1. The View Kant is Rejecting

In §3 of KU (5: 205–7) Kant argues that if feelings were sensations,Footnote ⁵ then the subjects in this counterfactual scenario would assess their engagement with what we in fact distinguish as the agreeable, the beautiful and the morally good according to one standard and with only one goal: the amount of ‘gratification’ they expect, that is, the amount of pleasure in the agreeable. These subjects would lack what Kant calls taste and, in particular, they would be unreceptive to the concept of duty. If we restrict the notion of happiness to gratification, we can say that the subjects deliberate exclusively according to the ‘principle of happiness’ (PoH); they would be a species of hedonists. The argument to this conclusion proceeds in two steps: (i) from the assumption that all pleasure is a sensation it follows that there can be only the pleasure of the agreeable; (ii) from this intermediary result, Kant infers further that the subjects would follow PoH in all their affairs. This is taken to be a reductio of the assumption about the nature of pleasure.

Here is the first step:

How is the ‘common confusion’ about pleasure to be understood? If feelings of pleasure were sensations, how would it follow that they all collapse into the agreeable? Wouldn’t it still hold that, for example, pleasure taken in the representation of a morally good deed is pleasure taken in a conceptual representation and therefore does not fall under the rubric of the agreeable? Kant seems to be claiming that, if pleasure is a sensation, then the three types of pleasure he generally distinguishes – in the good that ‘pleases … through the mere concept’ (5: 207), in the agreeable that pleases in sensation and in the beautiful that pleases through the mere form of a representation – cannot be differentiated any more according to their source; they all turn out to be ‘what pleases the senses in sensation’.

Kant’s result follows in this way. Pleasurable representations R – which can be ‘impressions of the senses’, ‘principles of reason’ or ‘merely reflected forms of intuition’ (5: 206) – are representations that we prefer to have rather than not have. Under the assumption that the pleasure we take in a representation R is a representation distinct or separate from whatever R occasions it, we would prefer to have this distinct representation (rather than not have it) and we would take pleasure in R only indirectly or mediately: as a means to achieve the pleasure representation. This means–end structure follows from the assumption of separateness; it leaves it open what sort of representation pleasure is. If we further assume that the pleasure representation itself is a sensation, then all pleasure falls under the notion of the agreeable as that which ‘pleases the senses in sensation’.

The target of Kant’s criticism then is the conjunction of the separateness and the sensation assumption:

If we read ‘occasions’ as ‘causes’, (A) expresses a view that has often been ascribed to Kant and in §3 of KU, at least on the face of it, Kant is explicitly rejecting this view. The result of the first step of the argument – all pleasure reduces to the agreeable – sounds very much like Beck’s presentation of Kant’s supposed view: ‘Whether the origin of the pleasure lies in some physical stimulation, the physical fulfilment of a desire, or some idea held in contemplation, the feeling is always an effect upon our sensibility … there is no place for qualitative differences’ (Beck Reference Beck1960: 93–4). Similarly, Guyer suggests that for Kant ‘all feelings of pleasure are a qualitatively uniform kind of sensation’ and that therefore pleasures are ‘internally opaque with regard to their divers … relations to their objects’ (Guyer Reference Guyer1997: 104–5).

It seems clear that Kant is arguing against a view – summarized in (A) – that prominent commentators have attributed to him, a view that Guyer admits ‘may strike us now as wildly archaic’ (Reference Guyer1997: 105). Indeed, I am not aware that any of the contemporary theories of pleasure could be characterized by (A). If we think of pleasure in terms of a ‘hedonic tone’ that colours experiences of objects or if we take pleasure to be an attitude that we take towards such experiences – views that reject the separateness thesis – Kant’s result that all pleasure reduces to the agreeable would not follow. (More on attitudinal theories in section 4.) It is, however, worth pointing out that (A) does have a tradition in eighteenth-century philosophy. One can point to Locke’s influential characterization of pleasure as a ‘simple idea’, distinct from other ideas that it may accompany, and to Hutcheson who – for the case of the pleasure of taste – postulated a special sense that enables us to perceive the ‘idea of beauty’, which is identified with the sensation of pleasure, while the other (external) senses receive other sensations of objects.Footnote ⁶ And closer to Kant’s time, Johann Nikolaus Tetens, in 1777, spent some time on refuting the view, held by some British and French philosophers, that feelings are sensations that are separate from the sensations that give rise to them and who suggested that this separation is based on a physiological difference between ‘nerves’ for receiving ordinary sensations and ‘nerves’ for receiving pleasure sensations.Footnote ⁷

Beck and Guyer support their interpretation of Kant by drawing on passages in KpV. If Kant indeed held the pleasure-as-sensation view two years before KU, there seems to be a choice: either the argument of §3, far from correcting a ‘common confusion’, is itself confused and should be ignored because it is in conflict with Kant’s real view (as expressed in KpV);Footnote ⁸ or §3 is to be taken seriously and Kant changed his mind in the short time between the second and third Critique.Footnote ⁹ A closer look at KpV shows, as I want to argue, that Kant did not reveal his real view there and that he did not undergo a change of mind between 1788 and 1790.

2. Did Kant Hold the Pleasure-as-Sensation View Before KU?

Although the textual evidence for the Beck-Guyer interpretation – that Kant affirmed the pleasure-as-sensation view in KpV – seems impressive, a preliminary hint that we should be cautious in reading him in this way is provided by the widely and highly regarded review of KpV by August Wilhelm Rehberg, published in 1788 (Rehberg Reference Rehberg and Bittner1975 [1788]). One of Rehberg’s main objections was that Kant does not explicitly come out in favour of the pleasure-as-sensation view and that he regards moral pleasure (the pleasurable aspect of respect) as a feeling and not a sensation. For Rehberg, only as a sensation could respect for the moral law have any influence on the faculty of desire. At least this reviewer thus read KpV differently than the commentators. (I return to Rehberg below.)

The relevant passages in KpV are found in the first Remark of §3 of the Analytic. Kant argues there that those philosophers who accept PoH as the only practical principle cannot legitimately introduce a distinction of a lower and a higher faculty of desire ‘by noting whether the representations which are associated with the feeling of pleasure have their origin in the senses or in the understanding [reason]’ (5: 22–3). The premises of the argument are (1) the doctrine that all desire is determined by the feeling of pleasure and displeasure (PoH), and (2) no prior distinction within the faculty of desire is assumed. The attempt to introduce a differentiation within the faculty of desire fails, because

Taken out of their context, these quotations do seem to show that Kant held that all feelings of pleasure are of ‘exactly the same kind’ as the agreeable. But the context in which he makes these pronouncements is decisive. All of the claims concerning the sameness of feelings stand, in the overall argument, under premises (1) and (2). That is, if we presuppose PoH, then all feelings are of the same kind with respect to their motivational role and further distinctions as to their origins cannot be taken to be relevant to this role. ‘If the determination of the will rests on the feelings of agreeableness or disagreeableness which he expects from any cause, it is all the same to him through which kind of representation he is affected’ (KpV, 5: 23; my emphasis). If it is ‘the agreeableness and enjoyment which one expects from the object which impels the activity toward producing it’, then ‘however dissimilar the representations of the objects … the feeling of pleasure, by virtue of which they constitute the determining ground of the will … is always the same… (ibid.). This has to be so because PoH requires that all incentives be comparable with respect to their degree or intensity; they cannot qualitatively differ from each other, which would render them incomparable.Footnote ¹⁰ The obvious differences, for example, between the pleasures from reading a book and from going hunting are made, as it were, inoperative so as to allow comparison of them, as determining grounds of desire, under PoH. Pleasures of taste, for instance, could well contribute to one’s overall happiness but how much they contribute, in comparison to other pleasures, can only be assessed once they have been rendered comparable, that is, once a ‘common currency’ has been introduced. Only then can a ranking of alternatives be established to which PoH can be applied.Footnote ¹¹

Since the homogeneity or qualitative sameness of pleasures follows from the requirements of the principle of choice, nothing can be said about the nature of pleasures if the principle is not assumed at the start. And since Kant rejects this assumption (in contrast to the philosophers addressed in the passage), he should not be understood as embracing the pleasure-as-sensation view in KpV.

If we go further back than 1788, we find numerous Reflexionen in which Kant makes it clear that feelings of pleasure are very different from each other, depending on their origins, and that only with respect to their affecting the faculty of desire do they have to be considered qualitatively the same. In 1776–8, for instance, Kant explicitly noted that ‘feelings (Fühlungen) are distinguished from sensations’ and that the former are ‘the subjective in my representations that cannot be objective’ (Refl. 552, 15: 240; cf. 551, 15: 240, and 1684, 16: 140).Footnote ¹² Throughout the 1770s he thought that even though pleasures are ‘of different kinds (verschiedener Art), even when they are sensual’, they ‘subjectively … agree among each other because each kind moves the subject’s desire; objectively [however] they cannot be compared, e.g. a meal and a virtuous action’ (Refl. 591, 15: 255 [1771–2?]). That is, feelings are ‘atomic’ when considered with respect to the faculty of desire; only in this respect do they all reduce to degrees of agreeableness:

These statements, I think, reflect Kant’s considered view on feelings which did not change between the 1770s and KU.

I return briefly to Rehberg, the acute reviewer of KpV, who thought Kant was mistaken in not treating the feeling of respect as a sensation. Since Kant was aware of the review and respected its author,Footnote ¹³ it is perhaps not unlikely that he responded in §3 to this particular line of criticism. An action, argued Rehberg, cannot be regarded as the effect of reason; to claim this would mean to commit the very mistake Kant had diagnosed in the Amphiboly section of KrV. The transition from reason to action rather has to be mediated through ‘something’ that is homogeneous with sensibility so as to subject pure reason to the determination of time, without reason becoming thereby sensible, that is, what is required here is something analogous to the schematism in the first Critique. This mediating instance is

Kant ‘accuses those of enthusiasm who take this feeling of pleasure in the law to be the moral incentive while the incentive has to consist of the law itself’ (ibid. 188). But, Rehberg concludes,

Rehberg’s criticism of the way Kant treats the feeling of respect culminates in the accusation that denying it the status of a sensation (as Kant appears to do) amounts to ‘enthusiasm’ because enthusiasm is the claim that we have an unmediated experience of the intelligible realm. Kant’s response to this, in KU, §3, may well have consisted in pointing out that the identification of feeling and sensation would not prevent enthusiasm but lead to hedonism, to precisely the scenario where rational beings are incapable of experiencing that feeling by which the law ‘immediately influences … [us] to obedience’ (KU, 5: 452).

3. How does the Assumption that Pleasure is a Sensation Lead to Hedonism?

I have argued that we cannot extract from the KpV passage any commitment on Kant’s part to the view that all pleasures are sensations. But even if this is granted, it does not shed light on his claim in KU that assuming pleasures are sensations would lead to hedonism. After all, the KpV argument presupposed PoH while the KU discussion is supposed to show that PoH follows from the assumption about pleasure.

The second step in Kant’s argument starts from the result that under (A), all pleasure is pleasure in the agreeable:

Although the argument here proceeds from the result of the first step (all pleasures are pleasures in the agreeable) to PoH as the only practical principle for the subjects in question, Kant is not narrowly concerned with moral deliberations. The somewhat obscure phrase that ‘in the end all the effort of our faculties is directed to what is practical and must be united in it as their goal’ suggests that the discussion is more generally concerned with the ways in which the deliverances of the faculties motivate us to engage with objects or actions in different contexts.Footnote ¹⁴ (‘Motivate to engage’ is chosen here to indicate that the discussion is not meant to be restricted to determinations of the faculty of desire, which would exclude the case of aesthetic response.) In the actual world – as opposed to the counterfactual scenario that Kant describes – the response to what is agreeable motivates us to procure the agreeable objects or to prolong their effect; in taste, the pleasure of taste induces us to continue contemplation of, or ‘linger’ over, the beautiful object (5: 222); and in moral contexts, the feeling of respect motivates us to do what is morally good. Our conduct with respect to these different contexts can be assessed and we assign value to what we engage in or accomplish. In the cases of taste and moral conduct we evaluate our engagement in terms of whether we appreciate the (form of) the object for its own sake and whether we perform an action that is judged to be morally good for its own sake and not as a means for some further end.

These are, for Kant, the standard ways of assessment. What happens to these practices when we assume (A) and its consequence? If the pleasures that motivate our engagement are all sensations, then what we immediately or intrinsically like are these (pleasure) sensations while the objects that occasion them are liked only as means; the liking of an object or action for its own sake is ruled out. This is the consequence of the separateness claim in (A). Assuming (A) makes it effectively impossible in the counterfactual scenario to draw the crucial distinctions of our actual evaluative practices between the agreeable, the beautiful and the good:

Under the assumption that pleasure is a distinct sensation, the pleasure in the good in itself and in the beautiful could not be immediate kinds of satisfaction in some representation R. In each of these cases what we would like immediately is the pleasure sensation rather than the having of R. The same holds for the agreeable itself: the taste sensation of the wine would not be the object of our immediate liking but rather the further pleasure sensation, produced by the taste sensation. (This is the reason why I understand Kant’s rejection of ‘all pleasures are sensation’ not as ‘there are at least some pleasures that are not sensations’ but as the stronger claim that ‘no pleasures are sensations’. The agreeable itself cannot be a distinct sensation because this would entail that agreeable objects or representations do not please immediately.)

Furthermore, according to the sensation claim in (A), pleasure is now supposed to be a sensation and from this Kant concludes that the pleasure sensations are ‘homogeneous’, that is, differences between them (if there are any) cannot be indicated by their different sources, which ‘are all entirely the same as far as the effect on the feeling of pleasure is concerned’. This does not exclude, of course, differences in intensity of the pleasure sensations. In fact, such differences – quantitative differences – seem to be the only feature that allows for discrimination of pleasures and for their comparative assessment. That intensity or degree of sensations is the only standard for assessment Kant had already argued in the passage from KpV discussed above (cf. also KU, 5: 266). But how is this claim to be defended in the present context? In KpV he had argued that such homogeneity is required by PoH, which, so I claimed, was the premise of the whole discussion. In the KU argument, however, this presupposition is not made; PoH as the only practical principle is rather supposed to be the conclusion of the argument. Kant therefore needs a different reason (independent of PoH) for the homogeneity of pleasures – which he does not provide.

I suggest that as a sensation that is occasioned by some other representation, pleasure shares with all sensations a lack of transparency as to its source (what Guyer called its ‘opaqueness’). A sensation, for Kant, does not have an intentional object, nor does it somehow indicate the object that causes it. It is a perception that ‘refers merely to the subject as a modification of its state’ (A320/B376) and only in conjunction with a concept can it become an ‘objective perception’, a representation of an object. But if neither causal origin nor intentional object can discriminate between the pleasure sensations, then it is hard to see what other standard for our liking could be applied than intensity or degree of the sensations.Footnote ¹⁵

From the homogeneity of pleasures Kant concludes that the subjects in the scenario under (A) can have only ‘one goal’ and that ‘one could not expect of them any assessment of things and their value than that which consists in the gratification that they promise’ (KU, 5: 206). The subjects can evaluate their conduct only with respect to the goal of maximized gratification, that is, in terms of how adept they are in procuring whatever means are suitable to reach that goal. In such evaluations, ‘the choice of means alone can make a difference’ and the subjects ‘could certainly blame one another for foolishness and incomprehension, but never for baseness and malice’. Thus engagement with the beautiful can be assessed, as a means, for what degree of the pleasure sensation it provides but not in terms of whether somebody has taste or not. Telling a lie can be evaluated as a means for the same goal but not as an intrinsically malicious act.

In the light of these results, how should we understand Kant’s explicit claim that in the counterfactual scenario ‘principles of reason’ have an effect on feeling? Under (A) such principles occasion sensations of pleasure or displeasure, feelings of the agreeable or disagreeable. The moral law might be one of these principles. The subjects then are aware of the moral law (the imperative to choose maxims that are universalizable); but they do not feel respect for the law, i.e. they do not feel that the law obliges unconditionally, that they are to execute actions for the sake of duty alone. Instead, under some circumstances the subjects might feel pleasure in obeying the law but this pleasure would have to be weighed against the pleasure associated with alternative action options. If it so happened that the pleasure attained from acting on principles of reason is greater than what the alternatives can deliver, then the subjects obey the law. Not, however, for its own sake, as the concept of duty requires, but for the sake of the agreeable state of mind that they expect from such obedience. The subjects would act in accordance with what moral duty requires but not from duty. In short: under (A), the moral law would lack its appropriate incentive.Footnote ¹⁶

This result – which started with (A), the claim that pleasures are sensations – is evidently a scenario in which subjects have reason (they calculate their maximal expected pleasure and are aware of the content of the moral law) but their actions cannot be imputed to them since they lack the feeling of respect for the moral law that would let them experience the obligating force of the law. In the terms of MS, these subjects possess reason (they are vernünftige Wesen) but they are not intelligible beings (they are not Vernunftwesen).

4. An Alternative View of Pleasure

If pleasure is not to be considered a distinct sensation, what is it then for Kant? KU, §3, does not contain further information about his positive view regarding feelings. For this we have to look at §10, the First Introduction of KU, lecture transcripts and indeed at the writings of some of his contemporaries. In the First Introduction and then again in §10 Kant attempted an admittedly ‘inadequate’, ‘transcendental explanation’ or ‘definition’ in terms of the influence or effect that a pleasurable representation has on the activity of the mind. This influence, he claimed, is the tendency of such a representation to maintain itself in the mind; in more technical terms, pleasure is ‘the consciousness of the causality of a representation with respect to the state of the subject, for maintaining it in that state’ (5: 220; cf. First Introduction, 20: 230).Footnote ¹⁷

This ‘maintenance’ definition was not original to Kant but can be found in a number of his contemporaries, among them Mendelssohn, Sulzer and Tetens. They all adopted what can be classified as ‘attitudinal’ theories of pleasure:Footnote ¹⁸ a pleasurable representation is one towards which we adopt an attitude that is variously described as preference or desire. This is accompanied by rejection of the view that pleasure is a distinct kind of sensation.

In 1755 Moses Mendelssohn discussed the similarities between pleasure and desire in his letters On Sentiments. In pleasure and desire, he claimed, ‘the determination of our power of representation is the same, differing only in degree’ (Mendelssohn Reference Mendelssohn and Altmann1971 [1755]: 107–8). They share a common structure because they are both directed at an object or a representation that we judge to be ‘good’ (perfect, beautiful, morally good). This judgement is followed either by the desire for the object (or for producing it) or by a ‘second judgement: I want to have this representation more than not have it’ (66). Mendelssohn has arrived here at what he took to be the commonly accepted ‘nominal definition’ (Worterklärung) of pleasure: an object gives us pleasure if we prefer having a representation R of the object over not having R (112). For the publication of the letters in the second edition of his Philosophische Schriften of 1771 Mendelssohn changed the definition: the judgement that this object is good is now followed either by the ‘desire to make the object of such a representation actual or by the striving of the mind to preserve the representation’ (Mendelssohn Reference Guyer and Altmann1971 [1771]: 258; my emphasis). This formulation is now very close to Kant’s disjunctive definition of pleasure in the First Introduction (20: 230)Footnote ¹⁹ and explicitly contains the definition in terms of the maintenance of a representation in the mind. Although the attitude we take towards pleasurable representations is described as ‘wanting it to continue’ or ‘preferring to have it’, Mendelssohn adds that this preference is based on a desire or interest. We have to ask: does this object or its representation ‘agree with the true need of a rational being?’ – where this need is our ‘inclination (Neigung) towards perfection’ (257).

This characterization of pleasure was taken up by other German philosophers in the 1770s (without attribution to a source). Johann Georg Sulzer, for instance, wrote that the sensation of warmth is agreeable (pleasurable) when we ‘desire to remain in this state or to enjoy the sensation more intensely. If the state displeases us, the force that we regard as our own essence manifests a strife for a different state’ (Sulzer Reference Sulzer1774: 1084) Johann Nikolaus Tetens, in his Philosophische Versuche über die menschliche Natur characterized the ‘most perfect satisfaction’ in having a representation of an object (i.e. pleasure) as ‘a tendency to maintain [or preserve] oneself in such a state’ (Tetens Reference Tetens1913 [1777]: 183).

The terms used by these writers suggest that they indeed took pleasure in a representation to consist in an attitude – better: a pro-attitude – we take towards having that representation.Footnote ²⁰ In his metaphysics lectures from the mid-1770s Kant made this quite vivid:

The attitude is characterized in slightly different ways – as a kind of desiring, as a preference, as a liking – but in all these cases no attempt is made to identify a special quality of feeling common to all pleasurable experiences or to define a distinct pleasure sensation. In fact, the widely held opinion among eighteenth-century philosophers that there cannot be a proper definition of pleasure may well be the result of failing to identify a common feature that would distinguish pleasure from other experiences.Footnote ²¹ Kant himself seems to refer to this difficulty when he insists that an ‘explanation of this feeling considered in general, without regard to the distinction whether it accompanies sensation, reflection or the determination of the will, must be transcendental’ (First Introduction, 20: 230), that is, in terms of a pro-attitude towards whatever kind of representation (sensation, form, concept) is in our mind.

In such an attitudinal framework Kant’s three variants of pleasure, in the agreeable, in the beautiful and in the good – which are ‘three different relations of representations to the feeling of pleasure and displeasure’ (5: 209) – are distinguished from each other by the different types of representations that give rise to them: sensations, concepts, reflected forms. What is common to them as pleasures is our attitude towards having the representations, or, in Kant’s transcendental terms, their causal role (‘the consciousness of the causality of a representation with respect to the state of the subject, for maintaining it in that state’: 5: 220).

As we saw, Mendelssohn tried to explicate the pro-attitude associated with pleasure further as involving a ‘true need’ or interest of rational beings that is satisfied by our having a certain representation. We find a similar characterization of pleasure in the second Critique:

This is apparently a quite different definition of pleasure than the one in KU. First, the KpV definition is tied to the faculty of desire, which would seem to exclude from the definition’s domain the pleasure of taste as a contemplative, or disinterested, pleasure. Second, instead of using maintenance of a representation in the mind as the characteristic mark, the KpV definition requires a pleasurable representation to satisfy the ‘subjective conditions’ of the faculty of desire. Kant does not elaborate at this point what such conditions might be. But it seems plausible to understand such a representation to ‘agree with’ or satisfy some pre-existing disposition (an interest) of the faculty of desire.Footnote ²² This agreement triggers a pro-attitude towards the representation (the feeling of pleasure) which in turn determines the faculty to produce the object of the representation.

Both definitions of pleasure can be combined if we take into account the generalization of ‘desire’ that Kant introduces later in the KpV and that Paul Guyer has discussed in his interpretation of the theory of pleasure in KU.Footnote ²³ This is the notion that faculties of the mind – not only the faculty of desire – have ‘interests’, or generalized desires, where such interests consist in ‘the conditions under which alone [the faculty’s] exercise is promoted’ (KpV, 5: 119). Agreement of the representation of an object with the ‘subjective conditions’ of the faculty of desire then means that the representation satisfies an interest of this faculty; hence, we desire to have the representation or realize its object. In the case of the contemplative pleasure of taste, we want the representation to continue in our mind: we desire to have it because it satisfies the interest of a faculty, i.e. the power of reflective judgement. Here the ‘consciousness of the merely formal purposiveness’ of the representation – its satisfying the faculty interest by inducing the free ‘play of the cognitive powers’ – ‘is the pleasure itself’ (KU, 5: 222).

Guyer has recently proposed to understand Kant’s view on the pleasure of taste in terms of a dispositional interpretation (Guyer Reference Guyer, Sorensen and Williamson2018). Although he does not refer to the distinction of §3, Guyer argues – thereby revising his own earlier view on how Kant thought about pleasure – that the pleasure of taste is best understood not as a ‘separate and unique sensation’Footnote ²⁴ but as a disposition we have: the tendency to remain in the state in which we have the representation of something beautiful. Such a dispositional interpretation, if generalized to all pleasures, rejects (A) and would therefore block the inference to hedonism in §3. But the advantage of a pro-attitude interpretation, I suggest, is that it lends itself to be joined with the theory of faculty interests. That is, the attitudinal view allows us to see how Kant’s maintenance definition of pleasure is consistent with the definition in KpV.

5. The Moral Significance of Feeling: the Wider Context of §3

I have argued that in KU Kant argued from the assumption that all pleasures are sensations to the conclusion that subjects would behave in accordance with PoH even though they are – in some sense – aware of the moral law. If the argument is granted, he has thus specified a condition under which rational subjects could fail to be moral subjects. This condition – that moral pleasure is not a sensation – is identified by considering a counterfactual scenario, a possible world in which all pleasures are sensations. This is surprising since Kant is usually taken to claim that the moral law holds, unconditionally, for all rational beings and that its reception by us as rational and sensual beings is in the form of the categorical imperative. No mention is made of a further condition concerning the nature of pleasure. Reason, even in sensually affected subjects like humans, is supposed to be practical of itself, i.e. it provides its own incentive, the feeling of respect. And Kant claims that this effect of reason on our feeling can be demonstrated a priori (KpV, 5: 73); hence it seems a necessary consequence of awareness of the law. Given this claim, it would seem that a scenario in which subjects are aware of the law and would nevertheless lack the appropriate incentive is ruled out. The blameless hedonists of the scenario are conceptually impossible and Kant’s intended reductio does not succeed.

It is true that the subjects are actually not aware of the law in the sense in which we are aware of it. We are aware of it as a categorical imperative, an unconditional command, because we are aware of both the universalizability requirement (the content) and the attendant feeling that makes the requirement obligating for us. The two ingredients, however, are conceptually distinct: the law represents some action as a duty which is, as Kant says in MS, ‘a merely theoretical cognition of the possible determination of Willkür’; to this has to be added an incentive which turns the theoretical cognition into an actual determination of Willkür (MS, 6: 218). It is a characteristic feature of Kant’s moral philosophy after KpV that he identifies a special disposition or capacity for feeling as a condition for subjects to be receptive for the incentive of the law. He made this disposition explicit for the first time in the Religion of 1792/3 where he presented three dispositions that can be assigned to human beings:

1. 1. The predisposition to the animality of the human being, as a living being;

2. 2. To the humanity in him, as a living and at the same time rational being;

3. 3. To his personality, as a rational and at the same time responsible being. (6: 26)

In a long footnote added to this distinction, Kant elaborated on the differentiation of dispositions 2 and 3. Personality is not ‘already included in the concept of’ humanity, but must be treated as a ‘special predisposition’. From the fact that a being has reason – theoretical reason, including the ability for prudential calculations – it does not follow that ‘this reason contains a faculty of determining the power of choice unconditionally by virtue of simply representing its maxims as suited to universal legislation’ (6: 23n.; trans. modified). Such a ‘merely’ rational being – even a maximally rational being – would not have an inkling of something like the moral law that commands unconditionally. A most rational being (allervernünftigstes Wesen) therefore could be, as Kant puts it in MS, ‘morally dead’ (6: 400), unreceptive to the obligating force of the moral law.

Thus a few years later Kant made the same point in different terminology: a human being considers herself ‘first as a sensible being, that is as a human being (a member of one of the animal species), and secondly as an intelligible being (Vernunftwesen) (not merely as a being that has reason (vernünftiges Wesen), since reason as a theoretical faculty could well be an attribute of a living corporeal being)’ (MS, 6: 418). By analogy to the predisposition to personality of the Religion, in MS Kant identifies ‘natural predispositions of the mind … for being affected by concepts of duty’, capacities for feeling that are given by nature to ‘every human being’ (6: 399). These capacities are, as it were, activated by consciousness of the moral law and generate what Kant now calls ‘moral feeling’ (among other kinds of morally relevant feeling). He explicitly denies that any actual human lacks these predispositions (6: 400), but it would seem to be a contingent fact about us that nature equipped us in this way. Considerations of counterfactual scenarios in which subjects lack these predispositions should therefore be legitimate. It becomes a conceptual possibility that subjects have ‘theoretical cognition’ of the law without experiencing it as a categorical imperative because they lack the required disposition.

It is a contingent fact about us that we are conscious of the law as a categorical imperative because we happen to have inclinations that can conflict with the law. But a holy will who is not afflicted in this way is a conceptual possibility. When Kant identifies a special capacity for feeling in Religion and MS as a contingent part of our nature, it becomes possible to consider further counterfactual scenarios about what happens when the condition is not satisfied. Only if the condition is in place does the necessary relation between the law and its incentive in the form of a feeling of respect hold.

That such a special capacity for feeling is required is an insight that Kant seems to have fully realized only in his later works. I suggest that the discussion in §3 of KU is in the spirit of this insight: Kant describes subjects that are (in the attenuated sense) aware of the moral law but are incapable of experiencing it as a categorical imperative because for them feelings are always sensations which, for the reasons given in section 3 above, prevents them from having the capacity for moral feeling. (It is interesting that at the end of the Dialectic of KpV Kant considers a counterfactual situation in which subjects are prevented from feeling respect for the law – and hence follow PoH – not out of a lack of the capacity for moral feeling but out of an excess of knowledge.Footnote ²⁵ )

I would like to sketch – very briefly and speculatively – how Kant’s views on the relation of theoretical and practical reason developed and why the topic of moral feeling became more prominent in his works after 1788. From the point of view of his moral writings before KpV, the distinction of rational beings who possess theoretical reason (or technical-practical reason, in the terminology of KU, 5: 172) but lack (morally) practical reason must seem strange indeed. In the Transcendental Dialectic of KrV and in GMS he had arguedFootnote ²⁶ that there is a ‘transition’ from the fact that a creature possesses theoretical reason – a spontaneous capacity of the mind – to the claim that such a creature must therefore also be subject to the moral law, i.e. possess practical reason. In these works, as well as in lectures and notes from the 1770s, Kant seems to have held that the spontaneity that comes with theoretical reason indeed is sufficient for having moral status. If he still held this view in 1790, the reductio of §3 could not succeed, because beings who are aware of principles of reason would inevitably possess the kind of spontaneity that makes them subject to the moral law.

In GMS III Kant argued that it is quite possible – ‘as speculative philosophy can show’ (in KrV) – to presuppose freedom of the will without creating a conflict with the principle of natural necessity (4: 461). And for every rational being that is conscious of itself as having a will it is, ‘without further conditions’, even necessary to think of its actions as free (ibid.). Speculative philosophy presumably has shown, in the resolution of the third antinomy: (i) that transcendental freedom can be thought without contradiction if an intelligible realm is admitted, though it is true that within the sensible world it cannot be thought coherently; and furthermore (ii) that it is also possible to attribute such freedom to human beings if it can be demonstrated that humans are not only sensible creatures but also participate in the intelligible world. The argument in KrV is as follows:

The argument is repeated in GMS III (4: 451–2) after the moral law has been established as binding for any purely rational will. Since humans are not purely rational wills – they are sensibly affected by inclinations – the question is: why should the law also be binding for them? As in KrV, Kant infers from the fact that we have theoretical reason (the ability to form ideas) that we are not merely sensible beings but also intelligible subjects, and that the moral law is therefore binding for us as an unconditional imperative.Footnote ²⁷

If this was Kant’s view on the relation of theoretical and practical reason up to and including GMS, then it seems clear that on this basis a dissociation of humanity and personality is not possible. By the time of KpV, however, Kant had clearly withdrawn this doctrine. He considered a scenario, for instance, in which the actions of a being ‘have their determining ground … in the causality of a supreme being which is distinct from him’, as a ‘marionette’. Such a being could well be equipped with self-consciousness, with the ability to think, but ‘the consciousness of his own spontaneity, if taken for freedom, would be a mere delusion” (KpV, 5: 101; cf. the automaton spirituale at 5: 97).Footnote ²⁸ That is, a being could have the spontaneity of apperception and still lack freedom in the transcendental sense.Footnote ²⁹

How did Kant come to realize that there is no ‘transition’ from spontaneity in the theoretical sense – the ability to form ideas, concepts without possible instantiations in experience – to spontaneity in the practical sense? Many commentators have suggested that Kant saw, after the publication of GMS, that the argument considered above is faulty. He indeed seems to have retracted the argument in KpV (5: 46) and presented instead the doctrine of the ‘fact of pure practical reason’ (5: 31).Footnote ³⁰

In this framework the scenario of §3 of KU becomes possible. Kant can now conceive of rational beings that are not able to feel the obligating force of the moral law because they lack the appropriate feeling ‘by which the moral law immediately influences [them] to obedience’ (5: 452). It would be a situation in which there are ‘principles of reason which determine the will’ (5: 206) but the subjects would only know the content of such principles without experiencing their special obligating force.

Kant thus explicitly introduces a gap into the spontaneous faculties. Theoretical reason (including the understanding) is dissociated from practical reason in a way that contrasts significantly with his earlier views. What takes on a more prominent role for Kant is the capacity for moral feeling because it is this capacity that, as it were, bridges the gap.

In sum, my suggestion is that only if we assume that, by the time of KU, Kant has rejected the view he held in GMS does the argument of §3 make sense. The argument shows that if we subscribe to the common confusion of pleasure and sensations, i.e. if we accept (A), then we would just have ‘humanity’ but not ‘personality’, we would have theoretical awareness of the moral law (we would know its content) but we would not feel obligated by it. The rejection of (A) is thus a condition for experiencing moral feeling.

This development, I suggest, explains why after 1788 Kant placed so much emphasis on ‘finding ways in which morality, based in freedom, can be made accessible to feeling as well as reason’ and that ‘every means for cultivating moral feeling … must be seized’ (Guyer Reference Guyer2005: 225). Thus increased focus on feeling is due not so much to his belated realization that the doctrine of KpV was too austere for humans. In my view it is more likely the loss of the ‘transition’ from theoretical spontaneity to moral status – the new split within the spontaneous faculties – that required the more prominent role of moral feeling as the missing link between the faculties. Kant of course had not earlier ignored the role of feeling in moral beings; but what is clearly new, after 1788, is the significance such feelings take on for him because the difference between rationality and morality consists precisely in the capacity for moral feeling.Footnote ³¹