I begin by stating the obvious: Paul Guyer’s Kant’s Impact on Moral Philosophy is an excellent book, of extraordinary richness, scope, and ambition.Footnote ¹ There is a huge amount to learn from it, and I have no doubt that readers will return to it repeatedly, myself certainly included. It is nonetheless most fruitful in an engagement of this kind to concentrate on our disagreements, so taking my appreciation of the book as read, that is what I will do. I home in on the immediate Kantian aftermath – the decade or so after the major practical works were published, ultimately stretching out as far as an ongoing line of nineteenth-century criticism – and one of the five general types of objection to Kantian ethics that Guyer usefully identifies as repeatedly emerging from his successors (Guyer Reference Guyer2024: 113; cited simply by page in what follows), the suggestion that Kant cannot give an adequate account of freedom and moral responsibility (115). Specifically, my focus is a perception that emerged over the period that Kant cannot intelligibly account for the possibility of wrongdoing, because it is a consequence of his theory that either acts contrary to duty are impossible, or else that they can only happen in ways that absolve those performing them of blame. I treat it as obvious that any such implication would be a disastrous reductio ad absurdum of Kant’s project, far too big a bullet for him to bite.

Guyer is well aware, of course, that these concerns were being raised but views them as largely overblown: mainly the result of confusion and addressable by clarifying certain distinctions that were always implicit in Kant and which he in fact makes clear via the Wille/Willkür distinction in the Religion (139). In what follows, I hope to persuade Guyer (or, failing that, the reader) that, first, the objection that Kant cannot give a satisfactory account of wrongdoing is weightier and more threatening to Kant’s project than he gives it credit for in the book. And second, that Kant became aware of this, at least to a degree. I will argue that this best explains a late disagreement between Kant and his disciple Reinhold, which, in my view, is still not well understood, where Kant puzzled and embarrassed Reinhold by rejecting his definition of the free will (MM, 6: 226–7), which the latter had previously offered as a helpful necessary clarification of Kant’s commitments (Reinhold Reference Reinhold1792: 263, 268), needed precisely because of the issue that had been raised about imputable wrongdoing (ibid.: 267). Guyer takes Kant’s riposte to Reinhold to concern a rather technical general issue of what it is appropriate to include within a definition (172 ff.), but I will suggest that Kant’s concern was a dialectically weightier one. Although Reinhold clearly perceives a particular peril for Kantian ethics and formulates his definition specifically in order to avoid it, there is a second peril which he doesn’t appreciate, a Charybdis as well as a Scylla, and in failing to notice it, he doesn’t grasp how accepting his definition would be to steer Kantianism away from the metaphorical dragon only to veer instead towards the whirlpool. The details of what this Charybdis is I will get to in due course, but to anticipate, the threat is to the normative authority and stringency of the moral law. Essentially, the suggestion will be that Reinhold’s proposal undermines the ground for claiming that the moral law is unconditionally binding. Since this would imply that the ‘moral law’ cannot really be a practical law at all and that the ‘categorical imperative’ would not be an imperative that is categorical, a ‘normativity crash’ of this nature would entirely upend the project of Kantian ethics. Rather a lot is potentially at stake, therefore, or so I will presently argue.

The problem regarding the possibility of wrongdoing is sometimes presented in what we might call a ‘metaphysical’ form and sometimes in a ‘motivational’ form. The metaphysical form is chronologically earliest, appearing with Schmid’s Reference Schmid1790 book that attempts to outline a systematic Kantian moral philosophy, at least as that work is usually interpreted (e.g., influentially by Kosch Reference Kosch2006: 50ff.). I make this qualification because – to complicate matters somewhat – Guyer’s understanding of the intellectual events leading up to the Reinhold-Kant dispute differs in some ways from what I take the usual understanding of them to be. I will comment on this difference shortly, but in the meantime, it is by way of this received reading that I will make my case.

On this reading, Schmid apparently notices that Kant frequently talks as if the moral law descriptively just is the will of an agent insofar as he is an intelligible being (e.g., Schmid Reference Schmid1788: 178-9/N&W 64-5) and correspondingly concludes that for Kant, insofar as an agent is free and rational, he just does act as the moral law dictates.Footnote ² But what about all those actions contrary to duty? Schmid reasons that in these cases something else must have occurred to inhibit what would otherwise have been the guaranteed decision of the intelligible self to act morally. But what? This could not be anything happening within the world of sense, as the intelligible order exists beyond the causality that governs the sensible world (Schmid Reference Schmid1790: 209/N&W 78). It follows that it must be some other intelligible occurrence, of a character inevitably inaccessible to our experience and so unknown to us (ibid.: 211/N&W 78). Hence the position he presented as the appropriate Kantian one became known as intelligible fatalism (ibid.), the implication being that from the agent’s point of view, it is a matter of sheer luck whether one acts in accordance with duty or not. Qua rational agent, your will is always disposed to do so, and you will in fact do so, unless you are somehow subject to mysterious and unpredictable noumenal inhibition. When someone’s rational will is so inhibited, they will follow the other secondary ‘source of normativity’ for human beings, sensible incentives, and may well perform a morally wrong act as a consequence. Nevertheless, the ultimate explanation for this will be neither the inclination in question itself nor a choice on their part to prioritise it, but instead the fact that the derailing noumenal event took place that undermined reason’s causal efficacy.

This rather strange proposal ultimately led to the controversy between Reinhold and Kant, because Reinhold’s definition that Kant ended up criticising was presented in response to it. Reinhold understood Schmid to be a self-identified ‘friend’ of Kant (Reinhold Reference Reinhold1792: 295) but took the intelligible fatalism with which he wanted to saddle Kantianism to be so unpalatable as to risk reducing it to absurdity, essentially because it simply recreated in another guise the very problem it was supposed to solve. After all, if it is a matter of ‘noumenal luck’ whether you act rightly or not, then how can you be responsible for it if you don’t? The impossibility of imputable wrongdoing apparently follows. Hence Kant suddenly seemed to find himself in need of defence against his own defender, an alternative account of how agents can perform morally blameworthy acts, which Reinhold provided by ascribing to human agents a free will, conceived of (and partly defined) as a capacity to determine oneself to choose either moral or immoral action (Reinhold Reference Reinhold1792: 272).

Contra the received reading, Guyer holds (143ff.) that Schmid took himself to be responding to an attack on Kantian ethics by J. A. H. Ulrich, whose position Schmid characterised as an ‘intelligible fatalism of nature’, endorseing personally a basically orthodox Kantian position on noumenal freedom in the process. Reinhold then mistakenly took Schmid to be embracing this ‘intelligible fatalism’ against Kant and steamed in with a defence of Kant’s non-fatalistic true position, which was essentially much the same as the one that Schmid was in fact proposing. When Kant unexpectedly repudiated the strategy, Reinhold’s puzzlement is perfectly understandable. But actually their dispute was a storm in a teacup, since Kant had only intended to challenge Reinhold’s presentation of the true Kantian position: specifically, his idea of what a definition might appropriately include, not his understanding of the types of actions that agents can actually choose, as Reinhold assumed. For Guyer, the whole thing is a comedy of errors (139), as, besides some minor differences of presentation, all three men basically agree on the core points. All endorse the correct and defensible Kantian position: that the free will legislates the moral law for itself (=Wille), but since its freedom also amounts to a power of choice (=Willkür), it also needs to exercise this power to determine itself accordingly, meaning that equally it might freely and imputably neglect to do so.

For what it’s worth, my interpretation of the historical engagement is closer to the received reading than to Guyer’s. Regarding Ulrich, I acknowledge that he raised against Kant the general problem of accounting for moral responsibility given the threat of determinism. But to properly grasp the dialectic we are interested in, we need to see that a distinctive objection – that Kantian moral theory implies that immoral action specifically cannot occur – was felt to need addressing. To my mind, Ulrich doesn’t do quite enough to win the honour of being the first to articulate this issue. For whilst he arguably circles around it (Ulrich Reference Ulrich1788: 33-5/N&W 13-14, and also 1788: 28, 29-31, 37ff.), he never outright states that Kant’s moral theoretical commitments imply that immoral actions distinctively either literally cannot be performed or could not be performed imputably. Instead, he always veers back to the more general suggestion that Kant has given no persuasive reason to think that agents can do other than they do, and hence that they can be morally responsible for what they do, on Kant’s conception of responsibility, whether what they do is good or evil. As I see it, it is Schmid who first suggests that a distinctive explanation of immoral actions is required of Kant when he presents the question as to why ‘the activity of reason and morality [does] not equally manifest itself in all perceptible actions’ as needing an answer (Schmid Reference Schmid1790: 208/N&W 77).

Regarding Schmid, I certainly agree that he intended to support Kant, albeit by resolving difficulties arising as much from his own reading of him as from anything in Ulrich. But I think that intelligible fatalism was intended to be the basis of that defence, as the sections following the question just mentioned (§255–257) can only reasonably be interpreted as embracing it. §255 tells us that if we want to escape groundlessness, which he has just deemed to be irrational, then to explain immorality we need to think assertorically a thought that we would otherwise only have to consider problematically, since we cannot have cognition of it by way of experience (Schmid Reference Schmid1790: 209/N&W 77). This is the thought that there might be something existing noumenally alongside the agent’s reason that is capable of restricting its operation as that would otherwise manifest itself in appearance, given that it is the noumenal as a whole that grounds the sensible as a whole. Schmid is therefore telling us straightforwardly that the theoretical explanation of immorality needs to be intelligible fatalism, as the received view understands it, or so it seems to me. None of this is countermanded by the discussion of the unfathomable character of freedom in §260 of his book (Schmid Reference Schmid1790: 220ff.), which Guyer references as evidence for Schmid’s orthodoxy (153). Nor by the fact that Schmid talks of an ‘intelligible natural fatalism’, which seems to be Guyer’s ground for claiming that Schmid means to deem Ulrich the intelligible fatalist, rather than claiming the mantle for himself (152). For Schmid straightaway emphasises that this intelligible Naturfatalismus involves noumenal rather than phenomenal causality (Schmid Reference Schmid1790: 211/N&W 78), and only the latter could be taken to apply to Ulrich, who is a straight-down-the-line soft determinist (143). Meanwhile, Ulrich had already licensed this way of characterising intelligible fatalism when he acknowledged that noumenal necessity would be ‘natural necessity understood another way’ (Ulrich Reference Ulrich1788: 32/N&W 12).

Where I do go beyond the received reading (e.g., contra Kosch Reference Kosch2006: 51–2) is in acknowledging that Schmid insisted that our ordinary concepts and practices of imputation are left unaltered by his innovation. His argument for this takes a first-person practical form, based precisely on our inability to know when or how the derailing noumenal agency might operate. According to Schmid, this means it could provide no principle or consideration relevant to deliberation about action, and hence no direct basis for an agent contemplating her moral failure to excuse it (Schmid Reference Schmid1790: 206-7/N&W 76). For Schmid, it follows that they can be explained in the usual manner and remain practically justified. Unfortunately, this argument is unpersuasive, since thinking the thought of intelligible fatalism assertorically itself amounts to an explanation for action contrary to duty, which, although lacking detail, is still sufficient to provide an indirect basis for self-exoneration. After all, if one knows that on any occasion where one’s reason is inefficacious and one doesn’t act dutifully, then ‘it could not be efficacious’ (Schmid Reference Schmid1790: 210/N&W 77), then one still seems able to judge that one cannot be blameworthy for this, despite not knowing what the source and character of the restriction on its efficacy is. Schmid’s insistence that this is merely a practically irrelevant ‘theoretical point of view’ (Schmid Reference Schmid1790: 211/N&W 77) is therefore hardly compelling.

At any rate, no one reading Schmid this way has ever thought so, as far as I’ve been able to determine. Instead, intelligible fatalism has always been considered a disastrous commitment of his version of Kantian ethics, from his initial readers down to our own time. For instance, its inability to handle imputable wrongdoing was the main complaint of an otherwise quite warm early review (Anonymous 1791), which was then reiterated repeatedly in the more straightforwardly hostile responses that followed it (e.g., Schwab Reference Schwab1792; Creuzer Reference Creuzer1793: 250–2; Fichte Reference Fichte1793: 204-5/N&W 211; Schwab Reference Schwab1794). A hostile review of the third edition of Schmid’s book focused on the same issue (Anonymous 1796). Etc. Nor have I found any contemporary source that understands Schmid as Guyer does. The upshot, I would suggest, is that it doesn’t especially matter whether Guyer or the received view is right about what Schmid intended to argue for. Either way, his role in the history of philosophy was fated to be spokesman for intelligible fatalism as the received view understands it, and as such to constitute a mere dialectical danger, a place that Kantianism must take care not to end up at as it attempts to reconcile its many commitments.

Of course, this is precisely the metaphorical Scylla that Reinhold set out to avoid. Ostensibly, he did so with ease. If you are prepared to ascribe to the human agent a will with an inherent capacity to freely determine itself to act contrary to the moral law, then you don’t need to postulate any unpredictable noumenal events to explain these actions, and the explanation you offer of them makes the actions imputable. This can appear quite straightforward and lacking any downside, as presumably it did to Reinhold. But as we know, one was perceived by Kant, who stepped up to deem this conception libertas indifferentiae (MM, 6: 226). It is true that, just like Reinhold, previous thinkers adopting this label took freedom to consist in a capacity to act otherwise than one does (e.g., Crusius Reference Crusius1744: 43-5, 54, 61-2/Wal. 50-1, 54, 55), which offers some explanation of this. But there is a definite sting in the tail to Kant’s use of it, since it was also generally taken to imply that the will exists in a condition of lawlessness (e.g., Creuzer Reference Creuzer1793: 119/N&W 167). And a will lacking its own laws could only be a heteronomous one, which indeed is the character of these thinkers’ moral theories (including that of Crusius, who tried to ground obligation in divine commands (Crusius Reference Crusius1744: 159-60, 210-2/Wal. 61-2, 68-9; G, 4: 443)). Naturally, Reinhold took himself to be a true Kantian in grasping that autonomy is the only possible ground for duty (Reinhold Reference Reinhold1792: 271) and continuing to insist that its form rather than any material factor is what makes a principle a binding categorical imperative (ibid.: 288). Likely, therefore, he found this insinuation most unfair. Certainly, he was mystified by Kant’s stubborn insistence that freedom cannot consist in any ability to act against the law, even though experience shows us that we have that (in)ability(!) (MM, 6: 226–7), protesting that the only way he could make sense of Kant’s stricture was ‘in accordance with the principles of intelligible fatalism’ (Reinhold Reference Reinhold1797: 387/N&W 245). But Kant never replied to his reply, leaving historians of philosophy with a puzzle as to why Kant had reprimanded him, since the reasons given were by no means a model of clarity.

It is therefore time for Charybdis to be properly revealed. The new peril arises because there is another dimension to the problem concerning wrongdoing, a motivational one as well as a metaphysical one, and distinctive difficulties and corresponding objections emerge along this dimension also. Our Kantian account of the free will must appropriately align with everything Kant is committed to saying about agents’ incentives, naturally including those motivating immoral action, and this is less straightforward a matter than one might initially think. Moreover, the incentives the agent needs to be presented with for such explanations to be viable must also be compatible with what Kant needs to say about the normativity of those actions, notably the key claim that the moral law is unconditionally binding upon practically rational and freely self-determining agents.

This intimate connection between incentives and practical reasons is embodied in some well-known interpretations of key aspects of Kant’s moral theory, especially his argument that ‘a free will and a will under moral laws are one and the same’ (G, 4: 446; CprR, 5: 29). Schmid notwithstanding, it is natural to interpret this argument as having a prescriptive conclusion, with the will ‘standing under moral laws’ meaning simply that it has unconditional and overriding reason to act according to the principle of the categorical imperative. All Kant’s statements of this argument are brief and obscure, but a number of commentators have thought that what is being implied is that the will has overriding reason to do so because this expresses and preserves its freedom, whilst the alternative does not (e.g., Fichte Reference Fichte1845: 54ff.; Carnois Reference Carnois1987: 72, quoting Gueroult Reference Gueroult1930: 42; Korsgaard Reference Korsgaard1996: 160–71; Morgan Reference Morgan2005: 73–8). For suppose one flouts the moral duty legislated by one’s own will and pursues the siren song of inclination instead. Then one abandons true self-determination and instead allows one’s actions to be the upshot of the causal laws of the phenomenal world, insofar as these have shaped one’s sensible desires. You thereby make yourself their puppet, the plaything of the metaphorical winds of causality. What could this offer to the rational will, which is the faculty that ultimately makes the choice between these two options? It abandons reason, it abandons freedom, so surely nothing. In which case, it would seem to follow that all the reason-grounding considerations lie on the side of morality, none on that of self-love, so acting morally is always what we have unconditional reason to do, whatever our desires might happen to be.

So understood, the argument has been subject to various challenges (e.g., Kosch Reference Kosch2006: 52ff.). But for our purposes, the main potential concern is that it threatens to prove too much, not just that dutiful action is rationally overriding, but also at the same time the impossibility of its contrary, immoral action. For it was precisely the aim of the argument to show that action contrary to duty has nothing going for it, at least from the perspective of the freely willing faculty of choice which actually makes the decision about what to do, whilst moral action has everything going for it from that perspective. But if immoral action really has nothing going for it as far as the will is concerned, how could it even find an incentive to act this way at all? Kant states in various places that no action can take place unless there is an incentive that can function as its determining ground (esp. RBMR, 6:36; G, 4:427–8), and to be incentivised is precisely to see an action as having something going for it. Admittedly, one might think the answer obvious: the incentive is the inclination that the action would gratify. But this would miss the point, as what needs incentivising is the will, and consistency with the normative argument requires that considerations of sensibility are silenced for it when compared to those of freedom.

How does this relate to the Reinhold-Kant contretemps? First, it is not clear that Reinhold has done enough to deal with this particular iteration of the problem about imputable wrongdoing. It is one thing to challenge Schmid’s presuppositions about how noumenal activity must take place, which are rooted in an assumption he shares with Ulrich that indeterminism in any form amounts to a ‘mere chance’ that is metaphysically absurd (Ulrich Reference Ulrich1788: 19; Schmid Reference Schmid1790: 187/N&W 73). It is by no means obvious that an opponent must concede this, and Reinhold in fact refused to do so when Maimon suggested that he should (correspondence reprinted in Maimon Reference Maimon1793: 229–36, at 235). But in raising a doubt about the motivational intelligibility of immoral action, the revised version of the objection places itself beyond Reinhold’s expedient of ascribing us the metaphysical capacity to φ or to ψ by questioning the possibility of the obligated will ever moving itself to exercise its capacity in opposition to morality, even if it abstractly possessed it. And second, the tension between the motivational aspects and the prescriptive aspects of Kant’s theory that this version draws out raises the uncomfortable possibility that, in principle, Reinhold’s approach to Kantian moral theory will not be able to adequately address it: specifically, because denying the will as such a fundamentally moral orientation so as to account for our acting wrongly might mean that it would then cease to be obligated not to do so.

The danger is one that the young Schelling perceives quite clearly:

The problem he raises here is the very one about autonomy we were just considering: Kantian ethics is supposed to be a distinctive and decisive advance over all previous moral systems because Kant understands that a moral precept must be a categorical one, and a principle of action can only be unconditionally binding upon the will if it is the will’s own law. Whatever the character of its proffered grounding – sentiment, divine command, ontological perfection, or whatever – every moral system presenting a heteronomous principle of morals allows a question to be raised about its authority over the agent (G, 4: 432–3), one which could only be positively answered by appeal to some condition; but of course, this would preclude it from being unconditional. To be sure, officially Reinhold accepts all this, but Schelling’s point is that he can sustain the autonomy of Kantian ethics only verbally (Schelling Reference Schelling1797: 156/N&W 255). For his approach to prioritising imputability precisely involves locating the supposed source of authoritative normativity outside of the will, in a volitionally inert faculty of reason (e.g., Reinhold Reference Reinhold1792: 294). In doing so, he cuts our presumed practical imperatives off from their root in the will’s self-legislation and allows the question to arise once again as to why the free will should take itself to be constrained by them, just as it does for the heteronomists.

For the Schelling of the 1790s, this is just a contingent error on Reinhold’s part, so it does not threaten Kantian ethics as such. As young Schelling reads him, a better proposal had already been made by Kant in the Religion (Schelling Reference Schelling1797: 169/N&W 261, citing what is now RBMR, 6: 22-3n), which is that instead of excluding the self-legislation of the law from the will to achieve the libertas aequilibrii that Reinhold seeks, we should locate both the law and its incentive and the corresponding pull towards evil internal to it. In itself, this doesn’t get us out of the woods, however, as plausibly we would then find ourselves threatened with the same categoricity collapse for a different but parallel reason. In an ethics of autonomy, the moral law is authoritative because it is the will’s own law, the principle of action generated by its own self-incentivisation to follow it. But if that is why it is supposed to be authoritative, then what happens to that authority if we acknowledge that there is another rival and contrary principle that has a similar basis? Surely it must drain away, because if the very same kind of grounding equally recommends something else than moral action, choosing one or the other of them looks optional. Indeed, within a decade or so, this is Schelling’s own view of the matter. For at least by the time of the 1809 Freiheitsschrift, he is arguing that the only way to make wrongdoing by a finite will possible is to ascribe to it an intrinsic orientation to both good and evil, and that this then makes impossible an ethics of autonomy (Schelling Reference Schelling1860: 352–3, 372).

The difficulty that Kant seems to face is that if he ascribes to the will an orientation towards lawful action only (as he appears to do in The Metaphysics of Morals, the earlier Religion notwithstanding), then wrongdoing will not be imputable, as action contrary to duty would then only be possible under conditions of intelligible fatalism. But if he ascribes to it an essential orientation either to both good and evil or to neither good nor evil, then categorical normativity collapses and heteronomy or arbitrariness follows. But what other option than one of these three does he have? The concern then is that the two problems together combine into the horns of a significant dilemma for Kantian ethics. Certainly, a particular line of post-Kantian thought, which includes the later Schelling, holds that Kant must get impaled on one or other of them and that this is enough to sink his whole project in practical philosophy (see Kosch Reference Kosch2006: esp. chs. 4–6).

Can Kant successfully address it? If he is to do so, I suggest he would need to argue that the will as such is offered incentives through which it can determine itself to act not only morally but also immorally, but in a manner where the moral incentive can be shown to carry a distinctive normative authority. He could then say that despite moving ourselves in the depths of our willing towards both good and evil, we nonetheless have reason to prefer the former to the latter. In a paper written some years ago, I offered a reading of Kant’s Religion in which his claim that there is a universal human propensity to evil is vindicated as the conclusion of an implicit transcendental argument: such a propensity is universal because its presence in the will is a condition of the possibility of us acting wrongly (Morgan Reference Morgan2005). In doing so, I unknowingly followed the lead of the Schelling of Reference Schelling1797. But albeit without properly stressing the seriousness of the corresponding threat to the law’s authority, I went somewhat further than he did then, because I did grasp that something needed to be said about why the moral incentive should be preferred to the one running counter to it, as this could not be taken for granted.

Very roughly, my argument was as follows: for an incentive to motivate the will as such, it must be an incentive of freedom, so that in pursuing the corresponding action the will could present its activity to itself as constituting it. Freedom’s true self-constitution is autonomy, Kant needs to be able to say. But there are other conceptions of freedom available, including that of ‘outer freedom’, which takes freedom to be the absence of limitation upon the faculty of choice. Acting without this constraint is valid insofar as one’s choices can cohere with those of others, but beyond that, its restriction is interpersonally justified (MM, 6: 396). It is nonetheless intelligible that the will might attempt to self-constitute its freedom as outer freedom outside of these bounds – as licence, the unlimited wilful indulgence of the agent’s own whims (Morgan Reference Morgan2005: 80–3).

It is not difficult to see how embracing unrestrained self-indulgence would make a person ‘evil’. But for someone’s action to be condemnable, they need to be obligated to avoid it, so what can be said about how the moral law can retain its authority when the wilful incentive equally emerges from the condition of freedom? My suggestion was that as a standalone notion of freedom, licence is self-undermining: first, because it cannot arrive at any determinate content for a maxim except insofar as it sources this from sensibility. ‘Let my whims be indulged!’ only amounts to a principle of action if there are grounds external to it on which whims can be based, and inclinations are the obvious candidate. But basing one’s reasons on sensible incentives throws one back into the unfreedom of heteronomy, which autonomy, by contrast, escapes (Morgan Reference Morgan2005: 82–3). And second, it is self-undermining because its own incentive is insatiable and its goal correspondingly unachievable, unlike autonomy once again (ibid.: 85). If some argument along these lines succeeds, then there may be a middle path for Kant between the Scylla of non-imputability and the Charybdis of heteronomy. There is something else I would also suggest: that his statement about freedom that bamboozled Reinhold is itself grounds to think that Kant realised this. For since assigning the moral and the immoral incentive the same status eliminates the bindingness of obligation, and Reinhold’s definition of the free will does exactly that, then rejecting it is quite understandable. Meanwhile, whilst accepting that immoral actions can indeed be chosen, his notion that this constitutes an ‘inability’ gives the self-defeating character of evil idiomatic expression. Now that we can see there are important substantive reasons for saying these things, it is reasonable to conclude they were in fact why Kant said them. Or so I would argue, space allowing.