of rationality. Such concepts individuate ends for us because they engage our inclinations (G 4:427). That is, they engage those aspects of our feelings made up of pleasures and pains which are contingent from the perspective of rationality (2C 5:21–26), though we must judge them permissible before incorporating them into our wills. Such ends are merely permissible ends (MPEs).

There is no puzzle about how I can be motivated by my MPEs. My concept of caffeinating myself contains no marks of law apart from permissibility, and I have innumerable permissible things I can do, so permissibility alone cannot give me a reason to caffeinate myself. But I experience the pleasures of being caffeinated and the pains of not being caffeinated despite their contingency, and it is because of them (and their incorporation) that caffeinating myself is an end for me.

However, the ends of others which I have an imperfect duty to adopt are often MPEs, and there is a puzzle about how their concepts of their MPEs can engage with my pleasures and pains so that their MPEs become ends for me. My friend’s end of stopping my pain is an MPE. She is vulnerable to pain because she has a sensible nature which is contingent from the perspective of moral law, and so her pain is not in itself morally bad (2C 5:60). If her pain is intense, then this end will be weighty and constant in her mind but, for all that, it is an MPE. There is no puzzle about how she can be motivated by it—the fact that it is her pain, and that she takes pleasure in the thought of its cessation, makes it an end for her.

The puzzle is about how I can adopt her MPE. If I am to adopt her end as my own, I must do so in terms of her concept. The fact that it has no mark of law apart from permissibility, and that I have innumerable permissible things I can do, means that permissibility cannot be a sufficient reason for me to set an end in terms of this concept—and crucially, it means that I cannot intelligibly set an end in terms of this concept out of respect for law since there is (as it were) simply not enough law in the concept for that to be possible. Since I have an imperfect duty to adopt her end, it is meritorious for me to adopt her end, and so we might suppose that I can add a mark of merit to her concept in order to set her end for myself out of respect for law. But this is not possible. Her end is individuated by her concept, and since it is an MPE, she cannot rationally include a mark of merit in it. This means that a concept which includes a mark of merit would be a different concept, and would therefore individuate a different end. Footnote ²³ Since I cannot set her end as my own out of respect for law, I must look to the only other motivational source available in Kantian moral psychology—that is, my contingent pains and pleasures. I must have the capacity to voluntarily prompt pains and pleasures in myself which correspond to hers so that her concept engages with my feelings in the way necessary for it to individuate an end for me. To do this, I must voluntarily step into an imagined version of her perspective in a way that is rationally sympathetic, not merely logical and heuristic. It must be emphasized that this voluntary step into her perspective is motivated by respect for law, so respect for law is a necessary part of rational sympathy, but the contingent feelings which arise in response to my voluntary imaginings are necessary too.

While Kant thinks we can know a priori that feelings about the moral law, beauty, and sublimity are universally communicable, he thinks we cannot know this about our other feelings (3C 5:213, 24), including the feelings that motivate us to pursue our MPEs. But the passages discussed above show that he thinks these feelings are in fact communicable on a frequent basis.Footnote ²⁴ As Alix Cohen notes, Kant thinks feelings which are not about morality, beauty, or sublimity can be comparatively universal (Cohen Reference Cohen2008, 316; AP 7:242; 3C 5:213). The fact that we can share feelings about MPEs with only comparative universality is compatible with the claim that sympathy is necessary to adopt others’ MPEs. The duty to adopt others’ ends is imperfect, so it accommodates imperfections in our abilities, including imperfections in our abilities to share the feelings required for its fulfillment. We do enough when we make a sincere effort to sympathize.

It may be objected that Kant says the duty of beneficence requires us to promote [befördern] the ends of others (G 4:430; also see 2C 5:34, MM 6:453, VE 27:544), and we can do things to promote others’ ends without sympathizing with them.Footnote ²⁵ This point is important, but not in conflict with the account presented here. Promoting and adopting others’ ends are distinct activities, and while the duty of beneficence makes both activities valuable, adoption requires sympathy and is necessary to truly make others’ ends our own. I may do things to alleviate my friend’s pain because (I) I adopt her end through rational sympathy, (II) I want to sleep soundly and her groans wake me, or (III) it is a way of fulfilling my duty of beneficence. I promote her end in all these ways. There may be no differences in the consequences I produce in cases (I), (II), and (III). But in cases (II) and (III), I promote her end as a means to different ends, neither of which are her end. Case (II) is different from case (III) since in case (III) I promote her end as a means to a distinct but nonetheless morally valuable end. We can recognize the activity in case (III) as morally valuable but still hold that there is an independent value in adopting her end, and thereby making her end my own. Kant uses this very language at MM 6:450, where he says the duty of beneficence can be expressed as the duty to make others’ ends my own [Anderer ihre Zwecke zu den meinen zu machen]. At G 4:430, Kant he writes that

Commentators have puzzled over what Kant means by “as far as possible” in this passage, in light of the latitude granted by the imperfect duty of beneficence (MM 6:393), and the interpretation offered here provides an explanation. That is, adopting another’s end as my own through rational sympathy, rather than merely promoting it in the service of some other end, is a plausible way to understand making it my own as far as possible, as well as ensuring “its full effect in me.”

Why does it matter whether we adopt others’ ends or merely promote them when both may yield the same consequences? The answer is that Kantians should attend to the quality of our wills when we adopt others’ ends, not just the consequences of our willing. This is a place where the value of care emerges in Kantian ethics, and though Kant does not emphasize this point, contemporary defenders of Kant should. Adopting others’ ends creates an immediate and intimate connection not only between our wills, but also between our perspectives and our feelings, and it makes good moral sense to embrace the idea that all this flows from the aspect of the duty to treat others as ends which is the duty to take their ends as our own.

The upshot of this section for the question of suffering in sympathy with friends we cannot help is as follows. Since sympathy is a necessary condition of adopting others’ MPEs, we can only cease suffering in sympathy with our friends if we cease to adopt some of the ends they care about most, and we thereby forsake our friends in an important sense. Though we do not withhold help since there is no help we could provide, we nonetheless withhold care. We might say that we forsake our friends in our hearts. Since we ought not forsake our friends, we ought to suffer in sympathy even when we cannot help. This holds not only for ordinary people but also for sages. Sages are different from ordinary people in that they are able to exercise perfect rational control over their feelings. But as idealized human beings, they too must sympathetically step into others’ perspectives to adopt others’ MPEs.Footnote ²⁶ It would seem that even God cannot adopt others’ MPEs without sympathy, so if God lacks sympathy (as 2C 5:34 suggests), then only beings like us can adopt others’ MPEs. This casts light on a special role we play in the moral world.

5. Responses to objections

This section responds to three objections to the interpretation presented above. First, in Friedländer’s Anthropology lecture notes, Kant can appear to suggest that we should not sympathize with suffering friends we cannot help because it makes things more difficult for them: “[I]f you cannot help him at all … then go away unperturbed. Weeping, mourning, lamenting … make[s] the other’s misfortune more acute and unbearable for him” (FA 25:612).Footnote ²⁷ It is clear that someone rationally sympathizing would not experience or express her sympathetic pain in the unmoderated ways depicted in this remark. The interpretation presented here thus makes it reasonable to suppose that Kant’s concern in this passage is just that affect-driven excesses of natural sympathy can distress the suffering friend. Admittedly, even moderated expressions of rational sympathy could probably be distressing sometimes. In some cases, this problem could be resolved by sympathizing without communicating one’s sympathy.Footnote ²⁸ But suppose my friend wishes to communicate to confirm that I am not sympathizing. She may say that she will rest easier if she knows I have stopped thinking of her pain with sorrow—that she cares more about my living a joyful life than she does about my sympathy. This now appears to be a case where there is something I can do to help—I can help her pursue one of her ends, that I live a joyous life—but according to the interpretation defended here, I can only do this by relinquishing her end of stopping her pain, in the sense of ceasing to adopt it (though I should still search for ways to promote it). In such a case, the best I can do is to adopt some but not all of her ends. The world constrains actualization of ideals.

The second objection is as follows. As noted near the outset, Kant says that “when another suffers and, although I cannot help him, I let myself be infected [anstecken lasse] by his pain, then two of us suffer, though the trouble really (in nature) affects only one,” and proceeds to say that “there cannot possibly be a duty to increase the ills in the world” (MM 6:457; also see a similar argument at MA 25:1321). He can appear to be invoking the consequentialist argument that there is no moral reason to experience pain unless that pain is outweighed by its diminution of some other pain. But Kant is typically understood to be an opponent of consequentialism, and while this sometimes leads to oversimplifications, we should be cautious about attributing consequentialist arguments to him. So it is important to carefully consider the particular terms Kant uses to explain what I do not have a duty to do in this case. He says I have no duty to let his pain infect [anstecken lasse] me. As mentioned above, Kant uses an adjectival form of anstecken earlier on the very same page of the Doctrine of Virtue to characterize what he calls “humanitas aesthetica,” which was shown to be a term for natural sympathy rather than rational sympathy: he explains the sense in which humanitas aesthetica is “communicable” by saying that it is “like receptivity to … infectious^t diseases [ansteckender Krankheiten]” (MM 6:457). Ansteckung and its derivatives appear rarely in Kant’s corpus, so this connection between passages in close proximity means we should read Kant as claiming that we have no duty to sympathize naturally with a suffering friend who we cannot help, but not claiming that we have no duty to sympathize rationally.

The third objection is based on a passage in Collins’s Ethics notes which appears in close proximity to the story of the sage, part of which we saw earlier, in which Kant can seem to be criticizing sympathy for friends we cannot help because it is a mere wish:

The first thing to note in interpreting this passage is Kant’s specification in the first sentence that he is criticizing “sehnliche” (“ardent”)Footnote ²⁹ wishes. The word sehnliche seems to appear exclusively as a modifier for Wünsche (“wishes”) in Kant’s corpus, in three different versions of students’ lecture notes from Kant’s Ethics classes (the other instances appear at ME 27:1543 and BE 253). “Ardent” suggests that Kant is specifically criticizing the affect-driven wishes which occur in natural sympathy. Additional support for this reading comes from a point we saw earlier in Herder’s Ethics notes: that Gleichgültigkeit is a “good trait” only insofar as it “holds sympathy in check and gives it the right degree^t” (HE 27:54). In light of the interpretation advanced here, we can say that Gleichgültigkeit is a good trait only insofar as we draw on it to experience rational sympathy rather than natural sympathy. As we also saw earlier, Herder’s Ethics notes identify the good sort of Gleichgültigkeit with Kaltblütigkeit (“coldbloodedness”). We should also recall the Collins’s Ethics distinction between Kaltsinnigkeit (“frigidity”) and Kaltblütigkeit, which defines the former as a “want of love” and the latter as a “want of Affekts in love” which “provides regularity and order” in love. That distinction appears at CE 27:420, in the paragraph immediately prior to the CE 27:421-2 critique of mere wishing currently under discussion. Together these points provide evidence that the Gleichgültigkeit Kant recommends at CE 27:422 as an alternative to the sympathy of mere wish is not a state devoid of sympathetic feeling, but rather a kaltblütig state of regular, orderly sympathy in the right degree. We should hold that he is criticizing natural sympathy as mere wish, but endorsing something like wishing which involves rational sympathy.

It may seem puzzling to claim that there is room for a distinction between natural and rational sympathy with respect to the concept of wish in cases where we cannot help, such that the former is mere wish which ought to be eradicated and the latter is a different and valuable mental activity. That is, in such cases, we cannot help, so it is common to both natural and rational sympathy that they cannot prompt useful action, and so they are both like wishing in an obvious way. Can Kantian ethics accommodate a distinction between (i) mere wish, which is morally empty and should be eradicated, (ii) a kind of conative state which is like wishing in that it can produce no useful action but is nonetheless morally valuable, and (iii) the kind of conative state that produces useful action?

The answer is yes; Kant quite unequivocally relies on such a threefold distinction in the famous “useless jewel” passage, in which he argues that goodness of will does not entail good actions:

This pivotal passage entails a threefold distinction between (i) mere wish, (ii) a kind of conative state which “achieve[s] nothing” but is nonetheless morally valuable because it is an activity of practical reason, and involves the “summoning all of all means insofar as they are in our control,” and (iii) a kind of conative state which is based on practical rationality and leads to useful action. Kant implies that both (ii) and (iii) are sources of moral worth, though (i) is not. Kant’s assertion that there is moral value in (ii)-type cases is fundamental for understanding the sense in which his ethics is nonconsequentialist. That is, to recognize the value in (ii)-type cases is to recognize that the value of the activity of good willing is independent of the value of the states of affairs it sometimes brings into being. We can draw a distinction between (i)- and (ii)-type cases of sympathizing when one cannot help which corresponds to the distinction between rational and natural sympathy. When, in rational sympathy, we wrack our brains to try to find ways to help, sifting through our knowledge of empirical laws to find relevant hypothetical imperatives we might follow, we are clearly summoning means. When we voluntarily and sympathetically put ourselves in the other’s place to understand the help they need, we are summoning means here too, and the text supports this claim. In section 34 of the Doctrine of Virtue, Kant tells us that “humanitas practica,” which was shown above to be rational sympathy, involves using our “receptivity” to “[s]ympathetic joy and sadness … as a means to promoting active and rational benevolence” (MM 6:456), and repeats the point in section 35: we should “cultivate the compassionate … feelings” and “make use of them as so many means to sympathy based on moral principles” (MM 6:457; my boldface). Rational sympathy is not the passive feeling-state which Kant criticizes as mere wish; it is action of the mind and will which has as much claim to being a (ii)-type case as anything in Kant’s ethics, and thus has the Kantian credential we need to recognize it as an aspect of good will.