Background

Kant's argument for FH plays a prominent role in the larger project he undertakes in his Groundwork for the Metaphysics of Morals. His stated aims there are to ‘seek out’ the supreme principle of morality, by arguing in the first two chapters that the various formulations of the categorical imperative are implicit in ordinary moral notions of good will and duty, and to ‘establish’ that principle, by showing in the third chapter that our common belief in morality is not illusory (G 4: 392). Kant's striking claim in section II is that what it is to be subject to duty is to be under a categorical imperative, which he defines as an unconditional, comprehensive and necessary requirement of reason addressed to beings who can, but might not, follow it (G 4: 413–14, 419–21). His initial statement of the supreme moral principle in that chapter is the formula of universal law (FUL): ‘Act only on that maxim by which you can at the same time will that it should become a universal law’ (G 4: 421). He goes on to argue that if there is to be a categorical imperative, there must be an ‘objective end’ or ‘end in itself’ that provides good and sufficient reasons for each and every rational being to follow it (G 4: 427–8). In two quick arguments, Kant contends that rational agents are just such ends in themselves, which leads him to conclude that, if there is a categorical imperative, then it can be expressed as FH: ‘Act in such a way that you treat humanity, whether in your person or in any other person, always at the same time as an end, never merely as a means’ (G 4: 429).Footnote ³

Before we examine these arguments for FH, I note a few basic points. First, there is widespread disagreement about what our humanity or rational nature consists in – some think it is our capacity to set ends, others our autonomy, still others our capacity for moral deliberation and action.Footnote ⁴ Second, Kant's commentators diverge over whether FH is essentially the same as the FUL – some think FH is equivalent to FUL by prohibiting maxims that others could not possibly share while others contend that FH requires us to respect, honour and cherish a special dignity or status had by all rational agents.Footnote ⁵ Third, the crucial clause in the humanity formula is the one enjoining us to treat humanity in ourselves and others as an end in itself, which includes, and is more comprehensive than, the other clause forbidding us from treating humanity as a mere means. Finally, there are long-standing puzzles about how to apply FH to real-world moral problems without generating moral dilemmas.Footnote ⁶

Kant's first argument that rational nature in persons is an end in itself is evidently an argument by elimination (G 4: 428). The objects of our inclinations, according to Kant, do not always provide overriding reasons for us to act so they do not have the unconditional and objective worth that would be necessary to ground the categorical imperative, which commands us to act whether we are inclined to or not. Similarly, Kant claims that our inclinations themselves are not objectively valuable in the way that would be needed to ground the categorical imperative. Non-rational ‘things’ are not ends in themselves either, according to Kant, because they do not give us unconditional and overriding reasons to act. The only remaining candidate is apparently rational nature in persons, so Kant concludes that humanity in ourselves and others is an end in itself.

His second argument for FH, which has received considerably more attention, is in some ways analogous to John Stuart Mill's ‘proof’ of the principle of utility, so I will refer to it as Kant's proof of the humanity formulation of the categorical imperative, or ‘Kant's proof’ for short. Mill argues that, because we each desire our own happiness, our happiness is good to each of us, so the general happiness is good to the aggregate of everyone.Footnote ⁷ Similarly, Kant contends that because we each value our own rational nature on the same grounds as everyone else values theirs, we must value the rational nature of all people, so our humanity is an end in itself that must always be treated as such (G 4: 429). Let's look more closely at this argument.

The Premise

Most of the discussion of Kant's proof has been focused on:

Why does Kant think that we all necessarily regard ourselves in this way, and for the same reasons? Various interpretations of his premise have been proposed, but each has its costs.

(P₁) One possibility is that we see our rational nature as an end in itself because our non-rational inclinations lead us to do so.Footnote ⁸ We may derive great pleasure and happiness from seeing our rational nature as an objective end, and thinking of our rational nature merely as a thing or tool may cause us great pain and unhappiness, so it is ‘necessary’ (G 4: 429) that we see ourselves as objective ends in the sense that our empirical psychological makeup inevitably moves us to do so. This would also be a ‘subjective’ (G 4: 429) principle in the sense that it is a principle that is acted on by a person and based on inclinations. As an empirical generalization, however, this claim is doubtful and anyway it cannot play the requisite role in Kant's proof, as we will see, so we have some reason to look for other interpretations of the premise.

(P₂) A second way of understanding the premise is that rational agents necessarily conceive of their own rational nature as an end in itself on the grounds that rational nature per se is an end in itself (Sayre-McCord Reference Sayre-McCord2001; Wood Reference Wood1999, Reference Wood2008). If rational nature is an objective, intrinsic and agent-neutral value – one that everyone has reason to respect – then perhaps it is part of being a rational agent that one see oneself as possessing that value. The point of Kant's proof, on this view, is not to derive the value of humanity from the way we regard ourselves; rather, the argument is meant to convince people of what is already true, namely that rational nature has objective worth. That is, we can distinguish between interpretations of Kant's proof that purport to reveal an epistemic priority of the premise over the conclusion and those that regard our rational self-conception as taking normative priority over how we ought to conceive of others. We may come to learn that the rational nature of others is an end in itself on the basis of reflecting on how we see ourselves, but proponents of P₂ claim that we are morally justified in valuing our rational nature as an end in itself because rational nature in general is an end in itself. This interpretation of Kant's proof may even fit his inferential vocabulary if we interpret those terms loosely, as part of an epistemic argument that is meant to show conscientious people that ‘the most reasonable way of understanding what we are doing and thinking’ is that we and others have an objective value (Wood Reference Wood2008: 90). Yet it may seem to some that Kant is using ‘therefore’ and ‘hence’ because he thinks there is a normative priority of the premise about how we conceive of ourselves over the conclusion about the value we must attribute to others, which fits with his suggestive theme that duties to oneself are the foundation of duties to others. A further difficulty with any reading of Kant that assumes an antecedent intrinsic value is Kant's well-known claims that the various statements of the categorical imperative, including FH, are not justified on the basis of objective values – he says, for example: ‘the concept of good and evil must not be determined before the moral law (for which, as it would seem, this concept would have to be made the basis) but only (as was done here) after it and by means of it’ (CPrR 5: 63).Footnote ⁹

(P₃) The most influential account of why Kant thinks rational agents necessarily view themselves as objective ends is the clever and vexing ‘regress argument’ of Christine Korsgaard and Allen Wood.Footnote ¹⁰ They interpret Kant as follows: When we set ends, which we must do in order for there to be rational action and a categorical imperative at all, we take those ends to be objectively good in the sense that all rational agents have reasons to value our own ends just because we set them. It is not just that we regard our ends as good for us, we see them as ‘the object of every rational will’ (Korsgaard Reference Korsgaard1996a: 120). We necessarily regard our ends as objectively valuable and take the source of their value to be our endorsement of them, so it follows that we must regard our capacity to set ends (i.e. our rational nature) as unconditionally and objectively good, that is, as an end in itself.

A lingering problem with this argument is whether we really do attribute objective value to the non-rational ends we set or whether instead we see our attaining our ends as good for us without yet presupposing that we automatically generate reasons for others to respond favourably to our personal plans and projects – having a project of finishing Robert Caro's multi-part biography of Lyndon Johnson may generate certain reasons for me to do so but if we do not add an additional moral principle of, say, beneficence then it is difficult to see why all rational agents have reason to help me get to reading or see my finishing the books as good tout court as opposed to good for me.Footnote ¹¹ A further issue is why we must regard our capacity to set ends as objectively and unconditionally valuable even if it is the source and condition of our objectively valuable ends – Elvis may have the power to make things objectively good by touching them, but it does not seem to follow that this capacity is objectively good in itself even if its deliverances are good in virtue of his touch.Footnote ¹² Finally, Korsgaard and Wood disagree about whether the regress argument is supposed to establish that someone's rational nature is objectively and unconditionally valuable or whether setting ends merely commits a rational person to regarding or seeing her rational nature as having that sort of value, which it has whether she recognizes it or not. In other words, does setting ends make that capacity an end in itself or is it already an end in itself and rational reflection in the form of the regress argument reveals this antecedent value to us?Footnote ¹³ Wood's epistemic rendering of the regress argument makes Kant's premise a more sophisticated version of the previous interpretation (P₂) whereas Korsgaard intends the regress argument as a moral justification of that claim.

(P₄) A fourth way of understanding the premise is that it is simply part of our rational nature to regard our humanity as an end in itself because we have autonomy of the will (Rawls Reference Rawls2000: 196–9; Hill Reference Hill1992, Reference Hill2002a). In the footnote to his proof of FH, Kant claims that the reason why we take our rational nature to be an end in itself is given in Groundwork III. There he argues that rational agents by necessity acknowledge that they have autonomy of the will. A rational person ‘necessarily conceives his own existence’ (G 4: 429) in this way, as an end in itself, and he does so, according to this interpretation, on the same grounds as other rational agents, namely that he possesses autonomy of the will. If we ask why it is a necessary part of being a rational person that one conceive one's rational nature as an end in itself because one has autonomy of the will, we may find that Kant did not attempt to argue directly for the premise or explicitly explain why he thinks it is true, choosing instead to leave its justification to ordinary moral understandings about the nature of a rational and reasonable person and perhaps also to the fact of reason (CPrR 5: 151–2, 155–6). Kant's view may have been that, once we understand the premise in this way, no further justification for it is needed.Footnote ¹⁴

The Inference

Kant's inference from the premise to his conclusion has also been understood in a variety of ways:

How (if at all) might we explain the move from each of the four interpretations of the premise we just discussed to the conclusion that we are all committed to regarding rational nature in others as an end in itself?

(C₁) Even if we assume that all rational agents unavoidably conceive of themselves as ends in themselves because doing so contributes to their happiness, as P₁ claims, it does not logically follow that they must regard one another as ends in themselves. Just because we may desire the same thing, say happiness, this does not give us any reason to satisfy that desire, according to Kant, which means that, while conceiving of one another as ends in ourselves may contribute to general happiness, this alone could not make us ends in ourselves (G 4: 431). Moreover, the Conclusion is supposed to be a universal principle, which Kant thinks cannot be established on the basis of empirical generalizations (G 4: 431; Pogge Reference Pogge1998: 197; Hill Reference Hill2002a: 124). In order for Kant's proof to succeed, therefore, his premise must say that we are rationally required to conceive of ourselves as ends in ourselves, rather than that we do so merely on the basis of empirical inclinations.Footnote ¹⁵

(C₂) Suppose next that we assume that rational agents necessarily conceive of their own rational nature as an end in itself on the grounds that rational nature per se is an end in itself. If rational nature is an intrinsic, agent-neutral and objective value then we have reasons to regard it in ourselves and others as such. The role of Kant's ‘proof’ would then be to elucidate and reveal this objective value to rational persons who may be biased and distracted by the splendour of their own rational nature and need reminding that the rational nature of others has objective worth as well. I have already suggested, however, some reasons why interpreters of Kant may be reluctant to understand him as appealing to a value of this sort as a starting-point in his moral framework.

(C₃) We can consider the inference from the remaining two versions of the premise (P₃ and P₄) together because the same problem arises when we try to derive the conclusion from either of them. Suppose it is part of our rational nature to conceive of our own rational nature as an end in itself on the basis of our capacity to set ends (P₃) or on the basis of our autonomy of the will (P₄). Kant's proof could then be understood as addressed to any rational agent, pointing out to him that other rational agents have the same characteristics that, as he sees it, make his rational nature an end in itself. Because there is no relevant difference between his rational nature and those of others, he must therefore conceive of their rational nature as an end in itself as well. If this argument were successful, it would show that it is a necessary feature of rational agency to conceive of humanity in general as an end in itself, and it would do so without appealing to an already existing, agent-neutral value.

This argument is not air-tight, however, because of the possibility of a rational agent who conceives of his humanity as an end in itself because it is his own. When someone is pressed by this version of Kant's proof to explain why he does not value rational nature in others given that he values his own, we can imagine him consistently claiming that there is indeed a relevant difference between his humanity and that of others, which is that his rational nature is his, it belongs to him, he is in charge of it. For such a person, part of what makes him regard his rational nature as an end in itself is that it is his, so valuing his humanity on these grounds does not commit him to valuing rational nature that is not his own. I will have more to say about this case in the next section, but for now what it shows is that if Kant's proof is interpreted in this way, it commits what might be called the self-regarding fallacy, which is to erroneously claim that if someone values a feature of themselves then they must value that same feature in others, for someone may value a personal characteristic on grounds that include an ineliminable back-reference to that agent. If I value my happiness and I can find no relevant difference between it and the happiness of others then I may be committed to valuing the happiness of others as well, but if I value my happiness because it is mine then I am not thereby committed to valuing the happiness of others. In order for Kant's proof to establish the conclusion that all rational agents must conceive of rational nature in others as an end in itself, we need rational grounds for incorporating this possibility into the argument or for setting it aside.

Commentators on Kant's proof are well aware of the self-regarding fallacy, so even when they do not discuss in detail someone who conceives of his rational nature as an end in itself for reasons that contain essential reference to himself, most of them assume there must be resources in Kant for handling this possibility (Rawls Reference Rawls2000: 197; Pogge Reference Pogge1998). One response has been to look back to the ‘regress argument’ and claim that, because our self-regarding agent has the capacity to set objective ends, he is rationally required to value his rational nature as an end in itself, but because others have that same power to set objective ends, he must conceive of their rational nature as an end in itself as well. This does not avoid the self-regarding fallacy, however. If we suppose for now that in setting ends this person makes them objectively valuable in the sense that his ends provide reasons to everyone, he may just consistently deny that others have this same power that, as he sees it, only he possesses. Other agents can set their own personal ends, make plans, and form commitments, he may admit, while denying that they can set objective ends, so he can conclude that their end-setting by itself does not provide him or anyone else with any reasons whatsoever. This would certainly be a ‘remarkable feat of egocentrism’, to judge that only he possesses the power to set objective ends on grounds that he set them, but if his stance is possible and so far not irrational then the regress argument would only commit him to regarding his own humanity as an end in itself, not the humanity of others.Footnote ¹⁶

A second response is to claim that this self-regarding agent is somehow irrational and so excluded from Kant's proof, which only describes how rational agents must regard humanity in themselves and others. But how is he irrational? It is conceivable that this person does not violate any of the most widely accepted formal requirements of rationality – his beliefs may be consistent and his self-regard may not involve failures of instrumental rationality, for example. Such a person, we can assume, satisfies all self-regarding rational requirements included in Kant's ethical system – he values his rational nature as an end in itself, he avoids servility, practises self-improvement and so on. This person certainly violates the formula of humanity, but that is the fundamental rational requirement that Kant is attempting to argue for in this context. As we will see, however, perhaps some progress can be made if we add that, in order to be rational an agent must also be committed to the formula of universal law, which Kant has immediately argued for prior to his proof of FH.

A third way to try to avoid the self-regarding fallacy is to appeal to common moral understanding in an attempt to set aside the outlier case of someone who genuinely regards his rational nature as an end in itself because it is his. Such a person is clearly morally defective, so he may not be the kind of person we have to convince that humanity in general is an end in itself – we may wonder whether a good person could sincerely have and maintain such an attitude. It may be enough for Kant's purposes to show reasonable, conscientious people that they have good and sufficient reasons to treat humanity as an end in itself. Kant's proof may not be an attempt to fend off every logical possibility; it may instead be meant as a reasonable interpretation of ourselves and how we think and act (Wood Reference Wood2008: 96–7).

Finally, Korsgaard (Reference Korsgaard1996a: 122) and Dean (Reference Dean2006: 128–9) draw on some vague suggestions Kant makes in the Critique of Practical Reason that ‘a universal law of nature makes everything harmonious’, to argue that if the categorical imperative merely required everyone to regard her own rational nature as an end in itself, but not others, then the ‘most extreme opposite of harmony would follow, the worst conflict’ because ‘the will of all has not one and the same object but each has his own’ (CPrR 5: 28; Korsgaard Reference Korsgaard1996a: 122; Dean Reference Dean2006: 128–9). My reconstruction of Kant's proof appeals to some of these same ideas, but it is worth noting that the context of the passage they cite is very different from the one in which Kant gives his proof for the formula of humanity, so it is difficult to see how one bears on the other. In the second Critique passages, Kant is concerned with how a universal principle of pursuing one's own happiness would bring disharmony and conflict because we will inevitably desire the same things as others. It is less clear what sort of harmony, if any, he thinks must exist when it comes to valuing rational nature. A moral principle allowing everyone to pursue her own happiness would, according to Kant, result in ‘complete annihilation of the maxim itself and its purpose’, which suggests that certain kinds of disharmony involve failures to satisfy the formula of universal law, so insofar as these passages are relevant to Kant's proof, they point towards a strategy that appeals to FUL. Kant could also be appealing to a sort of harmony that is involved in natural teleology, but there is deep disagreement about whether Kant's views about natural teleology play such a foundational role in his moral theory.

Having surveyed the main contenders for interpreting Kant's proof of the formula of humanity and noted potential problems with each, some of us may find ourselves in the following position. We would like, if at all possible, to find an interpretation of Kant's argument for FH that (1) does not rely on empirical premises about our psychological makeup or what we happen to desire, (2) does not assume that rational nature is already an intrinsic, agent-neutral and objective value but instead (3) attempts to justify that as a conclusion while (4) avoiding the self-regarding fallacy of inferring that everyone must regard rational nature as an end in itself because all rational agents conceive of their own rational nature in that way, for some of them may see their humanity as objectively valuable in part because it is theirs. Others may wish to pursue one of the other strategies, and perhaps there are other options available that (5) do not treat the proof as an utter failure. I now wish to suggest an alternative interpretation of Kant's argument for the formula of humanity that satisfies these five desiderata better than these others.

Rational Self-Regard and Regard for Others

Someone who treats her rational nature as an end in itself because it is hers is usually regarded as a relatively unimportant and outlier case, a mere logical possibility that need not trouble good and conscientious people who are sincerely interested in explaining the objective value of humanity. According to the interpretation of Kant's proof I propose, this sort of person lies at the heart of what Kant intends his argument to accomplish, which is to explain why someone who values her own rational nature as an end in itself because it is hers but who is also willing to reciprocate with those who are willing to reciprocate with her should also value the humanity of others as an objective end. As we shall see, Kant does not mean to refute the radical egoist, who values only himself and his own interests, or the solipsist, who denies that there are other rational agents besides herself – both types of person are excluded by the premise as irrational. Kant instead pursues a strategy familiar from the social contract tradition, which is to appeal to a thin principle of mutuality, in the form of the formula of universal law, to bridge the gap between substantive rational self-regard and substantive rational regard for others. The basic idea is as follows. From my perspective as a rational agent, I value my humanity as an end in itself because it is mine and you value yours in the same way because it is yours, but we are also rationally committed to acting only on maxims that each of us could will as universal laws. My rational self-regard requires that I reject any maxim that permits you or anyone else to disregard or disparage my rational nature and the same is true of you. Therefore, all rational agents could will the maxim that humanity in oneself and others must be treated as an end in itself.

Let's look again at the passage where Kant gives his proof of FH. Kant starts off with the claim that a representative rational agent, say R, necessarily conceives of his own rational nature as an end in itself. Doing so is necessary for R, not as a matter of his empirical psychology, but in the sense that it is rationally necessary for R to regard his own humanity as an end in itself. This means that R is rationally committed and disposed to conceive of his own rational nature as objectively valuable and if he were fully rational he would succeed at always treating himself accordingly. Kant describes this way of regarding himself as a subjective principle of human action in order to indicate that the principle is not only a principle on which R acts but is also about R himself. On my construal, R does not automatically make the non-rational ends he sets objectively valuable, as proponents of the regress argument claim, but it is part of his nature as a rational agent to conceive of his own humanity as an end in itself, which he does by treating it as ‘above all price’ and ‘without equivalents’, showing himself proper respect and avoiding servility, standing up for himself and insisting that others not ridicule or humiliate him, developing his talents, and otherwise satisfying his duties to himself. Kant's proof of FH therefore begins with duties to oneself as part of the foundation of duties to others.

Kant next claims that every other rational being conceives of his or her own rational nature as an end in itself. I have followed the common practice of formulating this part of Kant's proof in terms of the way all rational agents conceive of their humanity, but it is worth emphasizing that Kant actually begins the relevant passage by describing how a particular representative rational agent conceives of himself. He does so, I suggest, in order to signal that the argument is addressed to the first-person perspective of someone who conceives of his own humanity as having absolute worth but wonders why, if at all, he should view the rational nature of others in that way as well. If Kant can give this person sufficient reasons to have proper regard for humanity in others then he will have explained to each one of us, from our own rational standpoints, why we should do so as well, which is just the approach we would expect from a contractualist style of argument.

We are then told that each of us regards our own rational nature as an end in itself ‘on the same rational ground’ as everyone else. What is this ground supposed to be, why is it that rational agents, in virtue of their nature as such, supposedly regard our own humanity as objectively valuable? Kant's vague footnote tells us that these grounds will be more fully described in the final chapter, so we must briefly turn to Groundwork III. His line of reasoning there is notoriously difficult, but among the parts of it that Kant could be alluding to as the grounds that lead rational agents to regard their humanity as an end in itself, the most plausible one is his claim that rational agents have autonomy of the will.

Autonomy of the Will

Kant argues that rational agents must act ‘under the Idea of freedom’ (G 4: 448) by taking ourselves to be free in the negative sense that our wills can do things ‘independently of alien causes determining it’ (G 4: 446). Because our contingent interests and sensibilities cannot ground moral laws that are universal and rationally necessary (G 4: 446–7), Kant argues that having negative freedom entails having autonomy of the will, which is the capacity to will unconditional and rationally necessary principles for ourselves and others, independent of our inclinations, as part of a moral community in which we are all author and subject of those laws (G 4: 440). When Kant says that we each conceive of our rational nature as an end in itself on the same rational grounds and points us to Groundwork III for those grounds, he evidently means that a constitutive feature of having autonomy of the will is that an agent values his own rational nature as an end in itself, not because he values rational nature in general, but because it is his rational nature and he has autonomous control over it.

Why might it be a constitutive feature of an autonomous agent that she conceives of her humanity as an end in itself? Kant may not have said enough about why he thinks this claim is true, and perhaps he would refer to the fact of reason and urge that we treat it as a substantive yet ungrounded starting point – his view may be that moral inquiry begins with how we are rationally disposed to treat ourselves and proceeds to explain how we must treat others as well.

Yet Kant's conception of autonomy may provide some additional clues about why he thinks someone who has autonomy of the will necessarily values her humanity as an end in itself. Autonomous wills are in part self-regarding. An agent can only will for herself and set her own ends, but it is impossible to do this for others (MM 6: 381); she has most direct authority over herself rather than anyone else, much as a political constitution governs the society it belongs to rather than other societies that it can nonetheless stand in certain relations to; and an autonomous will can only develop and perfect its own rational powers, not those of others (MM 6: 444–5). An autonomous will is described as fundamentally ‘a law unto itself’ rather than a law governing the wills of other rational agents. If an autonomous will is, at least initially, concerned with itself then it is plausible that such a will is disposed to preserve, protect, develop and respect its own rational nature, and otherwise treat its humanity an end in itself.

An autonomous will is not wholly self-regarding, however, for it also accepts as its most fundamental principle to will moral laws for human beings that are justifiable to all in light of their self-regarding rational interests. Kant describes this principle of autonomy in Groundwork III as the ‘formula of the categorical imperative’ to act ‘on no other maxim than one that can also have [itself as] a universal law for its object’ (G 4: 447; translation slightly revised from that by Hill and Zweig). The unconditional moral imperatives that an autonomous will legislates are always the same because they are not willed on the basis of inclinations, personal tastes or preferences but instead are willed in virtue of essential rational interests of that rational, autonomous agent. These interests include protecting and developing our rational powers, securing the ability to exercise our autonomous will, preventing others from degrading, humiliating or subjugating us, and so on, all within the bounds of moral laws that are justifiable to all.

Kant's idea of autonomy therefore includes the two essential features of a contractualist theory, namely a set of self-regarding interests and a principle of reciprocity requiring each to act in ways that are acceptable to all in light of those interests.Footnote ¹⁷

Kant's reference to autonomy helps to clarify the purpose of his proof of the formula of humanity. Kant is not attempting to convince the solipsist that there are other minds or agents in the world, for he argues that we are rationally committed to recognizing others as having autonomy of the will, which means we must regard them as having essential self-regarding rational interests along with a fundamental disposition, which we share, to reciprocate with one another (G 4: 447–8). Nor is Kant addressing the radical egoist, who cares only for his own interests, because an autonomous will is fundamentally committed to reciprocating with others. His proof is instead meant to convince a rationally self-regarding and autonomous person who conceives of her own rational nature as an end in itself and is committed to acting in ways that are justifiable to others in light of their self-regarding rational interests, but wonders how more specifically she should treat them and, in particular, why she should regard their rational nature as an end in itself as well.

Formula of Universal Law

Supposing rational agents with autonomy of the will necessarily regard their own rational nature as an end in itself, why should they regard the humanity of one another as an end in itself as well? On my interpretation of Kant's proof, the transition from the premise to the conclusion is mediated by the formula of universal law, interpreted as a contractualist principle of reciprocity or mutuality.

Consider the place of Kant's argument for the formula of humanity in the overall argumentative structure of Groundwork II. In that chapter, Kant begins with our common notion of duty and concludes that we are subject to moral duties if and only if we are rational agents with autonomy of the will. Immediately prior to his discussion of the formula of humanity, Kant has concluded that our common notion of duty presupposes that the supreme moral principle is the formula of universal law: ‘Act only on that maxim by which you can at the same time will that it should become a universal law’ (G 4: 421). It is reasonable to expect that subsequent steps in Kant's overall Groundwork II argument could then appeal to that conclusion, which Kant takes to have been established. Therefore, having argued that rational agents conceive of their own rational nature as an end in itself, Kant can invoke FUL to explain why they should regard the humanity of others as objectively valuable as well.

Interpretations of FUL differ, but one common problem is to specify the standards that determine whether or not someone can will a maxim as a universal law. These standards are not merely a matter of what people happen to want or contingent factors about their psychology but instead specify reasons for what we can and cannot rationally will. Various standards have been proposed: a rational agent wills the ‘normal and predictable results’ of his actions, he avoids undermining his own purposes, he takes the necessary means to his ends or gives up his ends, or wills the capacity for effective willing (O'Neill Reference O'Neill1975: 63–93; Reference O'Neill1990b; Korsgaard Reference Korsgaard1996b; Herman Reference Herman1993: 121–2). Whatever these standards turn out to be, when deciding whether he can rationally legislate or will his proposed maxim as a universal law, a rational agent looks for an inconsistency between, on the one hand, his maxim, along with its typical and foreseeable results, underlying purposes, or any other necessary means it may involve, and, on the other, everyone acting or being permitted to act in that way.Footnote ¹⁸ Commentators disagree about how well these various views fit the text, whether all of them actually specify standards of practical reason, and whether they can generate all of the moral duties that seem evidently correct.

Thomas E. Hill (Reference Hill2002b) has proposed that standards of rational willing include, along with formal ones of consistency and coherence, substantive requirements about what ends a rational person must will (see also Guyer Reference Guyer1995). He argues that these substantive standards should be imported from other formulations of the categorical imperative, particularly FH. If rational agents necessarily treat humanity in themselves and others as an end in itself then any proposed maxim that, when universalized, would be incompatible with regarding humanity in this way, cannot be legislated or willed as a universal law by any rational agent. According to Hill, what a rational agent could will as a universal law is what she would will as a universal if she were fully rational. Hill supplements FUL with Kant's claim that rational agents are rationally committed to treating humanity as an end in itself, but our concern has been whether Kant has an argument that rational agents must regard one another in this way.

A different but related strategy for incorporating substantive requirements of rational willing is to interpret FUL as a principle of moral contractualism that require us to act only in ways that are justifiable to everyone, where the rational standards of justifiability are either formal or substantive and self-regarding. More specifically, a person should not act on a maxim that, if universally adopted, could not be rationally willed by each and every rational agent in light of what they each rationally will for themselves. If the universal counterpart of someone's maxim conflicts with the rational self-regard of another agent then that maxim is impermissible. Rational agents are committed to their own rational interests but they are also willing to reciprocate with other rational agents in the sense that they will constrain themselves by principles that are acceptable to everyone.

What, then, are the self-regarding and substantive standards of rational willing that determine whether a maxim can willed as universal law? As I see it, Kant fills in these rational interests by specifying, in his proof of FH, that a rational agent necessarily regards her humanity as an end in itself. Regarding oneself in this way not only involves acting accordingly towards oneself but also standing up for oneself with others, insisting that they treat one's humanity as an end in itself and refusing to allow oneself to be treated as a relative end. A rational agent is therefore committed to acting only on maxims that can be willed as universal law by all rational agents in virtue of their self-regarding rational standards, including the rational interest we all share in treating our own humanity as an end in itself.

The Argument

Kant claims that the principle of autonomy and FUL are somehow equivalent (G 4: 436), and interpretations of both formulations differ, but a person who has autonomy of the will is committed to the categorical imperative in both forms. As I read Kant's proof, a rational agent conceives of her humanity as an end in itself because she has autonomy of the will. As an autonomous agent, she is also committed to acting only on maxims that are acceptable to all rational agents in virtue of their self-regarding rational standards. Any principle that permitted the rational nature of someone to be treated as anything other than an end in itself is inconsistent with the self-regarding rational commitments of such a person – no one could rationally accept a moral principle that permitted others to treat her as a relative end. Each rational agent could rationally will or legislate, therefore, a principle that says that rational nature in oneself and others must be treated as an end in itself, for this principle is consistent with each person having proper regard for herself. A rational agent, in other words, is rationally disposed to ensure that others treat her rational nature as an end in itself, but she is also willing to act on principles that are, in a sense, acceptable to all, so rational agents must treat humanity in one another as an end in itself. That, as I see it, is how Kant's proof of FH proceeds.

More formally, we can express the argument as follows:

This argument satisfies the four criteria I defined. (1) It avoids empirical premises about what we contingently want, along with (2) assumptions about antecedent and agent-neutral values. (3) My reconstruction attempts to establish as a conclusion that all rational agents have reasons to treat one another's rational nature as an end in itself on the basis of an assumption about how agents regard themselves. (4) It avoids the self-regarding fallacy by assuming that rational agents are not only rationally self-regarding but are also deeply committed to reciprocating and giving the same to others. And (5) my reconstruction shows Kant's proof of FH to be forceful and philosophically interesting.