1. The axiom of justice and the universal law formula of the categorical imperative

How many self-evident axioms Sidgwick detected is open to debate, but on first sight, there appear to be three: the axiom of justice, the axiom of prudence and the axiom of benevolence.Footnote ⁵ The axiom of prudence will be left aside since Sidgwick does not claim convergence with Kant on this subject. The canonical formulationFootnote ⁶ of the axiom of justice reads as follows:

In short, this axiom claims that it cannot be right to treat two persons differently, unless a relevant difference between them can be identified as the decisive reason for this. Sidgwick claims that there is a literal logical dependency relationship between this principle, or axiom, of justice and the categorical imperative. He argues that Kant’s FUL, ‘duly restricted’, is ‘an immediate practical corollary’ of the axiom cited above (Sidgwick (Reference Sidgwick1981 [1907]: 386).Footnote ⁷ Kant’s FUL requires in Sidgwick’s own words: ‘acting on a maxim that one can will to be law universal’ (ibid.; corresponding to Kant, G, 4: 421). As quoted, Sidgwick clearly states that this formulation of the categorical imperative follows logically from the axiom.

Sidgwick explicitly claimed to be impressed by FUL, but he never accepted all of Kant’s examples of it and rejected its ultimate rationale. In particular, he did not believe that FUL could support a general prohibition against lying or the duty of beneficence.Footnote ⁸ Even more importantly, he denied that morality can ultimately be grounded in human freedom. In his introduction to the sixth edition of the Methods, he writes about Kant’s system:

The restrictions about the examples and about the metaphysical basis of Kant’s view which Sidgwick placed on his acceptance of FUL seem to support Kantians who are doubtful about the existence of the close relationship that Sidgwick sees between his axiom and Kant’s formula. Nonetheless, some possible attempts to formulate the central difference are less promising than others. In particular, it is often thought that utilitarians tend to hold a problematic, teleological interpretation of FUL, according to which the world would simply end up in a worse state if everyone acted on the basis of non-universalizable maxims.Footnote ⁹ It is possible that Mill is assuming a teleological interpretation of FUL when he says that ‘[t]o give any meaning to Kant’s principle, the sense put upon it must be, that we ought to shape our conduct by a rule which all rational beings might adopt with benefit to their collective interest’ (Mill Reference Mill and Robson1985 [1833]: 249, emphasis added). However, there is no evidence in the Methods, or anywhere else, that Sidgwick held such a teleological view. On the contrary, as he particularly emphasized in the History of Ethics, actions based on non-universalizable maxims cannot even be conceived or involve a contradiction of the will for Kant. To quote a central passage in full:

Passages such as these make it very unlikely that Sidgwick can actually be accused of having held a teleological interpretation of Kant’s FUL.

Another issue which could be raised is whether Sidgwick’s axiom of justice could be a version of the Golden Rule, which Kant discredited. If this were the case, then Sidgwick’s claim would actually mean that FUL is a logical corollary to a version of the discredited Golden Rule. Even though this would be an interesting position,Footnote ¹⁰ it is very unlikely that it can be attributed to Sidgwick. Immediately before his central characterization of the axiom, Sidgwick discusses various versions of the Golden Rule, agrees explicitly with Kant that the Rule is problematic and discusses some of the well-known reasons for this: the rule ‘Do to others as you would have them do to you’ could lead to absurd consequences if a person wishes that others would cooperate in his sins; he also rejects the negative version, according to which ‘you should not do to others what you would take to be wrong if done to yourself’, because there could be relevant differences between people which would render this version incorrect (Sidgwick Reference Sidgwick1981 [1907]: 379–80). In fact, it appears that Sidgwick’s declared goal in constructing the exact formulation of the axiom of justice was to provide a more precise and better principle than the various versions of the Golden Rule.

One difference lies in the fact that the exact formulation of the axiom of justice contains no reference to any subject’s wishes, will or opinions about what is right, as is the case in the discredited versions of the Golden Rule. Instead, it claims that in cases in which it is simply wrong for B to treat A in a certain way, irrespective of anyone’s beliefs or wishes about the matter, the reverse must also be false. A similarity which remains between the axiom of justice and these versions of the Golden Rule is the mirror-like comparison of particular kinds of scenarios, but the claim that what is right can be deduced from a particular individual’s wishes, will or opinions is abandoned in favour of an assumption of objective rightness.

A further, related objection to the allegedly close relationship between Sidgwick’s axiom of justice and FUL is that FUL was intended as a way of determining the moral rightness of a much larger variety of relevant maxims. In an earlier part of the Methods, Sidgwick explicitly states that an action that is acceptable according to FUL could in fact turn out to be wrong for other reasons:

As this passage demonstrates, Sidgwick did not accept the general law formula as a final, or a complete, test of the rightness or wrongness of all sorts of morally relevant maxims. In a similar way, Sidgwick only grants limited powers to the axiom of justice – the axiom is unable to settle all morally relevant questions on its own. Had he thought that the axiom could do so, then it could have remained the sole axiom – but he found others which are not identical in scope and meaning.

What remains somewhat puzzling is the reason Sidgwick gives for his assessment that FUL has only limited scope. Immediately after the passage quoted above, he explains that assuming that the categorical imperative can give complete guidance would lead to the obliteration of subjective and objective rightness: it would imply that anyone acting conscientiously could sincerely will his maxim to become a universal law, even if it were wrong. He writes:

It would indeed be strange if a conscientiously formed subjective belief – even a false one – according to which a certain maxim is universalizable, would in and of itself indicate that the respective act is permissible. If this were possible, there would no longer be a difference between subjective views about the right and objective rightness. Moreover, in cases in which there is moral disagreement, this would also lead to the absurd consequence that a particular act could turn out to be permissible and impermissible at the same time, depending on whose conscientiously formed belief we are looking at.

Kantians would have a number of possible responses to this very brief consideration. For example, in contrast to Sidgwick, a Kantian might be willing to simply bite the bullet and insist that the good will is decisive even if it is based on false beliefs and beliefs that others disagree about. The Kantian could also claim that merely believing that your maxim is universalizable does not amount to it being universalizable. Unrestricted, first-personal access to the rightness of a maxim is not a necessary feature of Kant’s moral metaphysics. Sidgwick may literally not have seen some of the strong argumentative resources that Kant could have utilized to rebut this charge: a charge that seems to have led Sidgwick to grant only limited scope to FUL.

Despite Sidgwick’s attempt to arrive at a contrary conclusion, the FUL probably achieves more than the axiom and thus has a broader scope.Footnote ¹² In particular, it is supposed to explain what makes an action right or wrong in the first place. The canonical formulation of the axiom, on the other hand, is a meta-rule which already presupposes that certain actions are right or wrong and merely demonstrates that mirror-like extensions to similar cases are obligatory. While some of the restrictions that Sidgwick places on his acceptance of FUL, especially regarding the lying and FUL’s support of beneficence, might be a matter of debate, one of the central reasons for his restrictions – the one about the obliteration of subjective and objective rightness – could rather easily be rejected by Kant. A stipulation of unrestricted subjective access to whether a maxim can rightly be willed is certainly not a feature of Kant’s view.

Overall, claiming convergence with Kant on the grounds that Kant’s view could be different in ways that he clearly rejects can be viewed as a strange argumentative approach. The question of whether the ‘duly restricted’ version of FUL that Sidgwick imagines and the axiom of justice are compatible has comparatively little to do with the relationship between the axiom and FUL as presented by Kant.Footnote ¹³

2. The axiom of benevolence as supported by Kant’s duty of beneficence

Sidgwick also believes that Kant would agree with the axiom of benevolence. The most famous formulation of this axiomFootnote ¹⁴ states:

According to this axiom, the welfare of all individuals is of equal value and one should promote general welfare rather than the welfare of any particular individual. Preference is to be given to individuals for whom one can promote more good than for others. Sidgwick argues that there is a close relationship between this axiom and Kant’s views. More precisely, he argues that ‘Kant’s conclusion appears to agree to a great extent with the view of the duty of Rational Benevolence that I have given’ (Sidgwick Reference Sidgwick1981 [1907]: 386).

Immediately before this remark about the presumed convergence with Kant, he elaborates on Kant’s virtue of beneficence. In particular, he says that rational benevolence, or the happiness of others, is the only ultimate end Kant recognizes, and that Kant takes it to be evident a priori that each person as a rational agent is bound to aim at the happiness of others (Sidgwick Reference Sidgwick1981 [1907]: 386). In an added footnote about the Doctrine of Virtue, Sidgwick admits that one’s own perfection is acknowledged by Kant as a second ultimate aim. However, in the realm of duties towards others, for Kant the ultimate purpose, as stated in the Doctrine of Virtue, is indeed their happiness (Glückseligkeit).Footnote ¹⁵

In a short passage in Sidgwick’s History of Ethics, there is a further, more explicit explanation of Kant’s duty to promote the happiness of others which refers to both the Groundwork and the Doctrine of Virtue.Footnote ¹⁶ In this monograph, Sidgwick correctly characterizes the humanity formula as the imperative to treat humanity, both in one’s own person as well as in the person of any other, as an end and never merely as a means (Sidgwick Reference Sidgwick1988 [1902]: 275, discussing G, 4: 429). He also highlights Kant’s claim that human rational nature constitutes an end in itself (p. 274, discussing G, 4: 429). Thereafter, he raises the notorious question of what exactly it might mean to treat someone as an end, and comes to the conclusion that it cannot mean to treat others in such a way that their own moral perfection is promoted, because perfection is something that a person can only ever accomplish on their own. Instead, he further concludes that it can only mean that one ought to promote the happiness of others – more specifically, happiness in the form in which it represents itself to these other people. He finally states that the subjective happiness of others must become my own end if I want to treat them as an end.

The elucidations in the History of Ethics are adequately supported by references to the relevant passages in the Doctrine of Virtue and the Groundwork.Footnote ¹⁷ The relevant passage in the Doctrine of Virtue to which Sidgwick refers in the History of Ethics is the one in which Kant says that there are two ultimate ends, happiness and virtue, and that my own happiness is a natural inclination rather than a duty, and that others’ perfection is not something that I can accomplish for them. In the realm of virtues towards others, Sidgwick continues, the promotion of their happiness is the only ultimate end acknowledged by Kant (Sidgwick Reference Sidgwick1988 [1902]: 275, discussing MM, 6: 385–6). The Groundwork passage to which Sidgwick refers in the History of Ethics is part of the discussion on why the humanity formula of the categorical imperative requires beneficence towards others. Kant admits that humanity could exist without anyone ever contributing to another person’s happiness but demands a positive contribution to humanity in the form that assisting others in their pursuit of ends has to become one’s own end to the largest extent possible (ibid., discussing G, 4: 430).

Overall, that Kant acknowledges a duty to promote the happiness of others is not controversial. Sidgwick’s modest formulation, that Kant ‘appears to agree to a great extent’ with the axiom of benevolence, lends further initial credibility to his claim. Perhaps Sidgwick never meant to say more than that Kant agrees that benevolence is one’s duty. Sidgwick’s apparent general desire for finding points of convergence rather than disagreements might support such a modest interpretation of his claim. However, it would also make the alleged convergence with Kant much less interesting.

Many Kantians’ primary objection to Sidgwick’s partial convergence claim about benevolence will be on other grounds. For example, they will argue that Kant never thought that the size of the benefit that one can bestow on someone could be a reason for this person’s preferential treatment and that Kant never thought that one should aim for the general good of all, rather than the individual good of particular others, presumably those with whom one is engaged in a rather immediate way. However, it is interesting that the passages which Sidgwick seems to have had in mind, namely the statement of the humanity formula and a further passage from the Groundwork which Sidgwick actually quotes in the History of Ethics, are written in a somewhat misleading way in this respect. The passage in question follows the central formulation of the humanity formula according to which you ought to act so that you use humanity, both in your own person and in the person of any other, as an end and never merely as a means.Footnote ¹⁸ The subsequent elucidatory passage, quoted by Sidgwick in the History of Ethics, contains a suggestive superlative:

In this passage, Kant – if taken literally – seems to say that one ought to promote any other person’s happiness to the greatest extent possible.Footnote ²⁰ If this is correct, it seems to claim that it is not sufficient to promote the good of others merely to some extent, but rather that the impartial maximization of the good of others is required: a thought that can often be found in general descriptions of the utilitarian doctrine.

There is a chance that Sidgwick fell prey to an overly utilitarian understanding of this passageFootnote ²¹ – an understanding that is incompatible with Kant’s actual view. This is confirmed by a passage in the Doctrine of Virtue where Kant says that there is discretion regarding the ways and extent to which the duty of practical love should be exercised. The duty of practical love allows for narrowing specification; for example, in form of the rule that one should honour one’s parents (MM, 6: 390).

By now, several different interpretative attempts seek to make sense of this latent disagreement between the two authors. As Roger Crisp argues, in accordance with Terence Irwin (Crisp Reference Crisp2015: 123, 126; Irwin Reference Irwin2009: 512–13), ethicists in Kant’s camp would likely object to the strong impartiality of Sidgwick’s formulation of the axiom, which seems to preclude such partiality towards friends and family. Nonetheless, as Anthony Skelton has recently demonstrated (Skelton Reference Skelton, Paytas and Henning2020: 22ff.), it should be remembered that Sidgwick explicitly considers the possibility that partiality towards close others might be beneficial overall.Footnote ²² This means that his general support of impartiality might have been weaker than the formulation of the axiom suggests and he might have only thought that there is no ultimate reason for partiality. This could imply that the divergence between the authors on this point is not as pronounced as one might think.

There is a further way in which Kant rejects impartiality, but which Sidgwick could, in principle, follow to some extent. Already on the first page of the Groundwork, Kant points out that there is no worth in the happiness of a morally undeserving being.Footnote ²³ Arguably, this view has an impact on Kant’s duty of benevolence: if one could choose either to make a morally deserving or undeserving person happy, one might have to be partial towards the morally deserving person. Even if human beings do not have a duty to distribute happiness according to moral desert and only God is in charge of this, the point remains that for Kant the value of happiness seems to depend on moral desert. In this sense, the value of happiness is not an impartial matter for Kant. However, arguably Sidgwick could also agree with this form of partiality. Just like partiality towards friends and family members, partiality towards the deserving might be indirectly conducive to the general good. Since Sidgwick’s support for partiality towards the deserving would likely be instrumental as well, while there does seem to be a difference on the level of justification, of course, the practical divergence seems less clear.

There are further reasons for the claim that reading an individual or a general happiness maximization requirement into the quoted passage from the Groundwork – as Sidgwick may have done – must be wrong. For instance, Kant would assume that the rational ends of human beings must, at a minimum, be permitted by the categorical imperative – a requirement that happiness, as an end, obviously does not automatically satisfy.Footnote ²⁴ Furthermore, for Kant, benevolence is an imperfect duty, so if there were a conflict with a perfect duty, Kantians typically argue that the duty of beneficence would actually be suspended.Footnote ²⁵ While Sidgwick seems to have been to some extent open to such a non-absolute view about benevolence in his own account (Sidgwick (Reference Sidgwick1981 [1907]: 421), he does not address the fact that it is clearly a feature of Kant’s account.

As a consequence, Thomas Hurka and Terence Irwin plausibly conclude that Sidgwick did not distinguish clearly enough between benevolence as an absolute principle and as an ‘other-things-equal’ principle (Hurka Reference Hurka2014: 149, 151–2; Irwin Reference Irwin2009: 518). Sidgwick might not have taken full account of the imperfect character of Kant’s duty of beneficence, which can be outweighed by perfect duties. Consequently, he might have assumed Kant would agree with the more broadly construed utilitarian form of benevolence.

However, as shall be demonstrated, the Hurka-Irwin thesis does not offer a complete picture of the divergence between Kant and Sidgwick when it comes to beneficence. As Alan Donagan points out, Sidgwick did not accept the view that humanity, with its capacity for rationality, can be viewed as an end in itself.Footnote ²⁶ For Sidgwick, the humanity formula primarily says that one should adopt the practical ends of others as one’s own. But this is not how Kant saw it. For Kant, an end is not merely something to be produced, a consequence to be brought about; rather an end in itself is an already existing rational individual for the sake of whom one ought to act. Treating someone as an end in itself is important because it means respecting this human individual’s inherent freedom. Adopting the aims of others, practising beneficence, is merely a plausible way of respecting what is really important to human beings: to be respected as rational free agents, as ends in themselves.

We see clearly Sidgwick’s worries about the role of rationality in the humanity formula in a lengthy footnote to the chapter containing the convergence claim. Here he explains his problems seeing or accepting Kant’s derivation of the duty to promote the happiness of others from the humanity formula. As he puts it toward the end of the footnote:

In addition to expressing his central concern about how to derive positive duties, in particular the duty of benevolence from an end understood as something already there, he is generally sceptical about Kant’s understanding of humanity as an end in itself.

Moreover, he did not see that Kant’s meaning of the term ‘end’ in the humanity formula provides a principled reason for excluding at least the morally irrational, and arguably even the prudentially irrational impulses of others.Footnote ²⁸ The passage about having to adopt other people’s ends as one’s own has to be read in light of the humanity formula according to which the reason for taking the ends of others as important is their character as rational human beings who are ends in themselves. Promoting morally (and perhaps also prudentially) irrational ends of others is not a plausible way of showing respect for their capacity as rational beings who are ends in themselves. It can certainly be argued that Sidgwick’s puzzlement in his footnote can be explained by an assumption on his part according to which he did not see the non-absolute character of Kant’s duty of benevolence, as Hurka and Irwin would argue. However, Sidgwick’s error can be even better explained by noting a fundamental oversight regarding the full Kantian meaning of an end and its implications for the duty of beneficence.

Sidgwick failed to take proper account of the decidedly non-absolute nature of beneficence in Kant. For Kant beneficence is merely an imperfect duty that can be overridden by perfect duties. To say this differently, beneficence is overridden whenever the ends of the other person that beneficence is meant to promote are morally irrational. In this case the ends do not express the rationality constitutive of a human being as an end in and of itself.

The analysis suggested here is strongly supported by a passage in the History of Ethics. As indicated earlier, in the History of Ethics Sidgwick explicitly raises the question of what it might mean to treat someone as an end. His response to this rhetorical question, in the exact context of the discussion of the humanity formula, very clearly indicates that he adopts an understanding of an end as being a goal to be produced, and that he rejects the idea that treating someone as an end might mean treating him as an end in itself, by respecting his rational human capacities:

Of course, one might admit that it is not immediately clear what respect for someone’s rational capacities entails in terms of actions, but this further and certainly important Kantian issue must not disguise the fact that Sidgwick did not consider, or even understand, this possible interpretation of the term ‘end’ in the humanity formula.

In short, while the Crisp-Irwin and Hurka-Irwin analyses are illuminating, they can also be called into question to some extent. Proponents of both of these views must acknowledge that there are brief passages in which Sidgwick considers views in the respective, more Kantian directions. As already mentioned, he explicitly considers the possibility that partiality towards one’s peers might, in practice, be beneficial overall. He also briefly indicates a willingness to consider non-absoluteness concerning benevolence in the chapter on the proof of utilitarianism, in which it is stated that the duty to promote the general good might not be the sole or supreme principle of ethics (Sidgwick Reference Sidgwick1981 [1907]: 421). These passages are very brief and they are by no means fully explicated. Nonetheless, Sidgwick’s editing process was extensive and he might have developed them further had there been sufficient time and energy for an eighth edition of the Methods. It is an advantage of my alternative analysis that it is not called into question even by any brief remarks in his texts. Moreover, it seems to provide a reason for why he did not see the non-absolute character of Kant’s duty of benevolence, rather than simply stating the point.

In other words, the tension between Sidgwick and Kant’s views on benevolence, about which Sidgwick claimed that there is partial convergence, lies only partly in a divergence about the proper extent of impartiality or the relevant grounds of partiality, or in the fact that beneficence has the form of a non-absolute rather than an absolute principle for Kant. Another important feature of the divergence lies in the fact that Sidgwick did not accept the full Kantian understanding of what constitutes an end.Footnote ²⁹

3. How this interpretation is tied to Sidgwick’s worries about metaphysics

A further advantage of this interpretative thesis over others consists in the fact it can provide a clear link to Sidgwick’s famous and central concerns about the metaphysical basis of Kant’s system, in particular in regard to the relationship between rationality and freedom and in regard to the existence of a Supreme Being. First, Sidgwick had well-known worries about Kant’s view of the relationship between rationality and freedom. The most well-known treatment of this is found in his paper ‘The Kantian Conception of Free Will’ (Sidgwick Reference Sidgwick1888). which first appeared in one of the early issues of the journal Mind and was later partly added as an appendix to the Methods. One of his worries regarding rationality and freedom concerns the way in which Kant derives beneficence from the humanity formula. As already noted, Sidgwick finds this argument perplexing, because he thinks that it requires adopting other people’s subjective ends determined by irrational impulses. However, as Paul Guyer argues (Guyer Reference Guyer2000: 156ff.), Sidgwick’s interpretative mistake consists in the fact that he mistook respect for ends as meaning respect for pure rationality, without any regard for desires and aversions. As Guyer insists, the ends of others that we must respect are the ones that are mutually compatible with the freedom of all others. In other words, respect-worthy ends must meet a respect for rationality test: they may indeed be derived from subjective and non-rational impulses, as long as they are not incompatible with the freedom (of others) to set their own ends. This freedom-related concern is another facet of Kant’s view of ends that Sidgwick fails to be fully cognizant of.

For Kant, the duty to treat humanity as an end in itself is not just due to the fact that humans tend to possess the capacity for rationality; it is also important because the exercise of this rationality is related to their freedom. For Kant, human beings exercise their moral freedom from sensuous inclinations by acting rationally (setting their own ends) in accordance with the moral law. This Kantian relationship between rationality and freedom appears problematic to Sidgwick, as he lays out in ‘The Kantian Conception of Free Will’. In this article, he argues that Kant uses the term ‘freedom’ in incompatible ways without noticing it. On the one hand, he argues, there is the notion of rational freedom in Kant, which means acting in accordance with the moral law, and on the other hand, there is neutral freedom, which means having a choice, including the choice not to act according to the moral law. The upshot of this criticism is that if Kant wants to stick with his strong tie between rationality, freedom and the moral law, then rational and deliberately wrong actions no longer appear to be possible. This would, of course, make ascriptions of moral responsibility for wrongdoing problematic.Footnote ³⁰ Sidgwick is not the only author who raises this concern. Reinhold advances a similar worry,Footnote ³¹ and Kant’s understanding of freedom and its various forms is still under discussion todayFootnote ³² – sometimes even in light of Sidgwick’s criticism.Footnote ³³

In other words, there are two aspects of humanity as an end that Kant usually highlights: the rationality and the freedom of human beings. Sidgwick saw a tension between these two aspects – a tension which he considers at the heart of the Kantian notion of respect for humanity as an end in itself. Representing the bigger picture of Sidgwick’s misunderstanding of Kant’s view provides a clearer and more comprehensive interpretation of the argument between them.

Secondly, there are some symmetries between freedom and the existence of a Supreme Being in Kant. Both themes play a central role in the Critique of Pure Reason, especially in the discussion of the antinomies, and Kant ultimately decides that they are postulates of practical reason. In the discussion of the third antinomy, Kant tries to establish that it is not contradictory to the causal laws of nature to assume that human volition can be a starting point for a chain of events in the world. There is evidence that Sidgwick found the relationship between freedom of the will and a Supreme Being in the antinomies to be problematic. As he argues in his rarely discussed Lectures on the Philosophy of Kant – which deal with Kant’s first Critique – this speculative argument in favour of transcendental freedom of the will seems to depend on the transcendental argument in favour of a necessary Being in order to be fully convincing. In a lecture addressing Kant’s third antinomy (on free will) and the fourth antinomy (about a necessary Being) in the Transcendental Dialectic, Sidgwick writes:

And a few pages later:

In other words, Sidgwick could not see how the argument in favour of transcendental human freedom could work without also accepting the transcendental argument in favour of the existence of a necessary Being at the very beginning of all further contingent free causes. Even though many Kant commentators are sceptical whether the claim that we have to assume a necessary Being to explain the assumption of further free causes in the world is convincing or was even intended by Kant,Footnote ³⁴ Sidgwick clearly saw a strong link there – which must have aggravated his reservations about the metaphysical basis or implications of Kant’s moral philosophy. It is well known that Sidgwick had deep-rooted concerns with theology.Footnote ³⁵ He could not possibly have been content with a view according to which both a Supreme Being and human freedom are necessary. This point becomes even harder to deny if one considers the fact that he took the whole issue of the freedom of the will to be rather irrelevant to ethics in the first place (Sidgwick Reference Sidgwick1981 [1907]: 57ff.).

Of course, Sidgwick is aware that the theoretical discussion in the first Critique is not the ultimate justification of the postulate of human freedom. It is the moral law – rather than speculative reason – that implies, or even requires, the existence of God and the afterlife in order to enable the reconciliation of happiness and moral merit. As Tyler Paytas (Reference Paytas, Paytas and Henning2020) argues, both Kant and Sidgwick worried deeply about the relationship between adhering to one’s duty and the possibility of reaching happiness, and both saw the possibility of reconciling the tension with an appeal to God and the afterlife. There is in fact a similarity here, and both authors ultimately reject a divine command theory of moral motivation. However, there is also some reason to believe that, regarding the motivational force of the moral law itself,Footnote ³⁶ Sidgwick’s Kant exegesis was focused on an overly strong reading of Kant. As he points out in both the Lectures on the Philosophy of Kant and in the History of Ethics, Kant’s first Critique says that ‘without a god and without a world not visible to us now but hoped for, the glorious ideas of morality would be indeed objects of applause and admiration, but not springs of purpose and action’.Footnote ³⁷ According to this passage, the moral law itself does not motivate us to act; it is the existence of a God who reconciles duty with happiness. It is interesting that Sidgwick repeatedly quotes this particularly strong passage from the first Critique, and does not mention any more moderate passages, especially from the second Critique. It can be maintained that, according to the second Critique, the existence of God is not literally necessary for the moral law to be persuasive, but his existence is merely implied by the already-given rational persuasiveness of the moral law.Footnote ³⁸ Sidgwick may either have not been aware of this change, or it could be that a mere implication of the existence of God was just as unpalatable to him as the stronger view about the alleged necessity of his existence that he quotes.

It is interesting that Sidgwick primarily discusses the stronger reading, according to which God is literally necessary to motivate moral behaviour, in the last section of the Methods, after acknowledging a fundamental dualism of practical reason. According to this unfortunate dualism, it is always rational to act morally, but at the same time also always rational to act egoistically. As he sees it, a Supreme Being who rewards us for obeying the rules of duty would resolve the dualism of practical reason, but he cannot bring himself to confirm the existence of such a Being. In fact, in a famous footnote to this point he goes as far as saying that ‘feeling bound to believe for purposes of practise what I see no ground for holding as a speculative truth’ appears to him as a ‘half-wilful irrationality, committed in a violent access [sic] of philosophic despair’ (Sidgwick Reference Sidgwick1981 [1907]: 507, n. 3). Given the immediate context of the footnote, according to which the idea is that God will punish and reward, this exclamation is probably directed at the stronger and more controversial reading of Kant that goes back to the first Critique only. Sidgwick’s desperation is at least partly due to an incomplete and one-sided interpretation of the argument in Kant.

At this point, followers of Sidgwick can nonetheless point out how close he comes to adopting a transcendental argument of his own in the very last paragraph of the Methods. Despite his worries about a transcendental argument involving a necessary Supreme Being, he concludes the Methods by pointing towards a transcendental argument according to which the ‘necessity to avoid a fundamental contradiction in one chief department of our thought’ might be a sufficient reason to accept the reconciliation of duty and self-interest (Sidgwick Reference Sidgwick1981 [1907]: 508–9). In the hope of finding a more scientific foundation for moral philosophy, Sidgwick pushes this option away. At the same time, though, he explicitly leaves open the possibility that a better foundation might not be available – perhaps not even in the world of the natural sciences.

As Sidgwick must have seen it, for Kant God and the afterlife provide the possibility of reaching the supreme good of merited happiness, and thereby ensure the much-needed reconciliation of duty and inclination. Arguably, the disagreement over whether this kind of reasoning is a satisfying basis for moral philosophy is a key difference between the two authors that lies behind their more specific disagreements. While it is central to the understanding of Kant’s practical philosophy as a whole, to Sidgwick it appeared to be only a second-best solution.

Kant’s understanding of humanity as an end is strongly tied to his views about the freedom of human beings and his related views about theology. A thesis about the most central difference between Kant’s categorical imperative and Sidgwick’s axioms that points in the direction of these views that Sidgwick certainly did not want to adopt and, in the case of theology, even severely struggled with, should be a competitive candidate for the right view.